ValCitGlyPro-dexamethasone antibody conjugates selectively suppress the activation of human monocytes

Abstract

Glucocorticoids (GCs) are effective in treating autoimmune and inflammatory disorders but come with significant side effects, many of which are mediated by non-immunological cells. Therefore, there is rapidly growing interest in using antibody drug conjugate (ADC) technology to deliver GCs specifically to immune cells, thereby minimizing off-target side effects. Herein, we report the study of anti-CD11a, anti-CD38, and anti-TNFα ADCs to deliver dexamethasone to monocytes. We found that anti-CD11a and anti-CD38 were rapidly internalized by monocytes, while uptake of anti-TNFα depended on pre-activation with LPS. Using these antibodies were attached to a novel linker system, ValCitGlyPro-Dex (VCGP-Dex), that efficiently released dexamethasone upon lysosomal catabolism. This linker relies on lysosomal cathepsins to cleave after the ValCit sequence, thereby releasing a GlyPro-Dex species that undergoes rapid self-immolation to form dexamethasone. The resulting monocyte-targeting ADCs bearing this linker payload effectively suppressed LPS-induced NFκB activation and cytokine release in both a monocytic cell line (THP1) and in human PBMCs. Anti-TNFα_VCGP-Dex and anti-CD38_VCGP-Dex were particularly effective, suppressing ∼60–80% of LPS-induced IL-6 release from PBMCs at 3–10 μg mL−1 concentrations. In contrast, the corresponding isotype control ADC (anti-RSV) and the corresponding naked antibodies (anti-CD38 and anti-TNFα) resulted in only modest suppression (0–30%) of LPS-induced IL-6. Taken together, these results provide further evidence of the ability of glucocorticoid-ADCs to selectively suppress immune responses, and highlight the potential of two targets (CD38 and TNFα) for the development of novel immune-suppressing ADCs.

A series of immune-suppressing antibody-drug-conjugates were prepared using a novel tetrapeptide self-immolative linker, ValCitGlyPro, that is rapidly cleaved by lysosomal proteases to release dexamethasone.

Introduction

Glucocorticoids (GCs) are a highly efficacious first-line treatment for a wide variety of autoimmune and inflammatory disorders. However, this efficacy comes with a significant risk. Studies have demonstrated that >90% of patients who take GCs for more than 60 days exhibit one or more known side-effects.¹ For example, doses as low as 2.5 mg per day of prednisolone are associated with increased risk of hip and vertebral fractures,² doses of 5–10 mg per day are associated with 4–8% mean body weight gain,³ and doses of >7.5 mg per day are associated with a 2–4 fold increase in the risk of heart failure.⁴ Importantly, most of these side effects are mediated by non-immunological cell types. For instance, GCs are known to inhibit osteoblast function resulting in reduced bone density, to block pituitary release of cortisol resulting in adrenal insufficiency, and to upregulate hepatic enzymes required for gluconeogenesis resulting in hyperglycemia.⁵ Indeed, GC receptors are found to be expressed in nearly all nucleated cells in the body.⁶ Thus, minimizing GC exposure in these tissues while maintaining exposure in immunological tissue provides a viable strategy for amelioration of many of the side-effects frequently associated with treatment.

With this in mind, we turned our attention to antibody drug conjugate (ADC) technology that is now widely recognized for its ability to promote tissue-specific delivery of otherwise highly cytotoxic drugs. Numerous ADCs have been designed to target various hematological malignancies, thus broadly validating the concept of selective delivery to lymphocytes and myeloid cells. Building on this, a number of groups have recently published strategies for delivering GCs to various cell types for the treatment of inflammatory disorders. This published work generally falls into three categories: 1) targeting the CD163 receptor to deliver an ester-linked dexamethasone to macrophages,^7,8 2) targeting CD70 or CD74, widely expressed on lymphocytes, to deliver phosphate-linked glucocorticoids,^9,10 3) targeting transmembrane TNFα to delivery ultra-potent budesonide derivatives to TNFα-expressing cells.^11,12 This latter approach has proven successful in a variety of preclinical models and has now advanced into phase II clinical studies.¹³

