COPZ1 depletion induces cell death through different mechanisms in in vitro murine models of thyroid cancer

Abstract

Objective

In the past years, the discovery of non-oncogene addiction (NOA) has expanded the cancer targets repertoire. In particular, novel strategies are aimed at counteracting the dependency of cancer cells on normal genes that are essential to sustain their stress phenotype. Of the NOA genes, discovered by our group, that are essential for human thyroid cancer cells, we extensively focused on COPZ1, a subunit of coatomer complex I. We previously reported that in in vitro human thyroid cancer models, COPZ1 depletion can induce IFN-I-mediated stimulatory effects, which culminates in immunogenic cell death. To evaluate in vivo the effects of COPZ1 depletion on immunity and inflammation, generation of syngeneic mouse models is needed. Toward this aim, in this work, we studied in vitro the dependency on COPZ1 in murine thyroid cancer cell lines.

Methods

Cell viability was assessed through crystal violet assay and a commercially available kit. qRT-PCR and western blot were used to evaluate the expression of genes and proteins of interest; commercially available kits were used for monitoring cytokine release and oxidative status.

Results

Analogous to the human counterpart, COPZ1 silencing impaired cell proliferation and induced ER stress in four cell lines. The transcription and secretory IFN-I-related program was activated in two of the four cell lines. In the other two cell lines, we detected an increase in ROS production, lipid peroxidation, and Ca²⁺ levels.

Conclusion

Activation of inflammatory effects after COPZ1 silencing is cell line-dependent.

Introduction

Thyroid cancer (TC) represents the ninth most common malignancy worldwide, whose incidence has increased in the past years, mainly due to the strengthening of diagnostic techniques (1). Papillary thyroid carcinoma (PTC) and follicular thyroid carcinoma (FTC) are well-differentiated thyroid cancer (DTC), accounting for 90% of all thyroid carcinomas; they are characterized by a good prognosis, even if about 23.5% of patients develop recurrence (2). Less common are the more aggressive subtypes, poorly differentiated thyroid carcinoma (PDTC) and anaplastic thyroid carcinoma (ATC), accounting for <5% and ∼2% of cases, respectively (3).

Surgery is the standard treatment for resectable TCs (4). Radioactive iodine ablation represents the main post-surgery treatment, especially for patients with high risk of tumor recurrence and with metastatic disease (5). Patients with advanced DTC and PDTC can be treated with antiangiogenic multikinase inhibitors (6). In Italy, dabrafenib and trametinib are the treatment options for patients with a resistant or metastatic ATC with BRAF mutation. Recently, immune checkpoint inhibitors have also been suggested as treatment; indeed, pembrolizumab showed efficacy as monotherapy in patients with DTC (7) and in combination with spartalizumab in patients with ATC (8). Collectively, the response is often modest and short-lasting, so limited or no therapeutic options are currently available for patients with aggressive and iodine-refractory TC. Thus, to improve the treatment of the most aggressive tumor forms not curable by standard therapy, a deep understanding of the mechanisms of thyroid carcinogenesis and the identification of novel therapeutic targets are needed.

The discovery of non-oncogene addiction (NOA) has paved the way for the identification of innovative therapeutically exploitable cancer ‘Achilles’ heels’ (9). NOA genes are fundamental to support the stress phenotype of cancer cells, but they are not required at the same level for normal ones (10, 11). Thus, NOA genes represent a tumor cell-specific vulnerability. By using an RNA interference-based screening, our group identified several thyroid tumor cell-specific vulnerabilities; the top ranked genes included COPZ1, a component of the heptameric coatomer protein complex I (COPI). Of note, our screening also identified other COPI subunits whose inhibition affected thyroid cancer cells’ viability. In particular, COPE represents a NOA target; on the contrary, COPA and COPB1 subunits were identified as lethal hits for both normal and cancer thyroid cells (12).

The COPI complex is involved in the trafficking of proteins between the Golgi apparatus and the endoplasmic reticulum (ER), lipid homeostasis, autophagy, and vesicular transport (13, 14). COPZ1, encoded by a gene on chromosome 12q13.13, participates to the COPI complex in alternative to COPZ2, encoded by the paralog gene on chromosome 17q21.32. Recently, COPZ1 gene mutations, leading to loss of function, have been identified in three patients with severe congenital neutropenia syndrome (15).

Shtutman et al. provided the first evidence that COPZ1 represents a canonical NOA vulnerability, since it is not mutated in tumor cells of different origin and its depletion has no effects on normal ones. Moreover, COPZ1 represents an example of paralog NOA gene: cancer cells are sensitive to COPZ1 depletion because the paralog COPZ2 is downregulated (16).

A similar scenario was found in TC. We identified COPZ1 as a TC cell-specific vulnerability, concomitantly with COPZ2 gene downregulation (12, 17). More recently, COPZ1 has been proposed as a target for breast cancer, lung adenocarcinoma, and glioblastoma (18, 19, 20). Of note, those studies were prompted by COPZ1 overexpression in tumors with respect to normal tissues; the expression of the paralog COPZ2 gene was not investigated.

