Literature DB >> 28473874

Regional differences in mitochondrial DNA methylation in human post-mortem brain tissue.

Matthew Devall1, Rebecca G Smith1, Aaron Jeffries1,2, Eilis Hannon1, Matthew N Davies3, Leonard Schalkwyk4, Jonathan Mill1,2, Michael Weedon1, Katie Lunnon1.   

Abstract

BACKGROUND: DNA methylation is an important epigenetic mechanism involved in gene regulation, with alterations in DNA methylation in the nuclear genome being linked to numerous complex diseases. Mitochondrial DNA methylation is a phenomenon that is receiving ever-increasing interest, particularly in diseases characterized by mitochondrial dysfunction; however, most studies have been limited to the investigation of specific target regions. Analyses spanning the entire mitochondrial genome have been limited, potentially due to the amount of input DNA required. Further, mitochondrial genetic studies have been previously confounded by nuclear-mitochondrial pseudogenes. Methylated DNA Immunoprecipitation Sequencing is a technique widely used to profile DNA methylation across the nuclear genome; however, reads mapped to mitochondrial DNA are often discarded. Here, we have developed an approach to control for nuclear-mitochondrial pseudogenes within Methylated DNA Immunoprecipitation Sequencing data. We highlight the utility of this approach in identifying differences in mitochondrial DNA methylation across regions of the human brain and pre-mortem blood.
RESULTS: We were able to correlate mitochondrial DNA methylation patterns between the cortex, cerebellum and blood. We identified 74 nominally significant differentially methylated regions (p < 0.05) in the mitochondrial genome, between anatomically separate cortical regions and the cerebellum in matched samples (N = 3 matched donors). Further analysis identified eight significant differentially methylated regions between the total cortex and cerebellum after correcting for multiple testing. Using unsupervised hierarchical clustering analysis of the mitochondrial DNA methylome, we were able to identify tissue-specific patterns of mitochondrial DNA methylation between blood, cerebellum and cortex.
CONCLUSIONS: Our study represents a comprehensive analysis of the mitochondrial methylome using pre-existing Methylated DNA Immunoprecipitation Sequencing data to identify brain region-specific patterns of mitochondrial DNA methylation.

Entities:  

Keywords:  5-Methylcytosine; 5-mC; Blood; Brain; DNA methylation; Epigenetics; MeDIP-seq; Mitochondria; NUMTs

Mesh：

Substances：

Year:  2017        PMID： 28473874      PMCID： PMC5415779          DOI： 10.1186/s13148-017-0337-3

Source DB:  PubMed          Journal:  Clin Epigenetics        ISSN： 1868-7075            Impact factor:   6.551


Introduction

Mitochondria are unique organelles in that they have their own circular genome, approximately 16.6 kb in size [1]. Mitochondrial DNA (mtDNA) consists of 37 genes, 22 encoding for transfer RNAs (tRNAs), two for ribosomal RNAs (rRNAs) and 13 encoding for proteins important in the electron transport chain. Each of these 13 proteins are directly involved in the regulation of cellular respiration, generating the majority of ATP required for the process. However, mitochondria have an array of other important cellular roles such as calcium homeostasis [2] and neural stem cell differentiation [3]. As such, abnormal mitochondrial function, dynamics and trafficking have been associated with a number of brain disorders including Alzheimer’s disease [4, 5], schizophrenia [6], bipolar disorder [7] and major depressive disorder [8]. Epigenetic processes mediate the reversible regulation of gene expression, occurring independently of DNA sequence variation, acting principally through chemical modifications to DNA and nucleosomal histone proteins and orchestrate a diverse range of important physiological functions. DNA methylation is the best characterized and most stable epigenetic modification modulating the transcription of mammalian genomes and, because it can be robustly assessed using existing genomic DNA resources, is the focus of most human epidemiological epigenetic research to date [9]. The most widely used method for epigenome-wide analysis of DNA methylation is the Illumina 450K methylation array, and a number of studies have recently shown differential DNA methylation of the nuclear genome (ncDNA), between different tissue types [10-12] and also in a range of complex diseases, from brain disorders such as Alzheimer’s disease [13-15] and schizophrenia [16, 17], to systemic diseases such as type 2 diabetes [18] and Crohn’s disease [19]. However, with no representation of the mitochondrial genome on this platform, as well as a lack of analysis on other genome-wide platforms, the role of mtDNA methylation has been largely neglected [20, 21]. Since the identification of 5-methylcytosine (5-mC) in mitochondria, research into mtDNA methylation as an independent and potentially relevant mark has received more regular attention [22, 23]. However, most research is either focussed on low resolution, global DNA methylation, or candidate gene DNA methylation changes using techniques such as bisulfite pyrosequencing [20]. These recent publications have indicated that differences in mtDNA methylation are present in a variety of different phenotypes [24-29] and may have potential utility as a biomarker [30]. In addition, a recent study has explored the use of Methylated DNA Immunoprecipitation Sequencing (MeDIP-seq) to investigate changes in mtDNA methylation across 39 cell lines and tissues from publicly available data [31]. At present, genome-wide sequencing technologies have not yet been used to interrogate alterations in the mtDNA methylome across tissues in the same individuals. A high proportion of current, publicly available, genome-wide DNA methylation data has been generated through the use of MeDIP-seq, a method designed to interrogate genome-wide changes in methylation at high throughput and low cost [32]. However, given the presence of nuclear-mitochondrial pseudogenes (NUMTs), regions of the nuclear genome that share a high sequence homology with their mitochondrial paralogue [33, 34], mitochondrial reads are often discarded from further analysis. The development of bioinformatic pipelines to investigate regions of differential mtDNA methylation from whole genome data would provide a novel way in which to interrogate the mtDNA methylome in publicly available data. Here, we control for the presence of NUMTs in a previously published MeDIP-seq dataset, to investigate differential DNA methylation across the mitochondrial genome in human post-mortem brain samples.

