Setting up a murine system to research the effects of
nsGSLs on the anti tumor immune response in vivo
Bachelor thesis by Aimée SeltenAbstract
Immunotherapy has revolutionized cancer treatment over the last decade, focussing on suppressing tumor expansion by inducing a productive anti tumor immune response. However, tumor immune evasion represents one of the major obstacles in achieving an efficient immunotherapeutic strategy. A leading tumor immune escape mechanism is the alteration of Major Histocompatibility Complex class I (MHC-1) surface expression, a protein essential for activating CD8+ T cells and tumor cell clearance. A multitude of regulatory mechanisms are involved in efficient HLA-1 antigen presentation and further research into regulatory layers may reveal novel immunotherapeutic opportunities. Recently, the neolactoseriesglycosphingolipids (nsGSL) synthesis pathway has been identified to regulate MHC-1 accessibility. Signal peptide peptidase like 3 (SPPL3) was described as a protease that targets B3GNT5, an enzyme involved in the nsGSL synthesis pathway. In the absence of SPPL3, nsGSL levels increased and MHC-1 accessibility decreased, resulting in impaired T cell activation. Furthermore, anti tumor immunity was improved in glioma cells upon drug intervention of the nsGSL pathway in vitro. However, the effects of nsGSL surface composition on tumor development and metastatic characteristics have not been described before in vivo. These recent findings regarding the nsGSL pathway indicate a potential new direction of immunotherapy but necessitate further research in the nsGSL pathway and anti tumor immunity in vivo, before a targeted immunotherapeutic strategy may be achieved. To research the effects of nsGSLs in vivo we aim to set up an in vitro model using murine tumor cell lines, modified in nsGSL regulators mSPPL3, mUGCG and mB3GNT5. Here, we utilized the OVA model peptide system to achieve MHC-1 and TCR interaction. We generated SIINFEKL expressing clones with a variety of TCR activation capacity. We have produced lentivirus and retrovirus to eventually achieve an absent nsGSL profile by generating a mUGCG KO and a high nsGSL profile by knocking out mSPPL3 and overexpressing mB3GNT5. We intend to transfer these murine tumor cell lines in mice, to investigate the effects of nsGSLs on immune activation, tumor development and potential immunotherapeutic opportunities. This in vivo mouse model is necessary to research this novel regulatory layer of MHC-1 accessibility and may facilitate an immunotherapeutic strategy targeting nsGSLs.
Introduction
Cancer remains one of the leading causes of death globally, accounting for approximately 1 in 6 deaths worldwide (WHO, 2021). Individuals and health care systems involved are being pushed to their physical, emotional and financial limits. Metastatic disease in some tumors in particular, can hardly be constrained by oncologic treatments such as surgery, radiation and chemotherapy (Koppulu & Vasigala, 2018; Ganesh et al., 2019; Peters et al., 2019; Urruticoechea et al., 2010). Hence research in tumor immunology has developed extensively over the last decade, leading to redefinition of the therapeutic landscape in oncology (Debele et al., 2020; Salama et al.,2020; Makuku et al., 2021; Dimberu & Leonhardt, 2011).
In the case of impaired innate or adaptive immunity, the immune system is no longer fully capable of controlling neoplastic tumor growth and tumor development may occur (Stewart & Abrams 2008). Immunotherapy focuses on attempting to suppress tumor expansion by inducing a productive anti-tumor immune response, which obstructs tumor growth.
However tumor cells may escape this immunologic pressure, similar to the ways in which these tumor cells often circumvent classic mechanisms of tumor suppression (Hanahan & Weinberg,
2000). This selective biological pressure often generates more aggressive tumor escape variants with reduced immunogenicity (Dunn et al., 2006).
As of today, tumor immune evasion continues to be one of the major obstacles in generating effective immunotherapeutic strategies. The genomic unstable nature of tumors drives the acquisition of genetic, as well as epigenetic abnormalities, providing tumors with passage to avoid immune recognition (Hicklin et al.,1999; Park et al., 2019). Immune escape mechanisms include immune checkpoint suppression, induction of an immunosuppressive environment and decreased antigen presentation (Sharma et al., 2017). The latter of which, is often achieved by alteration of Major Histocompatibility Complex class I (MHC-1) surface expression, a glycoprotein that plays a critical role in the regulation and initiation of the adaptive immune response (Garcia Lora et al., 2003; Cornel et al., 2020). MHC-1 presents endogenous derived antigens to the T cell receptor (TCR) on CD8+ T cells. This interaction activates CD8+ T cells and triggers a cascade of signalling events leading to target cell death and cytokine release. (Neefjes et al., 2011; Unanue & Cerottini 1989).
Upon binding of the TCR to antigen presented by MHC-1, a conformational change in the intracellular CD3 protein complex is induced, which activates and phosphorylates several other proteins (Alarcon et al., 2003; He et al., 2019). The activated molecules form a scaffold, triggering a multitude of signalling pathways in T cells, including protein kinase C (PKC), mitogen-activated protein kinase (MAPK) and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) (Huse, 2009). Ultimately, this signalling cascade enables CD8+ T cells to migrate and proliferate, as well as secrete cytokines and granules, resulting in target cell death.
In turn, MHC-1 surface expression is regulated by a multitude of mechanisms. MHC-1 antigen presentation is initially established in the endoplasmic reticulum (ER), where ER chaperone proteins, beta-2 microglobulin (B2M), the peptide loading complex (PLC) and peptide transporter TAP are involved in the assembly of a stable, mature MHC-1 heterotrimer complex (Blees et al., 2017; Trowitzsch & Tampé 2020; Rock et al., 2016).Subsequently, MHC-1 is trafficked to the cell surface by the Golgi Apparatus to cognate CD8+ T cells. Due to the rigorous complexity of MHC-1 antigen presentation, a multitude of regulatory mechanisms in this pathway are insufficiently investigated and remain to be elucidated..
Recently, the neolactoseriesglycosphingolipids (nsGSL) synthesis pathway was revealed to be a novel regulatory layer in HLA-1 accessibility (Jongsma et al., 2020). By performing a genome wide haploid genetic screen on HAP1 cells, Signal peptide peptidase like 3 (SPPL3) was uncovered as a positive regulator of HLA-1 antigen presentation. SPPL3 targets Beta-1,3-N-acetylglucosaminyltransferase (B3GNT5), a glycosyltransferase essential for the synthesis of cell surface nsGSLs. Enzymatic activity of B3GNT5 enzymatic activity increases in the absence of SPPL3, leading to an increase in nsGSL surface levels, which interferes with HLA-1 accessibility towards several immune receptors as well as CD8+ T cell activation. In addition, the nsGSL surface composition is relevant in regard to cancer research, as a high nsGSL profile is negatively associated with glioma patient’s survival prognosis. Moreover, drug intervention in the nsGSL pathway in glioma cells drives improved anti tumor immunity in vitro (Jongsma et al., 2020). Thus, the nsGSL composition may represent a possible therapeutic target to achieve a more defined TCR activation and anti tumor immune response.