We broadly studied the effects of COPZ1 silencing in in vitro and in vivo preclinical models of TC. In particular, we found that COPZ1 depletion in vitro is responsible for cell growth impairment in human TC cell models and that activates a series of inflammatory events culminating in immunogenic cell death (ICD), suggesting that COPZ1 depletion could induce an anti-tumor immune response (17, 21).

To translate in vivo, our previous results ad hoc syngeneic mouse models of TC are needed. Toward this aim, here we studied the COPZ1 dependency of murine TC cells and the molecular mechanisms involved. We found that the activation of the inflammatory program is not a hallmark of COPZ1 gene silencing, but it is cell line-dependent. Indeed, two cell lines recapitulate what we observed for the human counterpart after COPZ1 silencing, thus being suitable for the future generation of an immunocompetent mouse model of TC; in the other two TC cell lines, we provide evidence that COPZ1 depletion triggers cell death mechanisms not previously described in COPZ1-depleted tumor cells.

Materials and Methods

Cell lines

Murine TC cell lines are not commercially available; they were provided by the scientists who established them from genetically modified mice. In particular, T3531L and T4888M murine TC cell lines were obtained from Dr A Di Cristofano (Montefiore Einstein Comprehensive Cancer Center, New York, USA); They are derived from ATC tumors developed by PTEN^−/−/KRAS^G12D and PTEN^−/−/p53^−/− 129Sv mice, respectively (22, 23, 24). 3610R, 3743, and 3868 murine TC cell lines were obtained from Dr David G McFadden (Comprehensive Cancer Center, Dallas, USA); 3610R and 3743 are derived from ATC tumors developed by BRaf^V600E/WT/PTEN^−/− and BRaf^V600E/WT/p53^−/− B6129SF1/J mice, respectively; 3868 was derived from a PTC tumor developed by BRaf^V600E/WT/p53^−/− B6129SF1/J mice (25). All the murine cells, including mouse fibroblasts NIH/3T3, were maintained in DMEM supplemented with 10% (v/v) heat-inhactivated fetal bovine serum (FBS) at 37°C and 5% CO₂ and routinely tested for mycoplasma contamination (MycoStrip Mycoplasma Detection Kit, InvivoGene, USA).

siRNA transfection

siRNA transfection was performed using 20 nM of siRNA oligos (siCOPZ1: MISSION esiRNA EMU027941, siRNA Universal Negative Control #1 SIC001, Sigma-Aldrich, USA) and the Lipofectamine RNAiMAX reagent (5 μL/mL) (Invitrogen Life Technologies, USA), according to the manufacturer’s instructions.

Crystal violet assay

Cells (1.5 × 10⁴) were seeded in 24-well plates and transfected with 20 nM siRNAs the following day. From 48 to 96 h, cells were fixed with 3.7% formaldehyde solution (v/v) at room temperature for 15 min, washed with PBS, stained for 20 min with 0.1% crystal violet (w/v), and allowed to dry. Crystal violet was then dissolved in 1% SDS (v/v) for 30 min on a rocker. Optical density at λ = 570 nm was measured using a microplate reader (TECAN Ultra, TECAN Group, Switzerland).

Apoptosis assay

Cells were plated at 2 × 10³ cells/well for T3531L and 3610R or 3 × 10³ cells/well for T4888M and 3868, in 96-well plates; 24 h later, they were transfected with siRNA in the presence of 2 μM of the fluorogenic substrate for activated caspase-3/7 (CellEvent™ Caspase-3/7 Green Detection Reagent, Invitrogen Life Technologies). Cells were imaged at 10× magnification in an IncuCyte SX5 live-cell analysis system (Sartorius, Germany) at 37°C, 5% CO₂. Images (three/well) were acquired every 6 h from 24 to 72 h post-transfection. Data at 72 h post-transfection were analyzed using IncuCyte analysis software to detect and quantify green (apoptotic) cells. Each condition was performed in triplicate.

Cell viability assay

T3531L (1 × 10³), 3610R (1 × 10³), T4888M (3 × 10³), and 3868 (3 × 10³) cells were seeded in 96-well plates and transfected with 20 nM siRNAs the following day. Cell viability was measured 96 h later using CellTiter-Glo® luminescent cell viability assay (Promega Corporation, USA) according to the manufacturer’s instructions. Luminescence was measured using a microplate reader (TECAN Ultra, TECAN Group, Switzerland).

RNA purification and real-time PCR

Cells (6 × 10⁵) were seeded in 100 mm plates and transfected with 20 nM siRNAs the following day. Total RNA samples were purified from 24 to 72 h using the NucleoSpin®RNA isolation kit (Macherey-Nagel, Germany), and the reverse transcriptase-PCR was performed using SuperScript^® III first strand kit (Invitrogen Life Technologies, USA). For real-time PCR analysis, 20 ng of retrotranscribed RNA were amplified in PCRs carried out in triplicate on an ABI PRISM 7900 using the following TaqMan gene expression assays (Applied Biosystems, USA): COPZ1 (Mm00490769_m1), COPZ2 (Mm04203911_m1), Ddit3/CHOP (Mm07294308_m1), Ddx58 (Mm01216853_m1), Mx1 (Mm00487796_m1), TLR3 (Mm01207404_m1), IRF7 (Mm00516788_m1), ISG15 (Mm01705338_s1), IFIT1 (Mm07295796_m1), Atg5 (Mm01187303_m1), Atg7 (Mm00512209_m1), HO-1 (Mm00516005_m1), TFRC (Mm00441941_m1), and Actb (Mm02619580_g1).

