Published online 28 October 2022 Nucleic Acids Research, 2022, Vol. 50, No. 19 10981–10994 https://doi.org/10.1093/nar/gkac941 Vitamin C enhances NF-␬B-driven epigenomic reprogramming and boosts the immunogenic properties of dendritic cells Octavio Morante-Palacios1,2 , Gerard Godoy-Tena1 , Josep Calafell-Segura1 , Laura Ciudad1 , Eva M. Martı́nez-Cáceres3,4 , José Luis Sardina5 and Esteban Ballestar 1,2,6,* 1 Epigenetics and Immune Disease Group, Josep Carreras Research Institute (IJC), 08916, Badalona, Barcelona, Spain, 2 Germans Trias i Pujol Research Institute (IGTP), 08916, Badalona, Barcelona, Spain, 3 Division of Immunology, Germans Trias i Pujol Hospital, LCMN, Germans Trias iPujol Research Institute (IGTP), 08916, Badalona, Barcelona, Spain, 4 Department of Cell Biology, Physiology, Immunology, Autonomous University of Barcelona, 08193, Bellaterra, Barcelona, Spain, 5 Epigenetic Control of Haematopoiesis Group, Josep Carreras Research Institute (IJC), 08916, Badalona, Barcelona, Spain and 6 Epigenetics in Inflammatory and Metabolic Diseases Laboratory, Health Science Center (HSC), East China Normal University (ECNU), Shanghai 200241, China Received June 22, 2022; Revised September 23, 2022; Editorial Decision October 06, 2022; Accepted October 10, 2022 ABSTRACT INTRODUCTION Dendritic cells (DCs), the most potent antigen- Dendritic cells (DCs) play a central role in the immune sys- presenting cells, are necessary for effective activa- tem, bridging innate and adaptive immune responses. As tion of naı̈ve T cells. DCs’ immunological properties innate immune cells, they are able to recognize a plethora are modulated in response to various stimuli. Ac- of pathogen-associated molecular patterns (PAMPs) and tive DNA demethylation is crucial for DC differenti- damage-associated molecular patterns (DAMPs) through pattern recognition receptors (PRRs) (1). Moreover, they ation and function. Vitamin C, a known cofactor of are very efficient at antigen processing and presentation to ten-eleven translocation (TET) enzymes, drives ac- T cells and are, therefore, responsible for initiating antigen- tive demethylation. Vitamin C has recently emerged specific immune responses. as a promising adjuvant for several types of can- DCs are highly heterogeneous, and comprise plasmacy- cer; however, its effects on human immune cells are toid DCs (pDCs) and conventional DCs (cDCs). Further- poorly understood. In this study, we investigate the more, monocyte-derived DCs (moDCs) can be differenti- epigenomic and transcriptomic reprogramming or- ated in vitro from monocytes (MOs), with GM-CSF and chestrated by vitamin C in monocyte-derived DC dif- IL-4 (2). moDCs have been classically used as a convenient ferentiation and maturation. Vitamin C triggers ex- model that mimics blood DCs, especially cDC2 (3). How- tensive demethylation at NF-␬B/p65 binding sites, ever, increasing evidence indicates that MOs can extravasate together with concordant upregulation of antigen- to peripheral tissues and give rise to bona fide moDCs in vivo (4). presentation and immune response-related genes Changes in DNA methylation, mainly active DNA during DC maturation. p65 interacts with TET2 and demethylation, are involved in several differentiation pro- mediates the aforementioned vitamin C-mediated cesses from MOs, leading to macrophages (5), osteoclasts changes, as demonstrated by pharmacological inhi- (6) and DCs (7) and are crucial for immune cell dif- bition. Moreover, vitamin C increases TNF␤ produc- ferentiation, identity, and function (8). In general, DNA tion in DCs through NF-␬B, in concordance with the demethylation is more extensive during MO differentiation upregulation of its coding gene and the demethyla- than during subsequent maturation/activation (7). Addi- tion of adjacent CpGs. Finally, vitamin C enhances tionally, the maturation of moDCs with live bacteria pro- DC’s ability to stimulate the proliferation of autolo- duces DNA demethylation that follows gene activation, lim- gous antigen-specific T cells. We propose that vita- iting the potential direct regulatory effects of DNA methy- min C could potentially improve monocyte-derived lation in such a context (9). TET2, a member of the Ten- Eleven Translocation (TET) methylcytosine dioxygenases, DC-based cell therapies. * To whom correspondence should be addressed. Tel: +34 935572800; Fax: +34 934651472; Email: eballestar@carrerasresearch.org C The Author(s) 2022. Published by Oxford University Press on behalf of Nucleic Acids Research. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com 10982 Nucleic Acids Research, 2022, Vol. 50, No. 19 has been pointed out as the main enzyme involved in multi- verified by flow cytometry, which yielded >95% of CD14+ step active demethylation processes in terminal MO-related cells. differentiation processes (6,10). However, recent data indi- MOs were resuspended in Roswell Park Memorial Insti- cate that TET3 might complement TET2-mediated activ- tute (RPMI) Medium 1640+ GlutaMAX™ (Gibco, Ther- ity during MO differentiation (11). Recently, TET2 has also moFisher) and immediately added to cell culture plates. been implicated in glucocorticoid- and vitamin D-mediated After 20 min, monocytes were attached to the cell culture modulation of the immunogenic properties of DCs (12,13). plates, and the medium was changed with RPMI contain- Vitamin C (L-ascorbic acid) is an essential nutrient with ing 10% fetal bovine serum (Gibco, ThermoFisher), 100 pleiotropic functions. Its deficiency is associated with a dis- units/ml penicillin/streptomycin (Gibco, ThermoFisher), ease, namely scurvy, characterized by a plethora of symp- 10 ng/ml human GM-CSF (PeproTech) and 10 ng/ml hu- toms, including the malfunction of the immune system. For man IL-4 (PeproTech). In the case of the cells treated instance, the normal intracellular level of vitamin C in MO with vitamin C, 500 ␮M (+)-sodium L-ascorbate (Sigma- cytoplasm is ∼3 mM, 60 times higher than the plasma level, Aldrich) was also added to the medium. For dendritic cell reflecting a specific function of the molecule in immune maturation, LPS (10 ng/ml) was added to cell culture on cell biology (14). Vitamin C can act as a cofactor of Fe- day 5. containing hydroxylases such as TET enzymes and Jumonji Cells were collected on day 2 (designated as Ø/ØvitC , C domain-containing histone demethylases (JHMDs), en- given their incomplete differentiation status) and on day 7, hancing their catalytic activity (15). Some studies in mice including immature dendritic cells (iDC/iDCvitC ) and ma- suggest that vitamin C can stimulate DC capacity to pro- ture dendritic cells, with LPS stimulus (mDC/mDCvitC ). duce proinflammatory cytokines and promote differentia- tion of T cells (16). Moreover, vitamin C intravenous treat- Genomic DNA and total RNA extraction ment in mice has been shown to abrogate cancer progression through direct TET2 function restoration in cancer cells Genomic DNA and total RNA were extracted using the (17) and immune system modulation (18). Maxwell RSC Cultured Cells DNA kit (Promega) and the The in vivo modulation of DC migration and function, Maxwell RSC simplyRNA Cells kit (Promega), respec- and the administration of DC-based vaccines, are poten- tively, following the manufacturer’s instructions. tial strategies to treat different types of cancer (19). In par- ticular, autologous moDCs obtained ex vivo from patient- NF-␬B chemical inhibition derived blood MOs have been used in several clinical trials with mixed results (20–22). In this regard, the improvement Bay 11-7082 (Sigma-Aldrich) was used for NF-␬B inhibi- of moDC generation in vitro, and the use of molecules to tion. The compound was diluted in DMSO to 50 mM. Bay modulate MO differentiation in vivo may boost the clinical 11-7082 at 10␮M or an equivalent amount of diluent were outcome of cancer patients. used as final concentrations. In this work, we have investigated the effects of vitamin MOs were differentiated to iDCs/iDCvitC as previously C treatment during MO-to-DC in vitro differentiation and described. On day 5, LPS, Bay 11–7082 or equivalent maturation, identifying extensive DNA demethylation as- amounts of diluents were added to the cell culture for 2 sociated with the upregulation of migration, chemotaxis, days, yielding iDC/iDCvitC , iDC + Bay/iDCvitC + Bay, antigen presentation, and immune response-related genes. mDC/mDCvitC , and mDC + Bay/mDCvitC + Bay. Moreover, vitamin C-mediated DNA demethylation, gene upregulation, and increased TNF␤ production during DC DNA methylation profiling maturation were associated with p65, a component of the NF-␬B complex that interacts with TET2 in this context. 