Introduction
Immunotherapy, particularly immune checkpoint blockade (ICB), has transformed cancer patient care in recent years. The blockade of inhibitory immune checkpoints, such as programmed cell death 1 (PD‐1) and cytotoxic T lymphocyte‐associated protein 4 (CTLA‐4), unleashes a potent antitumor response with cytotoxic T cells (CTLs) for a growing number of patients (Hodi
et al,
2010; Larkin
et al,
2015,
2019; Schadendorf
et al,
2017; Wolchok
et al,
2017).
Cytotoxic T cells are activated when their T‐cell receptors (TCRs) encounter matching antigens presented by antigen‐presenting cells (APC) or tumor cells. This can result in the elimination of target cells by the secretion of cytotoxic molecules, death ligands, and cytokines triggering cell death signaling (Russell & Ley,
2002; Martínez‐Lostao
et al,
2015; Farhood
et al,
2019). Owing to the critical role of CTLs in cancer immunotherapy, a rapidly increasing number of immunotherapeutic studies have been launched focusing on reversing T‐cell dysfunction to improve treatment outcome (Wherry & Kurachi,
2015; Zarour,
2016; Jiang
et al,
2018; Thommen & Schumacher,
2018). However, tumor‐intrinsic mechanisms, too, often contribute to the escape of immune surveillance, commonly impairing durable responses to ICB (Spranger
et al,
2015; Gao
et al,
2016; Zaretsky
et al,
2016; Sharma
et al,
2017; Litchfield
et al,
2021; Vredevoogd
et al,
2021; Zhang
et al,
2022).
Among various resistance mechanisms, IFNγ signaling plays a crucial role in determining tumor sensitivity to T cells. For example, tumors with specific deficiencies in IFNγ signaling can be more resistant to immune checkpoint therapy (Gao
et al,
2016; Zaretsky
et al,
2016; Shin
et al,
2017; Apriamashvili
et al,
2022). Therefore, we previously set out to identify IFNγ signaling‐independent tumor determinants of T‐cell sensitivity. Specifically, we performed an unbiased genome‐wide CRISPR‐Cas9 knockout screen in IFNγ receptor‐deficient (
IFNGR1‐KO) melanoma cells under cytotoxic T‐cell attack, uncovering an important role of TRAF2 in determining tumor sensitivity to T‐cell elimination in an IFNγ‐independent tumor landscape (Vredevoogd
et al,
2019). In the present study, we reanalyzed the results of this screen and identified TSC2 as a novel regulator of tumor cell sensitivity to T‐cell killing, both
in vitro and
in vivo.
TSC1 and TSC2 are known to be crucial regulators of many biological processes by forming a complex that negatively regulates mTORC1 via the GTPase activation property of TSC2 toward RheB (Garami
et al,
2003; Tee
et al,
2003; Zhang
et al,
2003; Inoki
et al,
2003a). Their dysregulation contributes to tumor development (Adachi
et al,
2003; Jiang
et al,
2005; Menon & Manning,
2009; Xu
et al,
2009). This correlates with elevated mTORC1 signaling (Inoki
et al,
2002; Potter
et al,
2002), which in turn leads to increased cell metabolism and biosynthesis while inhibiting autophagy, ultimately resulting in enhanced cell growth (Kim
et al,
2002,
2011; Inoki
et al,
2003b,
2006; Hosokawa
et al,
2009; Düvel
et al,
2010; Valvezan
et al,
2017; He
et al,
2018; Liu & Sabatini,
2020). However, how TSC1 and TSC2 regulate tumor vulnerability to T‐cell toxicity has not yet been addressed to our knowledge. In addition to validating TSC2 as a key tumor cell determinant in the context of T‐cell susceptibility, here we investigated its mechanism of action and its clinical relevance for cancer immunotherapy.
Discussion
By re‐interrogating the hits from a genome‐wide CRISPR‐Cas9 knockout screen we performed previously for critical determinants of tumor cell sensitivity to T‐cell killing (Vredevoogd
et al,
2019), we identify here two negative mTOR regulators from the TSC complex, TSC1 and TSC2. Although these genes are established as tumor suppressors, their roles in determining tumor sensitivity to cytotoxic T‐cell challenge have not yet been described to our knowledge. We show that TSC2 plays a dominant role over TSC1 in regulating tumor T‐cell sensitivity and that this regulation is conducted, at least in part, through its control of mTORC1 and mTORC2 signaling balance; this regulation is impaired in
TSC2‐KO tumor cells upon T‐cell attack. We also demonstrate that by pharmacologically redirecting the mTOR downstream signal toward mTORC1 while inhibiting mTORC2 signaling (simultaneously sustaining phosphorylated ribosomal protein S6 and inhibiting Akt phosphorylation), tumor vulnerability to T‐cell killing can be induced (Fig
6A–D).
