Using an RNAi approach, we conducted in vitro gene knockdown studies of the 4 genes comprising an MPM prognostic test[20–22] and determined the effect, if any, on cell death and growth arrest using molecular markers for apoptosis and mitosis, respectively. Supporting data was obtained in select positive experiments using cellular morphology. We have found that three of the MPM prognostic genes (ARHGDIA, COBLL1, TM4SF1) are associated with a phenotype consistent with a negative regulator of apoptosis; rates of programmed cell death increased ~2- to 4-fold when mRNA expression levels for these genes were attenuated in cultured cells. Increased apoptosis was also associated with concurrent changes in nuclear morphology and a modest, but statistically significant, reduction in cell number. The fourth MPM prognostic gene (PKM2) could not be linked to either apoptosis or growth arrest in the current study.
We established the Hoechst-based Condensed Chromatin Index as an apoptotic indicator secondary to the detection of cleaved lamin A, a well established molecular marker. We found that rates of apoptosis measured using this index were of a generally lower magnitude compared to those obtained via detection of cleaved lamin A, and furthermore not all genes linked to apoptosis via cleaved lamin A were associated with a statistically significant change in the Condensed Chromatin Index (e.g., COBLL1, TM4SF1), despite a trend in the expected direction. This is likely due to the facts that i) the index is based on clearly detectable nuclear changes that are a hallmark of relatively early events in apoptosis, such as local chromatin condensation while cleaved lamin-A is a relatively late event, and ii) we collected readouts at a single timepoint only (72 hours post-siRNA transfection) by which point early apoptotic events may have already occurred.
A second morphological readout (Aberrant Nuclei Index) was collected to determine whether experimental knockdown cells were associated with abnormally shaped nuclei. The Aberrant Nuclei Index detects cells whose nuclei deviate from the norm using two parameters, circularity and elliptical fit, which measure how spherical and how oval nuclei are respectively. Changes in nuclear morphology can be elicited by a number of different factors including, but not limited to fundamental alterations to microfilament arrangement and the nuclear skeleton. In addition to providing basic nuclear shape and stability, nuclear support proteins are also required for chromatin organization, transcription regulation, DNA replication, nuclear assembly, nuclear positioning, and apoptosis. Thus, the aberrant nuclei classification parameter may be best described as a general marker for cell well-being. We hypothesized that knockdown of MPM prognostic genes that led to an increase in apoptosis would also lead to an increase in the numbers of aberrantly shaped nuclei. Although the frequency of aberrantly shaped nuclei was modestly elevated under these expected conditions, the results were not statistically significant. The implication of this negative result, if any, is not clear and may relate to timing issues and/or other factors since only 1 of the 2 apoptosis positive control knockdown cell lines displayed a statistically significant increase in the numbers of aberrantly shaped nuclei.
The MPM prognostic genes examined in the current study have in some cases been previously linked to cancer or cancer relevant processes. TM4SF1, a.k.a. L-6 tumor antigen, is a distant member of the transmembrane 4 superfamily of cell-surface proteins that are characterized by the presence of four hydrophobic domains. Although the precise function and physiological role of this gene is unknown, the fact that it is a cell surface antigen that is highly expressed in different carcinomas and expressed at relatively low levels (if at all) in many normal tissues has led to its evaluation as a candidate therapeutic target for radioimmunotherapy treatment, particularly for breast cancer. More recently, TM4SF1 has been linked to metastasis and angiogenesis, but not apoptosis (to our knowledge) prior to the current study.
The COBLL1 (COBL-like 1) gene was cloned in 1999 and originally designated KIAA0977. The deduced protein contains 1,166 amino acids and was found to be expressed at high levels in lung, liver, kidney, pancreas, ovary, spinal cord, brain, fetal liver, and all specific adult brain regions. In 2003, Carroll et al. renamed the gene COBBL1 based on homology to the newly discovered COBL (i.e., cordon-bleu homolog, mouse) gene. The specific role(s) of both genes are not known, but each is presumed to play a role in embryogenesis based on temporal expression patterns during development. Given the known role of apoptosis in determining tissue morphology during normal embryonic development and the results of the current study, it is reasonable to hypothesize that inappropriate expression in tumor cells acts to prevent programmed cell death and promotes tumor cell survival.