500 ng of genomic DNA was converted using the EZ DNA We have shown that the modulation of DNA methylation Methylation Gold kit (Zymo Research), using 4 biological changes during DC differentiation and maturation yields replicates for each group. Infinium MethylationEPIC Bead- functional and phenotypic changes in these cells, improv- Chip (Illumina) arrays were used to analyze DNA methyla- ing their immunogenicity. tion, following the manufacturer’s instructions. This plat- form allows around 850 000 methylation sites per sample to be interrogated at single-nucleotide resolution, covering MATERIALS AND METHODS 99% of the reference sequence (RefSeq) genes. Raw files (IDAT files) were provided for the Josep Carreras Research CD14+ monocyte purification and culture Institute Genomics Platform (Barcelona). Buffy coats were obtained from anonymous donors via Quality control and analysis of EPIC arrays were per- the Catalan Blood and Tissue Bank (CBTB). The CBTB formed using ShinyÉPICo (23) a graphical pipeline that follows the principles of the World Medical Association uses minfi (24) for normalization, and limma (25) for dif- (WMA) Declaration of Helsinki. Before providing blood ferentially methylated positions analysis. CpH and SNP samples, all donors received detailed oral and written in- loci were removed and the Noob + Quantile normaliza- formation and signed a consent form at the CBTB. tion method was used (26,27). After quality control, 831 421 PBMCs were isolated by density-gradient centrifuga- CpGs were preserved for the analysis. Donor ID was used as tion using lymphocyte-isolation solution (Rafer). Pure MOs a covariate, and Trend and Robust options were enabled for were then isolated from PBMCs by positive selection with the eBayes moderated t-test analysis. CpGs were considered magnetic CD14 MicroBeads (Miltenyi Biotec). Purity was differentially methylated when the absolute differential of Nucleic Acids Research, 2022, Vol. 50, No. 19 10983 methylation was >30% (␤ > 0.3) and the FDR was <0.05. mg/ml) in round bottom 96-well plates (ThermoFisher). Infinium MethylationEPIC BeadChip arrays were designed Co-culture was performed in the presence of SARS-CoV-2- using the hg19 human genome. For some downstream anal- S antigen or SARS-CoV-2-N control antigen (PepTivator). ysis, the genomic coordinates were converted to hg38 using T cell proliferation was analyzed by FACS and determined the LiftOver tool (28). by considering the proliferating of those where CFSE stain- For the genome-wide DNA methylation study after NF- ing had decreased compared to not co-cultured T cells. T ␬B chemical inhibition, the same experimental approach cells stimulated with anti-CD3/CD28 microbeads 5 ug/ml and normalization pipeline was followed. After quality con- (eBioscience) were used as a positive control. trol, 812 429 CpGs were preserved for the analysis. Flow cytometry RNA-seq To study cell-surface markers, cells were collected us- RNA-seq libraries of MOs, Ø/ØvitC , iDC/iDCvitC and ing Versene, a non-enzymatic dissociation buffer (Ther- mDC/mDCvitC were generated and sequenced by Novo- moFisher). Cells were resuspended in the staining buffer gene (Cambridge), in 150-bp paired-end, with the Illumina (PBS with 4% fetal bovine serum and 2 mM ethylenedi- NovaSeq 6000 platform, using three biological replicates aminetetraacetic acid (EDTA)). Cells were then incubated for each group. More than 40 million reads were obtained on ice with an Fc block reagent (Miltenyi Biotec) for 10 for each sample. Fastq files were aligned to the hg38 tran- minutes, and stained with the proteins of interest, using the scriptome using HISAT2 (29) with standard options. Reads following antibodies: CD8 (FITC) (#21270083, Immuno- mapped in proper pair and primary alignments were se- tools), and CD4 (APC) (#555349, BD Biosciences). After lected with SAMtools (30). Reads were assigned to genes staining, cells were analyzed using a BD FACSCanto™ II with featureCounts (31). Cell Analyzer (BD Biosciences). Data were analyzed with Differentially expressed genes were detected with DE- the FlowJo v10 software. Seq2 (32). The donor was used as a covariate in the model. The Ashr shrinkage algorithm was applied and only protein-coding genes with an absolute logFC >0.5 and an Bisulfite pyrosequencing FDR less than 0.05 were selected as differentially expressed. 500 ng of genomic DNA was converted using the EZ DNA For representation purposes, Variance Stabilizing Transfor- Methylation Gold kit (Zymo Research). PCR was per- mation (VST) values and normalized counts provided by formed using the bisulfite-converted DNA as input and DESeq2 were used. primers designed for each amplicon (Supplementary Table 1). These primers were designed using the PyroMark Assay Quantification of cytokine production Design 2.0 software (Qiagen). PCR amplicons were pyrose- Cell culture supernatants were collected after 7 days and di- quenced using the PyroMark Q48 system and analyzed with luted appropriately. Enzyme-linked immunosorbent assays PyroMark Q48 Autoprep software. (ELISA) were performed to detect TNF␤, following the manufacturer’s instructions (TNF beta Human ELISA Kit, Real-time quantitative reverse-transcribed polymerase chain ThermoFisher). reaction (qRT-PCR) 300 ng of total RNA were reverse-transcribed to cDNA T cell clonal expansion and proliferation assay with Transcriptor First Strand cDNA Synthesis Kit PBMCs from healthy donors were purified from blood by (Roche) following the manufacturer’s instructions. qRT- density gradient centrifugation. The PBMCs (1 ml; 3 × 106 PCR was performed in technical triplicates for each biolog- cells per well) were cultured in the presence of SARS- ical replicate, using LightCycler® 480 SYBR Green Mix CoV-2- S (9pmol) (PepTivator) in 24-well plates and main- (Roche), and 7.5 ng of cDNA per reaction. The average tained in IMDM medium (Gibco) supplemented with peni- value from each technical replicate was obtained. Then, the cillin (100 units/ml), streptomycin (100 mg/ml) and human standard double-delta Ct method was used to determine the serum (10%) (Millipore) in the absence of IL-2 for 3 days. relative quantities of target genes, and values were normal- After 3 days, 1 ml of medium with 80 U/ml of recombinant ized against the control genes RPL38 and HPRT1. Custom human IL-2 (PeproTech) was added to the wells, with a fi- primers were designed to analyze genes of interest (Supple- nal concentration of 1.5 × 106 cell/ml and 40 U/ml of IL- mentary Table 1). 2. After 7–10 days of culture, T cells were expanded in the presence of 30-Gy irradiated autologous PBMCs (3 × 106 Co-immunoprecipitation (Co-IP) cells/well) previously pulsed with 9 pmol of SARS-CoV-2- S (PepTivator). Antigen-specific T cells (mix of CD4+ and Co-IP assays were performed using mDCs and mDCvitC af- CD8+ T cells) were selected by performing the same proto- ter 5 days of differentiation from MOs and 24 h of stimu- cols two times to have a positive selection. lation with LPS. Cell extracts were prepared in lysis buffer After 7 days of differentiation and activation, DCs were [50 mM Tris–HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, washed to remove vitamin C and were co-cultured with 1% Triton-X-100, protease inhibitor cocktail (cOmplete™, antigen-specific autologous CFSE-stained T cells at a DC: Merck)] with corresponding units of Benzonase (Sigma) T cell ratio of 1:2 in 200 ␮l of RPMI 1640 medium contain- and incubated at 4◦ C for 4 h. 100 ␮l of supernatant was ing 10% FBS, penicillin (100 units/ml), streptomycin (100 saved as input and diluted with 2× Laemmli sample buffer 10984 Nucleic Acids Research, 2022, Vol. 50, No. 19 (5× SDS, 20% glycerol, 1 M Tris–HCl (pH 8.1)). Su- munoprecipitated DNA compared to input was calculated pernatants were first precleared with PureProteome™ Pro- with the following formula: 2((Ctinput – 6.64) – Ctsample) × 100% tein A/G agarose suspension (Merck Millipore) for 1 h. The lysate was then incubated overnight at 4◦ C with re- spective crosslinked primary antibodies. The cross-linking Western blotting was performed in 20 mM dimethyl pimelimidate (DMP) Cytoplasmic and nuclear protein fractions were obtained (Pierce, Thermo Fisher Scientific, MA, USA) dissolved in using hypotonic lysis buffer (Buffer A; 10 mM Tris pH 0.2 M sodium borate (pH 9.0). Subsequently, the beads were 7.9, 1.5 mM MgCl2 , 10 mM KCl supplemented with pro- quenched with 0.2 M of ethanolamine (pH 8.0) and resus- tease inhibitor cocktail (Complete, Roche) and phosphatase pended at 4◦ C in PBS until use. Beads were then washed inhibitor cocktail (PhosSTOP, Roche) to lyse the plasma three times with lysis buffer at 4◦ C. Sample elution was done membrane. Cells were visualized in the microscope to as- by acidification using a buffer containing 0.2 M glycine sure correct cell lysis. The nuclear pellets were resuspended (pH 2.3) and diluted with 2× Laemmli. Samples and in- in Laemmli 1× loading buffer. For whole-cell protein ex- puts were denatured at 95◦ C in the presence of 1% ␤- tract, cell pellets were directly resuspended in Laemmli 1× mercaptoethanol. Anti-p65 C15310256 (DIagenode) and loading buffer. control IgG C15410206 (Diagenode) were used for co-IP. Proteins were separated by SDS-PAGE electrophore- sis. Immunoblotting was performed on polyvinylidene di- fluoride (PVDF) membranes following standard proce- dures. Membranes were blocked with 5% Difco™ Skim Chromatin immunoprecipitation Milk (BD Biosciences) and blotted with primary antibod- On day 6, after 5 days of differentiation and 24h of matu- ies. After overnight incubation, membranes were washed ration with LPS or an equivalent amount of diluent, iDC, three times for 10 min with TBS-T (50 mM Tris, 150 mM iDCvitC , mDC and mDCvitC were fixed with Pierce™ fresh NaCl, 0.1% Tween-20) and incubated for 1 h with HRP- methanol-free formaldehyde (ThermoFisher) for 15 min conjugated mouse or rabbit secondary antibody solutions and prepared for sonication with the truChIP Chromatin (Thermo Fisher) diluted in 5% milk (diluted 1/10 000). Shearing Kit (Covaris), following the manufacturer’s in- Finally, proteins were detected by chemiluminescence us- structions. We performed chromatin immunoprecipitation ing WesternBright™ ECL (Advansta). The following anti- assays at 24 h of LPS-mediated maturation to capture the bodies were used: Anti-p65 C15310256 (DIagenode), Anti- binding occurring before the final time point (48 h). Chro- phosphorylated p65 (Ser536) 93H1 (Cell Signaling), Anti- matin was sonicated 18 min with the Covaris M220 in 1 ml GAPDH 2275-PC-100 (Trevigen), Anti-TET2 C15200179 milliTubes (Covaris). The size distribution of the sonicated (Diagenode), Anti-histone H1 ab4269 (Abcam), Anti-beta chromatin was checked by electrophoresis to ensure an ap- Actin ab8227 (Abcam). Protein quantification was per- propriate sonication, with a size around 200 bp. formed with ImageJ/Fiji (33). Magna Beads Protein A + G (Millipore) were blocked with PBS + BSA (5 mg/ml) for 1 h. Chromatin was pre- Data analysis and representation cleared with 25 ␮l of beads for 1.5 h and 10 ␮g of chromatin were incubated overnight with each antibody: 10 ␮l Anti- Statistical analyses were performed in R 4.0.3. Gene expres- p65 antibody ab16502 (Abcam), in a buffer with 1% Triton sion and DNA methylation heatmaps were created with the X-100, 150 mM NaCl and 0.15% SDS. Then, three washes heatmap.2 function of the gplots package. The findMotif- were performed with the Low Salt Wash Buffer (0.1% SDS, sGenome.pl function of HOMER (Hypergeometric Opti- 1% Triton X-100, 2 mM EDTA, pH 8.0, 20 mM Tris–HCl mization of Motif EnRichment) was used to analyze known pH 8.0, 150 mM NaCl), the High Salt Wash Buffer (0.1% motif enrichment, using the parameters ‘-size 200 -cpg. All SDS, 1% Triton X-100, 2 mM EDTA, pH 8.0, 20 mM Tris– EPIC array CpG coordinates were also used as background HCl pH 8, 500 mM NaCl) and the LiCl Wash Buffer (0.25 for the methylation data. GREAT software was used to cal- M LiCl, 1% Nonidet P-40, 1% deoxycholate, 1 mM EDTA culate CpG-associated genes and gene ontology (GO) en- pH 8, 10 mM Tris–HCl), followed by a last wash with TE richment (34). GO enrichment of gene expression data was buffer (pH 8.0, 10 mM Tris–HCl, 1 mM EDTA). Chro- performed using the clusterProfiler package (35). ChIP-seq matin was eluted for 45 min 65◦ C with 100 ␮l of elution peaks files of histone marks from MO, iDCs, and mDCs buffer (10 mM Tris–Cl, 1 mM EDTA, 1% SDS) and de- were downloaded from the BLUEprint webpage (http://dcc. crosslinked adding 5 ␮l 5 M NaCl and 5 ␮l 1 M NaHCO3 blueprint-epigenome.eu). Consensus peaks of the different (2 h 65◦ C). Next, 1 ␮l of 10mg/ml proteinase K (Invitrogen) replicates were obtained with the MSPC algorithm (36), us- was added, and samples were incubated at 37◦ C for 1 h. For ing the options ‘-r Biological -w 1E-4 -s 1E-8 -c 3’. DNA purification, iPure kit v2 (Diagenode) was used, fol- The chromatin state learning model for CD14+ mono- lowing the manufacturer’s instructions. 1% of the chromatin cytes was downloaded from the Roadmap Epigenomics input from each sample was purified by the same method. Project webpage, and chromatin state enrichments were cal- For ChIP-qPCR, samples were diluted 1/10, and 4 ␮l culated using Fisher’s exact test. and specific primers (Supplementary Table 1) were used for Public DNAse-seq and ATAC-seq bigwigs were aggre- each reaction. qRT-PCR was performed in technical trip- gated using wiggletools (37) to obtain the mean value of licates for each biological replicate, using LightCycler® several replicates. Heatmaps were generated with deeptools 480 SYBR Green Mix (Roche). The relative amount of im- (38). Nucleic Acids Research, 2022, Vol. 50, No. 19 10985 Proportional Venn diagrams were generated with the Gene Ontology enrichment analysis of M1 DMP- Meta-Chart webpage (https://www.meta-chart.com/). associated genes revealed categories related to positive reg- ulation of myeloid differentiation, regulation of JAK ac- tivation, regulation of defense response to virus, and vi- RESULTS tamin transport, among others. In contrast, M2 DMPs were enriched in terms related to LPS response and im- Vitamin C drastically enhances DNA demethylation during mune activation such as cellular response to molecules monocyte to dendritic cell differentiation and maturation of bacterial origin, leukocyte activation, response to bac- MOs isolated from peripheral blood of healthy donors were terium, regulation of IFN␥ production, and positive reg- differentiated in vitro to DCs for 7 days using GM-CSF ulation of NF-kB activity (Figure 1F). These functions and IL-4, in the presence or absence of vitamin C (vitC). are consistent with the respective association of M1 and Samples were collected on day 2, in the middle of the dif- M2 clusters with the DC differentiation and maturation ferentiation (designated as Ø/ØvitC given their incomplete steps. differentiation), and on day 7, including immature DCs Active demethylation in myeloid cells is often mediated (iDCs) (iDC/iDCvitC ), without further treatment, and ma- by transcription factors that recruit specific epigenetic en- ture DCs, exposed the last 48 h to lipopolysaccharide (LPS) zymes. In this regard, M1 DMPs were enriched in the con- (mDC/mDCvitC ) (Figure 1A). sensus binding motifs of transcription factors previously re- DNA methylation was profiled using Illumina Infinium lated to DC differentiation, such as EGR2(11), STAT6(7), MethylEPIC arrays, which covered 831 421 GpG sites and PU.1(11), whereas M2 DMPs were enriched in the con- in the human genome, after the quality control. First, sensus binding motifs of NF-␬B, AP-1 and IRF (Figure overall changes in DNA methylation were calculated be- 1G). Employing the average signal of public MO DNAse- tween groups pairwise (Supplementary Figure 1A). DNA seq (Blueprint database) (39) and iDC ATAC-seq triplicates demethylation was the most prominent change, as pre- (40), we found that M1 and M2 DMPs present low accessi- viously described (7). We then compared the demethy- bility in MOs (Supplementary Figure 1E). We also observed lated positions in MO-to-iDC differentiation, in compari- that M1 DMPs have greater accessibility than M2 DMPs in son with MO-to-iDCvitC differentiation, as well as iDC-to- iDCs (Supplementary Figure 1F). mDC maturation in comparison with iDCvitC -to-mDCvitC maturation. As it can be observed, the positions demethy- Vitamin C drives gene expression remodeling in dendritic cells lated in the differentiation and maturation processes in the presence of vitamin C include the majority of CpGs Given the extensive differences in DNA methylation medi- demethylated in the regular differentiation and maturation ated by vitamin C, we then performed RNA-seq of MOs, of DCs, as well as a vast number of additional CpGs (Figure Ø, ØvitC , iDC, iDCvitC , mDC and mDCvitC and checked for 1B). differences in their transcriptomes. In contrast with DNA Principal component analysis (PCA) revealed that on day methylation, transcriptome variance of principal compo- 2, most DNA methylation variance of the MO-to-iDC dif- nent (PC)1 and PC2 are mainly explained by the maturation ferentiation had already developed (Figure 1C), whereas of DCs and the differentiation of MO to DC, respectively. no differences with the vitamin C stimulus were found. In However, differences between iDC and iDCvitC and mDC contrast, on day 7, vast differences were observed for both and mDCvitC can also be observed in the PCA (Figure 2A). iDCvitC and mDCvitC , in comparison with their correspond- We first checked the expression patterns of the