Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Jun 1.
Published in final edited form as: Neuroepigenetics. 2016 Jun;6:1–9. doi: 10.1016/j.nepig.2016.03.001

Novel method to ascertain chromatin accessibility at specific genomic loci from frozen brain homogenates and laser capture microdissected defined cells

Elaine Delvaux a,b, Diego Mastroeni a,b,c, Jennifer Nolz a,b, Paul D Coleman a,b,✉
PMCID: PMC4856059  NIHMSID: NIHMS779456  PMID: 27158594

Abstract

We describe a novel method for assessing the “open” or “closed” state of chromatin at selected locations within the genome. This method combines the use of Benzonase, which can digest DNA in the presence of actin, with qPCR to define digested regions. We demonstrate the application of this method in brain homogenates and laser captured cells. We also demonstrate application to selected sites within more than one gene and multiple sites within one gene. We demonstrate the validity of the method by treating cells with valproate, known to render chromatin more permissive, and by comparison with classical digestion with DNase I in an in vitro preparation. Although we demonstrate the use of this method in brain tissue we also recognize its applicability to other tissue types.

1. INTRODUCTION

The epigenetic regulation of chromatin structure is a major component of the gene expression regulatory machinery and this regulation is accomplished by multiple mechanisms including DNA methylation and hydroxymethylation, multiple histone modifications, binding molecules and complex interactions among all these elements (Jaenisch and Bird, 2003; Mastroeni et al., 2011; Chouliaras et al., 2013). These complex interactions converge to modulate chromatin structure, much as multiple inputs converge on the final common path of the anterior horn cells (Sherrington, 1906). By analogy, we posit chromatin structure as the final common path of multiple epigenetic mechanisms.

Chromatin structure determines whether any selected segment of DNA is accessible to transcription factors (TFs), DNA binding proteins or other modifiers which govern a genes’ availability for transcription (Quina et al., 2006). There are multiple classes of methods for examining chromatin structure. Broadly speaking these are methods that utilize DNase I digestion, chromatin precipitation, matrix-assisted reader chromatin capture, etc., that are typically followed by methods to determine location of permissive or repressive chromatin structure.

In the past, permissive, or open chromatin, has been identified by DNase I nuclease sensitivity analysis (Krebs and Peterson, 2000) and more recently open chromatin data generated from DNase-seq has been able to predict cell-type-specific gene expression (Natarajan et al., 2012).

An overview of current methods available to ascertain chromatin structure yields a broad range of approaches, many of which are variations of chromatin immunoprecipitation (ChIP), an antibody-based methodology first described by Gilmore and Lis (1984, 1985). Fast carrier ChIP (Fast CChIP) can be used study to transcription factor binding in small amounts of fresh tissue (Hao et al., 2008). Another ChIP-based technique uses genome-wide chromatin immunoprecipitation in concert with tiling microarrays (ChIP-chip). ChIP-chip, along with ChIP-qPCR, showed that transcription factor FoxA1 is differentially recruited to cell-type specific enhancers (Lupien et al., 2008). ChIA-PET combines ChIP with paired-end DNA sequencing to determine global de novo chromatin interactions (Fullwood et al., 2009). PAT-ChIP, also a ChIP-directed technique, was used to isolate, extract and sequence chromatin from formalin-fixed, paraffin-embedded (FFPE) tissue samples (Fanelli et al., 2010; Fanelli et al., 2011; Amatori et al., 2014). A common method used to determine chromatin structure is DNase I digestion of DNA with subsequent sequencing of resulting fragments (DNase-seq). The sites of cleavage are called DNase I hypersensitive sites (DHS) suggesting areas of the chromatin accessible to DNase digestion, to be “open”, and poised for transcription. A recent international epigenetic collaborative study, ENCODE, created a genome-wide map of DHSs in 125 human cell lines and tissue-types and correlated those sites with ChIP-seq data (Thurman et al., 2012). In addition to genome-wide mapping of DHS, DNase I sensitivity and sequencing can be used to analyze receptor binding influences on chromatin structure at specific gene loci in vitro (Tewari, et al., 2012). More recently, a modification to ChIP-Seq, called Bar-ChIP, allowed multiple DNA-protein interactions to be simultaneously profiled by attaching molecular barcodes to chromatin fragments from S. cerevisiae (Chabbert et al., 2015).

Chromosome conformation capture (3C) technologies have been utilized for more than a decade. First described in 2002 (Dekker et al. 2002), 3C delivers 3-D chromatin structure information based on interaction frequencies between genomic loci. Some 3C technologies, such as e4C use ChIP within the protocol (Sexton et al., 2012). Capture-C, the most recent derivative of the 3C technique, modifies the 3C method into a multiplexed, high-throughput approach to analyzing cis-acting elements that control gene expression (Hughes et al., 2014).

A ChIP-less technique to examine the epigenome, called MARCC (matrix-assisted reader chromatin capture), is a platform that probes combinatorial histone modification patterns to determine functional chromatin conformation (Su et al., 2014). Finally, direct chromatin PCR (DC-PCR) is a method for gene specific chromatin analysis performed on cultured cells utilizing properties of standard PCR buffers and thermal cycling temperatures to amplify open regions of chromatin (Vatolin, et al., 2012).

Although ChIP and non-ChIP-based techniques have been broadly used in elucidating chromatin structure and its impact on gene regulation, caveats to their use exist. The ChIP-based techniques rely on antibodies, which involve problems with antibody specificity and optimization (Orlando, 2000). Moreover, most ChIP and non-ChIP-based methods need large amounts of cultured cells or fresh tissue to start with. In addition, these complicated protocols take several days to complete and can be expensive with labor, reagents, equipment and per sample core costs (Sheffield and Furey, 2012).

A major limitation of DNase I based methods is the susceptibility of DNase I to inhibition by actin, a component of many cells and tissues. Recently, a protocol was developed for determining chromatin accessibility in frozen tissue homogenates using Benzonase, a robust nuclease whose efficacy is not affected by the presence of actin (Grøntved et. al, 2012). Here we describe a protocol to assess chromatin structure at specific loci using Benzonase digestion and qPCR of DNA extracted from frozen tissue homogenates and laser capture dissected defined cells. This protocol is relatively quick, inexpensive and achievable in any research laboratory outfitted with standard real time PCR equipment. This novel method will allow a more focused examination of chromatin structure, its influence on gene expression and the impact it may have on gene expression in normal and diseased brain as well as other tissues.