Cancer cells often show elevated mTORC1 signaling (Sato
et al,
2010; Gerlinger
et al,
2012; Grabiner
et al,
2014; Saxton & Sabatini,
2017), resulting in enhanced cell growth associated with accelerated protein synthesis and metabolism (Düvel
et al,
2010). However, several studies have shown that when encountering stress signals, cells downregulate mTORC1 signaling to lower their energy consumption rate and release the inhibition of autophagy, allowing for resource turnover (Ng
et al,
2011; Aramburu
et al,
2014). At the same time, they upregulate mTORC2 survival signal to inhibit cell apoptosis. Aligning with this, we found that tumor cells skew their mTOR signaling toward a mTORC2 phenotype when responding to stress induced by T‐cell challenge. This mTOR signaling regulation may allow tumor cells to balance their energy requirement and enhance apoptosis resistance to survive under unfavorable conditions.
Acting as the central regulator of both mTOR1 and mTORC2 signaling, the TSC1‐TSC2 complex is considered to be a central integrator of external stress. It is essential for triggering proper stress responses through balancing the mTOR signaling level (Aramburu
et al,
2014; Demetriades
et al,
2014; Menon
et al,
2014). TSC1‐TSC2 complex inhibits mTORC1 signaling by regulating Rheb activity and activating mTORC2 signaling through direct interaction (Huang
et al,
2008). It also plays a crucial role in the crosstalk between mTORC1 and mTORC2 signaling through PI3K/Akt feedback regulation. Thereby, Akt directly phosphorylates TSC2 to suppress the inhibitory function of TSC2 on Rheb and mTORC1, limiting TSC2's inhibition of mTORC1 signaling (Manning
et al,
2002; Potter
et al,
2002; Cai
et al,
2006; Huang & Manning,
2009). In agreement with our observations in multiple tumor cell lines,
TSC2 depletion interrupts the feedback regulation of mTORC2/Akt on mTORC1 signaling, leading to constitutively hyperactivated mTORC1 while suppressing mTORC2 signaling. This result confirms the status of TSC2 as a core regulator of the mTOR signaling balance.
TSC‐deficient cells are more vulnerable to various cell death stimuli due to the impaired autophagy function caused by constitutive mTORC1 activation, while they are highly apoptotic due to diminished Akt signaling (Ng
et al,
2011). In this study, we show that once
TSC2‐ablated tumor cells encounter cytotoxic T‐cell stress, they are less capable of downregulating mTORC1 signaling and upregulating mTORC2 signaling. As a result,
TSC2‐depleted cells continue to display a higher mTORC1/mTORC2 signaling ratio than
TSC2‐proficient tumor cells. Hyperactivated mTORC1 signaling is known to induce apoptosis owing to a constantly high metabolism rate and suppressed resource turnover from autophagy inhibition (Düvel
et al,
2010; Ng
et al,
2011). When treating
TSC2‐KO cells with LY2584702 (an S6 kinase inhibitor), we observed downregulation of mTORC1 signaling, which was associated with reduced T‐cell sensitivity. Together with the established inhibition by mTORC1 of autophagy, our data support the finding that autophagy inhibition sensitizes tumor cells to T‐cell killing (Lawson
et al,
2020). On the contrary,
TSC2‐depletion directly inhibits mTORC2, thereby releasing Akt‐inhibited apoptosis (Kennedy
et al,
1997), similar to what we observed in this study. Of note, we found Akt phosphorylation to be more heterogeneously regulated among different cancer cell lines. This may be caused by other regulators that phosphorylate Akt independently of TSC/mTORC2 signaling (Bozulic
et al,
2008). Our study supports previous findings that TSC‐null cells are extremely sensitive to multiple stress signals, such as DNA damage, ER stress, energy starvation, and apoptosis (Kang
et al,
2010; Wang
et al,
2013). Our results are also in line with the finding in TSC and Lymphangioleiomyomatosis (LAM) animal models that treatment with anti‐PD‐1 antibody or the combination of anti‐PD‐1 and anti‐CTLA4 antibodies leads to the suppression of
TSC2‐null tumor growth and induces tumor rejection (Liu
et al,
2018).
