Opportunities and challenges to induce aging in a dish: Neurological aging strategies in iPSC models to study late-onset neurological and neurodegenerative diseases.

MSc Brain and Cognitive Sciences

Behavioural Neuroscience

Thesis

Opportunities and challenges to induce aging in a dish

Neurological aging strategies in iPSC models to study late-onset neurological and

neurodegenerative diseases.

by Iris Marchal 10995749 September – December 2016 Supervisor Co-assessor Dr. Frank Jacobs Dr. Miranda van Wonterghem ABSTRACT A decade ago the field of in vitro stem cell models was revolutionized by the introduction of iPSC technology. Since then giant leaps have been taken in the manipulations of cellular fate, enabling the re-creation of virtually any type of cell from pluripotent stem cells in a dish. A vast research field has now arisen in which iPSCs have been studied for multiple purposes ranging from drug discovery to regenerative medicine. Additionally, iPSCs have revolutionized research into neurological diseases, with an emphasis on late-onset disorders such as Alzheimer’s disease (AD), Parkinson’s disease (PD) and Schizophrenia. However, upon reprogramming iPSCs are rejuvenated, resulting in a cellular age that is similar to that of an embryonic stem cell. In order to create realistic in vitro replicas of human tissues, a new challenge has now arisen: the challenge to re-create age. Finding new methods to mimic age-related changes in iPSC models will not only provide new opportunities to study late-onset disorders, but will also be of particular importance to further unravel the mechanisms underlying human neurological aging. This review will describe current opportunities and challenges of this search to model neurological aging in a dish, and will provide suggestions of future research directions.

GRAPHICAL ABSTRACT LIST OF ABBREVIATIONS 5-mC 5-methylcytosine AD Alzheimer’s Disease ATF Atrificial transcription factor CNS Central nervous system CpG Cytosine-phosphate-guanine CS Cockayne syndrome DC Dyskeratosis congenita DNAm DNAmethylation DNMTS DNA methyl transferases DMRs Differential methylated regions ES cell Embryonic stem cell HDAC Histone deacetylase HGPS Hutchinson-Gilford Progeria Syndrome HP-1 Heterochromatin protein 1 iN Induced neuron iPSCs Induced pluripotent stem cells miRNA Micro RNA mRNA Modified RNA MS cell Mesenchymal stem cell MSN Medium spiny neurons mtROS Mitochondrial ROS NAD+ Nicotine adenine dinuclease NCC Nucleocutoplasmic compartementalization PcG Polycomb group protein PD Parkinson’s Disease PTB Polypyrimidine-tract-binding protein ROS Reactive oxygen species SMC Smooth muscle cell WS Werner Syndrome ZFP Zinc finger protein

TABLE OF CONTENT 1. INTRODUCTION 4 2. CELLULAR AGING 4 2.1 Genomic instability 5 2.1.1 Insights from premature aging diseases 5 2.2 Telomere Attrition 6 2.3 Epigenetic Regulation 6 2.3.1 DNA methylation 7 2.3.2 Histone modifications 8 2.3.3 Micro RNA 9 2.4 Summary 10 3. MECHANISMS OF REPROGRAMMING TO INDUCED PLURIPOTENCY 10 3.1 Epigenetic mechanisms of reprogramming 11 3.1.1 Is everything lost? 12 4. CURRENT STATE OF MODELING AGE IN VITRO 12 4.1 The classical aging paradigm 12 4.2 Premature aging diseases 13 4.2.1 HGPS 13 4.2.2 Other syndromes 14 4.3 Taking the shortcut: skipping pluripotency state 15 4.3.1 Maintaining age in directly reprogrammed neurons 15 5. DISCUSSION 17 5.1 Challenges 17 5.2 Promises 18 6. REFERENCES 20

1. INTRODUCTION

For decades, the availability of human brain material to study neurological diseases was mainly limited to post-mortem tissue and embryonic stem (ES) cells. This changed in 2006, when a major breakthrough was achieved by the introduction of a new reprogramming technique to create induced pluripotent stem cells (iPSCs) (Takahashi & Yamanaka, 2006). By the over-expression of only four reprogramming factors: OCT4, SOX2, KLF4, and c-MYC, adult somatic cells can be reprogrammed into a state of pluripotency, which provides the possibility to create neuronal tissue upon re-differentiation. Ten years since its introduction, a vast research field has arisen in which iPSCs have been studied for multiple purposes ranging from drug discovery to regenerative medicine (Compagnucci, Nizzardo, Corti, Zanni, & Bertini, 2014; Isobe et al., 2014; Wu & Hochedlinger, 2011). Additionally, iPSCs have revolutionized research into neurological diseases, with an emphasis on late-onset disorders such as Alzheimer’s disease (AD), Parkinson’s disease (PD) and Schizophrenia (Marchetto, Brennand, Boyer, & Gage, 2011). iPSC technology has provided an inexhaustible source of patient-specific neural material to recreate and manipulate these diseases. However, iPSC modelling of late-onset diseases in vitro has been suffering from a major limitation (Vera & Studer, 2015). Upon reprogramming iPSCs rejuvenate, obtaining a cellular age profile similar to ES cells (Mahmoudi & Brunet, 2012; Suhr et al., 2010). This reset of age has been found to obstruct the acquisition of age-related features of a disease phenotype (Srikanth & Young-Pearse, 2014). Furthermore, aging is considered to be the main risk factor for development of late-onset neurological diseases (Farooqui & Farooqui, 2009). Modeling of late-onset diseases in cells that are aged zero at a cellular scale is therefore found to be inappropriate. In order to overcome this problem, scientists have tried to develop strategies to induce aging in iPSCs and iPSC-derived neural cultures (Studer, Vera, & Cornacchia, 2015).

In this thesis, I will review the current opportunities and challenges of this search to model neurological aging in a dish. By doing so, I will touch upon the main controversies and uncertainties that need to be overcome: (1) what are the cellular and molecular mechanisms underlying neurological aging? (2) how are adult somatic cells reprogrammed into iPSCs and what happens to their aging profile during reprogramming on a cellular and molecular scale? (3) To what extent have current aging-paradigms been able to successfully recover the features of cellular age in iPSCs or iPSC-derived neural tissue? Furthermore, I will review the development of new reprogramming methods and their potential value for solving the problem of rejuvenation. Lastly, I will argue how aging should be incorporated as an independent factor for the in vitro modeling of late-onset diseases. Better aging models will not only provide us with the opportunity to study various age-related neurological diseases, but will contribute to new insights in the basic mechanisms of neurological aging and will ultimately create new opportunities to facilitate healthy aging.