Given our team's long-standing interest in ADC linker technology and immune-modulating ADCs, we recently undertook a project to design ADCs to deliver dexamethasone to broad sets of lymphocytes and myeloid cells.^14–16 For the purposes of optimizing the design of the ADC, we began with a focus on monocytes. Monocytes are a key component of the innate immune system and play an important role in a variety of immunological disorders. Monocytes respond to a diverse set of pathogen-associated molecular patterns (PAMPs) and danger associated molecular patterns (DAMPs) by virtue of various TLR receptors expressed on the cell surface and endosome. As such, they are rapidly recruited from the bloodstream into tissues in response to inflammatory stimuli, wherein they undergo differentiation into various antigen presenting cells (APCs), including dendritic cells and macrophages. Macrophages, in turn, can promote tissue remodeling and degradation associated with chronic inflammatory diseases while dendritic cells serve as APCs that engage with T-cells, kicking off the pathway towards adaptive immune response. Moreover, monocytes, and the cells derived from them, are widely known to release pro-inflammatory cytokines including interleukin-1 (IL-1), tumor necrosis factor-alpha (TNFα), and interleukin-6 (IL-6), among others. These cytokines amplify the inflammatory response by attracting other immune cells to the site of inflammation, furthering the cycle of chronic inflammation.

We envisioned designing monocyte-targeting ADCs that would function as outlined in Fig. 1. Activation of TLRs by DAMPs/PAMPs results in the NFκB-mediated upregulation of various surface markers and cytokines, including CD38, TNFα, and IL-6. Glucocorticoid-ADCs which target monocyte surface markers, such as transmembrane TNFα (tmTNFα), CD38, and CD11a, will be endocytosed by the classical clathrin-mediated uptake process, wherein the ADC is shuttled to the lysosome and undergoes proteolytic degradation. The released glucocorticoid (dexamethasone) binds to the glucocorticoid receptor (GR) wherein it dimerizes, translocates to the nucleus and binds to the glucocorticoid response element (GRE) and promotes the transcription of various anti-inflammatory proteins, including glucocorticoid-induced leucine zipper protein (GILZ). Simultaneously, the activated GR (monomer or dimer) binds to various pro-inflammatory response elements and thereby suppresses the expression of pro-inflammatory cytokines and pro-inflammatory surface markers.⁶

Fig. 1. Schematic illustrating the mechanism of immune-suppressing ADCs. The process begins (1) with the binding of a TLR ligand and subsequent activation of the NFκB pathway, resulting in (2) the production of cytokines, including tmTNFα. (3) The tmTNFα is cleaved by TACE to form soluble TNFα (sTNFα). (4) Binding of an anti-TNFα, anti-CD11a, or anti-CD38 ADC results in internalization and (5) release of dexamethasone directly inside TLR-activated cells. Note that the anti-TNFα ADCs will have a dual effect – both delivering a payload to TLR-activated cells and also neutralizing soluble TNFα that has been released by TACE.

The design of ADCs that function as outlined in Fig. 1 required us to overcome two key hurdles. First, we needed to identify suitable markers on monocytes that undergo rapid internalization. Second, we need to design a suitable linker that can be rapidly cleaved in the lysosome to release dexamethasone. Herein, we will describe our efforts to overcome both of these challenges. This work builds on our recently published work describing the stability of an ester-linked dexamethasone ADC.¹⁵ In our previous publication, we demonstrated that a CD11a-targeting ADC linked to dexamethasone via an ester linkage was not readily cleaved by lysosomal enzymes, but was able to modestly suppress LPS-induced TNFα production. Herein, we expand on this work by exploring various monocyte-targeting antibodies and by redesigning the linker to more efficiently release dexamethasone upon lysosomal uptake.