2. MATERIALS AND METHODS

2.1 Tissue (Samples)

All frozen brain tissue samples and frozen brain sections were obtained from the Brain Bank at Banner Sun Health Research Institute. Average postmortem delay at this Brain Bank is 2.6 hours. Homogenate studies used approximately 1 gram of middle temporal gyrus (MTG) and cerebellum (CBL) from each of 3 non-demented controls (ND) and 3 Alzheimer’s disease (AD) cases. LCM studies used 10 µm serial sections of middle temporal gyrus (area 22) from 1 ND and 1 AD case and 10 µm serial sections of hippocampus from 5 ND and 5 AD cases.

2.2 Primers

Primer Design for Chromatin Accessibility Analysis

Primers for KAT6B, APP and RSP28 were designed to cover the area 300 bp upstream of the transcription start site (TSS) and 100 bp downstream of TSS. The primer sequences for GAPDH, CTCF4 and CTCF10 were taken from He et al., (2014) and chosen to represent areas of the genome known to be constitutively open (GAPDH) and/or uniform in openness across sample-types (CTCF4, CTCF10) (Li et al., 2013). All primers were checked for specificity via NCBI’s BLAST search.

To examine chromatin structure in multiple regions of one gene, primers for ARC were designed based on ENCODE data (Thurman et al., 2012). ARC1 was located in the promoter region, an area identified with high significance as open chromatin and validated by DNase I HS (hypersensitivity) and FAIRE (Formaldehyde Assisted Isolation of Regulatory Elements) assays. This area contained DNase I hypersensitivity in 113 out of 125 cell lines/types. Transcription factor ChIP seq data shows it to be densely populated with DNA binding proteins (transcription factors, transcription machinery components, chromatin factors) with varying degrees of occupancy. ARC2 was located adjacent to ARC1, breaching the start of the coding sequence. This area was validated by both DNase I HS and FAIRE assays with DNase I hypersensitivity in 16 out of 125 cell lines/types. ChIP seq data identified several DNA binding proteins with varying degrees of occupancy. ARC3 was located near the end of the gene in the 3rd exon. This area contained DNase I hypersensitivity in 10 out of 125 cell lines/types. This region was considered high significance as open chromatin but not validated by both DNase I and FAIRE assays. Transcription factor CTCF occupancy is shown to be weak. ARC4 was designed in an area that could be considered a “control”; an intronic region between exon 2 and 3.

To determine the optimal annealing/extension temperatures for all primer sets designed, a gradient PCR with melting curve was run. Based on the results of the gradient, a standard curve PCR was performed to test the PCR efficiency. Efficiency greater than 90% was acceptable. After optimizing PCR protocols, the samples were run within those parameters. Primer sequences are found in Table 1.

Table 1.

Primer sequences used for qPCR to determine chromatin accessibility.

Gene FORWARD REVERSE
KAT6B CCCCAATGGGCTGCAGAGTGGTTAG CGCTGCCGCTGCCTGAGAAACT
APP CTCCCACTGTTCACGAAGCCCAGGT CTACCGCTGCCGAGGAAACTGACG
ARC1 GGGCCTCGCTGGCTGCATAAAGAG GTCCGGTGGTCCAGCTCCATCTGT
ARC2 ACTCGCAGCGCTGGAAGAAGTCCAT GCGTAGGGGCTGACGGTGTAGTCGT
ARC3 TGGTCCTTCACTGCCCACTCTCCTG CTAAGCTGGGGCTCCTGCCCTCTG
ARC4 ATGAGCTCCTCCCCAGACCCCAGAC GCAGTGAAGGACCACGCAGGACAGT
RPS28 AGAGCGAAGGGTCCCGGCTTAGGAG AGAGGAGTCACGTGCTTCGGGGAGA
GAPDH AAAAGCGGGGAGAAAGTAGG GCTGCGGGCTCAATTTATAG
CTCF4 CCCCAGAGAGTAGGGAACAG GGCACGCAAAGACATACTGA
CTCF10 AGAGCACCCCCTACTGGCTAA TAAGAAGCTGTGCGCGATGAC
Open in a new tab

2.3 Brain Homogenates

Benzonase Digestion (Adapted from Grøntved et al., (2012))

Approximately 600 mg of frozen brain tissue was pulverized with a mortar and pestle that had been pre-chilled on dry ice. The pulverized tissue was transferred to a pre-chilled 15 mL tube then suspended in 4 mL of lysis buffer (30 mM TrisHCL pH 8.0, 2 mM EDTA, 2 mM EGTA, 20 mM sodium butyrate, 2 mM sodium orthovanadate, 4 mM sodium fluoride, protease inhibitor cocktail (Pierce) using a syringe fitted with a 20 gauge needle. Volume was brought up to 6 mL with lysis buffer. Using a 23 gauge needle, the suspension was mixed and distributed in 500 µl equal volume aliquots into 2 mL microfuge tubes and incubated on ice for 5 minutes. Aliquots not needed were frozen at 20°C for later use. To the aliquots being used, 500 µl of nuclease digestion buffer (30 mM TrisHCL pH8.0, 14 mM MgCl2, 0.5% NP-40, 0.2% fatty acid-free BSA, 0.25 U/ µl Benzonase (EMD Millipore)) was added. A no nuclease control sample was also prepared. The tubes were gently inverted to mix and incubated at 37oC for 3 minutes. Immediately after the incubation, EDTA (at 50 mM final concentration) and SDS (at 0.1% final concentration) were added to stop the digestion reaction. 50 µl of RNaseA/RNaseT1 (Ambion/Life Technologies) was added and incubated overnight at 45°C. After incubation, SDS was added to final concentration of 0.75% and placed at 45°C for 2 hours. Proteinase K (Sigma) was then added to final concentration of 0.8 µg/mL and incubated over night at 45°C. Digested and undigested DNA reactions were purified by gel extraction (Wizard SV Gel and PCR Clean Up, Promega) or phenol chloroform extraction and ethanol precipitation then quantified with Quant-iT PicoGreen Assay (Life Technologies).