2. CELLULAR AGING

Modeling cellular aging in iPSC technology represents a considerable experimental challenge. In order to succeed, a mechanistic understanding of the aging process would be required. Understanding the biology of aging would have profound implications for human life, and it has therefore always been an object of extensive study (Rattan, 2006). Over the years, researchers have used various methods to measure and mimic aging in a dish. In order to evaluate which of these methods has been successful, one should first consider the basic mechanisms underlying the aging process. Aging can be defined as the time-dependent functional decline that affects most living organisms (López-Otín, Blasco, Partridge, Serrano, & Kroemer, 2013). Innumerable theories have been proposed to explain the mechanisms that underlie the process of

aging (Behl & Ziegler, 2014). Some theories argue that aging is a process that is intrinsically programmed into our system and is driven by either genetic, hormonal or immunological mechanisms. Others have argued aging to be merely a result of cumulative damage caused by environmental factors that organisms encounter during life (Jin, 2010). Of all the current theories available, none has yet been able to explain all the phenomena that are observed during aging. Despite this lack of one definitive explanation, it is nowadays believed that the processes described in each separate theory together interact in a complex way (Kirkwood, 2005). All key processes observed during aging seem to be hierarchically interconnected (Behl & Ziegler, 2014; Cornacchia & Studer, 2016). It is assumed that some molecular and cellular mechanisms are at the top of this hierarchy, and manipulating these components would trigger a broad downstream cascade mimicking physiological aging (Cornacchia & Studer, 2016). Research on aging has recently been focused on unraveling those molecular and cellular mechanisms. This has led to consensus about several key mechanisms that are involved in the aging process (López-Otín et al., 2013). These key mechanisms are known as the primary hallmarks of aging and mainly cover the levels of the genome and the epigenome. In the following section, I will describe the contribution of (1) genomic instability, (2) telomere attrition and (3) epigenetic regulation to the process of aging. It should be noted that these hallmarks only present the most prominent mechanisms of cellular and molecular aging, and are of particular interest with respect to the modeling of aging. An extensive review of all the hallmarks of aging is beyond the scope of this thesis. 2.1 Genomic instability

The first hallmark, long been implicated to be the main causal factor of aging, is genomic instability (Moskalev et al., 2013). As a response to different physiological conditions, genomes have a natural tendency to undergo alterations over time (Vijg & Suh, 2013). The

integrity of the genome is constantly challenged by many different DNA-damaging agents leading to the accumulation of genetic damage throughout life. Damaging agents can be both endogenous and exogenous. Endogenous agents are most frequent and involve DNA replication errors, spontaneous hydrolytic reactions and reactive oxygen species (ROS). Exogenous agents include ultraviolet (UV) light, ionizing radiation (IR) as well as many toxic chemicals. The genomic alterations caused by these agents can be divided into three types: chemical damage to the DNA, genomic mutations and mutations of the epigenome. There are some important differences between these types of damage. Firstly, chemical DNA damage leads to deviations from the structure of DNA. On the other hand, mutations lead to alterations of the information content. Additionally, DNA damage is reversible and can be corrected by numerous DNA repair mechanisms, whereas mutations are permanent. Under normal circumstances DNA repair mechanisms are highly efficient in counteracting DNA damage. It is therefore unlikely that the number of unrepaired DNA lesions will increase drastically when a cell ages. However, errors can be made during DNA repair activities, leading to irreversible mutations. The accumulation of these irreversible mutations over time is believed to be involved in the manifestation of human aging (Vijg & Suh, 2013).

2.1.1 Insights from premature aging syndromes

The notion that the accumulation of DNA mutation is related to aging is supported by observations from the human progeria disease Werner’s syndrome (WS) (Gregg et al., 2012; Hoeijmakers, 2009). WS is caused by mutations in the WRN gene, leading to a loss of function of the WRN protein (Kudlow, Kennedy, & Monnat, 2007). Defects in WRN protein lead to various disruptions in DNA repair mechanisms, resulting in a disease phenotype of chromosomal abnormalities, accelerated telomere shortening and early onset of aging characteristics (M. P. Cooper et al., 2000; D. K.

Singh, Ahn, & Bohr, 2009). Additionally, insights from another premature aging disease Hutchinson-Gilford Progeria Syndrome (HGPS) have shown that defects in the nuclear architecture of a cell also lead to genome instability and contribute to cellular aging (De Sandre-Giovannoli et al., 2003). HGPS is caused by mutations in the LMNA gene encoding the intermediate filament protein lamin A. Lamin A is a major component of the nuclear lamina. Mutations in the LMNA gene lead to miss-splicing of the LMNA mRNA and produces a mutant lamin A protein called progerin. Production of mutant progerin molecules and alterations in the nuclear lamina have also been reported during normal aging (Mattout, Dechat, Adam, Goldman, & Gruenbaum, 2006). Further evidence for the causal role of nuclear lamina deficiencies in aging are provided by the observation that decreasing lamin A and progerin levels leads to a delayed onset of progeroid features and extend life span in mouse models of HGPS (Varela et al 2008, Benson et al 2010). Although insights from premature aging syndromes can demonstrate a causal role of genomic instability in the process of aging, it should be noted that only some aspects of genomic alterations are covered by these diseases. Therefore, the relevance of these insights to normal aging processes remains uncertain. Future research should further investigate the entire complex apparatus of genomic integrity not only in diseased, but even more so in healthy aging conditions.

2.2 Telomere attrition

Telomeres are regions of the chromosome that are specifically susceptible to genomic alterations (Gonzalo & Eissenberg, 2016). Telomere-protective sequences at chromosome ends show a progressive decline with age. This phenomenon is caused by the fact that upon DNA replication, DNA polymerases are incapable to completely replicate the terminal ends of linear DNA molecules. Complete replication of terminal ends can only be performed by a specialized DNA polymerase called telomerase. However, most

mammalian cells do not express telomerase and therefore fail in this process (Olovnikov, 1996). Additionally, telomeres are protected from DNA repair mechanisms. Namely, if DNA repair mechanisms could access telomeres they would be recognized as DNA breaks and be ‘repaired’, leading to chromosome fusions. This protection is provided by a multiprotein complex known as shelterin (Palm & de Lange, 2008). Due to this restriction in DNA repair, DNA damage occurring at telomere regions are permanent in any way (Fumagalli et al., 2012). As a consequence, normal aging is accompanied by telomere attrition (Flanary & Streit, 2003; Harley, Futcher, & Greider, 1990). Telomere pathologies are associated with development of neurodegenerative diseases such as AD and can accelerate the aging process (Armanios et al., 2009; Johnson, 2008). Moreover, there seems to be a strong relation between short telomeres and mortality risk in humans (Boonekamp, Simons, Hemerik, & Verhulst, 2013). These data indicate a strong implication for telomere shortening in the process of aging.

2.3 Epigenetic regulation

Age-related alterations have also been described at the level of the epigenome (Booth & Brunet, 2016; Gravina & Vijg, 2010). Epigenetic regulation encompasses all mechanisms regulating gene expression without changing the genotype. Epigenetic regulation of gene expression occurs at three different levels: alterations in DNA methylation patterns, changes in histone modifications and non-coding (micro) RNA regulation (Lardenoije et al., 2015). Genetic alterations are usually permanent. In contrast, changes in the epigenome are mediated through processes that are in theory reversible. However, overall alterations in epigenetic regulatory patterns as a function of time have been observed in each level of gene expression (López-Otín et al., 2013), implicating involvement in the aging process.