Results and discussion

Antibody internalization and selection of antigens

THP1 is a monocytic leukemia cell line that is widely used for the study of monocytes and macrophages. This cell line expresses most of the common markers associated with monocytes and is widely employed as a model system for the study of inflammatory processes.¹⁷ Using THP1 cells as a model for human monocytes, we evaluated the lysosomal uptake of three antibodies that target surface markers known to be expressed on monocytes: anti-CD11a, anti-CD38, and anti-TNFα.

• CD11a is a subunit of leukocyte function-associated antigen-1 (LFA-1), an integrin that is expressed on various immune cells, including T cells, B cells, and monocytes.¹⁸ An antibody against CD11a, known as efalizumab, was approved by the FDA for the treatment of psoriasis, but withdrawn from the market in 2009 due to a rare brain infection that was observed in some patients. Efalizumab is known to be internalized into various cell types, and has been previously employed used for the design of an immune-suppressing ADCs.¹⁹

• CD38 is a cell surface protein that is widely expressed on various immune cells and hematopoietic cells, including B-cells, T-cells, NK cells, and monocytes. An anti-CD38 antibody, daratumumab, is currently approved for the treatment of multiple myeloma.

• TNFα is often considered to be the “master cytokine” in rheumatoid arthritis (RA).^20,21 It is initially produced as tmTNFα which is cleaved by TNFα converting enzyme (TACE) to produced soluble TNFα (sTNFα). Antibodies to tmTNFα are known to be internalized and trafficked to the lysosome.²² Anti-TNFα antibodies, such as adalimumab, have revolutionized the treatment of RA and related disorders by depleting the plasma levels of this pro-inflammatory cytokine, thus leading to a reduction in downstream cytokines, such as IL-1, IL-6, and IL-8.

The uptake of anti-CD11a (efalizumab), anti-CD38 (daratumumab), and anti-TNFα (adalimumab) into the lysosomes of THP1 cells was evaluated using pHAb-labeled tagged antibodies. pHAb (Promega) is a pH sensitive dye that dramatically increases in fluorescence at lysosomal pH. Antibodies were labeled with pHAb using traditional cysteine conjugation chemistry. THP1 cells were treated with the labeled antibody in the presence or absence of 10 ng mL⁻¹ of LPS. At various time points, cells were removed, fixed with formaldehyde, and analyzed by flow cytometry. Gating was performed, as shown in Fig. 2A, grouping cells into internalization “positive” and “negative” groups. Full results of the flow cytometry can be seen in Fig. S1.† As anticipated, anti-CD11a and anti-CD38 are rapidly internalized into THP1 cells while the nontargeted control antibodies, anti-RSV and anti-Her2, are not. Anti-RSV targets a viral capsid protein on respiratory syncytial virus (RSV) while anti-Her2 targets receptor tyrosine-protein kinase erbB-2 (Her2), expressed in a variety of breast and GI cancers. As anticipated, neither the anti-Her2 or anti-RSV exhibited any significant uptake into THP1 cells, validating their use as isotype controls in subsequent studies. Notably, the internalization of anti-CD38 increases slightly upon stimulation with LPS as evidenced by an increase in MFI (∼2.7× vs. 3.5×) and an increase in positivity (∼40% vs. ∼80%). However, most impressively, the internalization of anti-TNFα was shown to be completely dependent upon pre-activation with LPS. In the absence of LPS, virtually no internalization of anti-TNFα was observed while in the presence of LPS the MFI rapidly increased in just 2 h, plateauing at ∼4 h. (Fig. 2B and C) In the presence of LPS, ∼50% of cells were positive for the pHAb dye by 24 h, while in the absence of pHAb dye only ∼3% of cells were positive for the dye. (Fig. 2A and C).

Fig. 2. Internalization of 10 μg mL⁻¹ pHAb-labeled antibodies into THP1 cells in the presence and absence of 10 ng mL⁻¹ LPS. A) Flow cytometry illustrating the categorization of non-fluorescent and fluorescent (boxed) cells upon exposure to pHAb-labeled anti-TNFα or anti-CD11a; B) the increase in MFI over time in the presence (right) and absence (left) of LPS; C) the increase in fluorescent cells over time in the presence (right) and absence (left) of LPS.

