Structural Characterization of the Merkel Cell Polyomavirus Tumor Antigens

Abstract

The Merkel cell polyomavirus (MCPyV) is one of 14 human polyomaviruses (hPyV) and represents, together with JCPyV, BKPyV and Trichodysplasia spinulosa polyomavirus, one of four clinically relevant hPyVs. Although it persists asymptomatically in most of the population, in rare cases, such as in immunocompromised or elderly people, it can lead to the development of the very aggressive skin cancer, Merkel cell carcinoma (MCC). MCPyV is the only hPyV that causes tumors in its own host. Thereby, the small tumor antigen (sT) and a tumor-specific C-terminally truncated version of the large tumor antigen (tLT) play an essential role. For example, it has been shown that sT expression in rat fibroblasts leads to their in vitro transformation and that transgenic sT expression in p53-deficient mice provokes MCC-like tumors. Cell cycle dysregulation due to the inactivation of the LT LxCxE motif-bound pRb represents another important factor in tumorigenesis. In addition to truncating and tumor-specific LT mutations, leading to the loss of the origin binding domain (OBD) and the helicase which results in viral replication deficiency, the viral genome is monoclonal and tumor-specifically integrated into MCC cells. During the permissive life cycle, the full-length LT is involved in viral replication. It binds the viral origin of replication (ori) via its OBD and subsequently unwinds the viral genome by its helicase. MCC is an aggressive tumor with a 5-year survival rate of 62-25%, depending on the initial tumor stage. The introduction of immunotherapy has already substantially improved existing treatment options, however, a significant proportion of the MCC patients do not respond to the PD-1/PD-L1 immune checkpoint inhibitors. Since the T antigens play a crucial role in the viral life cycle and the development of MCC, their atomic structure can serve as a starting point for the development of new treatment methods. However, only a single domain of a MCPyV T antigen, the LT OBD, could be solved, and no complete protein structure of any PyV LT is known. For sT, only the SV40 structure is available. In addition to serving as a drug target, the protein structure can reveal previously unknown T antigen functions and novel protein-protein interactions. This is of particular interest for LT, as the MCPyV LT Merkel unique regions (MURs) have no sequence similarity to other PyV LTs, and their function is largely unknown. Therefore, in this work, the MCPyV T antigens, specifically LT, were biochemically and structurally analyzed. Since previous in vitro crystallization approaches of E. coli-expressed tLT were not successful, ascribed to low protein concentration and high polydispersity, in cellulo crystallization was used in this work. The crystalline matrix of the catalytic domain of the metazoan-specific kinase PAK4 and its endogenous inhibitor Inka1 was used as a crystallization aid for the MCPyV T antigens in various eukaryotic cell lines. In cellulo co-crystals of tLT, as well as sT and truncated sT versions, could be observed, but showed little to no X-ray diffraction. Based on preceding findings, transmission electron cryo-microscopy single particle analysis (cryoEM SPA) of LT was thereafter performed. The protein was expressed in insect cells, chromatographically purified, and the polydispersity and aggregation were reduced by additives. In addition, the LT-DNA binding behavior of multiple oligonucleotides was investigated. It could be shown that the viral ori stabilizes LT most effectively and reveals the highest binding specificity. Furthermore, it was demonstrated that the presence and absence of different oligonucleotides did not affect LT hexamer formation. In addition, LT was found to be stabilized in phosphate-containing buffers, which, together with the preceding optimization steps, did not result in an aggregate-free and monodisperse protein solution. Notwithstanding these limitations, the first MCPyV LT helicase model with a resolution of 6.8 Å has been generated by using cryoEM SPA. Interestingly, despite the low sequence identity of 33%, the hexamer shows a pronounced structural similarity to the SV40 LT helicase. Reasons for the intermediate resolution are preferential particle orientation, protein aggregation, and polydispersity, possibly caused by the MURs, which are predicted to be unstructured. In addition, the absence of the OBD and the LT N-terminus in the cryoEM SPA model may also be attributed to the MURs. In summary, this work provides the first MCPyV LT helicase model with a resolution of 6.8 Å, using cryoEM SPA. Despite its intermediate resolution, based on protein aggregation, polydispersity, and preferential particle orientation, the model represents an important initial step towards the development of specific antiviral therapies.