Therapeutic antibodies to intracellular targets in cancer therapy

Abstract

Therapeutic monoclonal antibodies (mAbs) are a proven therapeutic platform, but they cannot readily cross the cell membranes to bind intracellular antigens, while some of the most important disease-associated proteins are intracellular, protected from direct mAb attack. However, the cellular processes of necrosis and major histocompatibility complex (MHC) class I antigen presentation expose epitopes from intracellular proteins to the extracellular environment or cell surface. Antibodies that exploit these processes can therefore specifically target diseased cells based on their intracellular protein content. These strategies expose important new targets for mAb therapy and expand the potential for effective therapies.

Keywords::

• antibody therapy
• ESK1
• HLA
• intracellular antigens
• MHC class I presentation
• TCR mimic
• TCR-like
• tumor necrosis therapy

1. Introduction

Therapeutic monoclonal antibodies (mAbs) are highly specific and potent drugs, capable of marking cells for immunologic attack, or of modulating signaling pathways to reduce cell growth or induce apoptosis. Native mAbs can activate powerful immune effector mechanisms such as antibody-dependent cell-mediated cytotoxicity (ADCC) and complement-mediated cytotoxicity (CMC), while antibody drug conjugates (ADCs) and radioimmunoconjugates harness the specificity of mAbs as a vehicle to deliver their cytotoxic payload. However, unlike small molecule drugs that cross the cell membrane, mAbs cannot directly access intracellular proteins. All therapeutic mAbs currently marketed in the United States target extracellular or cell-surface molecules, while many important oncogenes and disease targets are intracellular.

Fortunately, biologic pathways expose many intracellular proteins to the cell surface or extracellular environment. When cells undergo necrosis, the intracellular components are exposed to recognition by antibodies; this can be exploited in the highly necrotic environment of solid tumors. Alternatively, the MHC class I presentation pathway has evolved as a process for all live nucleated cells to present peptide epitopes of internal proteins on the surface for T-cell surveillance by T-cell receptor (TCR) recognition. Hence, by mimicking the specificity of TCRs, antibodies can be designed to specifically bind cells expressing an intracellular target protein.

As with conventional antibody therapy, choice of target is crucial to efficacy and specificity. Ideal targets are disease-specific epitopes with limited expression in normal tissues. Further, the intracellular repertoire contains proteins that act as disease drivers, such as oncogenes or stem cell-associated proteins; therefore, selection of these targets could potentially lead to disease eradication not possible with therapies against extracellular molecules.

While the antibody therapy strategies discussed here have the goal of killing target cells, or associated stroma or vascular tissue, by engaging the immune system or delivering a cytotoxic payload, antibodies can also be used to block protein–protein interactions, such as for ligand–receptor interactions that drive growth. However, these mAbs still can only access secreted factors or cell-surface receptors, and lack tumor specificity.

2. Current approaches

2.1 Tumor necrosis therapy

Solid tumors can outgrow their blood and nutrient supply, leading to cell degeneration and necrosis. Tumor necrosis therapy (TNT) mAbs can theoretically target a nonspecific intracellular moiety that is selectively exposed in necrotic cells, including single-stranded DNA or histone complexes. Many TNT strategies employ ‘armed' antibodies to destroy live cancer cells surrounding the necrotic target [1]. Also, native mAbs can bind released tumor antigens, forming antigen–mAb complexes that promote DC cross-presentation to enhance CD8⁺ T-cell responses [2]. In addition, mAbs directed against cancer cell-specific intracellular proteins inhibited syngeneic tumor growth in a mouse model [3], though no clear mechanism was identified.

