TELOMERE EXTENSION USING MODIFIED TERT Mrna TO

TELOMERE EXTENSION

USING MODIFIED TERT mRNA

TO LENGTHEN HEALTHSPAN

A DISSERTATION

SUBMITTED TO THE NEUROSCIENCES PROGRAM

AND THE COMMITTEE ON GRADUATE STUDIES

OF STANFORD UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

John Ramunas

May 2014

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/vb798wq6556

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Helen Blau, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Michael Longaker

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Juan Santiago

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Robert Sapolsky

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost for Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii ABSTRACT

One of our long-term goals with this work is to extend the human health span, the period of life when humans are relatively free from age-related disease. Molecular mechanisms that limit the human health span include epigenetic drift, accumulation of cellular waste products,

DNA damage, and telomere shortening, and here we demonstrate a method to address telomere shortening.

Telomeres comprise DNA sequences that protect the ends of chromosomes but shorten over time due to oxidative damage and incomplete DNA replication during S phase of the cell cycle. Short telomeres can lead to activation of p53, cell cycle arrest or apoptosis, chromosome- chromosome fusions, or malignancy. Short telomeres are implicated not only in age-related diseases including cancer and heart disease, but in diseases in which cells are under high replicative demand, such as muscular dystrophy.

The nucleoprotein telomerase comprising the protein component TERT and RNA component TERC extends telomeres, and several efforts at extending telomeres have focused on increasing the amount of telomerase in cells, which conveys proliferative capacity to cells in culture and can reverse phenotypes of aging in animal models. Currently small molecule activators of telomerase are in use by humans, but have no detectable effects on telomere length in many subjects. Adeno-associated viral delivery of TERT is another promising approach, one that extends rodent lifespan without increasing the incidence of cancer possibly because it is usually episomal and diluted out in fast-dividing cells, but risks genomic integration and resulting constitutive TERT expression which may not be acceptable in longer-lived species.

Here we present an alternative method for telomere extension made possible by the recent discovery that delivery of mRNA comprising modified nucleotides such as pseudouridine

iv modulates the Toll-like receptor mediated innate immune response that is activated in most cells in response to unmodified mRNA. Nucleoside modifications occur naturally in mammalian

RNA and provide a means of distinguishing endogenous from exogenous RNA such as bacterial

RNA which have fewer or no such modifications.

We find that delivery of modified mRNA encoding TERT to fibroblasts and myoblasts results in transient elevation of telomerase activity, telomere extension, and increased proliferative capacity. All cells treated to date have eventually senesced and expressed markers of senescence to the same degree as untreated cells, important for the safety of our approach.

Repeated treatment increases proliferative capacity further, suggesting that the approach may be useful over a prolonged period. We have taken initial steps at delivering mRNA in vivo and this work continues.

ACKNOWLEDGEMENTS

I would like to thank my thesis advisor and mentor Helen Blau for her unwavering and strong support; creativity; highly energetic connecting of ideas, technologies, and people; excellent judgment about what matters and what does not in science and in life; for an exceptionally well-run and well-stocked lab; and for never shying from raising the bar higher. I am also deeply grateful to my co-advisor Juan Santiago for being a role model in applying engineering thinking to biological problems, for his outstanding intuition about how anything works, his excellent advice, strong support, and wisdom over the years. I would also like to thank my thesis committee members Robert Sapolsky and Michael Longaker for their excellent biological judgment and advice that has steered us well. Robert went to great lengths at the start of my graduate program to help guide my rejuvenation dreams in the right direction, and his advice was spot-on then and continues to be in each of our meetings. I would like to thank our key collaborator John Cooke for his perspective, insights, and positive energy, and for recruiting my scientific brother Eduard Yakubov. I would like to thank Eduard for his friendship, creativity, molecular biology expertise and transcribing many thousands of micrograms of mRNA, wisdom, and constancy. I am grateful to Jen Brady for expertly answering every one of my endless questions about molecular biology. I thank my father Tony Ramunas and mother

Susan Fletcher for their love that has shaped my perspective and default mental state well, for focusing my attention on health on a daily basis, for teaching me to aspire to change the human condition, and for their infinite support as we all follow that path together. I would like to thank my step-father John Freeman for his support and many insights, and my brother Alan and his wife Jen and their kids for their love. I would like to thank Colin Holbrook for his passion for perfection and deep interest in science, Moritz Brandt for inspiring me to know more, and Viktor

Shkolnikov for his friendship, deep knowledge of Russian and other science, technical skill, dedication, and the many successes and failures we have enjoyed and learned from together. I would also like to thank Robin Holbrook, Peggy Kraft, and Kassie Koleckar for your immense skills and your essential help that have, on a daily basis, helped immensely to make this project happen. I would like to thank the other Blau lab members for their valuable advice and insight, and for their good nature and humor that never ends: Glenn Markov, Russ Haynes, David Burns,

Faye Mourkioti, Matt Decker, Andrew Ho, Penney Gilbert, Karen Havenstrite, Paul Cook,

Stephane Corbel, Ben Cosgrove, Ermelinda Porpiglia, Erika Cornell, Nora Yucel, Srihari

Sampath, Srinath Sampath, and Eva Moreno. I would like to thank the members of the Santiago lab and the AntiHero team for your engineering skill, help, and camaraderie: Karl Stahl, Angus

Pacala, Mary Reynolds, Curran Kaushik, Francisco de la Paz, Ken Lopez, Michele Dragoescu,

Giancarlo Garcia, Anita Rogacs, and David Fenning. I would like to thank Dr. Zane Cohen for a surgery very well done. I would like to thank Ross Colvin for his expert help at every turn in my degree. Finally I would like to thank John Huguenard for being open to pursuing rejuvenation many years ago and for supporting me through the years. I am very happy to be able to continue to work on this project with the same wonderful, passionate, and skillful team.

May we see the products of our work in use.

vii

TABLE OF CONTENTS

ABSTRACT ...... iv ACKNOWLEDGEMENTS ...... vi List of Figures ...... x List of Tables ...... xii Chapter 1 Introduction to telomere shortening and extension ...... 1 Telomeres ...... 2 Mechanisms of telomere shortening ...... 3 Telomerase ...... 5 Telomere shortening and disease ...... 7 Genetic mutations in telomere maintenance genes ...... 8 Muscular dystrophy ...... 8 Cognitive decline ...... 9 Hypertension and heart disease ...... 9 Cancer ...... 10 Immunosenescence ...... 11 Limits imposed by telomere shortening on cell and tissue engineering ...... 13 Criteria for a therapeutically useful telomere extension method ...... 14 Critical evaluation of methods of extending telomeres ...... 16 Nucleoside-modified mRNA ...... 17 Exosomes ...... 21 Summary ...... 22 Chapter 2 Telomere extension in cells using TERT mRNA...... 23 Forward ...... 24 Introduction ...... 24 Results ...... 28 Discussion ...... 58 Materials and Methods ...... 60 Acknowledgements ...... 68 Chapter 3 In vivo delivery of mRNA ...... 69 Introduction ...... 70 Criteria for selection of delivery methods ...... 70 Survey of existing RNA delivery methods ...... 71 Results ...... 71

viii

Cationic lipids ...... 71 Naked mRNA...... 74 High-density lipoprotein ...... 74 Albumin ...... 77 Exosomes ...... 77 Synthetic exosomes ...... 86 Acknowledgements ...... 88 Chapter 4 Feedback-based drug delivery pumps for mouse artificial endocrine organs ...... 89 Forward ...... 90 Introduction ...... 90 Implantable remote-control pump...... 91 Reciprocating peristaltic pump with constrictions at fixed locations ...... 93 Reciprocating electrolytic pump ...... 95 Divertable constant-pressure pump ...... 107 Latching SMA-actuated 2-way valve ...... 109 Sliding rod SMA-actuated 3-way latching valve ...... 111 Watch stepper motor-actuated 3-way latching valve ...... 113 Control and feedback system ...... 116 Electronics...... 116 Visual feedback ...... 119 Running wheel feedback ...... 120 Bioacoustic feedback ...... 122 Cage-mounted remote-control pump ...... 124 Conclusion and recommendations ...... 125 Acknowledgements ...... 127 Chapter 5 Conclusions and future directions ...... 128 References ...... 131

List of Figures

Figure 1: Structure of telomeric DNA and the shelterin protein complex ...... 2 Figure 2: Telomeres form a loop ...... 3 Figure 3: Mechanisms of telomere shortening during DNA replication ...... 4 Figure 4: Telomerase structure ...... 5 Figure 5: Telomerase post-translational regulation ...... 7 Figure 6: Benefits of a rapid method to extend telomeres ...... 15 Figure 7: Examples of mammalian post-transcriptional RNA modifications ...... 18 Figure 8: Schematic of approach to telomere extension using nucleoside modified mRNA encoding TERT ...... 27 Figure 9: Transfection efficiency of modified mRNA into human myoblasts ...... 29 Figure 10: Electroporation of GFP mRNA into human leukocytes ...... 30 Figure 11: Amounts of exogenous CI TERT and TERT mRNA taken up by cells are equal, and do not perturb levels of endogenous TERT mRNA...... 31 Figure 12: Absolute quantitation of uptake of CI TERT and TERT mRNA ...... 32 Figure 13: Increases in TERT protein following modified TERT mRNA delivery ...... 34 Figure 14: Specificity of the TERT antibody ...... 36 Figure 15: Dose-dependent increase in TERT protein following TERT mRNA delivery ...... 37 Figure 16: Increases in telomerase activity in multiple human cell types transfected with modified mRNA encoding TERT ...... 38 Figure 17: Dose response of telomerase activity ...... 39 Figure 18: Time course of telomerase activity following TERT mRNA treatment...... 40 Figure 19: Mean telomere lengths in untreated fibroblasts decrease over time in culture...... 41 Figure 20: Increase in telomere length following modified TERT mRNA delivery...... 42 Figure 21: Sensitivity of Q-FISH microscope for measuring telomere lengths to temporal variation in illumination intensity and ambient temperature...... 43 Figure 22: Flat-field correction for the microscope and cameras used for Q-FISH image acquisition ...... 44 Figure 23: Representative Q-FISH images of metaphase spreads of fibroblasts...... 46 Figure 24: Telomeres in metaphase spreads are not within a single depth of field ...... 46 Figure 25: Telomere length analysis of fibroblasts treated with TERT mRNA...... 48 Figure 26: 184-point Q-FISH telomere length quantification...... 50 Figure 27: Telomeres shorten after treatment with TERT mRNA...... 51 Figure 28: Increase in proliferative capacity following modified TERT mRNA delivery...... 52 Figure 29: Growth curve of untreated MRC5 cells...... 53 Figure 30: Proliferation capacity of myoblasts treated with TERT mRNA...... 54 Figure 31: Increased proliferative capacity in cells treated with TERT and TERC ...... 55 Figure 32: Transient reduction of senescence-associated markers following modified TERT mRNA delivery...... 57 Figure 33: Watershedding telomere image to identify maxima for 3D Q-FISH ...... 67 Figure 34: Intramuscular injection of mRNA results in very limited biodistribution ...... 72 Figure 35: Intravenous injection of mRNA with cationic lipid results in delivery to spleen ...... 73 Figure 36: Schematic of our HDL mRNA delivery approach ...... 75 Figure 37: Stability of mRNA ligated to cholesterol ...... 76 Figure 38: Exosome production and fusion ...... 78 x

Figure 39: Schematic of our approach to generating targeted exosomes carrying mRNA ...... 80 Figure 40: Electron micrograph of our targeted exosomes isolated from engineered immature dendritic cells ...... 82 Figure 41: Size distribution of targeted exosomes ...... 82 Figure 42: Loading of GFP mRNA and luciferase-GFP mRNA into exosomes ...... 83 Figure 43: Schematic of design of exosomes engineered for in vivo delivery of mRNA ...... 84 Figure 44: Schematic of the mouse-implantable remote-control pump ...... 92 Figure 45: Pumping cycle of the reciprocating peristaltic pump ...... 93 Figure 46: An example of the single actuator peristaltic micropump actuated by SMA wire..... 95 Figure 47: External view of reciprocating electrolytic pump ...... 96 Figure 48: Two-channel pump with saddle configuration ...... 97 Figure 49: User-friendly pump filling procedure...... 97 Figure 50: An exploded view of the reciprocating electrolytic pump ...... 98 Figure 51: The pumping cycle of the reciprocating electrolytic pump ...... 99 Figure 52: Electrolysis phase of the pumping cycle ...... 101 Figure 53: Recombination phase of the pumping cycle...... 101 Figure 54: The check valve we developed to make our pump small and efficient ...... 102 Figure 55: Formation of the smooth dome of the check valve ...... 103 Figure 56: The process of making the dome check valves ...... 104 Figure 57: Mouse walking a few minutes after awakening from surgery ...... 105 Figure 58: Mouse running approximately one hour after having the pump implanted ...... 105 Figure 59: Testing the pump using bioluminescence imaging ...... 106 Figure 60: Simplified pump to increase reliability and user-friendliness ...... 107 Figure 61: Initial designs for pressurizing the drug reservoir ...... 108 Figure 62: Remote-control latching valves directing flow from an osmotic pump to either a target tissue or a waste reservoir ...... 109 Figure 63: Latching valve actuated by shape memory alloy (SMA) wire ...... 110 Figure 64: Testing the function of the valve in Figure 63 ...... 111 Figure 65: Dual-catheter valve actuated by SMA wire ...... 112 Figure 66: Close-up view of the valve in Figure 65 ...... 112 Figure 67: Three-way valve actuated by a watch stepper motor ...... 113 Figure 68: Close-up view of the conducting channel in the side of the rotor in Figure 67 ...... 114 Figure 69: Side view of a prototype of the rotor valve mechanism ...... 114 Figure 70: Top view showing the housing and rotor in place in a watch stepper motor ...... 115 Figure 71: Schematic of the feedback system for controlling the implantable pumps ...... 116 Figure 72: Circuit diagram for the implantable remote-control pump ...... 118 Figure 73: Video cameras mounted on a cage ...... 119 Figure 74: Detection of mouse location by infrared imaging ...... 120 Figure 75: Running wheel rotation sensor ...... 121 Figure 76: Running wheel activity from six cages ...... 122 Figure 77: Ultrasonic microphones and cages in acoustic isolation chambers ...... 123 Figure 78: Examples of recorded vocalizations ...... 123 Figure 79: Cage-mounted pump with plate mail catheter ...... 125 Figure 80: Mutation sites in our telomere-extending mutants ...... 129 Figure 81: Co-delivery of factors to enhance telomere extension ...... 130

List of Tables

Table 1: Telomerase activity in diverse cell types following delivery of TERT mRNA...... 39 Table 2: Primer sequences ...... 68 Table 3: Pathogen ligands ...... 84 Table 4: Fusogenic ligands ...... 85 Table 5: Endothelial and extravasation ligands ...... 85 Table 6: Targeting ligands ...... 85 Table 7: Initial evaluation of relative merits of implantable vs cage mounted pumps ...... 91 Table 8: Figures of merit for the three four pumps developed here ...... 126

xii

Chapter 1

Introduction to telomere shortening and extension

The first telomere was sequenced in Tetrahymena thermophile 36 years ago (Blackburn and Gall, 1978) and telomerase was discovered soon thereafter (Greider and Blackburn, 1985).

Since then several mechanisms by which telomere shortening causes or exacerbates disease have been elucidated and provide strong motivation to develop safe methods to extend telomeres.

Telomeres

Telomeres comprise DNA sequences at the ends of chromosomes that, in humans, are typically made of thousands of tandem repeats of the hexameric DNA sequence TTAGGG

(Szostak and Blackburn 1982). Telomeres form a nucleoprotein complex with the proteins collectively called shelterin (Figure 1), comprising telomere repeat binding factors (TRF1,

TRF2), TRF-interacting nuclear protein 2 (Tin2), tripeptidyl peptidase 1 (TPP1), protection of telomeres 1 (POT1), and repressor of activator protein 1 (Rap1) (de Lange 2009). The 3’ end of telomeric DNA is single-stranded and typically 50-500 nucleotides (nt) long.

Figure 1: Structure of telomeric DNA and the shelterin protein complex

Shelterin comprises TRF1 and TRF2, lack of which results in non-homologous end-joining.

POT1 binds the single-stranded 3’ overhang of telomeres, preventing access of telomerase, and also preventing damage processing factors from recognizing telomere ends as broken DNA. 2

TPP1 helps recruit telomerase to telomeres. TIN2 connects TRF1 and TRF2 to TPP1 which binds POT1. Telomeres also contain nucleosomes. TEN1, STN1, and CTC1 form the CST complex that binds the 3’ overhang and protects the 5’ end from resection. CST also helps regulate telomerase as described below.

When telomeres are sufficiently long, the 3’ overhang inserts itself into the double stranded region of telomeres forming a loop called the T-loop (Figure 2). The loop structure and shelterin protect telomeres, and thus the ends of chromosomes, from being mistaken for broken

DNA by the cell.

Figure 2: Telomeres form a loop

The single-stranded 3’ overhang at the end of telomeres can invade the double-stranded region to form a loop. TRF2 is important for formation of the T-loop, and lack of TRF2 results in absence of T-loops and occurrence of non-homologous end-joining.

Mechanisms of telomere shortening

Telomeres shorten over time due largely to oxidative damage and incomplete DNA replication (Figure 3).

Figure 3: Mechanisms of telomere shortening during DNA replication Telomeres shorten during DNA replication by two mechanisms that differ between the 3’ and 5’ telomere strands. On the 3’ strand, which is the lagging strand in telomeres, DNA replication is primed by 8-14 nt-long RNA primers to which DNA polymerase adds nucleotides at the 3’ end.

The most distal primer on the 3’ overhang of telomeres binds at a random position that is usually not exactly at the end, but rather on average 60 nt from the end (Chow et al. 2012). Thus there is no RNA primer to initiate duplication of the part of the 3’ overhang distal to the last primer, and so telomeres shorten by the distance from the last primer to the end of the 3’ overhang with each cell cycle. On the 5’ strand, which is the leading strand in telomeres, the initial product is blunt- ended, and then the 5’ end is resected to reform the 3’ overhang after duplication, resulting in net shortening.

Telomeres also shorten under oxidative stress, and conversely free radical scavengers slow telomere shortening (Thomas von Zglinicki 2002). Fibroblasts in growth arrest due to contact inhibition exposed to oxidative stress exhibit rapid telomere shortening once allowed to proliferate again.

Telomerase

The ribonucleoprotein telomerase extends telomeres and comprises two essential components: the protein telomerase reverse transcriptase (TERT) and the RNA telomerase RNA component (TERC or TR) (Greider and Blackburn 1985)(Greider and Blackburn 1989)(Lingner et al. 1997)(Artandi and DePinho 2010). Telomerase extends the 3’ overhang of telomeres using

TERC as a template (Figure 4).

Figure 4: Telomerase structure

Telomerase comprises a protein component, telomerase reverse transcriptase (TERT), and an

RNA component, telomerase RNA component (TERC). TERC contains the sequence

CAAUCCCAAUC, the first five nt of which bind to the telomere 3’ overhang, and the last six nt

5 of which act as a template for addition of the sequence GGTTAG, after which telomerase can either shift by 6 nt and repeat the process or dissociate. For easy nomenclature and memory the telomere sequence is typically expressed as repeats of TTAGGG or T2AG3 rather than the actual sequence added, GGTTAG. TERC has extensive secondary structure resulting in a Y-shape, with two arms of the Y binding TERT, and the stem of the Y binding dyskerin and telomerase

Cajal body protein 1 (TCAB1), which are essential for telomerase complex formation.

Mutations in TERT, TERC, dyskerin, or other proteins involved in telomere maintenance cause disease.

Telomerase forms a dimer, and inactive TERT or TERC prevent telomere extension in a dominant negative fashion. Telomerase usually extends telomeres during and after S phase of the cell cycle (Chow et al. 2012). Telomerase is active in many cell types including stem and progenitor cells. However over the lifetime of an individual telomeres shorten in most tissues

(Takubo et al. 2010)(Sahin and Depinho 2010)(Signer and Morrison 2013).

Telomerase is subject to a large number of regulatory mechanisms including several post- translational mechanisms. Assembly of TERT and TERC is facilitated by TCAB1, POT1, and

TPP1. Post-translational regulation is illustrated in Figure 5.

Figure 5: Telomerase post-translational regulation Telomerase is regulated in a cell cycle-dependent manner in part by translocation to Cajal bodies by Pinx1, and recruitment to telomeres by TPP1. In response to oxidative stress TERT is phosphorylated by Src1 and cAbl and exported from the nucleus by CRM1. POT1 is part of shelterin and blocks access of telomerase to telomeres. TERRA, which comprises RNA transcripts of telomeres, has been suggested to inhibit or promote telomerase activity (Cusanelli,

Romero, and Chartrand 2013).

Telomere shortening and disease

When telomeres become critically they are unable to form the protective T-loop, and the chromosome ends are exposed. The exposed ends can also participate in non-homologous end joining or homology-directed repair, resulting in chromosome-chromosome fusions that can lead to malignancy. The exposed ends can be recognized by DNA damage repair factors that can activate P53 leading to cell cycle arrest or apoptosis, and thus a decreased number of functional cells, and an increased number of senescent cells. Such senescence cells not only consume nutrients without aiding the organism but can produce senescence-associated secretory factors

7 that may have deleterious effects on other cells (Elizabeth H Blackburn 2011)(R. T. Calado et al.

2012)(Elizabeth H Blackburn 2005). p53 activation can also compromise mitochondrial function and biogenesis by reducing expression of PGC1-alpha and beta (Sahin et al. 2011). These mechanisms link telomere shortening to various diseases.

Genetic mutations in telomere maintenance genes

The strongest evidence of the causal role of telomere shortening in disease comes from genetic diseases in which mutation in TERT, TERC, dyskerin, shelterin genes, or other genes associated with telomere maintenance results in accelerated telomere shortening. Genetic mutations of TERT and TERC are linked to fatal inherited diseases of inadequate telomere maintenance including forms of idiopathic pulmonary fibrosis, dyskeratosis congenita, and aplastic anemia (R. T. Calado et al. 2012)(Armanios and Blackburn 2012). Families with these mutations exhibit genetic anticipation, in that in each successive generation with the mutation, the disease occurs earlier and with greater severity. This is due to the fact that telomere length is inherited, and thus successive generations have progressively shorter telomeres.

Muscular dystrophy

We (the Blau lab) showed that short telomeres lead to muscle stem cell (MuSC) replicative exhaustion, and consequent fatal myopathy in DMD (Sacco et al. 2010; Mourkioti et al. 2013). Dr. Blau first demonstrated a potential role of telomere shortening in DMD in her pioneering work showing decreased replicative capacity in DMD myoblasts (Blau, Webster, and

Pavlath 1983), and more recently, developed the first mouse model (mdx/mTR) to faithfully recapitulate the fatal myopathy of DMD, by crossing mice with a dystrophin mutation with mice with rapidly shortening telomeres.

Cognitive decline

Aging humans and mice with short telomeres exhibit reduced myelination and markers of hippocampal neurogenesis (Wijngaarden, Franklin, 2013). Activation of tamoxifen-inducible telomerase in adult short-telomere mice rescues or partially restores these and other deficits including olfactory function (Jaskelioff, DePinho,2010). In humans, short leukocyte telomere lengths correlate with Alzheimer’s disease (Thomas, Fenech, 2008), vascular dementia (200%)

(T von Zglinicki et al. 2000), and risk of post-stroke dementia (Martin-Ruiz, Zglinicki, 2006)

(Stern and Bryan 2008). Age-related cognitive decline may be mediated in part by shortening telomeres in multiple compartments including the nervous system, vascular endothelium, and immune system.

Hypertension and heart disease

Aged humans with short telomeres are approximately 200% more likely to develop myocardial infarction (Fitzpatrick et al. 2007). Hypertension is the major risk factor for cardiac failure, which in turn is the leading cause of death in the USA, and one quarter of the world’s adult population is estimated to be hypertensive (Dhaun et al. 2008). Humans with short telomeres are more likely to become hypertensive and suffer cardiac arrest (Z. Yang et al. 2009), and mice engineered to have short telomeres exhibit hypertension due to excess release of endothelin-1 by senescent endothelial cells (Pérez-Rivero et al. 2006). Similarly a major mechanism underlying hypertension in humans is excess endothelin-1 production (Dhaun et al.

