A cellular model to study HNRNPK essentiality

We aimed to identify genes whose loss alleviates, or exacerbates, the impaired cellular fitness caused by the depletion of hnRNP K. As model systems, we chose the human glioblastoma-derived LN-229 and U-251 MG cell lines which express high levels of HNRNPK [2,3], a key factor for optimizing our synthetic-viability screen. We employed a plasmid harboring quadruple non-overlapping guide RNAs (qgRNAs), driven by four distinct constitutive promoters, to target the human HNRNPK gene and maximize editing efficiency in polyclonal LN-229 and U-251 MG cells stably expressing Cas9 [32]. Seven days after qgRNA lentiviral delivery, we observed a substantial reduction in hnRNP K protein followed, as expected, by a drop in cell viability (Fig 1A-1D). A minor fraction of LN-229 cells exhibiting no Cas9 expression did not undergo HNRNPK ablation, resulting in incomplete cell death (Fig 1A and 1D). To address this issue, we isolated by limiting dilutions LN-229 single-cell clones expressing Cas9 (S1A Fig). We compared Cas9 activity in seven individual clones using an eGFP reporter and selected LN-229 clone C3 (S1B Fig). When we tested the HNRNPK ablation efficiency in LN-229 C3 cells, we observed complete protein depletion and cell death (Fig 1A and 1B). Interestingly, U-251 MG Cas9 cells showed delayed cell death compared to LN-229 C3 (Fig 1A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. A. Cell viability after delivery of HNRNPK or non-targeting (NT) qgRNAs (CellTiter-Glo assay).

Results are normalized against the NT condition. 10 individually treated wells. B-D. Cas9 and hnRNP K protein upon delivery of HNRNPK (+) or NT (-) qgRNA. E. Genome-wide CRISPR deletion screen. F. Volcano plot showing differential sgRNAs abundance in the HNRNPK vs. NT comparison at day 14. G-H. Cell viability after sequential co-deletion of each candidate gene and HNRNPK (CellTiter-Glo assay). Results are normalized against the seeded cell density before HNRNPK ablation and compared to the double NT+NT condition. Red empty circles: non-targeting control (NT) and non-specific genes (PCNA, GPKOW, PRNP). Black-filled circles: genes confirmed only in LN-229 C3 cells. Yellow-filled circles: genes confirmed in both LN-229 C3 and U251-Cas9 cells. n ≥ 3. Data information: n represents independent experiments. Mean ± SEM. **: p < 0.01; ***: p < 0.001; ****: p < 0.0001 (Two-way ANOVA Tukey’s test in A. One-way ANOVA Dunnett’s test in G-H).

https://doi.org/10.1371/journal.ppat.1014056.g001

To confirm that the observed lethality resulted from the absence of hnRNP K, we transduced LN-229 C3 cells with constructs encoding the HNRNPK ORF sequence under transcriptional control of the elongation factor 1α (EF-1α) promoter. We then utilized intron-targeting single-guide RNAs (sgRNAs) to selectively ablate the endogenous HNRNPK gene (S1 Table). The cell death resulting from HNRNPK deletion was suppressed by the exogenous constructs, confirming the specificity of the lethal phenotype and the reliability of this cellular model (S1C and S1D Fig).

Genome-wide CRISPR ablation screen for the identification of HNRNPK epistatic interactors

To identify functionally relevant epistatic interactors of HNRNPK, we conducted a whole-genome ablation screen in LN-229 C3 cells using the Human CRISPR Brunello pooled library [33], which targets 19,114 genes with an average of four distinct sgRNAs per gene, each expressed by a separate plasmid (total = 76,441 sgRNA plasmids). The lentiviral transduction of the Brunello library was followed by six days of antibiotic selection and subsequent lentiviral delivery of qgRNA vectors containing either HNRNPK-specific or non-targeting (NT) qgRNA guides. Cells underwent antibiotic selection for six more days before harvesting and gDNA extraction (Figs 1E and S2A). sgRNAs distribution was analyzed by next-generation Illumina sequencing (NGS) at the beginning of the screen after library transduction (Day 1) and at the screen endpoint (Day 14; S2A Fig).

Two independent screens were conducted on different days and yielded a robust correlation indicative of satisfactory technical performance (S2B Fig). When using the DepMap repository [30,31] to compare the representation of essential genes in LN-229 cells at Day 14 NT vs. Day 1 (S2 Table), we found that 75% of the sgRNAs targeting known essential genes were efficiently depleted (log₂ fold change ≤ -1, FDR ≥ 0.01) with >92% of those essential genes having ≥2 sgRNAs dropped below threshold (S2C-S2E Fig).

We then listed genes whose sgRNAs were over- or underrepresented in the HNRNPK vs. NT at day 14, reasoning that their deletion would modify the lethality resulting from hnRNP K removal. We obtained a list of 763 and 37 significantly enriched and depleted genes, respectively (log₂ fold change ≥1 or ≤ -1, FDR ≥ 0.01; S2F Fig and S3 Table). Pathway analysis of genes enriched with ≥2 sgRNAs yielded gene ontology (GO) terms related to ribosomal biogenesis, tRNA processing, non-coding RNA metabolism, and translation, consistent with the known roles of hnRNP K in RNA metabolism (S2G Fig). Accordingly, ablation of HNRNPK in LN-229 C3 cells showed a progressive reduction in global protein synthesis (S2H Fig). Also, the GO analysis highlighted “tRNA wobble base modification” as the most overrepresented GO term (S2G Fig). Genes encoding for Elongator complex proteins (ELPs), which are included in this pathway, were significantly enriched in the screen (Fig 1F and S3 Table), suggesting that their deletion counteracts the deleterious effects of HNRNPK ablation. Previous CRISPR screens showed that the absence of ELPs prevents apoptosis in metastatic gallbladder cancer (GBC) [34]. Accordingly, our screen also showed enrichment of other general proapoptotic factors, including AIFM1, MFN2, and FADD (S3 Table).

Among the most profoundly depleted genes were PCBP1, PCBP2, and HNRNPA1, all of which belong to the same genetic superfamily as HNRNPK (S3 Table) [35,36]. The synthetic lethality deriving from their deletion suggests that these genes cooperate with hnRNP K in cell-essential processes. Conversely, the screen was enriched for sgRNAs targeting CPSF6, NUDT21, and XRN2, which form protein complexes with hnRNP K and regulate RNA maturation processes (S3 Table) [7,37]. Hence, our screen successfully identified HNRNPK functional partners with both sensitivity and specificity despite the detection of additional, less specific cell death modulators.

TFAP2C ablation suppresses HNRNPK loss-of-function

To prioritize biologically relevant hits among the 763 enriched genes, we focused on those showing enrichment of all four sgRNAs (S3 Table). We applied the STRING database [38] to assess protein-protein interactions and biological pathways associated with these genes. Next, we examined whether any other genes enriched in the screen scored as interactors. This allowed us to identify and select hierarchical functional clusters among our hits (S4 Table). In parallel, we ranked the 763 enriched genes by multiplying their False Discovery Rate (-log₁₀ FDR) with their effect size (log₂ fold change; S4 Table), as previously described [39,40]. We focused on genes with ≥2 sgRNA scoring in the top 100 rankings, with ≥1 sgRNA among the top 35. The intersection of this ordered list with the STRING clusters yielded 19 genes (Fig 1F and S4 Table). We generated a second list including those genes that, independently from the ranking and clustering, had ≥ 2 sgRNAs enriched and scored as non-essential in the Day 14 NT vs. Day 1 comparison (maximally one sgRNA with log₂ fold change ≤ -1, FDR ≥ 0.01; Fig 1F and S4 Table).

Based on these two lists, 32 genes were selected for validation in LN-229 C3 cells. We ablated each gene individually using qgRNAs and then deleted HNRNPK. Two different non-targeting (NT) qgRNAs were used as controls for the two sequential ablations.