2.3.1 DNA methylation

The best described epigenetic modification is DNA methylation. It involves the addition of a methyl group at the 5’ position on the pyrimidine ring of cytosine, creating 5-methylcytosine (5-mC). These modifications primarily occur at cytosine-phosphate-guanine (CpG) islands. DNA methylation is associated with gene silencing and is mostly found in heterochromatin, a tightly packed form of DNA (Miller & Sweatt, 2007). Age-related alterations in DNA methylation have been described in different animal models as well as in humans (Jakovcevski & Akbarian, 2012). Whether the age-related changes in methylation are increased or decreased is mostly dependent on the studied tissue type (Richardson, 2003). Regarding the brain, age-related changes in DNA methylation are associated with a global decrease in DNA methylation. However, some CpG sites are found to be hypermethylated during aging (Hernandez et al., 2011; Siegmund et al., 2007) This hypermethylation of a small proportion of CpG sites is consistently present across all tissue types (Tra et al., 2002). However, which specific subgroup of CpG sites is hypermethylated seems to be tissue-specific (Ben-Avraham, Muzumdar, & Atzmon, 2012). Exactly why specific CpG sites show an age-related hypermethylation pattern in contrast to an overall hypomethylation remains to be resolved. Overall, studies show a significant role for DNA methylation patterns in the aging process. However, the tissue-specificity in DNA methylation patterns makes it hard to unravel the dynamics of this form of epigenetic regulation. Findings from non-brain tissues do not automatically apply to the brain. Full mapping of the epigenome of brain tissue should provide new insights into the dynamics of DNA methylation patterns that are specifically related to neurological aging. Despite the issue of tissue-specificity that impedes the exploration of why certain CpG-sites are hypermethylated in a certain tissue and others are not, researchers have attempted to examine how age-related changes in DNA

methylation might potentially be regulated. DNA methylation is normally regulated by DNA methyltransferases (DNMTs). There are four known subtypes of DNMTs called DNMT1, DNMT2, DNMT3A and DNMT3B. DNMT1 is the most abundant variant in somatic cells and is mainly involved in the process to preserve the DNA methylation profile after cell division, known as maintenance DNA methylation (Mastroeni et al., 2010). DNMT3A and DNMT3B regulate the addition of completely new DNA methylation marks known as de novo DNA methylation (Okano, Bell, Haber, & Li, 1999). Potential changes in DNMT regulation is one of the proposed underlying mechanisms of age-related alterations in DNA methylation. DNMT1 activity has been found to decrease significantly during the aging process, and this is seen as a possible model to explain the age-related patterns of overall hypomethylation (Casillas, Lopatina, Andrews, & Tollefsbol, 2003; Lopatina et al., 2002). DNMT1 expression is known to be influenced by levels of growth hormone. It was therefore proposed that DNMT1 decreases during aging might be due to an age-related reduction in growth hormone (Armstrong, Rakoczy, Rojanathammanee, & Brown-Borg, 2014). Other proposed mechanisms explaining the overall hypomethylation during aging include changes in metabolic factors – such as folate and vitamin B12 – , SIRT1 expression levels, potential involvement of the TET-proteins and exogenous influences from toxic environmental elements (Jung & Pfeifer, 2015; Richardson, 2003). Additionally, the potential mechanisms regulating local DNA hypermethylation with aging are less well understood. Recently, Jung and Pfeifer (2015) proposed a theory to explain de novo hypermethylation during aging that is specific for Polycomb group protein (PcG) target genes. PcGs normally target unmethylated DNA regions, forming complexes that lead to gene repression by application of a repressive H3K27me3 mark. As a result, access to DNA by DNMTs is obstructed. It was proposed that the efficiency of this PcG machinery to target

unmethylated sequences would degrade with age, resulting in an increased access of DNMTs to the target sequences allowing DNA hypermethylation of these specific sites upon aging (Jung & Pfeifer, 2015). In summary, there is a growing body of literature supporting age-specific changes in DNA methylation patterns. Although scientists have moved beyond observing correlative associations and have begun to unravel the precise underlying mechanisms regulating this age-dependent drift, definitive causal relations between methylation patterns and age-related functional decline and to which genes and pathways these changes are most important still needs to be determined (Jung & Pfeifer, 2015).

2.3.2 Histone modifications

Next to the alterations in DNA methylation, the neuronal histone code also experiences an extensive number of age-related changes. The accessibility of the DNA for transcription is largely determined by the conformation of histones, which can be adjusted through modifications of their N-terminal tails. There are endless possible combinations of various histone modifications involving many enzymes, allowing the histone code to be fine-tuned in a highly adaptable manner (Day & Sweatt, 2011). By the differential recruitment of histone-modifying enzymes, certain patterns of histone marks can be identified in the genome. Histone marks help to control genetic expression, for example by gene silencing via the formation of heterochromatin (H3K27me3) or by regulating genomic stability (H3K56ac) (Booth & Brunet, 2016). Recently, numerous histone marks have been identified that exhibit patterns of expression that change in an age-dependent manner. Histone marks showing changes in patterns during aging include H3K27me3, H3K56ac, H4K16ac, H3K4me3, H3K36me3 and H3K9me3 (Benayoun, Pollina, & Brunet, 2015). One of the most well-studied histone marks that is known to be consistently decreased with age is H3K9me3 (Booth & Brunet, 2016). H3K9me3 is

normally bound to heterochromatin. Global decrease in H3K9me3 – and loss of heterochromatin protein 1 (HP-1) – is therefore associated with a global loss of heterochromatin during aging (O’Sullivan & Karlseder, 2012). Loss of H3K9me3, HP-1 and heterochromatin has also been indicated in premature aging diseases such as WS and HGPS (Scaffidi & Misteli, 2006; Zhang et al., 2015). Therefore, H3K9me3 and HP-1 levels and the degree of heterochromatin are currently widely used as cellular markers of aging. However, results from several model organisms concerning other histone marks are contradicting, illustrating that these mechanisms and their role in the aging process are likely more complex. For example, levels of H3K27me3 show global decrease during aging in C. elegans (Ni, Ebata, Alipanahiramandi, & Lee, 2012). In contrast, global levels of H3K27me3 are increased in muscle stem cells of old mice and the brains of aged African Killifish (Baumgart et al., 2014; Liu et al., 2013). Moreover, whereas decreased expression of the H3K27me3 demethylase UTX-1 elongates lifespan in C.

elegans (Maures, Greer, Hauswirth, & Brunet, 2011), overexpression of another demethylase called