transcrip- ing controls without vitamin C suggesting that the vitamin tion factors related to the motifs enriched in the M1 and C-mediated boost in demethylation occurs later in time. The M2 DMPs (Figure 1G). We could observe that most of variable that explains most of the variance in DNA methy- these transcription factors are more expressed than the me- lation resides in the presence/absence of vitamin C during dian gene expression in the dataset, with the exception of differentiation. WT (WT1), SPIB and Nur77 (NR4A1) which are lowly ex- All differentially methylated positions (DMPs) associ- pressed in dendritic cells (Supplementary Figure 2A). ated with vitamin C (iDC versus iDCvitC and mDC ver- Since samples were collected on day 2 (Ø/ØvitC ), and sus mDCvitC ) were represented together, revealing two clus- on day 7, without (iDC/iDCvitC ) or with (mDC/mDCvitC ) ters of DMPs (M1 and M2) (Figure 1D and Supplementary LPS-mediated activation, three potential comparisons of Table 2). M1 corresponds with CpGs demethylated during vitamin C-treated cells can be performed, in relation to differentiation in the presence of vitamin C whereas M2 their respective controls. On day 2, very few differences are CpGs demethylated during LPS-mediated maturation were found between Ø and ØvitC (63 downregulated and in the presence of vitamin C. Both clusters were enriched 75 upregulated genes). On day 7, we found 163 downregu- in monocytic enhancers and regions flanking active tran- lated and 159 upregulated genes between iDC and iDCvitC , scription start sites (Figure 1E). That corresponds with pre- whereas most differences were found between mDC and dominant localization in intergenic regions, far from CpG mDCvitC (185 downregulated and 772 upregulated genes) islands (Supplementary Figure 1B and Supplementary Fig- (Figure 2B, Supplementary Table 3). Furthermore, most ure 1C). Of note, M1 and M2 DMPs are located in regions differentially expressed genes (DEGs) were not shared be- with subtle increases in H3K27ac and H3K4me1 histone tween comparisons (Figure 2C). marks from MOs to iDCs and mDCs, respectively. This sug- We then joined the differentially expressed genes of the gests that these regions are primed for activation, even in the three comparisons. We used hierarchical clustering and the absence of vitamin C (Supplementary Figure 1D). elbow method to decide an appropriate number of clusters 10986 Nucleic Acids Research, 2022, Vol. 50, No. 19 Figure 1. Vitamin C-mediated dendritic cell DNA methylome remodeling. (A) Scheme depicting the experimental setup. Monocytes (MO) were differenti- ated to dendritic cells (DCs) using GM-CSF and IL-4, in the presence or absence of vitamin C (vitC). Samples were collected on day 2, in the middle of the differentiation, and on day 7, including immature DCs (iDCs) and mature DCs (mDCs), exposed the last 2 days to lipopolysaccharide (LPS). (B) Area- proportional Venn diagrams comparing the demethylated CpG sets of MO-to-iDC versus MO-to-iDCvitC , and iDC-to-mDC versus iDCvitC -to-mDCvitC transitions. (C) Principal Component Analysis (PCA) of differentially methylated positions (DMPs) comparing all groups pairwise. Principal component 1 and principal component 2 are represented on the x- and y-axis, respectively. Solid and dashed lines represent the trajectories across MO-iDC-mDC and MO-iDCvitC -mDCvitC differentiation/maturation respectively, to make the plot easier to follow. (D) DNA methylation heatmap of DMPs comparing iDC to iDCvitC , and mDC to mDCvitC (␤ ≥ 0.3, FDR < 0.05). Scaled ␤-values are shown (lower DNA methylation levels in blue and higher methylation levels in red). On the right side, violin plots of clusters M1 and M2 depict scaled ␤-values. (E) Enrichment of M1 and M2 DMPs in ChromHMM 15-states categories of MOs (Roadmap Epigenomics Project). Fisher’s exact tests of M1 and M2 DMPs were calculated using all the CpGs annotated in the EPIC array as background. Significantly enriched categories (FDR < 0.05 and odds ratio > 2) are depicted with a black stroke, including TxFlnk (Flanking Active TSS), TxWk (Weak Transcription), EnhG (Genic Enhancers), and Enh (Enhancers). (F) GO (Gene Ontology) over-represented categories in M1 and M2 DMPs. Fold Change in comparison with background (EPIC array CpGs) and –log10 (FDR) is represented. (G) Bubble scatterplot of transcription factor binding motif enrichment for M1 and M2 DMPs. The x-axis shows the percentage of windows containing the motif and the y-axis shows the fold enrichment of the motif over the EPIC background. Bubbles are colored according to the transcription factor family. FDR is indicated by bubble size. Nucleic Acids Research, 2022, Vol. 50, No. 19 10987 Figure 2. Shifting in gene expression of dendritic cells triggered by vitamin C. (A) Principal Component Analysis (PCA) of gene expression. Principal component 1 and principal component 2 are represented on the x- and y-axis, respectively. Solid and dashed lines represent the trajectories across MO- iDC-mDC and MO-iDCvitC -mDCvitC differentiation/maturation respectively, to make the plot easier to follow. (B) Volcano plots of gene expression in the iDC vs. iDCvitC and the mDC versus mDCvitC comparisons. The binary logarithm of the fold change is represented on the x-axis, whereas the negative decimal logarithm of the FDR is represented on the y-axis. Downregulated genes are shown in blue (FDR < 0.05, Fold Change < –2) and upregulated genes are shown in orange (FDR < 0.05, Fold Change > 2). (C) Area-proportional Venn diagrams comparing the upregulated and downregulated gene sets of the Ø versus ØvitC , iDC versus iDCvitC and mDC vs. mDCvitC comparisons. (D) Gene expression heatmap of differentially expressed genes comparing Ø to ØvitC , iDC to iDCvitC , and mDC to mDCvitC (absolute log FC > 1, FDR < 0.05). Scaled expression VST values are shown (lower expression levels in green and higher expression values in orange). The division of the dendrogram created with the ward.D2 agglomeration method yielded six different expression clusters (E1–E6). (E) Temporal progression of gene expression of the expression clusters (E1–E6) during the differentiation process with (orange) or without (blue) vitamin C. The y-axis shows VST values, where a higher value means a higher gene expression, and the line ribbons represent the 95% confidence interval. (F) Gene Ontology (GO) over-representation of GO Biological Process categories in the E1–E6 clusters. Fold Change of genes over the background and –log10 (FDR) of the Fisher’s exact tests are shown. Significant categories (FDR < 0.01) are depicted with a black stroke. (Supplementary Figure 2B). We then divided the genes into and response to calcium ion, and E5 in type I interferon sig- six clusters (expression clusters E1-E6) with different behav- naling pathway and defense response to virus. iors (Figure 2D and Supplementary Table 4). E1, E2 and E3 clusters show a diminished downregulation trend in the iDCvitC to mDCvitC transition, in comparison with the iDC Vitamin C-mediated demethylation is linked to increased to mDC transition, whereas E5 genes are more upregulated gene expression during dendritic cell maturation in the iDCvitC to mDCvitC transition in comparison with the To identify potential functional effects of the vitamin C- iDC to mDC transition (Figure 2E). mediated demethylation, we linked each DMP with its clos- Gene Ontology Enriched terms were calculated for each est gene. When the global profile of M1 and M2 associ- expression cluster, obtaining distinctive categories (Figure ated genes was intersected with DEGs from Figure 2D, we 2F). For instance, E1 was enriched in dendritic cell migra- found that M1 associated genes are less downregulated in tion and chemotaxis, E2 in macrophage activation and cy- mDCs with vitamin C treatment, whereas M2 associated tokine secretion, E3 in antigen processing and presentation genes are upregulated in mDCs with vitamin C treatment 10988 Nucleic Acids Research, 2022, Vol. 50, No. 19 (Figure 3A). Concordantly, E1 and E2 clusters are enriched myeloid cells whose activity is enhanced by vitamin C. The in M1-associated genes, and the E5 cluster is enriched in co-immunoprecipitation of p65 revealed its interaction with M2-associated genes (Figure 3B). TET2 in both mDCs and mDCvitC (Figure 4G). Individual DMPs and their associated DEGs illustrate Additionally, we performed a ChIP-qPCR analysis of different relationships between DNA methylation/gene ex- p65 on day 6, comparing the binding in a negative control pression. M1-associated genes, C1QB and CD1C, which are amplicon around the INS gene with an amplicon around related to the complement system and antigen presentation the cg08639424, located close to LTA and TNF genes. p65 respectively, show