Aging is a major risk factor in many forms of late-onset neurodegenerative disorders. The ability to recapitulate age-related characteristics of human neurons in culture will offer unprecedented opportunities to study the biological processes underlying neuronal aging. Here, we show that using a recently demonstrated microRNA-based cellular reprogramming approach, human fibroblasts from postnatal to near centenarian donors can be efficiently converted into neurons that maintain multiple age-associated signatures. Application of an epigenetic biomarker of aging (referred to as epigenetic clock) to DNA methylation data revealed that the epigenetic ages of fibroblasts were highly correlated with corresponding age estimates of reprogrammed neurons. Transcriptome and microRNA profiles reveal genes differentially expressed between young and old neurons. Further analyses of oxidative stress, DNA damage and telomere length exhibit the retention of age-associated cellular properties in converted neurons from corresponding fibroblasts. Our results collectively demonstrate the maintenance of age after neuronal conversion.https://doi.org/10.7554/eLife.18648.001
As we age, so do our cells. When cells are used in the laboratory to study the biology of diseases, it is important that the age of the cells reflects the age at which the disease develops. This is particularly important for illnesses with symptoms that develop during old age, and where younger cells may appear to be relatively unaffected.
Aging is a major risk factor in many brain disorders that affect elderly individuals. These late-onset disorders can be difficult to study because it is rarely possible to collect diseased cells from patients. Recent experimental advances, however, now mean that unrelated cell types – typically cells called fibroblasts, taken from a patient’s skin – can be converted directly into brain cells instead. These new brain cells will have the same genetic makeup as the patients they came from, but whether these converted cells would reflect the patient’s age too remained to be determined.
By measuring a range of biological properties of the converted cells, Huh et al. now show that converted cells do indeed keep track of their age when they are changed from fibroblasts to brain cells. The age of the cells was tested by looking at age-linked markers attached to their DNA known as an “epigenetic clock”. In addition, Huh et al. measured the age of the cells by examining the expression of genes altered with aging. Other factors examined included the amount of damaged DNA and the size of DNA regions called telomeres, which become shorter with age. Together, all of these indicated that the converted brain cells retain the age of the fibroblasts that they were made from.
So far this work has only been done using fibroblasts collected from healthy people. The same tests now need to be done using cells from people with late-onset illnesses like Huntington’s disease and Alzheimer’s disease. If the converted brain cells show signs of illness, it may provide new ways to study these illnesses using cells from specific patients, which may eventually lead to the development of new treatments.https://doi.org/10.7554/eLife.18648.002
Increasing evidence suggests that in addition to genetic susceptibility, age-related neurodegeneration may be caused in part by cellular aging processes that result in accumulation of damaged DNA and proteins in neurons (Mattson and Magnus, 2006). However, because of the inaccessibility to neurons from elderly individuals, studying these age-related cellular processes in human neurons remains a difficult task. Cellular reprogramming approaches have explored generating populations of human neurons by inducing pluripotent stem cells (iPSC) from human fibroblasts and subsequent differentiation into neurons (Takahashi and Yamanaka, 2006; Takahashi et al., 2007; Hanna et al., 2008; Hu et al., 2010). Importantly, this induction of pluripotency in adult fibroblasts reverts cellular age to an embryonic stage (Lapasset et al., 2011; Patterson et al., 2012) which remains even after differentiation into neurons (Patterson et al., 2012; Miller et al., 2013). While this is useful for modeling early developmental phenotypes (Lafaille et al., 2012; Lee et al., 2009), iPSC-derived cells have been reported to be unsuitable in recapitulating phenotypes selectively observed in aged cells (Mattson and Magnus, 2006; Vera and Studer, 2015). Recently, experimental manipulations to accelerate aging in iPSC-derived cells have been explored, for instance, by overexpressing progerin, a mutant form of lamin A observed in progeria syndrome, to force the detection of age-related pathophysiology of neurodegenerative disease (Arbab et al., 2014; Cornacchia and Studer, 2015).
Alternatively, we previously described a reprogramming paradigm using neuronal microRNAs (miRNAs), miR-9/9* and miR-124 (miR-9/9*-124), that exert reprogramming activities to directly convert human fibroblasts to specific mature neuronal subtypes (Richner et al., 2015; Victor et al., 2014; Yoo et al., 2011). Because this neuronal conversion is direct and bypasses pluripotent/multipotent stem cell stages, we reasoned that miR-mediated directly reprogrammed neurons would retain the age signature of the original donor. To assess the cellular age, we used the epigenetic clock method, which is a highly accurate biomarker of age based on DNA methylation (Horvath, 2013). Further, we evaluated age-associated signatures based on gene expression levels, miRNAs, and cellular readouts considered to be hallmarks of aging (López-Otín et al., 2013). Our thorough investigation into multiple age-associated signatures collectively demonstrate the maintenance of cellular age of the original donor during neuronal conversion and strongly suggest that directly converted human neurons can be advantageous for studying age-related neuronal disorders.