2008). Endothelial cell telomeres are shorter at vascular bifurcations, where endothelial turnover is increased, and are shorter in atherosclerotic lesions than in adjacent tissue (Chang and Harley

1995). One of our collaborating labs (the John Cooke lab) has shown that telomere extension prevents endothelial cell senescence (Matsushita et al. 2001), and thus safe telomere extension in

9 endothelial cells has great potential to treat or prevent hypertension and associated cardiovascular diseases.

Cancer

The effects of premature cellular senescence and apoptosis due to short telomeres in these diseases are devastating in themselves, and may be compounded by increased risk of cancer

(Artandi and DePinho 2010)(Alter et al. 2009). In addition to abundant correlative data linking short telomeres to cancer (Wentzensen et al. 2011), the progression from aplastic anemia to cancer provides some of the first direct evidence that critically short telomeres and resulting chromosomal instability predispose cells to malignant transformation in humans, in that aplastic anemia patient bone marrow cells in culture manifest aneuploidy years before they develop clinically detectable aneuploidy (R. T. Calado et al. 2012). The increased cancer rates in patients with telomeres syndromes are consistent with abundant correlative data linking short telomeres to cancer in humans in general: in a recent analysis of 26 cancer studies, together studying over

11,000 subjects, the mean odds ratio for increased risk of cancer in people with shorter leukocyte telomere lengths was 1.35, with a 95% confidence interval of 1.14 - 1.60 (Ma et al. 2011). The increased risk of cancer may be due in part to chromosomal fusions, and may be compounded by reduced immune surveillance due to immunosenescence resulting from shortening telomeres.

Cancer requires a telomere extension mechanism to support proliferation sufficiently for malignancy, and 85-90% of cancers use telomerase, with the remainder using the alternative lengthening of telomeres (ALT) mechanism in which single-stranded telomeric DNA acts as a primer for DNA polymerase-mediated telomere extension (Cesare and Reddel 2010). Therefore the risk benefit-ratio of telomere extension may be more favorable before cancer starts, at which time it may be preventive by reducing the incidence of chromosomal rearrangements. In

10 contrast, the risk-benefit ratio of extending telomeres, even transiently, may be less favorable after telomeres have become critically short resulting in chromosomal rearrangements, and may support progression to malignancy by providing premalignant cells an opportunity to acquire additional mutations. A workaround may be to extend telomeres in a subset of hematopoietic progenitors that still have long telomeres, to help the immune system fight tumor formation.

Immunosenescence

Immune system cells can be highly subject to telomere shortening and replicative senescence due to high proliferative demand. Even in healthy individuals, on the order of one billion new immune cells are produced per hour (Armanios and Blackburn 2012). Proliferative demand is further increased in the context of chronic stimulation as in chronic viral infection

(Bestilny et al. 2000), autoimmune disease (Hohensinner, Goronzy, and Weyand 2011), graft- versus-host disease (Fukunaga et al. 2007), and cancer (Z.-Z. Yang et al. 2012). Replicative senescence results in reduced immune cell numbers and function, weakening the immune system

(Weng 2012). Telomere shortening in the immune system is linked causally and correlatively to disease. Patients with telomere syndromes develop epithelial and hematological cancers including acute myeloid leukemia and myelodysplastic syndrome at high rates (Armanios and

Blackburn 2012). Short leukocyte telomeres are not only correlated with, but predict, hypertension (Z. Yang et al. 2009), atherosclerosis (Willeit et al. 2010), heart disease (Fitzpatrick et al. 2007), stroke (Martin-Ruiz et al. 2006), and dementia (Martin-Ruiz et al. 2006). The mechanisms underlying these diseases may include chronic inflammation due to altered cytokine production in senescent immune cells, and reduced disposal of waste for example by macrophages. Telomere shortening in immune cells can also lead to another major killer, infection. Even in healthy adults, the common cold introduced experimentally to volunteers

11 more easily infects people with shorter leukocyte telomeres (Cohen et al. 2013). Short telomeres correlate with higher risk of infection in the middle-aged, though not necessarily the very aged

(Fitzpatrick et al. 2007)(Njajou et al. 2009). Pneumonia and influenza are the ninth most common cause of death in the US across all age groups (Panda et al. 2009). In a biracial population-based cohort study, increased leukocyte telomere length correlated with years of healthy life (Njajou et al. 2009). A vicious cycle may arise in which infection and inflammation shorten telomeres, and, in turn, telomere shortening causes immunosenescence which leads to additional telomere shortening, increased infection, and inflammation due to oxidative stress, reduced immune surveillance, and increased senescence associated cytokine production, respectively (Ilmonen, Kotrschal, and Penn 2008). Thus telomere shortening can initiate and propagate a spiral in which disease or aging causes immune cell dysfunction which in turn causes additional disease. Preventing or interrupting this spiral is of considerable clinical interest.

Since people with short telomeres in one tissue are likely to also have short telomeres in most of their other tissues, (Takubo et al. 2010), and short telomeres are implicated in multiple diseases, the cumulative risk over all diseases for individuals with short telomeres is large.

In addition to these major diseases, short telomeres are implicated in conditions associated with age even in the absence of overt disease. For example TERC-null and TERT- null mice bred to generations in which their telomeres become critically short exhibit multiple phenotypes mirroring age-related conditions in humans, including hypertension due to endothelial cell senescence, heart dysfunction, infertility, intestinal atrophy, alopecia and hair greying, and short lifespan (reviewed in (Blasco 2005)).

Limits imposed by telomere shortening on cell and tissue engineering

Short telomeres limit cell amplification for cell therapies and tissue engineering (Mohsin et al. 2012, -)(Bodnar et al. 1998)(Shay and Wright 2001). Short telomeres limit cell therapies using hematopoietic stem cells, cardiac progenitors, and iPSC-derived retinal cells (Zimmermann and Martens 2008)(Mohsin et al. 2013)(Suhr et al. 2009)(Kokkinaki, Sahibzada, and Golestaneh

2011). Myoblasts from teenage DMD patients and stem cells from the DMD mouse model only undergo a few (<10) divisions in culture before entering replicative senescence, in contrast to population doublings typical of myoblasts or stem cells from normal age-matched controls

(Webster and Blau 1990)(Sacco et al. 2010). Short telomeres destabilize stem cell differentiation

(Pucci, Gardano, and Harrington 2013) and, via p53 activation, inhibit reprogramming (Hong et al. 2009)(Marión et al. 2009)(Le et al. 2013). iPSC telomere lengths are heterogeneous and short compared to embryonic stem cells (R. Allsopp 2012)(Vaziri et al. 2010). Further, iPSCs derived from patients with diseases mediated by impaired telomere maintenance exhibit reduced self- renewal and survival (Batista et al. 2011)(Andrade et al. 2012). Thus a method to extend telomeres could facilitate reprogramming (Utikal et al. 2009)(R. Allsopp 2012) and enable amplification and transdifferentiation of autologous cells, even from aged or diseased patients, for cell therapy and tissue engineering applications.

The abundant correlative and causal evidence linking telomere shortening to disease has led many to strongly advocate the development of therapies to address telomere shortening

(Bodnar et al. 1998)(Harley 2005)(Westin et al. 2007)(Blasco 2005)(Bernardes de Jesus and

Blasco 2011)(Hohensinner, Goronzy, and Weyand 2011)(R. Calado and Young 2012)(López-

Otín et al. 2013)(E H Blackburn and Gall 1978; Greider and Blackburn 1985),(Elizabeth H

Blackburn 2011; Elizabeth H Blackburn, Tisty, and Lippman 2010),(Farzaneh-Far et al. 2010).

A safe method for telomere extension might delay, prevent, or ameliorate disease.

Criteria for a therapeutically useful telomere extension method

In order for telomere extension therapy to be useful for prevention or treatment of disease, it must meet several criteria. Not least, its benefits must outweigh its risks, whether for fatal genetic diseases of inadequate telomere maintenance or delaying or preventing cancer, heart disease, and other diseases in normal subjects. We hypothesize that the following features are desirable in a telomere extension method:

 Transient

 High rate of telomere extension

 Specific

 No risk of genomic integration

 Effective even in slow-dividing or quiescent cells

 Low or no immunogenicity (with a caveat mentioned below)

 Deliverable in vivo

Transience is desirable because constitutive telomerase expression can facilitate cancer, as shown in mice (Artandi et al. 2002). We hypothesize that a high rate of telomere extension during a brief, transient treatment is desirable so the treatment can be performed infrequently, so as to allow the normal anti-cancer telomere shortening mechanism to function in the interim

(Figure 6).

Figure 6: Benefits of a rapid method to extend telomeres We hypothesize that if telomere extension can be rapid enough, then between treatments, normal telomerase activity and telomere shortening is present, and therefore the anti-cancer safety mechanism of telomere shortening to prevent out-of-control proliferation will remain intact, while the risk of short telomere-related disease remains low.

Specificity is important for safety. Absence of risk of genomic insertion is important both for preventing insertional mutagenesis, and also for avoiding constitutive telomerase expression that could support oncogenesis.

The ability to extend telomeres even in slowly- or non-dividing cells is important for extending telomeres in some progenitor and stem cell populations.

Immunogenicity is generally not desirable as it can result in vaccination against TERT

(Saebøe-Larssen, Fossberg, and Gaudernack 2002).

Deliverability in vitro might be sufficient for some applications such as immune system rejuvenation, cell therapies and tissue engineering, but an ideal telomere extension therapy will

15 be preventive and minimally invasive, and thus an ideal treatment should be deliverable in vivo easily. We address delivery further in Chapter 3.

Critical evaluation of methods of extending telomeres

We surveyed available methods of extending telomeres and scored them against our criteria for a useful therapy. A summary of our survey and scores is presented in Table 1. A variety of treatments increase telomerase activity or extend telomeres (Table 1).

Table 1: Evaluation of existing approaches to increasing telomerase activity

Type Examples References Growth factors EGF, IGF-1, FGF-2, VEGF (Liu et al. 2010) Genetic Retroviral delivery of DNA encoding TERT (Bodnar et al. 1998) treatments Adeno-associated virus Plasmid encoding TERT Hormones Estrogen (Imanishi, Hano, and Nishio 2005) Erythropoietin (Akiyama et al. 2011) Physical UV radiation (Ueda et al. 1997) treatments Hypoxia (Gladych, Wojtyla, and Rubis 2011) Cytokines IL-2, IL-4, IL-6, IL-7, IL-13, and IL-15 (Akiyama et al. 2002) (Liu et al. 2010) Phytochemicals Resveratrol (Pearce et al. 2008) Compounds extracted from Astragalus (Zvereva, Shcherbakova, membranaceus including cycloastragenol and Dontsova 2010) (TAT2), TA-65, or TAT153 Inibitors of: Menin, SIP1 (Lin and Elledge 2003) pRB, p38 (Di Mitri et al. 2011) p53, p73 (Beitzinger et al. 2006) MKRN1, CHIP, Hsp70 (Lee et al. 2010) androgens (Nicholls et al. 2011) TGF-beta (Prade-Houdellier et al. 2007)

Current methods of extending telomeres include small molecule activators of telomerase which are easy to delivery but may have non-specific effects, and only results in a detectable change in number of cells with short telomeres (not average telomere length), and only in some

16 subjects (Bernardes de Jesus et al. 2011) (Harley et al. 2011)(Eitan et al. 2012). Viral delivery of

TERT, while possibly inducible (Jaskelioff et al. 2011), risks insertional mutagenesis and thus presents serious safety concerns. Treatments that involve continuous telomerase overexpression are potentially unsafe because they may support malignancy by enabling unlimited proliferation(1,13) in a cell with an oncogenic mutation, either due to critically short telomeres and resulting chromosomal instability (O’Sullivan and Karlseder 2010) or to another cause.

Perhaps the most promising of existing approaches are small molecules and adeno-associated virus (AAV)-based delivery of TERT (Bernardes de Jesus et al. 2012), because AAV is largely episomal, meaning that it integrates into the genome less frequently than, for example, lentivirus.

Nonetheless, it does integrate into the genome, in about 1 out of every 2,500 cells, when delivered in vivo to humans (Kaeppel et al. 2013), and this may pose an unacceptable risk especially in normal subjects. To our knowledge no one has yet reported delivery of recombinant TERT protein to cells, possibly because it has proven difficult to produce isolated functional TERT protein. Further, protein transduction domains such as TAT require denaturation of the attached protein, and thus folding post-delivery could also be problematic.

Unmodified mRNA encoding TERT was delivered to cells in 2002, but it resulted in a strong immune response (Saebøe-Larssen, Fossberg, and Gaudernack 2002).

Nucleoside-modified mRNA

Although the dream of delivering mRNA to cells and tissues to transiently increase the amount of the protein encoded by the mRNA has been discussed and attempted since the first in vitro transcription (IVT) reactions of cDNA to produce mRNA were performed, until recently

IVT-produced mRNA has only been used as a means of vaccination against the mRNA-encoded protein (Saebøe-Larssen, Fossberg, and Gaudernack 2002). Delivery of IVT mRNA comprising

17 the four canonical nucleotides U, A, C, and G in vivo results in only very brief (a few hours) expression immediately after injection (K Karikó et al. 2001).

Recently it was discovered that delivery of mRNA comprising modified nucleotides such as pseudouridine () or 5-methylcytosine (m5C) reduces the innate immune response that is activated in mammalian cells in response to in vitro transcribed (IVT) mRNA comprising only the four canonical nucleotides (Katalin Karikó et al. 2005).

In mammalian cells RNA is transcribed using the four canonical nucleotides after which some of the nucleotides in a given RNA strand are modified by enzymes in a partially directed manner. These enzymes include pseudouridine synthase and RNA methyltransferases, that convert U and C to m5C, respectively (Figure 7), as well as others.

Figure 7: Examples of mammalian post-transcriptional RNA modifications

There are hundreds of naturally-occurring nucleoside modifications in mammalian RNA, but  is the most abundant.  occurs in mammalian rRNA and tRNA, in which one of its roles

18 is to stabilize secondary structure. For example  affects the local structure of tRNA modulating binding to the 30S ribosomal subunit (Charette and Gray 2000). In newly transcribed mRNA, U is pseudouridinylated by mechanisms that are at least partly directed, for example by base-pairing of bases adjacent to a given U with snoRNA in the nucleolus thereby disrupting rRNA secondary structure and allowing  synthase access to U. Interestingly  is not found in mammalian mRNA, yet 100% replacement of U with  in mRNA reduces innate immunity to in vitro transcribed mRNA (Anderson et al. 2010).

m5C is found in mammalian tRNA, rRNA, and, unlike pseudouridine, mRNA. Like  m5C also affects the local structure of RNA, for example enabling tRNA to bind magnesium. m5C also reduces innate immune response to IVT mRNA, but substitution with both and m5C together reducing innate immune response more than either alone.

mRNA comprising modified nucleotides is produced in vitro by substituting the modified for the unmodified nucleotide triphosphates in an IVT reaction, in which an RNA polymerase is mixed with nucleotide triphosphates and a DNA template in which the coding sequence is downstream of a promoter recognized by the RNA polymerase and flanked by UTRs that are typically from relatively stable mRNA species (Pardi et al. 2013). The resulting mRNA may then be used directly or purified (Katalin Karikó et al. 2011). To generate the DNA template, the open reading frame (ORF) of the gene of interest is inserted into a plasmid containing the promoter, UTRs with a polyadenylation sequence, and in some cases the actual sequence of poly-T for the poly-A tail.

Unmodified IVT mRNA activates components of the cellular innate immune system that recognize pathogen-associated molecular patterns (PAMPs) resulting in toxicity, lack of stability,

19 and lack of translational efficiency. PAMP pattern recognition receptors (PRRs) mediating IVT unmodified mRNA immunogenicity include:

 Protein kinase R (PKR)

 Toll-like receptor 3 (TLR3)

 TLR7/TLR8

 2’-5’-oligoadenylate synthetase (OAS)

 Retinoic acid-inducible gene I (RIG-I)

PKR and TLR3 recognize double-stranded RNA (dsRNA) that can occur in IVT products due to RNA-dependent RNA transcription due to the high density of RNA products in the IVT mixture (Katalin Karikó et al. 2011), and can occur in single-stranded RNA (ssRNA) due to secondary structure (Anderson et al. 2010).

TLR3 activates TRIF that in turn activate cAMP-dependent binding protein (CREB), NF-

B, and activator protein 1 (AP1) leading to production of pro-inflammatory cytokines.

PKR activated by dsRNA phosphorylates translation initiation factor 2 ( eIF-2) which inhibits further mRNA translation.

Binding of ssRNA to TLR7 and TLR8 activate myeloid differentiation primary-response protein 88 (MyD88), leading to activation of interferon regulatory factors (IRF) and expression of Type I interferons IFN and IFN.

2’-5’-Oligoadenylate Synthetase (OAS) recognizes dsRNA and activates RNase L that cleaves mRNA (Kawai and Akira 2010)(Anderson et al. 2011). Modified mRNA not only does not activate OAS as much as does unmodified mRNA, but is also more resistant to cleavage by

RNase L, which cuts preferentially after UA and UU dinucleotides.

RIG-I recognizes 5’-triphosphate RNA (Hornung et al. 2006), which can occur in IVT products due to incomplete capping. Capping is typically performed by one of two means: incorporation of anti-reverse cap analog (ARCA) into the IVT mixture, or enzymatically. In the former case, the ARCA is incorporated into the majority, but not all, of the transcripts, and thus phosphatase treatment must be performed to remove 5’ phosphates from the remainder.

Nucleoside-modified mRNA was used for the first time as a therapeutic agent in mice in

2011 via muscle injection and aerosol delivery (Kormann et al. 2011b). Since mRNA treatment is transient, it is probably most suitable for treatments in which only a brief burst of activity is needed for a therapeutic response. In Chapter 2 we show, for the first time to our knowledge, that delivery of nucleoside-modified mRNA encoding TERT increases telomerase activity in cells, extends telomeres rapidly, and increases proliferative capacity following a brief treatment.

TERT mRNA satisfies many of the criteria for therapeutic usefulness listed above. In vivo, mRNA can be delivered in vivo to a small subset of tissues; however, as with most nucleic acid therapies, delivery to most tissues is a hurdle, and thus we are pursuing several methods for in vivo delivery of mRNA discussed in Chapter 3, including a novel mRNA delivery vehicle: exosomes.

Exosomes

To be useful in vivo for telomere extension, a delivery vehicle for TERT mRNA is needed, and fortunately it has been recently discovered that in the human body exosomes are natural carriers of mRNA,. Exosomes are lipid vesicles 40-150 nm in diameter present in most human body fluids including blood, urine, lymph, and saliva (van Dommelen et al. 2011). In

2007 exosomes were discovered to be natural carriers of mRNA between cells in humans, including over 1,300 types of mRNA and 121 types of non-coding microRNA (Lakhal and 21

Wood 2011)(Valadi et al. 2007). Exosomes contain membrane proteins such as Lamp2 which can be fused to targeting ligands (Lakhal and Wood 2011), and in 2011 such an approach was used to deliver targeted autologous exosomes carrying siRNA to brain via intravenous injection with therapeutic benefit (Alvarez-Erviti et al. 2011). To our knowledge exosomes have not been used to deliver mRNA. In Chapter 3 we present our progress toward delivering TERT mRNA in vivo using targeted exosomes, and our initial steps synthesizing synthetic exosomes that have the benefits of exosomes, without their endogenous cargoes and targeting decorations.

Summary

The telomere shortening-related diseases provide strong motivation for developing a method to extend telomeres that satisfies the safety criteria listed above. To this end this thesis presents a possible approach combining two novel technologies: TERT mRNA and mRNA delivery using exosomes.

Chapter 2

Telomere extension in cells using TERT mRNA

Forward

This work resulted in US patent application 14/187,265 Ramunas J, Yakubov E, Cooke J,

Blau H filed 2014 titled “Compounds, Compositions, Methods, and Kits Relating to Telomere

Extension”. Part of this chapter is being submitted to PLoS ONE this week (March 10-14, 2014) under the title “Telomere extension following delivery of TERT mRNA”. The authors are John

Ramunas, Eduard Yakubov (co-first author), Jennifer J. Brady, Stéphane Y. Corbel, Colin

Holbrook, Moritz Brandt, Jonathan Stein, Juan G. Santiago, John P. Cooke, and Helen M. Blau.

Introduction

Telomere extension has been proposed as a means to improve cell and tissue engineering and to treat disease. A treatment that elevates telomerase activity only briefly but extends telomeres rapidly during that brief period may be advantageous with respect to safety and convenience. Here we report that delivery of modified mRNA encoding TERT to human fibroblasts and myoblasts increases telomerase activity transiently (24-48 h) and rapidly extends telomeres. Successive transfections over a four-day period extended telomeres up to 0.9 kb in a cell type-specific manner in fibroblasts and myoblasts and conferred an additional 28 ± 1.5 and

3.4 ± 0.4 population doublings, respectively. These effects were not observed in cells treated with mRNA encoding a catalytically inactive form of TERT, indicating that they required telomerase activity. Notably, unlike immortalized cells, all treated cell populations eventually stopped dividing and exhibited senescence markers to the same extent as untreated cells. This rapid and transient method of extending telomeres and increasing cell proliferative capacity without risk of insertional mutagenesis should have broad utility in research and medicine.

Telomeres comprise tandem DNA repeats that, with associated proteins collectively named shelterin, protect chromosome ends from acting as damaged DNA (Szostak and

Blackburn 1982; de Lange 2009). Telomeres shorten over time due, in part, to incomplete chromosomal end replication as well as other factors including oxidative damage (Shay and

Wright 2001; Lansdorp 2005). Telomerase is a ribonucleoprotein that extends telomeres and consists of a protein component, telomerase reverse transcriptase (TERT), which complexes with an RNA component (TERC) (Greider and Blackburn 1985; Greider and Blackburn 1989; Lingner et al. 1997; Artandi and DePinho 2010). Telomerase is active in many cell types including stem and progenitor cells, however, over the lifetime of an individual telomeres shorten in most tissues (Takubo et al. 2010; Sahin and Depinho 2010; Signer and Morrison 2013).

When telomeres become sufficiently short, p53 and DNA damage response pathways are activated, levels of PGC1-alpha and -beta are reduced leading to mitochondrial dysfunction, chromosome-chromosome fusions can occur leading to malignancy, and cells may apoptose or senesce (Elizabeth H Blackburn 2005; Sahin et al. 2011; Elizabeth H Blackburn 2011; R. T.

Calado et al. 2012; Mourkioti et al. 2013). Genetic mutations in TERT and other genes involved in telomere length maintenance result in diseases such as aplastic anemia and dyskeratosis congenita (R. Calado and Young 2012; Armanios and Blackburn 2012). Further, we recently showed that shortened telomeres underlie the progression of Duchenne Muscular Dystrophy

(DMD) (Sacco et al. 2010; Mourkioti et al. 2013), and that telomere extension averts endothelial cell senescence, which is associated with atherosclerosis and hypertension (Matsushita et al.

2001; Pérez-Rivero et al. 2006).