Mechanistically, we demonstrate that
TSC2‐depleted tumor cells are highly susceptible to TRAIL‐induced cell toxicity through its binding to death receptors TRAIL‐R1 and TRAIL‐R2. Specifically, we found that
TSC2‐depleted tumor cells elevate their expression of TRAIL receptors. Correspondingly, by blocking TRAIL signaling during T‐cell attack, increased T‐cell sensitivity of tumor cells upon
TSC2‐depletion could be completely rescued. Our finding indicates the existence of crosstalk between TSC2/mTOR signaling and TRAIL sensitivity, which supports previous studies that activation of the Akt survival pathway leads to TRAIL resistance in tumor cells (Chen
et al,
2001; Nesterov
et al,
2001). Interestingly, loss of
TSC1/TSC2 was recently reported to induce tumor PD‐L1 expression and increase tumor mutational burden, which was accompanied by an inflamed TME (Huang
et al,
2022). Accordingly,
TSC1/TSC2‐deficient tumors benefited from immunotherapy in murine models and NSCLC patient cohorts. Together with our finding, these studies indicate that TSC2 regulates tumor sensitivity to immune challenge, not only intrinsically by tuning tumor susceptibility to T‐cell challenge, but also extrinsically by modulating PD‐L1 expression and reshaping the TME. Both studies suggest that cancer patients with no or low TSC2 expression might benefit from immunotherapies.
Numerous mTOR inhibitors have been developed and are being tested for cancer treatment in the clinic (Chiarini
et al,
2015; Hua
et al,
2019). However, long‐lasting anti‐cancer effects are rare and patients often relapse due to various resistance mechanisms. Potential mechanisms have been suggested, including an activation of PI3K/Akt/mTORC2 survival signaling (Shi
et al,
2005; Sun
et al,
2005) and upregulation of PD‐L1 expression in cancer cells (Lastwika
et al,
2016; Deng
et al,
2019). On the contrary, mTOR inhibition is associated with immunosuppressive properties, since both mTORC1 and mTORC2 are required for proper T‐cell activation (Colombetti
et al,
2006; Zheng
et al,
2007) and trafficking (Sinclair
et al,
2008). Moreover, mTORC1 and mTORC2 have distinct effects on fate decisions during immune cell differentiation (Delgoffe
et al,
2009,
2011; Rao
et al,
2010; Chi,
2012). These findings emphasize the importance of uncoupling mTORC1 and mTORC2 signaling in any future cancer treatment.
TRAIL agonists have shown promising clinical benefit in cancer treatment (Snajdauf
et al,
2021). Combination therapies are being explored, for example with mTOR inhibitors, aiming to boost the effect of TRAIL (Snajdauf
et al,
2021). In this study, we show that induction of tumor cell death by
TSC2 depletion could be recapitulated by treatment with Conatumumab, a TRAIL agonist monoclonal antibody that is being been evaluated in several clinical trials. This opens up a potential translational value of targeting TSC2 in combination with TRAIL cancer therapy. Although the detailed mechanism of how TSC2/mTOR signaling regulates TRAIL receptor expression remains to be explored, our finding provides in principle a rationale for selecting mTOR modulators as candidates for TRAIL combination therapy. Inhibiting mTOR signaling has a considerable impact on immune cell development and function (Dumont
et al,
1990; Grolleau
et al,
2002; Mills & Jameson,
2009). Therefore theoretically, targeting TSC2 in combination with TRAIL treatment may be sufficient to bypass the undesired immunosuppressive effect of combining mTOR inhibitors with ICB therapy, which is highly dependent on immune cell functions.
Overall, our study uncovers crosstalk between TSC2 regulation and TRAIL signaling and provides a novel concept for disrupting the mTORC1/mTORC2 balance to enhance tumor susceptibility to immune challenge. Further mechanistic study will be required to fully dissect the complexity of these signaling networks. This may allow us to identify specific targets for orchestrating an optimal mTORC1/2 signaling ratio in combination with TRAIL treatment in cancer therapy, aiming to avoid potential immunosuppressive effects.
