TRIM28 (Tripartite Motif-containing 28), also known as KAP1 (KRAB-Associated Protein 1) or TIF1-β (Transcriptional Intermediary Factor 1β), was previously shown to be involved in many aspects of cell homeostasis [1
]. As a cofactor for an abundant family of KRAB-ZNF (Krüppel associated box (KRAB) domain Zinc Finger Proteins) transcription factors, TRIM28 mediates the repression of a vast number of target genes [2
]. Furthermore, TRIM28 takes part in the DNA damage response pathway [4
], safeguards the genome stability through inhibition of retrotransposition [5
], stimulates the epithelial-to-mesenchymal transition (EMT) [6
], inhibits p53 activity [7
], and induces the autophagosome formation that facilitates cell survival [8
]. Moreover, TRIM28 is strictly associated with the maintenance of stem cell self-renewal [9
]. Recent studies revealed that TRIM28 safeguards stem cell phenotype at least partially by repressing the genes related to cell differentiation and inducing stemness markers [10
]. All the above-mentioned functions of TRIM28 are frequently harnessed by cancer cells to promote cancer development and progression.
The last decades of research have led to the identification of a specific population of cancer cells endowed with the capability to self-renew, differentiate, and be highly responsible for tumor growth and progression [11
]. These cells, known as cancer stem cells (CSCs), possess intrinsic resistance to chemo- and radiotherapy, a high metastatic potential, and provide tumor relapse after treatment. Previously considered to be a small population, CSCs appeared to be heterogeneous and sometimes numerous within specific types of cancer [12
]. Due to their high plasticity, CSCs may experience phases of transition between stem-like and non-stem-like states. Furthermore, increasing evidence demonstrates that bulk tumor cells can acquire stem cell-like phenotype in response to exogenous stimuli. This suggests that the process of cancer cell differentiation can be reversed and further adds to the cancer heterogeneity [11
]. Altogether, considerable controversy remains as to how unequivocally define CSCs and to which extent distinct tumor types possess a hierarchical organization. Nevertheless, a growing body of evidence indicates that stem cell-associated molecular features, collectively known as stemness, are biologically important in cancer development and progression [11
As TRIM28 emerges as a guard of the intrinsic state of cell differentiation, maintaining stem cells and somatic cells in the pluripotent and differentiated state, respectively, its’ role in cancer stem cell maintenance was not surprising. To date, several possible mechanisms were suggested for TRIM28 to facilitate the acquisition of stem-cell like phenotype in distinct cancer types, and therefore to contribute to worse patient outcomes [14
Recently, stem cell-associated molecular features, frequently referred to as stemness, were recognized as valuable predictive or prognostic characteristics [16
]. Stem cell gene signatures were utilized to develop gene expression-based biomarkers that ultimately proved a strong association of stem-like phenotype with worse patient outcomes across many types of cancer [19
]. Furthermore, molecular signatures capable of grading cancer stemness represent an essential step in designing novel therapeutic regimens that target cancer stem cells. These strategies may have more success in preventing long-term recurrence than currently used therapies evaluated based on their ability to reduce the overall size of a tumor.
Cancer stemness was also negatively associated with antitumor immunity with stemness-high cancers exhibiting reduced immune cell infiltration [19
]. Miranda A. et al. [20
] suggested that the stemness phenotype found in cancer cells, similar to that in normal stem cells, involves the expression of immunosuppressive factors that engender the formation of immune-privileged microenvironments in which tumor clone diversification can occur.
Here, we analyzed the association of TRIM28 expression with the stemness of patient-derived melanoma samples and evaluated potential value for TRIM28 in predicting stem-like phenotype. Using the sphere-forming assay as a gold standard assay to assess cancer cell stemness in vitro, we confirmed a role for TRIM28 in facilitating the stem cell-like phenotype of melanoma cells. Moreover, we revealed TRIM28 as a direct mediator of a low antitumor immune response in high stemness melanomas. Specifically, TRIM28 high expression correlated with significant depletion of interferon alpha (IFN-α) and interferon gamma (IFN-γ) response in melanomas, which resulted from significant downregulation of IRF (Interferon Regulatory Factor) transcription factor family members. TRIM28 emerged as a regulator of IRF expression, mediating epigenetic repression of IRF family members in stemness-high melanomas.
However, further studies are needed to determine whether the attenuation of interferon signaling mediated by TRIM28 is sufficient to acquire the stem cell-like phenotype in melanoma. Moreover, the role for TRIM28 as a druggable anticancer target should be clarified. Ultimately, it may pave the way to novel anticancer therapies that directly target cancer stem cell population while enhancing endogenous antitumor immune response.
