5-azacytidine affects TET2 and histone transcription and reshapes morphology of human skin fibroblasts

Phenotype definition is controlled by epigenetic regulations that allow cells to acquire their differentiated state. The process is reversible and attractive for therapeutic intervention and for the reactivation of hypermethylated pluripotency genes that facilitate transition to a higher plasticity state. We report the results obtained in human fibroblasts exposed to the epigenetic modifier 5-azacytidine (5-aza-CR), which increases adult cell plasticity and facilitates phenotype change. Although many aspects controlling its demethylating action have been widely investigated, the mechanisms underlying 5-aza-CR effects on cell plasticity are still poorly understood. Our experiments confirm decreased global methylation, but also demonstrate an increase of both Formylcytosine (5fC) and 5-Carboxylcytosine (5caC), indicating 5-aza-CR ability to activate a direct and active demethylating effect, possibly mediated via TET2 protein increased transcription. This was accompanied by transient upregulation of pluripotency markers and incremented histone expression, paralleled by changes in histone acetylating enzymes. Furthermore, adult fibroblasts reshaped into undifferentiated progenitor-like phenotype, with a sparse and open chromatin structure. Our findings indicate that 5-aza-CR induced somatic cell transition to a higher plasticity state is activated by multiple regulations that accompany the demethylating effect exerted by the modifier.

DNA methylation is essential for mammalian development, gene regulation, genomic imprinting, and chromatin structure 1 . Changes in methylation allow mature cells of adult organisms to acquire their differentiated state through a gradual loss of potency 2 and a progressive restriction in their options 3 .
The process is reversible and may be altered by biochemical and biological manipulation, making it an attractive target to reactivate hypermethylated pluripotency genes 4 and facilitate cell transition to a higher plasticity state 5 .
The epigenetic modifier 5-azacytidine (5-aza-CR) is a chemical analogue of cytosine. It is known to induce reversible cell cycle arrest 6,7 and acts as a direct inhibitor of methyltransferase activity, decreasing methylation in newly synthesised DNA. The molecule substitutes for cytosine, incorporates into DNA and RNA, during replication 8,9 , forms covalent adducts with DNA methyltransferase (DNMT) 1, thereby depleting the cells from enzyme activity, causing demethylation of genomic DNA 8,10 , as well as gene reactivation 11 . Thanks to its powerful effects, this compound may be used to increase chromatin plasticity and facilitate phenotype changes 12,13 . Several studies reported 5-aza-CR ability to facilitate adult somatic cell switch from one phenotype to a different one [14][15][16] . In particular, we demonstrated that a short exposure to 5-aza-CR allows a transient passage through a plastic chromatin state. This is sufficient to allow a complete directed differentiation of an adult mature cell into a different cell type [17][18][19][20] .
All these studies are very promising because they allow to obtain safe and viral vector free cells that may be used for regenerative medicine. However, the mechanisms underlying 5-aza-CR effects on cell plasticity and differentiation are still poorly understood and need to be better elucidated. In this manuscript we expose 5-aza-CR treated cells either to: 1) embryonic stem cell (ESC) medium (to promote and maintain cell plasticity); Human fibroblasts acquire a transient high plasticity state after epigenetic erasing with 5-aza-CR. Methylation decrease was accompanied by the up-regulation of the ten-eleven translocation 2 (TET2) gene (Fig. 1c, Post 5-aza-CR). However, the modifier did not affect all the TET family members, but rather it exerted a distinct effect on TET2 transcription, while no significant effect was observed for TET1 and TET3.
5-aza-CR also induced the onset of pluripotency gene expression [POU class 5 homeobox 1 (OCT4), sex determining region Y-box 2 (SOX2), Nanog homeobox (NANOG), and ZFP42 zinc finger protein (REX1)] (Fig. 1b, Post 5-aza-CR). ESC medium maintained high expression levels of all these genes for the first 2 days of culture. By contrast, pluripotency gene expression significantly decreased by day 2, when post 5-aza-CR cells were cultured in differentiation media ( Fig. 1b; 2d, 4d, 6d PANCR; 2d, 4d, 6d FB). Furthermore, even when maintained in ESC medium over a longer period of time, cells turned down the transcription of these genes by day 6 of culture ( Fig. 1b; 6d ESC). TET2 upregulation resulted in 5-Formylcytosine and 5-Carboxylcytosine increase. 5-aza-CR induced TET2 upregulation resulted in significant increase of 5fC and 5caC ( Fig. 1d and Suppl. Figure 1). These increased levels were also present in cells cultured in ESC medium for 24  TET2 downregulation affects 5-aza-CR induced global DNA demethylation. Global DNA demethylation induced by 5-aza-CR was decreased in response to TET2 RNA silencing (Fig. 1a). Significant differences in methylation levels were also detected in cells cultured in ESC medium (

5-aza-CR modulates expression of histones and histone acetylating enzymes in human fibroblasts.