In order to confirm these results, a set of cytotoxic ADCs prepared from the same antibodies were assessed in a 96 h cell killing assay using a well-studied tubulin inhibitor, vcMMAE, at an ADC payload (Fig. 3). vcMMAE was attached to the hinge cysteine residues of the corresponding antibodies using the same chemistry as used for the pHAb conjugates (endogenous cysteine conjugation). Consistent with the flow cytometry data, the anti-CD11a vcMMAE conjugate was highly cytotoxic against the THP1 cells, irrespective of the presence of absence of LPS (IC₅₀ ∼0.01 μg mL⁻¹). The non-targeted isotype control ADC (anti-Her2) was non-toxic (IC₅₀ >10 μg mL⁻¹) in both the presence and absence of LPS. Interestingly, however, the toxicity of the anti-TNFα vcMMAE conjugate was dependent on the presence of LPS. The addition of 10 ng mL⁻¹ LPS increased the cytotoxicity ∼5-fold, as seen in Fig. 3 (from 0.88 μg mL⁻¹ to 0.18 μg mL⁻¹).

Fig. 3. Cytotoxicity of MMAE ADCs in THP1 monocytes.

Linker design

Having identified a few suitable antigens for targeting THP1 monocytes, we next turned our attention to the design of the linker chemistry for releasing dexamethasone. The ADC linker is responsible for the release of the payload upon uptake to antigen expressing cells, but is also essential for the stability of the ADC in circulation. There are several ADC linker technologies designed to be cleaved by lysosomal enzymes such cathepsin B, beta-glucuronidase, legumain, sulfatase, and phosphatase.^6–10 The majority of these methods, however, have been focused on the release of amine-containing payloads. There are still relatively few methods and linker systems that release payloads containing a hydroxyl functional group.²³ Dexamethasone contains a C-21 alcohol which can be readily functionalized with an ester for the attachment of a linker and conjugation handle. Indeed, this is the strategy employed by Moestrup and colleagues in their preparation of anti-CD163 macrophage targeting ADCs.^7,8 However, we recently showed that simple esters (i.e.Fig. 4, Rxn 1) are not readily cleaved by lysosomal enzymes and thus are generally unsuitable for delivery of dexamethasone.¹⁵ Instead, we designed a small set of amine-containing self-immolative spacers that are sterically constrained in order to promote rapid lactam formation upon cleavage of a peptidic bond. In this manner, we would be able to rely on traditional proteases such as cathepsin B or legumain that are known to be highly active in lysosomes. The end results would, in effect, be a “protease cleavable ester”.

Fig. 4. The design of self-immolative spacers to release dexamethasone (red star). Simple esters are not readily cleaved by lysosomal enzymes (Rxn 1), and therefore the focus of this work is on protease-cleavable linkers that undergo spontaneous lactone formation to release an alcohol-containing payload (Rxns 2–4).