There are several major findings of this study: (i) TRIM28 robust upregulation is associated with a dramatically lower survival rate of melanoma patients, (ii) the transcriptome profile of the TRIM28HIGH melanomas is significantly enriched with stemness-associated markers, (iii) the TRIM28HIGH expressing tumors are robustly depleted with infiltrating leukocytes, (iv) TRIM28 expression can significantly discriminate the cohort of “stemness high/immune low” melanoma patients, and (v) significant attenuation of interferon signaling in TRIM28HIGH melanomas might result in the acquisition of stem cell-like melanoma phenotype.
gene is highly expressed in different cancer types, and higher expression frequently correlates with poor patient prognosis [14
]. Recently, Fernandez-Marrero Y. et al. [40
] have reported a significant association of TRIM28
expression with the survival of melanoma patients. Using three independent datasets (SKCM TCGA, GSE65904, and GSE19234), we also demonstrated that TRIM28
upregulation is associated with a dramatically lower survival rate of melanoma patients. Interestingly, the expression of other close-related TRIM family members’ (namely, TRIM24, TRIM33, and TRIM66) does not correlate with the survival of melanoma patients, suggesting that TRIM28 is specifically involved in melanoma progression.
Furthermore, using transcriptomic data from melanoma samples, we revealed distinct gene expression profiles of the TRIM28HIGH melanomas robustly enriched with genes involved in stemness-facilitating biological processes and significantly depleted with markers associated with immune cell infiltration. In contrast to TRIM24- or TRIM33- or TRIM66-associated gene expression profiles, the one correlated with TRIM28 expression is significantly enriched with stemness-related biological processes accompanied by robust depletion of the immune response. This confirmed previously reported relation between TRIM28 level and cancer dedifferentiation status and suggested a specific role for TRIM28 (and not other close-related TRIM family members) in cancer reprogramming.
To further evaluate this association, we implemented several stemness quantifiers [16
] and correlated it to TRIM28
expression. Previously, Ben-Porath I. et al. [16
] have found that histologically poorly differentiated tumors show preferential overexpression of genes normally enriched in embryonic stem (ES) cells. They demonstrated that this ES-like signature was associated with high-grade estrogen receptor (ER)-negative tumors, often of the basal-like subtype, and with poor clinical outcome. Moreover, the ES signature was present in poorly differentiated glioblastomas and bladder carcinomas, suggesting its’ versatility in the acquisition of cancer dedifferentiation status regardless of the tumor type. Moreover, Wong DJ. et al. [17
] have identified the embryonic stem cell (ESC) transcriptional program that is frequently activated in diverse human epithelial cancers and strongly predicts metastasis and death. They also identified the c-Myc oncogene as being sufficient to reactivate the ESC-like program in cancer cells. Also, we looked at the recently reported the mRNA Stemness Index, developed with the innovative one-class logistic regression (OCLR) machine learning algorithm. In their analyses, Malta T. et al. [19
] have observed that higher values of the mRNA Stemness Index (mRNA-SI) were significantly associated with known biological processes active in the CSCs and with greater tumor dedifferentiation, as reflected in the histopathological grade.
Here, we analyzed the correlation of TRIM28
expression with the ES-derived gene signatures and observed significant positive association. We also observed a positive correlation between TRIM28
expression and the mRNA-SI and significant enrichment of the TRIM28HIGH
transcriptome profile with markers of the mRNA-SI gene signature. Moreover, TRIM28
expression emerged as a good predictor of stem cell-associated melanoma phenotype, regardless of the stemness quantifier. The TRIM28HIGH
-expressing melanomas were significantly enriched with c-Myc associated gene signature. As c-Myc is sufficient to induce cancer stem cell phenotype in epithelial cancers [17
], it is possible that (at least partially) melanoma stem cell-like phenotype of the TRIM28HIGH
melanomas results from significant c-Myc activation.
We further suggest that TRIM28 high expression facilitates the acquisition of a stem cell-like phenotype. Using a panel of melanoma cell lines, we observed that TRIM28 high expressing cells possess greater melanosphere formation potential, which is reflective of stem cell phenotype. Moreover, using shRNA-mediated targeting of TRIM28 in three melanoma cell lines, we confirmed the involvement of TRIM28 in cancer stem cell maintenance. In SK-MEL28, WM9, and A375 melanoma cell lines with downregulated TRIM28 expression, the sphere formation was significantly reduced. As TRIM28 depletion did not affect cell proliferation or cell death directly (preliminary data for SK-MEL28, not shown), we suggest that it might attenuate the maintenance of stem cell-like population, therefore resulting in smaller sizes of melanospheres. Indeed, the expression of stemness markers was lowered in TRIM28-depleted melanoma cells, which strongly supports TRIM28-mediated acquisition of stem cell-like phenotype in cancer.