Chromatin Accessibility Analysis by qPCR (Adapted from Rao et al., (2001))

5 ng of digested and undigested DNA from each sample was amplified in triplicate for 50 cycles on an iCycler iQ (Bio-Rad) using EpiQ SYBR Green Supermix (Bio-Rad) (This supermix was discontinued in early stages of protocol design. iTaq Universal SYBR Green Supermix (Bio-Rad) was used for remaining experiments) and 2 µM primers under the optimized parameters for each primer set. In the resulting PCR analyses higher CT values indicate more “open” chromatin accessible to regulatory elements.

2.4 Laser Capture Microdissected Derived Cells

Benzonase Digestion of Many Cells

10 µm frozen sections of hippocampus from (5) Alzheimer (AD) and (5) non-demented control (ND) cases were mounted on PEN slides (Leica) and kept under desiccant at − 80°C until use. After a brief thaw, sections were placed in 100% EtOH then 95% EtOH and a final 70% ethanol for 5 seconds each then rinsed with molecular grade water. Neutral red stain (1%) was applied for 30 seconds, then rinsed twice with molecular grade water and dehydrated for 5 seconds in 70% EtOH then 95% EtOH and for 1 minute in 100 % ethanol. Laser capture microdissection of dentate gyrus granule cells was performed on an LMD 6500 (Leica). Briefly, 40 µl of lysis buffer (20 mM TrisHCL pH 7.5, 2 mM EDTA, 1 mM EGTA, 0.5% glycerol, 20 mM sodium butyrate, 2 mM sodium orthovanadate, 4 mM sodium fluoride, protease inhibitor cocktail (Pierce)) was added to the cap of the tube where the sample was captured. After a 1 minute spin, another 40 µl of lysis buffer was added to the cap. The tube was set upside down on a bench for 20 minutes then spun down in a bench top centrifuge for 1 minute, then split equally (40 µl) to two tubes for no enzyme control and enzyme digestion. For enzyme digestion, 40 µl nuclease digestion buffer (40 mM TrisHCL pH8.0, 6 mM MgCl2, 0.3% NP-40, 1% glycerol, 0.25 U/µl Benzonase (EMD Millipore)) was added. The no nuclease control tube received 40 µl of nuclease digestion buffer minus the Benzonase. Both tubes were placed @ 37°C for 2.5 minutes. To stop the reaction, EDTA (10 mM final) and SDS (0.75% final) was added to each tube. Then 4 µl RNaseA/RNase T1 was added and the tubes were placed at 45°C overnight. The next day SDS at 0.75% final concentration was added and incubated for 2 hours at 45°C. Next, Proteinase K (0.5 µg/mL final) was added and incubated overnight at 45°C. The digested and undigested DNA reactions were purified by phenol chloroform extraction and ethanol precipitation.

Chromatin Accessibility Analysis by qPCR of Many Cells (dentate gyrus granule cells)

Digested and undigested DNA from each sample was amplified in duplicate for 50 cycles on an iCycler iQ (Bio-Rad) using iTaq Universal SYBR Green Supermix (Bio-Rad) and 2 µM primers under the optimized parameters for each primer set.

Benzonase Digestion of Few Cells

10 µm frozen serial sections of middle temporal gyrus (MTG) from (1) Alzheimer (AD) and (1) non-demented control (ND) case were mounted on PEN slides (Leica) and kept under desiccant at −80°C until use. After brief thaw, sections were stained with 1% neutral red to identify pyramidal neurons. Neurofibrillary tangle-bearing neurons were identified by staining an adjacent section with 0.1 % Thioflavin S for 15 seconds. Non-tangle bearing pyramidal neurons were chosen after comparing the adjacent sections and excluding tangle-bearing cells. Laser capture microdissection of non-tangle neurons was performed on a PALM MicroBeam (Zeiss) where approximately 50 cells were captured into AdhesiveCap tubes (Zeiss). 20 µl of lysis buffer (20 mM TrisHCL pH 7.5, 2 mM EDTA, 1 mM EGTA, 0.5% glycerol, 20 mM sodium butyrate, 2 mM sodium orthovanadate, 4 mM sodium fluoride, HALT protease inhibitor cocktail (Pierce)) was added to the cap of the tube. The tube was set upside down on a bench for 20 minutes then spun down in a bench top centrifuge for 1 minute. An additional 10 µl of lysis buffer was added to the tube, then split equally (~ 15 µl) to two tubes for no enzyme control and enzyme digestion. For enzyme digestion, 15 µl nuclease digestion buffer (40 mM TrisHCL pH8.0, 6 mM MgCl2, 0.3% NP-40, 1% glycerol, 0.25 U/ µl Benzonase (EMD Millipore)) was added. The no nuclease control tube received 15 µl of nuclease digestion buffer minus the Benzonase. Both tubes were placed at 37°C for 2.5 minutes. To stop the reaction, EDTA (10 mM final) and SDS (0.75% final) was added to each tube. Then 1.5 µl RNaseA/RNase T1 was added and the tubes were placed at 45°C overnight. The next day SDS at 0.75% final concentration was added and incubated for 2 hours at 45°C. Next, Proteinase K (0.5 µg/mL final) was added and incubated overnight at 45°C. The digested and undigested DNA reactions were purified by phenol chloroform extraction and ethanol precipitation.

Chromatin Accessibility Analysis by qPCR of Few Cells (non-tangle bearing neurons)

Approximate equivalents of 5 cells underwent qPCR on an iCycler iQ (Bio-Rad) with the same APP primers and conditions as with frozen homogenates protocol except the cycle number was extended to 80 cycles due to the minute amount of starting material. A 1.5% agarose gel was run to verify amplification of 260 bp amplicon (data not shown).