These promising new findings introduce nsGSL surface levels as a regulator of HLA-1 accessibility and novel immunotherapeutic opportunities in tumor cells in vitro. However, it remains unclear how nsGSLs affect tumor growth, development and metastatic characteristics in vivo. Particularly, whether drug intervention in the nsGSL pathway instigates anti tumor immunity in vivo. Therefore, this novel regulatory layer of HLA-1 accessibility necessitates further research into this pathway in vivo, before a targeted immunotherapeutic strategy for cancer patients may be achieved.
We hypothesized that elevated levels of nsGSLs, due to the absence of SPPL3, would lead to impaired MHC-1 accessibility, TCR activation and tumor clearance in vivo. In contrast to an absent nsGSL profile, which would aid MHC-1 antigen presentation. To investigate this, we aimed to set up an in vitro mouse model utilizing murine tumor cell lines, genetically modified in nsGSL regulating proteins mUGCG, mSPPL3 and mB3GNT5 to achieve varying levels of nsGSLs. Furthermore, we used OT-1 hybridoma’s and expression of the SIINFEKL model peptide in our tumor cells to determine MHC-1 and TCR interaction. Eventually, we intend to transfer these murine tumor cells in mice to assess the effects of nsGSLs on the anti tumor immune response and tumor progression in vivo.
Here, we set up a mouse model to research the effects of nsGSL composition on tumor immunosurveillance. We generated AT3 and MC38 SIINFEKL transduced clones and B16 OVA clones with capacity to activate OT-1 hybridomas, validated with a mIL-2 ELISA. Three clones of each cell line were selected, each with a different level of TCR activation. We obtained efficient lentiviral transduction with the purpose of achieving a high or absent nsGSL profile by use of a mSPPL3 and mUGCG KO, a progenitor of mB3GNT5. We displayed a substantial decrease of ganglioside GM1 surface levels in mUGCG KO cells, whereas mSPPL3 KO cells only show a slight decrease in GM1 surface levels. Furthermore, we generated retrovirus containing the mB3GNT5 gene, to eventually achieve a high nsGSL profile by overexpressing mB3GNT5 in the mSPPL3 KO cells. Ultimately, we aim to transfer the tumor cells with either a high or absent nsGSL profile in mice to assess tumor progression and T cell mediated clearance in vivo.
Materials and methods. Cell culture.
B16 OVA, MC38, AT3, B16 F10, C1498, OT1 hybridoma’s and HEK293T cells were cultured in IMDM (Lonza) supplemented with 10% fetal calf serum (Serana) and 1% antibiotics (PenStrep by Invitrogen). Phoenix-Ampho cells were cultured in similarly supplemented DMEM (Gibco). All cell lines were cultured at 37°C and 5% CO2. Hek293T, B16 F10 and B16 OVA
cells were kindly gifted by Dr. Monica Wolkers (Sanquin Research). Plasmids
We used plasmids LentiCRISPR v2-blast (Addgene), LentiCRISPR v2-neo (Addgene), LentiCRISPR v2-mCherry (Addgene) and Pmx-RFP-puro (RFP cloned into pmxs sv40 puro, Addgene). Plasmids LentiCRISPR v2-blast and LentiCRISPR v2-neo were digested with Bsmb1 (Biolabs) in the presence of NEBuffer 3.1 (Biolabs). Digested plasmids were run on a 1% agarose gel (Ultrapure agarose, Thermo Scientific) with Sybr Safe (Invitrogen) and 4X loading dye (New England Biolabs) in TAE buffer (in house produced) at 120V for 40 minutes. Plasmid DNA was isolated using QIAquick Gel Extraction Kit (QIAGEN). mUGCG and mSPPL3 gRNA was generated by annealing oligo’s (Table 1) using T4 DNA ligase buffer (New England Biolabs) in a PCR run (Thermocycler, Bio-Rad). mUGCG and mSPPL3 gRNA was ligated into LentiCRISPR v2-blast and LentiCRISPR v2-neo using T4 DNA ligase (New England Biolabs) and T4 DNA ligase buffer (New England Biolabs). Plasmids were heat-shock transformed into 5-alpha Competent E. coli bacteria (New England Biolabs) and plasmid DNA was purified using NucleoSpin Plasmid (Macherey-Nagel) according to the manufacturer’s protocol. The plasmids Pmx-SIINFEKL-mCherry, pmx-mB3GNT5-puro and LentiCRISPR v2-GFP containing mUGCG or mSPPL3 KO gRNA were previously generated and validated by our group.
Sanger Sequencing
Plasmids LentiCRISPR v2-blast and LentiCRISPR v2-neo containing mUGCG or mSPPL3 gRNA were sequenced using BigDye Terminator Kit v1.1(Applied Biosystems), sequencing buffer Cycle Sequencing Ready Reaction Mix (Applied Biosystems) and primer 5’GAGGGCCTATTTCCCATGATTC’3. Subsequently, sequencing was performed by our in house sequencing facility (Sanquin). Sequences were analyzed using SnapGene® software (Insightful Science).
Genome editing
To generate genomic knock-out cells for murine SPPL3 or UGCG, we used the CRISPR-Cas9 system. LentiCRISPR v2-blast, LentiCRISPR v2-neo and LentiCRISPR v2-GFP plasmids containing the gRNA targeting mSPPL3 or mUGCG were transfected into HEK293T cells along with packaging plasmids psPAX2, pVSVg and pAdVAntage (Promega) using polyethylenimine (PEI; polyscience) for lentivirus production. Relative transfection efficiency was measured through flow cytometry on HEK293T cells transfected with the control lentiCRISPR v2-mCherry plasmid.
The Pmx-mB3GNT5-puro and Pmx-SIINFEKL-mCherry plasmids were transfected into Phoenix-Ampho cells supported by CaCl2 (11.5 mM, in house produced) and x HBSP (50 mM HEPES, pH 7.07, 10 mM KCl, 12 mM dextrose, 280 mM NaCl, 1.5 mM Na2PO4, in house produced). Relative transfection efficiency was measured using flow cytometry on Phoenix-Ampho cells transfected with the pmx-RFP-puro plasmid. After a minimal of two days following transfection, the lentivirus as well as the retrovirus containing supernatants were harvested and filtered through a 0,2 μm filter. Lentiviral and retroviral transduction was performed in the presence of 8 ug/ml protamine sulfate (in house produced), supported by spinoculation. Transduction efficiency of LentiCRISPR v2-GFP containing mUGCG or mSPPL3 KO gRNA was measured by performing flow cytometry on transduced target cells. For the other lentiviral and retroviral transductions, relative transduction efficiency was measured by performing flow cytometry on cells transduced with lentiCRISPR v2-mCherry or pmx-RFP-puro. Murine tumor cells transduced with plasmids pmx-mB3GNT5-puro, lentiCRISPR blast or lentiCRISPR v2-neo containing mUGCG/mSPPL3 KO gRNA were selected for three days using 2-8 μg/ml puromycin (Gibco), 4-20 μg/ml blasticidin (Gibco) or 2-20 mg/ml geneticin (Thermofisher), respectively. Murine tumor cells transduced with plasmids pmx-SIINFEKL-mCherry or lentiCRISPR v2-GFP containing mUGCG/mSPPL3 KO gRNA were purified through sorting based on their expression of mCherry or GFP, respectively
FACS
To obtain mCherry positive single cell derived clones, pmx-SIINFEKL-mCherry transduced MC38 cells were single cell sorted for mCherry positive cells on the Aria lll (BD Biosciences). B16 OVA and AT3 single cell derived clones were generated by limited dilution. Murine tumor cells transduced with lentiCRISPR v2-GFP containing mUGCG or mSPPL3 KO gRNA were sorted for GFP positive cells using Aria ll or Aria lll (BD Biosciences). Staining for MHC-1 was performed by incubating cells with diluted anti-H-2Kb (APC conjugated, AF6-88.5.5.3, eBioscience). Staining for ganglioside GM1 was performed by incubating cells with diluted Cholera toxin subunit B (FITC conjugated, C1655 (Sigma) or AF594 conjugated, C34777 (Invitrogen)). Cells were incubated with staining mixture diluted in PBS for 30min on ice. After washing, cells were fixated in PBS/1% paraformaldehyde (Merc) / DAPI (1 µM, Sigma-Aldrich) and analyzed on the Fortessa, or LSR ll (BD Biosciences). Data analysis was performed using FlowJo 10.7.1 (BD Biosciences).