We excluded genes whose deletion enhanced or impaired cell viability by >50% (S2I and S2J Fig) independently of hnRNP K ablation. 16 of the initial 32 hits increased viability of ΔHNRNPK cells by >2-fold (p < 0.01) (Fig 1G). We then tested these 16 genes also in U-251 MG Cas9 cells (henceforth abbreviated as U251-Cas9 cells) at a log₂ fold threshold of ≥ 0.5. We confirmed a total of 9 hits (Fig 1H), including the ELPs gene IKBAKP and the transcription factor TFAP2C, the two strongest hits identified in LN-229 C3 cells. However, in the U251-Cas9 the rescue effect did not always fall within the exact range observed in LN-229 C3 cells, likely due to intrinsic differences between the two cell lines.

TFAP2C (Transcription Factor AP-2γ) was particularly interesting because it regulates the expression of several long non-coding RNAs (lncRNAs) [41–43] and has critical roles in neurodevelopmental processes [44,45] similar to HNRNPK [9,17,18,37,46,47]. Moreover, both HNRNPK and TFAP2C have been described to modulate glucose metabolism [48–50]. Therefore, we elected to explore the epistatic interaction between these two genes.

To consolidate the observations above, we repeated the experiments described in Fig 2G and 2H by individually deleting only TFAP2C in 10 distinct replicas (S3A and S3B Fig). As an orthogonal means of confirmation, we assessed the clonogenic potential of the respective ablated cells (Fig 2A and 2B). Again, the deletion of TFAP2C suppressed the cell death induced by removing hnRNP K in both LN-229 C3 and U251-Cas9 cells, whereas TFAP2C ablation alone only slightly reduced their growth rate. Thus, the loss of TFAP2C did not induce any intrinsic pro-survival effect, pointing to a specific epistatic interaction between TFAP2C and HNRNPK.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. A-B. Clonogenic fitness of ΔTFAP2C or NT-unmodified cells after delivery of HNRNPK or NT qgRNAs (Crystal violet staining).

Western blot: hnRNP K and Tfap2c protein 7 (A) and 10 (B) days after qgRNAs delivery. n = 3. C, E. Tfap2c protein after HNRNPK ablation. n = 5. D, F. qRT-PCR showing TFAP2C RNA in LN-229 C3 (D) and U251-Cas9 (F) cells upon HNRNPK deletion. n = 3 and 5. G. Viability course of LN-229 C3 and U251-Cas9 cells untransduced or overexpressing TFAP2C or mCherry after delivery of HNRNPK (+) or NT (-) qgRNAs (CellTiter-Glo assay). Results are normalized against the NT condition. 10 individually treated wells. Western blot: hnRNP K, Tfap2c, and mCherry protein 7 (LN-229 C3) or 10 days (U251-Cas9) after qgRNAs delivery. H-I. Co-immunoprecipitation of Tfap2c and hnRNP K in low-detergent conditions after benzonase nuclease digestion. IP: Immunoprecipitated Protein; FT: Flow-Through after immunoprecipitation. Data information: qRT-PCR results are normalized against GAPDH expression. n represents independent experiments. f.c.: fold change. Mean ± SEM. *: p < 0.05, **: p < 0.01, ****: p < 0.0001 (Unpaired t-test in D and F. Two-way ANOVA Šídák’s test in G).

https://doi.org/10.1371/journal.ppat.1014056.g002

TFAP2C upregulation sensitizes cells to the loss of HNRNPK

Following HNRNPK ablation, we observed an increase in TFAP2C RNA and protein amount in LN-229 C3 cells (Fig 2C and 2D). This suggested that the toxicity caused by hnRNP K deletion may be due to TFAP2C upregulation. However, TFAP2C overexpression using quadruple-guide CRISPR activation [32] in HNRNPK^+/+ LN-229-dCas9-VPR cells did not impair viability (S3C Fig). In U251-Cas9 cells, HNRNPK ablation had the opposite effect and decreased both the RNA and protein levels of TFAP2C (Fig 2E and 2F), potentially explaining their relative resilience to HNRNPK ablation (Fig 1A) and the smaller protective effect mediated by TFAP2C deletion in this cell line (Figs 2A and 2B and S3A and S3B).

To test if TFAP2C overexpression sensitizes cells to the loss of HNRNPK, we produced stable LN-229 C3 and U251-Cas9 lines overexpressing TFAP2C or mCherry for control and measured their viability after hnRNP K removal. Overexpression of TFAP2C induced a significant acceleration of cell death in both lines (p < 0.0001; Fig 2G), confirming the genetic relationship between TFAP2C and HNRNPK and highlighting a causative dependency between their expression levels and cell death.

Although we efficiently ablated HNRNPK in LN-229 and U-251 MG cells, we were unable to overexpress it, possibly due to tight autoregulatory feedback loops. However, while Tfap2c levels changed upon HNRNPK deletion, neither the ablation nor the overexpression of TFAP2C significantly modified the levels of hnRNP K in LN-229 and U-251 MG cells (S3D-S3I Fig), disproving the existence of a transcriptional feedback loop between HNRNPK and TFAP2C.

Nuclear colocalization and interaction between hnRNP K and Tfap2c

We then asked whether hnRNP K and Tfap2c proteins physically interact and modulate their reciprocal subcellular localization. When staining hnRNP K and Tfap2c by immunofluorescence in LN-229 C3 and U251-Cas9 cells, we noticed a nuclear overlap between the two proteins but no change in their subcellular distribution after ablation of TFAP2C or HNRNPK (S4A and S4B Fig). Under no-detergent conditions, we observed a weak co-immunoprecipitation between hnRNP K and Tfap2c in LN-229 C3 and U251-Cas9 cells (S4C-S4E Fig). To validate this interaction, we employed low-detergent lysis conditions to enhance protein extraction and solubilization, along with benzonase nuclease treatment to exclude potential DNA-mediated associations (Figs 2H and 2I and S4F and S4G). These experiments confirmed a weak but specific reciprocal co-immunoprecipitation, indicating a potential direct interaction between hnRNP K and Tfap2c.

HNRNPK and TFAP2C control the activation of caspase-dependent apoptosis but not ferroptosis

We asked if the deletion of HNRNPK causes apoptosis. We detected increased levels of PARP and Caspase 3 cleavage in LN-229 C3 cells upon ablation of HNRNPK (Fig 3A). Conversely, the prior removal of TFAP2C limited the cleavage of these two proteins (Fig 3A), suggesting that TFAP2C ablation prevents HNRNPK-induced apoptosis. PARP cleavage had the same pattern in U251-Cas9 cells; however, these cells did not show cleavage of Caspase 3 upon deletion of HNRNPK, TFAP2C, or both (Fig 3B). We then asked whether the pan-caspase inhibitor Z-VAD-FMK reduced the lethality resulting from HNRNPK deletion. Z-VAD-FMK decreased cell death consistently and significantly in LN-229 C3 and U251-Cas9 cells transduced with HNRNPK ablation qgRNAs (Fig 3C and 3D), confirming that HNRNPK deletion promotes cell apoptosis. However, we observed that viability recovery plateaued already at the lowest concentration (2 nM) without further increase at higher doses, suggesting a saturation effect. This indicates that while caspase inhibition alleviates part of the cell death, HNRNPK loss triggers additional mechanisms beyond apoptosis. We then treated LN-229 C3^ΔTFAP2C and U251-Cas9^ΔTFAP2C cells with increasing concentrations of staurosporine, a potent apoptosis inducer. LN-229 C3^ΔTFAP2C cells were only mildly resistant to the apoptotic action of staurosporine, and U251-Cas9^ΔTFAP2C showed even lower and minimal recovery (Fig 3E and 3F). These results indicate that TFAP2C plays a limited role in apoptosis regulation and suggest that its suppressive effect on HNRNPK essentiality is not mediated through direct modulation of apoptosis but rather through upstream processes that eventually converge on it.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. A-B. PARP and Caspase 3 protein cleavage ratio after HNRNPK and TFAP2C deletion.