Western blot

Cells (6 × 10⁵) were seeded in 100 mm plates and transfected with 20 nM siRNAs the following day. Total protein samples were purified from 48 to 96 h and analyzed by western blot (WB) as previously described (26), using the following antibodies: anti-COPZ1 (catalog #sc-398081, B-12) and anti-GAPDH (catalog #sc-32233, 6C5; RRID AB_627679) from Santa Cruz Biotechnology, Inc. (USA); anti-BiP (catalog #31775S, C50B12; RRID AB_2119845) and anti-LC3B (catalog #2775S; RRID AB_915950) from Cell Signaling Technology Inc. (USA); anti-calreticulin (catalog #ab92516, EPR3924; RRID AB_10562796), anti-HMGB1 (catalog #ab18256; RRID AB_444360), anti-IL1β (catalog #ab9722; RRID AB_308765), anti p62 (catalog #ab56416, RRID AB_945626) from Abcam (UK); anti-FTH1 (catalog #STJ23718) and anti-NCOA4 (catalog #STJ91667) from St John’s Laboratory (UK); anti-pERK 1/2 (catalog #M8159), anti-ERK 1/2 (catalog #M5670), anti-actin (catalog #A2066; RRID AB_476693), and anti-vinculin (catalog #V9131, clone hVIN-1) from Sigma-Aldrich (USA).

Immunofluorescence

T4888M (2.2 × 10⁴) and 3868 (1.8 × 10⁴) cells were seeded in μ-slide 8 wells with high glass bottom (#80807, Ibidi, Germany) and transfected with 20 nM siRNAs the same day; 48 h after transfection, wells were processed as described in (27). p62 primary antibody (catalog #ab56416; RRID AB_945626) from Abcam (UK) was used at a 1:100 dilution. A goat anti-mouse IgG (H + L) cross-adsorbed Alexa Fluor-488 (catalog #A11001; RRID AB_2534069) from ThermoFisher Scientific (USA) secondary antibody was used. Nuclear staining was performed with DAPI. Fluorescence images were taken 48 h later with a microscope (Nikon Eclipse TE2000-S inverted microscope).

Enzyme-linked immunosorbent assay (ELISA)

Cells (6 × 10⁵) were seeded in 100 mm plates and transfected with 20 nM siRNAs the following day. Conditioned medium (CM) was harvested 72 h after transfection, incubating cell culture overnight in serum-free conditions. Quantification of IFNβ was performed using the Verikine™ mouse IFNβ ELISA kit (catalog # 42400-1) from PBL Assay Science (USA); other cytokines were detected using the RayBio® C-series mouse inflammation antibody array C1 (catalog # AAM-INF-1-8) from RayBiotech Life Inc. (USA), according to the manufacturer’s instructions.

IFNβ neutralizing antibody assay

T3531L and 3610R cells (7 × 10⁴) were seeded in 48-well plates and transfected with 20 nM siRNAs the following day. Six hours post-transfection, both siNT and siCOPZ1 cells were treated with IFNβ neutralizing antibody (nAb) (catalog # MAB8234, and clone # 1776D, R&D SYSTEMS, USA) 0.8 μg/mL. At 72 h post-transfection/treatment with IFNβ nAb, a crystal violet assay was performed.

ATP assay

Levels of extracellular ATP were detected in CM of cells using the ATP determination kit (#A22066) from Invitrogen Life Technologies (USA) according to the manufacturer’s instructions. Luminescence was measured using a microplate reader (TECAN Ultra, TECAN Group, Switzerland).

H₂O₂ assay

T4888M and 3868 cells (3 × 10³) were seeded in 96-well plates and transfected with 20 nM siRNAs the following day. Levels of H₂O₂ were detected using the Test ROS-Glo™ H₂O₂ Kit (#G8820) from Promega Corporation (USA) according to the manufacturer’s instructions. Luminescence was measured using a microplate reader (TECAN Ultra, TECAN Group, Switzerland).

GSH/GSSG assay

T4888M and 3868 cells (3 × 10³) were seeded in 96-well plates and transfected with 20 nM siRNAs the following day. Measurement of total glutathione and oxidized glutathione was accomplished using the luminescence-based system of GSH/GSSG-Glo^TM Assay (#V6611) by Promega Corporation (USA) according to the manufacturer’s instructions. Luminescence was measured using a microplate reader (TECAN Ultra, TECAN Group, Switzerland).

Lipid peroxidation assay

T4888M and 3868 cells (8 × 10⁵) were seeded in 100 mm plates and transfected with 20 nM siRNAs the following day. 5 × 10⁶ cells/sample were collected 72 h after transfection and solubilized in lysis buffer containing BHT, following RayBio® Lipid Peroxidation (MDA) Assay Kit procedure (#MA-MDA-2, RayBiotech, USA). Absorbance at 532 nm was read by using a microplate reader (TECAN Ultra, TECAN Group, Switzerland). MDA content was determined following the manufacturer’s instructions.