KDM6B/JMJD-3.1 has exactly the same effect (Labbadia & Morimoto, 2015). Similar to the observations in H3K27me3 mark, knockdown of H3K4me3 regulating enzymes can either increase or decrease lifespan dependent on the specific enzyme or context (Greer et al., 2010; McColl et al., 2008). Together, the association of these histone mark levels with both increased and decreased lifespan and the different effects of specific regulating enzymes on longevity suggests that these histone mark levels likely influence the aging process in a cell- and loci-specific manner (Benayoun et al., 2015). These examples of age-related changes in histone mark patterns only represent a very small part of the total number of known changes, and many more likely remain to be discovered. To fully understand the dynamics of histone mark patterns and their role in the aging process, the big challenge remains to uncover the

underlying mechanisms that regulate these patterns over time. The enzymes that regulate histone modifications act in large protein complexes that are recruited to the chromatin through the interaction with transcription factors. Histone mark patterns thus reflect the transcriptional activity of a locus, explaining the cell- and locus-specificity observed in the different animal studies. Changes in patterns of histone marks could thus be a result of age-related changes in histone modifying enzymes or differential recruitment of these enzymes to the chromatin (Booth & Brunet, 2016). However, the nature of these mechanisms is highly complex and are only beginning to be unraveled (Sen, Shah, Nativio, & Berger, 2016). A full comprehensive overview of all the histone-modifying enzymes indicated in the aging process is beyond the scope of this review. Therefore, only the best-known group of histone modifying enzymes implicated in the aging process, histone deacetylases (HDACs), will now be focused on. HDACs can fulfill many different roles from effecting gene expression to direct regulation of protein functions (Thiagalingam et al., 2003). One group of HDACs that has been implicated to play a specific role in aging is the SIRT family, also known as sirtuins (Mostoslavsky et al., 2006; Mouchiroud et al., 2013). These enzymes regulate lifespan in part by modulating caloric restriction pathways. Activity of SIRTs is dependent on nicotine adenine dinuclease (NAD+), an important cofactor regulating metabolic homeostasis. (Greiss & Gartner, 2009). Mammals have seven SIRT paralogs called SIRT1, -2, -3, -4, -5, -6, and -7. Several of them are known to improve aspects of healthy aging (Haigis & Guarente, 2006). For example, SIRT1 – which is downregulated during normal aging – improves aspects of general health in aging when overexpressed in mice (Herranz et al., 2010). Moreover, higher levels of SIRT1 have neuroprotective effects in mice models of AD (Godoy, Zolezzi, Braidy, & Inestrosa, 2014). SIRT1 deficiencies have shown to contribute to cognitive decline and neurodegenerative disorders (Cho et al., 2015). Additionally, mice

overexpressing SIRT6 have a longer lifespan compared to controls (Kanfi et al., 2012) and SIRT6 deficiencies lead to accelerated aging (Mostoslavsky et al., 2006). The impact of SIRT1 and SIRT6 deficiencies on aging is probably related to its role in DNA repair machinery. Upon DNA damage SIRT1 is relocated from the promotor sites to the site of DNA damage, changing gene expression in a way that features cellular aging (Oberdoerffer et al., 2008). Additionally, the recruitment of SIRT6 to sites of DNA damage remodels the local chromatin environment through deacetylation of H3K56 and thereby helps DNA repair complexes to access the DNA (Toiber et al., 2013). Changes in SIRT1, SIRT6 and H3K56 acetylation can thus influence the abilities of DNA damage repair upon aging. Overall, these studies indicate a potential role of the sirtuin family in the process of aging. However, it is important to note that despite the current knowledge on sirtuin function, the role of most histone modifying enzymes in the aging process remains largely unknown and is only beginning to be reported. Characterizing the specific mechanisms underlying histone modifying enzymes and their interactions with other compounds such as transcription factors will further clarify the exact impact and dynamics of histone modifications on aging.

2.3.3 Micro-RNA

The role of micro RNA (miRNA) in age-related epigenetic changes is also beginning to be unraveled. miRNA has a great variety of functions ranging from cell cycle regulation to tumor suppression (Grillari & Grillari-Voglauer, 2010). One single miRNA can have hundreds of targets, illustrating the complexity of miRNA transcription for genome function (Vergoulis et al., 2012). Recently, several studies have implicated a potential role of some specific miRNAs in neurological aging and the development of neurodegenerative diseases (Chen, Chiou, Chen, Li, & Chiou, 2010; Maciotta, Meregalli, & Torrente, 2013). Persengiev and colleagues (2011) performed a genome-wide analysis

of miRNA expression and found a potential role for miR-144 in neurological aging. In contrast to global miRNA downregulation, miR-144 was found to be upregulated in the aging cortex and cerebellum of humans, chimpanzees and rhesus macaques. MiR-144 targets the ataxin-1 gene, which is involved in the development of spinocerebellar ataxia type 1. Age-related dysregulation of miR-144 could thus be linked to the development of this neuro-degenerative disease. Global proteomic profiling of mice brains of different ages revealed a general upregulation of miRNAs over time (N. Li, Bates, An, Terry, & Wang, 2011). Of the 70 miRNA that were found to be upregulated, 27 were implicated in the downregulation of mitochondrial complexes III, IV, and F0F1-ATPase, providing a link between aberrant miRNA expression and age-related declines in mitochondrial respiration rates (N. Li et al., 2011). Although significant advances have been made in the potential role of miRNA in brain aging, additional research should be performed to create a thorough understanding of post-transcriptional gene regulation in normal brain aging.

2.4 Summary

To again depict the extreme complexity of the aging process, the above described mechanisms only cover a minority of all the changes observed with aging and even in the processes of these primary hallmarks a lot of questions remain unanswered. A thoroughly mechanistic understanding of the process remains to be obtained. Due to the highly pleiotropic nature of this process, it is practically impossible to differentiate between main causative and secondary effects. Furthermore, it is nowadays still debated whether aging is an effect of damage-accumulating events or an actively regulated state that is preprogrammed into an organism (Kirkwood & Melov, 2011). Despite these remaining uncertainties, the wealth of information that is available of the physiological phenomena on a cellular and molecular level has provided

opportunities to design new methods to measure and mimic cellular aging.

3. MECHANISMS OF REPROGRAMMING TO INDUCED PLURIPOTENCY

The aging profile of a somatic cell seems to be erased during reprogramming into induced pluripotency. In this process, cells are ‘rejuvenated’ and turned back into an ES cell like state in which all age-related features seem to be reset (Buganim, Faddah, & Jaenisch, 2013). For example, reprogramming seems to trigger an increase in telomere length (Lapasset et al., 2011; Marion et al., 2009), shows mitochondrial metabolism and properties comparable to young cells (Prigione et al., 2011; Suhr et al., 2010) and creates ES cell like gene expression profiles (H. Li et al., 2009). Although rejuvenation has been well-studied from a phenotypic perspective, little is known about the genomic dynamics that conduct this transition (Cornacchia & Studer, 2016). Induction of a state of pluripotency is accompanied by massive molecular changes. For a cell to become pluripotent, a whole set of somatic genes needs to be downregulated and a new set of stem cell genes needs to be expressed. In order to do so, cells need to reset their epigenetic state. This entails a genome-wide remodelling of chromatin modifications such as DNA and histone methylation (Mikkelsen et al., 2008). The exact mechanisms and temporal sequence of this process is not yet known (Papp & Plath, 2011; Stadtfeld & Hochedlinger, 2010). Epigenetic changes are one of the primary hallmarks of aging and it is therefore of great importance to take a close look at the process of rejuvenation and the epigenetic state of an iPSC after reprogramming. When mimicking or measuring cellular aging in vitro, one would ideally possess a thorough understanding of the age profile of the used iPSC cell cultures. The next section will therefore describe the epigenetic changes that have been observed during reprogramming, and to what extent this process of rejuvenation resets the cellular aging profile.