Expression analysis demonstrated 5-aza-CR ability to significantly upregulate the transcription of histones within the 1, 2, 3 and 4 families, and their encoded protein (Figs 2 and 3). Histones showed a significant increase post 5-aza-CR and persisted at high levels in cells cultured in ESC medium for 24 h and 48 h (24 h ESC; 48 h ESC). In contrast, their expression levels were significantly down-regulated when cells were transferred to PANCR (24 h PANCR; 48 h PANCR) and returned to values comparable to T0 after 48 h in FB medium (48 h FB). No significant change was detected in cells cultured in the different media, without prior exposure to 5-aza-CR.
5-aza-CR did not affect the expression of HIST1H2AC (H2A type 1-c) and HIST2H2AC (H2A type 2-C, Figs 3 and 4). Moreover, we detected a parallel downregulation of a fibroblast specific gene (COL6A3) as well as a decreased expression of a transcript involved in cell cycle control (CCNG1). Furthermore, increased expression of histones was supported by the upregulation of the acetylating enzyme Histone acetyltransferase 1 (HAT1) and by the decreased expression of the Histone deacetylase enzyme 1 (HDAC1).
Scientific RepoRts | 6:37017 | DOI: 10.1038/srep37017 Figure 1. (a) Global DNA methylation of cells before and after siRNA transfection, exposed to 5-aza-CR and to different treatments for 24 and 48 hours. Highest expression set to 1 and all other times relative to this. Bars represent the mean ± SD of two independent experiments with three independent replicates, obtained from three different fibroblast primary lines. Different superscripts denote significant differences between groups (P < 0.05). (b) 5-aza-CR treatment induced the onset of pluripotency gene expression, increasing the transcription of OCT4, SOX2, NANOG and REX1 (Post 5-aza-CR). Culture in ESC medium for two days (2d ESC) maintained upregulated expression of all the genes, that are turned down by day 6 of culture (6d ESC). 5-aza-CR treated cells (Post 5-aza-CR) display pluripotency gene transcription levels in the range of 45% of those detected in human ESC (hESC) and iPSC (hiPSC). Gene expression levels are reported with the T0 expression set to 1 and all other times relative to this. Different superscripts denote significant differences between groups (P < 0.05). (c) Effect of 5-aza-CR on TET gene family transcription. The modifier upregulated TET2, while leaving the other TET members unchanged. Gene expression levels are reported with the T0 expression set to 1 and all other times relative to this. Different superscripts denote significant differences between groups (P < 0.05). (d) DNA dot blot analysis of 5-Formylcytosine (5fC) and 5-Carboxylcytosine (5caC) in cells exposed to 5-aza-CR and to different treatments for 24 and 48 hours.

5-aza-CR affects cell morphology and chromatin organization. T0 cells displayed a typical fibro-
blast morphology and were characterized by nuclei with condensed areas close to the nuclear membrane and  panel f), which is likely to be responsible for a final reduction in cell size. These morphological features, that are typically related to a high plasticity phenotype 24 , were maintained by 84,24 ± 5,05% and 83,74 ± 5,52% of cells cultured in ESC medium for 24 h and 48 h respectively, as confirmed at higher magnification ( Fig. 5 panels g-i), possibly suggesting that, once acquired, the increased plasticity state can be sustained by signals derived from medium.
Interestingly, cells returned to FB medium after exposure to 5-aza-CR re-established the fibroblast phenotype ( Fig. 5 panels j-l), indicating that the effect of the modifier is reversible and allows the re-acquisition of the original cell phenotype.