2.5 Validation in vitro by Treatment of Cells with A Reagent Known to Lead to Open, Permissive Chromatin

Valproic Acid Treatment to Decondense Chromatin in SH-SY5Y Cells

SH-SY5Y neuroblastoma cells (ATCC) were cultured and maintained in a humidified 37°C incubator with 5% CO2 in complete DMEM (DMEM with high glucose (Life Technologies), 10% FBS (Gemini Bio-Products), 2% HEPES (Irvine Scientific), 1% sodium pyruvate (Thermo Scientific), 1% penicillin/streptomycin (Life Technologies), and 0.1% gentamycin (Life Technologies)). After 4 days in culture cells were seeded at a density of 2.6 × 105 cells/mL in 6-well plates. Two days later they were fed with complete DMEM and half the wells were treated with 1 mM valproic acid (Sigma), an HDAC inhibitor known to decondense chromatin (Felisbino et al., 2011; Felisbino et al., 2014) and half were left untreated. After incubating for 24 hours at 37°C and 5% CO2 they were washed twice in 1X PBS, collected by centrifugation then washed twice in ice cold wash buffer (20 mM TrisHCL pH 7.5, 137 mM NaCl, 1 mM EDTA, 10 mM sodium butyrate, 10 mM sodium orthovanadate, 2 mM sodium fluoride, 1X HALT protease inhibitor cocktail (Pierce)) and resuspended in 1 mL lysis buffer (20 mM TrisHCl pH 7.5, 2 mM EDTA, 1 mM EGTA, 20 mM sodium butyrate, 2 mM sodium orthovanadate, 4 mM sodium fluoride, 0.5% glycerol and 1X HALT protease inhibitor cocktail (Pierce)) and aliquot 500 µl to two tubes; one for enzyme digestion and one no enzyme control. 500 µl of nuclease digestion buffer (40 mM TrisHCl pH 8.0, 6mM MgCl2, 0.3% NP-40 and 1% glycerol) was added to each tube. Benzonase was added to appropriate tubes for a final concentration of 0.5U/mL. Tubes were incubated at 37°C for 3 minutes. To stop the reaction, EDTA (10 mM final) and SDS (0.75% final) was added to each tube. Next, Proteinase K (0.5 µg/mL final) was added and incubated overnight at 45°C. The digested and undigested DNA reactions were purified by phenol chloroform extraction and ethanol precipitation and quantified by Quant-iT PicoGreen Assay (Life Technologies). Digested and undigested DNA (5 ng) from each sample was amplified in duplicate on an iCycler iQ (Bio-Rad) using iTaq Universal SYBR Green Supermix (Bio-Rad) and 2 µM primers as per the optimized parameters determined by gradient and melting curve PCR.

2.6 Validation in vitro By Comparison of Benzonase with DNase I Digestion in SH-SY5Y Cells

SH-SY5Y cells were cultured and maintained as described above. Four concentrations of Benzonase (None, 0.33U, 1U, 3U) were used to digest 3.4×106 cells as detailed above. The DNase I digestion protocol was based on Song and Crawford (2010). For DNase I digestion, 3.7 × 107 cells were pelleted in a 15 mL tube then resuspended by gentle flicking in 1 mL ice cold RSB (10 mM TrisHCl pH 7.4, 10 mM NaCl, 3 mM MgCl2). Lysis buffer (RSB + 0.1% IGEPAL CA-630 (Sigma)) was poured slowly into the tube then centrifuged (500 × g) at 4°C for 10 minutes. Supernatant was completely removed and nuclei were resuspended in 600 µl of cold RSB by gentle flicking. The nuclei were aliquoted equally into 5 microfuge tubes containing different concentrations of DNase I (Sigma) (None, 0.12U, 0.4U, 1.2U, 4U). Tubes were incubated at 37°C for 11 minutes. To stop the reaction, 160 µl of 50 mM EDTA was added and tubes were inverted 5 times to mix. The digested and undigested DNA reactions were purified by phenol chloroform extraction and ethanol precipitation and quantified by Quant-iT PicoGreen Assay (Life Technologies). Digested and undigested DNA (5 ng) from each sample was amplified in duplicate on an iCycler iQ (Bio-Rad) using iTaq Universal SYBR Green Supermix (Bio-Rad) and 2 µM primers as per the optimized parameters determined by gradient and melting curve pcr.

3. RESULTS

3.1 Homogenate Samples

1 Case of Alzheimer’s Disease, Promoter Region of 2 Genes in 2 Different Brain Regions

For proof of concept, brain homogenates from middle temporal gyrus (MTG) and cerebellum (CBL) of 1 Alzheimer’s disease (AD) case were digested with Benzonase. The amplification curves of the digested and undigested samples for genes KAT6B and APP demonstrate that undigested DNA has the lowest CT values for both genes and brain regions, and the promoter regions of KAT6B and APP are more open in MTG than CBL (Figure 1). APP plays a prominent role in Alzheimer’s pathology and increased expression of KAT6B, a lysine acetyltransferase, has been associated with decreased expression of synaptophysin, a presynaptic molecule. Since the middle temporal gyrus is more affected in AD than is the cerebellum it can be expected that the promoter regions of these genes would be more permissive in the MTG, consistent with literature showing CBL to be less affected than MTG in Alzheimer’s disease.

Figure 1.

Figure 1

Open in a new tab

Mid temporal gyrus (MTG) and cerebellum (CBL) homogenates of frozen tissue from one Alzheimer’s (AD) case underwent Benzonase digestion then qPCR to determine relative openness of chromatin in the promoter region of APP and KAT6B. qPCR traces show that the promoter regions of KAT6B and APP appear to be more accessible in MTG than in CBL. The lower CT values of the uncut samples are consistent with intact/undigested DNA.

More Cases and Two Brain Regions With Comparison of Alzheimer’s and Control Samples

Subsequently, the number of cases was increased to middle temporal gyrus and cerebellum of 3 ND and 3 AD cases. In MTG, KAT6B and APP were more open in AD than ND (Figure 2A). The results also showed that in CBL, KAT6B and APP are more open in AD as compared to ND (Figure 2B), suggesting some effect of AD in cerebellum, though less than in middle temporal gyrus, replicating the conclusion from Figure 1 that middle temporal gyrus is more affected than cerebellum (compare relative heights of bars in Figure 2A, temporal gyrus and 2B, cerebellum). CT values indicate both genes are more open in MTG (Figure 2A) than in CBL (Figure 2B).

Figure 2.