4 hours 1 μM staurosoprine (STS) was used as positive control. n = 3. C-D. Viability of cells 7 (C) or 10 (D) days after HNRNPK or NT qgRNAs transduction and supplementation of Z-VAD-FMK (CellTiter-Glo assay). Results are normalized against the DMSO/NT condition. Multiple comparison was made between Z-VAD-FMK and DMSO treatments in ΔHNRNPK cells. 6 individually treated wells. E-F. Viability of ΔTFAP2C cells treated with staurosporine or DMSO (CellTiter-Glo assay). Results are normalized against the DMSO-treated cells. 4 individually treated wells. Data information: n represents independent experiments. Mean ± SEM. ns: p > 0.05, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001 (Two-way ANOVA Uncorrected Fisher’s LSD in A-B, Dunnett’s test in C-D, and Šídák’s test in E-F).

https://doi.org/10.1371/journal.ppat.1014056.g003

To investigate whether HNRNPK and TFAP2C ablation modulate additional cell death pathways, we explored their potential involvement in ferroptosis, a form of cell death marked by lipid peroxides accumulation, and reported to be modulated by TFAP2C [43,51]. We inquired whether ferroptosis was activated upon ablation of HNRNPK. In LN-229 C3 and U251-Cas9 cells, deletion of HNRNPK was associated with reduced glutathione peroxidase 4 (GPX4) protein abundance (although not statistically significant in LN-229 C3; p ≈ 0.08), whereas deletion of TFAP2C increased it (S5A and S5B Fig). This last result was interesting as a previous study reported that Tfap2c enhances GPX4 expression [51]. Thus, the observed increase upon TFAP2C deletion suggests additional layers of regulation, potentially involving compensatory mechanisms. Moreover, the ablation of HNRNPK led to higher lipid peroxidation in LN-229 C3 cells, which was increased in the absence of TFAP2C (S5C Fig). To investigate this further, we challenged LN-229 C3^ΔTFAP2C and U251-Cas9^ΔTFAP2C cells with increasing doses of erastin, a commonly used ferroptosis inducer. TFAP2C ablation did not prevent erastin toxicity (S5D and S5E Fig). Additionally, different anti-ferroptosis drugs did not suppress the lethality of hnRNP K-depleted LN-229 C3 cells (S5F Fig). These results suggest a role for HNRNPK and TFAP2C in modulating the levels of GPX4. However, ferroptosis seems only marginally connected to the essentiality of HNRNPK and is unlikely to be the primary toxic pathway activated by its removal.

HNRNPK deletion perturbs the transcriptional regulation of genes involved in lipid and glucose metabolism

Apoptosis is often activated downstream of various stress signaling pathways. To gain a mechanistic understanding of the early events shaping the genetic interaction between TFAP2C and HNRNPK, we sequenced total RNA in LN-229 C3 cells depleted of hnRNP K, Tfap2c, or both (S5 Table). We lysed cells four days after delivery of HNRNPK qgRNAs. At this time point, ablation was already extensive (Figs 1B and S6A), yet cell viability was mostly preserved (Figs 1A and 2A). Gene Set Enrichment Analysis (GSEA) showed significant downregulation of genes involved in sterol biosynthesis, secondary alcohol metabolism, and fatty acids catabolism (FDR < 0.05) after removal of hnRNP K (ΔHNRNPK vs. NT) (Fig 4A). Conversely, the most upregulated genes in LN-229 C3^{ΔTFAP2C;ΔHNRNPK} versus LN-229 C3^ΔHNRNPK cells pertained to lipid metabolism and lysosomal functions, including sterol and secondary alcohol metabolic processes (Fig 4B). The intersection of these upregulated genes (log₂ fold change ≥ 0.5) with genes downregulated upon depletion of hnRNP K (log₂ fold change ≤ -0.5) yielded a significant enrichment of GO terms related to energy production and catabolic functions, including carbohydrate metabolism (Fig 4C and 4D and S5 Table). Factors involved in lipid metabolism and cholesterol biosynthesis like FASN, SREBF1, LSS, and DHCR7, and genes related to glycolysis and the pentose phosphate pathway, including PGLS, G6PD, TPI1, H6PD, and GPI, underwent consistent bidirectional regulation (Fig 4C and S5 Table). These data suggest that the transcriptional regulation of lipid and glucose metabolism is imbalanced by the removal of hnRNP K and is partially restored by TFAP2C deletion.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. A-B. RNA-seq Gene Set Enrichment Analysis. Shown are the 20 most negatively (A) and positively (B) enriched GOBP terms (Gene Ontology Biological Process), respectively, in ΔHNRNPK vs. NT (A) and ΔTFAP2C;ΔHNRNPK vs. ΔHNRNPK (B). NES: Normalized Enrichment Score. C. Intersection of RNA-seq data resulting from the ΔTFAP2C;ΔHNRNPK vs. ΔHNRNPK and the ΔHNRNPK vs. NT comparisons. D. Gene enrichment of the biological process for the inversely regulated genes in C.

https://doi.org/10.1371/journal.ppat.1014056.g004

HNRNPK and TFAP2C regulate cell metabolism and bioenergetics via mTOR and AMPK

The deletion of TFAP2C and HNRNPK caused a broad transcriptional rewiring of cell bioenergetics and metabolism (Fig 4 and S5 Table). Accordingly, LN-229 C3 cells had reduced intracellular ATP four days after delivery of HNRNPK ablation qgRNAs. Conversely, ATP levels slightly increased after TFAP2C deletion and remained high when both genes were ablated (Fig 5A). AMP-activated protein kinase (AMPK) is a well-established ATP/AMP sensor, and it is phosphorylated and activated in the context of low cellular energy [52]. Accordingly, we observed increased AMPK phosphorylation (pAMPK/AMPK ratio and pAMPK/Actin) upon ablation of HNRNPK, with a trend toward reduction in LN-229 C3^ΔTFAP2C cells (S6B Fig). LN-229 C3^{ΔTFAP2C;ΔHNRNPK} cells also showed a reduction of pAMPK/AMPK ratio relative to LN-229 C3^ΔHNRNPK cells, although absolute AMPK phosphorylation (pAMPK/Actin) remained high (S6B Fig). These results suggest that hnRNP K depletion causes an energy shortfall, leading to cell death.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. A. Intracellular ATP level in ΔTFAP2C or NT-unmodified LN-229 C3 cells 4 days upon delivering HNRNPK and NT qgRNAs. 5 individually treated wells. B. Representative blot for LC3-II protein in LN-229 C3 cells after HNRNPK and TFAP2C ablation. Quantification includes blot shown in S6C Fig. 4 hours 100 μM chloroquine (CQ). n = 4. C-D. mTOR RNA (C) and protein (D) upon deletion of HNRNPK and TFAP2C in LN-229 C3 cells. 6h HBSS-starvation (Starv.) in D was used as a positive control. qRT-PCR: n = 6. WB: n = 3. E. 4EBP1 and S6 protein phosphorylation after HNRNPK and TFAP2C ablation in LN-229 C3 cells. 6h HBSS-starvation (Starv.) was used as positive control. n = 3. F. mTOR and 4EBP1 and S6 protein phosphorylation in LN-229 dCas9-VPR cells upon TFAP2C overexpression. n = 4. G. Hypothesized mechanism of TFAP2C and HNRNPK in cell metabolism regulation. Data information: qRT-PCR results are normalized against GAPDH expression. n represents independent experiments. f.c.: fold change. Mean ± SEM. ns: p > 0.05, ^*: p < 0.05, ^**: p < 0.01, ^***: p < 0.001, ^****: p < 0.0001 (Two-way ANOVA Uncorrected Fisher’s LSD in A-E. Unpaired t-test in F).