demethylation with vitamin C treatment was significantly enriched in mDCs/mDCvitC , but not in in iDCs, conjoined with a lower reduction in expression iDCs/iDCvitC (Figure 4H). after activation (Figure 3C). Furthermore, M2-associated Since the LTA gene, which encodes TNF␤, was found up- genes such as LTA, IKBKE and IRF8 related to TNF␤ pro- regulated and the adjacent CpGs become demethylated in duction, NF-␬B pathway regulation, and interferon regu- mDCvitC , we measured the TNF␤ protein levels secreted by lation respectively, depict a decreased methylation in mDC these cells in comparison with mDCs. mDCs produced little with vitamin C, concomitant with increased gene expression amounts of TNF␤ whereas mDCvitC supernatant contained (Figure 3D). considerably higher concentrations (Figure 4I). Moreover, Finally, a database of transcription factor regulation (41) Bay treatment damped TNF␤ production of mDCvitC , con- was used to infer potential transcription factors involved in sistent with the reversion in the cg08639424 demethylation regulating expression clusters (Figure 3E and Supplemen- and LTA upregulation (Supplementary Figure 4F). tary Figure 3A). Interestingly, PU.1 and RELA (p65) were associated with M1/E2 and M2/E5 clusters, respectively. Vitamin C produces dendritic cells with higher T cell stimu- lation capabilities NF-␬B/p65 orchestrates vitamin C-mediated DNA demethy- We then characterized the mDCvitC phenotype in compar- lation, gene upregulation and increased proinflammatory cy- ison with mDCs. In particular, we studied the T cell stim- tokine production ulation capabilities and antigen presentation of mDCvitC Since p65 was associated with both DNA demethylation in contrast to mDCs. First, we performed a clonal expan- (Figure 1G) and gene upregulation (Figure 3E) during sion of T cells from healthy donors with a SARS-CoV-2 mDCvitC maturation, we studied the protein expression and mix of antigens. After the process, we verified that 95% phosphorylation by Western Blot. First, we found that p65 of the resulting cells were T lymphocytes by staining them presents similar protein levels in iDCs, mDCs, iDCvitC and with anti-CD4 and anti-CD8 antibodies (Supplementary mDCvitC . However, phosphorylated p65 (Ser536) (p-p65) is Figure 5A and Supplementary Figure 5B). We then differ- increased in both mDCs and mDCvitC (Figure 4A and Sup- entiated autologous MOs in vitro to iDCvitC /iDCs. After plementary Figure 4A). Moreover, we also detected p-p65 in 48 h of maturation with LPS and 1 h of antigen loading the nuclear fraction (NF) of mDCs and mDCvitC , enabling with the same SARS-CoV-2 antigen mix, mDCvitC /mDCs it to act as a transcription factor (Figure 4B and Supple- were cocultured with Carboxyfluorescein succinimidyl ester mentary Figure 4A). (CFSE)-stained clonal T cells for 5 days (Figure 5A). To further explore the role of NF-␬B/p65 in the mDCvitC In Figure 5B, a selected example of this T cell prolifera- transcriptomic and epigenomic reprogramming, we uti- tion assay is shown. The histogram of the CFSE signal of T lized a chemical inhibitor of I␬B degradation (BAY 11- cells alone (C- ) or cocultured with mDC/mDCvitC loaded 7082, Bay) which reduces the nuclear translocation of p65 with a control antigen or with the specific set of antigens (42,43) (Supplementary Figure 4B). MOs were differenti- (An) is depicted. ated in DCs as previously described, adding Bay (10 ␮M) Overall, we observed a significant increase in T cell pro- or Dimethyl Sulfoxide (DMSO) on day 5 (Figure 4C). liferation when they are cocultured with mDCvitC loaded We then tested the effect of NF-␬B inhibition on M1 and with a specific set of antigens, in comparison with mDCs M2 DMPs. We observed that the NF-␬B inhibitor dampens loaded with the same antigens. This can be observed by the the demethylation of M2 DMPs, concordantly with their increase in the proliferation percentages and the decrease in enrichment in NF-␬B motifs. However, this effect is smaller the median intensity of fluorescence (MFI) of CFSE (Fig- in M1 DMPs (Figure 4D). Specifically, NF-␬B inhibition ure 5C and D). However, when T cells were cocultured with pervasively prevents the demethylation of most M2 DMPs mDCs/mDCvitC loaded with a control antigen, no differ- (P < 1-1 ) (Figure 4E). As a validation, we performed pyrose- ences in proliferation or MFI were found, indicating that quencing of selected DMPs from the M2 cluster, observing the mDCvitC increased T cell activation capabilities rely on the blockage of demethylation. Conversely, a DMP from the specific antigen presentation and not only indirect stimula- M1 cluster was not affected (Supplementary Figure 4C). tion (Figure 5C, D). Secondly, we also tested the expression of M2-associated genes revealing that, inversely to DNA demethylation, gene DISCUSSION expression decreases in mDCvitC when NF-␬B is inhibited (Figure 4F and Supplementary Figure 4D). However, the In this work, we demonstrate a substantial effect of vita- same trend was not found in M1-associated genes (Figure min C supplementation during the MO-to-DC in vitro dif- 4F and Supplementary Figure 4E). ferentiation and maturation. First, we show vast demethy- We then checked the potential interaction between p65 lation in DCs treated with vitamin C, consistent with its role and TET2, a key mediator of active demethylation in as a TET enzyme cofactor, being the number of additional Nucleic Acids Research, 2022, Vol. 50, No. 19 10989 Figure 3. Integration of DNA methylation and gene expression vitamin C-mediated remodeling. (A) M1 and M2 DMPs (from Figure 1D) (␤ ≥ 0.3, FDR < 0.05), were associated with the nearest transcription start site. Temporal progression of gene expression of the M1- and M2-associated genes intersected with vitamin C-mediated differentially expressed genes (Figure 2D) (absolute log FC > 1, FDR < 0.05) with (orange) or without (blue) vitamin C is represented. The y-axis shows VST values, where a higher value means a higher gene expression, and the line ribbons represent the 95% confidence interval. (B) Enrichments of M1- and M2-associated genes over the E1–E6 expression clusters were calculated using Fisher’s exact tests. Odds ratios ±95% confidence intervals are shown. Significant enrichments (FDR < 0.01) are shown in black. (C) Selected examples of M1 DMPs (␤ ≥ 0.3, FDR < 0.05) and associated genes (absolute log FC > 1, FDR < 0.05). Temporal progression of DNA methylation (␤-value) (below) and gene expression (VST) (above) during the differentiation process with (orange) or without (blue) vitamin C is depicted. (D) Selected examples of M2 DMPs and associated genes. Temporal progression of DNA methylation (␤-value) (below) and gene expression (VST) (above) during the differentiation process with (orange) or without (blue) vitamin C is depicted. (E) Enrichment of clusters with gene expression/DNA methylation correlation with genesets from CheA 2016 database (REF), containing genes putatively regulated by transcription factors. The odds ratio over the background and -log10(FDR) of the Fisher’s exact tests are shown. demethylated CpGs higher than the MO-to-DC demethy- tion should occur through an active mechanism, consistent lation without vitamin C. This epigenomic remodeling cor- with previous studies, and be catalyzed by TET enzymes, relates with increased expression of genes related to anti- the activity of which is enhanced by vitamin C. We and gen presentation, cytokine secretion, and immune response. others have shown that TET2, the most expressed TET en- Moreover, our analysis indicates that NF-kB is directly in- zyme in MOs, is the major driver of active demethylation volved in the epigenomic and transcriptomic reprogram- in this biological context (7,10,46). In this study, we show ming observed during the maturation of DCs in the pres- the physical interaction between TET2 and p65, one of the ence of vitamin C, together with the increase in TNF␤ pro- NF-kB subunits, that is associated with the CpGs that be- duction. Finally, vitamin C enhances the capacity of DCs to come demethylated during DC activation in the presence induce the proliferation of autologous T cells through spe- of vitamin C. However, we cannot discard the participation cific antigen presentation. of other TET enzymes in the demethylation process, whose Vitamin C is a well-established cofactor of TET pro- enzymatic activity could also be enhanced by vitamin C. In teins, with the ability to enhance TET-mediated oxidation fact, TET3 could compensate for TET2 activity during MO of 5-methylcytosine (5mC) into 5-hydroxymethylcytosine differentiation (11). (5hmC) and further oxidized methylcytosine derivatives Of note, the observed effects of vitamin C on DNA (44,45). Given the absence of proliferation in the