3.1 Epigenetic mechanisms of reprogramming In lineage-specific somatic cells, chromatin is mostly present in a highly condensed heterochromatin state (Efroni et al., 2008). During iPSC generation the heterochromatin needs to be remodelled to an ES cell like state, with mainly euchromatin and loosely organized heterochromatin (Fussner et al., 2011). Since iPSC reprogramming is accompanied by a large amount of partially reprogrammed cells, epigenetic status can be analysed at different stages of the reprogramming process. Although little is known about the mechanisms underlying epigenetic changes, remodelling seems to take place in a coordinated and sequential manner and is established by changes in DNA methylation and histone modifications. Current knowledge on DNA methylation dynamics during reprogramming is mainly focused on changes associated with activation of pluripotency genes (Schmidt et al., 2012). In a somatic cell CpG-rich promoters tend to be hypomethylated and CpG-poor ones hypermethylated (Meissner, 2010; Meissner et al., 2008). DNA methylation at promotor sites is associated with repressive chromatin environment, and thus inactivity of corresponding genes (Mohn et al., 2008). In ES cells, genes associated with development are hypomethylated – and thus activated - and tissue specific genes are hypermethylated – and thus inactive. During reprogramming, pluripotency genes that are hypermethylated in somatic cells must be demethylated and activated. Insufficient demethylation has shown to be a limiting factor of the generation of iPSCs, indicating that loss of DNA methylation is a crucial part of the reprogramming process (Papp & Plath, 2011). However, the underlying mechanisms of DNA demethylation and how this process is targeted to the relevant genes remains elusive (Liang & Zhang, 2013). More importantly, due to the lack of a thorough understanding of the mechanisms underlying DNA (de)methylation during reprogramming it is currently impossible to assess the impact of these changes on the age-related methylation

status of iPSCs. Future research should unravel the mechanisms underlying DNA (de)methylation during reprogramming, not only to gain a better understanding of the process of pluripotency induction, but in particular to gain new insights into the consequences of reprogramming on the age-related methylation state of the cells of origin. Concerning histone modifications, comparative analysis of partially and fully reprogrammed cells show that rearrangement of the heterochromatin is characterized by the presence of H3K9me3 and HP-1 and precedes the expression of the pluripotency gene NANOG. With NANOG activation, euchromatin marks occur. These findings indicate that activation of pluripotency circuitry seems to be crucial in chromatin remodelling (Mattout, Biran, & Meshorer, 2011). Although partially reprogrammed cells provide for a powerful tool to study intermediate steps of the reprogramming process detailed characterization of changes at a molecular level is difficult due to the low percentage of cells that successfully become iPSCs. In order to overcome this issue, Koche and colleagues (2011) have been examining epigenetic changes at the first cell cycles after iPSC induction. They observed rapid, genome-wide changes in euchromatin histone modifications at more than a thousand loci mainly coding for pluripotency-related or developmentally-related gene promotors and enhancers. These early events in chromatin remodelling preceded the onset of pluripotency gene expression, indicating that chromatin status of a somatic cell normally restricts the activity of reprogramming factors. Patterns of the repressive H3K27me3 modification remained largely unchanged, indicating that drastic transition of epigenetic landscapes has yet to occur after these initial changes (Koche et al., 2011). As for DNA methylation, changes in histone modifications have also mainly been studied to gain a better understanding of the reprogramming process. How and to what extent these observed molecular changes erase the aging signature therefore still remains to be

unravelled. However, based on the magnitude of the changes needed to induce pluripotency and the corresponding ES cell like age-phenotype of iPSC cultures, one could assume that most age-related epigenetic marks are erased during the reprogramming process.

3.1.1 Is everything lost?

Although many aspects of aging are reset during reprogramming into pluripotency, there are some aspects that are likely not reversible. For example, the accumulation of DNA damage that is observed in the aging process is presumably still present after reprogramming. Several studies have described iPSC lines from old individuals to contain genetic aberrations (Boulting et al., 2011; Prigione et al., 2011). However, the observed genetic aberrations in iPSC culture are not only found to be pre-existent in the cells of origin, but are also known to randomly accumulate as a result of the reprogramming process, the process of re-differentiation or when keeping iPSCs in cell cultures (Gore et al., 2011; Laurent et al., 2011). Whether – and to what extent – the observed genetic aberrations are truly a representation of the age profile of the tissue of origin, or merely a result of the reprogramming process and protocol remains to be clarified. Next to the possible irreversible DNA damage, recent evidence from mouse and human iPSC studies have shown that cells maintain an epigenetic memory upon reprogramming (Doi et al., 2009; Lister et al., 2011; Polo et al., 2010). This epigenetic memory is mainly present in the DNA methylation signature, retaining methylation patterns characteristic for the tissue of origin. Interestingly, this memory is not necessarily present at the pluripotency state, but manifests itself later, when iPSCs are re-differentiated into specific cell types (Ohi et al., 2011; Polo et al., 2010). Currently, no studies have yet characterized whether this epigenetic memory also contains age-associated marks. It would be interesting to further investigate whether iPSCs of old cells retain a memory

of their age upon reprogramming, and how this memory manifests itself in DNA methylation and/or histone modifications (Mahmoudi & Brunet, 2012). Conclusively, despite the implication that some of the primary hallmarks of aging – such as DNA damage and epigenetic memory – might still be present after the reprogramming process, current evidence indicates that the age profile of somatic cells is erased upon induction of pluripotency. The massive molecular changes necessary to obtain iPSC cultures appear to create cells that are aged zero at a molecular scale. It is therefore that a new challenge arises once successful reprogramming is accomplished and iPSCs are born: the challenge to get them old again.

4. CURRENT STATE OF MODELING AGE IN VITRO

Various methods have been proposed to retrieve the age-related features in iPSC models of neurological aging. In the following section I will critically review each of those paradigms. Additionally, I will discuss recent development of aging models within the field of direct reprogramming.