Results

MtDNA methylation patterns are correlated between the cortex, cerebellum and blood

To date, no study has investigated differences in mtDNA methylation across different matched regions of human brain and blood samples. Our sample (Table 1) consisted of MeDIP-seq data from three individuals, free of any neuropathology and neuropsychiatric disease, for five different regions of the cortex (Brodmann areas (BA) 8, 9 and 10, superior temporal gyrus (STG) and entorhinal cortex (ECX)), the cerebellum (CER) and pre-mortem blood [35]. Given that MeDIP-seq data has been generated from standardly extracted total genomic DNA and thus contains a mixture of ncDNA and mtDNA [36], we initially controlled for regions of high sequence homology between the two genomes within our data by realigning mtDNA reads to a series of custom reference genomes using an in-house pipeline (see the Methods section) to specifically analyze mtDNA methylation (Fig. 1). Briefly, after an initial alignment to the GRCH37 reference genome using BWA, uniquely mapped reads were extracted and aligned to a custom GRCH37 reference genome not containing the mitochondrial sequence. Reads that did not map to this custom genome were found to share less homology with the nuclear genome and were taken forward and realigned once more to the full reference genome. Initially, we were interested to investigate whether changes in mtDNA across the mitochondrial genome were highly correlated between different tissue types. Using principal component analysis (PCA), we found that mtDNA methylation patterns are highly correlated between different cortical regions (r > 0.99, p < 2.2E−16), with a slightly weaker correlation between the cerebellum and cortex (r > 0.97, p < 2.2E−16) (Fig. 2). Due to the small number of blood samples available, deriving a significance level for the correlations between the cerebellum and blood could not be made. Instead, in an attempt to explore the similarity between matched blood and cerebellum samples, the direction of differential methylation with respect to the cortex was used. Here, we found that 93.1% of the windows analyzed in the cerebellum and blood had the same direction of methylation difference with respect to the cortex, further suggesting a strong correlation between the two tissue types.
Table 1

Demographic information

IndividualAge at death (years)Age at bloods sampled (years)Post-mortem delay (hours)Gender
1827943Female
292N/A17Female
3787810Male

MeDIP-seq data was available from post-mortem brain samples obtained from three individuals free of any neuropathology and neuropsychiatric disease. Data was available for five different regions of the cortex (Brodmann areas (BA) 8, 9 and 10, superior temporal gyrus (STG), entorhinal cortex (ECX), the cerebellum (CER) and pre-mortem blood (BLD). MeDIP-seq data was available for all individuals from cortical and cerebellar samples; however, blood MeDIP-seq data was not available for individual 2. Data is freely available to download from http://epigenetics.iop.kcl.ac.uk/brain

Fig. 1

Overview of the analysis pipeline

Fig. 2

MtDNA methylation patterns are correlated between the cortex, cerebellum and blood. Samples were ordered based upon the similarity of their principal components, RPKM values, with r calculated for the correlations between each tissue. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, STG superior temporal gyrus

Demographic information MeDIP-seq data was available from post-mortem brain samples obtained from three individuals free of any neuropathology and neuropsychiatric disease. Data was available for five different regions of the cortex (Brodmann areas (BA) 8, 9 and 10, superior temporal gyrus (STG), entorhinal cortex (ECX), the cerebellum (CER) and pre-mortem blood (BLD). MeDIP-seq data was available for all individuals from cortical and cerebellar samples; however, blood MeDIP-seq data was not available for individual 2. Data is freely available to download from http://epigenetics.iop.kcl.ac.uk/brain Overview of the analysis pipeline MtDNA methylation patterns are correlated between the cortex, cerebellum and blood. Samples were ordered based upon the similarity of their principal components, RPKM values, with r calculated for the correlations between each tissue. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, STG superior temporal gyrus

Differentially methylated regions of the mitochondrial genome can be identified between anatomically distinct cortical regions and the cerebellum

Having identified correlated mtDNA methylation patterns across different brain regions, we were interested to investigate whether we could identify differentially methylated regions (DMRs) in the mitochondrial genome between different regions of the cortex and cerebellum. To identify such tissue-specific DMRs within the mitochondrial genome, paired t tests were performed across matched cortical and cerebellum samples at 100 bp windows across the mitochondrial genome (see the Methods section). In total, we identified 74 nominally significant DMRs (p < 0.05) between the five individual cortical regions and the cerebellum (Table 2; Fig. 3). Of these DMRs, seven (Table 2, bold face) were found to be present across all prefrontal cortex areas (BA8, BA9, BA10). Furthermore, the direction of methylation difference was maintained in all Brodmann area regions, with three conserved regions of hypomethylation and four conserved regions of hypermethylation, with respect to the cerebellum. Furthermore, four of the seven conserved regions were adjacent to each other within the mitochondrial displacement loop (D-Loop) (16201–16600 bp), a region associated with gene transcription and DNA replication.
Table 2

List of DMRs identified between five anatomically discreet cortical regions and cerebellum

