Constitutive Photomorphogenesis Protein 1 homolog (COP1) sustains nuclear factor-4 alpha function in human hepatocyte models

Constitutive Photomorphogenesis Protein 1 homolog (COP1) is a conserved E3 ligase with key roles in several biological systems. Prior work in hepatocyte derived tumors categorized COP1 as an oncogene but its role in untransformed hepatocytes remains largely unexplored. Here we have investigated the role of COP1 in primary human hepatocytes as well as in two transformed hepatocyte models, HepG2 and HuH-7 cells. Contrary to a previous report, COP1 suppression via siRNA had no noticeable impact on HepG2 and HuH-7 proliferation and was associated with contrasting rather than congruent transcriptome changes. Clustering analyses identified patterns indicative of perturbed metabolism in primary hepatocytes and HepG2 cells whereas patterns pointed to cell proliferation impacts in HuH-7 cells. In HepG2 and primary hepatocytes, COP1 suppression reduced the expression important hepatic regulators and markers, which could be restored by the introduction of a siRNA resistant COP1 transgene. Strikingly, COP1 downregulation reduced hepatic nuclear factor-4 alpha (HNF4A) abundance and function, as assessed by lower abundance of key HNF4A targets and reduced APOB secretion. HNF4A restoration partially rescued COP1 silencing in HepG2 cells. This study identifies COP1 as a key regulator of hepatocyte function, in part via HNF4A. COP1 was required to maintain HNF4A abundance and function in primary hepatocytes and in HepG2 cells, but not in HuH-7 cells. Lastly, by demonstrating contrasting roles of COP1 in HuH-7 and HepG2 cells, our findings also challenge previous work linking COP1 to hepatic tumorigenesis.

In addition, COP1 has also been linked to oncogenesis.Hypomorphic cop1 mice are susceptible to developing tumors 5 .COP1 deficiency in mouse prostate leads to increased cell proliferation and hyperplasia 3 .Moreover, COP1 induces acute myeloid leukemia in murine overexpression models 13 .In humans with gastric carcinoma, low COP1 levels are associated with a poor prognosis and have been shown to increase proliferation in a gastric cancer cell line 14 .However, previous studies have been inconsistent regarding the role of COP1 function in hepatic cancer, with tumor suppressor or oncogenic properties 5,11,15,16 .HepG2 and HuH-7 cells, two well characterized liver tumor derived cell populations, have been reported to undergo profound growth inhibition following COP1 suppression 15 .COP1 suppression was shown to elicit growth inhibition and apoptotic regression and, in a nude mouse model COP1 siRNA, reduced xenograft growth 15 HepG2 and HuH-7 cells were established over 40 years ago.HuH-7 is an established cell line derived from a 57-year-old Japanese hepatoma whereas the HepG2 cell line was derived from the lobectomy specimen of a 15-year old Caucasian male 17,18 .Importantly, while both cell lines have a neoplastic origin, HuH-7 are hepatocarcinoma cells (HCC) whereas HepG2 cells are hepatoblastoma (HBB), despite the latter being often erroneously referred to as hepatocarcinoma 19 .HBB typically share features reminiscent of fetal liver, are less aggressive and respond better to chemotherapy than HCC 20 .As expected for transformed cell lines, HuH-7 and HepG2 genomes are rearranged and mutated [21][22][23][24] .Notwithstanding their transformed phenotypes, they are widely used as surrogate primary hepatocyte models.But this comes at a price, as these cells are known to differ substantially from plated hepatocytes.For instance, they display impaired lipoprotein secretion 25,26 .The two cell lines also differ functionally from each other in numerous points including their dependence on glucose and lipogenic ability 27,28 .
In this work, COP1 is investigated in HepG2 and HuH-7 cells as well as in primary hepatocytes to shed light on its role in both pathological and physiological liver settings.We report several discrepancies with previously published work on HepG2 and HuH-7 cells.Finally, we identify HNF4A as a functional target of COP1.