mIFNg stimulation and OT-1 Hybridoma co-culture
Murine tumor cell lines were either unstimulated or stimulated with 4 ng/ml mIFNγ (Peprotech) for 48h to an overnight co-culture with OT-1 hybridoma’s in a 1:1 ratio. To quantify the OT-1 hybridoma activation, a mIL-2 ELISA (Biolegend) was performed on the co-culture supernatant following the manufacturers’ protocol. The ELISA standard curve was raised to 750 pg/ml for the first curve point. For the enzyme-substrate reaction, 1-Step™ Ultra TMB-ELISA Substrate Solution (Thermo Fischer) and stop solution 0,2M H2SO4 (in house produced) were used.
Optical density was measured at 450 nm using Synergy™ HTX Multi-Mode Microplate Reader (BioTek) and data was analyzed using Gen5 3.0 (BioTek) and GraphPad Prism 8 (GraphPad Software).
Table 1. Oligo sequences used for targeting mSPPL3 or mUGCG.
Targeted gene Oligo 1 Oligo 2
mSPPL3 5´CACCGCAGAAATGTCGACACTTGAC´3 5´AAACGTCAAGTGTCGACATTTCTGC´3
Results
nsGSL expression of B16 OVA, MC38 and AT3 is altered upon genetic modification of nsGSL regulating proteins.
For our mouse model setup, we would prefer to utilize murine tumor cell lines that are susceptible for genetic modification of nsGSL pathway regulators. We aim to use murine tumor cell lines that approach the nsGSL phenotype of our previously described HAP1 model (Figure 1A) (Jongsma et al., 2020). To identify suitable tumor cell lines, we directed to perform mass spectrometry on murine tumor cell lines AT3, B16 OVA, C1498, CT26, LLC, MC38 and MEB4 to determine their natural nsGSL surface expression (Tao Zhang, LUMC). Data revealed that the nsGSL expression was either very low or absent on almost all tested murine tumor cell lines, with the exception of MC38 expressing low (5%) nsGSL surface levels (Figure 1B). Next, we made an assessment to investigate whether we could alter the nsGSL surface composition in these murine tumor cell lines when genetically modifying nsGSL regulators. Upon knocking out mSPPL3 and overexpressing mB3GNT5 in the same murine tumor cell lines, we confirmed that the nsGSL expression increased in cell lines AT3, MC38 and B16 OVA (Figures 1C-1E), approaching the nsGSL phenotype of human HAP1 cells (1A). nsGSL levels of AT3 increased from 0,5% to 16%, MC38 from 5% to 100% and B16 OVA from 0.5 to 51%. We conclude that the nsGSL phenotype is altered substantially upon genetic modification of nsGSL regulators in cell lines AT3, MC38 and B16 OVA. Therefore, we selected these three cell lines as our parental cell lines to create an absent or high nsGSL profile and measure their TCR activation capacity by use of the OVA peptide model system.
Figure 1: LC-MS/MS data demonstrating GSL expression levels of WT or genetically modified HAP1 and murine tumor cells. A-E: LC-MS/MS data demonstrating levels of Gangliosides and B3GNT5 generated nsGSLs in various cell lines. The nsGSL spectrum is indicated with a red square. The x-axis represents the retention time (minutes), the y-axis represents the relative abundance (%). A: HAP1 WT (blue), HAP1 SPPL3 KO (red) and HAP1 SPPL3/B3GNT5 double KO (black) cells. Extracted from Jongsma et al., 2020 (Immunity). B: Murine tumor cell lines AT3, B16 OVA, C1498, CT26, LLC and MEB4 display an absent nsGSL profile. MC38 displays nsGSL cell surface expression of 5%. C: AT3 WT (top) and AT3 mSPPL3 KO/mB3GNT5 overexpressed (bottom). D: B16 OVA WT (top) and B16 OVA mSPPL3 KO/mB3GNT5 overexpressed (bottom). E: MC38 WT (top) and MC38 mSPPL3 KO/mB3GNT5 overexpressed (bottom).
Generating and selecting B16 OVA, MC38 and AT3 clones by TCR activation capacity To establish an immune activation in vivo, we utilize the ovalbumin model antigen system. This system relies on presentation of the SIINFEKL peptide by MHC-1 surface protein and therefore, achieving this SIINFEKL peptide expression is the first step in modifying our murine tumor cell lines. Upon expression of the SIINFEKL peptide, cells are presumably capable of activating OT-1 hybridoma’s, which express a TCR specific for the SIINFEKL peptide. To determine the TCR activating capacity of the murine tumor cells, we co-cultured them with OT-1 hybridoma’s, which produce mouse Interleukin-2 (mIL-2) upon TCR and MHC-OT-1 interaction. The co-culture supernatant’s mIL-2 concentration, which can be quantified through a mIL-2 specific ELISA, giving a sensible indication of the TCR activation capacity of our murine tumor cells.
The first step is to establish SIINFEKL presenting B16, MC38 and AT3 polyclonal cell lines. Fortunately, we already obtained a B16 cell line with validated SIINFEKL expression (Wolkers lab, Sanquin). The expression of SIINFEKL in MC38 and AT3 was achieved by retroviral transduction of a pmx vector containing the SIINFEKL construct and a mCherry marker gene. Transduction efficiency was assessed by comparing the mCherry signal of the transduced with the untransduced cell line (Figure 2A-B). MC38 and AT3 displayed approximately 11% and 33% transduction efficiency, respectively.
Second, we aimed to generate SIINFEKL expressing single cell derived clones for each cell line by limited dilution for B16 OVA and AT3 and single cell sorting for MC38. We started with approximately 30 clones for MC38 and AT3, of which we excluded cells we suspect to be polyclonal (Figure 2C) and mCherry negative (Figure 2D). Eventually, we obtained approximately 15 B16, AT3 and MC38 clones to include in further experiments in our selection process.