https://doi.org/10.1371/journal.ppat.1014056.g005

Autophagy is a central process modulated by AMPK downstream activity. Consistently, and as previously reported [51], LC3-II levels significantly increased (p < 0.0001) in LN-229 C3^ΔHNRNPK cells, indicating enhanced autophagosome accumulation, likely due to an energy crisis triggering autophagy. This increase was significantly (p < 0.01) less pronounced in LN-229 C3^{ΔTFAP2C;ΔHNRNPK} cells (Figs 5B and S6C). Furthermore, in the presence of chloroquine, LC3-II levels rose almost proportionally across all conditions (Figs 5B and S6C), suggesting that the effects of HNRNPK and TFAP2C on autophagy occur primarily at the level of autophagosome formation, rather than autophagosome-lysosome fusion and degradation. Together with AMPK, the mechanistic target of rapamycin (mTOR) is a key regulator of cell anabolic and catabolic processes that senses and integrates upstream inputs related to cellular oxygen, nutrients, and energy levels [53]. mTOR and AMPK control autophagy by regulating the phosphorylation of several downstream targets [52–54]. Under nutrient sufficiency, mTOR phosphorylates Ulk1 at Ser758 (pUlk1), preventing its interaction with AMPK and inhibiting autophagosome formation [54]. To test if mTOR directly modulated the observed autophagy alterations, we measured the phosphorylation of Ulk1. As anticipated, Ulk1 phosphorylation was reduced by HNRNPK ablation, homeostatically restored by concomitant ΔTFAP2C;ΔHNRNPK double ablation, and increased by single TFAP2C removal (S6D Fig).

These results suggest that mTOR activity is central to the functional interaction between HNRNPK and TFAP2C. Interestingly, RNA and protein levels of mTOR were downregulated in LN-229 C3^ΔHNRNPK cells but were partially rebalanced by the ΔTFAP2C;ΔHNRNPK double deletion (Figs 4C and 5C and 5D). RNA levels of the mTOR-associated proteins Rptor, mLST8, and Pras40 (AKT1S1), which assemble with mTOR in the mTORC1 complex, followed the same trend (Fig 4C). 4EBP1 and S6 are two downstream targets of mTORC1 kinase activity. Deletion of HNRNPK diminished the highest phosphorylated form of 4EBP1 (high p4EBP1, marked with an asterisk), mimicking the effect observed in starved cells (Starv.). This high p4EBP1 band was preserved in both LN-229 C3^ΔTFAP2C and LN-229 C3^{ΔTFAP2C;ΔHNRNPK} cells (Fig 5E). Interestingly, total expression of 4EBP1 was reduced in both HNRNPK and TFAP2C single ablation and only partially rescued upon double deletion (Fig 5E). Consistent with the effects on 4EBP1 phosphorylation, the S6 phosphorylation (pS6/S6 ratio and pS6/Actin) was also reduced in LN-229 C3^ΔHNRNPK cells and restored in the ΔTFAP2C;ΔHNRNPK double-ablated cells (Figs 5E and S6E).

We conclude that HNRNPK and TFAP2C play a role in co-regulating mTORC1 and AMPK expression, signaling, and activity.

TFAP2C promotes mTORC1 activity

The above results suggest that HNRNPK ablation causes an energy crisis parallel to, or followed by, the inhibition of mTORC1 activity and a shift toward catabolism. TFAP2C deletion did not induce energetic impairment, yet it affected the mTORC1 pathway by decreasing the expression of 4EBP1. To clarify these observations, we overexpressed TFAP2C in LN-229-dCas9-VPR cells. We found that both the phosphorylation of S6 and expression levels of 4EBP1 increased upon TFAP2C upregulation (Figs 5F and S6F). However, we did not observe changes in mTOR protein expression (Fig 5F). These data specify a role for TFAP2C in promoting mTORC1-mediated cell anabolism and suggest that its overexpression might hypersensitize cells to HNRNPK ablation by depleting the already limited ATP available, thus making its deletion advantageous (Fig 5G).

TFAP2C overexpression reduces prion levels and prevents HNRNPK-induced prion propagation

Transcriptional silencing of hnRNP K causes increased generation of infectious prions (PrP^Sc) in various cell lines [26], whereas psammaplysene A (PSA), a compound that interacts with hnRNP K [55], reduces prion levels. HNRNPK knockdown and PSA treatment induce an inversely symmetric transcriptional profile with downregulation and upregulation of genes related to glycolysis and energy metabolism [26]. These results suggest that cell stress and energy homeostasis may impact prion propagation and protein aggregation [56–64].

As we observed an inverse relationship between HNRNPK ablation and TFAP2C activation in mTORC1 metabolic signaling in LN-229 cells (Figs 5 and S6), we wondered whether TFAP2C overexpression might influence prion propagation in a manner opposite to HNRNPK downregulation. Therefore, we continued our experiments using LN-229 cells, which provide a relevant model for studying prions, as glial cells can propagate prions and contribute to prion-induced neuronal loss through non-cell-autonomous mechanisms [65]. However, because human prions are highly biohazardous and replicate poorly in most human cells, we used CRISPR/Cas9 to produce an LN-229^ΔPRNP subline devoid of human PrP^C (S7A and S7B Fig). We then expressed a vector plasmid encoding the VRQ allele of the ovine prion protein (ovPRNP) to generate an isogenic ovinized line termed “HovL” (for “human ovinized LN-229”; S7A and S7B Fig). As reported for other ovinized cell models [66], HovL cells were susceptible to infection by the PG127 strain of ovine prions and capable of sustaining chronic prion propagation, as shown by proteinase K (PK)-digested western blot and by detection of PrP^Sc using the anti-PrP antibody 6D11, which selectively stains prion-infected cells after fixation and guanidinium treatment [67] (S7C-S7E Fig). Consistent with most prion-propagating cell lines [68], HovL cells did not exhibit specific growth defects, susceptibilities, or overt phenotypes beyond their ability to propagate prions.

We then acutely downregulated HNRNPK and overexpressed TFAP2C in PG127-infected HovL cells: 96 hours after inducing TFAP2C overexpression (or mCherry for control), we delivered HNRNPK siRNA (or non-targeting NT siRNA for control) and incubated HovL cells for the next 96 hours. As expected, PK-digested western blots showed a significant increase in PrP^Sc level upon HNRNPK downregulation (Figs S8A and 6A and 6B), with cell viability only minimally affected (S8B Fig). Conversely, overexpression of TFAP2C consistently reduced PrP^Sc level and prevented its accumulation following HNRNPK suppression (S8A Fig and 6A and 6B). Importantly, neither HNRNPK silencing nor TFAP2C activation perturbed ovPRNP RNA or PrP^C protein level, respectively, in PG127- infected or uninfected HovL (Figs 6C and S8C), indicating that they modulate a post-translational step of the prion life cycle. To assess the kinetics of such effects, we measured PrP^Sc level using 6D11 staining. Single-cell detection was performed by flow cytometry analysis after overexpression of TFAP2C for 2 or 7 days, followed by HNRNPK downregulation for 72 or 96 hours, respectively (Figs 6D and S8D). At 2 days + 72 hours, changes in the % 6D11⁺ cells or total 6D11⁺ fluorescence (% 6D11⁺ cells multiplied for the 6D11⁺ mean intensity) were mild or absent (Fig 6D), whereas flow cytometry at 7 days + 96 hours showed prominent shifts in both % 6D11⁺ cells and total 6D11⁺ fluorescence (Fig 6D), consistent with the results observed by PK western blots.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. A. Proteinase K (PK) digested (bottom) and undigested (top) western blots showing, respectively, PrP^Sc and total PrP after HNRNPK silencing (96 hours) and TFAP2C overexpression (192 hours) in PG127-infected HovL cells. n = 3. B. Quantification of PrP^Sc from A and S8A Fig. n = 6. C. PrP^C protein in uninfected HovL cells upon HNRNPK knockdown (96 hours) and TFP2C overexpression (192 hours). n = 3. D. Flow cytometry quantification of anti-PrP^Sc 6D11 antibody signal in PG127-infected HovL cells upon HNRNPK knockdown and TFP2C overexpression at different time points. 3 individually treated wells. Data information: PG127-infected (PG127) and Not-infectious Brain Homogenate (NBH) were used for PK digestion positive and negative control, respectively. Non-targeting siRNA (siNT) and mCherry overexpression were used as controls. n represents independent experiments. f.c.: fold change. Mean ± SEM. ns: p > 0.05, ^*: p < 0.05, ^**: p < 0.01, ^***: p < 0.001, ^****: p < 0.0001 (Two-way ANOVA Uncorrected Fisher’s LSD in B-D).

https://doi.org/10.1371/journal.ppat.1014056.g006

These data show that TFAP2C overexpression and HNRNPK downregulation bidirectionally regulate prion levels in cell culture. Their combined effect was not additive, indicating a potential convergence on shared molecular pathways ultimately influencing prion propagation.