Establishing robust reprogramming efficiency is essential prior to assessing age-related phenotypes in reprogrammed neurons. We therefore elected to test our recently developed conversion approach that utilizes miR-9/9*-124 and transcription factors to robustly generate a highly enriched population of striatal medium spiny neurons (MSNs) from the fibroblasts of donors of varying ages (Richner et al., 2015; Victor et al., 2014). Fibroblast samples from donors ranging from 3 days to 96 years of age were collected and expanded to match population doubling level (PDL) to eliminate any confounding variability introduced by sequential passaging (Campisi and d'Adda di Fagagna, 2007; Pazolli and Stewart, 2008), then subsequently transduced with miR-9/9*-124 with CTIP2, DLX1/2, and MYT1L (CDM) following our established protocol (Richner et al., 2015; Victor et al., 2014) (Figure 1A). Reprogrammed cells were then stained for neuronal markers, MAP2, TUBB3, NeuN and MSN markers, DARPP32 and GABA (Figure 1B–C). MAP2 and TUBB3-positive reprogrammed neurons exhibiting extensive neurite outgrowth represented approximately 70–80% of the cell population (Figure 1D), demonstrating the consistency of reprogramming efficiency in all fibroblast samples. Neurons reprogrammed from young and old fibroblasts exhibited fast inward currents and action potentials in monocultures without necessitating coculturing with glial cells or primary neurons (n = 11) (Figure 1E). In addition, the consistent upregulation of neuronal genes, including MAP2, NCAM, and voltage-gated sodium channels, and downregulation of fibroblast-associated genes were observed in reprogrammed neurons from both young and old cells (Figure 1—figure supplement 1). These results suggest the consistent applicability of miRNA-based neuronal reprogramming in fibroblasts of all ages.
Aging largely influences the epigenetic landscape of cells (Oberdoerffer and Sinclair, 2007) with a number of genomic loci becoming differentially methylated with age (Horvath et al., 2012; Christensen et al., 2009). The epigenetic clock, which analyzes the methylation status of 353 specific CpG loci, has been shown to be a highly accurate age estimator that applies to all human organs, tissues, and cell types (Horvath, 2013). The epigenetic clock leads to an age estimate (in units of years) which is referred to as epigenetic age or DNA methylation (DNAm) age. We analyzed DNA methylation levels of neurons converted from 16 fibroblast samples aged three days to 96 years alongside 37 fibroblast samples aged three days to 94 years (Figure 2A). The actual chronological donor age was highly correlated with the estimated DNAm age of fibroblasts (correlation = 0.75) (Figure 2B) and of reprogrammed neurons (correlation = 0.82) (Figure 2B). Importantly, when the DNAm age of each reprogrammed neuron was compared to the DNAm age of the corresponding fibroblast, there was a near-perfect correlation (correlation = 0.91), which suggests that the epigenetic clock is unperturbed during miRNA-based neuronal reprogramming (Figure 2C) and supports the notion of age maintenance during direct neuronal conversion. By contrast, it is known that iPSC generation resets the epigenetic clock to an embryonic state since iPSCs have a DNAm age that is negative or close to zero (Horvath, 2013).
Given the broad variability in gene expression with age in multiple cell types (Berchtold et al., 2008; Lu et al., 2004; Fraser et al., 2005; Glass et al., 2013) and a recent demonstration of maintenance of age-associated transcriptomic changes in neurons directly converted by transcription factors (Mertens et al., 2015), we sought to determine whether age-associated transcriptomic changes could be detected after miR-9/9*-124-CDM-based neuronal conversion. The transcriptome and microRNA profiles of reprogrammed neurons from both young and old fibroblasts were analyzed alongside corresponding fibroblasts. Principal component analysis (PCA) of transcriptome revealed cell type-specific clustering of reprogrammed neurons versus fibroblasts, while age-associated segregation is observed in both fibroblasts and reprogrammed neurons (Figure 3A). A cohort of upregulated and downregulated genes with aging was commonly observed in both fibroblasts and converted neurons (Figure 3B), consistent with a previous report (Mertens et al., 2015). Gene ontology (GO) analysis of differentially expressed genes with age in reprogrammed neurons is enriched for terms associated with age-related biological processes (Figure 3—figure supplement 1), including vesicle-mediated transport (Wilmot et al., 2008), nervous system development (Lu et al., 2004) NF-kappaB transcription factor activity (Tilstra et al., 2011), regulation of apoptosis and inflammatory response (de Magalhães et al., 2009), and for genes previously identified to be associated with age in the human brain (Lu et al., 2004).