A means to safely extend telomeres would benefit cell and tissue engineering by increasing the number of population doublings and cumulative cell numbers achieved in culture

(Bodnar et al. 1998; Shay and Wright 2001). This need is underscored by reports that short telomeres destabilize stem cell differentiation (Pucci, Gardano, and Harrington 2013) and inhibit

25 reprogramming via p53 activation (Hong et al. 2009; Marión et al. 2009; Le et al. 2013). Short telomeres also limit replicative capacity essential to cell therapies using transplanted hematopoietic stem cells, cardiac progenitors, and iPSC-derived retinal pigment epithelial cells

(Zimmermann and Martens 2008; Suhr et al. 2009; Kokkinaki, Sahibzada, and Golestaneh 2011;

Mohsin et al. 2013). We found that myoblasts (progenitors) from teenage DMD patients and stem cells from the DMD mouse model were limited in that they typically underwent only a few divisions in culture before entering replicative senescence. This is in stark contrast to the extensive population doublings typical of myoblasts or stem cells from normal age-matched controls (Webster and Blau 1990; Sacco et al. 2010). iPSC telomere lengths are short compared to embryonic stem cells (Vaziri et al. 2010; R. Allsopp 2012). Further, iPSCs derived from patients with diseases mediated by impaired telomere maintenance exhibit reduced self-renewal and survival (Batista et al. 2011; Andrade et al. 2012). Moreover, due to a body of literature linking telomere shortening to several genetic and age-related diseases, several investigators have proposed the use of telomere extension as a preventive or therapeutic intervention (Bodnar et al. 1998; Harley 2002; Harley 2005; Blasco 2005; Elizabeth H Blackburn, Tisty, and Lippman

2010; Jaskelioff et al. 2011; Bernardes de Jesus and Blasco 2011; R. Calado and Young 2012;

López-Otín et al. 2013). Thus there is an unmet need for an efficacious and safe way to extend telomeres.

For cell therapy applications, avoiding the risk of cell immortalization is of paramount importance. To this end, transient, rather than constitutive, telomerase activity may be advantageous for safety, especially if the elevated telomerase activity is not only brief but extends telomeres rapidly enough that the treatment does not need to be repeated continuously.

Current methods of extending telomeres include viral delivery of TERT under an inducible

26 promoter, delivery of TERT using vectors based on adenovirus and adeno associated virus, and small molecule activators of telomerase (Bodnar et al. 1998; Weinrich et al. 1997; Jaskelioff et al. 2011; Rothe, Modlich, and Schambach 2013)(Bernardes de Jesus et al. 2011; Harley et al.

2011; Eitan et al. 2012)(Mogford et al. 2006)(Steinert, Shay, and Wright 2000). Here we provide an alternative that offers the benefits of highly transient telomerase activation combined with rapid telomere extension.

Modified nucleoside-containing mRNA is non-integrating and has recently been used by others to transiently elevate levels of diverse proteins encoded by the mRNA (Katalin Karikó et al. 2012; Kormann et al. 2011a; Y. Wang et al. 2013). Here we deliver nucleoside-modified mRNA, in which pseudouridine and 5-methylcytidine replace uridine and cytidine, encoding

TERT to human cells (Figure 8).

Figure 8: Schematic of approach to telomere extension using nucleoside modified mRNA encoding TERT

The TERT mRNA comprises the coding sequence of the full length functional form of TERT or a catalytically-inactive (CI) form of TERT, flanked by untranslated regions (UTRs) of HBB and a 151 nt poly-A tail, synthesized using modified nucleotides pseudouridine and 5- methylcytidine. The TERT mRNA is delivered to cells using standard methods for delivering

27 nucleic acids including as part of complex with a cationic lipid such as TransIT or RNAiMax, or by electroporation. Once inside the cells, the TERT mRNA is translated into functional protein.

The mRNA poses no risk of genomic insertion.

Here we show in two human cell types that delivery of modified mRNA encoding TERT avoids immortalization, yet transiently increases telomerase activity, rapidly extends telomeres, delays expression of senescence markers, and increases proliferative capacity.

Results

Increase in TERT protein levels following modified TERT mRNA transfection. To test the hypothesis that modified TERT mRNA could substantially increase telomere lengths and cell proliferative capacity, we synthesized and delivered modified mRNA comprising pseudouridine and 5-methylcytidine. To increase mRNA stability, the full-length human TERT open reading frame was flanked by the 5’ and 3’ UTRs of human beta-globin (HBB), a 5’ cap and a 151 nt 3’ poly-A tail (Tavernier et al. 2011). The mRNA was transfected via a cationic lipid into primary human fibroblasts and myoblasts (Figure 1A), cells known to have limited proliferative capacity

(Hayflick and Moorhead 1961; Webster and Blau 1990; Yakubov et al. 2010). Transfection efficiency was determined using flow cytometric single cell quantitation of fluorescence following delivery of GFP mRNA, which showed that most cells (>90%) were transfected even at relatively low concentrations of modified mRNA (0.1 µg/ml) (Figure 9).

Figure 9: Transfection efficiency of modified mRNA into human myoblasts

(A) GFP fluorescent of myoblasts (n=2000) transfected with 0.1–0.8 µg/ml of modified mRNA encoding GFP as measured by flow cytometry 24 h after start of treatment. (B) Transfection efficiency expressed as percentage of GFP+ cells. (C) Mean fluorescence of modified GFP mRNA-transfected myoblasts in response to increasing doses.

Although transfection with RNAiMax was effective in fibroblasts, myoblasts, keratinocytes, and endothelial cells, it was less effective in mononuclear cells and CD8+ T-cells, however for these cell types electroporation was effective (Figure 10).

Figure 10: Electroporation of GFP mRNA into human leukocytes

We found that most leukocytes, with the exception of macrophages, were resistant to transfection using cationic lipid. Left: Transfection of leukocytes (mononuclear cells) using electroporation was effective. Right: CD8+ T-cells are subject to great proliferative demand and telomere extension in these cells is therefore a high priority for us. As a first step we validated that CD8+ cells can be transfected with GFP mRNA, and later we measured telomerase activity in the transfected cells (below). Dark dots are CD3/CD28-coated beads used to activate the T-cells for subsequent TERT mRNA treatment.

Treatment of cells with equal concentrations of exogenous TERT mRNA or mRNA encoding a catalytically inactive (CI) form of TERT resulted in internalization of similar amounts of mRNA (Figure 11), as measured by RT-qPCR 24 h after the first treatment.

Figure 11: Amounts of exogenous CI TERT and TERT mRNA taken up by cells are equal, and do not perturb levels of endogenous TERT mRNA

(A) Quantification of exogenous TERT mRNA in fibroblasts 24 h after transfection with 1 µg/ml of TERT or CI TERT mRNA, as measured using RT-qPCR. Ratio of TERT to CI TERT was calculated using the Pfaffl method with RPL37A and GAPDH as reference genes (n=3). (B)

Quantification of endogenous TERT mRNA in fibroblasts 24 h after transfection with 1 µg/ml of

TERT or CI TERT mRNA, as measured using qPCR, calculated as in (D). All data are presented as means ± s.e.m.

We measured the average number of copies of mRNA taken up per fibroblast to be 8-9 (Figure

12).

Figure 12: Absolute quantitation of uptake of CI TERT and TERT mRNA

Top: Schematic of method used for absolute quantitation of mRNA uptake using RT-qPCR. To control for the efficiency of the reverse transcription step, a serial dilution of mRNA was added to cells in the RNA isolation buffer (RLT). This allowed a standard curve (bottom left) to be created relating the amount of mRNA harvested with the cells to the Ct value measured in the qPCR step. As expected the untreated (UT) and vehicle only-treated cells had Ct values higher than even the most dilute of the serial dilution controls, indicating absence of CI TERT or TERT mRNA in the cells. The CI TERT and TERT mRNA-treated cells had Ct values in the linear range of the standard curve, and by dividing by the number of cells per reaction, the average absolute number of copies of mRNA taken up per cell was determined.

CI TERT has a substitution mutation at one of the triad of metal-coordinating aspartates at the catalytic site of the reverse transcriptase domain of TERT. As a result, CI TERT cannot add nucleotides to telomeres, yet remains structurally intact, able to bind template DNA, and exhibits stability comparable to wild type TERT in reticulocyte lysates (Wyatt 2009). Neither

TERT nor CI TERT mRNA treatment affected levels of endogenous TERT mRNA relative to untreated cells as measured by RT-qPCR (Figure 11).

Transfection with 1 µg/ml of either TERT or CI TERT mRNA resulted in equivalent 50% increases (P<0.05 and <0.01, respectively) in the amount of TERT protein in fibroblasts (Figure

13). The presence of endogenous TERT protein in the untreated MRC5 cells is consistent with

TERT protein also observed in Western blots of WI I38 and BJ human fibroblasts (Wick, Zubov, and Hagen 1999; Ahmed et al. 2008)(Masutomi et al. 2003). TERT is subject to extensive post- translational inhibitory regulation that may explain why these cells have little or no telomerase activity (Yi, Shay, and Wright 2001; Cifuentes-Rojas and Shippen 2011).

Figure 13: Increases in TERT protein following modified TERT mRNA delivery Top: Total TERT protein levels were measured by quantitative infrared Western blot. TERT protein expression in fibroblasts harvested 24 h after start of treatment with 1 µg/ml TERT mRNA was measured by multiplexed infrared Western blot. The serial dilution of total protein

34 was used to generate a standard curve to compare relative amounts of TERT protein to controls.

The specificity of the TERT antibody used here is demonstrated below. The size of the band observed here, slightly larger than 120 kDa, is consistent with the estimated weight of human

TERT of 127 kDa (Wick, Zubov, and Hagen 1999). Note that like the MRC5 cells measured here, other human fibroblast cells with little or no endogenous telomerase activity such as WI I38 and BJ fibroblasts also exhibit TERT protein on Western blots (Masutomi et al. 2003). Because of the low background fluorescence at the wavelengths employed by the infrared imaging system used to acquire this image, the apparent amount of TERT protein and the signal-to-noise ratio are greater than can be obtained with a traditional Western blot, but note that in the serial dilution of total loaded protein in the center of the gel, the TERT signal becomes almost imperceptible while the tubulin signal remains strong, indicating a relatively low TERT abundance even in the treated cells. Bottom: Quantification of TERT protein levels in the Western blot.

We tested the specificity of the TERT antibody used herein (Abcam AB32020) using a human fibroblast cell line that uses the ALT mechanism to extend telomeres and has very little

TERT (Wu et al. 2006) (Figure 14).

Figure 14: Specificity of the TERT antibody

We used the Abcam AB32020 anti-TERT antibody after testing three different antibodies. Left:

293T cells served as positive controls. TERT was detected as a band at approximately 120 kDa.

Right: The GM847 human fibroblast line served as a negative control as it employs the alternative lengthening of telomeres (ALT) mechanism and has very little TERT protein.

Treatment with either 0.2 or 1.0 ug/ml of TERT mRNA results in detectable band of TERT in

GM847 cells at approximately 120 kDa.

Treatment with increasing amounts of TERT mRNA resulted in a dose-dependent increase of TERT protein expression as measured in single cell assays by flow cytometry (Figure

15).

Figure 15: Dose-dependent increase in TERT protein following TERT mRNA delivery Quantification of TERT protein in response to various doses of mRNA was measured at the single cell level by flow cytometry (n=10,000) (panel C, right). *P<0.05, **P<0.01 compared to untreated cells. Error bars represent s.e.m.

Telomerase activity is transiently increased. To test whether modified TERT mRNA delivery resulted in the generation of functional TERT protein, telomerase activity was quantified using a gel-based TRAP assay. Telomerase activity was detected in fibroblasts and myoblasts at all doses tested (0.25, 0.5, 1.0, and 2.0 µg/ml), and was not detected in untreated cells or cells treated with either vehicle only or modified mRNA encoding CI TERT, even at the highest dose of 2.0 µg/ml. Transfected endothelial cells, CD8+ T-cells, and keratinocytes, but not untransfected controls of each type, also exhibited telomerase activity (Figure 16).

Figure 16: Increases in telomerase activity in multiple human cell types transfected with modified mRNA encoding TERT Top row: Detection of telomerase activity in various cell types transfected with 1 µg/ml modified TERT mRNA, as measured using the telomere repeat amplification protocol (TRAP).

Arrow indicates internal controls for PCR efficiency. The cells were transfected using

RNAiMax except for the CD8+ T-cells which were electroporated. Bottom: Delivery of TERT mRNA using electroporation results in increased telomerase activity in human keratinocytes.

The TSR8 and 293T lanes are positive controls for the quantitative PCR reaction and the extension and PCR reactions combined, respectively. The CHAPS buffer-only lanes is a negative control.

The results of TERT and CI TERT transfections of various cell types are summarized in

Table 1.

Table 1: Telomerase activity in diverse cell types following delivery of TERT mRNA.

Human cell type Untreated Vehicle-only CI TERT TERT MRC5 fetal lung fibroblasts - - - + Myoblasts - - - + Microvascular endothelial cells - - - + Keratinocytes - - - +

Telomerase activity increased in fibroblasts transfected with TERT mRNA in a dose-dependent manner (Figure 17).

Figure 17: Dose response of telomerase activity

Fibroblasts electroporated with increasing concentrations of TERT mRNA (0, 10, 40, or 80 ug/ml) exhibited increasing telomerase activity up to 40 ug/ml. CHAPS served as a negative control and 293T cells as positive control.

A time course revealed that telomerase activity peaked at 24 hours and returned to baseline levels within 48 hours after a single transfection (Figure 18). For comparison, the half- life of human telomerase is typically approximately 1 day, but can be shorter or longer depending on cell type and conditions (Holt et al. 1997b). TERT is subject to multiple modes of post-translational regulation, including by targeted degradation and interaction with factors such as TERRA that affect its catalytic activity (Cifuentes-Rojas and Shippen 2011). The half-life of human TERT mRNA is 2-4 h; and that of human beta-globin 17-18 h (our exogenous TERT mRNA is flanked by beta-globin 5’ and 3’ UTRs) (Kabnick and Housman 1988; Holt et al.

1997a; Xu et al. 1999).

Figure 18: Time course of telomerase activity following TERT mRNA treatment.

Fibroblasts transfected with TERT mRNA complexed with cationic lipid (RNAiMax) exhibited a peak in telomerase activity 24 hours after treatment. The bottom band is an internal PCR

40 control. The band above that in the 72 h lane is due to primer dimers which occur sporadically in the TRAP assay.

Lengthening of telomeres. Telomere lengths in untreated fibroblasts declined over time

(3 months) as expected (Sitte, Saretzki, and von Zglinicki 1998) (Figure 19) and was quantified using two different methods. We used the monochrome multiplex qPCR method (MMqPCR) to assess length, and validated our measurements independently with a qPCR method performed by

SpectraCell Laboratories, Inc. (correlation coefficient 0.97, P<0.001).

Figure 19: Mean telomere lengths in untreated fibroblasts decrease over time in culture.

Population doubling number starts at zero after receipt of cells from the supplier, ATCC, who describe them as being at passage number 14 after isolation from human fetal lung. Total telomere content was measured by MMqPCR and by SpectraCell (correlation coefficient 0.97,

P<0.001).

Delivery of TERT mRNA three times in succession at 48-hour intervals to fibroblasts or myoblasts starting at population doubling (PD) 25 and 6, respectively, extended telomeres by 0.9

± 0.1 kb (22 ± 3%), and 0.7 ± 0.1 kb (12 ± 2%), respectively (Figure 20). Treatment with vehicle only or CI TERT mRNA had no significant effect on telomere length relative to untreated cells.

The average rate of telomere extension in fibroblasts was 135 ± 15 bp/PD.

Figure 20: Increase in telomere length following modified TERT mRNA delivery.

Mean telomere lengths increased in fibroblasts (left) and myoblasts (right) transfected with 1

µg/ml TERT mRNA three times in succession at 48 h intervals compared to vehicle only-treated cells. MMqPCR was performed by us and qPCR was performed by SpectraCell. MMqPCR T/S ratios were converted to absolute telomere lengths using a standard curve generated from MRC5 cells at 6 different population doubling numbers sent to SpectraCell and included in the

MMqPCR panel. **P<0.01, ***P<0.001

We optimized the Q-FISH procedure for measuring telomere lengths in several ways.

First, we controlled for temperature variation that affected the acquired fluorescence signal intensity, probably due to variation in focal position due to temperature variation (Figure 21).

Figure 21: Sensitivity of Q-FISH microscope for measuring telomere lengths to temporal variation in illumination intensity and ambient temperature.

The fluorescent intensity of a fluorescent plastic block measured at 5 min. intervals without or with photosensor-based correction for illumination intensity (blue and red lines, respectively) exhibited a perturbation lasting several minutes after the room door was opened briefly (<5 s).

We are conducting future Q-FISH sessions using a microscope in a dedicated insulated enclosure.

Next, we performed flat-field correction using the DeltaVision microscope system (Figure 22).

Figure 22: Flat-field correction for the microscope and cameras used for Q-FISH image acquisition

Left: The fluorescence intensity of a fluorescent plastic block was acquired, and the resulting image scaled to 2% of full range. To compensate for the variation the DeltaVision acquires images with a range of illumination intensities and stores the gain for each pixel. Right: Image representing the dark current and read error of the camera, scaled to 2% of full range. The

DeltaVision system also records and compensates for this variation for each pixel.

These corrections resulted in good quality images of metaphase spreads (Figure 23).

Figure 23: Representative Q-FISH images of metaphase spreads of fibroblasts.

Individual images of single metaphase spreads of untreated (top) and TERT mRNA-treated

(bottom) fibroblast were assembled into a collage. Each image represents one image from the stack of images at different focal depths acquired of each metaphase spread. Yellow is the signal from AlexaFluor 555 telomere probes, and blue is DAPI.

However, we observed that the telomeres in metaphase spreads were never all in the same focal plane in a given cell (Figure 24).

Figure 24: Telomeres in metaphase spreads are not within a single depth of field

Usually Q-FISH images are acquired using a 60X or 100X objective, and typically a single image of each metaphase spread is acquired at the focal position at which the user determines the most telomeres to be in best focus. However, this leaves many telomeres out of focus. For example, the two images above were acquired at different focal planes in the same metaphase

46 spreads. The difference in the intensities of the spots in the images acquired at different focal planes indicates that the telomeres cannot all be imaged at best focus in a single image, and indeed some telomeres may be entirely excluded from subsequent analysis due to being too dim due to poor focus.

Therefore we acquired stacks of images of each metaphase spread and wrote an ImageJ plug-in to allow us to perform 3D Q-FISH. The software identifies, for each telomere, the image in which that telomere is in best focus, and quantifies the fluorescence intensity of the telomere in that image. We used this software to measure telomere lengths in TERT mRNA treated fibroblasts (Figure 25).

Figure 25: Telomere length analysis of fibroblasts treated with TERT mRNA.

Left: Representative images of metaphase spreads in which yellow is the signal from

AlexaFluor 555 telomere probes, and blue is DAPI. Stacks of images of each metaphase were obtained and analyzed using custom software. Right: Telomere lengths for each population are shown as frequency distributions of telomere signal intensities. Median telomere signal intensities are indicated above each distribution and as vertical dashed blue lines. The arrow

48 indicates one of the two peaks in the bimodal distribution indicating a set of chromosome ends with very short telomeres. This peak is smaller in fibroblasts treated three times with TERT mRNA (bottom), suggesting either a diluting out of cells with very short telomeres in the 10-14 days between the initiation of treatment and the harvesting of cells for Q-FISH analysis, extension of these telomeres, or both.

Because our custom 3D Q-FISH software allows us to image every telomere in best focus, we observe almost all 184 telomeres in the treated cells, and most of the telomeres in the untreated cells. This allowed a novel representation of the change in telomere lengths by sorting the 184 telomeres from each cell from brightest to dimmest, and then for each ordinal position in the sorted list from 1 to 184, average the brightness of the cells in a given population (e.g. treated or untreated). The result is shown in (Figure 26). This representation makes clear which telomeres, the shorter or longer, are being extended. In this case, it is both, with a slight preference for shorter telomeres.

Figure 26: 184-point Q-FISH telomere length quantification.

By acquiring a stack of images of each metaphase spread so that every telomere is acquired in focus, and using custom software to quantify the fluorescence intensity of each telomere in the image in which it is in best focus, we are able to quantify the intensities of all or almost all 184 telomeres of each metaphase spread. In this representation we sort the intensities of the telomeres of each cell from brightest to dimmest, and then for each population we average the resulting curves to obtain the above graph. The benefit of such a graph is that it provides a clear indication of which range of telomeres, the shortest or longest, for example, differ most between populations.

After the treatment, we cultured the fibroblasts, and found that their telomeres shortened again over time (Figure 27).

Figure 27: Telomeres shorten after treatment with TERT mRNA.

Telomeres in fibroblasts treated three times at 48-h intervals starting at PD 40 exhibited shortening as measured using MMqPCR at the indicated population doublings. The population stopped expanding at PD 80.

Cell type-dependent increases in proliferative capacity. To test the effect of modified

TERT mRNA delivery and consequent telomere extension on cell proliferative capacity, we transfected human fibroblasts either once, twice, or three times in succession. Treatments were delivered at 48-hour intervals. Untreated, vehicle only-treated, and CI TERT mRNA-treated fibroblasts exhibited an equivalent plateau in cell number after approximately 50-60 PD, whereas cells treated three times with TERT mRNA continued to proliferate for a finite additional 28 ±

1.5 PD with an overall increase in cell number of 2.7x108 beyond untreated cells (Figure 28).

The effect was dose-dependent with each additional treatment conferring additional PD (Figure

28 inset). The incremental increase in proliferative capacity was greater with the first treatment than with the second or third treatments. Human myoblasts treated three times in succession every 48 hours gained 3.4 ± 0.4 PD, equivalent to a 10-fold increase in cell number compared to

51 untreated or vehicle treated controls (Figure 30). Such differences in PD between myoblasts and fibroblasts are not unexpected, as prior studies found similar limited effects of TERT overexpression to a few PD and showed that this limitation was due to a p16-mediated growth arrest in human myoblasts, in contrast to fibroblasts (Zhu et al. 2007; Bodnar et al. 1998).

Figure 28: Increase in proliferative capacity following modified TERT mRNA delivery.

Growth curves of fibroblasts transfected with 1 µg/ml TERT mRNA, CI TERT mRNA, or vehicle only, once, twice, or three times in succession at 48 h intervals, with blue arrows indicating treatments times. A fourth cohort was treated three times at 48 h intervals and then a fourth time nine weeks later. Growth curves were repeated twice with each population cultured in triplicate.

Inset: Replicative capacity increased in a dose-dependent manner. *P<0.05, **P<0.01 compared to vehicle only-treated cells.

The complete growth curve for the untreated cells in Figure 28 is shown below in Figure 29.

Figure 29: Growth curve of untreated MRC5 cells. MRC5 cells were received from ATCC at what they describe as passage 14. We began counting population doublings from that point. Each point represents one passage.

Figure 30: Proliferation capacity of myoblasts treated with TERT mRNA. Proliferation capacity of myoblasts treated as in Figure 28 (green arrows). Growth curves were repeated twice, with each population cultured in triplicate. All data are presented as means ± s.e.m.

Delivery of TERT mRNA and TERC RNA together three times resulted in a greater increase in proliferative capacity (33 ± 1.3 PD) than delivery of TERT mRNA alone (28 ± 1.5)

(P<0.05) (Figure 31), consistent with previous reports of rapid telomere extension using viral delivery of TERT and TERC (Bachand, Kukolj, and Autexier 2000)(Cristofari and Lingner

2006). In both fibroblasts and myoblasts, vehicle only or CI TERT mRNA had no effect on proliferative capacity compared to untreated controls. These data show that delivery of modified

TERT mRNA is an effective method for increasing PD in culture. Importantly, all of the treated cells studied exhibited a significant increase in cell numbers, but eventually reached a plateau in their growth curves, suggesting absence of immortalization.

Interestingly, the telomerase-associated protein dyskerin (Figure 4) also functions as a pseudouridine synthase which is not thought to play a role in telomerase (Artandi and DePinho

2010). However since we delivered TERC in which  entirely replaces U and the treatment

54 increases proliferative capacity effectively, it prompts the questions of whether TERC is heavily pseudouridinylated naturally, and whether this affects its effectiveness in extending telomeres.

Figure 31: Increased proliferative capacity in cells treated with TERT and TERC Left: Treatment with TERT and TERC together in a 1:5 molar ratio increased proliferative capacity of cells by a greater amount than delivery of TERT alone. Right: Quantification of the increased proliferative capacity observed in growth curves of fibroblasts treated with CI TERT or TERT mRNA.

Transient reduction in markers of senescence. As the fibroblast populations stopped growing they exhibited markers of senescence including senescence-associated beta- galactosidase (beta-gal) staining and enlarged size (Figure 32A-C) (V J Cristofalo and

Kritchevsky 1969; Dimri et al. 1995; Vincent J Cristofalo et al. 2004; Lawless et al. 2010).