Drug inhibition of mTOR partially mimics HNRNPK’s role in driving prion propagation

We showed above that mTORC1 might be a putative pathway at the interface between HNRNPK and TFAP2C (Figs 4C, 5, and S6B-S6F). HNRNPK silencing in PG127-infected HovL reproduced the effects of HNRNPK ablation in LN-229 C3 cells, showing reduced levels of mTOR and Rptor, but not Rictor, a scaffold protein required for the formation of the mTORC2 complex (Figs 7A and S9A). Accordingly, HNRNPK downregulation caused a consistent reduction in the phosphorylation of S6 and 4EBP1 (Fig 7B), although total 4EBP1 levels appeared increased rather than decreased as in the LN-229 C3^ΔHNRNPK cells, potentially reflecting the poor viability of the latter. Thus, we wondered whether impaired mTORC1 activity might mediate or contribute to the effects of HNRNPK on prion propagation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. A. qRT-PCR showing HNRNPK, MTOR, RPTOR, and RICTOR RNA upon HNRNPK silencing (96 hours) in PG127-infected HovL cells. n = 6. B. 4EBP1 and S6 protein phosphorylation upon HNRNPK silencing (96 hours) or treatment with 500 nM Torin-1 or Rapamycin (72 hours) in PG127-infected HovL cells. n = 2. C. PK digested (bottom) and undigested (top) western blots showing, respectively, PrP^Sc and total PrP after treatment with 500 nM Torin-1 or Rapamycin (72 hours) in PG127-infected HovL cells. n = 4. D. Flow cytometry quantification of anti-PrP^Sc 6D11 antibody signal in PG127-infected HovL cells upon HNRNPK knockdown (96 hours) and treatment with 500 nM Torin-1 or Rapamycin (72 hours). n = 3, each with 3 individually treated wells. Data information: PG127-infected (PG127) and Not-infectious Brain Homogenate (NBH) were used for PK digestion positive and negative control, respectively. Non-targeting siRNA (siNT) and DMSO were used as controls. qRT-PCR results are normalized against GAPDH expression. n represents independent experiments. f.c.: fold change. Mean ± SEM. ns: p > 0.05, ^*: p < 0.05, ^***: p < 0.001, ^****: p < 0.0001 (Multiple Unpaired t-test Holm- Šídák method in A. One-way ANOVA Dunnett’s test in C. Two-way ANOVA Šídák’s test in D).

https://doi.org/10.1371/journal.ppat.1014056.g007

To test this hypothesis, we treated HovL cells with Torin-1 and Rapamycin, two potent mTORC1/2 inhibitors with different mechanisms of action. Torin-1 inhibits both mTORC1 and mTORC2 complexes, whereas short-term administration of Rapamycin blocks only mTORC1 [53]. Additionally, Torin-1 strongly inhibits the phosphorylation of mTORC1 substrates S6 and 4EBP1, whereas Rapamycin has a less pronounced effect on 4EBP1 phosphorylation, which is often cell-type dependent [53] (Fig 7B). After three days, both drugs caused a consistent increase in PrP^Sc levels, resembling the effects of HNRNPK downregulation (Figs 7C and S9B). Consistently, siRNA-mediated silencing of RPTOR and MTOR for 96 hours also resulted in elevated PrP^Sc levels (S9C Fig).

To dissect the net contribution of mTORC1 inhibition and HNRNPK silencing on prion propagation, we treated PG127-HovL with HNRNPK siRNA for 24 hours, followed by Torin-1 or Rapamycin treatment for 72 hours. 6D11 immunostaining and flow cytometry analysis confirmed that HNRNPK silencing, Torin-1, and Rapamycin each significantly increased the total 6D11⁺ fluorescent signal and the fraction of 6D11⁺ cells (Fig 7D). Co-treatment with HNRNPK siRNA and the mTOR inhibitors did not further increase the % 6D11⁺ cells compared to the effects of the drugs alone (Fig 7D). However, the total 6D11⁺ fluorescence increased upon HNRNPK downregulation even in the presence of Torin-1 or Rapamycin (Fig 7D). Nonetheless, the relative effect of HNRNPK silencing when combined with the drugs was still smaller than that mediated by HNRNPK siRNA alone (Fig 7D). These findings suggest that HNRNPK influences prion propagation at least in part through mTORC1 signaling, although additional mechanisms may be involved.

Cell culture

The LN-229 cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco, 11965092) supplemented with 10% fetal bovine serum (FBS) (Clontech Laboratories) and 1% penicillin/streptomycin (P/S) (Gibco, 15140122). The U-251MG cell lines were grown in OptiMEM (Gibco, 31985047) supplemented with 10% FBS, 1% GlutaMax (Gibco, 35050061), 1% MEM Non-Essential Amino acids (MEM-NEAA) (Gibco, 11140050), and 1% P/S. The HEK-293T cell line for lentivirus production was cultured in DMEM supplemented with 10% FBS. For lentiviral delivery, all cells were seeded in antibiotic-free medium and placed under antibiotic selection 24 hours after transduction. All cells were maintained in culture at 37 °C in a 21% oxygen and 5% CO₂ atmosphere.

HovL cell line generation

The generation of the human ovinized LN-229 cell line, HovL, was performed as previously described for the ovinized SHSY-5Y cell line, HovS [66]. LN-229 wild-type cells were co-transfected with two plasmids encoding Cas9 (pSpCas9(BB)-2A-GFP, addgene #48138) and qgRNAs against human PRNP [32] to achieve PRNP ablation. A monoclonal line was isolated via limiting dilution, and the ablation was confirmed by PCR and sequencing. ΔPRNP cells were then transfected using Lipofectamine 3000 (Invitrogen) with an expression vector (EF1α promoter) (Sigma-Aldrich, OGS606-5U) containing the Ovis aries PRNP VRQ allele coding sequence (Genescript). The insert was cloned with the human ER localization signal and optimized for codon usage in human cells. Cells were kept under geneticin selection (400 μg/ml) (Thermo Fisher Scientific), and a stable monoclonal line was isolated via limiting dilution. Chronic PG127 prion-infected or mock-infected HovL were obtained as previously reported [66]. HovL were incubated either with PG127 prion-contaminated Brain Homogenate (PG127) or Non-infectious Brain Homogenate (NBH) and split for at least eight passages to dilute the original prion inoculum and enhance de novo PrP^Sc formation and propagation. PG127 and NBH HovL were also transduced with lentiCas9-Blast plasmid to stably express Cas9 [82]. Primers’ sequences for PCR (Fwd: 5’-GCACTCATTCATTATGCAGGAAACA-3’, Rev: 5’-AGACACCACCACTAAAAGGGC-3’) and sequencing (5’-GGACTCTGACGTTCTCCTCTTC-3’).

