Preventing Microbial Infection
The large ectoplasmic domain and complicated glycosylation pattern of polymorphic epithelial mucin (PEM) allows for its role as an inhibitor of pathogenic infection. In the gastrointestinal tract, steric inhibition prevents the invasion of normal flora as bacteria have been shown to interact directly with mucin structures at the epithelial surface. The result is a drastic decrease in tissue invasion (Vimal et al., 2000). One such example in murine models has been in preventing Helicobacter pylori (H. pylori) infection. Interaction between PEM and H. pylori drastically decreases infection, even in the case of PEM detachment from the epithelial surface. An increase of infection was demonstrated in the reverse situation (Lindén et al., 2009).
PEM also plays a vital role in protecting against infection is in the female reproductive organs where it also plays a role in blastocyst attachment to the uterine endometrial wall. PEM levels decrease resulting in intimate interactions between selectin proteins displayed by the blastocyst and those displayed by the endometrium (Hey et al., 1994). Knockout experiments in mice have indicated that a loss of MUC1 represents a considerable increase in uterine infections (DeSouza et al., 1999).
Influence on Cell Signalling
An example of PEM signalling comes in the form of one of the most notorious oncogenic protein classes to have ever been studied: Ras. Studies have indicated that a direct interaction between the cytoplasmic tail of PEM and a secondary messenger, Grb2, eventually leads to the activation of membrane-bound Ras proteins. The attachment of the guanine nucleotide exchange factor (GEF), Sos, to Ras allows for the full activation of the oncoprotein (Pandey, Kharbanda and Kufe, 1995). The influence of Ras in tumour occurrence has been well documented, with indications of its role in cell survival and growth when mutated or over-activated.
Human epidermal growth factor (EGFR) receptors and their respective ligands have also been identified as targets for PEM interaction in many murine models. Some have indicated an increase in Erk1/2 activation in the presence of increased PEM (Schroeder et al., 2001). Activation of a mitogen-activated protein kinase (MAPK) such as Erk1/2 would suggest an influence of PEM over a wide variety of intracellular activity associated with potentially oncogenic behaviour.
Further interactions have been found between PEM and β-catenin, a process that induces an epithelial to mesenchymal transition in epithelial cells. Such a transition increases the likelihood of metastasis and may explain the association between PEM and particularly aggressive pancreatic tumours (Roy et al., 2011). A similar interaction has been noted in breast tumours where the co-localisation of PEM and β-catenin occurred in primary and secondary tumours. Most notably however was the dramatic increase in PEM-β-catenin interactions in metastatic tumours (Schroeder et al, 2003), again hinting at a role for PEM in tumour invasiveness.
As well as promoting invasiveness, it seems that PEM also has an influence in apoptotic avoidance in cells. The cytoplasmic tail of PEM can directly interact with the regulatory domain of p53, preventing apoptosis induction (Wei, Xu and Kufe, 2005). Furthermore, PEM has been shown to interfere with the Bax-induced intrinsic pathway of apoptosis, directly binding to the active BH3 domain of the protein (Ahmad R. et al., 2012).
Inter- and Intracellular Processes
The repetitive nature of the PEM protein sequence has been identified as an immediate link to oncogenic function as a disruption to any of the small sequence repeats can render PEM structures ineffective inhibitors of cell-cell and cell-extracellular matrix (ECM) interactions. The normal architecture of the PEM sequence allows for a modulation of interaction between cell integrins – as PEM levels increase, integrin interaction decreases and vice versa (Wesseling et al., 1995). In cancers where PEM is overexpressed, a lack of cell-cell and cell-ECM interaction results, eventually leading to uncontrollable growth.
A further characteristic of oncogenic PEM is a state of hypoglycosylation. In this state of being, PEM has been shown to influence cancer formation through inflammatory processes. Some studies have directly linked an activation of the NF-κB inflammation pathway through interactions with specific promoter sequences. Interleukin-6 (IL-6) and tumour necrosis factor-α (TNF-α) have both been shown to be up-regulated by PEM through interaction of the NF-κB pathway. Interestingly, the higher presence of tandem repeats in the PEM structure caused a higher activation of the intracellular inflammation pathway (Cascio, Zhang and Finn, 2011). Further experiments looking into the influence of PEM on inflammation have identified PEM as a potential target in treating prostate cancer (Burke et al., 2006) and inflammatory disorders that are known to lead to higher cancer risks such as inflammatory bowel disease (IBD). The migration of MUC1 specific T-cells to sites of IBD or to the pancreas in diseased mice have suggested that targeting MUC1 may also prove useful in pancreatic disorders (Kadayakkara et al., 2010). One experiment even found that a MUC1-specific vaccine significantly reduced the progression of IBD and resultant tumour formation in mice. The resulting solution suggested inducing MUC1-specific immunity in order to prevent the progression of pre-malignant transformations (Beatty et al., 2010).
