We identified ATAD2 several years ago as a member of senescence-related 15-gene signature set providing high confidence prediction of HCC patients with 97.5% accuracy [9]. In the same report, we also showed that freshly isolated human hepatocytes lack detectable ATAD2 protein in contrast to nine different HCC cell lines all expressing it. Here we evaluated the value of ATAD2 as a disease biomarker and potential therapeutic target for HCC. Our observations clearly indicate that this protein is a good disease progression marker, but a conditional therapeutic target.
The value as a tumor marker was evaluated using a set of tissues representing normal liver, dysplastic nodules and well-, moderate- and poorly differentiated HCCs. We observed almost a perfect correlation between ATAD2-positive staining cell numbers and disease progression. Accordingly, ATAD2 was not detected in LGDNs but was positive in 17% of HGDNs. This positivity progressively increased up to 83% in poorly differentiated HCCs. Our observations correlate with data reported previously [20–22]. Here we provide novel data showing that ATAD2 expression in normal liver and HCC is closely correlated with Ki-67 expression, a well-known biomarker used to evaluate proliferation index in tumors [23]. This led us to consider ATAD2 as a proliferation biomarker. Next, we tested whether ATAD2 expression in liver is specific to malignancy. Thus, we analyzed the status of its expression following induction of liver regeneration by partial hepatectomy in rats. This analysis was done in parallel to MKI-67 and CCND1 expression. ATAD2, MKI-67 and CCND1 RNA levels were low in the initial phase of the liver regeneration and all three showed strong induction between 24-36h, followed by a gradual decrease until the end of proliferation process. The absence of ATAD2 expression in normal liver together with its increased expression in regenerating rat liver, as well as in malignant liver lesions as a function of tumor progression is reminiscent of Ki67 expression in these situations. Indeed, Ki67 labelling index(LI) is less than 5% in normal liver and inactive cirrhosis, but higher in chronic hepatitis (LI: 29–41) and very high (LI: 71) in HCC [23]. In confirmation of our observations, MKI-67 expression during rat liver regeneration is induced and remained high for at least 4 days [24]. Ki-67 is a nuclear protein expressed in proliferating (G1, S, G2 and M phases) cells, but not in resting (G0 phase) cells including hepatocytes [25].
Our in vitro studies confirmed that ATAD2 is indeed a specific nuclear marker for proliferating cells like Ki-67. As shown in Fig. 2, normal hepatocytes did not express ATAD2, but all tested HCC cell lines displayed nearly 100% positive staining which paralleled Ki-67 staining. Indeed, ATAD2 immunostaining performs better than Ki-67 staining because of it’s a homogenous nuclear staining pattern. The presence of nuclear ATAD2 in all proliferating cells despite its role in “co-chaperon” role in the organization of newly loaded histones during DNA synthesis [3] is not necessarily contradictory. Ki- 67, a well-known marker for cells in proliferation (G1, S, G2, M phases) serves as a biological surfactant to disperse mitotic chromosomes during M phase [26].
Tunicamycin which causes extensive protein misfolding and activation of the unfolded protein response (UPR) is an ER-stress inducer [19, 27]. One of the cellular responses to UPR is to exit from the cell cycle by downregulation of Cyclin D1 levels [18]. Cyclin D1 degradation is also necessary for exit of hepatocytes from cell cycle and termination of liver regeneration [28]. HCC cell lines like many cancer cell lines display uncontrolled cell proliferation. In order to force HCC cell lines to exit cell cycle by UPR, we treated them with Tunicamycin and tested ATAD2 protein levels. As shown in Fig. 3, following 12h and 24h treatment with Tunicamycin, Hep3B cells displayed early accumulation of CHOP and strong induction of PARP cleavage, as well as downregulation of ATAD2 levels particularly at 24h. This suggested that Tunicamycin treatment caused a cell cycle arrest as well as apoptosis induction. The response of HepG2 cells was milder in terms of PARP cleavage but ATAD2 downregulation was evident at 12h and strong at 24h, suggesting minor apoptosis and more pronounced cell cycle exit. In support of our conclusion, we noticed that ATAD2 was one of the significantly downregulated genes in G0 as compared to G1 cells [29]. Taken together our observations supported by other reports are in favor of ATAD2 being a biomarker for a non- resting (non G0) state independent of cell cycle phases.
Next, we examined HCC cell response to ATAD2 downregulation. Out of five HCC cell lines tested, only two displayed clear and objective survival response (Online Resource 3). Detailed phenotypic analysis of ATAD2 knock-down effects in four representative cell lines clearly demonstrated that ATAD2 dependency is highly heterogeneous. Although a few HCC cell lines experience decreased survival and cell death, many others don’t show any severe survival defect. Our compared global gene expression analysis also indicates extreme heterogeneity in response to ATAD2 deficiency. Cellular response varied between no detectable alterations in gene expression to more than 100 genes affected. Thus, our observations lead us to conclude that ATAD2 expression is not absolutely necessary for survival of HCC cell lines or its effects are dependent on other factors. In other words, the effects of ATAD2 deficiency appear unpredictable in the contexts we analyzed. In accordance with our previous studies we concluded that ATAD2 depletion related survival outcomes do not associate with metastatic (EMT) status of HCC cells, as well as mutations in p53 or Rb pathway. Two out of 4 epithelial cells (Hep3B and HepG2) responded to ATAD2 depletion but the other 2 epithelial lines (PLC/PRF/5 and Huh7) and SNU449, as the only EMT cell line in our panel, didn’t [30]. Among the responders HepG2 is p53wt, whereas Hep3B is p53 null. Also all the cell lines used in this study, responders and non-responders, are Rb pathway deficient [31]. Therefore, additional studies are necessary to identify markers of response to ATAD2 depletion.
In support of this conclusion, ATAD2 has been qualified previously as a global ‘helper’, which in growing cells can be compensated for by other factors [2]. When compared to previous reports on the effects of ATAD2 suppression on HCC cell lines, our observations correlate only partially. Indeed, similar observations were reported for Hep3B and HepG2 cells [21, 32]. But we could not confirm earlier reports on ATAD2 dependency of Huh7 and PLC/PRF/5 cell lines [21, 32, 33]. The reasons of this discrepancy are not known. Cell lines used in these studies may have undergone drifts during their long-term maintenance between different laboratories. Alternatively, experimental conditions may have affected the phenotypic outcomes. Nevertheless, it is of interest that ATAD2-independent HCC cell survival has not been noticed in previous reports [21, 32, 33]. Here we show that ATAD2 is not a critical gene for the survival of many different HCC cell lines. These findings strongly suggest that this gene is not a preferable target for development of new targeted therapy approaches for HCC unless its function is further characterized. Indeed, as a gene associated with both normal and malignant cell proliferation, even if successful results can be achieved such therapies might present undesirable side effects. Altogether, here we show that ATAD2 is an excellent proliferation marker for liver diseases including HCC. However, currently, it is not a preferable therapeutic target.