Start (bp)Stop (bp)Gene(s)BA8BA9BA10ECSTG
p valueΔ RPKM p valueΔ RPKM p valueΔ RPKM p valueΔ RPKM p valueΔ RPKM
1100D-Loop - - - - - - - - - -
101200D-Loop - - - - - - - - - -
201300D-Loop - - - - - - - - - -
301400D-Loop - - - - - - - - - -
401500D-Loop - - - - - - - - - -
501600MT-TF - - - - - - - - - -
601700MT-TF/MT-RNR1 - - - - - - - - - -
701800MT-RNR1------3.21E−0279156- -
801900MT-RNR14.31E−02942462.93E−02−43592--- - 1.79E−0295257
9011000MT-RNR1NDNDNDNDNDNDNDNDNDND
10011100MT-RNR1NDNDNDNDNDNDNDNDNDND
11011200MT-RNR1NDNDNDNDNDNDNDNDNDND
12011300MT-RNR1NDNDNDNDNDNDNDNDNDND
13011400MT-RNR1NDNDNDNDNDNDNDNDNDND
14011500MT-RNR1NDNDNDNDNDNDNDNDNDND
15011600MT-RNR1NDNDNDNDNDNDNDNDNDND
16011700MT-RNR1/MT-TV/MT-RNR2NDNDNDNDNDNDNDNDNDND
17011800MT-RNR2NDNDNDNDNDNDNDNDNDND
18011900MT-RNR2NDNDNDNDNDNDNDNDNDND
19012000MT-RNR2NDNDNDNDNDNDNDNDNDND
20012100MT-RNR2NDNDNDNDNDNDNDNDNDND
21012200MT-RNR2NDNDNDNDNDNDNDNDNDND
22012300MT-RNR2NDNDNDNDNDNDNDNDNDND
2301 2400 MT-RNR2 2.54E−02 −97351 3.20E−02 −29509 4.31E−02 −82048 ----
24012500MT-RNR28.90E−0370950--------
25012600MT-RNR23.30E−0351937--------
26012700MT-RNR2NDNDNDNDNDNDNDNDNDND
27012800MT-RNR2NDNDNDNDNDNDNDNDNDND
28012900MT-RNR2----------
29013000MT-RNR2----------
30013100MT-RNR2----------
31013200MT-RNR2NDNDNDNDNDNDNDNDNDND
32013300MT-RNR2/MT-TL------2.09E−02242424--
33013400MT-TL1/MT-ND1----------
34013500MT-ND1--1.54E−02864685------
35013600MT-ND14.41E−021180914--1.72E−02226066----
3601 3700 MT-ND1 4.73E−02 1250436 4.52E−02 1228936 3.73E−02 1183681 ----
37013800MT-ND1----2.16E−0220818----
38013900MT-ND1----------
39014000MT-ND1----------
40014100MT-ND12.41E−02−394981.55E−02−38704--4.68E−02−497531.56E−02243767
41014200MT-ND1NDNDNDNDNDNDNDNDNDND
42014300MT-ND1/MT-TINDNDNDNDNDNDNDNDNDND
43014400MT-TI/MT-TQNDNDNDNDNDNDNDNDNDND
44014500MT-TM/MT-ND2NDNDNDNDNDNDNDNDNDND
45014600MT-ND2NDNDNDNDNDNDNDNDNDND
46014700MT-ND2NDNDNDNDNDNDNDNDNDND
47014800MT-ND2NDNDNDNDNDNDNDNDNDND
48014900MT-ND2NDNDNDNDNDNDNDNDNDND
49015000MT-ND2NDNDNDNDNDNDNDNDNDND
50015100MT-ND2NDNDNDNDNDNDNDNDNDND
51015200MT-ND2NDNDNDNDNDNDNDNDNDND
52015300MT-ND2NDNDNDNDNDNDNDNDNDND
53015400MT-ND2NDNDNDNDNDNDNDNDNDND
54015500MT-ND2NDNDNDNDNDNDNDNDNDND
55015600MT-ND2/MT-TW/MT-TANDNDNDNDNDNDNDNDNDND
56015700MT-TA/MT-TNNDNDNDNDNDNDNDNDNDND
57015800MT-TN/MT-TCNDNDNDNDNDNDNDNDNDND
58015900MT-TC/MT-TYNDNDNDNDNDNDNDNDNDND
59016000MT-CO1NDNDNDNDNDNDNDNDNDND
60016100MT-CO1NDNDNDNDNDNDNDNDNDND
61016200MT-CO1NDNDNDNDNDNDNDNDNDND
62016300MT-CO1NDNDNDNDNDNDNDNDNDND
63016400MT-CO1NDNDNDNDNDNDNDNDNDND
64016500MT-CO1NDNDNDNDNDNDNDNDNDND
65016600MT-CO1NDNDNDNDNDNDNDNDNDND
66016700MT-CO1NDNDNDNDNDNDNDNDNDND
67016800MT-CO1NDNDNDNDNDNDNDNDNDND
68016900MT-CO1NDNDNDNDNDNDNDNDNDND
69017000MT-CO1NDNDNDNDNDNDNDNDNDND
70017100MT-CO1NDNDNDNDNDNDNDNDNDND
71017200MT-CO1NDNDNDNDNDNDNDNDNDND
72017300MT-CO1NDNDNDNDNDNDNDNDNDND
73017400MT-CO1NDNDNDNDNDNDNDNDNDND
74017500MT-CO1/MT-TS1NDNDNDNDNDNDNDNDNDND
75017600MT-TS1/MT-TD/MT-CO2NDNDNDNDNDNDNDNDNDND
76017700MT-CO2NDNDNDNDNDNDNDNDNDND
77017800MT-CO2NDNDNDNDNDNDNDNDNDND
78017900MT-CO2NDNDNDNDNDNDNDNDNDND
79018000MT-CO2NDNDNDNDNDNDNDNDNDND
80018100MT-CO2NDNDNDNDNDNDNDNDNDND
81018200MT-CO2NDNDNDNDNDNDNDNDNDND
82018300MT-CO2/MT-TKNDNDNDNDNDNDNDNDNDND
83018400MT-TK/MT-ATP8NDNDNDNDNDNDNDNDNDND
84018500MT-ATP8NDNDNDNDNDNDNDNDNDND
85018600MT-ATP8/MT-ATP6NDNDNDNDNDNDNDNDNDND
86018700MT-ATP6NDNDNDNDNDNDNDNDNDND
87018800MT-ATP6NDNDNDNDNDNDNDNDNDND
88018900MT-ATP6NDNDNDNDNDNDNDNDNDND
89019000MT-ATP6NDNDNDNDNDNDNDNDNDND
90019100MT-ATP6NDNDNDNDNDNDNDNDNDND
91019200MT-ATP6NDNDNDNDNDNDNDNDNDND
92019300MT-ATP6/MT-CO3NDNDNDNDNDNDNDNDNDND
93019400MT-CO3NDNDNDNDNDNDNDNDNDND
94019500MT-CO3NDNDNDNDNDNDNDNDNDND
95019600MT-CO3NDNDNDNDNDNDNDNDNDND
96019700MT-CO3NDNDNDNDNDNDNDNDNDND
97019800MT-CO3NDNDNDNDNDNDNDNDNDND
98019900MT-CO3----------
990110000MT-CO3/MT-TG----------
1000110100MT-TG/MT-ND3----------
1010110200MT-ND3----------
1020110300MT-ND34.68E−02−662126----3.14E−02−5910677.30E−03−57012
10301 10400 MT-ND3/MT-TR/MT-ND4L 3.61E−02 −132719 3.61E−02 −135759 4.02E−02 −115364 1.40E−02−730633.36E−02−105106
1040110500MT-ND4L--4.68E−02−71246------
1050110600MT-ND4L----------
1060110700MT-ND4LNDNDNDNDNDNDNDNDNDND
1070110800MT-ND4L/MT-ND4NDNDNDNDNDNDNDNDNDND
1080110900MT-ND4NDNDNDNDNDNDNDNDNDND
1090111000MT-ND4NDNDNDNDNDNDNDNDNDND
1100111100MT-ND4NDNDNDNDNDNDNDNDNDND
1110111200MT-ND4NDNDNDNDNDNDNDNDNDND
1120111300MT-ND4--1.94E−02−689031.84E−02−796254.71E−0258742--
1130111400MT-ND4--4.78E−02−139847------
1140111500MT-ND4NDNDNDNDNDNDNDNDNDND
1150111600MT-ND4NDNDNDNDNDNDNDNDNDND
1160111700MT-ND4NDNDNDNDNDNDNDNDNDND
1170111800MT-ND4NDNDNDNDNDNDNDNDNDND
1180111900MT-ND4NDNDNDNDNDNDNDNDNDND
1190112000MT-ND4NDNDNDNDNDNDNDNDNDND
1200112100MT-ND4NDNDNDNDNDNDNDNDNDND
1210112200MT-ND4/MT-THNDNDNDNDNDNDNDNDNDND
1220112300MT-TS2/MT-TL2NDNDNDNDNDNDNDNDNDND
1230112400MT-TL2/MT-ND54.71E−021787-−197592------
1240112500MT-ND5----------
1250112600MT-ND5----------
1260112700MT-ND5----------
1270112800MT-ND52.05E−02−5702--4.58E−02−388706--4.36E−02−394978
1280112900MT-ND52.26E−02−89668--------
1290113000MT-ND5--3.10E−02−130016------
1300113100MT-ND5NDNDNDNDNDNDNDNDNDND
1310113200MT-ND5NDNDNDNDNDNDNDNDNDND
1320113300MT-ND5NDNDNDNDNDNDNDNDNDND
1330113400MT-ND51.48E−02−175917--9.80E−03−1449493.40E−02−133694--
1340113500MT-ND53.64E−02−104010--3.40E−02−81264----
1350113600MT-ND5----------
1360113700MT-ND5----------
1370113800MT-ND51.31E−02−422610--1.20E−033714----
1380113900MT-ND5--3.72E−027087612.59E−02123249----
1390114000MT-ND5----2.17E−02−75118----
1400114100MT-ND5----------
1410114200MT-ND5/MT-ND6--------4.18E−02−534766
1420114300MT-ND6----------
1430114400MT-ND6----------
1440114500MT-ND6----------
1450114600MT-ND6--------4.86E−02−82767
1460114700MT-ND6/MT-TE----------
1470114800MT-TE/MT-CYB----------
1480114900MT-CYBNDNDNDNDNDNDNDNDNDND
1490115000MT-CYBNDNDNDNDNDNDNDNDNDND
1500115100MT-CYBNDNDNDNDNDNDNDNDNDND
1510115200MT-CYB----------
1520115300MT-CYB------1.20E−02254183--
1530115400MT-CYB----------
1540115500MT-CYB4.38E−0210628--2.10E−02−554527--3.28E−02−554225
1550115600MT-CYB3.73E−02−671707--1.23E−02−6856303.82E−0293013.51E−02−695515
1560115700MT-CYB----4.33E−02−746270----
1570115800MT-CYB----------
1580115900MT-CYB/MT-TT----------
1590116000MT-TT/MT-TP----------
1600116100MT-TP----------
1610116200D-Loop--3.40E−033014------
16201 16300 D-Loop 4.64E−02 380407 2.50E−02 1117943 3.68E−02 296890 ----
16301 16400 D-Loop 3.65E−02 −941720 1.38E−02 −112453 2.73E−02 −177146 ----
16401 16500 D-Loop 4.29E−02 444051 2.20E−02 1151402 1.68E−02 313156 ----
16501 16600 D-Loop 3.71E−02 784572 2.10E−02 764994 2.46E−02 836662 ----