Inferring Treatment Resistance / Avoiding Apoptosis
An interesting feature of MUC1 in cancer is its ability to infer resistance to some treatments. Much like many other oncoproteins, a major concern upon mutation is the ability of the protein to create a malignant environment that is impenetrable to modern medicine. Herceptin, for example, has been shown to completely lose efficacy due to a mutated form of the PEM protein. Studies have shown that a truncated form of PEM is responsible for acquired resistance to Herceptin resulting in 25 per cent of metastatic breast cancer patients becoming untreatable with the antibody. Furthermore, treatment with a MUC1 inhibitor with Herceptin has produced a regaining of Herceptin efficacy, implying that MUC1 is a legitimate target with regards to salvage therapy (Fessler et al., 2009). Further studies into breast cancer indicated that MUC1 overexpression caused a desensitisation of tumour cells to standard chemotherapy as well as targeted treatment. Drugs such as cyclophosphamide, taxol and doxorubicin were all rendered useless in treating the disease whilst MUC1 inhibitors produced a regaining of sensitivity to such therapies (Fessler et al., 2009).
In anaplastic thyroid cancer cells – an extremely aggressive form of the disease with a poor prognosis – RNA interference of MUC1 produced a gain of sensitivity to standard chemotherapy agents. In the same study, interleukin-4 (IL-4) and 10 (IL-10) were identified as promoters of PEM oncogenic behaviour, suggesting that they caused migration of PEM to the mitochondrial membrane. The process of migration has been shown to be as a result of further complex interactions involving fibroblast growth factor receptor 3 (FGFR3) and heat shock protein 90 (Hsp90) (Ren et al., 2006). The result is an interference with the induction of the intrinsic apoptotic pathway (Siragusa et al., 2007). The PI3K/Akt pathway was also identified as harbouring promoters of PEM overexpression as neutralising the pathway lead to a significant decrease in PEM expression.
A further interaction with the PI3K/Akt cell survival pathway has been observed in colon cancer cells. The transcription factor FOXO3a which, through a down-regulation of FOXO3a phosphorylation by Akt, can activate oxidative stress survival behaviours such as DNA repair (Yin, Huang and Kufe, 2004) resulting in a suppression of apoptosis. This study however indicates that PEM involvement in oncogenic pathways is contextual as, in this case, PEM suppresses PI3K/Akt activity rather than encourages it.
CD33 is a much more established tumour associated target antigen than PEM and has a more recognised infrastructure in its involvement in specific cancers. The most notable association is with AML, a form of leukaemia that has been targeted with the immunoconjugate gemtuzumab ozogamicin (Mylotarg). Although Mylotarg was withdrawn from the market due to an increase in patient death and lack of advantage over previously established treatments for the disease, clinical evidence has revealed the efficacy of the immunoconjugate in terms of its targeting. Because of this, the drug revealed the specificity that could be achieved with targeted therapies, legitimizing CD33 as a therapeutic target. To this day further studies into a calicheamicin conjugate are underway as well as novel targeting techniques.
Role in Signalling
CD33 has a particular structural characteristic that places it in the superfamily of immunoglobulins, specifically the sialic-acid-binding immunoglobulin-like lectins (SIGLECs). The outer N-terminus region contains two immunoglobulin-like domains and the cytoplasmic C-terminus region is thought to cause an inhibitory effect on signal transduction due to the presence of a tyrosine-based inhibitory motif (Ulyanova, 1999). In mutant phenotypes, CD33 is believed to be over-expressed and has a lower rate of internalisation (Walter et al., 2008). Signal transduction through CD33 is responsible for the inhibition of various oncogenes, most notably Fms-like tyrosine kinase 3 (Flt-3). Flt-3 has been identified as one of the most common mutations in AML and studies have found that over-expression of CD33 correlates to a high-risk of Flt-3 mutations (Pollard et al., 2012). Mutant Flt-3 receptors express a constant signalling cascade to various downstream survival pathways including Ras and PI3K pathways (Rosnet, 1999).
Validity as a Therapeutic Target
Although calicheamicin was found to be a highly toxic molecule, it has been shown to be efficient in treating a particular subset of AMLs, namely those that exhibit mature AML cells (Walter et al., 2012). In these cases, efficacy and selectivity of Mylotarg has been shown to be significantly high. Further anti-CD33 antibodies have been tested using different drug linkers that release the drug in certain pH environments. Amide linkers and carbohydrate linkers (Hamann et al., 2002a) established themselves as the two frontrunners with carbohydrate conjugates being replaced by Mylotarg upon fast-track approval of the antibody (Hamann et al., 2002b).
There have also been investigations into alternative conjugates, some in the field of nanomedicine. The addition of gold has shown particular efficiency as a form of bioprobe, selectively binding to CD33-expressing AML cells at percentages as high as 95.4% whilst maintaining a very low non-specific uptake in peripheral blood cells (8.2%) (Retnakumari et al., 2011). An anti-CD33-maytansine conjugate has also been tested; however this proved unsuccessful in treating patients with relapsed AML (Lapusan et al., 2012). A humanised, recombinant gelonin conjugate has shown promise in phase 1 trials, with only two out of 28 patients forming antibodies to the gelonin component – the antigenicity otherwise was very low and proved to be safe for administration (Borthakur et al., 2013). An even more promising conjugate involves a humanised anti-CD33 antibody linked to pyrrolobenzodiazepine (PBD), a DNA cross-linking agent that has proven successful in treating AML murine models showing multidrug resistance (Kung Sutherland et al., 2013). Further, the addition of bismuth-213, an alpha emitting isotope, has also been shown to defeat chemo- and radioresistance in AML patients (Friesen et al., 2013).