These changes were transiently reduced in fibroblasts treated with TERT mRNA relative to untreated cells and cells receiving CI TERT mRNA or vehicle only. In accordance with findings by others, not all cells in the populations that had entered a growth plateau expressed beta-gal at

55 detectable levels (Binet et al. 2009; Lawless et al. 2010). However, TERT mRNA-transfected fibroblasts and myoblasts expressed beta-gal to the same degree as the control cells of each type after the two populations reached a growth plateau. These data suggest that cells treated with

TERT mRNA eventually and predictably cease division and express markers of senescence, and are therefore unlikely to be transformed.

Figure 32: Transient reduction of senescence-associated markers following modified

TERT mRNA delivery.

(A) Quantification of beta-gal-expressing fibroblasts after modified TERT mRNA transfection three times in succession at 48 h intervals (green arrows). The control cells, comprising untreated, vehicle only, and CI TERT mRNA-treated populations, stopped expanding at PD 53, and the TERT mRNA-treated population stopped expanding at PD 80. Each experiment was conducted twice with >50 cells per sample scored manually. Representative images show ß-gal- stained TERT mRNA-treated fibroblasts at PD 53 (top) and PD 80 (bottom). Scale bar length is

200 microns. (B) Quantification of beta-gal expression in myoblasts treated as in (A). Controls are as in (A). The control and TERT mRNA-treated populations stopped expanding at PD 8 and

PD 11, respectively. Each experiment was conducted twice, with >50 cells per sample scored manually. Representative images show myoblasts at PD 2 (top) and TERT mRNA-treated myoblasts at PD 11 (bottom). (C) Quantification of enlarged cells associated with replicative senescence in fibroblasts transfected three times with modified TERT mRNA. Population plateaus are as in (A). Controls are vehicle only and CI TERT mRNA-treated. Each experiment was conducted twice, with >50 cells per sample scored manually. Representative images show untreated fibroblasts at PD 2 (top) and PD 53 (bottom). All data are presented as means ± s.e.m.

Scale bar length is 200 microns.

Discussion

Here, we report that transient delivery of TERT mRNA comprising modified nucleotides extends human telomeres and increases cell proliferative capacity without immortalizing the cells. The rate of telomere extension in fibroblasts observed here of 135 ± 15 bp/PD is comparable to rates reported using viral methods, from 94 to >150 bp/PD (Bodnar et al. 1998;

Vaziri and Benchimol 1998). Modified TERT mRNA extended telomeres in fibroblasts in a few days by 0.9 ± 0.1 kb. Fibroblast telomere lengths have been reported to shorten over a human

58 lifetime by approximately 1-2 kb on average (R. C. Allsopp et al. 1992). Thus, modified TERT mRNA is efficacious, yet transient and non-integrating, overcoming major limitations of constitutively expressed viral TERT mRNA delivery.

Human cells of greatest interest are often limited in number, including stem cells for use in experimentation or regenerative medicine. This problem is currently being addressed by various methods including somatic nuclear transfer, viral methods for gene delivery, and the use of culture conditions that lessen the rate of telomere shortening (Zimmermann and Martens

2008; Mohsin et al. 2013; Le et al. 2013). The modified TERT mRNA treatment described here provides an advantageous complement or alternative to these methods that is brief, extends telomeres rapidly, and does not risk insertional mutagenesis. The brevity of TERT mRNA treatment is particularly attractive in that it can avert the loss of stem cell phenotype that can occur over time in culture (Gilbert et al. 2010) and shorten the post-reprogramming stage of iPSC generation during which telomeres extend (F. Wang et al. 2012). Such a method of extending telomeres has the potential to increase the utility of diverse cell types for modeling diseases, screening for ameliorative drugs, and use in cell therapies.

A spectrum of effects on proliferative capacity was observed for the cell types tested, in agreement with previous studies demonstrating different effects of TERT overexpression on myoblast and fibroblast proliferative capacity (Bodnar et al. 1998; Zhu et al. 2007). Moreover, the amount of telomere extension did not correlate with proliferative capacity. Thus, cell context determines the efficacy of TERT expression on proliferative capacity and an understanding of the factors mediating this effect is of interest in overcoming this limitation. Factors that have been implicated in limiting myoblast proliferative capacity upon viral TERT overexpression include p16-mediated growth arrest, cell type and strain, and culture conditions (Zhu et al. 2007). More

59 generally, the effect may be mediated by non-telomeric DNA damage, age, and mitochondrial integrity (Sahin et al. 2011; Mourkioti et al. 2013; López-Otín et al. 2013). The absence of an increase in telomere length or cell proliferative capacity in CI TERT mRNA-transfected cells is consistent with the treatment acting through the catalytic site of TERT by which nucleotides are added directly to telomeres. TERT mRNA-treated cell populations increased in number exponentially for a period of time and then eventually ceased expanding and exhibited markers of senescence to a similar degree as untreated populations, consistent with the absence of immortalization.

Although the therapeutic potential of modified TERT mRNA delivery remains to be determined, the transient non-integrating nature of modified mRNA and finite increase in proliferative capacity observed here are likely to render it safer than currently used viral or DNA vectors. Further, the method extends telomeres rapidly so that the treatment can be brief, after which the protective telomere shortening mechanism remains intact. This method could be used ex vivo to treat cell types that mediate certain conditions and diseases, such as hematopoietic stem cells or progenitors in cases of immunosenescence or bone marrow failure. Although delivery of modified mRNA to certain tissues in vivo has been achieved, it remains a challenge for most tissues (Kormann et al. 2011a). In summary, the method described here for transient elevation of telomerase activity and rapid extension of telomeres, which leads to delayed senescence and increased cell proliferative capacity without immortalizing human cells, constitutes an advance that will enable biological research and medicine.

Materials and Methods

mRNA template generation and synthesis. To generate modified mRNA encoding GFP,

TERT, and CI TERT, their respective open reading frames (ORFs) were inserted into the MCS of

60 a sta -globin (HBB), the MCS, the 3’ UTR of HBB, a 151 bp poly-A sequence, and a restriction site for linearization with a class

II enzyme following the poly-A sequence. The resulting intermediate plasmids were sequenced, linearized, and transcribed using the buffer and RNA polymerase from the MEGAscript T7 Kit

(Ambion, Austin, Texas, USA), and a custom mix of canonical and non-canonical nucleotides

(TriLink BioTechnologies, San Diego, CA, USA) in which the final nucleotide concentrations per 40 µl IVT reaction were 7.5mM for each of adenosine-5’-triphosphate (ATP), 5- methylcytidine-5′-triphosphate (m5C), and pseudouridine-5′-triphosphate (Ψ), 1.5 mM for guanosine-5’-triphosphate (GTP), and 6mM for the cap analog (ARCA) (New England Biolabs,

Ipswitch, MA, USA), or a molar ratio of ATP:m5C:Ψ:GTP:ARCA of 1:1:1:0.2:0.8. In some cases the IVT products were treated with Antarctic Phosphatase (New England Biolabs). The size and integrity of the mRNA products were verified using denaturing agarose gel electrophoresis. The wild type human TERT ORF used to generate the DNA templates for mRNA synthesis is identical to the NCBI human TERT transcript variant 1 (reference sequence

NM_198253.2). The ORF was generated from the pBABE-neo-hTERT plasmid(Counter et al.

1998) (plasmid 1774, Addgene, Cambridge, MA, USA). The pBABE-neo-hTERT plasmid had a non-silent mutation at residue 516 in the QFP motif of TERT, a motif associated with multimerization and TERT interaction with TERC RNA, and thus to avoid the possibility of artifacts due to this mutation we made the sequence identical to the NCBI reference sequence by correcting the mutation with the change G516D. The CI TERT mutant was generated from the

TERT sequence by introducing the mutation D712A.

Cell culture and treatment. Human primary fetal lung MRC5 fibroblasts were obtained from ATCC (Manassas, VA, USA) at passage 14. ATCC does not indicate the PD number,

61 thus, our PD values cited herein refer to the number of PD after receipt of cells from ATCC.

MRC5 cells were cultured in DMEM with 20% FBS and penicillin-streptomycin. Human 30 year-old primary skeletal muscle myoblasts (Lonza, Allendale, NJ, USA) were cultured in

SkGM-2 media (Lonza) according to the vendor’s instructions. Population doublings were calculated as the base 2 log of the ratio between cells harvested and cells plated at the previous passaging, and were considered to be zero if fewer cells were harvested than plated. Cells were transfected with modified TERT mRNA using Lipofectamine RNAiMax (Life Technologies,

Grand Island, NY, USA) prepared in OptiMEM Reduced Serum Media (Life Technologies,

Grand Island, NY, USA) and added to the cells in a 1:5 v:v ratio with their normal media to achieve the final concentrations indicated herein.

Telomerase activity measurement. Twenty-four hours after the start of the transfection period, cells were harvested and lysed in CHAPS buffer. The TRAP assay was performed using a modified version of the TRAPeze kit (EMD Millipore, Billerica, MA, USA), in which the primers and polymerase were added after, rather than before, the step during which the artificial telomere substrate is extended. The PCR program was 94°C 30s/59°C 30s/72°C 45s for 30 cycles, and the products were run on a 15% polyacrylamide gel in 0.5X TBE stained with SYBR

Gold Nucleic Acid Gel Stain (Life Technologies, Grand Island, NY, USA). The time course of telomerase activity was performed using the TRAPeze RT kit (EMD Millipore, Billerica, MA,

USA).

Western blot. Protein was harvested by washing cells once with PBS and then lysing cells in RIPA buffer (Cell Signaling Technology, Danvers MA, USA). Protein was run on

NuPAGE Novex Tris-Acetate Gels (Life Technologies, Grand Island, NY, USA), transferred to

PVDF membrane for 2 h at 35V, then hybridized to anti-α tubulin (Sigma, St. Louis, MO, USA)

62 at 1:10,000 and anti-TERT antibody (ABCAM, Cambridge, MA, USA, 32020 at 1:1000; or

Rockland Immunochemicals, Gilbertsville, PA, USA, 600-401-252S at 1:500) and incubated overnight at 4oC. Detection was performed using infrared (680 nm and 800 nm) antibodies (LI-

COR, Lincoln, NE, USA) and the Odyssey imager (LI-COR). Total intensity of each band was quantified using ImageJ (NIH, Bethesda, MD, USA). The intensity of each TERT band was normalized by its corresponding α tubulin band.

Flow cytometry. Cells were harvested 24 h after transfection with the indicated doses

(Supplemental Figure S1A-C) of TERT mRNA and stained with anti-TERT antibody (Rockland

Immunochemicals, Gilbertsville, PA, USA; 600-401-252S) at 1:500.

Telomere length measurement by SpectraCell Laboratories, Inc. Genomic DNA was extracted using phenol chloroform and quantified using the Quant-iT™ PicoGreen® dsDNA

Assay Kit (Life Technologies, Grand Island, NY, USA). Telomere length analysis was performed at SpectraCell Laboratories Inc. (Houston, TX, USA) using a CLIA approved, high throughput qPCR assay, essentially as described by Cawthon et al. (Cawthon 2002; Cawthon

2009). The assay determines a relative telomere length by measuring the factor by which the sample differs from a reference DNA sample in its ratio of telomere repeat copy number to singe gene (36B4) copy number. This ratio (T/S ratio) is thought to be proportional to the average telomere length. All samples were run in at least duplicate with at least one negative control and two positive controls of two different known telomere lengths (high and low) and an average variance of up to 8% was seen. The results were reported as a telomere score equivalent to the average telomere length in kb.

Telomere length measurement by MMqPCR. Telomere length was measured using a modified version of the MMqPCR protocol developed by Cawthon (Cawthon 2009) with the

63 following changes: Additional PCR pre-amplification cycles were added to make the telomere product amplify earlier, widening the gap between telomere and single-copy gene signals; a mixture of two Taq polymerases was experimentally determined to result in better PCR reaction efficiencies than each on its own; reducing the SYBR Green concentration from 0.75X to 0.5X resulted in earlier signal. Genomic DNA was harvested from cells using the PureGene kit

(Qiagen Germantown, MD, USA) with RNase digestion, quantified using a NanoDrop 2000

(ThermoFisher Scientific, Waltham, MA, USA), and 10-40 ng was used per 15 μl qPCR reaction performed in quadruplicate using a LightCycler 480 PCR System (Roche, Basel, Switzerland).

A serial dilution of reference DNA spanning five points from 100 ng/μl to 1.23 ng/μl was included in each assay to generate a standard curve required for sample DNA quantification.

The final concentrations of reagents in each 15 μl PCR reaction were: 20 mM Tris-HCl pH 8.4,

50 mM KCl, 3 mM MgCl2, 0.2 mM each dNTP, 1 mM DTT, 1 M betaine (Affymetrix, Santa

Clara, CA, USA), 0.5X SYBR Green I (Life Technologies, Grand Island, NY, USA), 0.1875U

Platinum Taq (Life Technologies, Grand Island, NY, USA), 0.0625X Titanium Taq (Clontech), and 900 nM each primer (telg, telc, hbgu, and hbgd primer sequences specified in (Cawthon

2009)). The thermal cycling program was: 2 minutes at 95°C; followed by 6 cycles of 15s at

95°C, 15s at 49°C; followed by 40 cycles of 15s at 95°C, 10s at 62°C, 15s at 72°C with signal acquisition, 15s at 84°C, and 10s at 88°C with signal acquisition. The Roche LightCycler 480 software was used to generate standard curves and calculate the DNA concentrations of telomere and single-copy genes for each sample. T/S ratios were calculated for each sample replicate, and the result averaged to yield the sample T/S ratio which was calibrated using blinded replicate samples of reference cells sent to SpectraCell as described above. The independently obtained

64 relative values of T/S ratios measured using MMqPCR and by SpectraCell for the same samples were highly consistent (correlation coefficient = 0.97, P<0.001).

Reverse transcription qPCR. Primers were designed using Primer3 (Untergasser et al.

2012) and are listed in Supplemental Table 1 except where otherwise noted. Twenty-four hours after start of treatment, cells were washed three times with PBS before harvesting in Buffer RLT

(Qiagen, Germantown, MD, USA). RNA was converted to cDNA using High Capacity RNA-to- cDNA Master Mix (Life Technologies, Grand Island, NY, USA). Endogenous TERT mRNA was amplified using a forward primer in the open reading frame of TERT and a reverse primer in the 3’ UTR of endogenous TERT mRNA. Exogenous TERT mRNA was amplified using a forward primer in the open reading frame of TERT mRNA and a reverse primer in the 3’ UTR of

HBB present in our exogenous TERT and CI TERT mRNA, but not in endogenous TERT mRNA.

Relative levels were calculated using the Pfaffl method. Reference genes were RPL37A (using primers specified in (Greber et al. 2011)) and GAPDH, neither of which exhibited a significant change in Ct value in control or treated cells.

Telomere length measurement by Q-FISH

Q-FISH staining was performed on metaphase spreads prepared by exposing cells to colcemid for 2-6 hours, swelling them in 75 mM KCl for 30 min. at room temperature, gradually adding Carnoy’s fixative (3:1 methanol:acetic acid), centrifuging and resuspending in fixative dropwise three times, and dropping on slides which were then stained with AlexaFluor 555 telomere PNA probe and AlexaFluor 647 Centromere probe. Cells were mounted in SlowFade

Gold (Life Technologies) in DAPI. 3D image stacks of cells were acquired at 200 nm intervals on a DeltaVision photosensor-compensated microscope. Telomere intensities were quantified using 3D Q-FISH using custom software.

Custom software for 3D Q-FISH

An ImageJ plug-in was written in Java to quantify telomere intensities in 3D stacks of images of metaphase spreads and cells in interphase. The software first identifies the region containing the cell of interest by thresholding the DNA in the DAPI channel and allowing the user to define the boundary surrounding the DNA of the cell of interest. The software then identifies, for each telomere and centromere, the image plane in which that telomere or centromere is in best focus, by first performing a watershed segmentation step on the z-projected data (Figure 33), and then for each maximum in the segmented image, identifying the image in the 3D image stack in which that maximum occurred, and quantifying the total intensity of that maximum in that image. The total intensity of each maximum is quantified by accumulating the sum of intensities of the pixels encountered while proceeding down the “mountain” from the maximum by the smallest intensity steps possible among those pixels available immediately adjacent to the contiguous mass of pixels that have been traversed so far in the descent, starting from the maximum and ending when the noise threshold, defined as two standard deviations above the background level, is reached. During the descent, the traversal path never proceeds uphill to higher intensity pixels except when the uphill step is smaller than the intensity variability due to noise as measured previously in a flat-field image. The software then exports the intensities of all of the telomere and centromere maxima to a comma separated variable file for analysis in a spreadsheet program. A custom Excel macro was written to extract telomere lengths from the exported data.

Figure 33: Watershedding telomere image to identify maxima for 3D Q-FISH An early step in quantification of telomere signal intensity in 3D stacks is to identify the maxima in the z-projected maximum intensity image. Each maximum is then used to identify the image in which it occurred, and the spot intensity in that image is then quantified. In this way each telomere intensities are determined in focus.

Senescence-associated beta-galactosidase staining and cell size scoring. Beta-gal staining was performed using the Senescence beta-Galactosidase Staining Kit (Cell Signaling

Technology, Danvers MA, USA). At least 50 cells per population were scored in duplicate. Cell diameter was scored manually after trypsinization on a hemocytometer grid (V J Cristofalo and

Kritchevsky 1969). 67

Statistics. Student’s T-tests and Pearson correlation coefficient calculations were performed using Microsoft Excel. Error bars represent the mean ± s.e.m.

Table 2: Primer sequences

Target Forward primer (5'-3') Reverse primer (5'-3') Product length (bp) Exogenous hTERT (NM_198253.2) 3’ UTR of HBB 162 TERT GTCACCTACGTGCCACTCCT AGCAAGAAAGCGAGCCAAT Endogenous hTERT (NM_198253.2) 3’ UTR of hTERT 74 TERT GCCCTCAGACTTCAAGACCA (NM_198253.2) GCTGCTGGTGTCTGCTCTC GAPDH CAATGACCCCTTCATTGACC TTGATTTTGGAGGGATCTCG 159

Acknowledgements

The co-authors on the manuscript resulting from this work are John Ramunas, Eduard

Yakubov, Jennifer J. Brady, Stéphane Y. Corbel, Colin Holbrook, Moritz Brandt, Jonathan Stein,

Juan G. Santiago, John P. Cooke, and Helen M. Blau. We thank SpectraCell, Inc. for measuring telomere lengths using qPCR. We appreciate the technical assistance of Luis Batista, David

Burns, Benjamin Cosgrove, Andrew Ho, Foteini Mourkioti Stefan Oliver, and Ermelinda

Porpiglia. We thank Robert Weinberg for depositing the pBABE-neo-hTERT plasmid at

Addgene. This work was supported by Stanford Bio-X grant IIP5-31 to J.G.S. and H.M.B., NIH

Director’s Transformative Research Award R01AR063963 from the National Institute of

Arthritis and Musculoskeletal and Skin Diseases to J.P.C. and H.M.B., National Heart, Lung, and

Blood Institute grant U01HL100397 to H.M.B. and J.P.C, a grant from the National Institute on

Aging AG044815-01 to H.M.B. and J.P.C., a University of Maryland, Baltimore subaward

SR00002307 (on NHLBI grant U01HL099997) to H.M.B. and J.G.S., and funding from the

Baxter Foundation to H.M.B.

Chapter 3

In vivo delivery of mRNA

Introduction

We wish to deliver TERT mRNA in vivo to help prevent, delay, or treat the many diseases discussed in the introduction. For some tissues and cells in vivo, such as lung and macrophages, in vivo delivery of RNA is not a challenge, and we are pursuing these. However, these are the exception, and for most tissues safe delivery of RNA in vivo remains a challenge that we are taking on here.

As a first step we surveyed current RNA delivery methods and, thanks to widespread interest in delivery of siRNA, several RNA delivery methods have been developed, including some in clinical trials (reviewed in (Davidson and McCray 2011)). Here we are adapting some of these methods to delivery of mRNA, as opposed to siRNA. We identified a set of criteria desirable in an mRNA delivery method, used this list to identify the best candidates, have tested several of these, and continue to develop two in particular as the most promising: exosomes and synthetic exosomes.

Criteria for selection of delivery methods

We seek a delivery vehicle which:

 Is non- or hypo-immunogenic  Protects the mRNA contents during transport in blood  Delivers mRNA to the cytoplasm, rather than to the endosome-lysosome degradative pathway; or, alternatively, is able to deliver its cargo from the endosome-lysosome pathway to the cytoplasm  Can deliver mRNA to specific cell types or globally  Can be delivered intravenously and extravasate from the vasculature

Survey of existing RNA delivery methods

Current methods for delivering siRNA include:

 HDL  LDL  Albumin  Cationic lipid (e.g. DOTAP)  Naked RNA  Polymer nanoparticles  Liposomes  Hydrodynamic delivery  Exosomes

LDL does not result in as widespread delivery as HDL nor target endothelial cells, possibly important for addressing hypertension and vascular dementia (Kuwahara et al. 2011).

Hydrodynamic delivery is restricted to vasculature in which pressure can be raised without unacceptable risks. The other methods met most of the criteria listed above, and so we decided to pursue them, and one other not yet in use anywhere to our knowledge – synthetic exosomes.

Results

Cationic lipids

Cationic lipids have been used to deliver mRNA to lung in an aerosol formulation, and to muscle by direct injection, and we are pursuing these methods (Kormann et al. 2011c). We tested intramuscular and intravenous delivery of mRNA complexed with two commercial cationic lipid vehicles, RNAiMax and TransIT.

Figure 34: Intramuscular injection of mRNA results in very limited biodistribution

The tibialis anterior (TA) muscles of C57BL6/J mice were injected with 10 ul of OptiMEM media with 1 ug of nGFP mRNA complexed with RNAiMax and imaged whole or in cross section freshly harvested, 24 hours later on an inverted fluorescent microscope. Top: Control

TA showed minimal fluorescence, though individual fibers were discernable by autofluorescence differences. Middle: Cross sections of injected TA showed GFP fluorescence in a punctate pattern (nuclei) consistent with nuclear expression, limited to a few muscle fibers of the needle path. Bottom: Longitudinal view confirmed the limited distribution seen in cross section, and punctate distribution (arrows).

We tested delivery of GFP mRNA by cationic lipid via i.v. injection and found that the luciferase mRNA localized almost exclusively to the injection site (tail vein) and the spleen.

Figure 35: Intravenous injection of mRNA with cationic lipid results in delivery to spleen

NOD-SCID mice were injected with 100 ul of OptiMEM media containing 2 ug of luciferase mRNA complexed with TransIT cationic lipid, and imaged on an IVIS bioluminescence imager

24 hours later. Organs were then harvested and submerged in PBS supplemented with luciferin and imaged. Signal was only detectable at the tail injection site and in the spleen.

Naked mRNA

Naked mRNA injection into muscle or tumors results in expression of the mRNA, but delivery of mRNA complexed with cationic lipid results in greater expression of the encoded protein than delivery of naked mRNA (K Karikó et al. 2001). Consistent with these previous findings, when we injected 2 ug of naked luciferase mRNA in 100 ul of PBS via tail vein we did not observe higher bioluminescence in the injected than control mouse.

High-density lipoprotein

RNA is found naturally associated with various components of blood including low- and high-density lipoproteins (LDL, HDL) and exosomes, recently discovered to naturally transport mRNA in mammalian blood. HDL, but not LDL, -complexed siRNA is taken up by brain microvasculature endothelial cells (Kuwahara et al. 2011), and we are interested in extending telomeres in endothelial cells to help avert hypertension and heart disease as discussed earlier.

HDL takes up cholesterol, and therefore we undertook to ligate luciferase mRNA to cholesterol and then complex it with commercial human HDL. HDL delivery of siRNA results in wider biodistribution of siRNA than does LDL (Wolfrum et al. 2007), and thus we chose to focus on

HDL.