Cell culture
HepG2 and HuH-7 cells were maintained in 5 mM Glucose DMEM (Gibco) supplemented with 10% fetal bovine serum and antibiotic supplements (Gibco).HepG2 and HuH-7 cells were obtained from the American Type Culture Collection and the Japanese Collection of Research Biosources Cell Bank, respectively.All siRNAs were of Silencer™ Select grade (Ambion) to minimize off-targeting.HepG2 and HuH-7 cells were transfected at 30%-50% confluence in 0.5 ml media in 24 well plates using 10 nM siRNA (final) and 1 µl of RNAiMax.The siRNA transfection mix was prepared in 100 µl Optimem (Gibco).Treatment was continuous for 96 h.Cryopreserved primary hepatocytes (HMCPMS), media and media supplements were obtained from ThermoFisher Scientific.Two distinct lots, originating from 2 distinct donors, were used: lot Hu8306 (transcription array analysis of COP1 suppressed hepatocytes; 27 YO, BMI 29.5) and Hu8287 (replication with si33 and si91; 49 YO, BMI 19.6) and were thawed in thawing media and then plated in plating media for 6 h in 12-well plates (1 ml media per well) pre-coated with collagen I (1 vial was seeded in 10 and 14 wells, respectively).
Media was changed to culture medium which was replaced every 24 h.Replicates

Expression constructs
COP1 and HNF4A were synthesized in vitro by BioBasic (Markham, Ontario) and assembled in the lentiviral PLVX plasmid.COP1 and HNF4A are tagged with Cterminal FLAG epitope and Hemagglutinin tags, respectively.COP1 has been codon optimized to be compatible with gene synthesis.Silent mutations over the siRNA targeted region (sequence in Supplementary Materials) were included to confer resistance to COP1 siRNA.Endogenous HNF4A is generated from 2 distinct promoters and is alternatively spliced.
The variant used here corresponds to NP_000448.3/HNF4α1, the canonical and fully active form, and includes a C-terminal HA tag.Sequences are detailed in the supplemental materials section.

Viral particle generation and stable cell line generation
Viral particles were generated by co-transfection of 293FT cells with the empty PLVX plasmid or PLVXCOP1(FLAG) with psPAX2 and pMD2.G viral particles harvested during the 16-72 h window post-transfection were concentrated with Lenti-X (Clontech) and stored at -80°C.Stable pools were generated using incremental doses of virus and puromycin resistance (3 µg/ml).Unless specified otherwise, cell populations selected with the lowest doses of virus compatible with puromycin resistance were retained.

Bioinformatic analyses
Mapping of transcriptional consequences of COP1 suppression was performed using the WebGestalt interface (http://www.webgestalt.org/) 29.Over-representation analysis (ORA) and Gene Set Enrichment analyses (GSEA) were used.Affymetrix Gene IDs were mapped via WebGestalt.GSEA was performed using the complete transcriptome fold-change dataset.Nominal changes were used for ORA. Background list was "affy hugene 2.0 stv1".Only FDR significant hits (q < 0.05) are shown.Impacted transcripts were mapped to Kyoto Encyclopedia of Genes and Genomes (KEGG), Gene Ontology (GO) and Reactome categories [30][31][32][33] .For Ingenuity Pathway Analysis (IPA) core analysis, molecules undergoing fold changes > 1.5 (ANOVA p < 0.05) as determined by the Transcriptome Analysis Console (Affymetrix) output were selected.The Ingenuity Knowledge Base (Genes + Endogenous Chemicals) was used as reference set.High confidence and experimentally observed data were included in the analyses.For IPA Upstream Regulator analysis, the analysis examines known targets of upstream regulators present in the dataset provided and compares the direction of the changes observed to the expected direction.Fisher's exact test is used to measure statistical enrichment of impacted targets present in the sample list versus a database of regulator-targets.

Statistical analyses
Statistical significance was tested using (paired) Student's t-test for 2 groups or oneway ANOVA for multiple treatment groups, as indicated.