Figure 2

Open in a new tab

Mid temporal gyrus (MTG) and cerebellum (CBL) homogenates of frozen tissue from 3 ND 3 AD cases underwent Benzonase digestion then qPCR to determine relative openness of chromatin in the promoter region of APP and KAT6B. (A) In MTG, promoter regions of KAT6B and APP appear to be more open in AD than ND. * Student’s t-test p< 0.05. (B) In CBL, promoter regions of KAT6B and APP appear to be more open in AD than ND. * Student’s t-test p< 0.05. Key: ND, non-demented control; AD, Alzheimer’s disease; D, Benzonase digested; U, undigested; error bars signify S.E.M.

More Regions Within One Gene - Homogenates of Middle Temporal Gyrus

Up to this point, chromatin accessibility analysis was limited to the promoter region of genes. To demonstrate ability to examine multiple regions within one gene, primers were designed for four regions of the Arc gene. Arc was selected for study because of its role as an immediate early gene important for memory (Guzowski et al., 2000). Four regions within the Arc sequence were chosen based on DNase I sensitive sites and H3K4Me3 ChIP analysis from the ENCODE project (Thurman et al., 2012). In this brain region ANOVA shows differential relative openness among the four arc regions examined (p=0.01). The promoter region of Arc (ARC1) is significantly less open in AD than in ND (t-test p<0.05) (Figure 3), consistent with the deficient memory function characteristic of Alzheimer’s disease.

Figure 3.

Figure 3

Open in a new tab

Mid temporal gyrus (MTG) homogenates from 3 ND and 3 AD cases underwent Benzonase digestion and qPCR. Four regions of the ARC gene were examined for chromatin accessibility. ANOVA between primer groups was p=0.01. * Student’s t-test (p<0.05) showed AD is significantly less open than ND in the promoter region of the gene (ARC1). Key: ND (−), non-demented control undigested; AD (−), Alzheimer’s disease undigested; ND (+), non-demented control Benzonase digested; AD (+), Alzheimer’s disease Benzonase digested; error bars signify S.E.M.

3.2 Laser Capture Samples

Many Dentate Gyrus Granule Cells

To test compatibility of our protocol with laser capture microdissection we used a Leica LMD 6500 to dissect a portion of the dentate gyrus granule cell layer followed by benzonase digestion and qPCR for four regions of the ARC gene. Figure 4 shows representative qPCR traces of the promoter region of ARC from 1 ND and 1 AD case (Figure 4A) and of an intronic region of ARC in 1 ND and 1 AD case (Figure 4B). The CT values of all undigested samples are comparable and lower than digested samples, as expected. ANOVA analysis of the CT values showed no significant difference between primer groups and no significant difference in openness between ND and AD in any of the gene regions (t-test) (Figure 5). For example, in support of this analysis seen in Figure 4B, the qPCR traces of ARC4 in digested ND and AD samples show that chromatin appears to be equally accessible. This result differs from the result shown for homogenate data in Figure 3. We emphasize that this is a different brain region in which neurons do not typically show the neurofibrillary tangle pathology of Alzheimer’s disease and granule cells of the dentate gyrus are not lost in AD (West et al., 1994).

Figure 4.

Figure 4

Open in a new tab

Dentate gyrus granule cells from 5 ND and 5 AD cases were obtained via laser capture microdissection (LCM) and underwent Benzonase digestion and qPCR to determine relative openness of chromatin in four regions of the ARC gene. A. Graph is representative of qPCR traces in the promoter region of Arc (ARC1) from one ND and one AD case. B. Graph is representative of qPCR traces in an intronic region of Arc (ARC4) from one ND and one AD case. As predicted and seen in both A and B, undigested samples should have equivalent and lower CT values than Benzonase digested samples.

Figure 5.

Figure 5

Open in a new tab

Dentate gyrus granule cells from 5 ND and 5 AD cases were obtained via laser capture microdissection then underwent Benzonase digestion and qPCR. Four regions of the ARC gene were examined for chromatin accessibility. No significant difference was found between primer groups or in accessibility between ND and AD in any of the gene regions. Error bars signify S.E.M.

Few MTG Pyramidal Neurons

To test the lower limit of the protocol, layer 3 pyramidal neurons from one ND and one AD case with no tangle pathology were selected in middle temporal gyrus and captured via LCM. Samples equaling 5 cell equivalents were subjected to Benzonase digestion and chromatin accessibility for APP determined by PCR. The data indicate that the promoter region of the APP gene is less open in AD than in non-demented controls (Figure 6). A 1% agarose gel indicated PCR products at the appropriate size (data not shown). These data indicate this protocol can be used on as few as 5 laser captured cells but appear contradictory to the homogenate data. We can theorize as to why the difference in results from the heterogeneity of cells in homogenates masking the data (Wills et al., 2013) to the minute starting material from single cells confounding the data. Nevertheless, we assert that without additional samples and replicates we cannot make a determination of chromatin accessibility of APP in ND or AD in non-tangle bearing neurons of the MTG.

Figure 6.

Figure 6

Open in a new tab

Five cell equivalents of MTG layer 3 pyramidal neurons with no tangle pathology from 1 ND and 1 AD case were acquired via laser capture microdissection then underwent Benzonase digestion and qPCR to determine relative openness of chromatin in the promoter region of APP. qPCR traces show that the promoter region of APP appears to be more accessible in ND than in AD.

3.3 Validation

By Treatment Known to Open Chromatin

We treated SH-SY5Y cells with valproic acid (VPA), an HDAC inhibitor known to decondense chromatin (Marchion et al., 2005). The prediction was that VPA treated cells would show more open chromatin than cells not treated with VPA. Figure 7 shows this to be true for the promoter region of KAT6B and APP genes, with equivocal results in the promoter region of Arc (ARC1) (Figure 7). The lack of effect that VPA had on Arc could be inherent to SH-SY5Y cells. Past studies found that Arc expression in these cells is a fine-tuned balancing act between translation-dependent mRNA decay and proteasomal degradation and without specific stimulation Arc expression is negligible (Soulé et al., 2012). This leads us to consider that regulation of Arc may not be so dependent on histone acetylation, thus, VPA may not act on the chromatin in the promoter region of Arc the same way as with KAT6B and APP.