Figure 36: Schematic of our HDL mRNA delivery approach

We first attempted to ligate the cholesterol to a small piece of RNA or DNA, and then ligate the

RNA-cholesterol or DNA-cholesterol to mRNA, since the conditions for RNA-RNA ligation or

RNA-DNA ligation are less harsh than for ligation to cholesterol.

Figure 37: Stability of mRNA ligated to cholesterol

We tested the integrity of the luciferase mRNA after ligation to cholesterol and found that it was being degraded. Further, exposure to the HDL mixture resulted in degradation of mRNA. Gel image indicating intact luciferase mRNA (bright bands) before ligation to cholesterol, but not after, and degradation of mRNA by the HDL mixture, and adding RNase inhibitor to the HDL mixture.

Because our first attempts resulted in degradation of the RNA we have considered purifying the initial cholesterol-RNA or cholesterol-DNA hybrid after ligation before ligating it

76 to the mRNA. However, we are concerned that mRNA, even of a protein like luciferase, is quite large (approximately 10 nm long) relative to HDL (10 nm in diameter) which also already typically contains lipids and cholesterol. Indeed the largest RNA species found naturally in HDL are only about 200 bp long. Therefore we currently favor another natural carrier of RNA, exosomes. Exosomes are about an order of magnitude larger in diameter than HDL.

Albumin

siRNA complexed with albumin is widely distributed following intravenous injection

(Wolfrum et al. 2007). As with HDL, siRNA complexes more efficiently with albumin when bound to a lipophilic moiety such as cholesterol. Therefore once we have successfully produced mRNA-cholesterol we will try complexing and delivering it with albumin as well as HDL.

Fortunately RNase-free albumin is readily commercially available.

Exosomes

Exosomes are lipid vesicles 40-150 nm in diameter present in most human body fluids including blood, urine, lymph, and saliva (van Dommelen et al. 2011). Most cells are capable of producing exosomes. Exosomes are natural carriers of RNA between cells in humans, including over 1,300 types of mRNA and 121 types of non-coding microRNA (Lakhal and Wood

2011)(Valadi et al. 2007). mRNA carried by exosomes can be translated in the destination cells as evidenced by presence translation of mouse proteins in human mast cells exposed to mouse exosomes (Valadi et al. 2007).

Advantages of exosomes as mRNA delivery vehicles

Exosomes protect their RNA cargoes during transport in the blood and other body fluids.

Unlike synthetic nanoparticles, which are typically endocytosed and processed in lysosomes, exosomes fuse with the cell membrane and thereby release their contents directly into the cytoplasm of the destination cell, an ideal route for an mRNA cargo that needs to be translated to be effective. This fusion is mediated by the interaction of tetraspanins such as CD9 on the surface of exosomes that interact with cell surface proteins.

Figure 38: Exosome production and fusion

Exosomes are produced by invagination of early endosomes, producing multivesicular bodies, taking up cytosolic contents including mRNA in a process that is not random, as evidenced by enrichment for specific mRNA species in exosomes. The MVB then fuse with the producing cell membrane releasing the exosomes into the extracellular space. Exosomes are hypothesized to be able to traverse cells. Exosomes fuse with destination cells in a process mediated by exosome surface proteins including tetraspanins. The contents of the exosome are thereby released directly into the cytoplasm of the destination cell. mRNA carried by exosomes is translated by destination cells.

Thus exosomes have several key advantages over other approaches to nucleic acid delivery in general, and to the only previously demonstrated method of mmRNA delivery in vivo

(cationic lipids). Specifically, exosomes:

 Can be generated from a patient’s own cells, making them non-immunogenic – they are

not attacked by antibodies, complement, coagulation factors, or opsonins

 Can be loaded with nucleic acids by electroporation or by expressing the mRNA in the

exosome-producing cells

 Are naturally-occurring vehicles that carry mRNA and microRNA between human cells

 Protect RNA during transport

 Can extravasate from the blood stream to extravascular tissues, even crossing the blood-

brain barrier

 Can be targeted to specific cell types or globally. Exosomes contain specific membrane

proteins that can be fused to targeting ligands and this approach has been used to deliver

siRNA to brain via intravenous injection (Alvarez-Erviti et al. 2011)(Lakhal and Wood

2011). Exosomes can also be targeted to vascular cells, important for our hypertension

work (Al-Nedawi, Szemraj, and Cierniewski 2005).

To our knowledge exosomes have not been used to deliver mRNA. Our approach to mRNA delivery using exosomes is illustrated in Figure 39.

Figure 39: Schematic of our approach to generating targeted exosomes carrying mRNA

We have engineered immature dendritic cells to express targeting ligands fused to exosomes surface protein Lamp2B, and to express GFP mRNA and luciferase-GFP mRNA, which we have detected in the secreted exosomes. We are also testing electroporation of mRNA into exosomes and determining which approach, electroporation or cell-based exosome loading, results in the highest concentration of cargo mRNA in the exosomes. The targeted exosomes carrying GFP mRNA or luciferase-GFP mRNA will be tested on cells and in mice using bioluminescence imaging.

To date we have generated immature dendritic cells (DC2.4 cells) in which the exosomal surface protein Lamp2 is fused to targeting ligands including the RGD peptide,(Hölig et al.

2004) and the TAT targeting peptide, as well as four cell-type specific ligands. The cells have also been engineered to express cargo mRNAs, either GFP or GFP-luciferase, as surrogates for

TERT, which we plan to eventually transduce into the exosome-producing cells once we have identified targeting ligands that result in exosomal delivery. We are testing the ability of a consensus sequence identified in mRNA species enriched in exosomes to increase mRNA loading efficiency.

We have isolated exosomes from our engineered cells, and confirmed their identity and quality by electron microscopy and nanoparticle tracking analysis, and their GFP mRNA content by RT-qPCR.

Figure 40: Electron micrograph of our targeted exosomes isolated from engineered immature dendritic cells

The exosomes exhibit the characteristic enclosed shape and are intact, unlike broken exosomes which would be expected to exhibit a cup morphology.

Figure 41: Size distribution of targeted exosomes

Exosomes from the engineered dendritic cells were analyzed by nanoparticle tracking analysis and have the expected size distribution of 50-150 nm.

Figure 42: Loading of GFP mRNA and luciferase-GFP mRNA into exosomes

As expected exosomes from wild type dendritic cells did not contain GFP mRNA. Exosomes from GFP-expressing cells contained approximately 37 times more GFP mRNA sequence than those from luciferase-GFP-expressing cells, perhaps due to the smaller size of GFP mRNA.

The next step is to test the ability of the targeted exosomes to deliver GFP mRNA to cells in culture. We will then inject the targeted exosomes carrying luciferase-GFP mRNA into tail veins of mice, and perform bioluminescence imaging and histology to map biodistribution.

Targeted exosomes that result in biodistribution beyond liver will be analyzed using histology to identify a set of exosomes capable of delivering mRNA to specific cell types. Immunogenicity will be quantified using ELISA of cytokines including interferon-alpha in blood samples acquired when mice are sacrificed for histology.

We are generating a library of exosomal targeting ligands for targeting various tissues via i.v. injection. Such a library should have similar usefulness for delivering mRNA to diverse

83 tissues. To reach and fuse with extravascular target cells following i.v. injection, exosomes need to:

1. extravasate from the vasculature;

2. bind to specific cells; and,

3. fuse

Therefore we are adorning exosomes with ligands that are intended to, either alone or in combination, accomplish these three tasks.

Figure 43: Schematic of design of exosomes engineered for in vivo delivery of mRNA

We are engineering exosomes to have multiple functions (extravasation, cell-type specific binding, and fusion) by fusing an exosomal surface protein to ligands that are associated with these functions in other contexts.

We surveyed the literature for ligands that are associated with the desired functions in other contexts. Our current candidate ligands are listed in the following tables.

Table 3: Pathogen ligands ID Ligand Reference P1 Herpes simplex virus 1 targeting ligand (Friedman et al. 1981)

P2 AAV9, AAV7 (Friedman et al. 1981) P3 Measles virus targeting ligand (Friedman et al. 1981) P4 Parainfluenza virus type 3 targeting ligand (Friedman et al. 1981) P5 DEN-2 rEgp (Glycoprotein of Dengue virus 2) (Wei et al. 2003) P6 Dengue virus targeting ligand (Dalrymple and Mackow 2011) P7 HPAI H5N1 and other influenza viruses (Zeng et al. 2012) P8 RVG (Alvarez-Erviti et al. 2011)

Table 4: Fusogenic ligands ID Ligand Reference F1 Influenza HA-2 (Martin and Rice 2007) F2 Tat (48-60) (Martin and Rice 2007) F3 Transportan (Martin and Rice 2007) F4 Penatratin (Martin and Rice 2007) F5 Myomaker (Millay et al. 2013) F6 Syntaxin F7 Transferrin receptor (Tauro et al. 2012) F8 CD44 (Tauro et al. 2012) F9 VAMP2, VAMP3 (Tauro et al. 2012)

Table 5: Endothelial and extravasation ligands ID Ligand Reference E1 PSGL-1 E2 ESL-1 E3 L-selectin E4 LFA-1 E5 Mac-1 E6 Pecam E7 Sialyl-Lewis E8 Galectin (Thiemann and Baum 2011) E9 VLA-4 E10 Ezrin, Moesin, and Radixin (Hegmans, 2004) E11 Annexins I, II, V, VI (A1, A2, A5, A6) (Perretti, 1999), (Mears, 2004), (Hegmans, 2004).

Table 6: Targeting ligands ID Ligand Reference T1 Histidine-rich (Martin and Rice 2007) T2 Integrin binding (Martin and Rice 2007) T3 RGD (Martin and Rice 2007) T4 Endothelial (Nicklin et al. 2000)

T5 LOX-1 (Martin and Rice 2007), (White et al. 2001) T6 Cardiac (Zahid et al. 2010) T7 Cardiac/skeletal (Samoylova and Smith 1999) T8 Endothelial BBB - GLA (van Rooy et al. 2010) T9 Endothelial BBB - GYR (van Rooy et al. 2010) T10 Endothelial skin (Rajotte et al. 1998) T11 Endothelial lung (Rajotte et al. 1998) T12 Endothelial intestine (Rajotte et al. 1998) T13 Tspan8 (Nazarenko et al. 2010) T14 CD106 (Nazarenko et al. 2010) T15 CD49d (Nazarenko et al. 2010) T16 CD81 (Rana and Zöller 2011) T17 CD9 (Rana and Zöller 2011) T18 Integrin beta 4 subunit (Hood and Wickline 2012) T19 Fragment of antibody to CD105 (Vandendriessche and Chuah 2013)

Exosomes have two disadvantages in that they already have a cargo and targeting system, although we are using immature dendritic cells to minimize this problem. Therefore we are also pursuing synthetic exosomes.

Synthetic exosomes

We seek to develop delivery vehicles that have the advantages of exosomes described above without necessarily having their natural cargoes or natural targeting. To this end we have made the first steps toward construction and testing of synthetic exosomes. In this work we are aided by the BioADD lab at Stanford which has expertise in the construction of nanoparticle drug delivery vehicles.

Nanoparticle vehicles for delivery of nucleic acids are generally limited by several factors

(Lakhal and Wood 2011)(van den Boorn et al. 2011). First, they often elicit an immune response via attack by antibodies, complement, coagulation factors, or opsonins. Second, the nucleic acids they carry are often exposed and therefore subject to degradation by endogenous nucleases.

Third, they often accumulate preferentially in certain organs including liver, lung, spleen, and vascular endothelium. Fourth, they are often phagocytosed.

To overcome these limitations we are adorning nanoparticles with exosomal proteins.

We are starting construction of our synthetic exosomes using anionic liposomes, which are less likely to complex with albumin in the blood by virtue of their negative external charge. Further, the liposomes are extruded through a membrane with 100 nm pores to make them the size of exosomes. Particles of this size are smaller than the size range typically phagocytosed by macrophages. The liposomes are also stabilized against adsorption in vivo by incorporation of polyethylene glycol (PEG). Finally, the liposomes are targeted using targeting ligands.

The anionic liposomes are synthesized by BioADD by first complexing the mRNA with a cationic polymer polyethylenimine (PEI) forming a polyplex with a slight positive charge, allowing the mRNA-PEI polyplex to complex with anionic lipids. The polyplex is combined with phosphatidylcholine (POPC) which is neutral and phosphatidylglycerol (POPG) which is negatively charged, as well as a lipid covalently bound to PEG. The complex is extruded through a membrane and the lipids self-assemble into a liposome in some cases resulting in the polyplex being on the inside. Anionic liposomes have worked for delivery of DNA in vivo (Ko,

Bhattacharya, and Bickel 2009).

So far we have tested the first batch of anionic liposomes carrying GFP mRNA on cells, specifically cardiomyocytes, in which no fluorescence was observed. The next step is to measure the quantity and quality of mRNA in the liposomes using a Bioanalyzer and adjust the liposome synthesis process accordingly. If we determine that the mRNA is being degraded during liposome synthesis, options include speeding up the process, controlling the pH with buffers, and adding RNase inhibitors.

The novelty in our approach consists first of using mRNA (instead of DNA) in the anionic liposomes, and more significantly, in future iterations we will confer desirable exosome qualities on the anionic liposomes by incorporating exosome components including tetraspanins, the exosome proteins that mediate release of exosome contents into the cytoplasm, to avoid endosomal uptake and lysosomal degradation.

Acknowledgements

We would like to acknowledge Jayakumar Rajadas and Wenchao Sun, both of the

BioADD lab at Stanford University, for designing and synthesizing the liposomes. We would like to thank Moritz Brandt for applying the BioADD nanoparticles to cardiomyocytes and quantifying fluorescence.

Chapter 4

Feedback-based drug delivery pumps for mouse artificial endocrine organs

Forward

The work in this chapter is a first step toward an approach to rejuvenation complementary to telomere extension. In the process of performing this work, three events happened that inspired me to shift my focus entirely to telomere extension. First, the experience of implanting pumps in mice instilled a strong aversion to invasive therapies, and emphasized the need to develop therapies to prevent aging and the need for such interventions. Second, although no remote-control mouse-implantable pump existed when we began this project, two years into the project two companies, Ithetis and Alzet, announced remote-control and programmable mouse- implantable pumps, respectively, thus reducing the importance of the technical side of this project, which comprised the bulk of our progress up to that point. Third, the telomere extension project began to bear fruit. Therefore we decided to focus our time and resources on the telomere extension project. Here we present our progress on the pump project up to that point.

This work emphasizes work done by the author. Part of this work is reported in (Shkolnikov,

Ramunas, and Santiago 2010; Hsu et al. 2011; Strickland et al. 2011).

Introduction

Since hormones in young mammals are in some situations released at times that are related to internal and external events such as exercise (e.g. growth hormone) or social interaction (e.g. oxytocin) (Strathearn et al. 2009), we pursued feedback-controlled pumps with the goal of testing the hypothesis that old mice could be rejuvenated with respect to certain parameters by delivering hormones at the times they are naturally released in young mice relative to events such as running or social interaction. We pursued this technology in anticipation that the understanding of these systems will continue to grow, and in the hope that this pump technology might help in testing various hypotheses relating to aging, behavior, and the

90 neuroendocrine system. In the broader context a remote-control feedback-controlled mouse pump could also have application in a wide range of fields including preclinical pharmacology, regenerative medicine, and bioimaging.

We considered two general approaches to feedback-controlled drug release: implantable pumps and a cage-mounted pump compatible with standard mouse housing.

Table 7: Initial evaluation of relative merits of implantable vs cage mounted pumps Approach Pros Cons Implantable • Autonomous movement • Invasive remote-control • Group housing • Experimental artifacts due to pump • Compatible with standard stress of pump implantation housing • Must be very small to fit in mice • Battery limits experiment duration • Drug stability at 37oC

Cage-mounted • Unlimited drug volume • Only one mouse with pump per pump • Minimally-invasive (catheter cage (due to tangling) and anchor) • Risk of infection at entry site • Compatible with standard housing

Based on this initial evaluation both approaches seemed worth pursuing as they have advantages in different situations: implanted pumps when group housing of implanted animals is required, and cage-mounted pumps when minimal invasiveness is essential. Thus we decided to pursue both mouse-implantable and cage-mounted remote-control pumps.

Implantable remote-control pump

Our initial concept for the remote-controlled mouse-implantable pump is illustrated below.

Figure 44: Schematic of the mouse-implantable remote-control pump

The smallest commercial drug delivery pump as of the start of the project, the Insulet

OmniPod insulin pump, was too large for mouse implantation. Juan Santiago is author on what was the most cited review of micropumps, which was published three years before we started the project (Laser and Santiago 2004), and there did not exist a pump that satisfied our size (1x2x2 cm), voltage (3V), pressure (1 psi), and, critically, reliability requirements. The most promising of the existing pumps were electroosmotic pumps, however the generation of hydrogen and oxygen gas due to electrolysis requires making such pumps larger to accommodate mechanisms developed by the Santiago lab to help recombine or remove the gases independently of pump orientation (Yao et al. 2003), making reliability a trade-off for size, a parameter which the mouse body severely constrains. Although pumps in use by humans have high reliability, the smallest insulin pump at the start of the project was the Insulet OmniPod which was 41x61x18 mm and weighed 34 g, which is about the size of a mouse, and even now the smallest insulin pump,

Debiotech’s JewelPUMP 2 which has been announced, but not released as of March 2014, is too large for a mouse. Other commercially available micropumps such as the Bartels Microtecknik

92 pumps are too large or use piezo actuation which emits sound in the ultrasonic range that mice can hear.

To minimize the size of the implanted pump we developed three pumping mechanisms, the first two of which are novel, and the third of which uses a novel valve mechanism:

1. Reciprocating peristaltic pump with constrictions at fixed locations

2. Reciprocating electrolytic pump

3. Divertable constant-pressure pump

Reciprocating peristaltic pump with constrictions at fixed locations

We arrived at this design by setting out to create the simplest possible pump. The resulting pump, the reciprocating peristaltic pump (U.S. Patent 8,382,460) achieves its simplicity in part by imparting two functions to a single component (component 5 in Figure 45): valving and fluid compression.

Figure 45: Pumping cycle of the reciprocating peristaltic pump A manually-actuated version (not for mouse implantation) illustrates the pump function. The pumping cycle involves a compression step during which members 2 and 5 compress the tube

93 and sealing it off. Member 5 then compresses the flexible tube in a different region, expelling the fluid through the compression at point 4. Member 5 is then allowed to spring back to its relaxed position and the elasticity of the tube causes it to expand, refilling it with fluid, ready for the next cycle.

For implantation we changed the shape of the pump to a circle so that we could wrap a shape memory alloy (SMA) wire around it. SMA wire contracts when heated, so by wrapping it around a circle, it exerts even pressure all the way around, and by incorporating a hinge, we allow the circle to compress. This version of the pump, which has the same basic mechanism as the one in Figure 45 is shown in Figure 46.

Figure 46: An example of the single actuator peristaltic micropump actuated by SMA wire This version is actuated by shape memory alloy (SMA) wire, which contracts when heated. The

SMA wire is wrapped (twice) around the outside of the pump. The round shape reduces friction on the SMA wire, increasing energy efficiency and maximum pump pressure. When the SMA wire contracts, the primary lever moves down, closing the temporary constriction with one end of the constrictor-and-compressor. When the temporary constriction is closed, it starts to act as the fulcrum for the constrictor-and-compressor lever, and for the rest of the stroke of the actuator, the compression section of the tube is compressed as the constrictor-and-compressor lever rotates quickly about this fulcrum point. The advantage of this geometry is that with a very short range of travel of the actuator, the temporary constriction closes first and the compression section gets compressed second. This pump is about 1/2” in diameter.

The key disadvantage of the reciprocating peristaltic pump is its high power consumption, severely limiting battery life in an implanted pump. Thus we pursued other mechanisms.

Reciprocating electrolytic pump

Electrolytic pumps apply a voltage to water which dissociates into hydrogen and oxygen gas. Electrolytic pumps usually have the disadvantage that hydrogen leaks through easily flexible membranes so the pump pressure is not stable over time, and can even become negative.

Solutions to this problem such as metal bellows are typically bulky and difficult to interface to other pump components, especially in a micropump.

Therefore we invented the reciprocating electrolytic pump, which avoids or minimizes the problem of leaking hydrogen by recombining it with the evolved oxygen at the

95 end of each pumping cycle. We added a catalyst, such as palladium or platinum on hydrophobic microparticles that maintain contact of the catalyst with the gas, to accelerate the recombination.

Figure 47: External view of reciprocating electrolytic pump The pump is 5 mm deep by 12 mm wide by 18 mm long, weighs only 2.3 g (<10% of an average mouse) including batteries, delivers 2 +/- 0.2 microliters of drug solution per pulse (which can be adjusted), can pump over 500 pulses on one set of batteries, and has two channels, allowing independent control of timing and dosage of two different drugs. The gold component is the battery contact. The rectangle to the left is the infrared light emitter. The tubes from the left come from two reservoirs, and the tubes on the right are the catheters. The component at top right is the antenna, which wraps around the pump housing.

Figure 48: Two-channel pump with saddle configuration Here a pump capable of pumping two different drug solutions is connected to two reservoirs of

230 ul each (stacked on top of each other in the left oval) by a soft silicone saddle which rests across the back of the mouse under the skin. The reservoir has a rigid oval frame and flexible front and back surfaces which collapse as the drug solution empties, preventing negative pressure.

Figure 49: User-friendly pump filling procedure

A syringe with a blunt needle is inserted into a larger hole in the frame of our reservoir, displacing air bubbles as it fills the lumen. When the reservoir is full, a tube from the pump with the same diameter as the filling hole is used to fluidly connect the reservoir to the pump.

Figure 50: An exploded view of the reciprocating electrolytic pump On the right from bottom to top: batteries, radio frequency circuit and microcontroller, electrolysis chambers and check valves for each of two independently controlled pumps, flexible membrane, and pumping chamber. The reservoir components are at left.

Figure 51: The pumping cycle of the reciprocating electrolytic pump In this schematic, the reservoir is connected to the tube on the left, and the tube on the right is the catheter which carries the pumped drug solution to the target tissue in the mouse. The pump

99 consists of a pumping chamber and an electrolysis chamber separated by a flexible membrane.

When a voltage is applied to electrodes in the electrolysis chamber, hydrogen and oxygen gas are produced by the electrolysis of water, increasing the pressure and causing the flexible membrane to move towards the pumping chamber, displacing drug solution out of the pumping chamber and into the mouse through a check valve (the check valve prevents back flow). When the voltage is turned off, the hydrogen and oxygen gas recombine to reform water, causing the pressure to drop, drawing drug solution into the pumping chamber from the reservoir through a second check valve. A catalyst in the electrolysis chamber reduces the recombination time to a few minutes, instead of days. Our reciprocating pump has the significant advantage over conventional electrolytic pumps that it avoids hydrogen leakage by rapidly recombining the hydrogen gas, thereby increasing the reproducibility of the pumped volume. Further, by limiting the range of travel of the elastic membrane with a surface, we can achieve the same pumped volume with every pump cycle. These properties significantly simplify control of our pump. By angling the electrodes so that the space between them is tapered, and only placing catalyst outside the tapered space, we limit recombination until the bubbles grow large enough to be pushed out of the tapered space, thereby temporally separating the electrolysis and recombination phases of our pump cycle, increasing the efficiency of our pump.

100

Figure 52: Electrolysis phase of the pumping cycle

Electrolysis of water producing hydrogen and oxygen gas in the electrolysis chamber of an early version of our pump, before we started using a tapered space between the electrodes to temporally separate the electrolysis and recombination phases of the pumping cycle.

Figure 53: Recombination phase of the pumping cycle Traditional electrolytic pumps suffer from the ability of hydrogen gas to permeate most materials, causing a gradual change in pressure which depends on the random distribution of

101 hydrogen gas bubbles in the electrolytic chamber (bubbles touching the chamber wall will leak hydrogen faster). In contrast, our novel reciprocating electrolytic pump avoids this problem by rapidly recombining the hydrogen with oxygen to reform water using a catalyst, in this case palladium acetate. Our current pump uses platinum particles on rubber particles.