In order to test this design strategy, we looked at the immolation rate of esters of 2-amino phenylacetate and 2-aminomethyl benzoate (Fig. 4, Rxn 2–3). These compounds proved to be stable as TFA salts, but underwent rapid cyclization to the corresponding lactams at pH 7.4 and more slowly at pH 5.2 (Fig. S2†). Similarly, a model GlyPro ester underwent rapid diketopiperazine formation within ∼1 h at pH 7, releasing the pendant alcohol (Fig. S3†). Esters of X-Pro dipeptides are known to rapidly cyclize to form diketopiperazines at neutral pH.^24,25 In fact, SerPro esters have been used as a self-immolative spacer for a previously published prodrug of vinblastin.²⁴ Likewise, similar prolinol-derived spacers have also been shown to undergo rapid cyclization to release alcohol-containing payloads.²⁶ While this work gave us confidence in the self-immolation of the shown spacers, it was unknown how the addition of these spacers would impact cleavage by lysosomal proteases, such as CatB. With this in mind, we prepared three model compounds (1–3, Fig. 5) in which the amine of the self-immolative spacer was attached to a ValCit dipeptide, a sequence known to be rapidly cleaved by CatB and other cysteine proteases.²⁷ The rate of CatB cleavage was assessed by LCMS by monitoring the disappearance of starting material (1–3) and the formation of Fmoc-ValCit-OH followed later by Fmoc-Val-OH, as shown in Fig. 5. The cleavage rate of each linker was compared to the well-studied ValCit-PAB comparator (4). Compounds 1 and 2 underwent slow cleavage, requiring 24–48 h to achieve significant proteolysis, perhaps due to the steric hinderance imparted by the ortho substituent on the aromatic ring. In contrast, the ValCitGlyPro peptide (3) underwent rapid cleavage, achieving >50% proteolysis within a few hours.

Fig. 5. Evaluation of CatB cleavage rate of various self-immolative model systems.

Preparation and metabolism of ADCs

Having demonstrated the rapid cleavage of the ValCitGlyPro peptide, we proceed incorporate a dexamethasone payload onto the C-terminal end of the peptide linker and a maleimide conjugation handle on the N-terminal end (Scheme 1). Mal-ValCitGlyPro-OH (5, Genscript) was coupled to C-21 hydroxyl of dexamethasone using EDC/HOBt chemistry resulting in the isolation of VCGP-Dex (6) in 36% yield. Compound 6 was attached to three monocyte-targeting antibodies (anti-CD11a/efalizumab, anti-CD38/daratumumab, and anti-TNFα/adalimumab) and two non-targeted control antibodies (anti-Her2/trastuzumab and anti-RSV/palivizumab). The conjugation was accomplished using traditional TCEP-promoted cysteine conjugation to the endogenous disulfides connecting the heavy chain (HC) and light chains (LC). The final conjugates exhibited drug-antibody ratios (DAR) ranging from 2.5 to 7.5, as confirmed by LCMS (Importantly, the two nontargeted control ADCs exhibited the highest DAR, thus minimizing the risk that inactivity of the non-targeted controls is simply due to a lower drug-antibody ratio.). Notably, little or no aggregation was observed by SEC for any of the bioconjugates.

Scheme 1. Preparation of VCGP-Dex bioconjugates.

As mentioned, we previously demonstrated that simple esters of dexamethasone (i.e. mc-Dex) are highly stable in the lysosome and therefore are not ideally suited for ADC designs. In order to demonstrate the superiority of our newly designed VCGP-Dex linker system, we evaluated the lysosomal stability of one of the conjugates, anti-Her2_VCGP-Dex, in a head-to-head study with anti-Her2_mc-Dex (Fig. 6). ADCs were incubated with pre-activated rat liver lysosomes (Xenotech). Aliquots were removed at various time points, quenched with PBS, and then analyzed by LCMS for the presence of dexamethasone. As shown in Fig. 6, the VCGP-Dex ADC exhibited rapid release of dexamethasone while the simple ester ADC mc-Dex exhibited only trace release. This result clearly indicates the functional superiority of the protease-cleavable ValCitGlyPro linker as compared to a traditional esterase cleavable linker. Notably, complete release of Dex occurs within ∼1 h at pH 7.4 (Fig. S3†) while little or no release occurs at pH 2 (data not shown). This suggests that lysosomal escape of the GlyPro-Dex is essential for functional activity.

Fig. 6. Comparison of lysosomal release of dexamethasone in two ester-containing linker systems. (100% release corresponds to ∼800 nM).