4.1 The classical aging paradigm

The classical strategy to induce age-like features in iPSC cultures is stress induction. Researchers have modeled late-onset diseases by exposure to toxins which trigger mitochondrial stress and ROS production (Byers et al., 2011; O. Cooper et al., 2012; Nguyen et al., 2011; Seibler et al., 2011). For example, Nguyen and colleagues (2011) modeled neurodegeneration in iPSCs by exposure to hydrogen peroxide. Exposure to hydrogen peroxide was found to induce oxidative stress, expressed by an upregulation of associated genes. iPSCs derived from patients with PD showed a higher increase in levels of neurodegeneration when exposed to hydrogen peroxide as compared to iPSC cultures of healthy controls. Inducing oxidative stress created iPSC cultures exhibiting early phenotypes linked to PD. It was suggested that these cells could provide a valuable

platform for identification of novel pharmacological agents and diagnostics for modeling and alleviation disease phenotypes (Nguyen et al., 2011). However, it remains to be determined whether inducing neurodegeneration by hydrogen peroxide or using alternative methods of forcing disease phenotypes to ‘young’ cells appropriately mimics age-related disease susceptibility. It may be equally – if not more – important to unravel the dynamics of protective mechanisms that fend off disease. Additionally, researchers have not yet assessed to what extend stress paradigms act upon the hallmarks of cellular aging. Therefore, in vitro models would ideally elicit the hallmarks of cellular aging in iPSCs of healthy and patient-specific tissues without the application of toxicity. Hence, it is more desirable to consider more sophisticated methods targeting these age-related hallmarks (Studer et al., 2015).

4.2 Premature aging diseases

Insights from premature aging disorders such as HGPS, Dyskeratosis Congenital (DC), WS and Cockayne syndrome (CS) have led to the development of alternative strategies for in vitro models to study neurological aging in both health and disease (Kipling, Davis, Ostler, & Faragher, 2004). iPSC models of premature aging have demonstrated a reset of the age-associated phenotypes (e.g. loss of nuclear defects, absence of progerin and disease associated epigenetic alterations) during reprogramming. However, upon re-differentiation these age-associated cellular features are rapidly reacquired (Agarwal et al., 2010; Andrade, Nathanson, Yeo, Menck, & Muotri, 2012; Batista et al., 2011; Liu et al., 2011; Zhang et al., 2011). These findings suggest that the molecular mechanisms underlying premature aging diseases could potentially be utilized to mimic aging in iPSC-based models (Vera & Studer, 2015).

4.2.1 HGPS

The most well-studied premature aging disease with respect to the modeling of aging is HGPS. A study by Liu

et al. (2011) described the use of iPSCs derived from patients with HGPS (HGPS-iPSCs) to induce age-related features. They studied vascular aging by differentiating HGPS-iPSCs to vascular smooth muscle cells (SMCs). Upon differentiation, SMCs restored the altered structure of the nuclear envelope, progerin levels, as well as the epigenetic modifications that are associated with HGPS. As a result, HGPS-iPSC-derived SMCs reached senescence-related phenotypes earlier than their normal counterparts. As mentioned before, alterations in progerin levels are also observed in normal aging (Scaffidi & Misteli, 2006). It was therefore suggested that this model could be utilized in the study of vascular aging in both healthy and diseased conditions (Liu et al., 2011). This study was performed for iPSC-derived SMCs and it remains to be examined whether these results will stand for iPSC-derived neuronal tissue. Expression levels of progerin are found to differ across various tissues in patients with HGPS (Röber, Weber, & Osborn, 1989). This tissue specific expression is likely the reason why the disease does not affect all organ systems equally. Progerin expression in the CNS is thought to be relatively low, due to the expression of miR-9, which targets lamin A and progerin (Nissan et al., 2012). HGPS-iPSC-derived neurons might thus not be the most suitable strategy to mimic neurological aging. An alternative approach to use HGPS-associated accelerated aging was proposed by Miller and colleagues (2013). They described a strategy to induce aging-related features in hiPSC-derived lineages to study the impact of aging in a PD genetic background. Instead of using iPSCs derived from patients with HGPS they overexpressed progerin in healthy iPSC-derived neurons by using synthetic modified RNA (mRNA). Progerin overexpression induced multiple aging related markers and characteristics including nuclear morphology abnormalities, loss of LAP2a expression, formation of DNA double-strand breaks, loss of heterochromatin markers (H3K9me3 and HP-1), increased mtROS and shortened telomeres. Moreover, progerin expression

in PD iPSC-derived dopamine neurons revealed a disease phenotype that involved dendrite degeneration, progressive loss of tyrosine hydroxylase - an enzyme that converts L-tyrosine to L-3,4-dihydroxyphelylalanine (L-DOPA) and is therefore a DA marker – and enlarged mitochondria or Lewy-body-precursor inclusions. The emergence of this phenotype was dependent on both aging and genetic susceptibility. This model provides a more refined method to study the contribution of aging to the disease process. Therefore, progerin-induced aging offers a new paradigm to study late onset age-related disease features (Miller et al., 2013). An important limitation of the current studies is that an in-depth analysis of the molecular changes associated with progerin exposure or expression has not yet been performed. Therefore, the possibility that progerin may trigger age-related markers via molecular mechanisms that are distinct from normal physiological aging cannot be excluded. In order to further examine whether progerin expression does represent true physiological aging, future studies could include additional aging markers such as methylation patterns of CpG-sites. Moreover, additional fibroblast lines covering the aging spectrum on a higher resolution would refine the results and uncover whether observed differences in the aging markers occur gradually or as a binary ‘on/off’ switch (J. D. Miller et al., 2013)

4.2.2 Other syndromes

Some research has been conducted into the use of other premature aging syndromes such as WS, DC and CS. Although these studies did not model neurological aging in iPSC-derived neurons, they show some promising results that might be implicated in future research. A recent stem cell study demonstrated the use of the WS mutation to induce age-associated alteration in heterochromatin. Upon differentiation of WRN-null ES cells to mesenchymal stem (MS) cells, cell cultures showed features of cellular aging such as a

global loss of H3K9me3 and changes in heterochromatin architecture. Decrease in WRN and heterochromatin marks were also detected in MS cells from older individuals indicating that this phenomenon is also present in normal physiological aging (Zhang et al., 2015). These preliminary findings imply promising opportunities for the use of WS in modeling aging in vitro. Future research might further evaluate this paradigm by applying it to iPSCs and create differentiation protocols towards neuronal tissue. This research direction may however not be without obstacles. The authors claim to have tried to create WS-specific iPSCs, but found that WS patients’ fibroblast lines showed severe karyotypic abnormalities which obstructed reprogramming capacities (Zhang et al., 2015) . Nevertheless, it would be interesting to further examine the possibility of creating WS-iPSC-derived neurons to use as a model of neurological aging. Cockayne syndrome (CS), another premature aging disease, is caused by mutations in the excision-repair cross complementing group 6 or 8 gene (ERCC6 & ERCC8 respectively). Patients with CS show neurological and developmental abnormalities. At the molecular level, CS is characterized by a deficiency in the transcription-couple DNA repair pathway (Laugel et al. 2010). Andrade and colleagues (2012) characterized the role of this mutation in the generation and development of iPSCs. They found that CS-iPSCs showed greater cell death rate and higher production of ROS. This phenotype was accompanied by upregulation of TP53 and TXNIP. This study demonstrated the possibility of successfully creating iPSCs derived from patients with CS. Since CS normally causes abnormalities in neurological tissue, CS-iPSCs-derived neurons may be a promising tool to mimic neurological aging in a dish. Future studies should therefore further characterize the molecular mechanisms underlying CS-derived iPSCs, and examine whether these cultures can be successfully differentiated into neuronal tissue. Lastly, some studies have investigated the effects of dyskeratosis