MicroRNA profiling similarly revealed distinct sample segregation both with age (young versus old) and cell type (fibroblasts versus reprogrammed neurons) (Figure 3C). Interestingly, we detected fourteen microRNAs commonly regulated with age in both fibroblasts and reprogrammed neurons, including miR-10a, miR-497, and miR-195, whose expression increased with age (Figure 3D). Because microRNAs have been implicated as global regulators of aging-associated cellular processes (Liu et al., 2012; Harries, 2014) through repression of existing target transcripts (He and Hannon, 2004; Pasquinelli, 2012; Lewis et al., 2005), we reasoned that these age-upregulated microRNAs may target and repress classes of genes found to be downregulated in old reprogrammed neurons. Indeed, GO analysis of predicted targets amongst age-downregulated transcripts in reprogrammed neurons (Figure 3B) for miR-10a-5p and miR-497-5p revealed terms associated with age such as metabolism and cellular death and survival (Figure 3—figure supplement 2). Our results suggest the potential role of miR-10a and miR-497 in regulating genes involved in cell death and survival, metabolic pathways, and DNA repair. While the exact role of these microRNAs in aging is unknown, miR-10a and miR-497 have been previously implicated in aging-associated cellular processes including inflammation, senescence, metabolism and telomerase activity (Kondo et al., 2016; Qin et al., 2012; Fang et al., 2010). Importantly, the increased expression of miR-10a, miR-497 and miR-195 detected in reprogrammed neurons was also validated by qPCR to be concordantly upregulated in human striatum and cortex samples from old individuals in comparison to young individuals (Figure 3E). These results further support the validity of reprogrammed neurons for detecting age-associated changes in microRNA network, mirroring changes observed in human brain.
Directly converted neurons were additionally assayed for cellular hallmarks of aging, including oxidative stress, DNA damage and telomere erosion (López-Otín et al., 2013). Oxidative stress has been reported to increase with age, in part due to the accumulation of reactive oxygen species (ROS) (Keating, 2008; Prigione et al., 2010; Suhr et al., 2010; Cui et al., 2012). FACS analysis of ROS levels using fluorescent marker MitoSOX (Miller et al., 2013) revealed that old reprogrammed neurons have increased ROS levels compared to young reprogrammed neurons, mirroring the observed age-associated differences of ROS levels in fibroblasts (Figure 4A). Moreover, reprogrammed neurons were analyzed by Comet Assay, a single-cell gel electrophoresis technique that assesses DNA damage accumulation (Singh et al., 1990). Old reprogrammed neurons were found to have longer comet tail lengths, an age-associated property also observed in fibroblasts, that reflects more extensive DNA damage accumulation compared to young cells (Figure 4B). Additionally, neuronal conversion maintained the length of telomeres from starting fibroblast which is virtually unchanged (Figure 4C), in contrast to the progressive increase in length commonly observed with iPSC reprogramming, where telomeres reach a plateau of around 12–14 kilobases after a few cellular passages (Agarwal et al., 2010; Batista et al., 2011; Marion et al., 2009). Together, these cellular assays support the maintenance of aging marks in old reprogrammed neurons after direct conversion.
Our in-depth analyses of multiple age signatures provide evidence that neuronal conversion from human somatic cells sampled at different ages generates neurons that emulate the donors’ ages. In addition to the demonstration of age-associated transcriptomic changes reported previously (Mertens et al., 2015), our results provide novel insights into multiple key signatures associated with age— epigenetic, microRNA and cellular— that are consistently maintained in directly converted neurons. Because aging is a complex process affecting many hallmarks of a cell (López-Otín et al., 2013), our assessment of a broad spectrum of age-related markers suggests that directly converted neurons may serve as an alternative model of neuronal aging to iPSC-derived neurons, whose erasure of multiple aging-associated signatures precludes it from adequately modeling, especially, late-onset diseases. Whereas, future studies may investigate whether additional aging marks are similarly conserved in reprogrammed neurons to model different facets of aging. While miR-9/9*-124-based reprogramming can directly convert fibroblasts with similar efficiency to neurons, we note that the older fibroblasts have lower replicative potential in a culture. However, this does not impede in the conversion efficiency of old fibroblasts. MiRNA-mediated generation of aged neurons paves the road to direct conversion into specific neuronal subtypes to investigate the contribution of neuronal aging to late-onset neurodegenerative disorders.