Materials and Methods
Cell lines and cell culture
Human D10 (Zimmerer
et al,
2013), SK‐MEL‐23 (CVCL_6027), SK‐MEL‐28 (CVCL_0526), SK‐MEL‐147 (CVCL_3876), A375 (CVCL_0132), BLM (CVCL_7035), LCLC‐103H (CVCL_1375), and HEK293T (CVCL_0063) cell lines were retrieved from the Peeper laboratory cell line stock. A549 (CVCL_0023) cells were obtained from Prof. dr. Wilbert Zwart. Human melanoma and lung cancer cell lines without endogenous HLA‐A*02:01 or MART‐1 expression (SK‐MEL‐28, SK‐MEL‐147, A375, BLM, A549, LCLC‐103H) were transduced with lentiviral constructs encoding both components. The B16‐F10 (CVCL_0159) cell line was obtained from ATCC and was lentivirus‐transduced to express the full‐length ovalbumin (OVA) protein. OVA‐expressing cells were selected with hygromycin (250 μg/ml, 10687010, Life Technologies). All cell lines were cultured in DMEM (GIBCO), supplied with 10% fetal bovine serum (Sigma) and 100 U/ml of Penicillin–Streptomycin (GIBCO). All cell lines were regularly tested for mycoplasma by PCR (Young
et al,
2010) and were authenticated using the STR profiling kit from Promega (B9510).
Isolation, generation, and maintenance of MART‐1 TCR CD8 T cells
MART‐1 TCR CD8 T cells were generated as previously described (Vredevoogd
et al,
2019). Briefly, primary human CD8 T cells were isolated from fresh, healthy male or female donor buffycoats (Sanquin, Amsterdam, the Netherlands), activated for 48 h in human CD8 T‐cell media (RPMI Medium (GIBCO) containing 10% human serum (H3667, Sigma‐Aldrich), 100 U/ml of Penicillin–Streptomycin, 100 U/ml IL‐2 (Proleukin, Novartis), 10 ng/ml IL‐7 (11340077, ImmunoTools), and 10 ng/ml IL‐15 (11340157, ImmunoTools)) with plate‐coated αCD3 and αCD28 antibodies (16‐0037‐85 and 16‐0289‐85, eBioscience) and spinfected with MART‐1 TCR retrovirus on Retronectin coated (TB T100B, Takara) nontissue culture treated plates. Cells were harvested and maintained in human CD8 T‐cell media 24 h after transduction. Paired untransduced T cells, which are isolated from the same donor but do not recognize MART‐1 antigen, are used as control (Ctrl T cells). One week after retroviral transduction, MART‐1 TCR expression was confirmed by flow cytometry (α‐mouse TCR β chain, 553172, BD Pharmingen), and cells were cultured in RPMI containing 10% fetal bovine serum (Sigma), 100 U/ml of Penicillin–Streptomycin (GIBCO) and 100 U/ml IL‐2 (Proleukin, Novartis).
Knockout and overexpression cell line generation
Knocking out and overexpressing genes of interest in different cell lines were done by lentiviral transduction. For gene knockouts, sgRNAs were cloned into lentiCRISPR‐v2 (#52961, Addgene) plasmid using a SAM target sgRNA cloning protocol (S. Konermann, Zhang lab, 2014). For TSC2 reconstitution, full‐length TSC2 cDNA containing a CRISPR‐Cas9‐resistant silent mutation was cloned into pCDH‐blast plasmid. Lentivirus was produced by transfecting HEK293T cells with psPAX2 (#12260, Addgene) and pMD2.G (#12259, Addgene) using polyethylenimine. The media was refreshed with OptiMEM (31985062, GIBCO) containing 2% fetal bovine serum 24 h after transfection. Supernatant was harvested 72‐h post‐transfection, filtered and stored at −80°C. Tumor cells were transduced with lentivirus, together with polybrene (8 μg/ml), and cell media was refreshed 24 h later. Cells were selected with antibiotics for at least one week. TSC1/TSC2 double knockout cells or TSC2 reconstitution in knockout cells were done sequentially by using both puromycin‐selectable and blasticidin‐selectable plasmids. TNFRSF10A /TNFRSF10B double knockout cells were purified by cell sorting after antibiotic selection.