congenital (DC) on undifferentiated iPSC phenotypes. DC is a premature aging disease associated with mutations in genes controlling telomere homeostasis, leading to premature shortening of telomeres mainly affecting blood, pulmonary tissue, and epidermal tissues (Walne & Dokal, 2009). Contrasting results were found in which iPSC derived from patients with DC were reported to have restored telomere homeostasis due to the activation of pluripotency genes (Agarwal et al., 2010) or still contained defects in telomere length characteristic of the disease (Batista et al., 2011). These contradicting findings may be due to the fact that DC can be caused by several different mutations. Each distinct form of DC leads to a different degree of perturbations in telomere homeostasis. Further characterization of each form of DC might clarify to what extent telomere shortening can be modeled in iPSCs. Furthermore, it should be investigated whether iPSCs that show restored telomere lengths in undifferentiated conditions might retrieve disease phenotypes upon differentiation. A model of aberrant telomere homeostasis would be particularly of interest to study the role of telomere shortening in its susceptibility to late-onset disease development.

4.3 Taking the shortcut: skipping pluripotency state

As reprogramming of somatic cells has been a huge area of research, different methods have been developed to reach efficient conversion. More recently, scientists have developed methods of direct reprogramming, in which the state of pluripotency can be skipped. (Corti et al., 2012; Ring et al., 2012; Vierbuchen et al., 2010). This direct reprogramming can be achieved by overexpression of specific sets of lineage-specific reprogramming factors in combination with the appropriate culture conditions. For example, Vierbuchen and colleagues (2010) identified a combination of three transcription factors, ASCL1, BRN2, and MYT1L, that were together able to

rapidly and efficiently convert mouse fibroblasts into functional neurons – called induced neurons (iNs) – in

vitro. It was hypothesized that without the

intermediate state of pluripotency cells are not required to undergo rejuvenation, indicating that neuronal tissue from this reprogramming method might maintain its age-related (epi)genetic signature. Recently, researchers have begun to characterize the cellular profile of these neuronal tissues with respect to retaining age-related features.

4.3.1 Maintaining age in directly reprogrammed neurons.

Mertens et al (2015) were the first to assess the issue of aging in direct reprogrammed cells. They performed a comprehensive comparative analysis of the transcriptome of different cell types to compare the differences in age-associated signatures. Human fibroblasts, iNs, iPSCs, iPSC-derived neurons and brain samples of 19 individuals were collected with age ranging from 0 to 89 years. iNs were obtained by overexpressing the proneural genes ASCL1 and NGN2 together with a small-molecules cocktail to enhance direct conversion efficiencies (Ladewig et al., 2012). In line with expectations, iPSCs and iPSC-derived neurons did not retain age-associated gene signatures present in the donors’ fibroblasts. On the contrary, iNs maintained age-specific transcriptional profiles. Further examination of the transcriptional profiles revealed three genes that were found to be differentially expressed in an age-dependent manner, namely LAMA3, PCDH10, and RanBP17. The authors concentrated on RanBP17, a nuclear pore-associated transport receptor responsible for proper compartmentalization of proteins containing nuclear localization signals. Knock down of RanBP17 caused loss of nucleocytoplasmic compartmentalization (NCC) in young cells. This age-associated decline in RanBP17 and corresponding loss of NCC was present in human fibroblast, iNs, and brain tissue samples. iNs thus retained important age-associated signatures that are

lost during reprogramming into iPSCs. iN conversion represents a promising new technology in which skipping the pluripotent state provides possibilities to create human neurons that reflect important aspects of cellular age. Interestingly, only 4% of age-associated genes showed overlap in expression between fibroblasts and corresponding iNs. When comparing iNs to postmortem tissue of the cortex, a 7-fold more (49 genes) overlap was found, indicating that the aging transcriptome is cell-type-specific and might become in some way translated to a neuronal context (Mertens et al., 2015). The current study hypothesized that aging might be regulated by only a small number of master genes. If these genes get disrupted along time, important down-stream processes might be impaired that eventually unfold as phenotypical aging. RanBP17 might be one of these master genes (Mertens et al., 2015). However, exactly how cellular age is encoded during direct conversion of fibroblasts to induced neurons remains to be clarified. To unravel this process, it would be interesting to characterize the epigenetic age-associated changes on the level of DNA methylation at CpG sites, histone marks such as H3K9me3 and miRNA profiles in these iNs as compared to their corresponding fibroblasts. Another promising new reprogramming procedure was recently developed using neuronal microRNAs (miRNAs) miR-9/9* and miR-124 (miR-9/9*-124). This method successfully converted human fibroblasts into striatal medium spiny neurons (MSNs) (Richner, Victor, Liu, Abernathy, & Yoo, 2015). Since this neuronal conversion skipped state of pluripotency during reprogramming it was reasoned that the derived neurons would contain age-related signatures of the donor age. In a very recent study, neurons derived from this reprogramming paradigm were subjected to a comprehensive analysis of maintenance of cellular markers of aging (Huh, Zhang, Victor, Dahiya, Batista, Horvath, & Yoo, 2016). Analysis was performed at different levels assessing DNA methylation, miRNA profiles, DNA damage

accumulation, telomere length and ROS levels. Firstly, DNA methylation of these directly converted neurons was analyzed (n=16, aged 3 days to 96 years) and compared to methylation profiles of corresponding fibroblasts (n= 37, aged 3 days to 94 years). In order to assess DNA methylation levels, the authors used the epigenetic clock, a method that accurately measures chronological age in human tissue (Horvath, 2013). The methylation status of 353 specific CpG loci were analyzed, leading to an estimation of DNA methylation (DNAm) age. Comparison of converted neurons with corresponding fibroblasts of the same age revealed a near-perfect correlation (𝑝 = 0.91), indicating that the epigenetic clock was unaffected during direct reprogramming (Huh, Zhang, Victor, Dahiya, Batista, Horvath, Yoo, et al., 2016). Secondly, Huh and colleagues (2016) performed an analysis of the transcriptome and miRNA profiles of the two different cell types at young and old age. Both fibroblasts and converted neurons showed a cohort of upregulated and downregulated genes with aging that was in line with earlier findings (Mertens et al., 2015). Additionally, miRNA profiling detected fourteen miRNAs showing an age-dependent increase of expression in both fibroblast and neurons. MiRNAs that were most prominently upregulated during aging included miR-10a, miR-497, and miR-195. These miRNAs are implicated in age-related processes of metabolism, cellular death and survival. Although the exact role of these miRNAs in aging is not yet well understood, these results support the idea that reprogrammed neurons retain an age-associated signatures on the miRNA level (Huh, Zhang, Victor, Dahiya, Batista, Horvath, Yoo, et al., 2016) Thirdly, converted neurons were assayed for cellular hallmarks of aging including oxidative stress (ROS accumulation), DNA damage and telomere shortening. FACS analysis of ROS levels demonstrated increased ROS levels in old as compared to young reprogrammed neurons, a difference that was also found in fibroblasts. Next, DNA damage accumulation in old versus young

