DNAJB1-PRKACA in HEK293T cells induces LINC00473 overexpression that depends on PKA signaling

Fibrolamellar carcinoma (FLC) is a primary liver cancer that most commonly arises in adolescents and young adults in a background of normal liver tissue and has a poor prognosis due to lack of effective chemotherapeutic agents. The DNAJB1-PRKACA gene fusion (DP) has been reported in the majority of FLC tumors; however, its oncogenic mechanisms remain unclear. Given the paucity of cellular models, in particular FLC tumor cell lines, we hypothesized that engineering the DP fusion gene in HEK293T cells would provide insight into the cellular effects of the fusion gene. We used CRISPR/Cas9 to engineer HEK293T clones expressing DP fusion gene (HEK-DP) and performed transcriptomic, proteomic, and mitochondrial studies to characterize this cellular model. Proteomic analysis of DP interacting partners identified mitochondrial proteins as well as proteins in other subcellular compartments. HEK-DP cells demonstrated significantly elevated mitochondrial fission, which suggests a role for DP in altering mitochondrial dynamics. Transcriptomic analysis of HEK-DP cells revealed a significant increase in LINC00473 expression, similar to what has been observed in primary FLC samples. LINC00473 overexpression was reversible with siRNA targeting of PRKACA as well as pharmacologic targeting of PKA and Hsp40 in HEK-DP cells. Therefore, our model suggests that LINC00473 is a candidate marker for DP activity.


Response:
We appreciate the suggestions provided by Reviewer #1, which have further strengthen our manuscript. However, we DO NOT claim that our model is "a suitable model for the study of this carcinoma". In fact, in the discussion section of the manuscript we had clearly stated "One of the limitations of our cell model is that it is not a suitable model to study the oncogenic mechanism of DP in liver cells since the HEK293 cells are not of the same origin and at baseline have a different gene expression pattern." We also had stated the following in the Discussion section: "Therefore, our engineered HEK-DP cell lines may have the potential to provide a platform for understanding the regulatory link between DP and LINC00473 expression, which is present in FLC tumors, and provide the field with a cellular system for screening DP-specific inhibitors by using LINC00473 as a candidate marker for DP activity." While the underlying idea to generate these cell lines has merit, the study design is fundamentally flawed and most of the conclusions are therefore not supported by the data.
Major Concerns: 1) Throughout this manuscript, the authors directly compare two monoclonal cell lines, which have undergone transfection, FACS and single-cell selection, to an apparently completely untreated bulk population that has not undergone any of these procedures. That this is highly problematic is evident in the authors own data, as the variability between the individual clones, where examined, is extensive (e.g., Fig. 2, Fig. 3, Fig. 7b). An appropriate experimental setup must include several wildtype clones that have undergone the same treatment, including transfection with CRISPR-Cas9 plasmids that do not carry a sgRNA sequence. As these controls were not included, the vast majority of the experiments reported here are not sound and unfortunately cannot be used to draw conclusions.

Response:
We appreciate the input. We have subsequently developed 5 clones from wild-type HEK293 cells which have undergone transfection with vectors lacking guide RNA sequences used for construction of HEK-DP cells, FACS, and single-cell selection. These clones did not demonstrate an increase in LINC00473 expression compared to wild-type cells. Results have been added to the supplementary data.
2) For their microarray analysis, the authors do not use replicates. While suboptimal, this can be done but should at least be evaluated with a more stringent fold-change cut-off. Typically, genes at the lower end of expression fluctuate considerably in microarray analyses and should be excluded from analysis, which the authors apparently neglected to do. To provide an assessment of this variability, the authors should show an MA plot.

Response:
The purpose of the microarray was to perform a cursory examination of the transcriptome in HEK-DP cells. In Figure 2, we have only included genes with greater than 2-fold change. We have not performed any GO analysis or proposed any pathways of significance since we do not think this is directly relevant to FLC biology. However, through the microarray and subsequent qPCR at Vakili lab and independently at Sethupathy lab, we identified LINC00473 as an upregulated transcript leading us to pursue whether this could be a marker for DP activity. Even if we set our fold-change at a higher cut-off of 5, we would still identify LINC00473 as a significantly upregulated transcript.
3) The authors provide no statistical analysis of the data presented in Fig. 3.

Response:
Since the focus of our work is on clones A9 and A11, we have revised the figure. The updated figure demonstrates statistical LINC00473 expression in A9 and A11. Fig. 4 and 5, the authors do not include wildtype controls at all. It would also have been interesting to see if H89 treatment affects the levels of CGA.

Response:
Since there is a 10-fold difference in LINC00473 expression between wildtype HEK cells and the HEK-DP clones, we did not feel it would be relevant to include wildtype cells for these experiments.

5)
The difference in proliferation between wildtype and the deletion clones is wholly unconvincing, in particular in the absence of proper wildtype controls.

Response:
We agree that the difference is not robust in the first 3 days, however, statistically significant after 3 days in A11 cells. Our hypothesis is NOT that the presence of DP will increase proliferation. Therefore, we did not perform extensive experiments to measure proliferation rates of the clones in response to DP targeting due to small difference in the rates of proliferation between HEK-DP and HEK-WT cells. We do hypothesize that once a cell transforms into a cancer cell, then the proliferation rate changes significantly. However, our HEK-DP cells are not considered cancer cells and it may take several generations of the clones and continued exposure to DP before they start demonstrating significant alterations in their proliferation rates. We simply included this graph as we expected that reviewers will ask about the proliferation rates of the DP clones.
6) There is no validation of the mass spectrometry results shown in Fig. 7.

Response:
We have now included a validation experiment in order to assess the interaction of BAG2 and DP based on the proteomic data. Please see Supplementary Figure S3. The co-IP experiments demonstrate interaction between BAG2 and DP as well as between BAG2 and PKA-C Based on the proteomic data and the intensity of the bands on the blots, the affinity of DP to form a complex with BAG2 may be higher than PKA-C. We will study this interaction in more detail in the future. We have outlined the potential importance of DP-BAG2-HSP70 interaction in the Discussion section of the manuscript.
7) The data on mitochondrial fission would be considerably more convincing had the authors shown some sort of rescue experiment, e.g. with a knockdown of the DNAJB1-PRKACA gene.