Next, we set out to determine the TCR activation capacity of each clone by performing an OT-1 hybridoma co-culture. MHC-OT-1 surface expression is essential for SIINFEKL presentation and subsequent OT-1 hybridoma activation capacity of the murine tumor clones. However, the cell line B16 has been described to be low or deficient in MHC-1 surface expression (Seliger et al., 2001). Therefore, we performed a H-2kb stain on the B16 clones following mouse IFNγ (mIFNγ) stimulation, to determine their MHC-1 surface expression. As a representative of the H-2Kb staining of the B16 OVA clones, the result of the B16 OVA polyclonal cell line is displayed in Figure 2E. In accordance with literature, we observed a low or absent H-2kb staining for non-stimulated cells whereas stimulated cells did have increased H-2Kb expression. To identify whether OT-1 hybridoma activation capacity was limited by low MHC-1 surface expression, we included mIFNγ stimulation on BMHC-16 clones prior to the OT-MHC-1 hybridoma co-culture and subsequent mIL-2 ELISA. Results indicate that, despite their low MHC-1 expression, all non-stimulated clones are capable of activating OT-1 hybridoma’s (Figure 2F). Furthermore, the non-stimulated clones show a variety of TCR activation capacity. Logically, most clones display increased OT-1 hybridoma activation capacity when stimulated with mIFNγ, as their MHC-1 surface expression is increased. From these results, we confirm that a low MHC-1 surface expression in B16 clones does not cease their OT-1 hybridoma activation capacity. Since MC38 and AT3 cell lines generally display a moderate MHC-1 surface expression (Haynes N.M. et al., 2010), we did not expect that this would be an issue in our co-culture. To validate this, we performed mIFNγ stimulation prior to a H-2Kb stain on the MC38 and AT3 clones. Indeed, the H-2Kb stain revealed the general MHC-1 expression of AT3 clones was moderate, whereas MC38 clones displayed a higher level of MHC-1 expression (Figure 3 Panel B). For each cell line, there were no particular clones that were deviant in terms of MHC-1 expression with and without mIFNγ stimulation.
The SIINFEKL presentation and TCR activation capacity of AT3 and MC38 clones was further validated by performing an OT-1 hybridoma co-culture with murine tumor clones and a mIL-2 ELISA on the co-culture supernatant. AT3 clones displayed a wide variety of TCR activation capacity (Figure 2G), whereas MC38 clones presented less variation in co-culture mIL-2 concentration (Figures 2H).
For all three cell lines, we categorized low, medium and high TCR activation of the murine tumor clones by assessing the mIL-2 concentration in the co-culture supernatant. We aim to include clones in our mouse model setup, that are as similar as possible to the mother cell line.
Therefore, we performed an additional selection step based on the forward scatter average (FSC-A), indicative of cell size, and the side scatter average (SSC-A), indicative of cell granularity or complexity (REF). From each category; low, medium or high TCR activation capacity, one clone was selected by comparing the FSC-A and SSC-A histograms of the clone to the mother cell line (Figure 2, Panel C-D).
For B16, the selected clones include; no. 16 for low, no. 25 for medium and no. 10 for high TCR activation capacity. For AT3, the selected clones include; no. 10, for low, no. 18 for medium and no. 28 for high TCR activation capacity. For MC38, the selected clones include; no. 30 for low, no. 40 for medium and no. 39 for high TCR activation capacity. To give an overview of the selected murine tumor clones, we presented the data regarding mCherry signal (for MC38 and AT3), H-2Kb staining, FSC-A SSC-A similarity and co-culture mIL-2 concentration in Figure 3. In conclusion, three SIINFEKL transduced clones per cell line were generated and validated.
Figure 2: Flow cytometry and ELISA data obtained in the process of generating and selecting suitable SIINFEKL transduced murine tumor clones. A-D: Flow cytometry data demonstrating the mCherry signal of murine tumor cells. The y-axis represents the normalized count. The x axis represents the mCherry signal. WT cell line indicated in grey. A: mCherry signal of MC38 WT (grey) and MC38 pmx-mCherry-SIINFEKl transduced polyclonal cell line (red). B: mCherry signal of AT3 WT (grey) and AT3 pmx-mCherry-SIINFEKl transduced polyclonal cell line (red). C: mCherry signal of MC38 WT (grey) and MC38 Clone 13 (red). This clones was excluded for future experiments. D: mCherry signal of AT3 WT (grey) and AT3 Clone 24 (red). This clones was excluded for future experiments. E: Flow cytometry data demonstrating the H-2Kb staining (AF6-88.5.5.3) of B16 OVA polyclonal cell line. Unstained (grey), non-stimulated (blue) and mIFNγ stimulated (orange) are displayed. The y-axis represents the normalized count. F-H: ELISA data demonstrating the mIL-2 concentration in murine tumor clones 24h co-culture with OT-1 hybridoma’s in a 1:1 ratio. The y-axis represents the supernatants mIL-2 concentration in pg/ml. N=3. Mean is indicated and error bars represent the standard deviation. F: B16 clones co-culture mIL-2 concentrations. Clones were non-stimulated or mIFNγ stimulated (48h). G: AT3 clones co-culture mIL-2 concentrations. H: MC38 clones co-culture mIL-2 concentrations.
Figure 3: Flow cytometry and ELISA data from the selected low, medium and high TCR activating clone from murine tumor cell lines AT3, MC38 and B16. Cell lines and clones indicated on the left. Panel A: Flow cytometry data demonstrating a normalized histogram of the mCherry signal of the selected MC38 and AT3 clones (red) and the WT cell line (grey). Panel B: Flow cytometry data demonstrating H-2Kb staining (AF6-88.5.5.3) on murine tumor clones. Unstained negative control (grey), unstimulated clone (blue) and 48h mIFNγ stimulated clone (orange) are indicated. The y-axis represents the normalized cell count. Panel C: Flow cytometry data demonstrating normalized FSC-A histograms of the polyclonal mother cell line (grey) and the selected murine tumor clones (blue). Panel D: Flow cytometry data demonstrating normalized SSC-A histograms of the polyclonal mother cell line (grey) and the selected murine tumor clones (blue). Panel E: ELISA data demonstrating the mIL-2 concentration of a 24h co-culture supernatant including the selected murine tumor clones and OT-1 hybridoma’s in a 1:1 ratio. The y-axis represents the supernatants mIL-2 concentration in pg/ml. N=3. Mean is indicated and error bars represent the standard deviation. The selected low, medium and high TCR activating clone for each cell line is represented from left to right, respectively.
Transduction efficiency indicates efficacious lentivirus production for LentiCRISPR v2 vectors containing gRNA targeting mUGCG or mSPPL3
With the purpose of achieving an absent nsGSL profile, we aimed to knock out mUGCG and a high nsGSL profile will be obtained by knocking out mSPPL3 and subsequently overexpressing mB3GNT5. To obtain a mUGCG as well as a mSPPL3 KO, we generated vectors containing mUGCG or mSPPL3 KO gRNA in a LentiCRISPR v2-blast, LentiCRISPR v2-neo and LentiCRISPR v2-GFP vector. By using multiple plasmids, we intent to keep our options open and chose with which vector we want to continue our experiments later in the experimental setup.