Shown is the location of the DMR within the mitochondrial genome (ChrM) (based on GENCODE), the gene(s) residing within the 100 bp window, and p value from paired t tests between each of the five cortical regions: Brodmann areas 8, 9 and 10 (BA8, BA9, BA10), entorhinal cortex (ECX) and superior temporal gyrus (STG) compared to the cerebellum (CER). Results are displayed in order of genomic position. RPKM and corresponding p values are shown for windows if p < 0.05. Key: - denotes data not significant (p > 0.05); ND denotes not determined as the window was not included in analysis due to removal in NUMT pipeline. Results shown in bold represent those found to be present across all prefrontal cortex areas (BA8, BA9, BA10)

Fig. 3

DNA methylation differences are seen in the mitochondrial genome between brain regions and blood. Average raw RPKM values across the mitochondrial genome for each individual cortical brain region alongside matched blood and cerebellum samples are shown in the top panel, with gene positions downloaded from GENCODE shown in the middle panel. For each 100 bp window, paired t tests were performed to compare each cortical brain region and the cerebellum, with -log10 (p) shown in the bottom panel. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, RPKM reads per kilobase of transcript per million mapped reads, STG superior temporal gyrus. Red dashed line denotes the Bonferroni significance, whilst blue dashed line denotes p < 0.05 in the lower panel

List of DMRs identified between five anatomically discreet cortical regions and cerebellum Shown is the location of the DMR within the mitochondrial genome (ChrM) (based on GENCODE), the gene(s) residing within the 100 bp window, and p value from paired t tests between each of the five cortical regions: Brodmann areas 8, 9 and 10 (BA8, BA9, BA10), entorhinal cortex (ECX) and superior temporal gyrus (STG) compared to the cerebellum (CER). Results are displayed in order of genomic position. RPKM and corresponding p values are shown for windows if p < 0.05. Key: - denotes data not significant (p > 0.05); ND denotes not determined as the window was not included in analysis due to removal in NUMT pipeline. Results shown in bold represent those found to be present across all prefrontal cortex areas (BA8, BA9, BA10) DNA methylation differences are seen in the mitochondrial genome between brain regions and blood. Average raw RPKM values across the mitochondrial genome for each individual cortical brain region alongside matched blood and cerebellum samples are shown in the top panel, with gene positions downloaded from GENCODE shown in the middle panel. For each 100 bp window, paired t tests were performed to compare each cortical brain region and the cerebellum, with -log10 (p) shown in the bottom panel. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, RPKM reads per kilobase of transcript per million mapped reads, STG superior temporal gyrus. Red dashed line denotes the Bonferroni significance, whilst blue dashed line denotes p < 0.05 in the lower panel