2.2 TCR-like/TCR-mimic antibodies

Intracellular proteins are processed by the proteasome and presented on the cell surface as small peptides in the pocket of MHC class I molecules (in humans, also called human leukocyte antigen [HLA]) allowing recognition by TCRs on T-cells. Therefore, mAbs that mimic the specificity of TCRs can bind cell-surface complexes specific to cells expressing an intracellular protein. ‘TCR-mimic' (TCRm) antibodies can be generated by immunization with recombinant MHC/peptide complex, or by phage display, as first reported in 1996 [4]. In this case, the phage library was generated from mice immunized with the antigen. However, human naïve libraries of sufficient size can also contain ScFv reactive to self-peptide/HLA antigens, allowing generation of Fab, mAb or bifunctional constructs. For a review of recent advances in the field, see Dahan and Reiter [5].

TCRm antibodies are especially interesting in oncology, because many of the most important tumor-associated and oncogenic proteins are nuclear or cytoplasmic. A fully human IgG1 mAb (ESK1) that specifically targets RMFPNAPYL (RMF), a peptide derived from Wilms' tumor gene 1 (WT1), presented in the context of HLA-A0201 was recently reported [6]. This was the first fully human antibody directed against cells expressing WT1, an important, immunologically validated oncogenic target. WT1 is a zinc finger transcription factor with limited expression in normal adult tissues, but is expressed in the majority of leukemias and a wide range of solid tumors [7]. ESK1 mAb specifically bound to leukemias and solid tumor cell lines that are both WT1+ and HLA-A0201+ and showed potent ADCC activity in vitro and efficacy in vivo against several WT1+ HLA-A0201+ leukemias.

Several other groups have produced TCRm or Fab antibodies against intracellular cancer cell targets, such as TARP, MART-1 and gp-100 peptides in the context of HLA-A0201 [8,9]. As immunotoxin conjugates, these constructs have shown activity against breast and prostate cancer, and melanoma xenografts in mice. Mouse IgG2a mAbs specific for human chorionic gonadotropin (h-CG)-β, p68 RNA helicase and Her2 peptides presented by HLA-A0201 are also active in vivo [10]. Several mechanisms of action were observed in vitro, including ADCC, and surprisingly, CMC (which normally requires high antigen density) and direct induction of apoptosis through the JNK signaling pathway. Interestingly, the TCRm against p68-helicase peptide/ HLA-A0201 also internalized, colocalizing with endosomal vesicles. Both ADCC and CMC activity were seen with a mouse IgG2a mAb directed against the PR1 peptide/HLA-A0201 complex [11]. The varied repertoire of mAb mechanisms of action suggests that species and choice of Ig as well as target characteristics may influence effector function.

3. Expert opinion

Monoclonal antibodies are increasingly important therapies with demonstrated clinical success in a large number of cancer types. Accessing intracellular proteins through the new strategies described here will greatly expand the potential uses of mAbs against previously inaccessible targets. However, several challenges remain in translating these mAbs into clinical therapies. Targets for TCRm mAbs are expressed in much lower numbers than many current cancer cell-surface targets, and therefore efficacy of native mAbs alone may be limited. Enhancing antibody potency may overcome this obstacle. Further, it will be important to discover novel, highly presented peptide antigens derived from important oncogenes, as well as strategies to modulate presentation levels. Finally, specificity is a concern with these therapies, as TCR mAbs must avoid targeting peptide/HLA complexes presented on healthy cells or significantly cross reacting with HLA in the absence of the specific peptide epitope.

3.1 Enhancing antibody potency

As TCRm antibodies target short peptides derived from one of tens of thousands of intracellar proteins, it is likely that TCRm epitopes will be expressed in extremely low abundance on the cell surface. Despite this calculation, some cancer cell lines bound 5000 –8000 TCRm mAbs per cell, which could account for nearly 0.5% of the total surface HLA-A2 molecules [6,11]. This level of expression may be limited to certain, highly immunogenic peptides. Therefore, selection of highly potent constructs for clinical development will be crucial to success for most targets.