COP1 suppression is well tolerated in HepG2 and HuH-7 hepatoma cells
While examining the role of COP1 in transformed liver models, we found that COP1 suppression (~85% in HuH-7 and ~70% in HepG2) had little impact on HepG2 and Over-representation analyses of transcriptome changes following COP1 suppression in transformed and primary hepatocytes.
HepG2 and HuH-7 transcriptomes following COP1 silencing showed relatively little overlap, sharing only 423 nominally significant hits (out of 6057 (7% of nominal hits) and 2861 (15% nominal hits) for HepG2 and HuH-7, respectively), indicating divergent trajectories with respect to the role of COP1 (Fig S3).These hits were mapped to pathways (Reactome and the Kyoto Encyclopedia of Genes and Genomes (KEGG)) and Gene Ontology (GO) terms by over-representation analysis (ORA), which identified enrichment of terms linked to cell cycle regulation and DNA metabolism (Table S1).
Comparison of nominally impacted transcripts within each cell type revealed dichotomous responses for HepG2 and HuH-7: transcriptome changes in HepG2 mainly mapped to metabolic processes while in HuH-7, changes were prominently consistent with cell cycle regulation terms (Table 1, Table S2).
These results pointed to divergent roles of COP1 in transformed hepatocyte models.To delineate the physiological role of COP1, COP1 was next targeted in primary human hepatocytes.Consistent with transformed models, siRNA mediated COP1 knock-down resulted in the suppression of COP1 at the mRNA (reduced to 28 ± 10% of control value, 95% C.I.) and protein levels (reduced to 25 ± 28% of control value, 95% C.I.) (Fig 2).The suppression resulted in 4262 nominally changed transcripts, of which only a small fraction was also impacted in HepG2 (848) and HuH-7 cells (299) (Fig S3).Over-representation analysis indicated that primary hepatocytes underwent transcriptome changes mapping primarily to metabolic processes (Table 1, Table S2).Mapping specificity was confirmed by transcript randomization (Table S3).
To obtain a more granular view of these findings, ORA was repeated on nominal transcripts segregated by direction of effect.Enrichment analysis of reduced abundance transcripts revealed an extensive overlap between HepG2 and primary hepatocytes, with an enrichment of metabolic and lipid-related terms (Table S4).
Moreover, the analysis uncovered opposite patterns in HepG2 versus HuH-7: proliferation related gene sets were upregulated in HepG2 but downregulated in HuH-7 (Table S4, pink highlights).By comparison, only one proliferation pathway ("epithelial cell proliferation") was positively enriched in primary hepatocytes.Rather, COP1 suppression resulted in the upregulation of transcripts enriched in immunity related gene sets ("acute inflammatory response", "lymphocyte mediated immunity", etc.) suggesting that COP1 contributes to the regulation of the hepatocyte immune environment.
Gene Set Enrichment Analysis analyses of transcripts in COP1-suppressed transformed and primary hepatocytes.
Transcriptome changes were then submitted to Gene Set Enrichment Analysis (GSEA).