The following fibroblast cell lines ranging in age from three day old to 96 year old were obtained from the NIA Aging Cell Repository at the Coriell Institute for Medical Research, Coriell ID, RRID#: AG08498, RRID:CVCL_1Y51; AG07095, RRID:CVCL_0N66; AG11732, RRID:CVCL_2E35; AG04060, RRID:CVCL_2A45; AG04148, RRID:CVCL_2A55; AG04349, RRID:CVCL_2A62; AG04379, RRID:CVCL_2A72; AG04056, RRID:CVCL_2A43; AG04356, RRID:CVCL_2A69; AG04057, RRID:CVCL_2A44; AG04055, RRID:CVCL_2A42; AG13349, RRID:CVCL_2G05; AG13129, RRID:CVCL_2F55; AG12788, RRID:CVCL_L632; AG07725, RRID:CVCL_2C46; AG04064, RRID:CVCL_L624; AG04059, RRID:CVCL_L623; AG09602, RRID:CVCL_L607; AG16409, RRID:CVCL_V978; AG06234, RRID:CVCL_2B66; AG04062, RRID:CVCL_2A47; AG08433, RRID:CVCL_L625; AG16409, RRID:CVCL_V978; GM00302, RRID:CVCL_7277; AG01518, RRID:CVCL_F696; AG06234, RRID:CVCL_2B66. We routinely check all our cell cultures and confirm it to be free of mycoplasma contamination. Authentication was completed by LINE and PCR-based techniques. The International Cell Line Authentication Committee (ICLAC) lists none of these primary cells are commonly misidentified cell lines. Fibroblast cell lines were cultured and expanded in DMEM media (high glucose, Invitrogen) supplemented with 10% or 15% fetal bovine serum (Gibco), sodium pyruvate, non-essential amino acids, GlutaMAX (Invitrogen), Pen/Strep solution, and Beta-mercaptoethanol. Fibroblast cell lines were expanded to a population doubling level (PDL) of ~19–21. Formula to calculate PDL = 3.32*log (cells harvested/cells seeded) + previous PDL. Cells were kept frozen at −150°C in the above culture medium with additional 40% FBS and 10% DMSO.
Human fibroblasts ranging in age from 3 days to 96-year old were transduced with a lentiviral preparation of the Doxycline-inducible synthetic cluster of miR-9/9* and miR-124 (miR-9/9*-124), alongside transcription factors CTIP2, DLX1, DLX2, and MYT1L as previously described (Richner et al., 2015; Victor et al., 2014). Briefly, transduced fibroblasts were maintained in fibroblast media for two days with doxycycline prior to selection with Puromycin (3 μg/ml) and Blasticidin (5 μg/ml) at day three, then were plated onto poly-ornithine, fibronectin and laminin-coated coverslips at day five. Cells were subsequently maintained in Neuronal Media (ScienCell, Carlsbad, CA) supplemented with valproic acid (1 mM), dibutyryl cAMP (200 μM), BDNF (10 ng/ml), NT-3 (10 ng/ml), and RA (1 μM) for 30–35 days before analysis.
Reprogrammed neurons were fixed with 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) for 20 min at room temperature (RT), then permeabilized with 0.2% Triton X-100 for 10 min at room temperature. Cells were blocked with 1% goat serum, incubated with primary antibodies at 4°C overnight, then incubated with secondary antibodies for 1 hr at RT. Primary antibodies used for immunocytochemistry included chicken anti-MAP2 (Abcam Cat# ab5392 RRID:AB_21381531; 1:10,000 dilution), mouse anti-β-III tubulin (Covance Research Products Inc Cat# MMS-435P RRID:AB_2313773; 1:5000), rabbit anti-β-III tubulin (Covance Research Products Inc Cat# PRB-435P-100 RRID:AB_291637; 1:2000), chicken anti-NeuN (Aves Labs Cat# NUN RRID:AB_2313556; 1:500), rabbit anti-GABA (Sigma-Aldrich Cat# A2052 RRID:AB_477652; 1:2000), mouse anti-GABA (Sigma-Aldrich Cat# A0310 RRID:AB_476667, 1:500), and rabbit anti-DARPP32 (Santa Cruz Biotechnology Cat# sc-11365 RRID:AB_639000; 1:400). The secondary antibodies included goat anti-rabbit, mouse, or chicken IgG conjugated with Alexa-488, −594, or −647 (Thermo Fisher Scientific, Waltham, MA). Images were captured using a Leica SP5X white light laser confocal system with Leica Application Suite (LAS) Advanced Fluorescence 22.214.171.12423.