A number of differentially methylated regions in mtDNA can be observed between the cortex and cerebellum

We were also interested to see whether total cortical tissue was significantly different to matched cerebellum samples. Given the paired nature of the different anatomical regions of the cortex, we used a multilevel mixed effects model to compare total cortex to cerebellum (see the Methods section). This analysis revealed 48 nominally significant (p < 0.05) windows (Table 3; Fig. 4), of which eight passed the Bonferroni correction (Table 3, bold face). Interestingly, three of these eight were adjacent to each other, lying between 10301 and 10600 bp and covering MT-ND3/MT-ND4L and MT-TR. We also saw a Bonferroni significant difference in DNA methylation in the D-Loop, where we earlier noted DNA methylation changes across all three Brodmann area regions.
Table 3

List of DMRs identified between total cortex and cerebellum

Start (bp)Stop (bp)Gene(s) p value
1100D-Loop-
101200D-Loop-
201300D-Loop-
301400D-Loop-
401500D-Loop7.99E−03
501600MT-TF4.49E−03
601700MT-TF/MT-RNR11.16E−02
701800MT-RNR13.22E−03
801 900 MT-RNR1 1.91E−04
9011000MT-RNR1ND
10011100MT-RNR1ND
11011200MT-RNR1ND
12011300MT-RNR1ND
13011400MT-RNR1ND
14011500MT-RNR1ND
15011600MT-RNR1ND
16011700MT-RNR1/MT-TV/MT-RNR2ND
17011800MT-RNR2ND
18011900MT-RNR2ND
19012000MT-RNR2ND
20012100MT-RNR2ND
21012200MT-RNR2ND
22012300MT-RNR2ND
23012400MT-RNR28.10E−03
24012500MT-RNR27.50E−03
25012600MT-RNR25.14E−03
26012700MT-RNR2ND
27012800MT-RNR2ND
28012900MT-RNR2-
29013000MT-RNR2-
30013100MT-RNR2-
31013200MT-RNR2-
32013300MT-RNR2/MT-TL1.51E−03
33013400MT-TL1/MT-ND1-
34013500MT-ND11.37E−02
35013600MT-ND14.27E−03
36013700MT-ND15.56E−03
37013800MT-ND18.21E−03
38013900MT-ND1-
39014000MT-ND1-
4001 4100 MT-ND1 3.07E−06
41014200MT-ND1-
42014300MT-ND1/MT-TIND
43014400MT-TI/MT-TQND
44014500MT-TM/MT-ND2ND
45014600MT-ND2ND
46014700MT-ND2ND
47014800MT-ND2ND
48014900MT-ND2ND
49015000MT-ND2ND
50015100MT-ND2ND
51015200MT-ND2ND
52015300MT-ND2ND
53015400MT-ND2ND
54015500MT-ND2ND
55015600MT-ND2/MT-TW/MT-TAND
56015700MT-TA/MT-TNND
57015800MT-TN/MT-TCND
58015900MT-TC/MT-TYND
59016000MT-CO1ND
60016100MT-CO1ND
61016200MT-CO1ND
62016300MT-CO1ND
63016400MT-CO1ND
64016500MT-CO1ND
65016600MT-CO1ND
66016700MT-CO1ND
67016800MT-CO1ND
68016900MT-CO1ND
69017000MT-CO1ND
70017100MT-CO1ND
71017200MT-CO1ND
72017300MT-CO1ND
73017400MT-CO1ND
74017500MT-CO1/MT-TS1ND
75017600MT-TS1/MT-TD/MT-CO2ND
76017700MT-CO2ND
77017800MT-CO2ND
78017900MT-CO2ND
79018000MT-CO2ND
80018100MT-CO2ND
81018200MT-CO2ND
82018300MT-CO2/MT-TKND
83018400MT-TK/MT-ATP8ND
84018500MT-ATP8ND
85018600MT-ATP8/MT-ATP6ND
86018700MT-ATP6ND
87018800MT-ATP6ND
88018900MT-ATP6ND
89019000MT-ATP6ND
90019100MT-ATP6ND
91019200MT-ATP6ND
92019300MT-ATP6/MT-CO3ND
93019400MT-CO3ND
94019500MT-CO3ND
95019600MT-CO3ND
96019700MT-CO3ND
97019800MT-CO3ND
98019900MT-CO3-
990110000MT-CO3/MT-TG-
1000110100MT-TG/MT-ND31.81E−02
1010110200MT-ND31.39E−02
10201 10300 MT-ND3 3.53E−04
10301 10400 MT-ND3/MT-TR/MT-ND4L 1.19E−05
10401 10500 MT-ND4L 2.61E−04
1050110600MT-ND4L9.05E−04
1060110700MT-ND4LND
1070110800MT-ND4L/MT-ND4ND
1080110900MT-ND4ND
1090111000MT-ND4ND
1100111100MT-ND4ND
1110111200MT-ND4ND
11201 11300 MT-ND4 8.86E−05
1130111400MT-ND41.65E−03
1140111500MT-ND4ND
1150111600MT-ND4ND
1160111700MT-ND4ND
1170111800MT-ND4ND
1180111900MT-ND4ND
1190112000MT-ND4ND
1200112100MT-ND4ND
1210112200MT-ND4/MT-THND
1220112300MT-TS2/MT-TL2ND
1230112400MT-TL2/MT-ND5-
1240112500MT-ND5-
1250112600MT-ND5-
1260112700MT-ND5-
1270112800MT-ND51.81E−03
1280112900MT-ND51.05E−02
1290113000MT-ND58.23E−03
1300113100MT-ND5ND
1310113200MT-ND5ND
1320113300MT-ND5ND
1330113400MT-ND51.44E−03
1340113500MT-ND59.13E−04
1350113600MT-ND51.61E−02
1360113700MT-ND53.89E−02
13701 13800 MT-ND5 2.77E−04
1380113900MT-ND52.80E−03
1390114000MT-ND51.88E−02
1400114100MT-ND59.31E−03
1410114200MT-ND5/MT-ND6-
1420114300MT-ND6-
1430114400MT-ND61.04E−02
1440114500MT-ND61.99E−02
1450114600MT-ND62.26E−02
1460114700MT-ND6/MT-TE3.82E−03
1470114800MT-TE/MT-CYB1.92E−02
1480114900MT-CYBND
1490115000MT-CYBND
1500115100MT-CYBND
1510115200MT-CYB3.30E−02
1520115300MT-CYB-
1530115400MT-CYB-
1540115500MT-CYB8.52E−04
1550115600MT-CYB7.43E−04
1560115700MT-CYB1.16E−02
1570115800MT-CYB2.24E−02
1580115900MT-CYB/MT-TT-
1590116000MT-TT/MT-TP-
1600116100MT-TP-
1610116200D-Loop2.23E−03
16201 16300 D-Loop 5.02E−04
1630116400D-Loop1.84E−03
1640116500D-Loop1.12E−03
1650116600D-Loop2.20E−03