Immunoblotting

Cell extracts were prepared in lysis buffer (50 mM Tris–HCl, pH 8, 150 mM NaCl, 0.5% sodium deoxycholate, and 0.5% Triton-X 100) supplemented with protease inhibitors and PhosStop (Sigma-Aldrich). In case of Proteinase K (PK) (Roche AG) digestion, protease inhibitors and PhosStop were avoided. The bicinchoninic acid assay (BCA) was used to measure the total protein concentrations according to the manufacturer’s instructions (Pierce). Immunoblots were performed using standard procedures. The samples, boiled at 95 °C in NuPAGE LDS Sample Buffer 1x (Invitrogen) supplemented with 1 mM DTT (Sigma-Aldrich), were loaded onto precast gels (Invitrogen), and blotted onto PVDF membranes (Invitrogen). Proteinase K (PK) digestion was performed at a final concentration of 2.5 μg/ml for 30 minutes at 37°C. Following are the antibodies used and their relative dilutions: anti-Tfap2c 1:20000 (Abcam, ab76007), anti-hnRNP K 1:2000 (Abcam, ab70492), anti-hnRNP K 1:1000 (Abcam, ab39975), anti-Cas9 (S. pyogenes) (Santa Cruz Biotechnology, 7A9-3A3) 1:1000 (Cell Signaling Technology, 14697), anti-Puromycin 1:1000 (Kerafast, EQ0001), anti-PARP 1:1000 (Cell Signaling Technology, 9542S), anti-Cleaved Caspase 3 1:1000 (Cell Signaling Technology, 9661S), anti-Caspase 3 1:1000 (Novus Biologicals, NB100–56708), anti-GPX4 1:1000 (Abcam, ab125066), anti-mCherry 1:1000 (Abcam, ab213511), anti-LC3 1:1000 (Cell Signaling Technology, 2775), anti-Actin-HRP 1:10000 (Sigma-Aldrich, A3854), anti-Vinculin 1:2000 (Abcam, ab129002), anti-PrP POM1 300 ng/ml (for PrP^Sc detection) [83] or POM2 300 ng/ml (for total PrP or PrP^C detection) [83], anti-mTOR 1:1000 (Cell Signaling Technology, 29835S), anti-pAMPK 1:1000 (Cell Signaling Technology, 2535S), anti-AMPK 1:1000 (Cell Signaling Technology, 2532S), anti-pUlk1 1:1000 (Cell Signaling Technology, 6888T), anti-Ulk1 1:1000 (Cell Signaling Technology, 8054T), anti-4EBP1 1:1000 (Cell Signaling Technology, 9452S), anti-pS6 1:1000 (Cell Signaling Technology, 2215), anti-S6 1:1000 (Cell Signaling Technology, 2217), anti-Rabbit 1:5000 or 1:10000 (Jackson ImmunoResearch, 111.035.045), anti-Mouse 1:5000 (BIO-RAD, STAR87P), anti-Mouse 1:5000 or 1:10000 (Jakson, 115-035-003).

Lentivirus production

All the lentiviral vectors were produced as follows: HEK-293T cells were seeded at 40% confluency, and 24 hours later the target plasmid was co-transfected together with the pCMV-VSV-G (Addgene plasmid # 8454) [84] and psPAX2 plasmids (Addgene plasmid # 12260) using Lipofectamine 3000 transfection reagent (Invitrogen, ThermoFisher Scientific). After 6 hours, the medium was replaced with DMEM supplemented with 10% FBS and 1% HyClone Bovine Serum Albumin (Cytiva). 72 hours after the transfection, the supernatant was collected, centrifuged at 1500 RCF at 4°C for 5 minutes, filtered through a 0.45 μm filter (Whatman), aliquoted, and stored at -80°C. Viral titer was measured as previously reported [85]. Cells were seeded at known numbers, infected with different volumes of lentivirus, and 24 hours later selected with the relevant antibiotics. The lentiviral Titer Unit (TU) was extrapolated by measuring the fraction of viable cells with CellTiter-Glo 2.0 and GloMax Plate reader (Promega). For plasmids bearing a fluorescent probe, the viral titer was measured by flow cytometry: briefly, cells were seeded in a 24-well plate and infected with different volumes of lentivirus after 6 hours; 72 hours after the infection, the cells were harvested, and the percentage of fluorescent positive cells was acquired by flow cytometry (BD LSRFortessa, Cell Analyzer) and analyzed by FlowJo 10 (Tree Star).

Whole-genome CRISPR-Cas9 screen and analysis

The human Brunello CRISPR ablation pooled library (Addgene #73178) [33] was amplified as previously reported [85], packaged into a lentiviral vector, and titrated as described above. The library was transduced into 280 million LN-229 C3 cells with a multiplicity of infection (MOI) of 0.3 at an estimated coverage of around 1100 cells expressing each sgRNA. The screen was performed as follows. Day 0: 1 million cells/ml were seeded in a final volume of 31.25 ml per flask in 9 T-300 flasks. Cell density was defined based on the original titration of the lentiviral packaged Brunello Library. 6 hours later, the cells were transduced with the packaged library. Day 1: 24 hours after the delivery of the library, half of the culture (280 million cells) was transferred into new T-300 flasks in DMEM supplemented with 15 μg/ml blasticidin and 1 μg/ml puromycin (Thermo Fisher Scientific). The other half (280 million cells) was harvested, pelleted, and frozen at -80°C. Day 5: At this point, all the uninfected cells were depleted, and 80 million cells were re-seeded to maintain a library coverage > 1000x. Day 7: 160 million cells were seeded at a concertation of 1 million cells/ml. 6 hours later, half of the culture was transduced either with qgRNAs against the HNRNPK gene or with the non-targeting control qgRNAs (NT) at an MOI of 10. Day 8: 24 hours after the delivery of the HNRNPK or NT qgRNAs, 160 million cells were transferred into 20 T-300 flasks and maintained under selection with 15 μg/ml blasticidin, 1 μg/ml puromycin and 1.5 mg/ml geneticin (Thermo Fisher Scientific). HNRNPK-targeted cells were incubated for six more days, changing the medium every three days. Day 11: To avoid over-confluency, the cells transduced with the NT qgRNAs were split according to the library coverage. Day 14: 80 million cells from both conditions were harvested, pelleted, and frozen at -80°C.

The extraction of the genomic DNA (gDNA), the preparation, and the purification of the NGS library were performed as previously reported [85]. Briefly, the gDNA was harvested from each of the collected samples using the Zymo Research Quick-gDNA MidiPrep (Zymo Research) according to the manufacturer’s protocol. To amplify the sgRNAs for NGS, a PCR reaction was set up for 23 cycles in 96 PCR plates (Sarstedt). Each gDNA sample was processed by using a mix of eight P5 primers in combination with one unique P7 primer. For the purification of the amplified product, the PCR reaction was mixed with five volumes of DNA Binding Buffer (Zymo Research), transferred into the Zymo-Spin V column with Reservoir (Zymo Research), and centrifuged at 500 RCF for 5 minutes at room temperature. The column was washed twice with 2 ml DNA Wash Buffer (Zymo Research) by a second step of centrifugation. The column was transferred to a 2 ml collection tube and spun again at maximum speed for 1 minute to remove residual wash buffer, and finally, the purified PCR reaction was eluted with Nuclease-Free Water (Invitrogen). The PCR product was quantified from agarose gel, and the samples sequenced by Illumina Novaseq 6000 Full SP Flowcell.

The Sushi data analysis platform [86] (FGCZ, University of Zurich) or R (version 3.5.2) were utilized for data analysis and graphical visualizations. Reads quality was assessed using FastQC. Reads were aligned to the Human Brunello CRISPR ablation pooled library [33] with targeted gene symbols. Read counts were generated using the featureCounts function [87] from the Rsubread package in R. A generalized linear model applying Trimmed Means of M-values (TMM) normalization was implemented using the EdgeR package in R [88] for differential expression analysis. Clustering analysis was performed with the hclust function from the R stats package. Differential gene enrichment was performed using Edge R (version 3.5.2) by providing the log₂ fold change and false discovery rate (FDR). sgRNAs have been grouped according to their targeted gene for the generation of the final gene lists.