Whole-cell patch-clamp recordings were performed at four weeks after transduction with miR-9/9*-124-CDM. Intrinsic neuronal properties were studied using the following solutions (in mM): Extracellular: 140 NaCl, 3 KCl, 10 Glucose, 10 HEPES, 2 CaCl2 and 1 MgCl2 (pH adjusted to 7.25 with NaOH). Intracellular: 130 K-Gluconate, 4 NaCl, 2 MgCl2, 1 EGTA, 10 HEPES, 2 Na2-ATP, 0.3 Na3-GTP, 5 Creatine phosphate (pH adjusted to 7.5 with KOH). Membrane potentials were typically kept at −60 mV to −70 mV. In voltage-clamp mode, currents were recorded with voltage steps ranging from +10 mV to +80 mV. In current-clamp mode, action potentials were elicited by injection of step currents that modulated resting membrane potential from −20 mV to +80 mV. Local application of TTX (Sigma-Aldrich#T8024) was achieved using a multibarrel perfusion system with a port placed within 0.5 mm of the patched cell.
Reprogrammed neurons were harvested after 30 days of ectopic expression of miR-9/9*-124-CDM. DNA was extracted using phenol/chloroform/isoamyl alcohol followed by ethanol precipitation with a final concentration of 0.75M NaOAc and 2 μg of glycogen. DNA concentration was quantified using a standard curve with the Quant-iT dsDNA Assay Kit, broad range (Thermo Fisher Scientific, Waltham, MA) according to manufacturer’s instruction, while the DNA quality was determined by the ratio of absorbance of 260 nm and 280 nm at approximately 1.7–2.0.
The bisulfite conversion was performed for fibroblasts and reprogrammed neurons using the Zymo Research EZ-96 DNA Methylation-Gold Kit (catalog #D5008). DNA methylation data were generated on the HumanMethylation450k Bead Chip (Illumina, San Diego, CA) according to the manufacturer's protocols. Scanning was performed via Illumina’s iScan system in conjunction with the Illumina Autoloader 2 robotic arm. DNA methylation levels (β values) were established by calculating the ratio of intensities between methylated (signal A) and un-methylated (signal B) sites. The β value was calculated from the intensity of the methylated (M corresponding to signal A) and un-methylated (U corresponding to signal B) sites, as the ratio of fluorescent signals β = Max(M,0)/[Max(M,0)+Max(U,0)+100]. β values range from 0 (completely un-methylated) to 1 (completely methylated). The data were normalized using the 'Noob' normalization method (Triche et al., 2013).
The epigenetic clock method is an accurate measurement of chronological age in human tissues (Horvath, 2013). Epigenetic age was estimated using the published software tools (Horvath, 2013). An online age calculator can be found at the webpage, https://dnamage.genetics.ucla.edu. The epigenetic clock has been shown to capture aspects of biological age: the epigenetic age is predictive of all-cause mortality even after adjusting for a variety of known risk factors (Marioni et al., 2015; Christiansen et al., 2016; Horvath et al., 2015a). The utility of the epigenetic clock method has been demonstrated in applications surrounding cognitive function (Levine et al., 2015), obesity (Horvath et al., 2014), Down syndrome (Horvath et al., 2015b), HIV infection (Horvath and Levine, 2015), and Parkinson's disease (Horvath and Ritz, 2015).