Ca²⁺ assay

T4888M and 3868 cells (1 × 10⁴) were seeded in μ-slide 8 wells with high glass bottom (#80807, Ibidi, Germany) and transfected with 20 nM siRNAs the following day. Fluo-4 AM cell permeant (#F14217, ThermoFisher, USA) of 1 μM was added directly to wells containing cultured cells and incubated for 30 min at 37°C. The fluorescence images of live cells were taken with microscope (Nikon Eclipse TE2000-S inverted microscope).

Results

COPZ2 gene is downregulated in murine TC cell lines

As reported in the ‘Introduction' section, COPZ1 was proposed as an example of paralog dependence in human tumors, being caused by the tumor-specific downregulation of the paralog gene encoding the COPZ2 isoform, which is an alternative to COPZ1 in the coatomer assembly (16). This concept was further corroborated by us, showing a frequent downregulation of COPZ2 gene in human TC specimens and a panel of TC cells (12, 17). Thus, COPZ2 gene downregulation can predict tumor cell susceptibility to COPZ1 inhibition. Based on this issue, to study COPZ1 dependency in five murine TC cell lines derived from tumors in genetically modified mice (see the section titled ‘Materials and Methods’), we first investigated the expression level of COPZ2 gene by qRT-PCR analysis. We observed reduced COPZ2 mRNA level in all the TC cell lines with respect to non-tumoral NIH/3T3 cells (Fig. 1A). The highest COPZ2 downregulation was observed for the 3868 cells, whereas the 3743 cells showed the lowest one. These results allow predicting sensitivity of the murine TC cells to COPZ1 inhibition.

Figure 1.

Susceptibility of murine TC cell lines to COPZ1 silencing. (A) Real-time PCR analysis of murine TC cells for COPZ2 gene expression. Results are presented as relative quantity (RQ) normalized for β-actin housekeeping gene expression. Data represent the mean ± SD of three independent experiments. To compare COPZ2 levels of TC cells with NIH3T3, an unpaired Student’s t-test was applied (**P < 0.01; ***P < 0.001; ****P < 0.0001). (B) Growth assay of murine TC cells 48–96 h after siRNA transfection; the values represent the optical density of crystal violet normalized to control values and are indicated as mean ± SD of three independent experiments; to compare siCOPZ1 with siNT, an unpaired Student’s t-test was applied (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

COPZ1 silencing affects proliferation and viability and induces ER stress in murine TC cell lines

The murine TC cells were transfected with siCOPZ1 or siNT. In siCOPZ1-transfected cells, a complete downregulation of COPZ1 protein was observed at all the time points analyzed (Supplementary Fig. 1A (see section on Supplementary materials given at the end of the article)). Cell growth analysis showed a modest effect in 3743 cells at 48 and 72 h after COPZ1 silencing (Fig. 1B), not further assessable at 96 h due to cell overgrowth; for our work, the 3743 cells were not further considered. In the remaining four cell lines, we observed a significant reduction in proliferation at the different time points analyzed (Fig. 1B), although never complete. This may be related to a residual COPI activity as a consequence of the residual COPZ2 expression and/or incomplete COPZ1 knockdown. Indeed, by blocking the COPI complex through silencing of the COPA subunit, which cannot be complemented by other isoforms, we observed a drastic reduction in cell growth (data not shown).

We next investigated the occurrence of cell death upon COPZ1 silencing. T3531L, T4888M, 3610R, and 3868 cells were transfected with siCOPZ1 or siNT siRNAs in the presence of the fluorescent substrate for activated caspase-3/7. Measurements of fluorescence signal, corresponding to apoptotic cells, were monitored 24–72 h post-transfection through the IncuCyte SX5 live-cell analysis system (Sartorius, Germany). In Fig. 2A and B, we reported the results obtained 72 h post-transfection. In all COPZ1-depleted cells, we observed a consistent increase in caspase-3/7 activity with respect to the siNT control. The increase was variable among the different cell lines, ranging from 3.25 folds in T4888M to 8 folds in T3531L. Moreover, CellTiter-Glo assay performed 96 h after COPZ1 silencing showed a reduction in cell viability in all the siCOPZ1-transfected cells with respect to the siNT control (Supplementary Fig. 1B).

Figure 2.

COPZ1 depletion impairs viability of murine TC cell lines. (A) Murine TC cells were transfected with siCOPZ1 or siNT siRNAs with 2 μM of the fluorogenic substrate for activated caspase-3/7; the images shown (10× magnification) were acquired 72 h after siRNA transfection with the IncuCyte analysis system and represent one of three independent experiments. (B) The graph represents the fold change value of the relative mean fluorescence signal of apoptotic cells in siCOPZ1 vs siNT and is indicated as mean ± SD of three independent experiments; to compare siCOPZ1 with siNT, an unpaired Student’s t-test was applied (*P < 0.05; **P < 0.01; ****P < 0.0001).