The ARHGDIA (Rho GDP dissociation inhibitor [GDI] alpha) gene, first described in 1993, belongs a family of genes whose members regulate (Ras superfamily) Rho genes by keeping them in the inactive GDP-bound state. ARHGDIA was the only gene in our RNAi studies found to impact apoptosis in both normal and tumor cells, with a generally greater magnitude of effect in tumor cells compared to the other 2 positive genes. Other investigations of ARHGDIA in the context of cancer have indicated a role in estrogen receptor signalling (and estrogen responsiveness) in breast cancer[35, 36] and association with poor prognosis in colorectal cancer consistent with our previous, similar findings in MPM[20–22, 37]. The normal physiological role of ARHGDIA is likely to be tissue-specific since ARHGDIA (-/-) knockout mice are viable, but suffer severe and progressive kidney and reproductive system impairment and typically die within a year of birth. While the precise manner in which ARHGDIA regulates apoptosis/cell survival in human cancer is not known, studies in cultured rodent insulinoma cells show that overexpression of ARHGDIA increased cell viability and decreased activated c-Jun N-terminal kinase (JNK) expression following exposure to the apoptosis promoter mycophenolic acid (MPA), whereas knockdown of ARHGDIA (via RNAi) enhanced MPA-induced cell death and increased the activation of JNK.
Many types of cancer cells express the M2 isoform of the glycolytic enzyme pyruvate kinase (PKM2), the only gene for which we could not provide evidence linking to either an apoptosis and/or growth arrest phenotype. The PKM2 gene product has been previously linked to cell proliferation in vivo , presumably by increasing cellular metabolism, and is being evaluated for selective therapeutic targeting based on preferential expression in tumor, and not normal, cells. PKM2 knockdown cell lines in our hands did not display levels of apoptosis or mitosis that differed meaningfully from controls which may not be entirely unexpected given the known role of this enzyme in metabolism and its hypothesized importance in tumor spreading, a phenotype we did not measure in the current study. However, as would be expected in a scenario of reduced metabolic activity and decreased proliferation, knockdown of PKM2 was associated with a modest (~15-20%) decrease in cell number (via the Cell Number Index, as a surrogate measure of proliferation) in two of three experiments employing different siRNA molecules, although the results were not statistically significant due primarily to excessive variability.
Although experimentally valid, a clear limitation of our study is in the choice of cell lines modeled, neither of which are mesothelioma-derived, although we have no reason to believe that the basic mechanism of action of any of these genes will differ dramatically among different tumor cell lines, although the strength of effect may. WI38 cells are normal, human, lung-derived fibroblasts. A549 cells are derived from an alveolar (lung) tumor. Several issues governed the choice of these cell lines. MPM tumor cells are not easily amenable to high-throughput assay conditions which we used to rapidly screen a set of four genes to determine which, if any, deserved further analysis in a traditional ("low throughput") one gene/one assay/one cell line experimental design. A549 cells, unlike mesothelioma tumor cell lines, are relatively easy to transfect and are more appropriate for use in a high-throughput assay. Also, there are no established (normal) pleural mesothelial cell lines available to use as a control. The general lack of knowledge surrounding the precise function of the four mesothelioma prognostic genes (ARHGDIA, COBLL1, PKM2, TM4SF1) indicated the importance of examining a normal cell line and WI38, while not ideal, was chosen because it is of similar embryonic origin (i.e., mesoderm) to mesothelial cells lining the thoracic pleura. Finally, despite these limitations, we assert these studies as described are still useful in that they add to the body of knowledge for a set of relatively understudied genes.
Surgical resection for MPM has been reported to improve survival in a subset of patients identified who can be identified pre-operatively using proposed pathological and molecule criteria[20–22]. Our results are consistent with previous data that anti-apoptotic genes are activated in MPM resulting in chemotherapy resistance and tumor cell survival[42, 43]. For example, the fact that two of the genes (TM4SF1, COBLL1) are molecular markers for relatively good prognosis patients while the other one gene (ARHGDIA) is a molecular marker for relatively poor prognosis patients suggest that therapies could be tailored and tested in a rational manner to target specific pathways activated in different mesothelioma subsets.