Total RNA was extracted from reprogrammed neurons from young fibroblasts aged three days, five months, and one year and from old fibroblast aged 90, 92, and 92 years alongside corresponding starting fibroblast samples using TRIzol (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instruction and extracted using chloroform and ethanol precipitation. RNA quality was determined by the ratio of absorbance at 260 nm and 280 nm to be approximately 2.0. Samples for RNA microarray were then standardly prepped and labeled with Illumina TotalPrep kits (Thermo Fisher Scientific, Waltham, MA) for Agilent Human 4x44Kv1, while samples for microRNA microarray were prepared using Genisphere Flashtag labeling kits designed for Affymetrix miRNA 4.0 microarray. Standard hybridization and imagine scanning procedure were performed according to the manufacturer's protocol at Genome Technology Access Center at Washington University School of Medicine, St. Louis. The intensity of probes was imported into R environment and normalized by using package 'oligo'. Differentially expressed mRNA transcripts were identified by using package 'limma' with cut-off at adjusted p-value<0.01 and over logfc >1 fold expression change. For miRNA, the intensity of human-specific probes was isolated by using in-house python script, and were imported into R environment. Quantile normalization was performed by using 'preprocessCore' package, and differentially expressed miRNAs were identified by using package 'limma' with cut-off at adjusted p-value<0.01 and over one-fold expression change.
cDNA was generated from 4 ng of RNA using specific primer probes from TaqMan MicroRNA Assays (Thermo Fisher Scientific, Waltham, MA) and subsequently analyzed on a StepOnePlus Real-Time PCR System (AB Applied Biosystems, Foster City, CA). Expression data were normalized to RNU44 and analyzed using the 2−ΔΔCT relative quantification method. QPCR validation of miRNA expression was conducted in reprogrammed neurons from old fibroblasts aged 89, 90, 91, 92, 92, 94 compared to reprogrammed neurons from young fibroblasts aged three days, five months, one, two, 12 years of age. QPCR experiments were conducted with human striatum and human cortex slices acquired from young individuals aged 9, 11, and 19 years compared to those from older individuals aged 83, 85, and 87 years.
MitoSOX Red Mitochondrial superoxide indicator (Thermo Fisher Scientific, Waltham, MA) was diluted to 15 μM and incubated with cells for 15 min at 37°C. Cells were washed three times with PBS, dissociated with 0.25% Trypsin, then stained with DAPI. If FACs was not conducted on the same day, cells were fixed with 4% paraformaldehyde for 20 min at room temperature. Samples were compared to untreated (unstained) fibroblast and reprogrammed neurons. Cell sorting was performed on a FACSCalibur and LsrFortessa (BD Biosciences), while quantification of the percent of the population of MitoSOX positive cells was performed using FlowJo X 10.0.7r2. Each plot on the graph represents an individual experiment with multiple reprogrammed neurons. Unpaired t-test analysis of 3 sets of experiments of reprogrammed neurons from young fibroblasts compared to reprogrammed neurons from old fibroblasts. Young samples included reprogrammed neurons from fibroblasts aged three days, five month, one year and two years, while old samples were from donors aged 86, 90, 91, 92A, and 92B years. Analyzed fibroblast samples include 1, 2, 91, 72, 74, and 94-year-old samples. P-values were calculated with the student t-test.
Cells were prepared and analyzed using the CometAssay Kit (Trevigen) according to manufacturer’s instruction. Cells were harvested after 30 days of neuronal reprogramming using 0.25% Trypsin, then whole cells were embedded in molten LMAgarose onto slides prior to overnight incubation in lysis buffer. Slides were then run in gel electrophoresis at 20 volts for 30 min, then stained with SYBR Green and visualized by epifluorescence microscopy. Tails lengths were measured by drawing a region of interest and p-values were calculated with student t-test for reprogrammed neurons from old fibroblasts aged 91, 92A and 92B compared to reprogrammed neurons from young fibroblasts aged three days, five months, and one year old, while analyzed fibroblasts were aged five months, one year, 12, 72, 86, and 92 years of age.
Genomic DNA was collected from reprogrammed neurons from fibroblasts from one year, 56, and 86 year old donors and the corresponding fibroblasts. The isolated genomic DNA was then digested with RsaI and HinfI and fractionated as described previously (Tomlinson et al., 2008). Membranes were prepared by Southern transfer and hybridized to a radioactively end-labelled (TTAGGG) 4 oligonucleotide probe as described previously (Batista et al., 2011).