sgRNA targeting sequences:
sg_hCtrl: 5′‐ GGTTGCTGTGACGAACGGGG ‐3′
sg_hTSC1: 5′‐ CGAGATAGACTTCCGCCACG ‐3′
sg_hTSC2‐1: 5′‐ CAGAGGGTAACGATGAACAG ‐3′
sg_hTSC2‐2: 5′‐ TCCTTGCGATGTACTCGTCG ‐3′
sg_hTSC2‐3: 5′‐ ATTGTGTCTCGCAGCTGATG ‐3′
sg_hTNFRSF10A: 5′‐ AGCCTGTAACCGGTGCACAG ‐3′
sg_hTNFRSF10B: 5′‐ AGGTGGACACAATCCCTCTG ‐3′
sg_mCtrl: 5′‐ AAAAAGTCCGCGATTACGTC ‐3′
sg_mTsc2‐1: 5′‐ TCATTCGGATGCGATTGTTG ‐3′
sg_mTsc2‐2: 5′‐ AGTTCTTGAGAGAGTAGAGC ‐3′
sg_mTsc2‐3: 5′‐ GGTCAGCAGGTCATGGACGA ‐3′.
In vitro competition assay
Parental or gene‐modified cells were stained with either the CellTrace CFSE Cell Proliferation Kit (C34554, CFSE; Thermo Scientific) or the CellTrace Violet Cell Proliferation Kit (C34557, CTV; Thermo Scientific) following the manufacturer's instructions. Stained cells were mixed at a 1:1 ratio and challenged with either MART‐1 T cells or Ctrl T cells for 3 days. For drug treatment, indicated compounds were added in the media together with T‐cell co‐culture, LY2584702 (S7704, SelleckChem), Nec‐1 s (50‐429‐70001, Sigma‐Aldrich), and Q‐VD‐Oph (S7311, SelleckChem). For TRAIL treatment competition assay, 100 ng/ml sTRAIL/Apo2L (310‐04, Peprotech) was added to the culture media. The percentage CFSE‐ and CTV‐positive cells was analyzed by flow cytometry. Sensitivity was calculated by the ratio of control cells to gene‐modified cells under MART‐1 T‐cell challenge normalized to their corresponding Ctrl T‐cell condition to exclude tumor cell‐intrinsic impact. Fold sensitization was calculated by further normalizing to the sgCtrl‐sgCtrl tumor mixing or no treatment groups, as specified for each experiment.
In vitro cytotoxicity assay
1 × 104 tumor cells were seeded per well into 96‐well culture plates (Greiner). Recombinant Human IFNα‐1b (11343594, ImmunoTools), IFNβ‐1b (11343543, ImmunoTools), IFNγ (Peprotech), TNFα (300‐01A, Peprotech), TNFβ (300‐01B, Peprotech), sFas Ligand (310‐03H, Peprotech), sTRAIL/Apo2L (310‐04, Peprotech), Tautomycin (580551, Sigma‐Aldrich), Triciribine (Akt Inhibitor V, 124012, Merckmillipore), Conatumumab (TAB‐203, Creative Biolabs Inc.) or T cells were added at indicated concentrations or ratio. Cells were incubated for 3 days before viability analysis unless specifically indicated. Drugs were washed away and cell viability was read using Cell Titer Blue Viability Assay (G8081, Promega) according to the manufacturer's instruction. For staining, plates were fixed and stained for 1 h with crystal violet solution (0.1% crystal violet (Sigma) and 50% methanol (Honeywell)). Quantification was done by dissolving remaining crystal violet in 10% acetic acid (Sigma). Absorbance of the solution was measured on an Infinite 200 Pro spectrophotometer (Tecan) at 595 nm. For Incucyte (Incucyte Zoom, Essen Bioscience) experiments, 1 × 104 tumor cells were seeded per well in 96‐well culture plates (Greiner). CD8 T cells were added in indicated ratios and a Caspase‐3/7 dye (4440, Sartorius) was added at 1:1,000 dilution. Growth of these co‐cultures was followed for 72 h.
In vivo competition assay and mouse model
D10 cells were first lentivirally transduced with sgCtrl or sgTSC2 and selected with puromycin for one week as described above. Then, cells were lentivirally transduced with eGFP (pLX304‐EGFP‐Blast) or mCherry (pLX304‐mCherry‐Blast) expression plasmids and sorted. Cells were mixed at 1:1 ratio prior to injection. 1 × 106 mixed cells per mouse were subcutaneously injected into immune‐deficient NSG‐B2m mice (n = 10, The Jackson Laboratory, Strain #:010636) with Matrigel (354230, Corning). Tumor growth was monitored three times per week. Mice were randomized 12 days after tumor injection based on tumor size and gender, and either 5 × 106 MART‐1 or Ctrl (untransduced, non‐matching) human CD8 T cells were intravenously injected into the tail vein, followed by daily 100,000 U IL‐2 (Proleukin, Novartis) intraperitoneal injection for three consecutive days. Researchers were blinded for treatment given. Tumors were harvested 8‐day post‐ACT and digested into single cell suspensions. EGFP‐ and mCherry‐positive cells were analyzed by flow cytometry. Mice without tumor outgrowth or failed to receive proper ACT were excluded.