Recently, Lee AK. et al. [41
] have demonstrated that TRIM28 affected intratumoral lymphocyte infiltration in melanoma patients. Higher levels of TRIM28
negatively correlated with different immune effector subsets as determined using the CIBERSORT. Here, using the gene signatures reflective of distinct immune cell subpopulations [33
], we revealed significant depletion of cytotoxic T cell, helper T cells, B cells, macrophages, and eosinophils in the TRIM28HIGH
melanomas, further supporting the association between the TRIM28
expression and “immune cold” melanoma phenotype.
Moreover, analyses of the tumor microenvironment performed by Malta T. et al. [19
] have revealed unanticipated correlation of cancer stemness with immune checkpoint expression and infiltrating immune cells. This was further supported by Miranda A. et al. [20
]. They observed that cancer stemness is associated with suppressed immune response, higher intratumoral heterogeneity, and dramatically worse outcome for most TCGA cancers. Our results are in line with previous reports, suggesting that TRIM28 protein is a direct link between cancer stemness acquisition and suppression of antitumor immune response. TRIM28 may serve as a good predictor, efficiently discriminating the “stemness high/immune low” melanoma phenotype. It would be critical to determine whether there is a role for TRIM28 in attenuation of immune cell infiltration, or rather TRIM28 is a mediator of the “immune cold” phenotype of melanomas.
TRIM28 maintains both normal and cancer stem cells in the pluripotent state at least partially by repressing the genes associated with differentiation and inducing expression of stemness markers [15
]. TRIM28 uses KRAB-ZNFs to cause epigenetic silencing of target differentiation genes via H3K9me3 deposition and DNA methylation [1
], although interactions with several other non-KRAB-ZNF transcription factors were also reported [46
]. Together with the EZH2, a member of Polycomb Repressor 2 (PRC2) Complex, TRIM28, co-regulates a set of genes associated with stem cell maintenance [46
]. On the other hand, together with the MAGE-A3/6 (Melanoma Antigen member A3, A6), TRIM28 forms a cancer-specific ubiquitinase that regulates the AMPK level in cancer cells, enhancing the oxidative phosphorylation and maintaining stem cell traits of breast cancer [15
]. This mechanism requires the involvement of RING-mediated ubiquitin E3 ligase activity of TRIM28 protein and might be utilized by other cancer types.
Here, we analyzed the profile of transcription factors significantly associated with the upregulation of TRIM28
to verify the potential mechanism of cancer stemness acquisition. Interestingly, we observed significant downregulation of the IRF transcription factor family members (IRF1, IRF2, IRF5, IRF8) corresponding to attenuated interferon signaling in TRIM28HIGH
melanomas, probably as a consequence of low immune cell infiltration. Recently, Alavi S. et al. [50
] have reported that downregulation of interferon signaling occurred in almost 70% of immunotherapy-naïve melanomas, although they did not link this phenomenon with TRIM28
expression. As TRIM28 is a well-known transcriptional co-repressor, we suggest that low IRF1
, and IRF8
expression (in the absence of stimulating signaling) result from TRIM28-mediated epigenetic modifications within the IRF promoters. Indeed, TRIM28 was previously reported to interact with the STAT1 transcription factor and to regulate IFN/STAT1-mediated IRF1
gene repression. As reported by Kamitani S. et al. [34
], siRNA-mediated reduction in TRIM28
expression enhanced IFN-induced STAT1-dependent IRF1 gene expression. These results indicated that TRIM28 acts as an endogenous regulator of the IFN/STAT1 signaling pathway [34
Enhanced IRF1 promoter methylation observed in TRIM28HIGH expressing melanoma patients is in line with the previous report, although further studies are needed to prove the versatility of TRIM28-mediated repression of other IRF members in “stemness high/immune low” melanomas. Moreover, it should be clarified whether IRF family members’ repression mediated by TRIM28 is sufficient for the acquisition of stem-cell like phenotype in melanoma.