Cellular senescence: when bad things happen to good cellsNature Reviews Molecular Cell Biology 8:729–740.https://doi.org/10.1038/nrm2233
Back and forth in time: Directing age in iPSC-derived lineagesBrain Research S0006-8993(15)00857-4.https://doi.org/10.1016/j.brainres.2015.11.013
Oxidative stress, mitochondrial dysfunction, and agingJournal of Signal Transduction 2012:646354.https://doi.org/10.1155/2012/646354
MicroRNAs: small RNAs with a big role in gene regulationNature Reviews Genetics 5:522–531.https://doi.org/10.1038/nrg1379
HIV-1 infection accelerates age according to the epigenetic clockJournal of Infectious Diseases 212:1563–1573.https://doi.org/10.1093/infdis/jiv277
The role of nuclear architecture in genomic instability and ageingNature Reviews Molecular Cell Biology 8:692–702.https://doi.org/10.1038/nrm2238
MicroRNAs and their targets: recognition, regulation and an emerging reciprocal relationshipNature Reviews Genetics 13:271–282.https://doi.org/10.1038/nrg3162
Defining the nature of human pluripotent stem cell progenyCell Research 22:178–193.https://doi.org/10.1038/cr.2011.133
Senescence: the good the bad and the dysfunctionalCurrent Opinion in Genetics & Development 18:42–47.https://doi.org/10.1016/j.gde.2007.12.002
Role of microRNAs in endothelial inflammation and senescenceMolecular Biology Reports 39:4509–4518.https://doi.org/10.1007/s11033-011-1241-0
DNA damage and rpair with age in individual human lymphocytesMutation Research/DNAging 237:123–130.https://doi.org/10.1016/0921-8734(90)90018-M
NF-κB in aging and diseaseAging and Disease 2:449–465.
Low-level processing of Illumina Infinium DNA Methylation BeadArraysNucleic Acids Research 41:e90.https://doi.org/10.1093/nar/gkt090
Translational gene mapping of cognitive declineNeurobiology of Aging 29:524–541.https://doi.org/10.1016/j.neurobiolaging.2006.11.008
Jeremy NathansReviewing Editor; Johns Hopkins University School of Medicine, United States
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your manuscript "Maintenance of age in human neurons generated by micro-RNA-based neuronal conversion of fibroblasts" to eLife. Three experts reviewed your manuscript, and their assessments, together with my own, Jeremy Nathans (the Reviewing editor), form the basis of this letter. As you will see, overall, the reviewers were impressed with the importance of your work.
I am including the three reviews at the end of this letter, as there are a variety of specific and useful comments in them. One part of the data that appears problematic is the comet assay. See the detailed comments by reviewer #1. A second point is the extent to which your work is distinct from or is an advance beyond that reported by Mertens et al. [Cell Stem Cell 17: 705-718. (2015)]. See the comments by reviewer #2. A fuller description of the Mertens et al. data and its relation to yours would be useful for the reader. This could be done in the Discussion section.
Huh et al. used microRNA based reprogramming to generate neurons from fibroblasts, starting with fibroblasts taken from humans over a range of ages, and claim that the reprogrammed neurons reflect the age of the donor based on epigenetic assays, DNA damage assays, gene expression patterns, microRNA expression patterns, telomere length, and other parameters. They convincingly show that the reprogramming works with cells of a range of ages, showing good staining of neuron-specific markers, and impressive changes in gene expression patterns from a fibroblast pattern to a neuronal pattern. They use a variety of assays to support the interpretation that the conversion maintains the "age" of the donor cells. In general, the data support their conclusions well. They use a DNA methylation assay as an assay of "epigenetic age" and show a reasonable correlation between fibroblast age and epigenetic age as measured in the converted neurons.
In their gene expression analysis, the PCA analysis seems to nicely separate converted neurons from fibroblasts as expected, but does not clearly separate older from younger cells of either type. The analysis of microRNA expression patterns does seem to separate older from younger of each type, though based upon a very small number of microRNA.
Regarding their GO analysis (Figure 3) of differentially expressed genes, they appear to refer to "GO terms associated with aging"; some of these terms obviously relate to aging (for example, aging), while many others are not obviously related to aging (negative regulation of metabolic processes, nervous system development, etc.) and it is not clear to me or well explained in the text where these come from or how they are related to aging per se. Please explain.
Comet assay: I am not an expert, but I found this unconvincing. There appears to be a big difference in loading that at least contributes to, if not accounts for, the claimed difference in comet tail length between young and old neurons. Both of the young assays (I assume these are "young" re-programmed neurons) seem underloaded compared to the older cells; and if we were to control in our mind's eye for this loading difference, I am not sure the lengths of the comet tails would really differ much or at all. At best, the difference seems extremely subtle, much less than is illustrated in the figure.
This figure would be stronger if the young re-programmed neurons were run against the young fibroblasts, and then old fibroblasts run against old neurons-this might be what is illustrated but the figure legend does not give enough information to understand it. But then they need to load equal amounts of DNA in each experiment, or explain why the assay is such that this is not a problem.