First, we ligated mUGCG or mSPPL3 KO gRNAs into digested LentiCRISPR v2-blast and LentiCRISPR v2-neo vector by means of Bsmb1 restriction sites (Figure 4A). To validate the KO gRNA was successfully inserted, we performed Sanger sequencing (Figures 4B-E). Sequencing results indicated that the gRNA for the mUGCG as well as the mSPPL3 KO were successfully ligated in both the LentiCRISPR v2-blast and the LentiCRISPR v2-neo vector. For lentivirus production, LentiCRISPR v2 vectors containing a blasticidin or neomycin resistance gene and mUGCG or mSPPL3 KO gRNA were transfected into HEK293T cells. To illustrate the relative transfection efficiency HEK293T cells were concurrently transfected with a LentiCRISPR v2 empty vector containing a mCherry signal which can be measured through flow cytometry. Figure 4F shows that we obtained a transfection efficiency of the empty vector control of approximately 6%. So we conclude we accomplished virus production.
To validate virus production and efficiency, supernatant containing the empty vector control (mCherry) was transduced to achieve a relative understanding of the transduction efficiency. Since the selected MC38 and AT3 clones were mCherry positive through their SIINFEKL transduced vector, we simultaneously transduced the mCherry negative WT mother cell line with the empty vector mCherry virus to obtain a general impression of transduction efficiency. Lentiviral transduction efficiency ranged between 11% and 28% (Figure 4G), varying between cell lines and clones. Subsequent to transduction, mUGCG and mSPPL3 KO gRNA transduced clones were validated by performing blasticidin or geneticin (neomycin) selection. In addition to generating a mUGCG or mSPPL3 KO by use of LentiCRISPR v2-Blast and LentiCRISPR v2-Neo, we also utilized the LentiCRISPR v2-GFP vector. Virus production and lentiviral transduction was performed likewise to the LentiCRISPR v2-Blast and LentiCRISPR v2-Neo vectors described earlier. Transduction efficiency fluctuated between 25% and 90% between the cell lines and clones (Figure 5A). AT3 and MC38 clones displayed a relatively high transduction efficiency and B16 clones obtained a relatively low transduction efficiency (no data for MC38 clone 39 and AT3 clone 10).
A homogenous population of LentiCRISPR v2-GFP transduced cells was achieved through cell sorting based on their GFP expression. Sorted AT3 clones were approximately 100% GFP positive after one sorting round, though all B16 OVA and most MC38 clones displayed a small GFP negative population. An example for each cell line after 1 sorting round is displayed in Figure 5B. A second sorting round could eliminate the GFP negative MC38 and B16 clones, however, this additional step is outside the scope of this thesis. In conclusion, we validated virus production and transduction efficiency of vectors LentiCRISPR v2-blast, LentiCRISPR v2-neo and LentiCRISPR v2-GFP containing gRNA for mUGCG or mSPPL3.
Figure 4: Data obtained in the process of virus production and lentiviral transduction with vectors LentiCRISPR v2-blast and LentiCRISPR v2-neo containing gRNA for mUGCG or mSPPL3. A: Image demonstrating LentiCRISPR v2-blast (left) and LentiCRISPR v2-neo (right) after digestion with Bsmb1 restriction enzyme. Slot 1: 1 kb ladder. Slot 2: undigested LentiCRISPR v2 plasmid. Slot 3: digested LentiCRISPR v2- plasmid. Run performed with 1% Agarose gel + Sybr safe DNA stain for 40 min at 120V.
B-E: Sanger sequencing data demonstrating sequences of LentiCRISPR v2-blast and LentiCRISPR v2-neo vectors after ligation of mUGCG or mSPPL3 KO gRNA. B: LentiCRISPR v2-blast mUGCG KO gRNA. C: LentiCRISPR v2-blast mSPPL3 KO gRNA. D: LentiCRISPR v2-neo mUGCG KO gRNA. E: LentiCRISPR v2-neo mSPPL3 KO gRNA. F: Flow cytometry data demonstrating HEK293T cells untransfected (left) and transfected with LentiCRISPR v2-mCherry (right). The y-axis represents the FSC-A and the x-axis represents the v2-mCherry signal. Transfection efficiency is indicated with the percentage of mCherry positive cells. G: Flow cytometry data demonstrating murine tumor cells untransduced (left) and transduced with LentiCRISPR v2-mCherry (right). Transduction was performed on MC38 WT, AT3 WT and selected B16 clones, indicated left of the plots. The y-axis represents the FSC-A and the x-axis represents the mCherry signal. Transduction efficiency is indicated with the percentage of mCherry positive cells.
Figure 5: Flow cytometry data demonstrating GFP signal of LentiCRISPR v2-GFP transduced murine tumor cells.
A: Flow cytometry data demonstrating the GFP signal of murine tumor clones untransduced (grey) and transduced with LentiCRISPR v2-GFP containing gRNA for mUGCG KO (blue) and mSPPL3 KO (green). Transduction was performed on selected MC38, AT3 and B16 clones indicated left of the plots. The y-axis represents the normalized count and the x-axis represents the GFP signal. Transduction efficiency is indicated with the percentage of GFP positive cells. B: Flow cytometry data demonstrating the GFP signal of LentiCRISPR v2-GFP mUGCG KO gRNA transduced cells after one sorting round for GFP expression. GFP negative (grey) and sorted clones (purple) are displayed. AT3 clone 28 mUGCG KO (left). B16 clone 16 mUGCG KO (middle). MC38 clone 39 mUGCG KO (right).
mUGCG and mSPPL3 KO gRNA transduced populations display altered ganglioside GM1 expression.
Earlier findings describe the alteration of ganglioside GM1 surface expression in mUGCG and mSPPL3 KO clones (Jongsma et al., 2020). They confirmed HAP1 UGCG KO cells displayed a significant decrease in GM1 expression, whereas HAP1 SPPL3 KO cells only showed a slight decrease in GM1 expression.
Therefore, our first step in validating our mUGCG and mSPPL3 KO gRNA transduced cells, is performing a stain with Cholera Toxin Subunit B (CTB), which stains for GM1. We included mUGCG and mSPPL3 KO gRNA transduced AT3 cells as a representative for all three murine tumor cell lines. CTB staining indicated a minimal difference in GM1 expression for the LentiCRISPR v2-blast vectors transduced AT3 cells (Figure 6A-B). The LentiCRISPR v2-neo vectors transduced AT3 cells indicated a negative CTB staining for the mUGCG KO, whereas the mSPPL3 KO cells displayed a slight decrease in GM1 staining in comparison to the untransduced mother clone.
When analysing the LentiCRISPR v2-GFP transduced cells, the mUGCG knock out AT3 cells had a CTB staining similar to the unstained negative control, whereas the mSPPL3 KO indicated a CTB stain slightly lower than the mother clone (Figure 6C).