Figure 7.

Figure 7

Open in a new tab

To validate Benzonase efficacy: SH-SY5Y cells were treated with valproic acid (VPA), an agent known to relax chromatin, and then digested with Benzonase. qPCR results indicate that VPA relaxed the chromatin in the primer-specific areas of KAT6B and APP with significant difference in relative openness between VPA treated cells with and without Benzonase digestion with Student’s t-test of p<0.05*. Key: U-B, untreated and no Benzonase digestion; VPA-B, VPA treated and no Benzonase digestion; VPA+B, VPA treated and Benzonase digested. Error bars signify S.E.M.

The results of the chromatin accessibility analysis showed no difference in relative openness in the no Benzonase digestion in VPA treated and untreated cells, but large increases in openness in VPA treated cells digested with Benzonase, as expected (Figure 7). Additionally, it appeared the VPA relaxed the chromatin in the primer-specific areas of KAT6B and APP, with a significant difference in relative openness between VPA treated without Benzonase digestion and VPA treated plus Benzonase digestion with student’s t-test of p< 0.05 (Figure 7).

By Comparison With DNase I

To further validate our Benzonase protocol we compared it with conventional DNase I digestion, the standard method for determining active, or “open”, areas of chromatin (Krebs and Peterson, 2000). To compare Benzonase digestion to DNase I digestion, we analyzed chromatin accessibility in 7 distinct regions of the genome after digesting SH-SY5Y cells with selected concentrations of each enzyme. Comparison of the peaks in the Benzonase digested cells (Figure 8A) to those in the DNase I digested cells (Figure 8B) showed similar patterns of relative openness, substantiating the two enzymes function comparably. In order to determine the effect of concentration of DNase I and Benzonase on our data we converted the data to standard scores. The resulting standard scores are presented in Figure 8C, which indicates that under the conditions examined and over the concentration ranges used, Benzonase and DNase I act very similarly.

Figure 8.

Figure 8

Open in a new tab

To compare Benzonase to DNase I: SH-SY5Y cells were digested with various concentrations of Benzonase (A) or DNase I (B). qPCR indicated that Benzonase digestion exhibited similar patterns of relative openness to DNase I digestion in 7 distinct regions of the genome. (C) Standard scores (x-m/sd) of CT data validate that DNase I and Benzonase are comparable under the conditions examined. Error bars signify S.E.M.

4. DISCUSSION

We have described a new method for examining chromatin structure that digests DNA with Benzonase to avoid the problems posed by actin when using DNase I. We have demonstrated the application of this method to homogenates of brain tissue, to laser microdissected brain tissue, to small numbers of neurons obtained by LCM and to in vitro cell preparations. We have demonstrated the applicability of this method to examination of multiple genes as well as multiple sites within one gene. We have experimentally demonstrated the validity of our method by treating cells with valproate, a reagent known to render chromatin more permissive, and by comparison with DNase I digestion.

The premise behind our chromatin accessibility analysis using a nuclease and qPCR is that “open” chromatin will be subject to enzyme cleavage; therefore less available to amplify in qPCR; leading to higher CT values. The higher the CT value the more accessible the chromatin is for initiation of transcription. Our novel protocol uses Benzonase, a nuclease that can successfully digest chromatin and whose efficacy is not inhibited by actin, as is the case with DNase I (Lazarides and Lindberg, 1974).

Initial experiments in frozen brain homogenates from one AD case demonstrated differential chromatin “openness” between two brain regions, as well as between two genes of interest. In this case, the promoter region of KAT6B and APP appeared more open in MTG than in CBL. When the number of cases was increased to compare chromatin accessibility between brain regions and disease states, the results showed that defined regions of both genes were more open in AD than ND in both brain regions and that both genes are more open in MTG than CBL, confirming the previous study.

When we tested the lower limits of our protocol on 5 neurons obtained by LCM we yielded seemingly different results than shown in the homogenates. We can speculate that cell heterogeneity in homogenates can mask the results that we may observe down to single-to-few cells level (Wills et al., 2013). Nevertheless, we emphasize that without additional samples and replicates we cannot draw a valid conclusion on chromatin accessibility from our 5 cell data.

We also used our method to examine chromatin structure at multiple sites within one gene. ARC was interrogated at four sites in MTG homogenate samples. Results showed that the promoter region of the ARC gene was significantly less open in AD than ND. When done on the dentate gyrus of LCM samples, there was no significant difference found in any of the four regions of the gene. We stress that this is a different brain region in which neurons do not typically show the neuron loss (West et al., 1994) or neurofibrillary tangle pathology of Alzheimer’s disease. Therefore, the finding of no difference between AD and ND in this region is to be expected.

Validation experiments began with treating SH-SY5Y cell with VPA, an agent known to decondense chromatin (Marchion et al., 2005). The treatment relaxed the chromatin as expected, validated by showing that the chromatin in VPA treated cells digested with Benzonase were more open than those not digested. Finally, when comparing chromatin structure (accessibility) in SH-SY5Y cells that had been digested with Benzonase to those digested with DNase I, the gold standard used for chromatin studies in cultured cells, similar patterns emerged. This concluding validation step demonstrates the efficacy and usefulness of Benzonase for chromatin structure analysis studies.

New methods to determine chromatin accessibility seem to arise at a consistent pace. One of the most recent examines chromatin structure at the single-cell level, however, only in cultured cells (Buenrostro et al., 2015). Up until now, no method existed to determine chromatin structure in frozen tissue samples at the level of defined cells. The protocol detailed here bridges that gap. This proposed novel method for frozen tissue demonstrates the capability to choose specific cell-types to interrogate based on disease vs. non-disease; or cells that express a particular protein; cells of a certain morphology or cells in particular regions of tissue. This potential could translate into the ability to identify specific regulatory mechanisms implicated in disease or normal cellular processes.

Numerous epigenetic mechanisms converge to modulate chromatin structure and many methods exist to delineate this structure. Most of these methods require considerable amounts of starting material, personnel time, reagent costs and expensive equipment. The protocol we describe investigates chromatin structure at distinct genomic loci from frozen tissue and defined cells obtained by laser capture microdissection. Additionally, the protocol is rapid, inexpensive and easily executed in a typical research laboratory setting.

