Detection of Modified Forms of Cytosine Using Sensitive Immunohistochemistry.

Methylation of cytosine bases (5-methylcytosine, 5mC) occurring in vertebrate genomes is usually associated with transcriptional silencing. 5-hydroxylmethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) are the recently discovered modified cytosine bases produced by enzymatic oxidation of 5mC, whose biological functions remain relatively obscure. A number of approaches ranging from biochemical to antibody based techniques have been employed to study the genomic distribution and global content of these modifications in various biological systems. Although some of these approaches can be useful for quantitative assessment of these modified forms of 5mC, most of these methods do not provide any spatial information regarding the distribution of these DNA modifications in different cell types, required for correct understanding of their functional roles. Here we present a highly sensitive method for immunochemical detection of the modified forms of cytosine. This method permits co-detection of these epigenetic marks with protein lineage markers and can be employed to study their nuclear localization, thus, contributing to deciphering their potential biological roles in different experimental contexts.


Introduction
Methylation of cytosine bases in DNA (5mC) represents a major epigenetic mark found in vertebrates genomes associated with transcriptional silencing 1 . 5mC is being introduced and maintained by DNA methyltransferases [2][3][4][5] , and has been shown to play important roles in a number of biological processes including genomic imprinting, X-chromosome inactivation, cellular differentiation, and development 3,6 . Consequently, the disruption of 5mC genomic patterns is associated with a number of diseases 7,[8][9][10][11] . Despite the progress in understanding 5mC role in development and disease, it is still remain largely unknown how this mark is being removed in developing and adult tissues. Several potential mechanisms of DNA demethylation have been recently proposed including active and passive demethylation mechanisms 12,13,14,15 . Discovery of the products of 5mC sequential oxidation mediated by ten-eleven translocation enzymes (tet1/2/3) such as 5-hydroxylmethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) in eukaryotic DNA 16,17,18,19 prompted speculations whether they may serve as intermediates of DNA demethylation processes or act as stable epigenetic marks in their own right 13 . While the demonstration that the component of base excision repair, thymine-DNA glycosylase (TDG) can bind and remove both 5fC and 5caC from DNA 19,20 suggests a role for the modified 5mC derivatives in an active DNA demethylation. Recent evidence showing that 5fC/5caC can modulate the rate of RNA II processivity points to potential involvement of these marks in transcriptional regulation 29 . Due to this potential biological importance of oxidized forms of 5mC, a range of biochemical and antibody based techniques have been employed for studying their genomic distribution and global content 16,[19][20][21][22][23][24] .
Given that most of the vertebrate organs consist of different cell types and that the distribution of modified cytosine bases is tissue and cell-type specific 16-18, 20, 23, 25-27 , determining the spatial distribution of oxidized 5mC derivatives in different tissues becomes an important experimental task necessary for unveiling their biological functions. Most of the biochemical and antibody based approaches do not provide any spatial information regarding the distribution of the modified forms of 5mC in different tissue and cell-types. In contrast, immunochemistry based techniques can provide a rapid tool for assessing the spatial distribution and nuclear localization of 5mC 28 and its oxidized derivatives 20 . That is said, the reported very low abundance of 5fC (20 in every 10 6 cytosine) and 5caC (3 in every 10 6 cytosine) in the mouse genome 18 represents a significant challenge for standard immunochemistry.

Protocol
All the animal-involved procedures were performed in accordance with the University of Nottingham's ethical review board.

Selecting Suitable Tissue Preparation for Immunostaining
1. Generate paraffin-embedded section of wild-type CD1 mouse embryos and adult brain tissues as described previously 25 . Use brain tissue sections fixed with either 4% formaldehyde (FA) or 4% paraformaldehyde (PFA) for immunostaining method 25 .