Flow cytometry
For cell surface staining, cells were harvested and stained with fluorescent‐conjugated antibodies. For cytokine production, cells were stimulated with 20 ng/ml PMA (P1585, Sigma) and 1 μg/ml Ionomycin (I9657, Sigma) for 4 h before harvesting for analysis, and Golgiplug (555029, BD Biosciences) was added 1 h after PMA/Ionomycin was added. Surface staining was performed by staining cells in PBS containing 0.1% Bovine Serum Albumin (Sigma) and fluorescent‐conjugated antibodies for 30 min on ice. Intracellular staining was performed with Foxp3/transcription factor staining buffer set (00‐5523‐00, Life Technologies) according to the manufacturer's instructions. Annexin V staining was performed using Annexin Binding Buffer (V13246, ThermoFisher) according to the manufacturer's instructions. Samples were analyzed with Fortessa flow cytometer (BD Bioscience). Antibodies against human CD261 (307207, Biolegend), CD262 (307405, Biolegend), TRAIL (308205, Biolegend), CD69 (310914, Biolegend), HLA‐A2 (561339, BD Biosciences), IFNγ (554702, BD Biosciences), TNFα (557068, BD Biosciences), Granzyme B (560213, BD Biosciences), IL‐2 (500325, Biolegend) and Live/Dead Fixable Near‐IR Dead Cell Stain Kit (L34976, Thermo) were used.
Immunoblotting
Cells were washed with PBS, scrape‐harvested, and lysed for 30 min on ice with RIPA buffer (50 mM TRIS pH 8.0, 150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with Halt Protease and Phosphatase inhibitor cocktail (78444, Fisher Scientific) for phosphoprotein blotting. Samples were centrifugated at 17,000 g, supernatant was collected and protein concentration was measured by Bradford Protein Assay (500‐0006, Bio‐Rad). To prepare immunoblot samples, protein concentration was normalized and 4×LDS sample buffer (15484379, Fisher Scientific) containing 10% β‐Mercaptoethanol (final concentration 2.5%) was added, following by 5‐min incubation at 95°C. Samples were size‐separated on 4–12% NuPAGE Bis‐Tris polyacrylamide‐SDS gels (Invitrogen) and transferred on nitrocellulose membranes (IB301031, Invitrogen, iBlot™ Transfer Stack). Blots were blocked in 4% milk powder in 0.2% Tween/PBS (PBST) and incubated at 4°C overnight with primary antibodies. After washing by PBST, secondary antibodies were applied for 1 h at room temperature. Blots were then washed by PBST and developed with SuperSignal West Dura Extended Duration Substrate (34075, Thermo Scientific), and luminescence was captured Luminescence signal was detected by either Amersham Hyperfilm high‐performance autoradiography film or Bio‐Rad ChemiDoc imaging system with default settings. Primary antibodies against TSC1 (6935, Cell Signaling Technology), TSC2 (4308, Cell Signaling Technology), Akt (sc‐8312, Santa Cruz Biotechnology), pAktSer473 (4060, Cell Signaling Technology), S6R (2217, Cell Signaling Technology), pS6Ser240/244 (2215, Cell Signaling Technology), pS6Ser235/236 (2211, Cell Signaling Technology), Caspase 3 (9665, Cell Signaling Technology), cleaved Caspase 3 (9664, Cell Signaling Technology), Caspase 8 (4790, Cell Signaling Technology), cleaved Caspase 8 (9748, Cell Signaling Technology), RIPK1 (3493, Cell Signaling Technology), Vinculin (4650, Cell Signaling Technology), α‐Tubulin (T9026, Sigma), HSP90 (sc‐7947, Santa Cruz Biotechnology), cyclophilin B (43603, Cell Signaling Technology) were used. Horseradish peroxidase‐conjugated secondary antibodies against mouse IgG (G21040, Thermo Scientific), rabbit IgG (G21234, Invitrogen) were used.