Overall, we conclude that the presumably mUGCG and mSPPL3 KO cells generated by LentiCRISPR v2-neo and LentiCRISPR v2-GFP vectors are altered substantially in GM1 expression, similar to UGCG KO or SPPL3 KO HAP1 cells (Jongsma et al., 2020). We believe the mUGCG and mSPPL3 KO generated with the LentiCRISPR v2-blast vector would require additional validation steps in order to include the KOs in future experiments.
The mUGCG and mSPPL3 KOs cells generated by LentiCRISPR v2-neo and LentiCRISPR v2-GFP vectors however, are most likely to reveal a successful KO following further validation.
Figure 6: Flow cytometry data demonstrating CTB stain on AT3 clones mUGCG
or mSPPL3 KO gRNA transduced.
A-C: The y-axis represents the normalized count and the x-axis represents the CTB
staining. AT3 clone 18 (left) and AT3 clone 28 (right) are displayed. Unstained mother
clone (grey), mother clone (green), mUGCG KO (red) and mSPPL3 KO (blue) are
indicated. A: LentiCRISPR v2-neo transduced AT3 clones. CTB-FITC conjugated was
used. B: LentiCRISPR v2-blast transduced AT3 clones. CTB-FITC conjugated was
used. C: LentiCRISPR v2-GFP transduced AT3 clones. CTB-AF594 conjugate was
used.
Retrovirus production reveals efficacious transduction of the mB3GNT5 gene.
In order to achieve a high nsGSL profile, we wish to overexpress mB3GNT5 in the mSPPL3 KO clones by retroviral transduction. To produce retrovirus, Phoenix-Ampho cells were transfected with a pmx-puro vector containing mB3GNT5 gene (pmx-puro-mB3GNT5) or a pmx-puro vector containing a RFP marker gene (pmx-puro-RFP) as a control vector. We performed flow cytometry on the pmx-puro-RFP transfected Phoenix-Ampho cells to determine the general transfection efficiency (Figure 7A). Approximately 32% of the transfected cells were RFP positive in comparison to the untransfected Phoenix-Ampho cells, indicating a successful transfection. To validate virus production and activity, before using it for our mSPPL3 KO clones, we transduced B16 OVA cells and MC38 WT cells. We performed flow cytometry on pmx-puro-RFP virus transduced cells, to determine the general transduction efficiency. Flow cytometry analysis showed a transduction efficiency of 33% and 81% for MC38 and B16 OVA respectively (Figures 7B-C). To validate the retroviral transduction further, we performed puromycin selection on transduced and untransduced cells. Cells transduced with the Pmx-puro-mB3GNT5 vector survived puromycin selection, in contrast to the untransduced cell lines as they presumably have no acquired puromycin resistance gene. In conclusion, we produced retrovirus containing the mB3GNT5 gene capable of efficient transduction in murine tumor cells. The next step is transducing the mSPPL3 KO cells, which is out of the scope of this thesis.
Figure 7: Flow cytometry data obtained during retrovirus production and transduction for mB3GNT5 overexpression.
A-C: The y-axis represents the normalized count and the x-axis represents the RFP signal. A: Phoenix-Ampho cells untransfected (grey) and transfected (blue) with pmx-puro-RFP. RFP+ percentage represents the relative transfection efficiency. B: B16 OVA polyclonal mother cell line untransduced (grey) and transduced (blue) with pmx-puro-RFP. RFP+ percentage represents the relative transduction efficiency. C: MC38 WT cells untransduced (grey) and transduced (blue) with pmx-puro-RFP. RFP+ percentage represents the relative transduction efficiency
Discussion
MHC-1 antigen presentation is a complex multifactorial pathway and a topic of interest with research focusing on tumor immune evasion strategies. Recently, the nsGSL synthesis pathway and proteins involved such as SPPL3 and B3GNT5 have been described to regulate in vitro MHC-1 accessibility (Jongsma et al., 2020). To investigate this further, we generated an experimental setup that potentially allows us to research the effects of nsGSLs on the anti-tumor immune response in vivo. We generated and selected SIINFEKL presenting clones for murine tumor cell lines MC38, AT3 and B16 with a variety of TCR activation capacity. Furthermore, we achieved lentiviral transduction of mUGCG or mSPPL3 KO gRNA in these clones to alter their nsGSL profile. We produced retrovirus to overexpress mB3GNT5 to achieve a high nsGSL profile in combination with the mSPPL3 KO. Taken together, this report describes an experimental setup to research a novel regulatory layer of immune regulation, nsGSLs, in vitro and may be applicable to in vivo research.
We derived single cell clones from a polyclonal SIINFEKL transduced population. Due to clonal variety, we were capable of obtaining clones with a range of TCR activation capacity. It is important to note that the use of single cell clones has their benefits and disadvantages. Clonal variety may include unwanted variation that could affect the experimental outcome. Possibly, mutations in the clones were introduced that minimize cell growth or cell viability, factors that may interfere with TCR activation capacity. Alterations in cell size and cell granularity are indicative of cell senescence, apoptosis, autophagy (Haynes M.K. et al., 2009). To minimize the effects of mutations in the clonal populations, we have selected clones by comparing the FSC-A and SSC-A to the polyclonal mother cell line. By including clones that align in cell size and granularity with the polyclonal cell line, we presumably exclude a great number of clonally induced mutations that may affect cell viability. Furthermore, by including multiple clones per cell line, we do not expect clonal variety to affect our final conclusions.
In the process of generating a mUGCG and mSPPL3 KO, we used a LentiCRISPR v2-GFP vector. AT3 KO cells were approximately 100% GFP positive after one sorting round. However, for B16 and MC38 KO cells there was a small GFP negative population of approximately 3% present following the first sorting round, despite a narrow gate. The inclusion of GFP negative cells could be explained by different reasons. For example, since we did not include a staining that enables elimination of dead cells, autofluorescence induced by cell death or cell stress may limit the enrichment of the GFP positive population (Daugherty et al., 2000; Surre et al., 2018).
Since the B16 and MC38 cells displayed a negative GFP population after one sorting round, we have to consider our experimental options and their benefits and detriments. Two options remain for the B16 and MC38 KO cells; (1) performing a second round of sorting to achieve a homogenous GFP population, resembling the AT3 cell line or (2) using the heterogenous GFP population for future experiments. Benefits of performing another round of sorting include obtaining a higher percentage of GFP positive cells, which will most likely enrich the KO population. Nonetheless, performing an additional sorting round on only a number of selected clones may introduce a bias as this results in differentially treated clones, thus possible unwanted variation. Continuing our experimental setup with the heterogenous approximately 97% GFP positive population will avoid possible experimental bias but yield a smaller KO population than a homogenous GFP positive population. As of now, we gravitate towards using B16 and MC38 KOs cells after two sorting rounds, when a homogenous GFP positive population is achieved. Since all three selected clones of the B16 and MC38 cell lines will be sorted twice, there is no experimental bias introduced between the clones because we intent to compare the clones within one cell line and not between cell lines. By using double sorted MC38 and B16 KOs, we will obtain a more enriched KO population.