To investigate the occurrence of ER stress in COPZ1-depleted cells, we evaluated the presence of the ER stress markers CHOP and BiP. qRT-PCR analysis demonstrated an increase of CHOP mRNA in siCOPZ1 with respect to siNT cells (Fig. 3A). Similarly, the level of BiP protein increased in COPZ1-depleted cells, although with variable extent in the different cell lines (Fig. 3B and Supplementary Fig. 2). On the whole, these results demonstrate that, similarly to the human counterpart, COPZ1 depletion in murine TC cell lines reduces cell proliferation and viability and induces ER stress.

Figure 3.

COPZ1 depletion induces ER stress in TC cell lines. (A) Real-time PCR analysis of murine TC cells for COPZ1 and CHOP gene expression 24–72 h after siRNA transfection. Results are presented as relative quantity (RQ Log10) normalized for β-actin housekeeping gene expression and considering control siNT as baseline (=0). Data represent the mean ± SD of three independent experiments. (B) Western blot analysis of murine TC cells for COPZ1 and BiP protein expression 48–96 h after siRNA transfection; actin was used for normalization of gel loading. Images shown are one representative of three experiments.

Cell line-dependent activation of a transcription and secretory IFN-related program and DAMP release upon COPZ1 silencing

ER stress is well known to be associated with inflammatory events (28). In human TC cells, we have previously demonstrated that ER stress triggered by COPZ1 silencing is associated with type I IFN-induced inflammation, resulting in immunogenic cell death (ICD) (21). To assess if this is recapitulated in the murine counterpart, we analyzed the expression of genes associated with the IFN pathway and viral mimicry response in COPZ1-depleted murine TC cells. qRT-PCR performed 48 h after COPZ1 silencing showed upregulation of several genes involved in the IFN-I pathway (Ddx58, Mx1, TLR3, IRF7, ISG15, and IFIT1) in T3531L and 3610R siCOPZ1, with respect to control siNT (Fig. 4A). On the contrary, in T4888M and 3868 cells, COPZ1 depletion did not induce upregulation/any significant modulation of those genes (Supplementary Fig. 3A).

Figure 4.

COPZ1 depletion activates a transcriptome and secretome IFN-related program in T3531L and 3610R cell lines. (A) Real-time PCR analysis of T3531L and 3610R cells for COPZ1 and IFN-related gene (Ddx58, Mx1, TLR3, IRF7, ISG15, and IFIT1) expression 48 h after siRNA transfection. Results are presented as relative quantity (RQ Log10) normalized for β-actin housekeeping gene expression, with control siNT as baseline (=0). Data represent the mean ± SD of three independent experiments. (B) ELISA for IFNβ release in CM of T3531L and 3610R cells 72 h after siRNA transfection; the values represent the fold change of amount (pg/mL) of IFNβ and are indicated as mean ± SD of three independent experiments; to compare siCOPZ1 with siNT, an unpaired Student’s t-test was applied (**P < 0.01). (C) Growth assay of T3531L and 3610R cell lines 72 h after siRNA transfection/treatment with IFNβ nAb; the values represent the optical density of crystal violet assay normalized to siNT control values and are indicated as mean ± SD of three independent experiments. (D) ELISA cytokine array performed on CM of T3531L and 3610R 72 h after siRNA transfection; the values represent the fold change of densitometric analysis of siCOPZ1 array vs siNT array and are indicated as mean of three independent experiments. (E) Upper panel: western blot analysis of T3531L and 3610R cells for COPZ1 protein expression 48–96 h after siRNA transfection; actin was used for normalization of gel loading. Lower panel: western blot analysis of T3531L and 3610R cells for IL1β release in CM 48–96 h after siRNA transfection. Images shown are one representative of three experiments.

To investigate the occurrence of IFN pathway in depth, we analyzed the release of IFNβ in CM of COPZ1-depleted cells, 72 h post-transfection. The results of ELISA showed that T3531L and 3610R COPZ1-depleted cells release more IFNβ compared with control siNT cells (Fig. 4B). On the contrary, no differences were observed in CM of T4888M and 3868 COPZ1-depleted cells, compared with siNT control (Supplementary Fig. 3B), in accordance with qRT-PCR data.

Type I interferon cytokines, such as IFNβ, are known to carry out anticancer/cytotoxic functions (29). To investigate the contribution of IFNβ to the anti-proliferative effects of COPZ1 depletion in T3531L and 3610R cells, we neutralized IFNβ with a specific nAb after transfection, and performed a proliferation assay 72 h later. We observed an increase in proliferation of siCOPZ1 cells treated with the IFNβ nAb with respect to untreated: 3-fold for T3531L and 2.2-fold for 3610R cells. These results suggest that IFNβ can contribute to the cytotoxic effects induced by COPZ1 depletion.

We analyzed the release of other inflammatory cytokines by COPZ1-depleted T3531L and 3610R cells by employing a mouse inflammation antibody array. We observed that both COPZ1-depleted cell lines release with variable extent 13 cytokines (GM-CSF, IL1α, IL4, IL12, CXCL1, LIX, CCL2, M-CSF, CCL5, CCL1, TNFα, TNF RI, and TNF RII), of which 7 and 11 are upregulated (fold change greater than 1.4) in T3531L and 3610R COPZ1-silenced cells, respectively (Fig. 4D and Supplementary Fig. 4). Moreover, the increased release of IL1β in COPZ1-depleted cells compared with control siNT was detected by WB analysis (Fig. 4E).