When post 5-aza-CR cells were transferred to PANCR medium, which has been previously shown to address them towards the pancreatic endoderm differentiation 17 , they adopted a completely different spatial organization. After 24 hours in this medium, 80,12 ± 4,41% cells were variously shaped and showed numerous and long cytoplasmic projections. At higher magnification, nuclei were characterized by indented profiles, clustered dense chromatin close to the nuclear envelope and prominent nucleoli ( Fig. 6 panels a, b, d, e). Cytoplasm contained mitochondria, lysosomes, granules and numerous vacuoles which enclosed materials of different electron density ( Fig. 6 panels c, d, f). Numerous RER dilated cisternae, wrapping fine reticular network could be appreciated close to the cell surface ( Fig. 6 panels a, c). At 48 hours, the same rate of cells had spotted contacts ( Fig. 6 panels g, i) or became tightly attached due to the presence of gap-like junctions ( Fig. 6 panel h). Cytoplasmic bridges among adjacent cells were visible ( Fig. 6 panel j). Vacuoles with heterogeneous material, granules and autophagosomes were present in the inner region of cytoplasm, while bundles of filaments were localized at the periphery of the cells (Fig. 6 panel k).

Discussion
The epigenetic modifier 5-aza-CR has been shown to incorporate into DNA during cell replication, leading to a global demethylation and gene reactivation 11 . Thanks to this powerful effects, the compound has been used to increase chromatin plasticity and facilitate phenotype changes in different models 12,13,[17][18][19][20] . Although the DNA demethylation capabilities of 5-aza-CR have been well characterized, both in vitro and in vivo, there is also reason to believe that this drug exhibits alternative mechanisms of action 25,26 . Here we confirm the global effect on methylation and, as shown in Fig. 1a, we detect a decreased methylation in fibroblasts exposed to 5-aza-CR. These levels are maintained in cells cultured in ESC medium, suggesting that DNA methylation levels may be modulated by the milieu of the cells, including signals deriving from the medium. These findings are in agreement with Blaschke et al. 27 and Yin et al. 28 that demonstrated changes in DNA methylation in response to specific media formulation and the contained nutritional factors. This possibility is also confirmed by the observation that cells, treated with the modifier, but exposed to FB medium and or PANCR medium, restored their methylation levels. Interestingly, the changes described were accompanied by an increased expression of several genes. Following preliminary NGS analysis of untreated fibroblasts vs. 5-aza-CR exposed cells, we here confirm upregulated transcription of the TET2 gene, which has been recently shown to support 5-aza-CR dependent demethylation  (HIST1H2AC, HIST2H2AC and HIST1H4E), or resulted in downregulation of COL6A3 and CCNG1. Furthermore, a sharp upregulation of HAT1, and a decrease of HDAC1 was detected after exposure to 5-aza-CR. Different superscripts denote significant differences between groups (P < 0.05).
process in hepatocarcinoma cell lines 29 . In particular, we show that upregulation of TET2 results in an increase of both 5fC and 5caC (Fig. 1d), indicating 5-aza-CR ability to activate a direct and active demethylating effect, possibly mediated via TET2 protein. Furthermore, the possibility of a synergistic action between direct and indirect demethylation processes is also demonstrated thank to TET2 siRNA experiments, where we demonstrate that downregulation of TET2 protein results in an overall decrease of global DNA demethylation (Fig. 1a), that however still persist in response to 5-aza-CR, possibly due to the unaffected action of 5-aza-CR on DNMTs.
TET2 upregulation is also in agreement with the acquisition of the higher plasticity state detected in post 5-aza-CR cells, since TET2 has been reported to have a role in pluripotency induction and maintenance 30 . Interestingly, 5-aza-CR did not affect all the members of the TET family, but rather it exerted a distinct effect on TET2, leaving the other members unchanged. Based on previous observations, showing that TET2 does not affect methylation levels at gene promoters and transcription start sites, but, in contrast, mainly regulates gene bodies and exon boundaries, our results suggest the possibility of a specific effect of the modifier on the latter.
Exposure to 5-aza-CR also caused the onset of pluripotency gene expression, such as OCT4, NANOG, REX1 and SOX2 (Fig. 1b). As shown in Fig. 1b, these genes were induced by 5-aza-CR, displaying transcription levels in the range of 45% of those detected in human ESCs and iPSCs. This result is also consistent with the molecule ability to reactivate previously silent genes and to alter the differentiation state of eukaryotic cells, as well as with recent data showing that 5-aza-CR demethylation of one or more loci facilitates a rapid and stable transition of partially reprogrammed iPSCs to a fully reprogrammed state 31 .
Interestingly, culture in ESC medium maintained upregulated expression of all the genes described, suggesting the acquisition of a high plasticity state. However, this condition was transient and reversible, since expression of pluripotency related genes decreased gradually when post 5-aza-CR cells where exposed to differentiation media (24 h and 48 h PANCR; 24 h and 48 h FB). Furthermore, even when maintained in ESC medium over a longer period of time, cells turned down the transcription of these genes by day 6 of culture (Fig. 1b).