Shown is the location of the DMR within ChrM (based on GENCODE), the gene(s) residing within the 100 bp window, and p value from a multilevel mixed effects model. Results are displayed in order of genomic position. RPKM and corresponding p values are shown for windows if p < 0.05. Key: - denotes data not significant (p > 0.05); ND denotes not determined as the window was not included in analysis due to removal in NUMT pipeline; bold denotes windows that reached our Bonferroni significant threshold of p < 7.04E−04

Fig. 4

DNA methylation differences are seen in the mitochondrial genome between the cerebellum and cortex. RPKM values in the total cortex and cerebellum across the mitochondrial genome are shown in the top panel, with gene positions downloaded from GENCODE shown in the middle panel. For each 100 bp window, paired t tests were performed to compare the cortex to the cerebellum, with -log10 (p) shown in the bottom panel. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, RPKM reads per kilobase of transcript per million mapped reads, STG superior temporal gyrus. Red dashed line denotes the Bonferroni significance, whilst blue dashed line denotes p < 0.05 in the lower panel

List of DMRs identified between total cortex and cerebellum Shown is the location of the DMR within ChrM (based on GENCODE), the gene(s) residing within the 100 bp window, and p value from a multilevel mixed effects model. Results are displayed in order of genomic position. RPKM and corresponding p values are shown for windows if p < 0.05. Key: - denotes data not significant (p > 0.05); ND denotes not determined as the window was not included in analysis due to removal in NUMT pipeline; bold denotes windows that reached our Bonferroni significant threshold of p < 7.04E−04 DNA methylation differences are seen in the mitochondrial genome between the cerebellum and cortex. RPKM values in the total cortex and cerebellum across the mitochondrial genome are shown in the top panel, with gene positions downloaded from GENCODE shown in the middle panel. For each 100 bp window, paired t tests were performed to compare the cortex to the cerebellum, with -log10 (p) shown in the bottom panel. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, RPKM reads per kilobase of transcript per million mapped reads, STG superior temporal gyrus. Red dashed line denotes the Bonferroni significance, whilst blue dashed line denotes p < 0.05 in the lower panel

MtDNA methylation patterns can distinguish between tissue types

Although we have shown that mtDNA methylation patterns are highly similar between distinct anatomical regions of the human brain and blood, we were also interested to identify whether mtDNA methylation patterns could distinguish between these tissue types. Through unsupervised hierarchical clustering, we showed that average mtDNA methylation patterns can segregate these tissues (Fig. 5a). Importantly, ncDNA methylation profiles in the same samples have also been previously shown to separate the cortex, cerebellum and blood [35]. Interestingly, when we performed unsupervised hierarchical clustering on the individual samples, we found that, in most cases, intra-individual differences across tissue types are greater than inter-individual differences within each tissue type, as the cortex, cerebellum and blood samples clustered with their own tissue type, respectively (Fig. 5b).
Fig. 5

MtDNA methylation patterns can distinguish between tissue types. a Average RPKM values for each cortical brain region, cerebellum and blood samples were clustered based upon the Euclidean distance, identifying two major clusters; the cortex and blood-cerebellum. b When clustering RPKM values in the individual samples from the cortex, cerebellum and blood, we observed that individual cortex samples clustered together, whilst cerebellum and blood samples formed separate clusters. This highlights that tissue-specific differences between the cortex, cerebellum and blood are greater than intra-individual variability within a tissue. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, RPKM reads per kilobase of transcript per million mapped reads, STG superior temporal gyrus

MtDNA methylation patterns can distinguish between tissue types. a Average RPKM values for each cortical brain region, cerebellum and blood samples were clustered based upon the Euclidean distance, identifying two major clusters; the cortex and blood-cerebellum. b When clustering RPKM values in the individual samples from the cortex, cerebellum and blood, we observed that individual cortex samples clustered together, whilst cerebellum and blood samples formed separate clusters. This highlights that tissue-specific differences between the cortex, cerebellum and blood are greater than intra-individual variability within a tissue. BLD blood, BA8 Brodmann area 8, BA9 Brodmann area 9, BA10 Brodmann area 10, CER cerebellum, CTX cortex, ECX entorhinal cortex, RPKM reads per kilobase of transcript per million mapped reads, STG superior temporal gyrus

Discussion

The availability of publicly available epigenomic data provides a great resource for mitochondrial epigenetics, a field that is relatively nascent and has yet to be thoroughly investigated in a range of complex diseases. Here, we present evidence that mtDNA methylation patterns across mtDNA are brain region specific. However, data such as that presented here is confounded by a lack of isolation of mtDNA prior to antibody enrichment and sequencing. As such, the potential of including NUMTs in datasets derived from data generated using total genomic DNA could lead to misleading results. Here, we controlled for regions of high sequence homology between the nuclear and mitochondrial genomes. However, this approach is likely over-conservative and does lead to the generation of a somewhat truncated consensus sequence. PCA of the mitochondrial epigenome after corrections for nuclear homology was able to separate individuals belonging to the three main tissue types, the blood, cortex and cerebellum based on mtDNA methylation variation among tissue types. This tissue specificity is further highlighted by the identification of eight DMRs that pass the Bonferroni correction for multiple testing between total cortex and cerebellum. MtDNA methylation has been shown to be cell line dependent in the past. [31] Although overall DNA methylation levels were low in all tissues, it is worth noting that the study was performed on non bisulfite-treated DNA. As such, the low percentage of mtDNA methylation is not a pitfall due to a lack of a total bisulfite treatment efficiency. One limitation of the current study is the unavailability of publicly available MeDIP-seq datasets of matched cortical and cerebellum tissue from other cohorts for validation purposes. Future work would aim to replicate our findings in additional study cohorts and also to investigate the relationship between mitochondrial DNA methylation and gene expression. Despite a number of nominally significant windows being identified between each individual cortical region and the cerebellum, these did not pass the Bonferroni correction, although it is likely this method is too stringent. Nevertheless, the conservation of seven nominally significant windows across each Brodmann area is interesting to note. Four of these windows lie adjacent to each other and correspond to the mitochondrial D-Loop, a region containing the only two mitochondrial promoters which is typically associated with gene transcription and DNA replication. However, one limitation of this study is owed to the use of antibody-based enrichment, resulting in the analysis being limited to a window-based approach. Despite this, studies of the nuclear genome have shown high correlation between window-based approaches and, more sensitive, single-site assays such as the Illumina 450K beadarray [32]. However, given the small size of the mitochondrial genome and that 23 of the 37 genes present in the genome are below 100 bp in size, this window-based approach may not be the most appropriate for future studies designed to specifically assess mtDNA methylation as it can result in a window intersecting two genes in the polycistronic transcript.