Suppression of NFκB activation and cytokine release

The immune-suppressing activity of the monocyte-targeting VCGP-Dex ADCs was initially assessed using a THP-1 monocyte reporter cell line, THP-1 Dual (Invivogen). THP-1 dual monocytes contain both a NK-kB reporter (secreted embryonic alkaline phosphatase, SEAP) and an interferon regulatory factor (IRF) reporter (secreted luciferase, Lucia), allowing facile monitoring of these two pro-inflammatory pathways. NFκB and IRF are transcription factors that are downstream of multiple TLR receptors, and are both known to promote the release of various pro-inflammatory cytokines. Glucocorticoids are known to inhibit the NFκB pathway by inducing the expression of I kappa B alpha, a protein which directly inhibits NFκB by masking its nuclear localization sequence.²⁸ The THP-1 NFκB/IRF reporter cell line was incubated with test ADCs in the presence of 10% human serum and was subsequently stimulated for 16 h with 1 ng mL⁻¹ LPS to induce NFκB and IRF activation. As seen in Fig. 7, NFκB signaling was increased ∼7 fold over background (non-stimulated) levels. The LPS stimulation only resulted in a slight (∼2×) activation of the IRF pathway, and therefore the focus of subsequent experiments was on the NFκB pathway. Initial ADC screens were performed at 6 μg mL⁻¹ and 30 μg mL⁻¹. (Fig. 7) We anticipated that the monocyte-targeting ADCs would block the NF-κB signaling pathways while the naked mAb and the control (nontargeted) ADC would not. As can be seen in Fig. 7, the situation we observed was a bit more complicated. The unmodified anti-CD11a antibody unexpectedly activated the NFκB pathway and promoted the production of IL-6 and IL-8. The corresponding anti-CD11a_VCGP-Dex ADC induced significantly less NFκB activation and cytokine production than the naked antibody, but still higher than the non-treated control. In contrast, the anti-CD38_VCGP-Dex and anti-TNFα_VCGP-Dex ADCs suppressed NFκB, IL-6, and IL-8 as compared to the non-treated control. Importantly, this effect was not driven by the antibody alone, as evidenced by the fact that the ADCs were significantly more active than the corresponding unmodified anti-CD38 and anti-TNFα antibodies (with the exception of IL-6 production in the 30 μg mL⁻¹ dose of anti-CD38_VCGP-Dex). As might be anticipated, the unmodified anti-TNFα exhibited some immune-suppressing activity on its own, but this activity was significantly enhanced by the addition of the dexamethasone payload. The anti-RSV_VCGP-Dex did not significantly suppress the NFκB pathway, and only exhibited a slight suppression of IL-6 and IL-8.

Fig. 7. Evaluation of immune suppression of test ADCs in LPS-stimulated THP1-dual cells (24 h incubation) as assessed by A) NFκB/SEAP induction compared to a non-stimulated control; B) IL-8 release; and C) IL-6 release.

The upregulation of the NFkB pathway and induction of cytokine production caused by the unmodified anti-CD11a antibodies was unexpected, but not without precedent. CD11a is one subunit of LFA-1, an integrin that is involved in leukocyte adhesion and migration. Studies have shown that cross-linking of LFA-1 or depletion of LFA-1 subunits can result in NFκB activation, TNFα production, and IL-1 production in monocytes and macrophages.^29–31 However, it was unclear whether the activation by anti-CD11a shown in Fig. 7 was perhaps exaggerated due to the use of a reporter cell line that was specifically designed to be highly sensitive to NFκB activation. Therefore, we evaluated the anti-CD11a_VCGP-Dex ADC on the parental THP1 cell line, using TNFα release as a surrogate for NFκB activation. As can be seen in Fig. 8a, the unmodified anti-CD11a antibody exhibited only a slight (∼20–25%) induction of TNFα while the anti-CD11a_VCGP-Dex ADC completely ablated the LPS-induced TNFα release at concentration of ≥3 μg mL⁻¹. Notably, the IC₅₀ of the ADC for suppression of LPS-induced TNFα was ∼0.3 μg mL⁻¹, while the IC₅₀ of the corresponding isotype control ADC (anti-Her2_VCGP-Dex) was ∼3 μg mL⁻¹. The activity of the isotype control ADC may arise from nonspecific internalization via micropinocytosis or Fcγ mediated uptake, as is widely known in the ADC field. Alternatively, small amounts of Dex may be released from the ADC during the 24 h incubation, thus driving some of the non-targeted mediated activity (Note Fig. S4†). Importantly, neither anti-CD11a_VCGP-Dex or anti-Her2_VCGP-Dex exhibited significant cytotoxicity either in the presence or absence of LPS, indicating that the ADC is blocking the NFκB or TNFα signalling cascade rather than simply killing the TNFα expressing cells. (Fig. 8B).