Moving away from conjugates, Lintuzumab, a humanised anti-CD33 antibody, has shown relatively little effect against overt AML but at the same time, has shown significant success at eliminating residual acute promyelocytic leukeamia (APL) (Jurcic, 2008). It has also been shown that colon cancer is influenced by CD33 as an increase in CD33+ myeloid deprived suppressor cells (MDSCs) has been correlated with a cancer stage and metastases (Zhang et al., 2013).
It is clear that CD33 is a valid target and is also one that can be targeted for various reasons. Although treating cancer is of utmost importance, it has been shown that targeting CD33 is as effective at screening for disease and prognosis – something that is equally imperative and attractive for a cell surface antigen. The flexibility in manufacturing an anti-CD33 antibody also argues in favour of investigating the antigen further. Proof can be seen in the linkage of a particular drug or radioisotope to effectively defeat AML in many of its forms. The affinity and specificity achieved with previous antibodies and CD33 also points to an antigen that is readily available to be targeted.
Single Chain Fragment Variable Antibodies
Single chain fragment variable antibodies (scFvs) consist of the variable heavy and variable light chain segments of an antibody structure fused together using a peptide linker. The Fc region of a normal antibody is not present, meaning the resulting antigen binding regions are highly unlikely to induce side effects; something that intact antibodies are usually responsible for.
Advantages of an scFv
Numerous features of an scFv separate them from other antibody formations currently used in therapy. Aside from their low immunogenicity, their size alone allows them to diffuse further into tumour tissue than perhaps a whole antibody could. Further, their clearance time in tumour tissue is much longer, while at the same time they have a rapid blood clearance time. The overall outcome of an scFv is a highly specific antibody that can reach further, persist longer in tumour tissues and be cleared faster from the blood than a standard immunoglobulin structure (Ahmad et al., 2012). In terms of reducing the immunogenicity and increasing circulation time of the molecules further, polyethylene glycol (PEG) can be used (Sorefan, 2011). Their easy manipulation in terms of conjugation for either visualisation or therapeutic techniques make the scFv a highly desirable formation for most tumour types.
It should be noted that an scFv can exist in two basic formations: VH-linker-VL and VL-linker-VH. Neither formation has proven significantly more efficient although some systems specifically favour one over the other – the latter recombinant form being specifically favoured in P. pastoris, for example (Luo et al., 1995). As well as the overall conformation, the peptide linker formation is vital to the antibody’s success. The length and amino acid composition are the two most important features, with the length of the linker dictating the folding ability of each domain and therefore the formation of an antigen binding site. Amino acid composition determines overall solubility and flexibility (Ahmad et al., 2012). Some studies have investigated linker length and determined that lengths below 12 amino acids were detrimental to proper scFv formation; however, associations between otherwise separate domains can result in bivalent dimers or tetramers forming (Guo & Cai, 2003). Certain dimeric formations have proven to be successful in terms of in vivo stability and efficacy and have advanced through pre-clinical trials. The association of two different scFvs (bi-specific diabodies) has been shown to be an effective means of T cell recruitment and viral retargeting (Kortt et al., 2001), both of which methods are involved in certain cancer therapies. Furthermore, dimeric anti-HER2-scFv has shown a much greater tumour retention in vitro and in vivo in murine models (Adams et al.,1998).
Research into the scFv formation has indicated many instances of clinical efficiency in cancer treatment. Much success has seen the development of conjugated scFv’s in treating disease due to the rapid clearance time of the protein and therefore exposure of healthy tissue to potentially DNA damaging agents. Technetium-99m (99mTc)-labelled anti-Tac (targetting interleukin-2 receptors) has shown little renal uptake when chemically altered through glycolation (Kim et al., 2002), therefore avoiding kidney damage whilst effectively treating cancerous tissue. The conjugation of an immunotoxin (Pseudomonas exotoxin A) with an scFv targeting a foetal acetylcholine receptor (fAChR) has shown great efficiency in treating rhabdomyosarcoma (RMS) cells in vitro and delaying RMS onset in murine models. As well as treating the highly resistant and most common malignant soft tissue cancer in children, the expression of fAChR in approximately 20 per cent of all metastatic malignant melanomas suggests the antibody may be applicable to treating further cancers (Gattenlöhner et al., 2010). Small interfering RNAs (siRNAs) have also been shown to be much more efficient when conjugated to an scFv vehicle to treat HER2+ breast cancers by suppressing targeted gene expression (in this case Polo-like kinase 1), reducing proliferation and inducing apoptosis. The use of the same fusion protein intravenously produced a decrease in metastasis and halted tumour growth. Both intravenous cell line models produced little toxicity (Yao et al., 2012).