Figure 54: The check valve we developed to make our pump small and efficient

102

Our pump requires very small check valves with low forward burst pressure and no reverse leakage, even at very low reverse pressures. Commercial check valves are too large, so we tried several of our own designs until we found a design which meets our criteria. The valve consists of a flexible membrane with a very small (50 micron diameter) hole which is stretched over a very smooth dome. Pressure applied in one direction lifts the membrane away from the dome, opening the hole (top left); whereas pressure applied in the opposite direction pushes the membrane against the dome, sealing the hole (middle left). The valve takes advantage of every square micron of pressure gradient exerted on the membrane to lift or lower the membrane, resulting in very efficient operation and zero reverse leakage, making our pump very efficient.

Our check valves can also be very small, on the order of 2 mm in diameter (bottom left).

Figure 55: Formation of the smooth dome of the check valve It is critical that the dome be perfectly smooth so that it can form a tight seal with the flexible silicone membrane. This is accomplish by casting the dome with a drop of glue.

103

Figure 56: The process of making the dome check valves The silicone membrane is cast on a micromachined mold (top left, background) which results in a disk shape with a hole in the center approximately 50 microns thick and 1.2 mm in diameter

(top right). The disk is lowered onto the smooth dome formed by the glue drop in the middle of the base piece (bottom left). The edges of the disk are then compressed against the outside rim of the base by friction-fitting the top cover (bottom right at right) which has an overhanging lip that presses down on the edge of the silicone disk. The assembly is then glued together.

104

Figure 57: Mouse walking a few minutes after awakening from surgery The pump is implanted under the skin of the upper back in a procedure which takes about 10 minutes per mouse.

Figure 58: Mouse running approximately one hour after having the pump implanted The pump weighs less than 10% of the average body weight of a young C57BL/6 mouse.

105

Figure 59: Testing the pump using bioluminescence imaging The mouse at left has been implanted the remote-control pump, which delivered luciferin resulting in a similar temporal profile of bioluminesence as when injected with the same volume using a syringe. The mouse at right is a control without luciferin.

Although the reciprocating electrolytic pump is small, consumes little power, and can be operated at a low voltage, it also has disadvantages revealed by testing. First, it takes several minutes for the oxygen and hydrogen to recombine even in the presence of a catalyst such as palladium acetate solution and platinum on hydrophobic microparticles. Second, the valves are vulnerable to being jammed open by particles, and although this can be addressed by filtering the drug solution, it requires a lot of care which slows down experiments. Third, sometimes the hydrogen and oxygen bubbles are located apart from each other and do not recombine during the recombination phase of the pump cycle, resulting in reduced pump capacity. Our experiences with the reciprocating peristaltic and reciprocating electrolytic pumps motivated us in our further

106 designs to prioritize reliability above all else. This approach resulted in the most promising of the three implantable pump solutions, the divertable constant-pressure pump.

Divertable constant-pressure pump

Figure 60: Simplified pump to increase reliability and user-friendliness To achieve our reliability goal of greater than 99.9% of doses within 5% of the target dose we pursued a simplified design in which flow from a pressurized reservoir is gated by a valve controlled using our radio control circuit. This approach has an extremely smooth fluid flow path from reservoir to target tissue, avoiding the complicated shapes that trapped bubbles in our electrolytic pumps, and reduces battery size since the energy to drive the fluid is stored in a mechanical or osmotic form, rather than in the batteries of the pump.

107

Figure 61: Initial designs for pressurizing the drug reservoir Each of these mechanical mechanisms resulted in highly variable flow rates. Complicated mechanical linkages can provide more constant pressure over time but are bulky. Therefore we turned to the most tested constant-pressure mouse-implantable source, osmotic pumps.

We found that Alzet osmotic pumps generate several atmospheres of pressure, and therefore our initial approach of simply opening or closing a single valve on the output of the osmotic pump resulted in burst valves. Therefore we altered the design so that rather than throttling the flow out of the osmotic pump, we diverted it with a 3-way valve either to the target tissue in the animal or to a waste reservoir (Figure 62).

108

Figure 62: Remote-control latching valves directing flow from an osmotic pump to either a target tissue or a waste reservoir Existing valves are too large and require too high a voltage to operate; for example Clark

Solutions sells what they advertise as the smallest commercial solenoid valve, which is 1.5 cm long and 5 mm in diameter and requires positive and negative 5V to operate. Therefore we tested several designs for low voltage microvalves. We considered a valve actuated by electroosmotic flow to displace a membrane and thereby either open or seal an orifice as a valve mechanism, however we did not prioritize this approach due to the need for a relatively bulky mechanism to remove gas bubbles resulting from electrolysis of water. Rather we focused on

SMA and watch stepper motor actuators, as both are extremely reliable, and the latter is also extremely energy-efficient.

Latching SMA-actuated 2-way valve

We attempted to fit the components of a latching valve into the smallest possible volume by wrapping an SMA wire actuator around several posts in a serpentine path, and making the member that compresses a tube to close it also engage or disengage a nearby tab to latch it or unlatch it. The smallest size we were able to achieve with a functional valve was 8x8x4 mm

(Figure 63).

109

Figure 63: Latching valve actuated by shape memory alloy (SMA) wire SMA wire contracts when heated by passing an electric current through it. Two SMA wires are wrapped around the circular member in the upper right. The circular member bears a tooth that can seal the catheter against an opposing surface, sealing the valve (middle, top). Activating one of the wires pulls the circular member and tooth away from the catheter, opening the valve and latching a catch on the round shape against a ledge (far right, bottom). Activating the other wire pulls the round member in a perpendicular direction, unlatching it and allowing the spring to push it back towards the catheter, resealing the catheter and closing the valve.

We tested the cycle time and sealing capacity of the valve and found that it allowed some leakage (Figure 64), and therefore we need to strengthen the spring that exerts the sealing force on the tube, or use a thinner tube. The current tube is 250 micron thick polyurethane with a 125 micron lumen.

110

Figure 64: Testing the function of the valve in Figure 63 The valve changes state from >95% opened to >95% closed or vice versa within approximately 5 s., and reaches 100% open or closed state within 10 s.

To avoid the need for two separate valves we tried making a 3-way version of the SMA-actuated

2-way latching valve by spreading the system out for easy manufacturing and testing.

Sliding rod SMA-actuated 3-way latching valve

In this valve, a single rod seals one of two catheters at a time (Figure 65).

111

Figure 65: Dual-catheter valve actuated by SMA wire This mechanism also had the advantage that the rod rolled rather than slide, reducing the required actuation force.

Figure 66: Close-up view of the valve in Figure 65 The catheters pass over two recesses in the bottom surface, such that the metal rod seals either one or the other catheter. The rod is pulled left and right by two SMA wires (out of frame, but visible in Figure 65). A future version will hopefully be collapsed down to the same footprint of the 2-way valve in Figure 63.

112

Watch stepper motor-actuated 3-way latching valve This valve employs a watch stepper motor to rotate a rod with a channel etched in one side within a Y-shaped microchannel, resulting in flow being directed from the source branch of the

Y to one of the two other branches (Figure 67 and Figure 68).

Figure 67: Three-way valve actuated by a watch stepper motor The blue piece is rotated by a watch stepper motor, causing a channel etched in its side to conduct fluid from the top channel to only one of the two horizontal channels at a time.

113

Figure 68: Close-up view of the conducting channel in the side of the rotor in Figure 67 The T-shaped channel is made of a soft silicone polymer which seals against the rotor and the rigid housing, shown in Figure 69.

Figure 69: Side view of a prototype of the rotor valve mechanism The rotor with the channel etched in its side projects into PDMS cast in the polycarbonate housing. The casting is done with three rods (not shown) place where the channels for conducting fluid will be.

114

Figure 70: Top view showing the housing and rotor in place in a watch stepper motor A current applied to the watch solenoid rotates the rotor by 180 degrees. This will allow the flow to switched from one output channel to the other.

At this point the resolution of our milling machine, 25 microns, was not sufficient to machine channel in the rotor for directing flow between the two destination tubes. Therefore the next step is to use a technique such as electrochemical machining, which applies a voltage in a conductive bath to corrode metal, or electrical discharge machining which applies a voltage in a dielectric bath resulting in erosion by arcing, to manufacture and test prototype rotors. Ultimately we hope to outsource manufacturing of the rotor to a company that makes rotors for watch motors.

115

Control and feedback system

To deliver hormones by remotely actuating the implanted pumps in response to events such as social interactions or exercise, we developed a feedback control system comprising an implantable transceiver, running, position, and bioacoustics sensors, and a custom software program to interpret the sensor data and send wireless signals to the pumps at appropriate times to deliver drugs. A schematic of the system is shown in Figure 71.

Figure 71: Schematic of the feedback system for controlling the implantable pumps

Electronics

The implantable circuit was designed around a Zarlink ZL70101 transceiver, the same chip used in the PillCam used clinically for remote gastrointestinal imaging. The ZL70101 uses the 402-405 MHz medical implant communications service (MICS) band allocated for local

116 medical communication. The chip reduces power consumption by shutting down the chip when not in use and allowing wake-up using a 2.4 GHz signal, which can be much stronger than the

MICS band signal is legally allowed to be in the US. We designed a circuit (Figure 72) in which the ZL70101 received control signals wirelessly and turned on or off one of two MOSFET transistors, enabling us to control the various pumps and valves described below. We laid out the circuit board using EagleCAD and had it made by a company. We then soldered on the surface mount components using a reflow oven. An example of the completed circuit board is visible above in Figure 50.

In some prototypes we connected one of the outputs to an infrared LED which we used to track the location of the mouse by emitting coded flashes detected by a camera mounted at the top of the mouse cage.

117

Figure 72: Circuit diagram for the implantable remote-control pump

118

The bottom portion of the circuit up to the chip is the filter that only allows MICS and 2.4 GHz frequencies through to the chip. The outputs to the left of the chip are connected to LEDs in this version for testing. The component on the right is the clock crystal for the chip.

Visual feedback

To enable tracking of multiple mice and social interactions we implanted infrared LEDs that could emit coded flashes and placed infrared and visible light cameras in the cage (Figure 73).

Figure 73: Video cameras mounted on a cage One of these two CCD cameras has had its infrared filter removed to allow detection of signals from infrared LEDs on the implanted pumps

The infrared signal was readily detected by the camera through the skin of the mice (Figure 74).

119

Figure 74: Detection of mouse location by infrared imaging The mouse in the running wheel in the visible light image at left is implanted with one of our pumps which is emitting identifying infrared flashes from its onboard infrared emitter. The flashes are detected by our custom software that outlines the infrared signal and the corresponding region on the visible image.

Running wheel feedback

To enable drug delivery at times relative to exercise we installed sensors on running wheels and connected their output to the computer controlling the remote-control implantable pumps (Figure 75).

120

Figure 75: Running wheel rotation sensor Rare earth magnets are mounted on the running wheels and detected by fixed Hall sensors on the cages. The signal from the Hall sensor is detected by an Arduino that relays rotation information via USB to the controlling computer. The yellow antenna is for communication with implanted pumps.

The circadian pattern of running wheel activity of the mice is evident in the output of a single day of the system tracking six mice (Figure 76).

121

Figure 76: Running wheel activity from six cages To trigger drug release, algorithms such as detection of increments of running exceeding a certain distance within a certain time period (arrows) can be used.

Bioacoustic feedback To monitor social interaction between mothers and pups we mounted ultrasonic microphones on the tops of sound isolation boxes enclosing mouse cages (Figure 77).

We found that the most effective method for distinguishing mother from pup vocalizations was cepstrum analysis which quantifies the ratios of overtones above the fundamental frequency. These ratios are highly consistent for a given speaker or musical instrument. Other parameters such as frequency, bandwidth, duration, and so forth were not as effective.

122

Figure 77: Ultrasonic microphones and cages in acoustic isolation chambers Left: Sound-insulated chambers with microphones in lids and computer running spectrum analysis and pump control software. Right: Mothers and pups are housed in cages in the chambers. The floor of the cages are covered with cotton batting, as we found that mouse movements in standard wood chips are loud in the ultrasound range.

The acoustic isolation chamber allowed us to record mother and pup vocalizations even in the relatively loud environment of the mouse room in which fan noise is constant (Figure 78).

Figure 78: Examples of recorded vocalizations Although at first glance there seemed to be differences between mother and pup vocalizations that could be detected by eye, in fact over hundreds of vocalizations the variety of sounds emitted by a given animal precluded such a method of discerning speakers. Rather it was

123 cepstrum analysis, in which the ratios of harmonics are quantified, that allowed mothers to be distinguished from pups.

Cage-mounted remote-control pump

In collaboration with Stanford veterinarians we developed the first remote-control cage- mounted drug delivery pump that is compatible with standard mouse housing (Figure 79). This housing connects to filtered air supply ports in the cage rack in the mouse housing facility, and therefore has limited space for attachments. Existing cage pumps sit on a table outside the cage and have a tether that dangles from a boom suspended over an open cage. We made our pump mountable on the front of the cage by creating a protective insert to allow the catheter to run under the lid of the cages without allowing air leakage. We also developed a plate-mail sheath that allows the tether to flex more than a spring, while completely protecting the enclosed catheter. Together these features allow the pump and tether to be used with standard housing. It will be important to add a commercially-available mouse-proof micro-swivel to allow the mouse to rotate freely without twisting the catheter.

124

Figure 79: Cage-mounted pump with plate mail catheter The cage-mounted pump was constructed using a 3D printer and commercial peristaltic pump. It is controlled by a custom Arduino board that connects to a computer via a USB or wireless connection. The system attaches to Stanford’s new cages with suction cups. The drug chamber is Peltier-cooled to ~10oC. The catheter has a diameter of 250 microns OD. The plate mail catheter armor with small (2 cm) radius of curvature survived one week coated with cheese in mouse cage (Kevlar and braided steel did not survive).

Conclusion and recommendations

125

We developed and evaluated four remote-control pumps, three implantable and one cage- mounted compared in Table 8, evaluated at approximately 3V because batteries capable of high discharge current (silver oxide and lithium polymer) produce approximately that voltage in series or alone, respectively.

Table 8: Figures of merit for the three four pumps developed here Pump Divertable type: Reciprocating constant pressure Peristaltic Watch Cage- Parameter (SMA) Electrolytic SMA motor mounted Energy (J at per ul 0.2 0.05 ** ** Unlimited 3V) per 2 0.5 1 <0.01 Unlimited cyc. Pressure (psi) 10+ 4 * * Unlimited Max. cycles 7200 tested 5 Unknown Unknown Unlimited Min. size (cm3) 2 1 2 2 Large Cycle time (s) 3 300 3 <1 <1 Reliability Medium Low High High High Failure modes Degassing Incomplete Tube Debris in causing recombination; crimping valve bubbles; Debris in check debris in valves check valves Major Power Reliability; Tube None Single weaknesses Slow cycling crimping known mouse per time cage * The maximum pressure of the pump is the pressure of the constant pressure source because one of the two output branches of the 3-way valve always provides a direct route from the constant pressure source to either the waste chamber or the delivery site in the animal. In the case of osmotic pumps, this pressure can reach several atmospheres. ** The divertable constant pressure pumps use latching valves, so the energy per cycle is meaningful but energy per ul is not i.e. once the valve is latched it does not consume energy.

Given their relative merits, we believe that the two most useful pumps for rodent studies are likely to be the cage-mounted and the divertable constant pressure watch motor-actuated pumps. The cage-mounted pump avoids implantation of everything except the catheter, minimizing animal stress. The divertable pump combines two highly reliable, low power, relatively miniature mechanisms: osmotic pumps and watch stepper motors.

126

Acknowledgements

I would like to thank my advisor Juan Santiago for his creativity and engineering skills, strong support over many years, inspiring level of knowledge and rigor, and good humor. I would like to thank my thesis committee member Robert Sapolsky for his keen insights into the feasibility of an artificial endocrine organ as an approach to rejuvenation and his many years of support in its pursuit and then in pursuit of telomere extension. I would like to thank my advisor

Helen Blau for having faith in this project and her constant support. I would like to thank

Michael Longaker for his helpful input. I would like to thank Eirik Ravnan for suggesting cepstrum analysis during his stay in the Santiago lab. I would like to thank the members of the

Santiago lab for their help, engineering and skills and judgment, and support. I would like to thank the AntiHero team for their dedication, shared vision, and camaraderie: Karl Stahl, Paul

Cooke, Angus Pacala, Mary Reynolds, Khaesha Hall, Giancarlo Garcia, Anita Rogacs, Curran

Kaushik, Francisco de la Paz, Ken Lopez, Michele Dragoescu, David Fenning, and Taher Ezzi.

127

Chapter 5

Conclusions and future directions

128

To date we have shown that in proliferating cells in culture TERT mRNA extends telomeres rapidly, enabling the equivalent of decades of telomere shortening in vivo to be reversed in a few days. Key milestones on the way to clinical translation will include demonstrating:

 Telomere extension in slowly-dividing and non-dividing cells.

 Telomere extension in vivo in animal models of aging and disease

 Safety

There are a variety of animal models of telomere-shortening related dysfunction and disease, including the mdx/mTR model for DMD, the TERT model which exhibits hypertension and a variety of aging phenotypes that are rescued by viral delivery of TERT, and the idiopathic pulmonary fibrosis model in which fibrosis is rescued by delivery or telomerase activator. To address these and other conditions and diseases it will be important to demonstrate telomere extension in stem cells and progenitors. Because stem cells are often slowly dividing or quiescent, and because telomerase is most active during S phase of the cell cycle, we have synthesized TERT mutants to avoid cell cycle regulation.

Figure 80: Mutation sites in our telomere-extending mutants

We have synthesized mRNA encoding a mutated form of telomerase, in order to bypass the post- translational regulatory mechanisms that suppress endogenous TERT in most cell types and in all cell types during interphase.

129

Figure 81: Co-delivery of factors to enhance telomere extension

To facilitate telomere extension in slow-dividing or quiescent cells we have synthesized the template to synthesize mRNA encoding a form of POT1 known to expose telomeres to telomerase even during interphase (Loayza and De Lange 2003, -1). We will test the ability of this construct to extend telomeres in slowly-dividing or quiescent cells such as freshly harvested

CD34+ hematopoietic progenitors.

The safety criteria we enumerated in the introduction have so far been satisfied by TERT mRNA, but our hypothesis that telomere extension performed infrequently using a brief, rapid treatment will reduce the overall risk of cancer remains to be tested.

Aging is the result of multiple mechanisms including epigenetic drift, DNA damage, accumulation of cellular waste, and telomere shortening (López-Otín et al. 2013). Therefore rejuvenation research is likely to converge on a combination of multiple preventive therapies and lifestyle practices to simultaneously address multiple of these mechanisms to prevent any one of them from curtailing human freedom, potential, and happiness.

130

References

Ahmed, Shaheda, João F Passos, Matthew J Birket, Tina Beckmann, Sebastian Brings, Heiko Peters, Mark A Birch-Machin, Thomas von Zglinicki, and Gabriele Saretzki. 2008. “Telomerase Does Not Counteract Telomere Shortening but Protects Mitochondrial Function under Oxidative Stress.” Journal of Cell Science 121 (Pt 7): 1046–53. doi:10.1242/jcs.019372. Akiyama, Masaharu, Teru Hideshima, Toshiaki Hayashi, Yu-Tzu Tai, Constantine S Mitsiades, Nicholas Mitsiades, Dharminder Chauhan, Paul Richardson, Nikhil C Munshi, and Kenneth C Anderson. 2002. “Cytokines Modulate Telomerase Activity in a Human Multiple Myeloma Cell Line.” Cancer Research 62 (13): 3876–82. Akiyama, Masaharu, Takeshi Kawano, Yoko Mikami-Terao, Miyuki Agawa-Ohta, Osamu Yamada, Hiroyuki Ida, and Hisashi Yamada. 2011. “Erythropoietin Activates Telomerase through Transcriptional and Posttranscriptional Regulation in Human Erythroleukemic JAS-REN-A Cells.” Leukemia Research 35 (3): 416–18. doi:10.1016/j.leukres.2010.11.002. Al-Nedawi, Khalid, Janusz Szemraj, and Czeslaw S Cierniewski. 2005. “Mast Cell-Derived Exosomes Activate Endothelial Cells to Secrete Plasminogen Activator Inhibitor Type 1.” Arteriosclerosis, Thrombosis, and Vascular Biology 25 (8): 1744–49. doi:10.1161/01.ATV.0000172007.86541.76. Allsopp, R C, H Vaziri, C Patterson, S Goldstein, E V Younglai, A B Futcher, C W Greider, and C B Harley. 1992. “Telomere Length Predicts Replicative Capacity of Human Fibroblasts.” Proceedings of the National Academy of Sciences of the United States of America 89 (21): 10114–18. Allsopp, Richard. 2012. “Telomere Length and iPSC Re-Programming: Survival of the Longest.” Cell Research 22 (4): 614–15. doi:10.1038/cr.2012.6. Alter, Blanche P, Neelam Giri, Sharon A Savage, and Philip S Rosenberg. 2009. “Cancer in Dyskeratosis Congenita.” Blood 113 (26): 6549–57. doi:10.1182/blood-2008-12-192880. Alvarez-Erviti, Lydia, Yiqi Seow, Haifang Yin, Corinne Betts, Samira Lakhal, and Matthew J A Wood. 2011. “Delivery of siRNA to the Mouse Brain by Systemic Injection of Targeted Exosomes.” Nature Biotechnology 29 (4): 341–45. doi:10.1038/nbt.1807. Anderson, Bart R, Hiromi Muramatsu, Babal K Jha, Robert H Silverman, Drew Weissman, and Katalin Karikó. 2011. “Nucleoside Modifications in RNA Limit Activation of 2’-5’- Oligoadenylate Synthetase and Increase Resistance to Cleavage by RNase L.” Nucleic Acids Research 39 (21): 9329–38. doi:10.1093/nar/gkr586. Anderson, Bart R, Hiromi Muramatsu, Subba R Nallagatla, Philip C Bevilacqua, Lauren H Sansing, Drew Weissman, and Katalin Karikó. 2010. “Incorporation of Pseudouridine into mRNA Enhances Translation by Diminishing PKR Activation.” Nucleic Acids Research 38 (17): 5884–92. doi:10.1093/nar/gkq347. Andrade, Luciana Nogueira de Sousa, Jason L Nathanson, Gene W Yeo, Carlos Frederico Martins Menck, and Alysson Renato Muotri. 2012. “Evidence for Premature Aging due to Oxidative Stress in iPSCs from Cockayne Syndrome.” Human Molecular Genetics 21 (17): 3825–34. doi:10.1093/hmg/dds211. Armanios, Mary, and Elizabeth H Blackburn. 2012. “The Telomere Syndromes.” Nature Reviews. Genetics 13 (10): 693–704. doi:10.1038/nrg3246.