Fig. 8. A) Anti-CD11a_VCGP-Dex suppressed TNFα release in THP1 monocytes far more effectively than the naked mAb and a non-targeted control ADC; B) VCGP-Dex ADCs were non-toxic to THP1 monocytes at doses up to 30 μg mL⁻¹; C and D) anti-TNFα ADCs suppressed LPS-induced NFκB activation in THP1 monocytes more effectively than either the naked mAb or a non-targeted control ADC, both in the absence (C) and presence (D) of a TACE inhibitor.

Perhaps the most interesting findings from the study of THP1-dual cells in Fig. 7 was the finding that the anti-TNFα_VCGP-Dex ADC suppressed NFκB activity and cytokine production significantly more than the unmodified antibody. In order to further understand and verify this finding, we performed a dose titration of anti-TNFα_VCGP-Dex, unmodified anti-TNFα, and anti-Her2_VCGP-Dex both in the presence and absence of a TACE inhibitor. (Fig. 8C and D) We hypothesized that inhibiting the cleavage of transmembrane TNFα with a TACE inhibitor may enhance the uptake of the anti-TNFα ADC, thereby enhancing the anti-inflammatory effects. As seen in Fig. 8C, stimulation of the THP-1 monocytes with 1 ng mL⁻¹ LPS increased the production of NF-κB activity ∼8–10 fold. Consistent with the data previously described, the anti-TNFα_VCGP-Dex ADC resulted in a dose-dependent reduction in NF-κB signalling. The naked anti-TNFα mAb also suppressed NF-κB signalling, but not nearly as significantly as the ADC. The isotype control anti-RSV_VCGP-Dex did not reduce NFκB signalling. Addition of 10 μM TAPI1, a well-studied TACE inhibitor, only slightly increased the suppression of NFκB, suggesting that internalization of the ADC is not significantly limited by TACE-mediated cleavage of the tmTNFα. (Fig. 8D).

The experiments performed to this point employed a monocytic leukemia line, THP1, that is frequently employed for studying monocyte activation. However, it was important that we establish the activity of these ADCs in primary human cells. To this end, we evaluated the ability of the three monocyte-targeting ADCs to inhibit LPS-induced IL-6 production in human PBMCs. We specifically focused on IL-6 production for two reasons. First, the majority of IL-6 production in PBMCs is mediated by myeloid cells (monocytes), allowing a way to minimize interference from lymphocyte-mediated cytokine production. Second, our interest in anti-TNFα ADCs precluded the study of TNFα release, which would already be completely suppressed by the antibody alone. With this in mind, analogous to our study of THP1 cells, we pre-treated the PBMCs with the ADC of interest (or corresponding naked mAb) and after two hours of incubation stimulated the cells with LPS (1 ng mL⁻¹). After overnight incubation, the supernatant was removed and the IL-6 production was evaluated by ELISA. Consistent with the THP1 data, the naked anti-CD11a antibody induced IL-6 release, while the corresponding anti-CD11a_VGCP-Dex suppressed ∼50% of IL-6 release compared to the non-treated control. (Fig. 9A and B) Anti-TNFα_VCGP-Dex and anti-CD38_VCGP-Dex suppressed ∼60 and ∼80% of IL-6 release, respectively, at the 3–10 μg mL⁻¹, in both cases far more effective than the corresponding naked antibodies. At the highest dose (10 μg mL⁻¹), the anti-RSV isotype control ADC exhibited modest (∼30%) suppression of IL6 production. However, as seen in Fig. 9B, the enhanced IL-6 suppression observed for the anti-TNFα and anti-CD38 VCGP-Dex ADCs was statistically significant.