MO-to- methylation occur in the last days of the differentiation DC differentiation process (13), the observed demethyla- process, since no DNA methylation differences were found 10990 Nucleic Acids Research, 2022, Vol. 50, No. 19 Figure 4. Role of p65 in the transcriptomic and epigenomic remodeling of vitamin C-treated dendritic cells. (A) Western blot of phosphorylated p65 (p- p65) (Ser536) and total p65 in whole-cell lysates. ␤-actin was used as a loading control. (B) Western blot of p-p65 in the nuclear (NF) and cytoplasmic fractions (CF). GAPDH and histone H1 proteins were used as loading control for cytoplasms and nuclei, respectively. (C) Scheme depicting the inhi- bition of p65. Monocytes (MO) were differentiated to dendritic cells (DCs) using GM-CSF and IL-4, in the presence or absence of vitamin C (vitC). On day 5, lipopolysaccharide (LPS) and BAY 11–7082 (Bay) or equivalent amounts of diluent were added to the cell culture. On day 7, iDC/iDCvitC , iDC + Bay/iDCvitC + Bay, mDC/mDCvitC and mDC + Bay/mDCvitC + Bay were obtained. (D) Scatter plots of M1 and M2 DMPs showing DNA methy- lation (␤-values) of Bay-treated (y-axis) versus untreated mDCs/mDCvitC (x-axis). Contour lines represent a 2D kernel density estimation. Labels with selected examples indicate the DMPs associated with these previously shown genes (Figure 3C, D). (E) Methylated CpG set enrichment analysis (mCSEA) of mDCvitC versus mDCvitC + Bay, using M2 DMPs as CpG set. The running enrichment score is represented and the normalized enrichment score (NES) and FDR are shown above. (F) The average gene expression of C1QB (M1-associated gene) and M2-associated genes (LTA, IKBKE and IRF8) obtained with RT-qPCR are represented with points, and the black lines indicate the standard error of the mean. P-values of paired t-tests are shown (n = 6) (ns: P > 0.05, *P < 0.05, **P < 0.01). (G) Western blot of the co-immunoprecipitation of p65, showing the signal of p65 and TET2 proteins. Below, the image signal quantifications of three independent western blots are shown for each protein. P-values of paired t-tests are shown (n = 3) (mean ± standard error of the mean) (*P < 0.05, **P < 0.01). (H) Chromatin Immunoprecipitation (ChIP) signal of p65 binding to a negative control locus (INS) and around an M2 DMPs (cg08639424) associated with LTA and TNF. The RT-qPCR signal relative to the ChIP input is shown (n = 3). P-values of t-tests are shown (n = 3) (mean ± standard error of the mean) (ns: P > 0.05, *P < 0.05). (I) TNF␤ production of mDCs and mDCvitC , after 5 days of differentiation and 48h of maturation with LPS. The P-value of a Wilcoxon rank-sum test is shown (n = 3) Nucleic Acids Research, 2022, Vol. 50, No. 19 10991 Figure 5. Functional and phenotypic alterations of vitamin C-treated dendritic cells. (A) Scheme depicting the T cell proliferation assay. PBMCs were obtained from healthy donors. T cell clones reacting to the specific antigen (SARS-CoV-2 protein S) were selected through several rounds of clonal expan- sion. On the other hand, mDCs and mDCvitC were obtained from the same donor and charged with the specific antigen. Finally, the Carboxyfluorescein succinimidyl ester(CFSE)-stained T cells were cocultured with mDCs/mDCvitC for 5 days. (B) Selected example of a histogram of CFSE signal from T cells, alone without stimulation (C− ) or cocultured with mDC/mDCvitC treated with a control antigen or a specific set of antigens of SARS-CoV-2 (An). When T cells proliferate, the CFSE signal is diminished. The statistical analysis with all the replicates can be observed in panels C and D. (C) The proliferation of T cells cocultured with mDC/mDCvitC (2:1 proportion), with or without the loading with a specific set of antigens of SARS-CoV-2. Negative control from each donor was used to calculate the proliferation percentage. P-values from two-tailed paired t-tests are shown (n = 6) (ns: P > 0.05, ** P < 0.01). (D) Median Fluorescence Intensity (MFI) of T cells alone (C- ), cocultured with mDC/mDCvitC (2:1 proportion), with or without the loading with a specific set of antigens of SARS-CoV-2, or stimulated with CD3/CD28 activation beads (C+ ). P-values from two-tailed paired t-tests are shown (n = 6) (ns: P > 0.05, * P < 0.05). with vitamin C treatment at day 2 of differentiation. In rect CpG-binding of the other TETs (TET1 and TET3). In contrast, most variance in methylation occurring in the this respect, its targeting depends on transcription factors MO-to-iDC transition occurs before day 2 suggesting that, or their co-factors for locus-specific recruitment (11,48). In without vitamin C, the function of TET enzymes in this addition, the effect of vitamin C of preexisting active sig- model is progressively diminished. Alternative reducing naling pathways is highlighted by the fact that M1 and M2 agents present in the culture medium such as glutathione DMPs are enriched in regions with increasing active chro- are less efficient as TET cofactors (45,47). Thus, the pro- matin marks (H3K27ac and H3K4me1) in the MO-to-iDC gressive oxidation of these reducing agents may explain the and the iDC-to-mDC transitions, respectively (Supplemen- impairment of TET function over time in the absence of tary Figure 1D). We cannot rule out that vitamin C pro- vitamin C. duces changes in histone modifications or DNA accessibil- Vitamin C-mediated demethylation occurs during the dif- ity that may contribute to the phenotype. In this regard, ferentiation (M1 cluster) or the maturation of DCs (M2 vitamin C also increases the enzymatic activity of histone cluster). Interestingly, the sets of transcription factor mo- demethylases (15). tifs enriched in the M1 and M2 clusters are equivalent to Since vitamin C can potentiate demethylation triggered the transcription factors involved in DC differentiation and by preexisting pathways, we hypothesize that this treatment LPS-mediated signaling, respectively. This suggests that, may have diverse effects depending on the specific context overall, vitamin C does not promote the recruitment of TET and active transcription factors. In this regard, vitamin C enzymes through new pathways but boosts the demethy- could potentiate demethylation in other models of mono- lation triggered by preexisting active signaling pathways. cyte differentiation to DCs with clinical use, such as tolero- In fact, TET2 lacks the CXXC domain that enables di- genic DCs (12,13). 10992 Nucleic Acids Research, 2022, Vol. 50, No. 19 The functional relationship between DNA demethylation mainly related to differentiation and activation might con- and gene expression has been extensively studied in several fer a better outcome. biological contexts (7,9,12,13). Genes more expressed in Therapies based on autologous DCs (DC vaccines) has mDCvitC than mDC, from E1 and E2 clusters are enriched been extensively investigated, with >200 completed clini- in M1-associated genes. This establishes a clear temporal re- cal trials to date (19). Most efforts have been focused on lationship, suggesting that prior demethylation could pro- cancer, but some clinical trials have also been initiated to tect some genes from downregulation during DC matura- treat infectious diseases such as COVID19 (NCT04685603, tion. On the other hand, genes from the E5 cluster are en- NCT05007496) (58). The use of moDCs differentiated ex riched in M2-associated genes. In this case, demethylation is vivo from monocytes of the same donor is a common and occurring at the same step as upregulation. Then, we cannot straightforward approach to generating DC vaccines, given discern whether demethylation or upregulation occurs first. the relatively high abundance of these cells in the human However, since the primary mechanism of action of vitamin blood. However, the lower antigen presentation capabilities C through DNA and histone demethylation is well known, of moDCs in comparison with blood DCs is a bottleneck we hypothesize that epigenetic modifications could mediate for the efficacy of these treatments (59). gene upregulation in mDCvitC . This assumption starts from Here, we show that mDCvitC loaded with SARS-CoV- a different point than other works that established DNA 2 antigens can stimulate the proliferation of autologous demethylation during DC maturation as a consequence of antigen-specific T cells more efficiently than mDCs. This gene upregulation because in that case, the differential stim- indicates that vitamin C induces an increase of the anti- ulus is live bacteria, that can activate a plethora of signaling gen presentation capabilities of mDCs in an antigen-specific pathways not necessarily linked directly to DNA demethy- fashion, and could be a promising strategy for generat- lation (9). ing DC vaccines towards specific tumor antigens. These re- NF-␬B