4. Materials and Methods
4.1. SKCM TCGA Genomic and Clinical Data
In the current study, we used publicly available TCGA data from the cBioportal (www.cbioportal.org
]. For skin cutaneous melanoma (SKCM) TCGA patients (n
= 469), the genomic data (copy number alteration, mutation spectrum, and fraction genome altered) and clinical data (sample type, sex, age at diagnosis, tumor location, CLARK level, Breslov depth, and overall survival status) were available for 66.4% to 100% of SKCM samples (Table 1
and Table S1
). All data is available online, and access is unrestricted and does not require patients’ consent or other permissions. The use of the data does not violate the rights of any person or any institution.
4.2. Transcriptomic Data
The RNA sequencing-based mRNA expression data were directly downloaded from the cBioportal. RNASeq V2 from TCGA is processed and normalized using RSEM [51
]. Specifically, the RNASeq V2 data in cBioPortal corresponds to the rsem.genes.normalized_results file from TCGA. Differentially Expressed Genes (DEGs) were cut off at p-
value < 0.05 and FDR < 0.05.
4.3. Gene Set Enrichment Analysis
The Gene Set Enrichment Analysis [52
] was used to detect the coordinated expression of a priori defined groups of genes within the tested samples [53
]. Genesets are available from the Molecular Signatures Database [54
]. All significantly differentially expressed genes (previously ranked based on their log2FC between analyzed groups) were imported to GSEA. The GSEA was run according to the default parameters: each probe set was collapsed into a single gene vector (identified by its HUGO gene symbol), permutation number = 1000, and permutation type = “gene-sets”. The FDR was used to correct for multiple comparisons and gene set sizes.
4.4. Methylation Data
Analyses of the methylation of IRF family members (CpG islands near the 3′UTR or 5′UTR of analyzed genes) were performed with the MEXPRESS database, a straightforward and easy-to-use web tool for the integration and visualization of the expression, DNA methylation, and clinical TCGA data on a single-gene level [55
]. In this study, the probes with the highest negative correlation of methylation (presented as beta values) with the gene expression were analyzed.
4.5. Stemness-Associated Scores
Novel stemness indices for assessing the degree of oncogenic dedifferentiation were previously determined using transcriptomic (mRNA Stemness Index) or epigenetic (mDNA Stemness Index) features of non-transformed pluripotent stem cells and their differentiated progeny [19
]. In the current study, we used publicly available information for 469 SKCM samples. For further details, see in [19
]. All stem cell-derived scores used in this study are shown in Table S2
4.6. Immune-Associated Scores
The Lymphocyte Scores were assessed by a pathologist based on the density (scored 0–3) and the distribution (scored 0–3) of lymphocytes in tumor samples, as previously reported [30
]. The Immune Scores were estimated based on the transcriptome profiles of each sample, as previously reported [31
]. The Leukocyte Infiltration Scores (LIS) were calculated based on the expression level of 25 genes, as previously reported [32
]. All immune-associated scores used in this study are shown in Table S4
4.7. Validation Sets
Additional datasets used in this study (GSE65904, melanoma, n
= 214 samples [26
]; GSE19234, metastatic melanoma, n
= 44 samples [27
]) were obtained from the R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl
). We selected these sets based on the availability of patients’ survival data. GSE65904 and GSE19234 datasets were analyzed online using the R2: Genomic Analysis and Visualization Platform to find genes that correlate with TRIM28 expression. All data is freely available online, and access is unrestricted and does not require patients’ consent or other permissions.
4.8. Statistical Analyses
Statistical analyses were carried out with GraphPad Prism 8.0 software (GraphPad Software, Inc., La Jolla, CA, USA). Single comparisons between two groups were performed with the Student’s t-test. Multiple comparisons were performed with the ANOVA followed by Tukey’s multiple comparisons post-test. The correlation between the two variables was assessed with Spearman’s rank correlation coefficient (r). For differential gene expression analysis, the statistical significance was tested with the Student’s t-test followed by a False Discovery Rate (FRD) correction with Benjamini–Hochberg procedure.
Survival analysis was performed according to the Kaplan–Meier analysis and log-rank test. Overall survival (OS) was defined as the time between the date of surgery and the date of death or the date of the last follow-up.
4.9. Melanoma Cell Lines
The original cell lines were obtained from American Type Culture Collection (ATCC, Manassas, VA, USA) or Rockland Immunochemicals, Inc. (Limerick, PA, USA). All cell lines were grown in Dulbecco’s Modified Eagle’s Medium (DMEM), 4.5 g/L of glucose, with 10% fetal bovine serum (FBS), 50 units/mL penicillin, and 50 μg/mL streptomycin (all from Invitrogen, Carlsbad, CA, USA) in a humidified atmosphere at 37 °C, 21% O2, and 5% CO2.