An additional note to be stated is that we used a suboptimal control in performing the CTB stain on mUGCG and mSPPL3 KO cells generated by LentiCRISPR v2-neo and LentiCRISPR v2-blast vectors. The LentiCRISPR v2-neo mSPPL3 KO and even more so the mUGCG KO cells displayed a substantial decrease in GM1 expression, in contrast to the LentiCRISPR v2-blast mUGCG and mSPPL3 KO cells that showed a slight decrease in GM1 expression
compared to the mother clone. The staining was performed immediately after antibiotic selection, whereas the control cell line was not selected prior to the experiment. There is a possibility that the antibiotic environment influenced GM1 expression of these cells resulting in a differential CTB stain. GSL surface composition is highly dependent on the environmental state of the cell, as environmental factors may alter the expression of glycosyltransferases, which are involved in GSL synthesis (D’Angelo et al., 2013; Zhang et al., 2019). Furthermore, antibiotics such as tunicamycin and Brefeldin A have been described before to inhibit ganglioside synthesis (Van Echten et al., 1990; Guarnaccia et al., 1983). Blasticidin or geneticin have not been described before in publications to alter nsGSL or other GSL levels. However, since GSL levels are heavily influenced by environmental factors, we can not exclude possible interference of the antibiotic environment prior to the CTB stain on GSL composition. If the reduction in CTB staining was caused by antibiotic selection and was independent of the genetic modification, we would have most likely observed an equal shift for both mUGCG and mSPPL3 KO cells as they were both exposed to the antibiotic environment prior to staining. Since there is a difference, we believe this is caused by the genetic alteration rather than the selection. To support this further, we observed a similar shift in CTB staining in the LentiCRISPR v2-GFP transduced cells. As they were not exposed to an antibiotic environment prior to staining, we presume the antibiotics did not influence GM1 surface expression. Therefore, we believe that having a suboptimal control should not lead to rejection of our initial conclusion, which is that the presumably mUGCG KO and mSPPL3 KO cells display an altered GM1 expression.
For lentiviral transduction, we utilized vectors LentiCRISPR v2-neo, LentiCRISPR v2-blast and LentiCRISPR v2-GFP. We initially started with three vectors to spread our chances in achieving a successful KO and, since they could all be successfully transduced, we are in the luxury position to choose the vector that is best for our future experiments. CTB staining revealed a defined alteration of the GM1 composition in LentiCRISPR v2-neo and LentiCRISPR v2-GFP transduced cells, similar to the HAP1 model described previously (Jongsma et al., 2020), suggesting we achieved KOs by using these vectors. Furthermore, lentiviral transduction with the LentiCRISPR v2-GFP vectors generated a higher transduction efficiency compared to the relative transduction efficiency of the LentiCRISPR v2-neo vectors, represented by the empty vector mCherry control. Due to the progress and validation thus far on the LentiCRISPR v2-GFP vector, we aim to continue our experimental setup by using the GFP positive KOs.
By researching the effects of nsGSLs on tumor development in vivo, we aim to propose a novel immunotherapeutic opportunity that possibly includes nsGSL synthesis inhibitors. We established inhibiting the nsGSL synthesis pathway will increase MHC-1 accessibility and TCR activation, generating a more efficient anti tumor immune response (Jongsma et al., 2020). However, GSLs mediate cellular interaction and modulate signal transduction pathways as their biological function and their existence could be significant (Hakomori, 2003; Zhuo et al., 2018). Luckily, two FDA approved GSL synthesis inhibitors are registered, eliglustat (Cerdelga) and milgustat (Zavesca), which are used for the treatment of Gaucher's disease. Although this is promising, further research should gain more information whether these inhibitors could also safely be applied to treat tumors. Also, by targeting GSL surface composition as an immunotherapeutic strategy, we have to consider how this treatment affects patients’ quality of life and how this weighs up against the curative or palliative benefits of immunotherapy. We based our hypothesis on earlier findings indicating that an increase in the nsGSL profile, hinders MHC-1 and TCR interaction as possible means to evade immune surveillance in vitro in human cells (Jongsma et al., 2020). This theory is further supported by research indicating that CD8+ T cells display reduced responsive capability to glioma cells in a GSL dependent manner (Furukawa et al., 2015). However, this theory regarding the nsGSL synthesis pathway and tumor immune evasion has not been described before in an in vivo setting. Here, we showed an experimental setup to investigate the effects of nsGSLs on immunogenicity in modified murine tumor cells in vitro and eventually applicable in vivo.
The primary cause of cancer related mortality is metastasis, accounting for approximately 90% of cancer deaths (Seyfried & Huysentruyt, 2013). Due to the limitations in researching
metastatic development in vitro, gaining insight into the effects of nsGSL expression on tumor development in vivo is critical to explore the opportunities for therapeutic intervention.
For future studies, we are finishing the in vitro experimental setup and, after proper validation, plan to start in vivo mouse experiments. First, we need to achieve a homogenous GFP positive population for all three murine cell lines transduced with mUGCG and mSPPL3 KO gRNA. Second, we will validate our mUGCG and mSPPL3 KOs cells by performing deep sequencing to anticipate the percentage of KO cells. We will carry out another validation step by performing mass spectrometry to identify expected variation in the nsGSL surface composition of the mUGCG and mSPPL3 KO cells in comparison to the mother clone. Furthermore, to create a high nsGSL profile, we will establish overexpression of mB3GNT5 in the mSPPL3 KO cells. Once we have realized the various nsGSL profiles of SIINFEKL expressing murine tumor cells, an OT-1 hybridoma co-culture will follow to determine the effects of nsGSLs on MHC-1 and TCR interaction. If our setup is in accordance with previous literature, we expect our OT-1 hybridoma’s to be less responsive to high nsGSL murine tumor cells compared to tumor cells with an absent nsGSL profile in vitro. Eventually, we aim to transfer the SIINFEKL expressing murine tumor cells with diverse nsGSL profiles in mice and determine the effects of nsGSLs on tumor development and immunosurveillance in vivo.
We speculate that the nsGSL surface composition affects the anti-tumor immune response and represents a targetable layer for immunotherapy. FDA approved GSL synthesis inhibitors eliglustat (Cerdelga) and milgustat (Zavesca) may be applicable to improve immunogenicity in tumors and develop an immunotherapeutic strategy in the future.
.
References
Alarcón, B., Gil, D., Delgado, P., & Schamel, W. W. (2003). Initiation of TCR signaling: regulation within CD3 dimers. Immunological reviews, 191(1), 38-46.
Blees, A., Januliene, D., Hofmann, T., Koller, N., Schmidt, C., Trowitzsch, S., ... & Tampé, R. (2017). Structure of the human MHC-I peptide-loading complex. Nature, 551(7681), 525-528.
Cornel, A. M., Mimpen, I. L., & Nierkens, S. (2020). MHC Class I downregulation in cancer: underlying mechanisms and potential targets for cancer immunotherapy. Cancers, 12(7), 1760.
D'Angelo, G., Capasso, S., Sticco, L., & Russo, D. (2013). Glycosphingolipids: synthesis and functions. The FEBS journal, 280(24), 6338-6353.
Daugherty, P. S., Iverson, B. L., & Georgiou, G. (2000). Flow cytometric screening of cell-based libraries. Journal of immunological methods, 243(1-2), 211-227.