mediates both demethylation and upregulation sults can lead to the generation of new in vitro protocols during the maturation of DCs in the presence of vitamin for the generation of moDC vaccines with higher perfor- C (Figure 4D and E). Since that pathway is associated with mance. However, multiple considerations should be taken toll-like receptor 4 (TLR4), which recognizes LPS, it is not into account to consider the viability of these treatments surprising that it plays a role in the process. Intriguingly, in the cancer context, including the survival of these cells this factor is present in the nucleus in mDCs and mDCvitC , and the maintenance of their enhanced immunogenic re- interacts with TET2, and binds in both cases to CpGs that sponses. Further works using animal models and human in become demethylated only in mDCvitC . We postulate that vivo moDCs from patients treated with high doses of vita- NF-kB genomic binding in mDCs and mDCvitC is proba- min C should shed light on the specific clinical implications bly similar, but vitamin C potentiates TET2 function allow- of these insights. ing the demethylation of genomic loci, which may condition the expression of the associated genes. Moreover, NF-␬B DATA AVAILABILITY also drives vitamin C-mediated production of TNF␤. This cytokine can signal through tumor necrosis factor recep- All DNA methylation and expression datasets for this pub- tor (TNFR)I and TNFRII to activate the NF-␬B pathway lication have been deposited in the NCBI Gene Expression in both DCs and T cells and has shown anti-carcinogenic Omnibus and are accessible through GEO SuperSeries ac- properties in animal models (49–51). cession number GSE203463. Linus Pauling proposed vitamin C as a potential cancer treatment >40 years ago, but the negative results of fur- SUPPLEMENTARY DATA ther clinical trials diminished the enthusiasm (52,53). How- ever, during the last few years, increasing interest has arisen Supplementary Data are available at NAR Online. around vitamin C as a treatment or adjuvant for several types of cancer. For instance, intravenous vitamin C treat- ACKNOWLEDGEMENTS ment in mice abrogates cancer progression through direct TET2 function restoration in cancer cells (17). Moreover, We thank CERCA Programme/Generalitat de Catalunya clinical remission following vitamin C treatment was found and the Josep Carreras Foundation for institutional sup- in a case of acute myeloid leukemia with mutations in TET2 port. (54). Furthermore, in mice models of different types of can- Author contributions: O.M.-P. and E.B. conceived and de- cer, a fully competent immune system was required to max- signed the study; O.M.-P., G.G.-T., J.C.-S., L.C., E.M.- imize the antiproliferative effects of vitamin C, suggesting C. and J.L.S. performed experiments; O.M.-P. performed an effect of that molecule in the modulation of the immune the bioinformatic analyses; O.M.-P. and E.B. wrote the system (18). Vitamin C has already been reported to play a manuscript; all authors participated in discussions and in- role in actively demethylating CNS2 in iTregs by activating terpreting the results. TET2. Vitamin C treated iTregs could maintain high levels of Foxp3 (55–57). However, those iTregs failed to survive FUNDING and maintain proper immune responses in complex GvHD models because CpG islands of non-targeted genes, proba- E.B. was funded by the Spanish Ministry of Sci- bly pro-apoptotic genes, were also demethylated. Thus, clin- ence and Innovation [PID2020-117212RB-I00, ical use of those iTregs is impractical. In this regard, in the AEI/10.13039/501100011033] and Instituto de Salud context of DCs the observation that the implicated TFs are Carlos III (ISCIII), Ref. AC18/00057, associated with Nucleic Acids Research, 2022, Vol. 50, No. 19 10993 i-PAD project (ERARE European Union program); J.L.S. epigenome remodeling in dendritic cells. Nucleic Acids Res., 50, was funded by Instituto de Salud Carlos III through 108–126. 13. Català-Moll,F., Ferreté-Bonastre,A.G., Godoy-Tena,G., the project CP19/00176 (co-funded by European So- Morante-Palacios,O., Ciudad,L., Barberà,L., Fondelli,F., cial Fund, ‘Investing in your future’); Spanish Ministry Martı́nez-Cáceres,E.M., Rodrı́guez-Ubreva,J., Li,T. et al. (2022) of Science, Innovation and Universities (MICINN) Vitamin d receptor, STAT3, and TET2 cooperate to establish [PID2019-111243RA-I00/AEI/10, 13039/501100011033]; tolerogenesis. Cell Rep., 38, 110244. E.M.C. is funded by project PI20/01313, integrated in 14. Levine,M., Wang,Y., Padayatty,S.J. and Morrow,J. (2001) A new recommended dietary allowance of vitamin c for healthy young the Plan Nacional de I + D + I and co-supported by women. Proc. Natl. Acad. Sci. U.S.A., 98, 9842–9846. the ISCIII-Subdirección General de Evaluación and 15. Ang,A., Pullar,J.M., Currie,M.J. and Vissers,M.C.M. (2018) Vitamin the Fondo Europeo de Desarrollo Regional (FEDER); c and immune cell function in inflammation and cancer. Biochem. O.M.-P. holds an i-PFIS PhD fellowship [IFI17/00034] Soc. Trans., 46, 1147–1159. 16. Jeong,Y.-J., Kim,J.-H., Hong,J.-M., Kang,J.S., Kim,H.-R., Lee,W.J. from Acción Estratégica en Salud 2013–2016 ISCIII, and Hwang,Y. (2014) Vitamin c treatment of mouse bone co-financed by Fondo Social Europeo. Funding for open marrow-derived dendritic cells enhanced CD8(+) memory t cell access charge: Spanish Ministry of Science and Innovation production capacity of these cells in vivo. Immunobiology, 219, [PID2020-117212RB-I00, AEI/10.13039/501100011033]. 554–564. Conflict of interest statement. None declared. 17. Cimmino,L., Dolgalev,I., Wang,Y., Yoshimi,A., Martin,G.H., Wang,J., Ng,V., Xia,B., Witkowski,M.T., Mitchell-Flack,M. et al. (2017) Restoration of TET2 function blocks aberrant self-renewal and leukemia progression. Cell, 170, 1079–1095. 18. Magrı̀,A., Germano,G., Lorenzato,A., Lamba,S., Chilà,R., REFERENCES Montone,M., Amodio,V., Ceruti,T., Sassi,F., Arena,S. et al. (2020) 1. Morante-Palacios,O., Fondelli,F., Ballestar,E. and High-dose vitamin c enhances cancer immunotherapy. Sci. Transl. Martı́nez-Cáceres,E.M. (2021) Tolerogenic dendritic cells in Med., 12, eaay8707. autoimmunity and inflammatory diseases. Trends Immunol., 42, 19. Wculek,S.K., Cueto,F.J., Mujal,A.M., Melero,I., Krummel,M.F. and 59–75. Sancho,D. (2020) Dendritic cells in cancer immunology and 2. Sallusto,F. and Lanzavecchia,A. (1994) Efficient presentation of immunotherapy. Nat. Rev. Immunol., 20, 7–24. soluble antigen by cultured human dendritic cells is maintained by 20. Schadendorf,D., Ugurel,S., Schuler-Thurner,B., Nestle,F.O., Enk,A., granulocyte/macrophage colony-stimulating factor plus interleukin 4 Bröcker,E.-B., Grabbe,S., Rittgen,W., Edler,L., Sucker,A. et al. and downregulated by tumor necrosis factor alpha. J. Exp. Med., 179, (2006) Dacarbazine (DTIC) versus vaccination with autologous 1109–1118. peptide-pulsed dendritic cells (DC) in first-line treatment of patients 3. Collin,M. and Bigley,V. (2018) Human dendritic cell subsets: an with metastatic melanoma: a randomized phase III trial of the DC update. Immunology, 154, 3–20. study group of the DeCOG. Ann. Oncol., 17, 563–570. 4. Coillard,A. and Segura,E. (2019) In vivo differentiation of human 21. Liau,L.M., Ashkan,K., Tran,D.D., Campian,J.L., Trusheim,J.E., monocytes. Front. Immunol., 10, 1907. Cobbs,C.S., Heth,J.A., Salacz,M., Taylor,S., D’Andre,S.D. et al. 5. Rodriguez,R.M., Suarez-Alvarez,B., Lavı́n,J.L., Ascensión,A.M., (2018) First results on survival from a large phase 3 clinical trial of an Gonzalez,M., Lozano,J.J., Raneros,A.B., Bulnes,P.D., autologous dendritic cell vaccine in newly diagnosed glioblastoma. J. Vidal-Castiñeira,J.R., Huidobro,C. et al. (2019) Signal integration Transl. Med., 16, 142. and transcriptional regulation of the inflammatory response mediated 22. Godoy-Tena,G. and Ballestar,E. (2022) Epigenetics of dendritic cells by the GM-/M-CSF signaling axis in human monocytes. Cell Rep., in tumor immunology. Cancers (Basel), 14, 1179. 29, 860–872. 23. Morante-Palacios,O. and Ballestar,E. (2021) shinyÉPICo: a graphical 6. de la Rica,L., Rodrı́guez-Ubreva,J., Garcı́a,M., Islam,A.B.M.M.K., pipeline to analyze illumina DNA methylation arrays. Bioinformatics, Urquiza,J.M., Hernando,H., Christensen,J., Helin,K., 37, 257–259. Gómez-Vaquero,C. and Ballestar,E. (2013) PU.1 target genes 24. Aryee,M.J., Jaffe,A.E., Corrada-Bravo,H., Ladd-Acosta,C., undergo Tet2-coupled demethylation and DNMT3b-mediated Feinberg,A.P., Hansen,K.D. and Irizarry,R.A. (2014) Minfi: a flexible methylation in monocyte-to-osteoclast differentiation. Genome Biol., and comprehensive bioconductor package for the analysis of infinium 14, R99. DNA methylation microarrays. Bioinformatics, 30, 1363–1369. 