The expression of histones belonging to 1, 2, 3 and 4 families was also affected by the exposure to the epigenetic eraser. The genes and their encoded proteins showed increases (between 3 to 11 fold), in cells treated with 5-aza-CR, and persisted at high levels in cells cultured in ESC medium for 24 and 48 hours (Figs 2 and 3). The effect on transcription was evident in replication-dependent as well as in replication-independent histones, indicating that it may not only be a secondary indirect consequence of 5-aza-CR on the cell cycle. Furthermore, histones belonging to the same cluster was either upregulated or unchanged by 5-aza-CR, suggesting a histone specific effect rather than a cluster specific one. We have no ready explanation for this, although it is known that separate sets of histone genes are transcribed in a precise temporal sequence during development and differentiation of lower species 32 . On the other hand, the contribution of histone transcription to the maintenance of high plasticity versus differentiation in mammalian is presently still to be fully understood and further studies are required.
We postulate that the 5-aza-CR effect on histone gene transcription is unlikely to be accounted for a general demethylating effect, since exposure to the modifier, resulted in parallel downregulation of fibroblast specific genes (COL6A3) as well as of transcripts involved in the control of cell cycle (CCNG1) and did not affect the expression of several histones genes (HIST1H2AC, HIST2H2AC and HIST1H4E, Fig. 4). In agreement with this hypothesis, Komashko et al. demonstrated that several changes in gene expression after 5-aza-CR treatment were related to an alternative DNA-hypomethylating independent effect 33 . Furthermore, consistent with the idea that this epigenetic modifier has additional modes of action, several works indicated that 5-aza-CR can induce expression of genes lacking CpG methylation 34,35 . Interestingly, Takebayashi et al. demonstrated a DNA methylation independent mechanism, directly related to an increase in histone hyperacetylation 25 . This is also in agreement with our observation that HAT1 transcription was significantly higher in cells exposed to 5-aza-CR, suggesting the possible correlation among the exposure to the modifier, the overexpression of HAT1, the consequent histone hyperacetylation and increased transcription. This possibility is further supported by the parallel decreased expression of HDAC1 (Fig. 4), reported in our manuscript. It is also coherent with previous works that show how modulation of histone acetylation is an integral component of gene expression control 36 and demonstrate a key role of the specific enzymatic activity of HDACs in transcriptional regulation 37 .
Morphological analysis further unravelled the impact of 5-aza-CR on cell plasticity and differentiation. In particular, it is interesting to underline that adult skin fibroblasts changed after 5-aza-CR exposure, became smaller in size and reshaped into an undifferentiated progenitor-like phenotype. In this condition, cells showed a nuclei/cytoplasm ratio similar to what previously described in hESCs 38 and in hiPSCs obtained from human dermal fibroblasts 39 . 5-aza-CR increased plasticity was also accompanied by autophagic phenomena, that have been reported to play a critical role in the maintenance of adult stem cells 40 , possibly reducing senescence via the active removal of oxigen reactive species 41 . Interestingly the presence of autophagic vacuoles have been previously reported in ultrastructural analysis of hiPSC 42 . Furthermore, post 5-aza-CR cell nuclei were characterized by sparse chromatin and appeared devoid of compact formations. This is consistent with the observation that an open chromatin structure is required for a high plasticity state 43 and is a common feature of iPSC and, more in general, one of the hallmark of stem cells 43,44 . Interestingly, this condition was maintained by ESC medium, possibly suggesting that, once acquired, the decreased methylation state can be sustained by factors contained in the medium such as LIF, which has been recently demonstrated to induce murine cell hypomethylation 45 . The acquisition of a transient high plasticity state is also consistent with post 5-aza-CR cell ability to respond to differentiation protocols and acquire a differentiated phenotype [17][18][19][20] . Indeed, when 5-aza-CR treated cells were transferred to PANCR medium, which address them towards the pancreatic endoderm differentiation 17 , they adopted a different phenotype. Nuclei were characterized by indented profiles, prominent nucleoli and clustered dense chromatin, mostly locating close to the nuclear envelope. These morphological changes, involving a packed chromatin conformation, and its clustering near the nuclear envelope, have been previously described and considered to be distinctive of a differentiated state 24 and suggest that post 5-aza-CR cells may reverse their high plasticity state and respond to differentiation stimuli. This is consistent with the methylation and transcriptional changes described above and indicates that the mechanisms exerted by the modifier have also deep effects at the phenotype level. Such hypothesis is further supported by the observation that the majority of cells showed gap-like junctions and cytoplasmic bridges, all features required to guarantee a correct differentiation processes 46,47 .