Sample preparation for generating TSC2 immune challenge signature
D10 melanoma and A549 lung cancer cell lines expressing sgCtrl or sgTSC2 were seeded into 10 cm tissue culture dishes at 70% for 48 h. Tumor cells were challenged with CFSE (423801, Biolegend) prelabeled MART‐1 T cells, with T cell to tumor ratio causing 50% tumor cell killing, or sTRAIL/Apo2L (310‐04, Peprotech; D10: 10 ng/ml; A549: 100 ng/ml) overnight. Supernatant containing cell debris and T cells was discarded and attached cells were harvested by trypsinization. Samples were washed with PBS and stained with DAPI. Pure viable tumor cells were sorted by FACSAria Fusion Cell Sorters gating on DAPI‐, CFSE‐ populations and sent for RNA sequencing.
Whole‐genome CRISPR‐KO screen data analysis
Count data from the whole‐genome screen (Vredevoogd
et al,
2019) was reanalyzed using MAGeCK (v0.5.7) using the second best sgRNA method (Li
et al,
2014). To make this analysis more robust, sgRNAs with low read counts (< 50) were filtered from this analysis.
Data resources and bioinformatic analysis
Count data of the TCGA SKCM patient cohort were obtained using the GDC query from the TCGAbiolinks package (1.15.1) in R (4.0.2). Read count data were preprocessed and normalized using DESeq2 (1.30.0). Expression data of the pan‐cancer TCGA cohorts were obtained using query option on the cBioPortal website. Survival analysis was performed on the disease‐specific survival (DSS) data and the progression‐free interval (PFI) data in months (Liu
et al,
2018), using the top and bottom quantiles of the
TSC2 expression for grouping the samples.
For the anti‐PD‐1‐treated patient cohort (Riaz
et al,
2017), the raw counts were downloaded from NCBI's GEO (GSE91061). For a second patient cohort, containing patients treated with either anti‐PD‐1 monotherapy or combined anti‐PD‐1 and anti‐CTLA‐4 (Gide
et al,
2019), the RNA sequencing data were downloaded from the European Nucleotide Archive (ENA) under project PRJEB23709. The fastq files were mapped using STAR (2.6.0c) with default settings on two‐pass mode. The raw counts were generated using HTSeq (0.10.0). For both cohorts, the raw read count data were preprocessed and normalized using DESeq2 (1.30.0).
Z‐scores were obtained from the normalized read counts by subtracting the row means and scaling by dividing the columns by the SD.
For the data on the control and TSC2‐depleted cell lines after cytotoxic T cell or TRAIL challenge used for generating the TSC2 immune response signature (TSC2‐IRS), the fastq files were mapped to the GRCh38 human reference genome (Homo.sapiens.GRCh38.v82) using STAR (2.7.3a) with default settings on two‐pass mode. Count data were generated with HTSeq (0.12.4) and preprocessed and normalized using DESeq2 (1.30.0). The genes from the TSC2‐IRS signature were significantly differentially expressed genes (DEGs) between the TSC2‐depleted tumor cell lines versus the control tumor cell lines, and showed up in both T‐cell treatment and TRAIL treatment groups. Significant DEGs were defined by an adjusted P‐value of < 0.01 and a minimum fold change (fc) of 0.15 or maximum fc of −0.15, for genes that were either up or down in TSC2‐depleted cell lines respectively. For clinical data analysis, TSC2‐IRS expression score was generated by first calculating separately the average expression level of the up signature (consists of 44 upregulated genes) and down signature (consists of 78 downregulated genes), and then dividing the up by the down expression score.
Statistics
Sample size was estimated, and the number of samples used for each experiment is indicated. When comparing two groups, a Two‐tailed Student's
t‐test was performed for normally distributed data, or by two‐tailed Mann–Whitney test with Bonferroni correction for data that was not normally distributed. When comparing more than one group of data to one control group, one‐way ANOVA with Holm–Sidak's multiple comparisons test was performed when data are normally distributed, or Kruskal–Wallis test with Dunn's
post hoc test was used when data were not normally distributed. Tukey's
post hoc analysis was used for multiple comparisons between all groups. Data distribution normality was analyzed by Shapiro–Wilk test. All analyses were performed by Prism (Graphpad Software Inc.).
P‐value lower than 0.05 was defined as statistically significant. For
in vivo experiments, sample size estimation for experimental study design was calculated by G*Power (Faul
et al,
2007).