The mechanisms of action of mAbs can be altered and enhanced through protein engineering [12]. Many ADCC-enhanced mAbs have been produced, either by glyco-engineering or point mutations, and several are in clinical trials with promising results [13]. For antibodies such as ESK1, that work primarily through ADCC, Fc-modified antibodies could be appropriately potent constructs for clinical development. For tumor types or targets that are not easily accessible by effector cells, antibodies can be linked to drugs, toxins or radioisotopes and used to selectively deliver these cytotoxic payloads. While radioimmunoconjugates can work from outside the cell, most ADCs require internalization of the conjugate and large numbers of cell-surface epitopes to be effective. Internalization has been reported for some antibodies, but appears to be epitope and construct specific. On the other hand, radioimmunoconjugates that rely on alpha emission are so potent that a single alpha particle can kill the cell from outside [14]. Finally, bispecific antibodies or fragments can recruit potent immune cells, such as cytotoxic T-cells. Bispecific constructs with an anti-CD3 fragment (BiTEs) could overcome limitations of native mAb therapy against cells with very few targets.

Unlike TCRm antibodies, TNT evokes a cytotoxic response limited to the surrounding solid tumor. TNT mAbs will likely be much more effective as radioimmunoconjugates or bispecifics, with a linked cytokine or chemoattractant moiety, since engagement of NK-cells and other immune effectors through the Fc receptor will likely be cell specific. Additionally, selection of appropriate antibody-linked cytokines and chemoattractants could recruit the immune system with the specificity of TNT mAbs [1].

3.2 Modulating TCRm binding sites

Target gene expression and translation, target protein degradation, expression of proteasome components and efficiency of processing, and HLA expression and surface level could all govern the level of peptide/HLA complex presented on the target cell surface. Agents that increase peptide/HLA complex presentation could be used to enhance mAb efficacy in vivo, thereby allowing a therapeutic response in patients who display low antigen levels and would otherwise not respond to TCRm mAb therapy. For instance, interferon gamma (IFNγ) is known to regulate expression of both HLA-A and several proteasome components. Conversely, it is important to identify agents that decrease peptide/HLA levels, as these could abrogate TCRm therapy.

3.3 Discovering new TCRm targets and HLA specificities

The peptide and the HLA molecule together govern the specificity of a TCRm antibody. Peptides targeted must be specific to the desired intracellular target, and also capable of binding HLA. Screening for peptides of a target protein, such as WT1, presented by a variety of HLA types [15], will yield new and possibly improved TCRm mAb targets. Such studies will also be crucial for discovering novel peptide targets for each HLA type and can supplement in silico methods.

Most TCRm antibody research to date has focused on peptide epitopes presented by HLA-A0201, as HLA-A0201 is the most common allele in the US population. However, HLA haplotypes are diverse worldwide, which poses a problem for TCRm antibody development. Certain peptides are only presented by certain HLA-types, representing only a fraction of patients in a population. Therefore, multiple TCRm mAbs will be needed to treat a broad population base. In the United States, ∼5 different HLA types are found in about 85% of the population. In addition, as all nucleated cells present peptides on HLA, TCRm antibodies must not react with the HLA complex alone. HLA-A0201 and HLA-A2402 transgenic mouse models are available to assess possible toxicity, though unless the peptide antigen of interest is conserved between human and mouse, murine cells may not present the same HLA/peptide epitope as humans (in which case the model can only assess non-specific binding to HLA alone). Further, higher mammals and primates do not express HLA, so conventional toxicology studies may be less useful for TCRm mAbs. In summary, monoclonal antibodies are increasingly important therapeutics with demonstrated clinical success in a large number of cancer types. Accessing intracellular proteins through the new strategies described here will greatly expand the potential uses of mAbs against previously inaccessible targets.

Declaration of interest

The authors have received funding from the Leukemia and Lymphoma Society, and from the National Cancer Institute. No funding was received in the preparation of this article. D Scheinberg and T Dao are also inventors of related technology. The authors have no other competing interests to declare.