Finally, we also investigated the release of specific damage-associated molecular patterns (DAMPs), namely HMGB1, calreticulin, and ATP, that are key molecules in the occurrence of immunogenic cell death, acting as ‘find me’ and ‘eat me’ signals for immune cells. Our analysis showed that in T3531L and 3610R COPZ1-depleted cells, there was no modulation in the intracellular level of calreticulin and HMGB1; on the contrary, COPZ1 silencing induced a consistent increase in their release in the CM (Fig. 5A). Similarly, the release of ATP in CM was found upregulated in siCOPZ1 cells with respect to siNT control (Fig. 5B). Different results were observed in T4888M and 3868 cells. The increase of calreticulin release in siCOPZ1 compared with siNT was modest in both cell lines; HMGB1 release decreased 72 h after COPZ1 silencing in T4888M cells, and was not detected in CM of 3868 cells (Supplementary Fig. 3C). Only ATP release was found upregulated in both COPZ1-depleted cell lines, with respect to control (Supplementary Fig. 3D).

Figure 5.

COPZ1 depletion induces DAMP release in T3531L and 3610R cell lines. (A) Upper panel: western blot analysis of T3531L and 3610R cells for COPZ1, calreticulin, and HMGB1 protein expression 48–96 h after siRNA transfection; actin was used for normalization of gel loading. Lower panel: western blot analysis of T3531L and 3610R cells for calreticulin and HMGB1 release in CM 48–96 h after siRNA transfection. Images shown are one representative of three experiments. (B) ATP assay performed on CM of T3531L and 3610R cells 48–96 h after siRNA transfection; amount (nM) of ATP is reported as mean ± SD of three independent experiments; to compare siCOPZ1 with siNT, an unpaired Student’s t-test was applied (ns = not significant; *P < 0.05; **P < 0.01).

On the whole, these results indicate that only in T3531L and 3610R cells, COPZ1 depletion triggers the same effects observed in human TC cell lines: activation of a transcription and secretory program associated with the IFN-I pathway that culminates in the release of several DAMPs.

COPZ1 depletion alters the oxidative status and causes accumulation of Ca²⁺ in T4888M and 3868 cell lines

As reported in the previous paragraph, T4888M and 3868 cells did not show any significant upregulation of IFN-related genes after COPZ1 silencing, even if in the presence of ER Stress, suggesting that other mechanisms could take place in these cells. Of note, two papers showed that COPZ1 silencing in in vitro models of glioblastoma and lung adenocarcinoma induces autophagy and ferroptosis, an iron-dependent form of regulated cell death (19, 20). In particular, COPZ1 depletion induced overexpression of the selective autophagy cargo receptor, nuclear receptor coactivator 4 (NCOA4), which binds ferritin and targets it to the autophagosome, thus promoting ferritinophagy and Fe²⁺ accumulation. We investigated the presence of autophagy and ferroptosis markers in COPZ1-depleted T4888M and 3868 cells. As shown in Fig. 6A, we detected upregulation of ATG5 (autophagy-related protein 5) and ATG7 (autophagy-related protein 7), that are essential for autophagosome biogenesis, and the ferroptosis markers HO-1 (heme oxygenase-1) and TFRC (transferrin receptor 1) that are involved in the changes of intracellular iron levels (30). Autophagy activation was also documented by the increase of LC3BII levels as a consequence of the conversion of LC3BI to LC3BII (Fig. 6B and Supplementary Fig. 5A). We analyzed p62 levels by immunofluorescence and WB, and the results revealed an intracellular accumulation of p62 (Fig. 6C and Supplementary Fig. 5B and C), indicating a blockade of autophagy. This suggests that ferroptosis, if any, could eventually be autophagy-independent.

Figure 6.

Analysis of ferroptosis and autophagy markers following COPZ1 silencing. (A) Real-time PCR analysis of T4888M and 3868 cells for COPZ1, Atg5, Atg7, HO-1, and TFRC gene expression 48 h after siRNA transfection. Results are presented as relative quantity (RQ Log10) normalized for β-actin housekeeping gene expression, with control siNT as baseline (=0). (B) Western blot analysis of T4888M and 3868 cells for COPZ1 and LC3B protein expression 48–96 h after siRNA transfection; actin was used for normalization of gel loading. Images shown are one representative of three experiments. (C) Analysis of p62 expression by immunofluorescence 48 h after siCOPZ1 transfection. p62-positive foci were counted in at least 10 cells per sample. Data are presented as mean ± SD. (D) Western blot analysis of NCOA4, FTH1, and COPZ1 expression 48–96 h after transfection; actin or GAPDH was used for normalization of gel loading. Images shown are one representative of three experiments.

As mentioned above, in ferroptosis induced by COPZ1 depletion, the intracellular Fe²⁺ accumulation is related to increased ferritinophagy promoted by NCOA4 overexpression. This is not the case in our setting: in both COPZ1-depleted cells, we observed neither an increase in NCOA4 nor a decrease in ferritin (Fig. 6D and Supplementary Fig. 6A and B). In keeping with these results, the determination of Fe²⁺ content performed with two different assays (colorimetric and fluorescent) did not show intracellular Fe²⁺ accumulation in COPZ1-depleted cells with respect to control. Finally, the expression levels of GPX4 and ACSL4 in T4888M and 3868 cells also were not modulated after COPZ1 silencing (data not shown). Thus, we excluded ferroptosis as the underlying mechanism of cell death in T4888M and 3868 cells.