In contrast to ORA, GSEA leverages a ranking approach rather than nominal significance of individual hits to map transcripts of interest to gene sets 34 .Importantly, the analysis provides directionality through a normalized effect size estimate.GSEA analyses were consistent with ORA findings.Findings in primary hepatocytes and HepG2 cells centered on impaired metabolic processes whereas terms related to inhibition in DNA replication and cell cycling were identified in HuH-7 (Table S5).In contrast to HuH-7 cells, transcripts mapping to DNA replication and proliferation were globally increased in HepG2 cells.
Ingenuity pathway analysis points to HNF4A dysfunction in primary hepatocytes and HepG2 cells following COP1 knock-down.
To further delineate the role of COP1, we then performed Ingenuity pathway analysis  S6).
In addition, IPA can infer effects on master regulators ("Upstream Regulators") in line with transcriptome perturbations.The analysis identified several targets of interest (Table S7).Changes in line with reduced HNF4A activity ("HNF4A", "HNF4alpha dimer") and HNF1A/B, two master regulators of liver function, were observed in HepG2 and primary hepatocytes but not in HuH-7 cells.COP1 suppression results in reduced HNF4A and MTTP protein abundance in HepG2 cells and primary hepatocytes.
Western blot analyses confirmed that reduced HNF4A transcript abundance was accompanied by lower protein abundance in both primary hepatocytes, HepG2 but not in HuH-7 cells (Figure 2).MTTP is a major regulator of very low density lipoprotein assembly and its expression requires HNF4A.Consistently, COP1 silencing resulted in a statistically significant reduction of MTTP mRNA in both HepG2 and primary hepatocytes (3.5-fold and 1.69-fold, respectively, in GSE206116).Again, effects in HuH-7 cells differed with a modest upregulation (1.2-fold).Western blot analysis confirmed that MTTP protein abundance was concordantly reduced in hepatocytes and HepG2 cells, but unchanged in HuH-7 cells (Figure 2).
To ensure specificity, COP1 suppression was repeated in a distinct hepatocyte lot, using the other two siRNAs used earlier (si33 and si91).In addition to HNF4A and Limited impact of higher COP1 levels on HepG2 cells.
The consequences of increased COP1 expression were examined next by transducing HepG2 cells with COP1.Silent mutations and a FLAG tag were introduced to render the transcript siRNA-resistant and to facilitate downstream experiments.TRIB1, which was upregulated in the transcriptome analysis and plays a role in sustaining HNF4A function was also included 35,36 .Except for COP1 itself, moderate COP1 HNF4A silencing partially phenocopies COP1 suppression in HepG2 cells.
COP1 suppression reduced HNF4A protein levels in HepG2 and primary hepatocytes.
Since HNF4A is a master regulator of hepatocyte function we hypothesized that the impacts of COP1 suppression might be through impaired HNF4A function 37,38  Identification of pathways co-regulated by HNF4A and COP1.
To estimate the degree to which processes were co-regulated by HNF4A and COP1, expression profiles of COP1 and HNF4A suppressed cells were compared.Data from GSE15991, reporting transcriptome changes in HNF4A silenced HepG2, were extracted and compared with our COP1 silenced findings 39 S8).These results suggest that HNF4A and COP1 jointly regulate a wide range of biological pathways central to liver function.