Acknowledgments

The authors thank the Banner Sun Health Research Institute Brain and Body Donation Program of Sun City, Arizona, for providing human brain samples. The Brain and Body Donation Program is supported by the National Institute of Neurological Disorders and Stroke (U24 NS072026 National Brain and Tissue Resource for Parkinson’s Disease and Related Disorders), the National Institute on Aging (P30 AG19610 Arizona Alzheimer’s Disease Core Center), the Arizona Department of Health Services (contract 211002, Arizona Alzheimer’s Research Center), the Arizona Biomedical Research Commission (contracts 4001, 0011, 05-901, and 1001 to the Arizona Parkinson’s Disease Consortium), and the Michael J. Fox Foundation for Parkinson’s Research. This work was supported by NIH R01 AG036400 and AARC DHS Award FY 2012 to Paul D. Coleman.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Disclosure statement

The authors have no conflicts of interest to disclose.

References

  1. Amatori S, Ballarini M, Faversani A, Belloni E, Fusar F, Bosari S, Fanelli M. PAT-ChIP coupled with laser microdissection allows the study of chromatin in selected cell populations from paraffin-embedded patient samples. Epigenetics & Chromatin. 2014:7. doi: 10.1186/1756-8935-7-18. 18-8935-7-18. eCollection 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Buenrostro JD, Litzenburger UM, Ruff D, Gonzales ML, Snyder MP, Chang HY, Greenleaf WJ. Single-cell chromatin accessibility reveals principles of regulatory variation. Nature. 2015;523(7561):486–490. doi: 10.1038/nature14590. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Chabbert CD, Adjalley SH, Klaus B, Fritsch ES, Gupta I, Pelechano V, Steinmetz LM. A high-throughput ChIP-seq for large-scale chromatin studies. Molecular Systems Biology. 2015;11(1):777. doi: 10.15252/msb.20145776. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Chouliaras L, Mastroeni D, Delvaux E, Grover A, Kenis G, Hof PR, van den Hove Daniel LA. Consistent decrease in global DNA methylation and hydroxymethylation in the hippocampus of alzheimer’s disease patients. Neurobiology of Aging. 2013;34(9):2091–2099. doi: 10.1016/j.neurobiolaging.2013.02.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Dekker J, Rippe K, Dekker M, Kleckner N. Capturing chromosome conformation. Science (New York, N.Y.) 2002;295(5558):1306–1311. doi: 10.1126/science.1067799. [DOI] [PubMed] [Google Scholar]
  6. Fanelli M, Amatori S, Barozzi I, Soncini M, Dal Zuffo R, Bucci G, Minucci S. Pathology tissue-chromatin immunoprecipitation, coupled with high-throughput sequencing, allows the epigenetic profiling of patient samples. Proceedings of the National Academy of Sciences of the United States of America. 2010;107(50):21535–21540. doi: 10.1073/pnas.1007647107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Fanelli M, Amatori S, Barozzi I, Minucci S. Chromatin immunoprecipitation and high-throughput sequencing from paraffin-embedded pathology tissue. Nature Protocols. 2011;6(12):1905–1919. doi: 10.1038/nprot.2011.406. [DOI] [PubMed] [Google Scholar]
  8. Felisbino MB, Tamashiro WM, Mello ML. Chromatin remodeling, cell proliferation and cell death in valproic acid-treated HeLa cells. PloS One. 2011;6(12):e29144. doi: 10.1371/journal.pone.0029144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Felisbino MB, Gatti MS, Mello ML. Changes in chromatin structure in NIH 3T3 cells induced by valproic acid and trichostatin A. Journal of Cellular Biochemistry. 2014;115(11):1937–1947. doi: 10.1002/jcb.24865. [DOI] [PubMed] [Google Scholar]
  10. Fullwood MJ, Liu MH, Pan YF, Liu J, Xu H, Mohamed YB, Ruan Y. An oestrogen-receptor-alpha-bound human chromatin interactome. Nature. 2009;462(7269):58–64. doi: 10.1038/nature08497. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gilmour DS, Lis JT. Detecting protein-DNA interactions in vivo: Distribution of RNA polymerase on specific bacterial genes. Proceedings of the National Academy of Sciences of the United States of America. 1984;81(14):4275–4279. doi: 10.1073/pnas.81.14.4275. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gilmour DS, Lis JT. In vivo interactions of RNA polymerase II with genes of drosophila melanogaster. Molecular and Cellular Biology. 1985;5(8):2009–2018. doi: 10.1128/mcb.5.8.2009-2018.1985. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Grøntved L, Bandle R, John S, Baek S, Chung HJ, Liu Y, Levens D. Rapid genome-scale mapping of chromatin accessibility in tissue. Epigenetics & Chromatin. 2012;5(1) doi: 10.1186/1756-8935-5-10. 10-8935-5-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Guzowski JF, Lyford GL, Stevenson GD, Houston FP, McGaugh JL, Worley PF, Barnes CA. Inhibition of activity-dependent Arc protein expression in the rat hippocampus impairs maintenance of long-term potentiation and the consolidation of long-term memory. J Neurosci. 2000;20:3993–4001. doi: 10.1523/JNEUROSCI.20-11-03993.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hao H, Liu H, Gonye G, Schwaber JS. A fast carrier chromatin immunoprecipitation method applicable to microdissected tissue samples. Journal of Neuroscience Methods. 2008;172(1):38–42. doi: 10.1016/j.jneumeth.2008.04.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. He HH, Meyer CA, Hu SS, Chen MW, Zang C, Liu Y, Brown M. Refined DNase-seq protocol and data analysis reveals intrinsic bias in transcription factor footprint identification. Nature Methods. 2014;11(1):73–78. doi: 10.1038/nmeth.2762. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hughes JR, Roberts N, McGowan S, Hay D, Giannoulatou E, Lynch M, Higgs DR. Analysis of hundreds of cis-regulatory landscapes at high resolution in a single, high-throughput experiment. Nature Genetics. 2014;46(2):205–212. doi: 10.1038/ng.2871. [DOI] [PubMed] [Google Scholar]