Essential genes identification

To identify the set of essential genes in LN-229 cells, the Cancer Dependency Map’s “2021-Q3” release [51] served as the primary resource. Genes whose “gene_effect” scored below -1 were classified as essential. The gene symbols were aligned with those referenced in the Brunello Library. The genes missing in the Brunello Library were excluded from the final list and the following analysis (S2 Table).

CRISPR ablation and activation qgRNAs

Unless otherwise specified, the effects of CRISPR ablation were always analyzed 7 days after transduction of ablation qgRNAs in LN-229 Cas9 cells, and after 10 days in U-251MG Cas9 cells. CRISPR activation was always evaluated five days after delivery of transactivating qgRNAs.

HNRNPK intron-targeting sgRNAs were designed with a customized algorithm, and the specificity and efficacy metrics were calculated using the GPP sgRNA designer [33,89], GuideScan [90], and CRISPOR [91] tools. The selected guides were then cloned into the pYJA5_G418 backbone [92], a modified version of the pYJA5 vector [32], containing the geneticin resistance marker instead of puromycin resistance. The sequence of the intron-targeting sgRNAs and their score parameters are reported in S1 Table.

The HNRNPK ablation qgRNAs and their non-targeting control (NT) were provided by Jiang-An Yin [32] and cloned into the pYJA5_G418 backbone. All the other ablation, activation, and control qgRNAs were cloned into the pYJA5, stocked in-house, and kindly provided by Jiang-An Yin [32] in the form of glycerol stocks. For double-ablation experiments, cells were always co-treated with pYJA5_G418 and pYJA5 plasmids expressing, respectively, HNRNPK or NT and TFAP2C (or other hits) or NT qgRNAs in the four possible combinations.

The non-targeting control NT qgRNAs used for CRISPR ablation and CRISPR activation are listed as Control_3 and Control_13 (for the ablation) and Control_5 (for the activation) in Jiang-An Yin et al. [32].

Cell survival analysis

For the clonogenic assay, cells were seeded in 6-well plates (3 x 105 U-251 MG cells and 1 x 105 LN-229 cells per well) and treated with qgRNAs against different genes the day after. The cells were incubated for different time points, washed twice with PBS, fixed with 4% paraformaldehyde (Thermo Fisher Scientific), stained with 0.5% crystal violet (Sigma-Aldrich) for 1 hour, washed three times with PBS, and dried before imaging.

For cell viability assay, cells were seeded in 96-well plates (1 x 104 U-251 MG cells and 3 x 103 LN-229 cells per well) and treated with drugs or qgRNAs against different genes the following day. The cells were incubated for different time points and the viability was measured using CellTiter-Glo 2.0 and GloMax Plate reader (Promega).

Translation activity measurement

Translation activity was measured by puromycin labeling as previously described [93]. Briefly, cells were pulse-chased with 10 µg/ml puromycin (Thermo Fisher Scientific) for 10 minutes and washed three times with PBS before harvesting. Proteins were immunoblotted with anti-Puromycin 1:1000 (Kerafast, EQ0001).

RNA preparation and qRT-PCR

RNA was isolated using the RNeasy Mini kit (Qiagen). RNA quality and concentration were measured with a NanoDrop spectrophotometer (Thermo Fisher Scientific). Reverse transcription was carried out with the Quantitect Reverse Transcription kit (Qiagen) as per the manufacturer’s guidelines. For each sample, 10 ng of cDNA was loaded into 384-well PCR plates (Life Systems Design) in triplicates, and the detection was conducted using SYBR green (Roche). Readout was executed with ViiA7 Real-Time PCR systems (Thermo Fisher Scientific). The ViiA7 Real-Time PCR system (Thermo Fisher Scientific) was used for the readout, and the qRT-PCR data were processed using the 2^-ΔΔCT method. Following are the primers’ sequences: HNRNPK (Fwd: 5’-TTCAGTCCCAGACAGCAGTG-3’, Rev: 5’-TCCACAGCATCAGATTCGAG-3’), TFAP2C (Fwd: 5’-GCCGTAATGAACCCCACTGA-3’, Rev: 5’-TTCTTTACACAGTTGCTGGGC-3’), GAPDH (Fwd: 5’-TGCACCACCAACTGCTTAGC-3’, Rev: 5’-GGCATGGACTGTGGTCATCAG-3’), MTOR (Fwd: 5’-AAAGAGCAGAGTGCCCGCAT-3’, Rev: 5’-TCC AGG CCA CTA ACC TGT GC-3’), ovPRNP (Fwd: 5’-AACCACCACAAAGGGCGAGA-3’, Rev: 5’-GCACATCTGCTCCACCACTC-3’), RPTOR (Fwd: 5’-CTCGCAGTGGACAGCTCGTG-3’, Rev: 5’-CACGGCGAGAATGAAAGCCG-3’), RICTOR (Fwd: 5’-TGGATCTGACCCGAGAACCT-3’, Rev: 5’-TCCTCATAGTGAAAGCCCAGTT-3’).

Immunofluorescence imaging

For immunofluorescence imaging, cells were seeded and grown in a 96-well plate or 8-well Culture Slide (Corning) and fixed in 4% paraformaldehyde for 20 minutes at room temperature. After three washes in PBS, the cells were blocked and permeabilized for 1 hour in FBS 10% supplemented with Triton X-100 0.2%. Cells were then incubated with primary antibodies diluted in the blocking/permeabilizing solution for 2 hours at room temperature. Cells were washed in PBS three times before incubation with the indicated secondary antibody for 1 hour at room temperature. After three more washes, cells were stained with DAPI diluted 1:10000 (Sigma-Aldrich) for 10 minutes at room temperature, washed again in PBS, and mounted. 8-well Culture Slides were imaged by Confocal Laser Scanning microscopy (Leica Stellaris 5 inverse), while 96-well plate images were captured by InCell Analyzer 2500HS high-throughput imaging microscope using a 20x water objective with 2x digital camera pixel binning for a total of 12 fields of view per well. Fluorescence per well was quantified with a custom-made CellProfiler pipeline and normalized to the total number of nuclei per well.

Following are the antibodies used and their relative dilutions: anti-Tfap2c 1:500 (Abcam, ab76007), anti-hnRNP K 1:500 (Abcam, ab39975), anti-Rabbit Alexa488 1:500 (Invitrogen, A-11008), anti-mouse Alexa488 1:500 (Invitrogen, A-11001), anti-Mouse Alexa647 1:500 (Invitrogen, A-21236).

RNA sequencing

RNA was isolated with the RNeasy Mini Kit (Qiagen). The Illumina Truseq Total RNA protocol (ribosomal depletion) was used for the preparation of the libraries. The quality of both the RNA and the libraries was assessed using the Agilent 4200 TapeStation System (Agilent). Libraries were pooled equimolar and sequenced on an Illumina NovaSeq 6000 sequencer (single-end 100 bp), achieving an approximate depth of 40 million reads per sample. The experiment was conducted in triplicate, however, due to low RNA quality, downstream analysis was performed only for two replicates, except for the TFAP2C;HNRNPK double-knockout condition, for which three replicates could be analyzed. Transcript alignment was carried out using the STAR package. After normalizing the library data (Transcripts Per Million, TPM), we conducted differential gene expression analysis using the DESeq2 package [94]. We identified upregulated and downregulated genes considering the log₂ fold change and Bonferroni-corrected p-value.

RNAi silencing

HovL cells were seeded at a density of 2 x 10⁵ cells/well in 6‐well plates. The next day, culture media was exchanged with OptiMEM (Gibco) without antibiotics, and siRNAs pre-mixed with RNAiMAX (0.3% final concentration) (Thermo Fisher Scientific) were delivered in a dropwise manner at a final concentration of 10 μM. Cells were incubated for 72 or 96 hours. Media containing siRNAs was aspirated, and cells were washed once with PBS and then lysed. Pools of three different siRNAs were used to downregulate HNRNPK (s6737, s6738, s6739. Thermo Fisher Scientific, #4392420).