Discussion
Here we have explored COP1 function in both normal and transformed hepatocytes.
Our transcriptomics findings in HepG2 and HuH-7 differ considerably from those previously published 15 .
In that report, 7 transcripts linked to p53 (including RFWD2/COP1) and a list of 78 genes that exhibited directionally coherent changes were reported to be impacted by COP1 silencing.No such pattern was evident from our data (Fig S9, Table S9).Reasons for this discrepancy are unclear.One possibility is that the previously reported changes resulted from apoptosis engagement, possibly through p53.Indeed, a separate report demonstrated (albeit in non-liver models) that COP1 could mediate p53 turnover, preventing p53 mediated cell death 7 .Thus, COP1 downregulation combined with other stressors, rather than COP1 suppression per se, may activate p53 and lead to apoptotic death.By contrast, we did not find evidence that reduced COP1 level is cytotoxic in either cell model, consistent with lack, or insufficiency, of p53 engagement.
COP1 silencing had significant but dichotomous effects on HepG2 and HuH-7.
Whereas changes elicited by COP1 suppression matched a prominently proliferative profile in HuH-7, perturbations in HepG2 revealed a metabolic portrait reminiscent of primary hepatocytes.Importantly, this work does not answer whether these different requirements for COP1 can be attributed to cell line singularities (e.g., specific mutations) or broader features (e.g., hepatoblastoma vs hepatocarcinoma).One limitation to these findings is that this 3-way transcriptome-wide comparison hinges on a single siRNA type, albeit designed to minimize off-targeting.While two additional siRNAs had similar impacts on a subset of targets of relevance to hepatocyte function, their transcriptome scale impacts were not investigated, and the extent of off-targeting remains unknown.However, it seems unlikely that off-targeting could account for the lack of cytotoxic impact of COP1 silencing observed here or the contrasted transcriptomic signatures.
Interestingly, transcripts could be mapped to the same terms and pathways despite HepG2 cells and hepatocytes sharing only a subset (~8 %) of nominally impacted transcripts.Most were directionally consistent by GSEA (e.g., cellular ketone metabolic processes, (reduced) lipid homeostasis, (reduced) small molecule catabolic process (reduced) etc..), although exceptions were evident.For instance, IPA indicated that COP1 suppression had opposite effects on transcripts linked to cholesterol synthesis (e.g., "Superpathway of Cholesterol Biosynthesis") in HepG2 and hepatocytes.Uniquely, COP1 suppression in hepatocytes resulted in increased abundance of transcripts that were mapped by ORA to gene sets linked to inflammation.
Although how it contributes to inflammation is still unclear, several mediators of liver inflammation featured prominently, including IL1B (1.9-fold increase) and IL1R1 (1.6fold increase), suggesting that COP1 plays an anti-inflammatory role.This is supported by IPA which predicts an activation of IL1 (IL1, IL1A and IL1B) signaling following COP1 knock-down (Table S6).
Leveraging siRNA experiments in HepG2 cells, HNF4A was identified as an important effector regulating a subset of COP1 targets.Importantly, COP1 restoration using a siRNA resistant expression construct could largely normalize HNF4A levels in HepG2 cells, ensuring siRNA specificity.Although restoration experiments were not performed in primary hepatocytes, our siRNA results indicated that COP1 is also instrumental in supporting HNF4A function therein.Comparison of FDR significant hits in HNF4Asi and COP1si samples identified unique and shared gene targets.
Transcripts impacted by COP1 silencing represented a small subset (215/4785) of HNF4A responsive genes while a large fraction (215/515) of COP1si impacted genes were also sensitive to HNF4A silencing.This analysis was conservative, as several transcripts of potential biological importance were excluded because they did not reach FDR significance.Indeed, as the extent of HNF4A reduction associated with COP1 knock-down is less pronounced than that achieved via a targeted HNF4A approach, sufficient HNF4A may persist in COP1 silenced cells to sustain near normal or not FDR significant changes in the expression of target genes.A more permissive approach using nominally significant hits (not corrected for the number of targets investigated) suggests that 20-25% of the transcriptome requires both COP1 and HNF4A (data not shown).
In view of the essential role of HNF4A in hepatocyte function, future work will aim to clarify how COP1 increases HNF4A levels 40,41 .The COP1-HNF4A relationship appears largely unidirectional as COP1 expression was unaffected by HNF4A targeting.
Whether HNF4A can affect COP1 function in other ways remains untested.COP1 is an E3 ligase that targets cognate proteins for degradation and although HNF4A can be ubiquitylated, our data are inconsistent with COP1 directly promoting HNF4A degradation 42,43 .For one, COP1 suppression led to decreased, rather than increased, HNF4A protein abundance.Moreover, COP1 overexpression did not reduce HNF4A protein level.Thus, the physical interaction we previously reported between HNF4A and TRIB1, a COP1 adaptor protein involved in CEBPA degradation, is unlikely to contribute to HNF4A degradation; indeed, TRIB1 was positively correlated with HNF4A abundance in that study 36 .Rather, COP1 may promote the degradation or inactivation of HNF4A transcription repressors, such as SNAI1 or NF-kappaB [44][45][46] .Of note, GSEA found NF-kappaB signaling to be increased in COP1 suppressed HepG2 cells (Table S5).
Furthermore, changes consistent with increased "TNF" (i.e., TNFA) activity were identified by IPA in HepG2 and hepatocytes, consistent with activation of NF-kappaB signaling (Table S7).Future studies will examine this possibility.
In summary, we demonstrated that HepG2 and HuH-7 could proliferate overtly normally despite reduced COP1 expression.Moreover, COP1 was required to maintain prominent hepatocyte transcripts, including HNF4A, in primary hepatocytes and HepG2 cells, but not in HuH-7 cells.Thus, HepG2 cells represent a better model system to further clarify the role of COP1 in liver physiology. in Supplementary Tables (Table S7).