  18. Jaenisch R, Bird A. Epigenetic regulation of gene expression: How the genome integrates intrinsic and environmental signals. Nature Genetics. 2003;(33 Suppl):245–254. doi: 10.1038/ng1089. [DOI] [PubMed] [Google Scholar]
  19. Krebs JE, Peterson CL. Understanding “active” chromatin: A historical perspective of chromatin remodeling. Critical Reviews in Eukaryotic Gene Expression. 2000;10(1):1–12. [PubMed] [Google Scholar]
  20. Lazarides E, Lindberg U. Actin is the naturally occurring inhibitor of deoxyribonuclease I. Proceedings of the National Academy of Sciences of the United States of America. 1974;71(12):4742–4746. doi: 10.1073/pnas.71.12.4742. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Li Y, Huang W, Niu L, Umbach DM, Covo S, Li L. Characterization of constitutive CTCF/cohesin loci: A possible role in establishing topological domains in mammalian genomes. BMC Genomics. 2013;14 doi: 10.1186/1471-2164-14-553. 553-2164-14-553. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lupien M, Eeckhoute J, Meyer CA, Wang Q, Zhang Y, Li W, Brown M. FoxA1 translates epigenetic signatures into enhancer-driven lineage-specific transcription. Cell. 2008;132(6):958–970. doi: 10.1016/j.cell.2008.01.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Marchion DC, Bicaku E, Daud AI, Sullivan DM, Munster PN. Valproic acid alters chromatin structure by regulation of chromatin modulation proteins. Cancer Research. 2005;65(9):3815–3822. doi: 10.1158/0008-5472.CAN-04-2478. [DOI] [PubMed] [Google Scholar]
  24. Mastroeni D, Grover A, Delvaux E, Whiteside C, Coleman PD, Rogers J. Epigenetic mechanisms in alzheimer’s disease. Neurobiology of Aging. 2011;32(7):1161–1180. doi: 10.1016/j.neurobiolaging.2010.08.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Natarajan A, Yardimci GG, Sheffield NC, Crawford GE, Ohler U. Predicting cell-type-specific gene expression from regions of open chromatin. Genome Research. 2012;22(9):1711–1722. doi: 10.1101/gr.135129.111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Orlando V. Mapping chromosomal proteins in vivo by formaldehyde-crosslinked-chromatin immunoprecipitation. Trends in Biochemical Sciences. 2000;25(3):99–104. doi: 10.1016/s0968-0004(99)01535-2. [DOI] [PubMed] [Google Scholar]
  27. Quina AS, Buschbeck M, Di Croce L. Chromatin structure and epigenetics. Biochemical Pharmacology. 2006;72(11):1563–1569. doi: 10.1016/j.bcp.2006.06.016. [DOI] [PubMed] [Google Scholar]
  28. Rao S, Procko E, Shannon MF. Chromatin remodeling, measured by a novel real-time polymerase chain reaction assay, across the proximal promoter region of the IL-2 gene. Journal of Immunology (Baltimore, Md.: 1950) 2001;167(8):4494–4503. doi: 10.4049/jimmunol.167.8.4494. [DOI] [PubMed] [Google Scholar]
  29. Sexton T, Kurukuti S, Mitchell JA, Umlauf D, Nagano T, Fraser P. Sensitive detection of chromatin coassociations using enhanced chromosome conformation capture on chip. Nature Protocols. 2012;7(7):1335–1350. doi: 10.1038/nprot.2012.071. [DOI] [PubMed] [Google Scholar]
  30. Sheffield N, Furey T. Identifying and characterizing regulatory sequences in the human genome with chromatin accessibility assays. Genes. 2012;3(4):651–670. doi: 10.3390/genes3040651. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Sherrington C. The integrative action of the nervous system CUP Archive. 1906 [Google Scholar]
  32. Song L, Crawford GE. DNase-seq: A high-resolution technique for mapping active gene regulatory elements across the genome from mammalian cells. Cold Spring Harbor Protocols. 2010;2010(2) doi: 10.1101/pdb.prot5384. pdb.prot5384. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Soulé J, Alme M, Myrum C, Schubert M, Kanhema T, Bramham CR. Balancing arc synthesis, mRNA decay, and proteasomal degradation: Maximal protein expression triggered by rapid eye movement sleep-like bursts of muscarinic cholinergic receptor stimulation. The Journal of Biological Chemistry. 2012;287(26):22354–22366. doi: 10.1074/jbc.M112.376491. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Su Z, Boersma MD, Lee JH, Oliver SS, Liu S, Garcia BA, Denu JM. ChIP-less analysis of chromatin states. Epigenetics & Chromatin. 2014;7 doi: 10.1186/1756-8935-7-7. 7-8935-7-7. eCollection 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Tewari AK, Yardimci GG, Shibata Y, Sheffield NC, Song L, Taylor BS, Febbo PG. Chromatin accessibility reveals insights into androgen receptor activation and transcriptional specificity. Genome Biology. 2012;13(10):R88. doi: 10.1186/gb-2012-13-10-r88. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Thurman RE, Rynes E, Humbert R, Vierstra J, Maurano MT, Haugen E, Stamatoyannopoulos JA. The accessible chromatin landscape of the human genome. Nature. 2012;489(7414):75–82. doi: 10.1038/nature11232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Vatolin S, Khan SN, Reu FJ. Direct chromatin PCR (DC-PCR): Hypotonic conditions allow differentiation of chromatin states during thermal cycling. PloS One. 2012;7(9):e44690. doi: 10.1371/journal.pone.0044690. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. West MJ, Coleman PD, Flood DG, Troncoso JC. Differences in the pattern of hippocampal neuronal loss in normal ageing and Alzheimer’s disease. Lancet. 1994;344:769–772. doi: 10.1016/s0140-6736(94)92338-8. [DOI] [PubMed] [Google Scholar]
  39. Wills QF, Livak KJ, Tipping AJ, Enver T, Goldson AJ, Sexton DW, Holmes C. Single-cell gene expression analysis reveals genetic associations masked in whole-tissue experiments. Nature Biotechnology. 2013;31(8):748–752. doi: 10.1038/nbt.2642. [DOI] [PubMed] [Google Scholar]

RESOURCES