131

Artandi, Steven E, Scott Alson, Maja K Tietze, Norman E Sharpless, Siqin Ye, Roger A Greenberg, Diego H Castrillon, et al. 2002. “Constitutive Telomerase Expression Promotes Mammary Carcinomas in Aging Mice.” Proceedings of the National Academy of Sciences of the United States of America 99 (12): 8191–96. doi:10.1073/pnas.112515399. Artandi, Steven E, and Ronald A DePinho. 2010. “Telomeres and Telomerase in Cancer.” Carcinogenesis 31 (1): 9–18. doi:10.1093/carcin/bgp268. Bachand, F, G Kukolj, and C Autexier. 2000. “Expression of hTERT and hTR in Cis Reconstitutes and Active Human Telomerase Ribonucleoprotein.” RNA (New York, N.Y.) 6 (5): 778–84. Batista, Luis F Z, Matthew F Pech, Franklin L Zhong, Ha Nam Nguyen, Kathleen T Xie, Arthur J Zaug, Sharon M Crary, et al. 2011. “Telomere Shortening and Loss of Self-Renewal in Dyskeratosis Congenita Induced Pluripotent Stem Cells.” Nature 474 (7351): 399–402. doi:10.1038/nature10084. Beitzinger, M, C Oswald, R Beinoraviciute-Kellner, and T Stiewe. 2006. “Regulation of Telomerase Activity by the p53 Family Member p73.” Oncogene 25 (6): 813–26. doi:10.1038/sj.onc.1209125. Bernardes de Jesus, Bruno, and Maria A Blasco. 2011. “Aging by Telomere Loss Can Be Reversed.” Cell Stem Cell 8 (1): 3–4. doi:10.1016/j.stem.2010.12.013. Bernardes de Jesus, Bruno, Kerstin Schneeberger, Elsa Vera, Agueda Tejera, Calvin B Harley, and Maria A Blasco. 2011. “The Telomerase Activator TA-65 Elongates Short Telomeres and Increases Health Span of Adult/old Mice without Increasing Cancer Incidence.” Aging Cell 10 (4): 604–21. doi:10.1111/j.1474-9726.2011.00700.x. Bernardes de Jesus, Bruno, Elsa Vera, Kerstin Schneeberger, Agueda M. Tejera, Eduard Ayuso, Fatima Bosch, and Maria A. Blasco. 2012. “Telomerase Gene Therapy in Adult and Old Mice Delays Aging and Increases Longevity without Increasing Cancer: TERT Alone Extends Lifespan of Adult/old Mice.” EMBO Molecular Medicine 4 (8): 691–704. doi:10.1002/emmm.201200245. Bestilny, L J, M J Gill, C H Mody, and K T Riabowol. 2000. “Accelerated Replicative Senescence of the Peripheral Immune System Induced by HIV Infection.” AIDS (London, England) 14 (7): 771–80. Binet, Romuald, Damien Ythier, Ana I Robles, Manuel Collado, Delphine Larrieu, Claire Fonti, Elisabeth Brambilla, et al. 2009. “WNT16B Is a New Marker of Cellular Senescence That Regulates p53 Activity and the Phosphoinositide 3-kinase/AKT Pathway.” Cancer Research 69 (24): 9183–91. doi:10.1158/0008-5472.CAN-09-1016. Blackburn, E H, and J G Gall. 1978. “A Tandemly Repeated Sequence at the Termini of the Extrachromosomal Ribosomal RNA Genes in Tetrahymena.” Journal of Molecular Biology 120 (1): 33–53. Blackburn, Elizabeth H. 2005. “Telomeres and Telomerase: Their Mechanisms of Action and the Effects of Altering Their Functions.” FEBS Letters 579 (4): 859–62. doi:10.1016/j.febslet.2004.11.036. ———. 2011. “Walking the Walk from Genes through Telomere Maintenance to Cancer Risk.” Cancer Prevention Research (Philadelphia, Pa.) 4 (4): 473–75. doi:10.1158/1940- 6207.CAPR-11-0066.

132

Blackburn, Elizabeth H, Thea D Tisty, and Scott M Lippman. 2010. “Unprecedented Opportunities and Promise for Cancer Prevention Research.” Cancer Prevention Research (Philadelphia, Pa.) 3 (4): 394–402. doi:10.1158/1940-6207.CAPR-10-0051. Blasco, Maria A. 2005. “Telomeres and Human Disease: Ageing, Cancer and beyond.” Nature Reviews. Genetics 6 (8): 611–22. doi:10.1038/nrg1656. Blau, H M, C Webster, and G K Pavlath. 1983. “Defective Myoblasts Identified in Duchenne Muscular Dystrophy.” Proceedings of the National Academy of Sciences of the United States of America 80 (15): 4856–60. Bodnar, A G, M Ouellette, M Frolkis, S E Holt, C P Chiu, G B Morin, C B Harley, J W Shay, S Lichtsteiner, and W E Wright. 1998. “Extension of Life-Span by Introduction of Telomerase into Normal Human Cells.” Science (New York, N.Y.) 279 (5349): 349–52. Calado, R T, J N Cooper, H M Padilla-Nash, E M Sloand, C O Wu, P Scheinberg, T Ried, and N S Young. 2012. “Short Telomeres Result in Chromosomal Instability in Hematopoietic Cells and Precede Malignant Evolution in Human Aplastic Anemia.” Leukemia: Official Journal of the Leukemia Society of America, Leukemia Research Fund, U.K 26 (4): 700– 707. doi:10.1038/leu.2011.272. Calado, Rodrigo, and Neal Young. 2012. “Telomeres in Disease.” F1000 Medicine Reports 4: 8. doi:10.3410/M4-8. Cawthon, Richard M. 2002. “Telomere Measurement by Quantitative PCR.” Nucleic Acids Research 30 (10): e47. ———. 2009. “Telomere Length Measurement by a Novel Monochrome Multiplex Quantitative PCR Method.” Nucleic Acids Research 37 (3): e21. doi:10.1093/nar/gkn1027. Cesare, Anthony J, and Roger R Reddel. 2010. “Alternative Lengthening of Telomeres: Models, Mechanisms and Implications.” Nature Reviews. Genetics 11 (5): 319–30. doi:10.1038/nrg2763. Chang, E, and C B Harley. 1995. “Telomere Length and Replicative Aging in Human Vascular Tissues.” Proceedings of the National Academy of Sciences of the United States of America 92 (24): 11190–94. Charette, M, and M W Gray. 2000. “Pseudouridine in RNA: What, Where, How, and Why.” IUBMB Life 49 (5): 341–51. doi:10.1080/152165400410182. Chow, Tracy T, Yong Zhao, Sabrina S Mak, Jerry W Shay, and Woodring E Wright. 2012. “Early and Late Steps in Telomere Overhang Processing in Normal Human Cells: The Position of the Final RNA Primer Drives Telomere Shortening.” Genes & Development 26 (11): 1167–78. doi:10.1101/gad.187211.112. Cifuentes-Rojas, Catherine, and Dorothy E Shippen. 2011. “Telomerase Regulation.” Mutation Research, October. doi:10.1016/j.mrfmmm.2011.10.003. http://www.ncbi.nlm.nih.gov/pubmed/22032831. Cohen, Sheldon, Denise Janicki-Deverts, Ronald B Turner, Margaretha L Casselbrant, Ha-Sheng Li-Korotky, Elissa S Epel, and William J Doyle. 2013. “Association between Telomere Length and Experimentally Induced Upper Respiratory Viral Infection in Healthy Adults.” JAMA: The Journal of the American Medical Association 309 (7): 699–705. doi:10.1001/jama.2013.613. Counter, C M, W C Hahn, W Wei, S D Caddle, R L Beijersbergen, P M Lansdorp, J M Sedivy, and R A Weinberg. 1998. “Dissociation among in Vitro Telomerase Activity, Telomere Maintenance, and Cellular Immortalization.” Proceedings of the National Academy of Sciences of the United States of America 95 (25): 14723–28.

133

Cristofalo, V J, and D Kritchevsky. 1969. “Cell Size and Nucleic Acid Content in the Diploid Human Cell Line WI-38 during Aging.” Medicina Experimentalis. International Journal of Experimental Medicine 19 (6): 313–20. Cristofalo, Vincent J, Antonello Lorenzini, R G Allen, Claudio Torres, and Maria Tresini. 2004. “Replicative Senescence: A Critical Review.” Mechanisms of Ageing and Development 125 (10-11): 827–48. doi:10.1016/j.mad.2004.07.010. Cristofari, G., and J. Lingner. 2006. “Telomere Length Homeostasis Requires That Telomerase Levels Are Limiting.” The EMBO Journal 25 (3): 565–74. Cusanelli, Emilio, Carmina Angelica Perez Romero, and Pascal Chartrand. 2013. “Telomeric Noncoding RNA TERRA Is Induced by Telomere Shortening to Nucleate Telomerase Molecules at Short Telomeres.” Molecular Cell 51 (6): 780–91. doi:10.1016/j.molcel.2013.08.029. Dalrymple, Nadine, and Erich R Mackow. 2011. “Productive Dengue Virus Infection of Human Endothelial Cells Is Directed by Heparan Sulfate-Containing Proteoglycan Receptors.” Journal of Virology 85 (18): 9478–85. doi:10.1128/JVI.05008-11. Davidson, Beverly L, and Paul B McCray Jr. 2011. “Current Prospects for RNA Interference- Based Therapies.” Nature Reviews. Genetics 12 (5): 329–40. doi:10.1038/nrg2968. De Lange, Titia. 2009. “How Telomeres Solve the End-Protection Problem.” Science (New York, N.Y.) 326 (5955): 948–52. doi:10.1126/science.1170633. Dhaun, Neeraj, Jane Goddard, Donald E Kohan, David M Pollock, Ernesto L Schiffrin, and David J Webb. 2008. “Role of Endothelin-1 in Clinical Hypertension: 20 Years on.” Hypertension 52 (3): 452–59. doi:10.1161/HYPERTENSIONAHA.108.117366. Di Mitri, Diletta, Rita I Azevedo, Sian M Henson, Valentina Libri, Natalie E Riddell, Richard Macaulay, David Kipling, Maria V D Soares, Luca Battistini, and Arne N Akbar. 2011. “Reversible Senescence in Human CD4+CD45RA+CD27- Memory T Cells.” Journal of Immunology (Baltimore, Md.: 1950) 187 (5): 2093–2100. doi:10.4049/jimmunol.1100978. Dimri, G P, X Lee, G Basile, M Acosta, G Scott, C Roskelley, E E Medrano, et al. 1995. “A Biomarker That Identifies Senescent Human Cells in Culture and in Aging Skin in Vivo.” Proceedings of the National Academy of Sciences of the United States of America 92 (20): 9363–67. Eitan, Erez, Ailon Tichon, Aviv Gazit, Daniel Gitler, Shimon Slavin, and Esther Priel. 2012. “Novel Telomerase-Increasing Compound in Mouse Brain Delays the Onset of Amyotrophic Lateral Sclerosis.” EMBO Molecular Medicine 4 (4): 313–29. doi:10.1002/emmm.201200212. Farzaneh-Far, Ramin, Jue Lin, Elissa S Epel, William S Harris, Elizabeth H Blackburn, and Mary A Whooley. 2010. “Association of Marine Omega-3 Fatty Acid Levels with Telomeric Aging in Patients with Coronary Heart Disease.” JAMA: The Journal of the American Medical Association 303 (3): 250–57. doi:10.1001/jama.2009.2008. Fitzpatrick, Annette L, Richard A Kronmal, Jeffrey P Gardner, Bruce M Psaty, Nancy S Jenny, Russell P Tracy, Jeremy Walston, Masyuki Kimura, and Abraham Aviv. 2007. “Leukocyte Telomere Length and Cardiovascular Disease in the Cardiovascular Health Study.” American Journal of Epidemiology 165 (1): 14–21. doi:10.1093/aje/kwj346. Friedman, H M, E J Macarak, R R MacGregor, J Wolfe, and N A Kefalides. 1981. “Virus Infection of Endothelial Cells.” The Journal of Infectious Diseases 143 (2): 266–73.

134

Fukunaga, Akiko, Takayuki Ishikawa, Masako Kishihata, Takero Shindo, Toshiyuki Hori, and Takashi Uchiyama. 2007. “Altered Homeostasis of CD4(+) Memory T Cells in Allogeneic Hematopoietic Stem Cell Transplant Recipients: Chronic Graft-versus-Host Disease Enhances T Cell Differentiation and Exhausts Central Memory T Cell Pool.” Biology of Blood and Marrow Transplantation: Journal of the American Society for Blood and Marrow Transplantation 13 (10): 1176–84. doi:10.1016/j.bbmt.2007.06.009. Gilbert, P. M., K. L. Havenstrite, K. E. G. Magnusson, A. Sacco, N. A. Leonardi, P. Kraft, N. K. Nguyen, S. Thrun, M. P. Lutolf, and H. M. Blau. 2010. “Substrate Elasticity Regulates Skeletal Muscle Stem Cell Self-Renewal in Culture.” Science 329 (5995): 1078–81. doi:10.1126/science.1191035. Gladych, Marta, Aneta Wojtyla, and Blazej Rubis. 2011. “Human Telomerase Expression Regulation.” Biochemistry and Cell Biology = Biochimie Et Biologie Cellulaire 89 (4): 359–76. doi:10.1139/o11-037. Greber, Boris, Philippe Coulon, Miao Zhang, Sören Moritz, Stefan Frank, Arnoldo José Müller- Molina, Marcos J Araúzo-Bravo, Dong Wook Han, Hans-Christian Pape, and Hans R Schöler. 2011. “FGF Signalling Inhibits Neural Induction in Human Embryonic Stem Cells.” The EMBO Journal 30 (24): 4874–84. doi:10.1038/emboj.2011.407. Greider, C W, and E H Blackburn. 1985. “Identification of a Specific Telomere Terminal Transferase Activity in Tetrahymena Extracts.” Cell 43 (2 Pt 1): 405–13. ———. 1989. “A Telomeric Sequence in the RNA of Tetrahymena Telomerase Required for Telomere Repeat Synthesis.” Nature 337 (6205): 331–37. doi:10.1038/337331a0. Harley, Calvin B. 2002. “Telomerase Is Not an Oncogene.” Oncogene 21 (4): 494–502. doi:10.1038/sj.onc.1205076. ———. 2005. “Telomerase Therapeutics for Degenerative Diseases.” Current Molecular Medicine 5 (2): 205–11. Harley, Calvin B, Weimin Liu, Maria Blasco, Elsa Vera, William H Andrews, Laura A Briggs, and Joseph M Raffaele. 2011. “A Natural Product Telomerase Activator as Part of a Health Maintenance Program.” Rejuvenation Research 14 (1): 45–56. doi:10.1089/rej.2010.1085. Hayflick, L, and P S Moorhead. 1961. “The Serial Cultivation of Human Diploid Cell Strains.” Experimental Cell Research 25 (December): 585–621. Hohensinner, Philipp J, Jörg J Goronzy, and Cornelia M Weyand. 2011. “Telomere Dysfunction, Autoimmunity and Aging.” Aging and Disease 2 (6): 524–37. Hölig, Peter, Miriam Bach, Tina Völkel, Thomas Nahde, Sven Hoffmann, Rolf Müller, and Roland E Kontermann. 2004. “Novel RGD Lipopeptides for the Targeting of Liposomes to Integrin-Expressing Endothelial and Melanoma Cells.” Protein Engineering, Design & Selection: PEDS 17 (5): 433–41. doi:10.1093/protein/gzh055. Holt, S E, D L Aisner, J W Shay, and W E Wright. 1997a. “Lack of Cell Cycle Regulation of Telomerase Activity in Human Cells.” Proceedings of the National Academy of Sciences of the United States of America 94 (20): 10687–92. Holt, S. E, D. L Aisner, J. W Shay, and W. E Wright. 1997b. “Lack of Cell Cycle Regulation of Telomerase Activity in Human Cells.” Proceedings of the National Academy of Sciences 94 (20): 10687. Hong, Hyenjong, Kazutoshi Takahashi, Tomoko Ichisaka, Takashi Aoi, Osami Kanagawa, Masato Nakagawa, Keisuke Okita, and Shinya Yamanaka. 2009. “Suppression of

135

Induced Pluripotent Stem Cell Generation by the p53-p21 Pathway.” Nature 460 (7259): 1132–35. doi:10.1038/nature08235. Hood, Joshua L, and Samuel A Wickline. 2012. “A Systematic Approach to Exosome-Based Translational Nanomedicine.” Wiley Interdisciplinary Reviews. Nanomedicine and Nanobiotechnology 4 (4): 458–67. doi:10.1002/wnan.1174. Hornung, Veit, Jana Ellegast, Sarah Kim, Krzysztof Brzózka, Andreas Jung, Hiroki Kato, Hendrik Poeck, et al. 2006. “5’-Triphosphate RNA Is the Ligand for RIG-I.” Science (New York, N.Y.) 314 (5801): 994–97. doi:10.1126/science.1132505. Hsu, L, J Ramunas, J Gonzalez, JG Santiago, and DG Strickland. 2011. “Toward an Electrolytic Micropump Actuator Design with Controlled Cyclic Bubble Growth and Recombination.” ECS Transactions 35 (30): 3–11. Ilmonen, Petteri, Alexander Kotrschal, and Dustin J Penn. 2008. “Telomere Attrition due to Infection.” PloS One 3 (5): e2143. doi:10.1371/journal.pone.0002143. Imanishi, Toshio, Takuzo Hano, and Ichiro Nishio. 2005. “Estrogen Reduces Endothelial Progenitor Cell Senescence through Augmentation of Telomerase Activity.” Journal of Hypertension 23 (9): 1699–1706. Jaskelioff, Mariela, Florian L Muller, Ji-Hye Paik, Emily Thomas, Shan Jiang, Andrew C Adams, Ergun Sahin, et al. 2011. “Telomerase Reactivation Reverses Tissue Degeneration in Aged Telomerase-Deficient Mice.” Nature 469 (7328): 102–6. doi:10.1038/nature09603. Kabnick, K S, and D E Housman. 1988. “Determinants That Contribute to Cytoplasmic Stability of Human c-Fos and Beta-Globin mRNAs Are Located at Several Sites in Each mRNA.” Molecular and Cellular Biology 8 (8): 3244–50. Kaeppel, Christine, Stuart G Beattie, Raffaele Fronza, Richard van Logtenstein, Florence Salmon, Sabine Schmidt, Stephan Wolf, et al. 2013. “A Largely Random AAV Integration Profile after LPLD Gene Therapy.” Nature Medicine 19 (7): 889–91. doi:10.1038/nm.3230. Karikó, K, J M Keller, V A Harris, D J Langer, and F A Welsh. 2001. “In Vivo Protein Expression from mRNA Delivered into Adult Rat Brain.” Journal of Neuroscience Methods 105 (1): 77–86. Karikó, Katalin, Michael Buckstein, Houping Ni, and Drew Weissman. 2005. “Suppression of RNA Recognition by Toll-like Receptors: The Impact of Nucleoside Modification and the Evolutionary Origin of RNA.” Immunity 23 (2): 165–75. doi:10.1016/j.immuni.2005.06.008. Karikó, Katalin, Hiromi Muramatsu, Jason M Keller, and Drew Weissman. 2012. “Increased Erythropoiesis in Mice Injected with Submicrogram Quantities of Pseudouridine- Containing mRNA Encoding Erythropoietin.” Molecular Therapy: The Journal of the American Society of Gene Therapy 20 (5): 948–53. doi:10.1038/mt.2012.7. Karikó, Katalin, Hiromi Muramatsu, János Ludwig, and Drew Weissman. 2011. “Generating the Optimal mRNA for Therapy: HPLC Purification Eliminates Immune Activation and Improves Translation of Nucleoside-Modified, Protein-Encoding mRNA.” Nucleic Acids Research 39 (21): e142. doi:10.1093/nar/gkr695. Kawai, Taro, and Shizuo Akira. 2010. “The Role of Pattern-Recognition Receptors in Innate Immunity: Update on Toll-like Receptors.” Nature Immunology 11 (5): 373–84. doi:10.1038/ni.1863.

136

Ko, Young Tag, Raktima Bhattacharya, and Ulrich Bickel. 2009. “Liposome Encapsulated polyethylenimine/ODN Polyplexes for Brain Targeting.” Journal of Controlled Release: Official Journal of the Controlled Release Society 133 (3): 230–37. doi:10.1016/j.jconrel.2008.10.013. Kokkinaki, Maria, Niaz Sahibzada, and Nady Golestaneh. 2011. “Human Induced Pluripotent Stem-Derived Retinal Pigment Epithelium (RPE) Cells Exhibit Ion Transport, Membrane Potential, Polarized Vascular Endothelial Growth Factor Secretion, and Gene Expression Pattern Similar to Native RPE.” Stem Cells (Dayton, Ohio) 29 (5): 825–35. doi:10.1002/stem.635. Kormann, Michael S D, Günther Hasenpusch, Manish K Aneja, Gabriela Nica, Andreas W Flemmer, Susanne Herber-Jonat, Marceline Huppmann, et al. 2011a. “Expression of Therapeutic Proteins after Delivery of Chemically Modified mRNA in Mice.” Nature Biotechnology 29 (2): 154–57. doi:10.1038/nbt.1733. ———. 2011b. “Expression of Therapeutic Proteins after Delivery of Chemically Modified mRNA in Mice.” Nature Biotechnology 29 (2): 154–57. doi:10.1038/nbt.1733. ———. 2011c. “Expression of Therapeutic Proteins after Delivery of Chemically Modified mRNA in Mice.” Nature Biotechnology 29 (2): 154–57. doi:10.1038/nbt.1733. Kuwahara, Hiroya, Kazutaka Nishina, Kie Yoshida, Tomoko Nishina, Mariko Yamamoto, Yukari Saito, Wenying Piao, Masayuki Yoshida, Hidehiro Mizusawa, and Takanori Yokota. 2011. “Efficient in Vivo Delivery of siRNA into Brain Capillary Endothelial Cells along with Endogenous Lipoprotein.” Molecular Therapy: The Journal of the American Society of Gene Therapy 19 (12): 2213–21. doi:10.1038/mt.2011.186. Lakhal, Samira, and Matthew J A Wood. 2011. “Exosome Nanotechnology: An Emerging Paradigm Shift in Drug Delivery: Exploitation of Exosome Nanovesicles for Systemic in Vivo Delivery of RNAi Heralds New Horizons for Drug Delivery across Biological Barriers.” BioEssays: News and Reviews in Molecular, Cellular and Developmental Biology 33 (10): 737–41. doi:10.1002/bies.201100076. Lansdorp, Peter M. 2005. “Major Cutbacks at Chromosome Ends.” Trends in Biochemical Sciences 30 (7): 388–95. doi:10.1016/j.tibs.2005.05.004. Laser, DG, and JG Santiago. 2004. “A Review of Micropumps.” J. Micromech. Microeng, 14: R35–R64. Lawless, Conor, Chunfang Wang, Diana Jurk, Alina Merz, Thomas von Zglinicki, and João F Passos. 2010. “Quantitative Assessment of Markers for Cell Senescence.” Experimental Gerontology 45 (10): 772–78. doi:10.1016/j.exger.2010.01.018. Le, Rongrong, Zhaohui Kou, Yonghua Jiang, Ming Li, Bo Huang, Wenqiang Liu, Hui Li, et al. 2013. “Enhanced Telomere Rejuvenation in Pluripotent Cells Reprogrammed via Nuclear Transfer Relative to Induced Pluripotent Stem Cells.” Cell Stem Cell, November. doi:10.1016/j.stem.2013.11.005. Lee, Ji Hoon, Prabhat Khadka, Seung Han Baek, and In Kwon Chung. 2010. “CHIP Promotes Human Telomerase Reverse Transcriptase Degradation and Negatively Regulates Telomerase Activity.” The Journal of Biological Chemistry 285 (53): 42033–45. doi:10.1074/jbc.M110.149831. Lin, Shiaw Yih, and Stephen J Elledge. 2003. “Multiple Tumor Suppressor Pathways Negatively Regulate Telomerase.” Cell 113 (7): 881–89.