Several types of cell death, including ferroptosis, can be triggered by an imbalance between ROS generation and the cell’s ability to repair or eliminate the resulting oxidative damage (31). We investigated these mechanisms concomitantly with the above analyses.

To assess the effect of COPZ1 silencing on intracellular ROS levels, we measured intracellular hydrogen peroxide (H₂O₂) concentrations in siCOPZ1- and siNT-transfected cells. In both cell lines, we observed increased H₂O₂ levels in siCOPZ1 cells relative to siNT cells. This increase was statistically significant 48 h after silencing in T4888M cells and 72 h after silencing in 3868 cells (Fig. 7A).

Figure 7.

COPZ1 depletion alters the oxidative status and calcium homeostasis in T4888M and 3868 cells. (A) Analysis of intracellular H₂O₂ levels in T4888M and 3868 cells transfected with siCOPZ1 or siNT. Relative luminescence units (RLU) are presented as mean ± SD from three independent experiments; statistical comparison between siCOPZ1 and siNT was performed using an unpaired Student’s t-test (*P < 0.05). (B) Analysis of intracellular MDA concentration in T4888M and 3868 cells 72 h after siRNA transfection. Relative luminescence units (RLU) are presented as mean ± SD from three independent experiments; statistical comparison between siCOPZ1 and siNT was performed using an unpaired Student’s t-test (ns = not significant; *P < 0.05). (C) Measurement of the GSH/GSSG ratio in T4888M and 3868 cells 72 h after transfection with siCOPZ1 or siNT. Relative luminescence units (RLU) are presented as mean ± SD from three independent experiments; statistical comparison between siCOPZ1 and siNT was performed using an unpaired Student’s t-test (ns = not significant; *P < 0.05). (D) Assessment of intracellular Ca²⁺ accumulation in T4888M and 3868 cell lines at 48–72 h after COPZ1 silencing. Live cells were stained with the fluorescent probe Fluo-4, and images were acquired using a fluorescence microscope fitted with a 40× objective. The images shown are from a single biological experiment representative of three independent biological replicates. (E) Western blot analysis of pERK 1/2, ERK 1/2, and COPZ1 expression 48–72 h after transfection; vinculin was used for normalization of gel loading. Images shown are one representative of three experiments.

ROS trigger the lipid peroxidation reaction, which can be assessed by the accumulation of its final product, malondialdehyde (MDA). In both cell lines, we observed an increasing trend of MDA levels following COPZ1 silencing, with a statistically significant increase observed in 3868 cells 72 h after silencing (Fig. 7B).

Under physiological conditions, lipid hydroperoxides (LOOH) are reduced into non-toxic lipid alcohols (LOH) by GPX4 (glutathione peroxidase 4) that uses glutathione (GSH) as a substrate and produces GSSG. Under oxidative stress conditions, GSH levels decrease, GPX4 activity is impaired, and GSSG (oxidized glutathione) increases (32). In both COPZ1-depleted cells, we detected a decrease in the GSH/GSSG ratio, 72 h after COPZ1 silencing (Fig. 7C), indicating a depletion of antioxidant capacity and an accumulation of oxidative stress.

Glutathione depletion and ROS production are markers of several types of cell death, such as oxytosis and paraptosis, which are also characterized by an increase of intracellular Ca²⁺ concentration (33, 34). We investigated the effect of COPZ1 silencing on intracellular calcium levels by using a Ca²⁺ fluorescent probe. Cells transfected with siCOPZ1 showed a marked increase of Ca²⁺ compared with siNT-transfected cells both 48 and 72 h post-transfection (Fig. 7D). These results suggest a possible role of Ca²⁺ in the reduction of proliferation and viability of T4888M and 3868 cells consequent to COPZ1 silencing. Moreover, they unveil Ca²⁺ accumulation as a novel, not previously described effect of COPZ1 depletion in tumor cells.

It has been proposed that paraptosis is triggered by the MAPK pathway (35, 36). Therefore, we also analyzed the MAPK pathway status monitoring ERK 1/2 activation. We observed that both in COPZ1-depleted T4888M and 3868 cells, ERK 1/2 is active, with a slight decrease compared with siNT control for T4888M (Fig. 7E and Supplementary Fig. 6C), thus suggesting the possible occurrence of paraptosis cell death.

Discussion

COPZ1 is emerging as a target for different tumor types. Several reports have documented that COPZ1 depletion affects cell growth in different tumor preclinical models (18, 19, 20). In human TC, we have identified COPZ1 as a non-oncogene target. According to the classic definition of NOA, in our models of TC, COPZ1 gene is expressed at the same level of normal cells. Indeed, COPZ1 silencing induces TC cells’ death while having no effects on normal ones. Moreover, our previous work showed that COPZ1 targeting in human TC cells triggers ICD. This allowed to speculate that COPZ1 targeting, in addition to a direct effect on tumor growth, may have potent immune-stimulatory effects through the establishment of a systemic inflammatory response, thus turning ‘cold’ immunosuppressive thyroid tumors into ‘hot’ inflamed tumors.