To produce lentiviral vectors (LV-shTRIM28, LV-puro-ctrl, LV-shTRIM28-TRIM28 (rescue)), HEK-293T cells were co-transfected with psPAX2, pMD2.G and lentiviral plasmid pWPTS-shTRIM28, pWPTS-puro-ctrl, or pWPTS-shTRIM28-TRIM28. The culture supernatant was collected 48 h post-transfection and passed through 0.45 µm filters, and aliquots were stored at −80 °C. Three melanoma cell lines were infected with lentiviruses, and 1 µg/mL puromycin was added 72 h after infection. The cells were selected using puromycin (Sigma-Aldrich, St. Louis, MO, USA) for 8–10 days and were subsequently tested for TRIM28 expression.
4.10. RT-qPCR Analyses
Total RNA was extracted using TRI Reagent RNA Isolation Reagent (Sigma Aldrich, St. Louis, MO, USA) according to the manufacturer’s protocol. The quality and quantity of RNA were evaluated by the NanoDrop 2000 UV-Vis Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and was stored in −80 °C. Reverse transcription was performed using the iScript cDNA Synthesis Kit (Bio- Rad Laboratories, Inc. Hercules, CA, USA) with 1 μg of total RNA per reaction. Gene-specific primers and probes (Universal ProbeLibrary Set, Human, Roche, Basel, Switzerland) were used for real-time qPCR. PCR amplification and fluorescence detection were performed using a LightCycler® 96 Real-Time PCR detection system (Roche, Basel, Switzerland), and the threshold cycles were determined using Light Cycler® 96 Software. Fold inductions were determined using the 2(−ΔΔCt) method against the GAPDH gene.
4.11. Western Blot
Whole-cell lysates were prepared by lysing the cells with radioimmune precipitation assay (RIPA) buffer (Sigma-Aldrich, St. Louis, MO, USA) completed with Complete Protease Inhibitor Mixture (Roche) according to the manufacturer’s protocol. Protein lysates were stored in −80 °C for further analyses. Protein concentration was measured with PierceTM BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA), and then samples were subjected to SDS-PAGE (using Mini-PROTEAN® Tetra Cell System, Bio-Rad) followed by immunoblotting with antibodies for TRIM28 (ab10483 or ab10484, Abcam, Cambridge, UK) and GAPDH (ab9485, Abcam). The blots were visualized using an enhanced chemiluminescence detection kit (ECL-Plus, Amersham Biosciences, Little Chalfont, UK) and a G:BOX F3 Gel Documentation System (Syngene, Bangalore, India).
4.12. Cell Proliferation Assay
The cells were seeded into 96-well plates in a number of 1 × 103 per well in triplicates. A 20 μL aliquot of full medium containing 1 μCi of 3H-thymidine (specific activity 70–90 Ci/mmol, 2590–3330 GBq/mmol, Perkin Elmer, Waltham, MA, USA) was added to each well 16–18 h before the termination of culture by freezing. Incorporated 3H-thymidine was assessed using a Micro Beta TriLux scintillation counter (Perkin Elmer, Waltham, MA, USA).
4.13. Sphere Formation Assay
For Soft Agar Colony Formation Assay, 12-well culture plates were coated with 0.66% low melting agarose (Lonza SeaPlaqueTM Agarose, Lonza Group, Basel, Switzerland) in a serum-free DMEM/F12 (350 μL per well) supplemented with B27 supplement (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), 20 ng/mL basic fibroblast growth factor (bFGF, Invitrogen, Carlsbad, CA, USA), 20 ng/mL epidermal growth factor (EGF, Invitrogen, Carlsbad, CA, USA), and 1% penicillin–streptomycin (Sigma-Aldrich, St. Louis, MO, USA). After the bottom layer solidified, the top layer was prepared by mixing 0.66% agarose/DMEM/F12 (supplemented as described above) with cell suspension to obtain 0.33% concentration of agarose, and then the top layer was added at the bottom layer (350 μL containing 1 × 103 of cells). Each well was checked under the microscope to confirm single-cell suspension with no cell clusters. The plates were incubated in a humidified atmosphere at 37 °C, 5% CO2, and 1% O2 (hypoxia) for two weeks. To determine the efficiency of colony formation, spheres were stained with 250 μL of 0.005% crystal violet for 1 h, and then were counted under the microscope (Leica DMI 3000B inverted microscope). Besides the number of outgrowing spheres, also the size and the morphology was monitored.