Debele, T. A., Yeh, C. F., & Su, W. P. (2020). Cancer Immunotherapy and Application of Nanoparticles in Cancers Immunotherapy as the Delivery of Immunotherapeutic Agents and as the Immunomodulators. Cancers, 12(12), 3773.
Dimberu, P. M., & Leonhardt, R. M. (2011). Cancer immunotherapy takes a multi-faceted approach to kick the immune system into gear. The Yale journal of biology and medicine, 84(4), 371.
Dunn, G. P., Koebel, C. M., & Schreiber, R. D. (2006). Interferons, immunity and cancer immunoediting. Nature Reviews Immunology, 6(11), 836-848.
Furukawa, J. I., Tsuda, M., Okada, K., Kimura, T., Piao, J., Tanaka, S., & Shinohara, Y. (2015). Comprehensive glycomics of a multistep human brain tumor model reveals specific glycosylation patterns related to malignancy. PloS one, 10(7), e0128300.
Ganesh, K., Stadler, Z. K., Cercek, A., Mendelsohn, R. B., Shia, J., Segal, N. H., & Diaz, L. A. (2019). Immunotherapy in colorectal cancer: rationale, challenges and potential. Nature reviews Gastroenterology & hepatology, 16(6), 361-375.
Garcia‐Lora, A., Algarra, I., & Garrido, F. (2003). MHC class I antigens, immune surveillance, and tumor immune escape. Journal of cellular physiology, 195(3), 346-355.
Guarnaccia, S. P., Shaper, J. H., & Schnaar, R. L. (1983). Tunicamycin inhibits ganglioside biosynthesis in neuronal cells. Proceedings of the National Academy of Sciences, 80(6), 1551-1555.
Hakomori, S. (2003). Structure, organization, and function of glycosphingolipids in membrane. Current opinion in hematology, 10(1), 16-24.
Hanahan, D., & Weinberg, R. A. (2000). The hallmarks of cancer. cell, 100(1), 57-70.
Haynes, M. K., Strouse, J. J., Waller, A., Leitao, A., Curpan, R. F., Bologa, C., ... & Thompson, T. A. (2009). Detection of intracellular granularity induction in prostate cancer cell lines by small molecules using the HyperCyt® high-throughput flow cytometry system. Journal of biomolecular screening, 14(6), 596-609.
Haynes, N. M., Hawkins, E. D., Li, M., McLaughlin, N. M., Hämmerling, G. J., Schwendener, R., ... & Smyth, M. J. (2010). CD11c+ dendritic cells and B cells contribute to the tumoricidal activity of anti-DR5 antibody therapy in established tumors. The Journal of Immunology, 185(1), 532-541.
He, Q., Jiang, X., Zhou, X., & Weng, J. (2019). Targeting cancers through TCR-peptide/MHC interactions. Journal of hematology & oncology, 12(1), 1-17.
Hicklin, D. J., Marincola, F. M., & Ferrone, S. (1999). HLA class I antigen downregulation in human cancers: T-cell immunotherapy revives an old story. Molecular medicine today, 5(4), 178-186.
Huse, M. (2009). The T-cell-receptor signaling network. Journal of cell science, 122(9), 1269-1273. Jongsma, M. L., Raaben, M., de Waard, A. A., Zhang, T., Cabukusta, B., Platzer, R., ... & Spaapen, R. (2020). The SPPL3-defined glycosphingolipid repertoire regulates immune responses by improving HLA class I access.
Makuku, R., Khalili, N., Razi, S., Keshavarz-Fathi, M., & Rezaei, N. (2021). Current and Future Perspectives of PD-1/PDL-1 Blockade in Cancer Immunotherapy. Journal of Immunology Research, 2021.
Neefjes, J., Jongsma, M. L., Paul, P., & Bakke, O. (2011). Towards a systems understanding of MHC class I and MHC class II antigen presentation. Nature reviews immunology, 11(12), 823-836.
Park, H. S., Cho, U., Im, S. Y., Yoo, C. Y., Jung, J. H., Suh, Y. J., & Choi, H. J. (2019). Loss of human leukocyte antigen class I expression is associated with poor prognosis in patients with advanced breast cancer. Journal of pathology and translational medicine, 53(2), 75.
Peters, S., Reck, M., Smit, E. F., Mok, T., & Hellmann, M. D. (2019). How to make the best use of immunotherapy as first-line treatment of advanced/metastatic non-small-cell lung cancer. Annals of Oncology, 30(6), 884-896.
Rock, K. L., Reits, E., & Neefjes, J. (2016). Present yourself! By MHC class I and MHC class II molecules. Trends in immunology, 37(11), 724-737.
Salama, E. A., Adbeltawab, R. E., & El Tayebi, H. M. (2020). XIST and TSIX: Novel cancer immune biomarkers in PD-L1-overexpressing breast cancer patients. Frontiers in oncology, 9, 1459.
Seliger, B., Wollscheid, U., Momburg, F., Blankenstein, T., & Huber, C. (2001). Characterization of the major histocompatibility complex class I deficiencies in B16 melanoma cells. Cancer research, 61(3), 1095-1099.
Sharma, P., Hu-Lieskovan, S., Wargo, J. A., & Ribas, A. (2017). Primary, adaptive, and acquired resistance to cancer immunotherapy. Cell, 168(4), 707-723.
Stewart, T. J., & Abrams, S. I. (2008). How tumours escape mass destruction. Oncogene, 27(45), 5894-5903.
Surre, J., Saint-Ruf, C., Collin, V., Orenga, S., Ramjeet, M., & Matic, I. (2018). Strong increase in the autofluorescence of cells signals struggle for survival. Scientific reports, 8(1), 1-14.
Thor Straten, P., & Garrido, F. (2016). Targetless T cells in cancer immunotherapy. Journal for immunotherapy of cancer, 4(1), 1-5.
Trowitzsch, S., & Tampé, R. (2020). Multifunctional chaperone and quality control complexes in adaptive immunity. Annual review of biophysics, 49, 135-161.
Unanue, E. R., & Cerottini, J. C. (1989). Antigen presentation 1. The FASEB journal, 3(13), 2496-2502. Urruticoechea, A., Alemany, R., Balart, J., Villanueva, A., Vinals, F., & Capella, G. (2010). Recent advances in cancer therapy: an overview. Current pharmaceutical design, 16(1), 3-10.
Van Echten, G., Iber, H., Stotz, H., Takatsuki, A., & Sandhoff, K. (1990). Uncoupling of ganglioside biosynthesis by Brefeldin A. European journal of cell biology, 51(1), 135-139.
World Health Organization, 2021. A report about cancer. Retrieved from https://www.who.int/health-topics/cancer#tab=tab_1
Zhang, T., de Waard, A. A., Wuhrer, M., & Spaapen, R. M. (2019). The role of glycosphingolipids in immune cell functions. Frontiers in immunology, 10, 90.
Zhuo, D., Li, X., & Guan, F. (2018). Biological roles of aberrantly expressed glycosphingolipids and related enzymes in human cancer development and progression. Frontiers in physiology, 9, 466.