This manuscript is a follow-up to a previous work from the same authors in which neuronal micro RNAs (miRNAs) were used to directly reprogram human fibroblasts into specific neuronal subtypes. The goal of this paper is to demonstrate that miRNA-derived neurons retain the age profile of their donors. Interestingly, fibroblasts from donors sampling a wide age range (from a few days old to 94 years old), after being reprogrammed into neurons retained the DNA methylation levels, telomere size, transcription profile, levels of DNA damage, etc. This result is in contrast to neurons differentiated from induced pluripotent stem cells (iPSCs) which reset their clock in terms of telomere size, gene expression profile, oxidative stress, etc.
This manuscript is straightforward and technically sound. The authors were thorough in testing all the important age markers and convincingly demonstrate the key point of the paper. However, this work is purely a cell characterization study and therefore only incremental (the actual miRNA conversion itself has been published in several articles before). More importantly, it has already been shown that fibroblasts reprogrammed into neurons (via a different route) retain the age-specific transcriptomic profile, see Mertens et al., Directly Reprogrammed Human Neurons Retain Aging-Associated Transcriptomic Signatures and Reveal Age-Related Nucleocytoplasmic Defects, Cell Stem Cell, 2015. This published paper has only casually been referenced. Along the same lines, in the Introduction, the study was motivated by the following sentence: "Because this neuronal conversion is direct and bypasses pluripotent / multipotent stem cell stages, we reasoned that directly reprogrammed neurons would retain the age signature of the original donor." This statement makes it seem as if this phenomenon has not already been shown before.
At this stage the paper is therefore only incremental and does not warrant publication in eLife. The miRNA-converted neurons could indeed potentially be a powerful tool in studying age-related diseases (just like the reprogramming method published by Mertens et al) and so, at this stage, it would be very valuable to already see a functional comparison between examples of iPSC and miRNA-generated neurons from a diseased patient or a more thorough study on the onset and specific causes of age-related changes.
A comment regarding the data in Figure 4B: the data points for the 1-yo and 56-yo examples are not clear and the retention of telomere lengths of fibroblasts and miRNA-generated neurons does not appear convincing. That said, the 86-yo point, which indeed is the most important point, is convincing. Moreover, a more detailed description of the image is warranted in the legend, for instance, what do the numbers on the left stand for?
This manuscript builds on prior work from the Yoo lab to test whether miRNA coverted neuronal cells derived from fibroblasts retain age-related signatures. I have a few comments regarding this work:
1) It would be of interest to quantitatively examine the expression levels of INK4A locus genes (p16 and ARF) since their upregulation has also been associated with aging.
2) Regarding telomere rejuvenation upon iPS reprogramming, since the referenced papers were published (Agarwal et al., 2010 and Marion et al., 2009), it has been shown that the original telomere length is only incrementally reset with iPS reprogramming (Batista Nature 2011, Moon Nat Gen 2015 and several others). This should be corrected.
3) It may be important to acknowledge that the retention of molecular marks of aging may make it more difficult to do in vitro with these cells. For example, shorter telomere length from older individuals may limit the replicative potential. Along the same lines, I suppose it's important to acknowledge that there may be other features of aging that were not assayed here that may be altered during the conversion process.https://doi.org/10.7554/eLife.18648.018
- Bo Zhang
- Luis FZ Batista
- Luis FZ Batista
- Luis FZ Batista
- Luis FZ Batista
- Steve Horvath
- Steve Horvath
- Andrew S Yoo
- Andrew S Yoo
- Andrew S Yoo
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank the Genome Technology Access Center in the Department of Genetics at Washington University School of Medicine for help with processing transcriptome and microRNA microarrays. We also thank UCLA Neuroscience Genomics Core and Helen Ibsen for processing Illumina Methylation450K chips. We are grateful to Shin-ichiro Imai for helpful suggestions on the manuscript. LB is supported by the NIH K99/R00 award (4R00HL114732-03), Washington University DDRCC (NIDDK P30 DK052574) and grants from the V Foundation For Cancer Research, Edward Mallinckrodt Jr. Foundation, and the AA&MDS International Foundation. SH is supported by National Institutes of Health NIH/NIA 1U34AG051425-01 and 5R01 AG042511-02. ASY is supported by the NIH Director's Innovator Award (DP2NS083372-01) and Presidential Early Career Award for Scientists and Engineers, and grants from the Ellison Medical Foundation and Cure Alzheimer’s Fund.
- Jeremy Nathans, Reviewing Editor, Johns Hopkins University School of Medicine, United States
© 2016, Huh et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.