Co-immunoprecipitation

Cell extracts were prepared in cold no-detergent conditions (PBS) or low-detergent extraction buffer (1% NP-40, 1% Glycerol, 20 mM Tris-Base, 137 mM NaCl) supplemented with protease inhibitors (Sigma-Aldrich) using mechanical disruption. BCA assay was used to measure the total protein concentrations according to the manufacturer’s instructions (Pierce). The protein sample was processed in Protein LowBind Tube (Eppendorf) during each step. For each condition, 1 or 3 mg of protein was incubated for 1 hour at 4°C with 10 μl pre-washed dynabeads (Invitrogen) to clear the lysate from unspecific interactions. The cleared lysate was then incubated overnight at 4°C with 50 μl antibody-conjugated dynabeads and the sample was kept mixing on a spinning wheel. For benzonase nuclease digestion, the procedure was modified according to a previously published protocol [95]. Briefly, extracted samples were supplemented with 2 mM MgCl₂ (final concentration) and pre-treated with benzonase nuclease (210 U/ml) for 1 hour at 4 °C. After centrifugation, samples were pre-cleared using dynabeads, as described above, and then incubated overnight at 4 °C with 50–100 μl of antibody-conjugated dynabeads, resulting in a final benzonase concentration of 125 U/ml. 1% agarose gel electrophoresis was used to check for DNA digestion. The day after, the flow-through was collected and the dynabeads were washed three or five times in cold PBS or extraction buffer supplemented with protease inhibitors. The samples, boiled at 95 °C in LDS 2x (Invitrogen) supplemented with 2 mM DTT (Sigma-Aldrich), were loaded onto 10% or 12% gels (Invitrogen), and blotted onto a PVDF membrane (Invitrogen). Following are the antibodies used and the relative dilutions. For immunoprecipitation: anti-Tfap2c 1:30 ~ 4–5 μg (Abcam, ab218107), anti-hnRNP K 8 μg (Abcam, ab39975), Normal Rabbit IgG (Merck, 12–370) diluted to match the concentration of anti-Tfap2c (Abcam, ab218107), Normal Mouse IgG 8 μg (Merck, 12–371); for immunoblotting: anti-hnRNP K 1:1000 (Abcam, ab39975), anti-hnRNP K 1:2000 (Abcam, ab70492), anti-Tfap2c 1:5000–1:20000 (Abcam, ab76007), anti-Tfap2c HRP 1:250 (Santa Cruz Biotechnology, sc-12762 HRP), anti-Actin HRP 1:10000 (Sigma-Aldrich, A3854), VeriBlot 1:10000 (Abcam, ab131366), anti-Rabbit 1:10000 (Jackson ImmunoResearch, 111.035.045), anti-Mouse 1:10000 (Jakson, 115-035-003).

Lipid peroxidation detection

Lipid peroxidation was measured by flow cytometry using Liperfluo (Dojindo, L248-10) according to the manufacturer’s instructions. ΔTFAP2C and NT-unmodified LN-229 C3 cells were seeded in a 24-well plate, transduced with NT or HNRNPK ablation qgRNAs, and the following day put under antibiotic selection for three more days. On day 4, the cells were stained with Liperfluo, detached, acquired by flow cytometry (BD LSRFortessa, Cell Analyzer), and analyzed with FlowJo 10 (Tree Star).

ATP level quantification

Intracellular ATP was measured via CellTiter-Glo 2.0 and GloMax Plate reader (Promega) and normalized to total protein synthesis by BCA assay (Pierce). ΔTFAP2C and NT-unmodified LN-229 C3 cells were seeded in a 96-well plate, transduced with NT or HNRNPK ablation qgRNAs, and the following day put under antibiotic selection for three more days. On day 4, the cells were lysed, and protein and ATP levels were quantified as specified above.

6D11 staining for imaging and flow cytometry

The immunostaining protocol was based on a previous publication [67]. For imaging, cells were washed with PBS and fixed with 4% paraformaldehyde for 12 minutes. After PBS washing, cells were incubated with 3.5M guanidine thiocyanate (Sigma-Aldrich) for 10 minutes. Following, cells were washed five times with PBS and incubated with mouse monoclonal (6D11) anti-PrP antibody (BioLegend, 808001) diluted 1:1000 in 25% Superblock (Thermo Fisher Scientific) at room temperature for 1 hour. After washing once with PBS, cells were labeled with secondary antibody (dilution 1:1000, anti-Mouse Alexa488, Invitrogen, A-11001) along with DAPI (Sigma-Aldrich) in 25% Superblock for 1 hour at room temperature. Cells were washed once with PBS and imaged with Fluoview FV10i confocal microscope (Olympus Life Science)

For flow cytometry, HovL cells were dissociated and fixed using the Cyto-Fast Fix/Perm buffer set (BioLegend, 426803) according to the manufacturer’s instructions. Additionally, after fixation, samples were incubated in 3.5M guanidine thiocyanate (Sigma-Aldrich) for 10 minutes and then immediately washed with 1 ml 1x Cyto-Fast Perm wash solution to ensure PrP^Sc-specific binding [67]. Staining was performed using AlexaFluor-647 mouse monoclonal anti-PrP antibody (6D11) (BioLegend, 808008) diluted 1:200 in 1x Cyto-Fast Perm wash solution. Data were acquired on SP6800 spectral analyzer (Sony Biotechnology Inc), and analysis was performed using FlowJo 10 (Tree Star).

Plasmids

pCMV-VSV-G was a gift from Bob Weinberg (Addgene plasmid # 8454; http://n2t.net/addgene:8454; RRID:Addgene_8454) [84]. psPAX2 was a gift from Didier Trono (Addgene plasmid # 12260; http://n2t.net/addgene:12260; RRID:Addgene_12260). pXPR_011 was a gift from John Doench & David Root (Addgene plasmid # 59702; http://n2t.net/addgene:59702; RRID:Addgene_59702) [96]. Human Brunello CRISPR knockout pooled library was a gift from David Root and John Doench (Addgene # 73178) [33]. pXPR_120 was a gift from John Doench & David Root (Addgene plasmid # 96917; http://n2t.net/addgene:96917; RRID:Addgene_96917) [97]. lentiCas9-Blast was a gift from Feng Zhang (Addgene plasmid # 52962; http://n2t.net/addgene:52962; RRID:Addgene_52962) [82]. HA-tagged or untagged full-length HNRNPK plasmids were a gift from Ralf Bartenschlager [98]. TFORF1330 was a gift from Feng Zhang (Addgene plasmid # 143950; http://n2t.net/addgene:143950; RRID:Addgene_143950) [99]. TFORF3550 was a gift from Feng Zhang (Addgene plasmid # 145026; http://n2t.net/addgene:145026; RRID:Addgene_145026) [99]. pSpCas9(BB)-2A-GFP (PX458) was a gift from Feng Zhang (Addgene plasmid # 48138; http://n2t.net/addgene:48138; RRID:Addgene_48138) [100]. PSF-EF1-UB-NEO/G418 ASCI - EF1 ALPHA PROMOTER G418 SELECTION PLASMID (Sigma-Aldrich, OGS606-5U).

Drugs

Erastin (Merck, E7781-1MG); Baicalein (Merck, 465119–100MG); Liproxstatin-1 (MedChemExpress, HY-12726); Ferrostatin-1 (MedChemExpress, HY-100579); Staurosporin (Abcam, ab120056); Z-VAD(Ome)-FMK (Cayman Chemical, Cay14463–1); Rapamycin (Selleckchem, S1039); Torin-1 (Sigma-Aldrich, 475991–10MG).

Curve Fitting

For sigmoidal curve fitting, we used GraphPad Prism (version X, GraphPad Software). Data in Fig 2G were fitted using nonlinear regression with a least squares regression model. For Figs 3E and 3F and S5D and S5E, data fitting was performed using an asymmetric sigmoidal model with five parameters (5PL) and log-transformed X-values (log[concentration]).