In this work, we characterized the COPZ1 dependency of a panel of five murine TC cell lines. Based on the downregulation of COPZ2 gene expression, all the cell lines were predicted to be sensitive to COPZ1 depletion. Upon COPZ1 silencing, the 3743 cell line showed a modest effect on proliferation, most likely related to a not efficient COPZ2 downregulation; in the remaining cell line proliferation was markedly reduced, concomitantly with reduction of viability and induction of ER stress. In human TC cell lines, we have previously reported that ER stress triggered by COPZ1 silencing is associated with type I IFN-induced inflammation. In murine TC cells, this event is cell line-dependent, as it was observed in two out of the four cell lines, namely T3531L and 3610R. The activation of the IFN-I program was documented by the upregulation of several key genes of the pathway and the concomitantly release of several inflammatory molecules that finally culminate in the release of the main DAMPs. Thus, COPZ1-depleted T3531L and 3610R cells recapitulated the effects observed in the human TC counterpart. Our results suggest that 3610R and T3531L COPZ1-depleted cells represent valid models for future studies investigating the occurrence of immunogenic cell death, as well as for the generation of ad hoc syngeneic immunocompetent mice to evaluate a possible anti-cancer immune response triggered by COPZ1 depletion (Fig. 8, upper panel).

Figure 8.

Different cell death molecular mechanisms induced by COPZ1 depletion in murine TC cells. Schematic representation of the effects induced in murine TC cells after COPZ1 silencing. The figure was created using Biorender.com.

In the other two cell lines lacking of IFN-I pathway activation, we observed the accumulation of LC3BII and p62 autophagy markers and a significant increase in intracellular ROS. Our results are consistent with those of Gasparian’s work (37): the authors found that in prostate cancer cells, following COPZ1 depletion, autophagy markers accumulate in lipid droplets in a ROS-dependent manner. Moreover, for T4888M and 3868, we also identified molecular mechanisms not previously described in COPZ1-depleted tumor cells, such as enhanced lipid peroxidation (as assessed by MDA accumulation) and a reduction in the GSH/GSSG ratio. These events are consistent with the activation of an oxidative stress response compatible with ferroptosis. However, contrary to the classical features of this form of cell death, we did not observe an increase in intracellular Fe²⁺, a key element thought to amplify oxidative damage during ferroptosis. This discrepancy led us to consider alternative mechanisms of oxidative cell death, such as oxytosis and paraptosis, both involving Ca²⁺ homeostasis (34, 35). Of note, our COPZ1-depleted cells displayed some features of oxytosis (glutathione decrease) and paraptosis (ER stress, ROS production, p62 accumulation, and ERK1/2 activation). The accumulation of Ca²⁺ also represents a novel, not-yet-described effect of COPZ1 depletion. Coatomer dysfunction, due to COPZ1 depletion, induces Golgi apparatus fragmentation that leads to ER stress (16, 17) and, in turn, to Ca²⁺ accumulation. Indeed, ER is the main site of Ca²⁺ storage (38), and as well documented, whatever the source, ER stress causes cytosolic Ca²⁺ accumulation through the opening of ER Ca²⁺ channels (39).

In summary, our findings suggest that COPZ1 silencing in T4888M and 3868 induces a cell death program dependent on intracellular oxidative status, potentially involving mixed mechanisms (Fig. 8, lower panel). Further studies are needed in order to identify the molecular determinants addressing one or other mechanisms.

The results of this work show that, following COPZ1 depletion, IFN-I activation, possibly leading to ICD, is cell line-dependent, even though the activation of type I IFN signaling is a common feature for COPI inhibition, also in other pathological conditions. For example, COPB1 knockdown induced type I IFN signaling activation, leading to inhibition of Chlamydia psittaci intracellular proliferation (40); pathogenic COPA variants cause chronic activation of the type I IFN signaling (41, 42, 43, 44); in two patients affected of congenital neutropenia, a COPZ1 stop codon mutation generating a truncated form of the protein was found; at the intracellular level, this was associated with induction of STING and interferon-stimulated genes (15); in intrahepatic cholangiocarcinoma, COPI targeting activates the STING–IFN-I pathway, triggering an anti-tumor T-cell response (45). However, in human COPZ1-addicted tumors, only our studies investigated the effects of COPZ1 signaling on IFN and ICD. In other human tumor types dependent on COPZ1 activity, the effects of COPZ1 inhibition on the IFN pathway and possible ICD remain to be investigated. Nevertheless, assessing the possible immunostimulatory effect of COPZ1 depletion would strengthen the validity of COPZ1 as anti-tumor target.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the work reported.

Funding

This work was supported by the Italian Ministry of Health (Grant RF2019 – 12369158 to A. Greco) and ‘Ricerca corrente’ funds.

Author contribution statement

TDM and AG conceptualized and designed the study, conducted formal analysis, supervised the study, and wrote the original draft. DV investigated and validated the study results and wrote the original draft. MM and BM supervised the study and investigated, and validated the study results. EM investigated and validated the study results. SP investigated and validated the study results and provided technical support.