Conclusions

This method provides a conservative approach to determine mtDNA methylation across the genome for data previously generated using next-generation sequencing approaches such as MeDIP-seq. Its conservative nature reduces the risk of the inclusion of NUMTs in the final analysis of whole genome data but may also lead to the inclusion of false negatives as well as potential gaps in the reference sequence. As such, it is best suited to analyzing previously generated whole genome data and is not a replacement for the isolation of mitochondrial DNA [36] prior to targeted methylation studies, which would be the optimal approach for investigating mitochondrial epigenomics. However, our method has allowed the identification of novel brain-region-specific DMRs in a previously generated publicly available dataset. Furthermore, the identification of brain region-specific mtDNA methylation patterns across the mitochondrial epigenome suggests the importance of a focussed, tissue-specific study design when investigating mtDNA methylation. As previously discussed, one caveat when utilizing MeDIP-seq data is the segregation of data into neighbouring windows, meaning that determining the exact corresponding gene of a DMR is difficult and, as such, future studies should aim to sequence the mitochondrial DNA methylome at single-base resolution to address this.

Methods

Data collection

We utilized publicly available MeDIP-seq data from Davies et al. [35]. In brief, this data was generated using 5 μg fragmented gDNA, which, following end repair base addition and adaptor ligation, was immunoprecipitated using an anti-5-mC antibody (Diagenode, Liège, Belgium). MeDIP DNA was purified and then amplified using adaptor-mediated PCR, with DNA fragments between 220 and 320 bp subjected to highly parallel 50 bp paired-end sequencing on the Illumina Hi-Seq platform. The paired-end, raw fasta files were provided by the authors and quality checked using FastQC. Sample information is provided in Table 1.

Quality control and NUMT exclusion

Fasta files were subjected to adaptor and Phred score (q < 20) trimming. In an attempt to remove any potential contamination of possible NUMTs, multiple alignments to the reference genome were undertaken. Paired fasta files were aligned to GRCH37 using BWA. Unique and mapped reads aligning to the mitochondria were then re-mapped to a custom GRCH37 reference without the mitochondrial chromosome. Reads not mapping to the custom reference were then taken forward and realigned to the full GRCH37 reference to eliminate the possibility of homologous regions mapping falsely to the mitochondrial genome (Fig. 1). All alignments were carried out using BWA mem and default settings. Reads per kilobase of transcript per million mapped reads (RPKM) values for each sample were calculated using the MEDIPS package [37]. Methylation was averaged across 100 bp non-overlapping windows (default parameter setting in MEDIPS), and only windows with read counts >10 were considered for analysis. Due to the non-normal distribution of all cohorts, RPKM values were log2 transformed before statistical analysis.

Statistical analyses

All analyses were performed in the R statistical environment version 3.2.1 [38]. For all analyses, a nominally significant threshold of p < 0.05 and a Bonferroni significant threshold of p < 7.04E−04 were used. Given the matched sample nature of this cohort, two-tailed, paired t tests were performed at each window along the mitochondrial genome to identify DMRs between the individual cortical regions and cerebellum. To compare the total cortex to cerebellum, we performed a multilevel mixed effects model in the Lme4 package in R [39], using the brain region as the random effect and individual as the fixed effect. To assess the similarity of the brain regions, we used the R function “hclust” to cluster average RPKM values for the brain regions using the Euclidean distance. We used the R function “corrgram” within the corrgram package [40] to order samples based upon the similarity of their principal components.
  38 in total

1.  Impairment of methyl cycle affects mitochondrial methyl availability and glutathione level in Down's syndrome.

Authors:  Vittoria Infantino; Alessandra Castegna; Francesco Iacobazzi; Iolanda Spera; Iris Scala; Generoso Andria; Vito Iacobazzi
Journal:  Mol Genet Metab       Date:  2010-12-09       Impact factor: 4.797

2.  Type 2 diabetes and leucocyte DNA methylation: an epigenome-wide association study in over 1,500 older adults.

Authors:  Ines Florath; Katja Butterbach; Jonathan Heiss; Melanie Bewerunge-Hudler; Yan Zhang; Ben Schöttker; Hermann Brenner
Journal:  Diabetologia       Date:  2015-10-03       Impact factor: 10.122

3.  Sequence and organization of the human mitochondrial genome.

Authors:  S Anderson; A T Bankier; B G Barrell; M H de Bruijn; A R Coulson; J Drouin; I C Eperon; D P Nierlich; B A Roe; F Sanger; P H Schreier; A J Smith; R Staden; I G Young
Journal:  Nature       Date:  1981-04-09       Impact factor: 49.962

4.  Mitochondrial DNA damage level determines neural stem cell differentiation fate.

Authors:  Wei Wang; Ying Esbensen; David Kunke; Rajikala Suganthan; Lyudmila Rachek; Magnar Bjørås; Lars Eide
Journal:  J Neurosci       Date:  2011-06-29       Impact factor: 6.167

Review 5.  Epigenetic studies in Alzheimer's disease: current findings, caveats, and considerations for future studies.