The reversibility of the effects exerted by 5-aza-CR, was confirmed by the re-acquisition of the original phenotype when cells were returned to FB medium, indicating that a short exposure to the modifier allows a transient high plasticity chromatin state, sufficient to allow cell reshaping.
Altogether, the data presented in this manuscript, indicate 5-aza-CR ability to increase TET2 gene expression, modulate transcription of histones and histone acetylating enzymes, which result in specific morphological changes. Moreover, the results obtained, also suggest the molecule ability to interfere with DNA methylation through a direct TET2-mediated mechanism, that accompanies the well-known indirect DNMT related one. siRNA transfection. The targeting siRNA used in the present study was TET2 siRNA (Santa Cruz) and siRNA induced protein knock-down was carried out as previously described 49 . More in detail, cells were transfected 24 hours before exposure to the modifier and during 5-aza-CR treatment, with 90 pM of siRNA, using Lipofectamine RNAi MAX Reagents diluted in Opti-MEM ® Medium, following the manufacturer's instructions. For negative controls, cells were incubated with Lipofectamine alone or with Lipofectamine plus 90 pM non-silencing siRNAs. The success of the silencing occurrence was assessed by monitoring the decrease of TET2 protein. After transfection global DNA methylation was assessed as described below. All experiments were performed three times in triplicate samples.
Global methylation analysis. Genomic DNA was extracted from untreated fibroblasts and after their siRNA transfection at different time points: T0, after 5-aza-CR treatment (Post 5-aza-CR), 24 and 48 hours after culture in fibroblast (FB), ESC or pancreatic induction (PANCR) media. PureLink ® Genomic DNA Kits was used according to the manufacturer's instructions. DNA was converted to single-stranded DNA by incubation at 95 °C for 5 min, followed by rapid chilling on ice. Samples were then digested to nucleosides by incubating the denatured DNA with nuclease P1 for 2 h at 37 °C in 20 mM sodium acetate (pH 5.2). Alkaline phosphatase was added and incubated for 1 h at 37 °C in 100 mM Tris (pH 7.5). After centrifugation, the supernatant was used for ELISA assay using Global DNA Methylation ELISA Kit (5′ -methyl-2′ -deoxycytidine Quantitation; CELL BIOLABS) according to the manufacturer's protocol. DNA dot blot. Genomic DNA was extracted with PureLink ® Genomic DNA Kits as previously described.
DNA concentration was assessed with NanoDrop 8000 (Thermoscientific). Aliquots of 200 ng total DNA were prepared in a total volume of 2 μ L per sample and spotted onto nylon membranes (Hybond-N+ , Amersham). The DNA spots were air dried for 15 min, UV-crosslinked for 1 min and probed with primary antibodies against 5-Carboxylcytosine (Active Motif, 1:1500) and 5-Formylcytosine (Active Motif, 1:1500). Dots were visualized with a WesternBreeze chemiluminescent kit. Signal intensity was quantified by densitometric analysis, using the Image J analysis software (National Institutes of Health).
Scientific RepoRts | 6:37017 | DOI: 10.1038/srep37017 number of amplification cycles for the fluorescent reporter signal to reach a common threshold value (Ct) were estimated and then normalized against the Ct value obtained for β -actin of the same sample to give the Δ Ct value.
Quantification of histone protein levels. Cells were detached from the culture dish with trypsin and collected by centrifugation. They were then washed three times with cold PBS (1x) and subjected to ultrasonication. After centrifugation and dilution, samples were used for ELISA assay using the commercially available Enzyme-linked Immunoassorbent Assay Kit (MyBioSource) for the specific histone (Table 2) Electron microscopy. Samples were fixed for 2 hours in 0.1 M cacodylate buffer pH 7.2, containing 2% glutaraldehyde. Specimens were then washed in the same buffer and post-fixed for 2 hours with 1% osmic acid in cacodylate buffer. After standard serial ethanol dehydration, specimens were embedded in an Epon-Araldite 812 mixture. Sections were obtained with a Reichert Ultracut S ultratome (Leica, Austria). Semi-thin sections were stained by conventional methods (crystal violet and basic fuchsin) and observed under a light microscope (Olympus, Japan) to score cells for different morphological changes. Thin sections were stained by uranyl acetate and lead citrate and observed with a Jeol 1010 EX electron microscope (Jeol, Japan).