NOTE:
The use of paraffin embedded tissue sections requires de-waxing prior to antibody labelling. Since the treatment with 2 -4 M hydrochloric acid (HCl) employed in the protocol for DNA denaturing is not compatible with most antigen retrieval strategies, we recommend the use of either cryo-or microtome sections for co-detection of 5mC oxidation derivatives with protein markers.

Dewaxing Paraffin Embedded Tissue Sections
1. In a running class II safety cabinet, wash the paraffin embedded tissue sections in a Coplin jar filled with xylene, 2 times for 10 min each at RT. NOTE: It is important to use fresh xylene as incomplete paraffin removal can lead to inconsistent staining patterns. 2. After de-waxing, rapidly rehydrate the tissue sections by washing consecutively in 95, 75, and 50% ethanol for 10 min each at RT.

Representative Results
To determine the distribution of 5hmC in brain tissue sections, we performed co-detection of this epigenetic modification with a marker for postmitotic neurons, NeuN, employing commercial anti-5hmC antibody that specifically interacts with this mark but not with other forms of modified cytosine 20,25 . Immunohistochemical analysis of 5hmC and 5caC distributions in the adult brain revealed that whereas the prominent 5hmC staining co-localizes with NeuN positive cells, NeuN-negative glial cells possess lower levels of genomic 5hmC (Figure 1) 20 .
We recently showed that Tet-dependent 5mC oxidation is operative during lineage specification of neural stem cells (NSCs) 20 . Despite being immunochemically undetectable in NSCs, 5fC and 5caC exhibit prominent immunostaining at early stages of NSCs differentiation towards neuronal and glial lineages. Both these marks transiently accumulate concurrently with appearance of the markers of early neuronal and glial differentiation 20 . To determine the distribution of 5caC in differentiating NSCs we performed co-staining of this mark with a glial marker GFAP on the fixed cultures of NSCs at 3 days of induction of glial differentiation. Unlike in NSCs or mature astrocytes (data not shown), we observed strong 5caC signal in relatively large proportion of the cells expressing GFAP in these cultures (Figure 2) 20 .

Discussion
Although the reported low abundance of 5mC oxidation derivatives, 5fC and 5caC in some tissues would present significant limitations for a standard immunochemistry protocol, the incorporation of peroxidase-conjugated secondary antibodies allowed the detection of these cytosine modifications in fixed tissues and cells (Figure 2). However, the optimal incubation time with the tyramide solution should be optimized experimentally for each individual batch of tyramide signal amplification kit where a linear relationship between signal intensity and duration of tyramide based signal amplification is observed, for details refer to Almeida et al. 2012 26 . In addition, the signal/background ratio can be significantly enhanced by carrying the washes in a Coplin jar to allow efficient removal of excess antibodies. Following incubation with tyramide solution, it is important to immediately stop the reaction by washing in PBT solution to decrease background staining. It is critical not to allow the sections to dry at any point during the procedure.
The efficiency of DNA depurination can be improved by carrying out the depurination reaction at 37 °C. While using 4 N HCl instead of 2 N enhances the staining of the modified forms of 5mC using 4 N HCl for DNA depurination would not permit co-staining with DAPI, as it interacts exclusively with double stranded DNA.
Although this technique can provide robust semi-quantitative assessment of the modified forms of cytosine bases where detectable, it cannot be used for evaluating the absolute levels of 5mC or its oxidation derivatives. Therefore, we recommend the use of other complementary but quantitative approaches 16, 19 -24 . Since the discovery of 5mC oxidation derivatives in mammalian genomes, several approaches have been developed to study their biological roles , this technique can be crucial to understanding the biological implications and the fate of oxidized derivatives of 5mC in various biological contexts where these modified derivatives of 5mC can be detected including cellular differentiation, development and disease. In addition, the method we describe here can give semi-quantitative assessments of the oxidized derivatives of 5mC in different tissues by assessing the kinetics of peroxidase reaction (which is proportional to the staining intensity) at different incubation times with tyramide 26 .

Disclosures
The authors declare no conflict of interest.