Authors:  Katie Lunnon; Jonathan Mill
Journal:  Am J Med Genet B Neuropsychiatr Genet       Date:  2013-09-13       Impact factor: 3.568

6.  Mitochondria DNA change and oxidative damage in clinically stable patients with major depressive disorder.

Authors:  Cheng-Chen Chang; Shaw-Hwa Jou; Ta-Tsung Lin; Te-Jen Lai; Chin-San Liu
Journal:  PLoS One       Date:  2015-05-06       Impact factor: 3.240

7.  Two-stage genome-wide methylation profiling in childhood-onset Crohn's Disease implicates epigenetic alterations at the VMP1/MIR21 and HLA loci.

Authors:  Alex T Adams; Nicholas A Kennedy; Richard Hansen; Nicholas T Ventham; Kate R OʼLeary; Hazel E Drummond; Colin L Noble; Emad El-Omar; Richard K Russell; David C Wilson; Elaine R Nimmo; Georgina L Hold; Jack Satsangi
Journal:  Inflamm Bowel Dis       Date:  2014-10       Impact factor: 7.290

8.  DNA methylome profiling of human tissues identifies global and tissue-specific methylation patterns.

Authors:  Kaie Lokk; Vijayachitra Modhukur; Balaji Rajashekar; Kaspar Märtens; Reedik Mägi; Raivo Kolde; Marina Koltšina; Torbjörn K Nilsson; Jaak Vilo; Andres Salumets; Neeme Tõnisson
Journal:  Genome Biol       Date:  2014-04-01       Impact factor: 13.583

9.  Methylation QTLs in the developing brain and their enrichment in schizophrenia risk loci.

Authors:  Eilis Hannon; Helen Spiers; Joana Viana; Ruth Pidsley; Joe Burrage; Therese M Murphy; Claire Troakes; Gustavo Turecki; Michael C O'Donovan; Leonard C Schalkwyk; Nicholas J Bray; Jonathan Mill
Journal:  Nat Neurosci       Date:  2015-11-30       Impact factor: 24.884

10.  Variation in 5-hydroxymethylcytosine across human cortex and cerebellum.

Authors:  Katie Lunnon; Eilis Hannon; Rebecca G Smith; Emma Dempster; Chloe Wong; Joe Burrage; Claire Troakes; Safa Al-Sarraj; Agnieszka Kepa; Leonard Schalkwyk; Jonathan Mill
Journal:  Genome Biol       Date:  2016-02-16       Impact factor: 13.583
View more
  14 in total

Review 1.  Cause or casualty: The role of mitochondrial DNA in aging and age-associated disease.

Authors:  E Sandra Chocron; Erin Munkácsy; Andrew M Pickering
Journal:  Biochim Biophys Acta Mol Basis Dis       Date:  2018-11-09       Impact factor: 5.187

2.  Single-molecule mitochondrial DNA sequencing shows no evidence of CpG methylation in human cells and tissues.

Authors:  Iacopo Bicci; Claudia Calabrese; Zoe J Golder; Aurora Gomez-Duran; Patrick F Chinnery
Journal:  Nucleic Acids Res       Date:  2021-12-16       Impact factor: 16.971

3.  Mitochondrial genome undergoes de novo DNA methylation that protects mtDNA against oxidative damage during the peri-implantation window.

Authors:  Yuan Yue; Likun Ren; Chao Zhang; Kai Miao; Kun Tan; Qianying Yang; Yupei Hu; Guangyin Xi; Gang Luo; Mingyao Yang; Jingyu Zhang; Zhuocheng Hou; Lei An; Jianhui Tian
Journal:  Proc Natl Acad Sci U S A       Date:  2022-07-18       Impact factor: 12.779

4.  Mitochondrial DNA methylation profiling of the human prefrontal cortex and nucleus accumbens: correlations with aging and drug use.

Authors:  Chia-Hung Huang; Man-Chen Chang; Yung-Chun Lai; Chun-Yen Lin; Cho-Hsien Hsu; Bo-Yuan Tseng; Chuhsing Kate Hsiao; Tzu-Pin Lu; Sung-Liang Yu; Sung-Tsang Hsieh; Wei J Chen
Journal:  Clin Epigenetics       Date:  2022-06-25       Impact factor: 7.259

5.  Human mitochondrial DNA is extensively methylated in a non-CpG context.

Authors:  Vibha Patil; Cyrille Cuenin; Felicia Chung; Jesus R Rodriguez Aguilera; Nora Fernandez-Jimenez; Irati Romero-Garmendia; Jose Ramon Bilbao; Vincent Cahais; Joseph Rothwell; Zdenko Herceg
Journal:  Nucleic Acids Res       Date:  2019-11-04       Impact factor: 16.971

Review 6.  Mitochondrial DNA Methylation and Human Diseases.

Authors:  Andrea Stoccoro; Fabio Coppedè
Journal:  Int J Mol Sci       Date:  2021-04-27       Impact factor: 5.923

Review 7.  Mitoepigenetics and Neurodegenerative Diseases.

Authors:  Fabio Coppedè; Andrea Stoccoro
Journal:  Front Endocrinol (Lausanne)       Date:  2019-02-19       Impact factor: 5.555

8.  Altered methylation pattern of the SRD5A2 gene in the cerebrospinal fluid of post-finasteride patients: a pilot study.

Authors:  Roberto Cosimo Melcangi; Livio Casarini; Marco Marino; Daniele Santi; Samantha Sperduti; Silvia Giatti; Silvia Diviccaro; Maria Grimoldi; Donatella Caruso; Guido Cavaletti; Manuela Simoni
Journal:  Endocr Connect       Date:  2019-08-01       Impact factor: 3.335

9.  The strand-biased mitochondrial DNA methylome and its regulation by DNMT3A.

Authors:  Xiaoyang Dou; Jerome D Boyd-Kirkup; Joseph McDermott; Xiaoli Zhang; Fang Li; Bowen Rong; Rui Zhang; Bisi Miao; Peilin Chen; Hao Cheng; Jianhuang Xue; David Bennett; Jiemin Wong; Fei Lan; Jing-Dong J Han
Journal:  Genome Res       Date:  2019-09-19       Impact factor: 9.043

10.  Epigenetic interventions for brain rejuvenation: anchoring age-related transposons.

Authors:  Adonis Sfera; Lisa Fayard; Carolina Osorio; Amy Price
Journal:  Neural Regen Res       Date:  2018-04       Impact factor: 5.135
View more

北京卡尤迪生物科技股份有限公司 © 2022-2023.