reprogrammed neurons and fibroblasts was assessed by Comet Assay (Singh, Danner, Tice, Brant, & Schneider, 1990). Both old neurons and old fibroblast showed increased comet tail length, displaying more extensive DNA damage accumulation over time. Lastly, telomere length of converted neurons as compared to the age-matched fibroblasts was practically identical (Huh, Zhang, Victor, Dahiya, Batista, Horvath, Yoo, et al., 2016). Conclusively, this recent report provides new opportunities in modeling neurological aging in a dish via the use of miRNA conversion. This reprogramming paradigm has provided new insights in key signatures associated with aging. A major strength of this report is that in reprogrammed neurons age-associated signatures are consistently preserved on the genomic, epigenetic, and cellular level. It should be noted that the current study was performed on a particular subtype of neurons, namely MSNs. Future studies should investigate whether miRNA-based reprogramming can be applied when differentiating into other neuronal subtypes and if so, whether maintenance of key age-associated signatures is consistent in other differentiation protocols.

5. DISCUSSION

A decade ago the field of in vitro stem cell models was revolutionized by the introduction of iPSCs. Since then giant leaps have been taken in the manipulations of cellular fate, enabling the re-creation of virtually any type of cell from pluripotent stem cells in a dish. In order to create realistic in vitro replicas of human tissues, a new challenge has now arisen: the challenge to re-create age. The ability to control cellular aging in a dish will boost the field of stem cell models even further, creating new possibilities to research the effects of aging on late-onset neurological and neurodegenerative diseases.

5.1 Challenges

Manipulating cellular age has so far been found to be an exceptionally bold endeavor, mostly due to the fact

that the processes of biological aging remain poorly understood. Although the process of aging is considered to be highly complex including changes at the levels of genomic instability, epigenetic regulation, proteostasis, telomeres, mitochondrial functioning, nutrient sensing mechanisms, cellular senescence, and changes in intercellular communication (López-Otín et al., 2013), most researchers have only been able to manipulate one aspect at a time of these cellular hallmarks of aging. It therefore often remains to be determined whether certain manipulations replicate true physiological aging, or whether the observed phenotype is just limited to inducing one specific associated hallmark, or is even constructed via pathways independent of aging processes. Additionally, iPSC technology represents some experimental shortcoming that are currently complicating the interpretation of results. Firstly, during reprogramming random accumulation of epigenetic and genetic abnormalities have been observed. These abnormalities can be different in iPSCs derived from the same donor, indicating that the process of reprogramming does not rejuvenate cells in an invariable manner (Nasu et al., 2013). Moreover, the observed accumulation of epigenetic and genetic abnormalities during the reprogramming process is thought to create iPSCs that might have an aberrant functionality and do not represent cells in in vivo conditions. Next to the observed differences of pluripotency state that seem to occur randomly as a result of the reprogramming process, the pluripotency state of iPSCs also seems to contain an epigenetic memory determined by factors such as the tissue type of origin and the donors age (Frobel et al., 2014; Horvath, 2013). These factors can influence the (epi)genetic state of pluripotent stem cells and subsequent iPSC-derived neuronal cultures. These observed differences in epigenetic memory are of particular importance when considering proper control cell lines. Comparing iPSCs and iPSC-derived neuronal tissue from patients versus healthy subjects,

young versus old donors or from different cell types-of-origin is far from ideal, since each iPSC cell line carries a different set of epigenetic memories. These differences in epigenetic memory make it difficult to analyze and compare the effects of certain manipulations affecting epigenetic regulation that are performed to model aging. In order to overcome the experimental variability in pluripotency induction, future studies should concentrate on gaining new insights in the actual processes underlying reprogramming and the associated reset of age. A limitation in this field of study is again the lack of knowledge on the basic mechanisms of aging. Namely, assessment of the pluripotency state is currently based on only a small set of age-related alterations. In order to investigate age-related signatures in iPSCs as a result of reprogramming failures, future research should look into all aspects of aging. 5.2 Promises Although the process of aging is not well understood, sometimes processes can be manipulated even without a thorough understanding of the mechanisms. This is shown for example by the progerin-induced strategy as proposed by Miller et al. (2013), illustrating a more sophisticated approach in which both aging and genetic susceptibility are required to reveal a PD disease phenotype. Future studies should further characterize the molecular pathways of progerin-induced aging. When progerin-progerin-induced aging does indeed represent true physiological aging, research should work towards applying this strategy, in which aging can be incorporated as a risk factor, to multiple disease models. Additionally, the successes achieved by methods of direct reprogramming (Huh, Zhang, Victor, Dahiya, Batista, Horvath, Yoo, et al., 2016; Mertens et al., 2015) also illustrate successful manipulations of age without thorough understanding of the aging and reprogramming processes. The strength of these studies lies within the extensive nature of the analyzes. By either performing a

thorough comparison between direct reprogrammed neurons and iPSC-derived neurons (Mertens et al., 2015) or by assessing multiple hallmarks of cellular aging – epigenetic regulation; miRNA status; telomere lengths; ROS-production; DNA damage – the authors demonstrate direct reprogramming strategies to show great potential in retaining age-related features in in

vitro models. Although the technique of direct

reprogramming is still in its infancy, these first studies open the door to investigate preservation of age-related features in previously proposed methods of direct reprogramming. For example, one successful direct reprogramming paradigm that would be interesting to further investigate on age-related features is conversion of fibroblast to functional neurons by repression of RNA binding polypyrimidine-tract-binding (PTB) protein as proposed by Xue et al. (2013). Next to the exciting advancements in the field of direct reprogramming, the great potential of iPSC aging models should not be underestimated. Future research directions should aim to apply the rapid gain of understanding the mechanisms of human aging in iPSC technology. So far, iPSC strategies have mainly focused on the application of premature aging diseases targeting the aging hallmark of genomic instability by induction of a mutation (Andrade et al., 2012; G.-H. Liu et al., 2011; W. Zhang et al., 2015) or defects in DNA repair systems and nuclear defects (Miller et al., 2013) or by targeting telomere attrition (Agarwal et al., 2010; Batista et al., 2011). However, the hallmark of epigenetic regulation has not yet been investigated as a possible target to induce aging. Aging is associated with overall epigenetic depression, represented by a decrease in DNA methylation and histone modifications (Gonzalo, 2010; Hernandez et al., 2011; Siegmund et al., 2007). Additionally, an increase of DNA methylation is observed at particular sites (Hernandez et al., 2011; Siegmund et al., 2007). Exploring the possibility to induce chromatin alterations to model aging might therefore open up new avenues in understanding the mechanisms of