HuH- 7
cell abundance (Fig 1A, B).This observation contrasts with previously published results, showing that a comparable COP1 suppression strongly impaired proliferation of both cell types 15 .Western blotting for COP1 confirmed COP1 suppression at the protein level (Fig 1C).To ensure specificity, HuH-7 and HepG2 cells were treated with 2 additional siRNAs, neither of which showed toxicity despite comparable COP1 suppression (Fig S1).As suppression of COP1 mRNA expression was comparable across both studies, other sources of discrepancy were considered, including the use of different transfection reagents, Lipofectamine 2000 versus RNAiMAX, used in the current study.We reasoned that Lipofectamine 2000 may synergize with COP1 suppression to be cytotoxic.This was tested in HuH-7; however, findings were comparable to RNAiMax (Fig S2).While the reasons behind this discrepancy were not further investigated, COP1 suppression in the absence of significant cell death warranted a fresh assessment of COP1 function in these 2 cell lines.

(
IPA).Like GSEA, IPA leverages changes in direction and magnitude to infer functionality but it uses its own set of curated pathways (canonical pathways) and a different analysis pipeline.Mapping transcriptome changes to canonical pathways following COP1si was largely supportive of earlier analyses (Fig S4; Fig S5; Table MTTP, CEBPA, another major liver transcription factor, reduced in GSE206116 (-1.8fold, p = 7.8e-5), was also measured.COP1 suppression resulted in a trend toward reduced HNF4A expression, reduced CEBPA and MTTP abundance, replicating the earlier findings (Fig S6A).Importantly, compromised HNF4A function was ascertained by a reduction of Apolipoprotein B-100 abundance in the media in silenced cells (Fig S6B).
overexpression did not alter transcripts tested, although TRIB1 and MTTP were slightly increased (Fig S7) with no effects on protein levels other than that of COP1.Introduction of a siRNA resistant COP1 form partially normalizes transcript abundance.Since COP1 overexpression had limited downstream effects, we next verified that the siRNA-elicited changes were specific to COP1.COP1 silencing was repeated in pools of HepG2 cells expressing a siRNA resistant COP1.Introduction of the siRNA resistant COP1 (resulting in a 2-3-fold increase over basal level) provided a partial protection as assessed by a return towards baseline of HNF4A, MTTP, APOC3 (another prominent target of HFN4A) andTRIB1 (Fig 3A, B).Changes at the protein level were confirmed by western blot for HNF4A, MTTP and COP1, for which suitable antibodies were available (Fig 3 C).

Figure 1 .
Figure 1.COP1 suppression in HepG2 and HuH-7 cells does not affect proliferation.HuH-7 and HepG2 cells were transfected with a COP1 targeting siRNA (COP1si) or a non-specific control (NT1) for 96 h prior to analysis.A, RNA isolation and quantification by qRT-PCR.B, cell abundance estimation by Alamar blue of siRNA treated cells.Three biological repeats (and means ± SD) are shown.Statistical significance of NT1si vs COP1si values was assessed using Students t-test.** p<0.01.C, Western blot of HuH-7 and HepG2 cells treated with siRNA or transduced (for HepG2) with a flagCOP1 expressing virus for 96 h.Quantification representative of 3 biological repeats.TUBB, Tubulin beta chain.

Figure 2 .Figure 4 .Figure 5 .
Figure 2. Western blots of COP1si treated cells.Primary hepatocytes (Hepato), HepG2 cells and HuH-7 cells were treated with COP1si or a Non-target control (NT1) for 96 h.Western blot is representative of 2 independent biological repeats for primary hepatocytes, and 4 for HepG2 and HuH-7 cells.In each cell type, the same blot was probed sequentially (COP1, MTTP, HNF4A, TUBB).Quantification is shown after correcting for TUBB (Tubulin beta chain) intensity and normalizing to the corresponding NT1si value.

Table 2 ,
. HNF4A suppression led to perturbations of about a third of the gene transcripts.A comparison of the FDR Table

Table 1 . Over-representation analysis of transcripts nominally affected in COP1 silenced HepG2, HuH-7 and primary hepatocytes.
Transcriptome data were mapped Gene Ontology terms (Biological_Process_noRedundant).The top five FDR significant hits are shown for each cell type.A complete table (including KEGG and Reactome sets) is included as TableS2.