Fig. 9. LPS-induced IL-6 release from human PBMCs (24 h) was suppressed by treatment with VCGP-Dex ADCs. A) Comparison of the VCGP-Dex ADCs with the corresponding naked mAb; B) comparison of the VCGP-Dex ADCs against a nontreated or isotype control ADC.

Conclusions

Herein, our team studied the immune suppressing effects of a series of monocyte-targeting ADCs that are coupled to dexamethasone via a novel cleavable linker. We found that CD11a and CD38 are constitutively expressed on a monocyte cell line, THP1, and antibodies that target these markers are rapidly internalized to lysosomal compartments. (Fig. 2) Transmembrane TNFα was not constitutively expressed, but could be induced by the addition of LPS, also resulting in rapid lysosomal uptake. In contrast, a pair of isotope control antibodies (anti-Her2 and anti-RSV) showed minimal uptake. The three monocyte targeting antibodies were conjugated to ValCitGlyPro-Dex, a novel linker payload designed to release dexamethasone upon proteolytic cleavable of the Cit–Gly bond by lysosomal cathepsins (Fig. 5 and 6). Lysosomal stability studies verified that this linker was rapidly cleaved, resulting in Dex release, while a simple ester linkage (mc-Dex) was not cleaved under the same conditions. The resulting monocyte-targeting ADCs (anti-CD11a_VCGP-Dex, anti-CD38_VCGP-Dex, and anti-TNFα_VCGP-Dex) were evaluated in LPS-induced NFκB and cytokine release assays. Unexpectedly, the anti-CD11a naked antibody induced NFκB activation and cytokine release in THP1 duals cells and, to a lesser extent, in THP1 parental cells and PBMCs. Importantly, however, the corresponding anti-CD11a_VCGP-Dex overcame this activation and was shown to result in suppression of TNFα release in THP1 cells and suppression of IL-6 release in hPBMCs (Fig. 8A and 9A). More impressively, both the anti-CD38 and anti-TNFα ADCs suppressed NFκB activation and IL-6/IL-8 production in THP1 cells, and also suppressed IL-6 production in PBMCs (Fig. 7 and 9).

As ADC technology matures, there is rapidly growing interest in expanding the repertoire of ADCs to therapeutic applications beyond oncology. Herein, we described our initial efforts to this end. Building on existing work in the field, we demonstrated that glucocorticoids can be delivered to THP1 monocytes by targeting CD11a, CD38 or TNFα, something hitherto not reported in literature. Our study of anti-TNFα glucocorticoid ADCs adds to the existing body of research by showing that monocyte uptake is clearly induced by LPS stimulation, even in the absence of TACE inhibitors. Further, we disclose the design of a glycine-proline self immolative spacer that may have broad utility in the delivery of alcohol-containing payloads using ADC technology. Unlike other methods that rely on phosphatase cleavage or esterase cleavage, our technology relies simply on the well-studied cathepsin B cleavage of ValCit peptides. Future studies will be designed to understand the utility of this linker system in in vivo disease models, with a particular focus on the stability of the ester linkage towards plasma esterases in circulation. Further, the relatively modest difference (∼10×) in activity between the targeted ADC and the isotype control suggests that our ADC designs would benefit from the incorporation of higher potency payloads. Evaluation of these and other design strategies are ongoing and results will be described in due course.

Author contributions

JH performed antibody conjugation, THP1 assays, and metabolism studies. SF and FX performed cell based assays and ELISA analysis. KW prepared synthesis, bioconjugation, and metabolism studies. SB performed bioconjugation and cytotoxicity studies. NT oversaw the research and coordinated all writing.

Conflicts of interest

There are no conflicts to declare.