7. Vento-Tormo,R., Company,C., Rodrı́guez-Ubreva,J., de la Rica,L., 25. Ritchie,M.E., Phipson,B., Wu,D., Hu,Y., Law,C.W., Shi,W. and Urquiza,J.M., Javierre,B.M., Sabarinathan,R., Luque,A., Esteller,M., Smyth,G.K. (2015) limma powers differential expression analyses for Aran,J.M. et al. (2016) IL-4 orchestrates STAT6-mediated DNA RNA-sequencing and microarray studies. Nucleic Acids Res., 43, e47. demethylation leading to dendritic cell differentiation. Genome Biol., 26. Touleimat,N. and Tost,J. (2012) Complete pipeline for infinium(®) 17, 4. human methylation 450K beadchip data processing using subset 8. Álvarez-Errico,D., Vento-Tormo,R., Sieweke,M. and Ballestar,E. quantile normalization for accurate DNA methylation estimation. (2015) Epigenetic control of myeloid cell differentiation, identity and Epigenomics, 4, 325–341. function. Nat. Rev. Immunol., 15, 7–17. 27. Triche,T.J. Jr, Weisenberger,D.J., Van Den Berg,D., Laird,P.W. and 9. Pacis,A., Mailhot-Léonard,F., Tailleux,L., Randolph,H.E., Siegmund,K.D. (2013) Low-level processing of illumina infinium Yotova,V., Dumaine,A., Grenier,J.-C. and Barreiro,L.B. (2019) Gene DNA methylation beadarrays. Nucleic Acids Res., 41, e90. activation precedes DNA demethylation in response to infection in 28. Hinrichs,A.S., Karolchik,D., Baertsch,R., Barber,G.P., Bejerano,G., human dendritic cells. Proc. Natl. Acad. Sci. U.S.A., 116, 6938–6943. Clawson,H., Diekhans,M., Furey,T.S., Harte,R.A., Hsu,F. et al. 10. Klug,M., Schmidhofer,S., Gebhard,C., Andreesen,R. and Rehli,M. (2006) The UCSC genome browser database: update 2006. Nucleic (2013) 5-Hydroxymethylcytosine is an essential intermediate of active Acids Res., 34, D590–D598. DNA demethylation processes in primary human monocytes. 29. Kim,D., Paggi,J.M., Park,C., Bennett,C. and Salzberg,S.L. (2019) Genome Biol., 14, R46. Graph-based genome alignment and genotyping with HISAT2 and 11. Mendes,K., Schmidhofer,S., Minderjahn,J., Glatz,D., Kiesewetter,C., HISAT-genotype. Nat. Biotechnol., 37, 907–915. Raithel,J., Wimmer,J., Gebhard,C. and Rehli,M. (2021) The 30. 1000 Genome Project Data Processing Subgroup, Li,H., epigenetic pioneer EGR2 initiates DNA demethylation in Handsaker,B., Wysoker,A., Fennell,T., Ruan,J., Homer,N., Marth,G., differentiating monocytes at both stable and transient binding sites. Abecasis,G. and Durbin,R. (2009) The sequence alignment/map Nat. Commun., 12, 1556. format and SAMtools. Bioinformatics, 25, 2078–2079. 12. Morante-Palacios,O., Ciudad,L., Micheroli,R., de la 31. Liao,Y., Smyth,G.K. and Shi,W. (2014) featureCounts: an efficient Calle-Fabregat,C., Li,T., Barbisan,G., Houtman,M., Edalat,S.G., general purpose program for assigning sequence reads to genomic Frank-Bertoncelj,M., Ospelt,C. et al. (2022) Coordinated features. Bioinformatics, 30, 923–930. glucocorticoid receptor and MAFB action induces tolerogenesis and 10994 Nucleic Acids Research, 2022, Vol. 50, No. 19 32. Love,M.I., Huber,W. and Anders,S. (2014) Moderated estimation of 46. Pacis,A., Tailleux,L., Morin,A.M., Lambourne,J., MacIsaac,J.L., fold change and dispersion for RNA-seq data with DESeq2. Genome Yotova,V., Dumaine,A., Danckaert,A., Luca,F., Grenier,J.-C. et al. Biol., 15, 550. (2015) Bacterial infection remodels the DNA methylation landscape 33. Schindelin,J., Arganda-Carreras,I., Frise,E., Kaynig,V., Longair,M., of human dendritic cells. Genome Res., 25, 1801–1811. Pietzsch,T., Preibisch,S., Rueden,C., Saalfeld,S., Schmid,B. et al. 47. Blaschke,K., Ebata,K.T., Karimi,M.M., Zepeda-Martı́nez,J.A., (2012) Fiji: an open-source platform for biological-image analysis. Goyal,P., Mahapatra,S., Tam,A., Laird,D.J., Hirst,M., Rao,A. et al. Nat. Methods, 9, 676–682. (2013) Vitamin c induces Tet-dependent DNA demethylation and a 34. McLean,C.Y., Bristor,D., Hiller,M., Clarke,S.L., Schaar,B.T., blastocyst-like state in ES cells. Nature, 500, 222–226. Lowe,C.B., Wenger,A.M. and Bejerano,G. (2010) GREAT improves 48. Ravichandran,M., Jurkowska,R.Z. and Jurkowski,T.P. (2018) Target functional interpretation of cis-regulatory regions. Nat. Biotechnol., specificity of mammalian DNA methylation and demethylation 28, 495–501. machinery. Org. Biomol. Chem., 16, 1419–1435. 35. Yu,G., Wang,L.-G., Han,Y. and He,Q.-Y. (2012) clusterProfiler: an r 49. Bauer,J., Namineni,S., Reisinger,F., Zöller,J., Yuan,D. and package for comparing biological themes among gene clusters. Heikenwälder,M. (2012) Lymphotoxin, NF-B, and cancer: the dark OMICS, 16, 284–287. side of cytokines. Dig. Dis., 30, 453–468. 36. Jalili,V., Matteucci,M., Masseroli,M. and Morelli,M.J. (2015) Using 50. Browning,J.L., Miatkowski,K., Sizing,I., Griffiths,D., Zafari,M., combined evidence from replicates to evaluate chip-seq peaks. Benjamin,C.D., Meier,W. and Mackay,F. (1996) Signaling through Bioinformatics, 31, 2761–2769. the lymphotoxin beta receptor induces the death of some 37. Zerbino,D.R., Johnson,N., Juettemann,T., Wilder,S.P. and Flicek,P. adenocarcinoma tumor lines. J. Exp. Med., 183, 867–878. (2014) WiggleTools: parallel processing of large collections of 51. Lukashev,M., LePage,D., Wilson,C., Bailly,V., Garber,E., genome-wide datasets for visualization and statistical analysis. Lukashin,A., Ngam-ek,A., Zeng,W., Allaire,N., Perrin,S. et al. (2006) Bioinformatics, 30, 1008–1009. Targeting the lymphotoxin-beta receptor with agonist antibodies as a 38. Ramı́rez,F., Dündar,F., Diehl,S. and Grüning,B.A., (2014) Manke potential cancer therapy. Cancer Res., 66, 9617–9624. T.deepTools: a flexible platform for exploring deep-sequencing data. 52. Cameron,E. and Pauling,L. (1976) Supplemental ascorbate in the Nucleic Acids Res., 42, W187–91. supportive treatment of cancer: prolongation of survival times in 39. Fernández,J.M., de la Torre,V., Richardson,D., Royo,R., terminal human cancer. Proc. Natl. Acad. Sci. U.S.A., 73, 3685–3689. Puiggròs,M., Moncunill,V., Fragkogianni,S., Clarke,L. and 53. Cabanillas,F. (2010) Vitamin c and cancer: what can we BLUEPRINT ConsortiumBLUEPRINT Consortium and Flicek,P. conclude–1,609 patients and 33 years later? P. R. Health Sci. J., 29, et al. (2016) The BLUEPRINT data analysis portal. Cell Syst, 3, 215–217. 491–495. 54. Das,A.B., Kakadia,P.M., Wojcik,D., Pemberton,L., Browett,P.J., 40. Johnson,J.S., Lucas,S.Y., Amon,L.M., Skelton,S., Nazitto,R., Bohlander,S.K. and Vissers,M.C.M. (2019) Clinical remission Carbonetti,S., Sather,D.N., Littman,D.R. and Aderem,A. (2018) following ascorbate treatment in a case of acute myeloid leukemia Reshaping of the dendritic cell chromatin landscape and interferon with mutations in TET2 and WT1. Blood Cancer J., 9, 82. pathways during HIV infection. Cell Host Microbe, 23, 366–381. 55. Sasidharan Nair,V., Song,M.H. and Oh,K.I. (2016) Vitamin c 41. Lachmann,A., Xu,H., Krishnan,J., Berger,S.I., Mazloom,A.R. and facilitates demethylation of the foxp3 enhancer in a tet-dependent Ma’ayan,A. (2010) ChEA: transcription factor regulation inferred manner. J Immunol, 196, 2119–2131. from integrating genome-wide ChIP-X experiments. Bioinformatics, 56. Kasahara,H., Kondo,T., Nakatsukasa,H., Chikuma,S., Ito,M., 26, 2438–2444. Ando,M., Kurebayashi,Y., Sekiya,T., Yamada,T., Okamoto,S. et al. 42. Lee,J., Rhee,M.H., Kim,E. and Cho,J.Y. (2012) BAY 11-7082 is a (2017) Generation of allo-antigen-specific induced treg stabilized by broad-spectrum inhibitor with anti-inflammatory activity against vitamin c treatment and its application for prevention of acute graft multiple targets. Mediators Inflamm., 2012, 416036. versus host disease model. Int. Immunol., 29, 457–469. 43. de la Rica,L., Garcı́a-Gómez,A., Comet,N.R., Rodrı́guez-Ubreva,J., 57. Iamsawat,S., Tian,L., Daenthanasanmak,A., Wu,Y., Nguyen,H.D., Ciudad,L., Vento-Tormo,R., Company,C., Álvarez-Errico,D., Bastian,D. and Yu,X.Z. (2019) Vitamin c stabilizes CD8+ iTregs and Garcı́a,M., Gómez-Vaquero,C. et al. (2015) NF-␬B-direct activation enhances their therapeutic potential in controlling murine GVHD of microRNAs with repressive effects on monocyte-specific genes is and leukemia relapse. Blood Adv., 3, 4187–4201. critical for osteoclast differentiation. Genome Biol., 16, 2. 58. Galati,D., Zanotta,S., Capitelli,L. and Bocchino,M. (2022) A bird’s 44. Wu,X. and Zhang,Y. (2017) TET-mediated active DNA eye view on the role of dendritic cells in SARS-CoV-2 infection: demethylation: mechanism, function and beyond. Nat. Rev. Genet., perspectives for immune-based vaccines. Allergy, 77, 100–110. 18, 517–534. 59. Garg,A.D., Coulie,P.G., Van den Eynde,B.J. and Agostinis,P. (2017) 45. Minor,E.A., Court,B.L., Young,J.I. and Wang,G. (2013) Ascorbate Integrating next-generation dendritic cell vaccines into the current induces ten-eleven translocation (Tet) methylcytosine cancer immunotherapy landscape. Trends Immunol., 38, 577–593. dioxygenase-mediated generation of 5-hydroxymethylcytosine. J. Biol. Chem., 288, 13669–13674.