137

Lingner, J, T R Hughes, A Shevchenko, M Mann, V Lundblad, and T R Cech. 1997. “Reverse Transcriptase Motifs in the Catalytic Subunit of Telomerase.” Science (New York, N.Y.) 276 (5312): 561–67. Liu, Jun-Ping, Shi-Ming Chen, Yu-Sheng Cong, Craig Nicholls, Shu-Feng Zhou, Ze-Zhang Tao, and He Li. 2010. “Regulation of Telomerase Activity by Apparently Opposing Elements.” Ageing Research Reviews 9 (3): 245–56. doi:10.1016/j.arr.2010.03.002. Loayza, Diego, and Titia De Lange. 2003. “POT1 as a Terminal Transducer of TRF1 Telomere Length Control.” Nature 423 (6943): 1013–18. doi:10.1038/nature01688. López-Otín, Carlos, Maria A Blasco, Linda Partridge, Manuel Serrano, and Guido Kroemer. 2013. “The Hallmarks of Aging.” Cell 153 (6): 1194–1217. doi:10.1016/j.cell.2013.05.039. Ma, Hongxia, Ziyuan Zhou, Sheng Wei, Zhensheng Liu, Karen A Pooley, Alison M Dunning, Ulrika Svenson, et al. 2011. “Shortened Telomere Length Is Associated with Increased Risk of Cancer: A Meta-Analysis.” PloS One 6 (6): e20466. doi:10.1371/journal.pone.0020466. Marión, Rosa M, Katerina Strati, Han Li, Matilde Murga, Raquel Blanco, Sagrario Ortega, Oscar Fernandez-Capetillo, Manuel Serrano, and Maria A Blasco. 2009. “A p53-Mediated DNA Damage Response Limits Reprogramming to Ensure iPS Cell Genomic Integrity.” Nature 460 (7259): 1149–53. doi:10.1038/nature08287. Martin, Molly E, and Kevin G Rice. 2007. “Peptide-Guided Gene Delivery.” The AAPS Journal 9 (1): E18–29. doi:10.1208/aapsj0901003. Martin-Ruiz, Carmen, Heather O Dickinson, Barbara Keys, Elise Rowan, Rose Anne Kenny, and Thomas Von Zglinicki. 2006. “Telomere Length Predicts Poststroke Mortality, Dementia, and Cognitive Decline.” Annals of Neurology 60 (2): 174–80. doi:10.1002/ana.20869. Masutomi, Kenkichi, Evan Y Yu, Shilagardy Khurts, Ittai Ben-Porath, Jennifer L Currier, Geoffrey B Metz, Mary W Brooks, et al. 2003. “Telomerase Maintains Telomere Structure in Normal Human Cells.” Cell 114 (2): 241–53. Matsushita, H, E Chang, A J Glassford, J P Cooke, C P Chiu, and P S Tsao. 2001. “eNOS Activity Is Reduced in Senescent Human Endothelial Cells: Preservation by hTERT Immortalization.” Circulation Research 89 (9): 793–98. Millay, Douglas P, Jason R O’Rourke, Lillian B Sutherland, Svetlana Bezprozvannaya, John M Shelton, Rhonda Bassel-Duby, and Eric N Olson. 2013. “Myomaker Is a Membrane Activator of Myoblast Fusion and Muscle Formation.” Nature 499 (7458): 301–5. doi:10.1038/nature12343. Mogford, Jon E, W Robert Liu, Russell Reid, Choy-Pik Chiu, Hakim Said, Shu-Jen Chen, Calvin B Harley, and Thomas A Mustoe. 2006. “Adenoviral Human Telomerase Reverse Transcriptase Dramatically Improves Ischemic Wound Healing without Detrimental Immune Response in an Aged Rabbit Model.” Human Gene Therapy 17 (6): 651–60. doi:10.1089/hum.2006.17.651. Mohsin, Sadia, Mohsin Khan, Jonathan Nguyen, Monique Alkatib, Sailay Siddiqi, Nirmala Hariharan, Kathleen Wallach, et al. 2013. “Rejuvenation of Human Cardiac Progenitor Cells with Pim-1 Kinase.” Circulation Research 113 (10): 1169–79. doi:10.1161/CIRCRESAHA.113.302302. Mohsin, Sadia, Mohsin Khan, Haruhiro Toko, Brandi Bailey, Christopher T Cottage, Kathleen Wallach, Divya Nag, et al. 2012. “Human Cardiac Progenitor Cells Engineered With

138

Pim-I Kinase Enhance Myocardial Repair.” Journal of the American College of Cardiology, July. doi:10.1016/j.jacc.2012.04.047. http://www.ncbi.nlm.nih.gov/pubmed/22841153. Mourkioti, Foteini, Jackie Kustan, Peggy Kraft, John W Day, Ming-Ming Zhao, Maria Kost- Alimova, Alexei Protopopov, et al. 2013. “Role of Telomere Dysfunction in Cardiac Failure in Duchenne Muscular Dystrophy.” Nature Cell Biology 15 (8): 895–904. doi:10.1038/ncb2790. Nazarenko, Irina, Sanyukta Rana, Alexandra Baumann, Jessica McAlear, Andrea Hellwig, Michael Trendelenburg, Günter Lochnit, Klaus T Preissner, and Margot Zöller. 2010. “Cell Surface Tetraspanin Tspan8 Contributes to Molecular Pathways of Exosome- Induced Endothelial Cell Activation.” Cancer Research 70 (4): 1668–78. doi:10.1158/0008-5472.CAN-09-2470. Nicholls, Craig, He Li, Jian-Qiu Wang, and Jun-Ping Liu. 2011. “Molecular Regulation of Telomerase Activity in Aging.” Protein & Cell 2 (9): 726–38. doi:10.1007/s13238-011- 1093-3. Nicklin, S A, S J White, S J Watkins, R E Hawkins, and A H Baker. 2000. “Selective Targeting of Gene Transfer to Vascular Endothelial Cells by Use of Peptides Isolated by Phage Display.” Circulation 102 (2): 231–37. Njajou, Omer T, Wen-Chi Hsueh, Elizabeth H Blackburn, Anne B Newman, Shih-Hsuan Wu, Rongling Li, Eleanor M Simonsick, Tamara M Harris, Steve R Cummings, and Richard M Cawthon. 2009. “Association between Telomere Length, Specific Causes of Death, and Years of Healthy Life in Health, Aging, and Body Composition, a Population-Based Cohort Study.” The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 64 (8): 860–64. doi:10.1093/gerona/glp061. O’Sullivan, Roderick J, and Jan Karlseder. 2010. “Telomeres: Protecting Chromosomes against Genome Instability.” Nature Reviews. Molecular Cell Biology 11 (3): 171–81. doi:10.1038/nrm2848. Panda, Alexander, Alvaro Arjona, Elizabeth Sapey, Fengwei Bai, Erol Fikrig, Ruth R Montgomery, Janet M Lord, and Albert C Shaw. 2009. “Human Innate Immunosenescence: Causes and Consequences for Immunity in Old Age.” Trends in Immunology 30 (7): 325–33. doi:10.1016/j.it.2009.05.004. Pardi, Norbert, Hiromi Muramatsu, Drew Weissman, and Katalin Karikó. 2013. “In Vitro Transcription of Long RNA Containing Modified Nucleosides.” Methods in Molecular Biology (Clifton, N.J.) 969: 29–42. doi:10.1007/978-1-62703-260-5_2. Pearce, V P, J Sherrell, Z Lou, L Kopelovich, W E Wright, and J W Shay. 2008. “Immortalization of Epithelial Progenitor Cells Mediated by Resveratrol.” Oncogene 27 (17): 2365–74. doi:10.1038/sj.onc.1210886. Pérez-Rivero, Gema, María P Ruiz-Torres, Juan V Rivas-Elena, Mirjana Jerkic, María L Díez- Marques, José M Lopez-Novoa, María A Blasco, and Diego Rodríguez-Puyol. 2006. “Mice Deficient in Telomerase Activity Develop Hypertension because of an Excess of Endothelin Production.” Circulation 114 (4): 309–17. doi:10.1161/CIRCULATIONAHA.105.611111. Prade-Houdellier, N, E Frébet, C Demur, E-F Gautier, F Delhommeau, A-L Bennaceur-Griscelli, C Gaudin, et al. 2007. “Human Telomerase Is Regulated by Erythropoietin and Transforming Growth Factor-Beta in Human Erythroid Progenitor Cells.” Leukemia:

139

Official Journal of the Leukemia Society of America, Leukemia Research Fund, U.K 21 (11): 2304–10. doi:10.1038/sj.leu.2404874. Pucci, Fabio, Laura Gardano, and Lea Harrington. 2013. “Short Telomeres in ESCs Lead to Unstable Differentiation.” Cell Stem Cell 12 (4): 479–86. doi:10.1016/j.stem.2013.01.018. Rajotte, D, W Arap, M Hagedorn, E Koivunen, R Pasqualini, and E Ruoslahti. 1998. “Molecular Heterogeneity of the Vascular Endothelium Revealed by in Vivo Phage Display.” The Journal of Clinical Investigation 102 (2): 430–37. doi:10.1172/JCI3008. Rana, Sanyukta, and Margot Zöller. 2011. “Exosome Target Cell Selection and the Importance of Exosomal Tetraspanins: A Hypothesis.” Biochemical Society Transactions 39 (2): 559–62. doi:10.1042/BST0390559. Rothe, Michael, Ute Modlich, and Axel Schambach. 2013. “Biosafety Challenges for Use of Lentiviral Vectors in Gene Therapy.” Current Gene Therapy, November. Sacco, Alessandra, Foteini Mourkioti, Rose Tran, Jinkuk Choi, Michael Llewellyn, Peggy Kraft, Marina Shkreli, et al. 2010. “Short Telomeres and Stem Cell Exhaustion Model Duchenne Muscular Dystrophy in mdx/mTR Mice.” Cell 143 (7): 1059–71. doi:10.1016/j.cell.2010.11.039. Saebøe-Larssen, Stein, Ellen Fossberg, and Gustav Gaudernack. 2002. “mRNA-Based Electrotransfection of Human Dendritic Cells and Induction of Cytotoxic T Lymphocyte Responses against the Telomerase Catalytic Subunit (hTERT).” Journal of Immunological Methods 259 (1-2): 191–203. Sahin, Ergün, Simona Colla, Marc Liesa, Javid Moslehi, Florian L Müller, Mira Guo, Marcus Cooper, et al. 2011. “Telomere Dysfunction Induces Metabolic and Mitochondrial Compromise.” Nature 470 (7334): 359–65. doi:10.1038/nature09787. Sahin, Ergün, and Ronald A Depinho. 2010. “Linking Functional Decline of Telomeres, Mitochondria and Stem Cells during Ageing.” Nature 464 (7288): 520–28. doi:10.1038/nature08982. Samoylova, T I, and B F Smith. 1999. “Elucidation of Muscle-Binding Peptides by Phage Display Screening.” Muscle & Nerve 22 (4): 460–66. Shay, J W, and W E Wright. 2001. “Telomeres and Telomerase: Implications for Cancer and Aging.” Radiation Research 155 (1 Pt 2): 188–93. Shkolnikov, Viktor, John Ramunas, and Juan G Santiago. 2010. “A Self-Priming, Roller-Free, Miniature, Peristaltic Pump Operable with a Single, Reciprocating Actuator.” Sensors and Actuators. A, Physical 160 (1-2): 141–46. doi:10.1016/j.sna.2010.04.018. Signer, Robert A J, and Sean J Morrison. 2013. “Mechanisms That Regulate Stem Cell Aging and Life Span.” Cell Stem Cell 12 (2): 152–65. doi:10.1016/j.stem.2013.01.001. Sitte, N, G Saretzki, and T von Zglinicki. 1998. “Accelerated Telomere Shortening in Fibroblasts after Extended Periods of Confluency.” Free Radical Biology & Medicine 24 (6): 885– 93. Steinert, S, J W Shay, and W E Wright. 2000. “Transient Expression of Human Telomerase Extends the Life Span of Normal Human Fibroblasts.” Biochemical and Biophysical Research Communications 273 (3): 1095–98. doi:10.1006/bbrc.2000.3080. Stern, J L, and T M Bryan. 2008. “Telomerase Recruitment to Telomeres.” Cytogenetic and Genome Research 122 (3-4): 243–54. doi:10.1159/000167810.

140

Strathearn, L., P. Fonagy, J. Amico, and P. R Montague. 2009. “Adult Attachment Predicts Maternal Brain and Oxytocin Response to Infant Cues.” Neuropsychopharmacology 34 (13): 2655–66. Strickland, D, J Ramunas, J Gonzalez, and JG Santiago. 2011. “Electrolytic Micropump Actuator with Controlled Cyclic Bubble Growth and Recombination.” 219th Meeting of the Electrochem. Society May 1-6. Suhr, Steven T, Eun Ah Chang, Ramon M Rodriguez, Kai Wang, Pablo J Ross, Zeki Beyhan, Shashanka Murthy, and Jose B Cibelli. 2009. “Telomere Dynamics in Human Cells Reprogrammed to Pluripotency.” PloS One 4 (12): e8124. doi:10.1371/journal.pone.0008124. Szostak, J W, and E H Blackburn. 1982. “Cloning Yeast Telomeres on Linear Plasmid Vectors.” Cell 29 (1): 245–55. Takubo, Kaiyo, Junko Aida, Naotaka Izumiyama-Shimomura, Naoshi Ishikawa, Motoji Sawabe, Rie Kurabayashi, Hiroaki Shiraishi, Tomio Arai, and Ken-Ichi Nakamura. 2010. “Changes of Telomere Length with Aging.” Geriatrics & Gerontology International 10 Suppl 1 (July): S197–206. doi:10.1111/j.1447-0594.2010.00605.x. Tauro, Bow J, David W Greening, Rommel A Mathias, Hong Ji, Suresh Mathivanan, Andrew M Scott, and Richard J Simpson. 2012. “Comparison of Ultracentrifugation, Density Gradient Separation, and Immunoaffinity Capture Methods for Isolating Human Colon Cancer Cell Line LIM1863-Derived Exosomes.” Methods (San Diego, Calif.) 56 (2): 293–304. doi:10.1016/j.ymeth.2012.01.002. Tavernier, Geertrui, Oliwia Andries, Jo Demeester, Niek N Sanders, Stefaan C De Smedt, and Joanna Rejman. 2011. “mRNA as Gene Therapeutic: How to Control Protein Expression.” Journal of Controlled Release: Official Journal of the Controlled Release Society 150 (3): 238–47. doi:10.1016/j.jconrel.2010.10.020. Thiemann, Sandra, and Linda G Baum. 2011. “The Road Less Traveled: Regulation of Leukocyte Migration across Vascular and Lymphatic Endothelium by Galectins.” Journal of Clinical Immunology 31 (1): 2–9. doi:10.1007/s10875-010-9460-z. Ueda, M, A Ouhtit, T Bito, K Nakazawa, J Lübbe, M Ichihashi, H Yamasaki, and H Nakazawa. 1997. “Evidence for UV-Associated Activation of Telomerase in Human Skin.” Cancer Research 57 (3): 370–74. Untergasser, Andreas, Ioana Cutcutache, Triinu Koressaar, Jian Ye, Brant C Faircloth, Maido Remm, and Steven G Rozen. 2012. “Primer3--New Capabilities and Interfaces.” Nucleic Acids Research 40 (15): e115. doi:10.1093/nar/gks596. Utikal, Jochen, Jose M Polo, Matthias Stadtfeld, Nimet Maherali, Warakorn Kulalert, Ryan M Walsh, Adam Khalil, James G Rheinwald, and Konrad Hochedlinger. 2009. “Immortalization Eliminates a Roadblock during Cellular Reprogramming into iPS Cells.” Nature 460 (7259): 1145–48. doi:10.1038/nature08285. Valadi, Hadi, Karin Ekström, Apostolos Bossios, Margareta Sjöstrand, James J Lee, and Jan O Lötvall. 2007. “Exosome-Mediated Transfer of mRNAs and microRNAs Is a Novel Mechanism of Genetic Exchange between Cells.” Nature Cell Biology 9 (6): 654–59. doi:10.1038/ncb1596. Van den Boorn, Jasper G, Martin Schlee, Christoph Coch, and Gunther Hartmann. 2011. “SiRNA Delivery with Exosome Nanoparticles.” Nature Biotechnology 29 (4): 325–26. doi:10.1038/nbt.1830.

141

Van Dommelen, S M, P Vader, S Lakhal, W W van Solinge, M J A Wood, and R M Schiffelers. 2011. “Microvesicles and Exosomes: Opportunities for Cell-Derived Membrane Vesicles in Drug Delivery.” Journal of Controlled Release: Official Journal of the Controlled Release Society, November. doi:10.1016/j.jconrel.2011.11.021. http://www.ncbi.nlm.nih.gov/pubmed/22138068. Van Rooy, Inge, Serpil Cakir-Tascioglu, Pierre-Olivier Couraud, Ignacio A Romero, Babette Weksler, Gert Storm, Wim E Hennink, Raymond M Schiffelers, and Enrico Mastrobattista. 2010. “Identification of Peptide Ligands for Targeting to the Blood-Brain Barrier.” Pharmaceutical Research 27 (4): 673–82. doi:10.1007/s11095-010-0053-6. Vandendriessche, Thierry, and Marinee K Chuah. 2013. “Targeting Endothelial Cells by Gene Therapy.” Blood 122 (12): 1993–94. doi:10.1182/blood-2013-08-518266. Vaziri, H, and S Benchimol. 1998. “Reconstitution of Telomerase Activity in Normal Human Cells Leads to Elongation of Telomeres and Extended Replicative Life Span.” Current Biology: CB 8 (5): 279–82. Vaziri, H, K B Chapman, A Guigova, J Teichroeb, M D Lacher, H Sternberg, I Singec, et al. 2010. “Spontaneous Reversal of the Developmental Aging of Normal Human Cells Following Transcriptional Reprogramming.” Regenerative Medicine 5 (3): 345–63. doi:10.2217/rme.10.21. Von Zglinicki, T, V Serra, M Lorenz, G Saretzki, R Lenzen-Grossimlighaus, R Gessner, A Risch, and E Steinhagen-Thiessen. 2000. “Short Telomeres in Patients with Vascular Dementia: An Indicator of Low Antioxidative Capacity and a Possible Risk Factor?” Laboratory Investigation; a Journal of Technical Methods and Pathology 80 (11): 1739– 47. Von Zglinicki, Thomas. 2002. “Oxidative Stress Shortens Telomeres.” Trends in Biochemical Sciences 27 (7): 339–44. Wang, Fang, Yu Yin, Xiaoying Ye, Kai Liu, Haiying Zhu, Lingling Wang, Maria Chiourea, et al. 2012. “Molecular Insights into the Heterogeneity of Telomere Reprogramming in Induced Pluripotent Stem Cells.” Cell Research 22 (4): 757–68. doi:10.1038/cr.2011.201. Wang, Yuhua, Hsing-Hao Su, Yang Yang, Yunxia Hu, Lu Zhang, Pilar Blancafort, and Leaf Huang. 2013. “Systemic Delivery of Modified mRNA Encoding Herpes Simplex Virus 1 Thymidine Kinase for Targeted Cancer Gene Therapy.” Molecular Therapy: The Journal of the American Society of Gene Therapy 21 (2): 358–67. doi:10.1038/mt.2012.250. Webster, C, and H M Blau. 1990. “Accelerated Age-Related Decline in Replicative Life-Span of Duchenne Muscular Dystrophy Myoblasts: Implications for Cell and Gene Therapy.” Somatic Cell and Molecular Genetics 16 (6): 557–65. Wei, Hui-Yong, Li-Fang Jiang, Dan-Yun Fang, and Hui-Yu Guo. 2003. “Dengue Virus Type 2 Infects Human Endothelial Cells through Binding of the Viral Envelope Glycoprotein to Cell Surface Polypeptides.” The Journal of General Virology 84 (Pt 11): 3095–98. Weinrich, S L, R Pruzan, L Ma, M Ouellette, V M Tesmer, S E Holt, A G Bodnar, et al. 1997. “Reconstitution of Human Telomerase with the Template RNA Component hTR and the Catalytic Protein Subunit hTRT.” Nature Genetics 17 (4): 498–502. doi:10.1038/ng1297- 498. Weng, Nan-ping. 2012. “Telomeres and Immune Competency.” Current Opinion in Immunology 24 (4): 470–75. doi:10.1016/j.coi.2012.05.001. Wentzensen, Ingrid M, Lisa Mirabello, Ruth M Pfeiffer, and Sharon A Savage. 2011. “The Association of Telomere Length and Cancer: A Meta-Analysis.” Cancer Epidemiology,

142

Biomarkers & Prevention: A Publication of the American Association for Cancer Research, Cosponsored by the American Society of Preventive Oncology 20 (6): 1238– 50. doi:10.1158/1055-9965.EPI-11-0005. Westin, Erik R, Elizabeth Chavez, Kimberly M Lee, Francoise A Gourronc, Soraya Riley, Peter M Lansdorp, Frederick D Goldman, and Aloysius J Klingelhutz. 2007. “Telomere Restoration and Extension of Proliferative Lifespan in Dyskeratosis Congenita Fibroblasts.” Aging Cell 6 (3): 383–94. doi:10.1111/j.1474-9726.2007.00288.x. White, S J, S A Nicklin, T Sawamura, and A H Baker. 2001. “Identification of Peptides That Target the Endothelial Cell-Specific LOX-1 Receptor.” Hypertension 37 (2 Pt 2): 449– 55. Wick, M, D Zubov, and G Hagen. 1999. “Genomic Organization and Promoter Characterization of the Gene Encoding the Human Telomerase Reverse Transcriptase (hTERT).” Gene 232 (1): 97–106. Willeit, Peter, Johann Willeit, Anita Brandstätter, Silvia Ehrlenbach, Agnes Mayr, Arno Gasperi, Siegfried Weger, et al. 2010. “Cellular Aging Reflected by Leukocyte Telomere Length Predicts Advanced Atherosclerosis and Cardiovascular Disease Risk.” Arteriosclerosis, Thrombosis, and Vascular Biology 30 (8): 1649–56. doi:10.1161/ATVBAHA.110.205492. Wolfrum, Christian, Shuanping Shi, K Narayanannair Jayaprakash, Muthusamy Jayaraman, Gang Wang, Rajendra K Pandey, Kallanthottathil G Rajeev, et al. 2007. “Mechanisms and Optimization of in Vivo Delivery of Lipophilic siRNAs.” Nature Biotechnology 25 (10): 1149–57. doi:10.1038/nbt1339. Wu, Ying-Li, Charles Dudognon, Eric Nguyen, Josette Hillion, Frédéric Pendino, Ilona Tarkanyi, Janos Aradi, et al. 2006. “Immunodetection of Human Telomerase Reverse- Transcriptase (hTERT) Re-Appraised: Nucleolin and Telomerase Cross Paths.” Journal of Cell Science 119 (Pt 13): 2797–2806. doi:10.1242/jcs.03001. Wyatt, Haley. 2009. “Structure-Function Analysis of the Human Telomerase Reverse Transcriptase”. Calgary, Alberta: University of Calgary. Xu, D, A Gruber, M Björkholm, C Peterson, and P Pisa. 1999. “Suppression of Telomerase Reverse Transcriptase (hTERT) Expression in Differentiated HL-60 Cells: Regulatory Mechanisms.” British Journal of Cancer 80 (8): 1156–61. doi:10.1038/sj.bjc.6690480. Yakubov, Eduard, Gidi Rechavi, Shmuel Rozenblatt, and David Givol. 2010. “Reprogramming of Human Fibroblasts to Pluripotent Stem Cells Using mRNA of Four Transcription Factors.” Biochemical and Biophysical Research Communications 394 (1): 189–93. doi:10.1016/j.bbrc.2010.02.150. Yang, Zhiwei, Xin Huang, Hong Jiang, Yanrong Zhang, Hongxia Liu, Chuan Qin, Gilbert M Eisner, et al. 2009. “Short Telomeres and Prognosis of Hypertension in a Chinese Population.” Hypertension 53 (4): 639–45. doi:10.1161/HYPERTENSIONAHA.108.123752. Yang, Zhi-Zhang, Deanna M Grote, Steven C Ziesmer, Toshiro Niki, Mitsuomi Hirashima, Anne J Novak, Thomas E Witzig, and Stephen M Ansell. 2012. “IL-12 Upregulates TIM-3 Expression and Induces T Cell Exhaustion in Patients with Follicular B Cell Non- Hodgkin Lymphoma.” The Journal of Clinical Investigation 122 (4): 1271–82. doi:10.1172/JCI59806.

143

Yao, Shuhuai, David E Hertzog, Shulin Zeng, James C Mikkelsen Jr, and Juan G Santiago. 2003. “Porous Glass Electroosmotic Pumps: Design and Experiments.” Journal of Colloid and Interface Science 268 (1): 143–53. Yi, X, J W Shay, and W E Wright. 2001. “Quantitation of Telomerase Components and hTERT mRNA Splicing Patterns in Immortal Human Cells.” Nucleic Acids Research 29 (23): 4818–25. Zahid, Maliha, Brett E Phillips, Sean M Albers, Nick Giannoukakis, Simon C Watkins, and Paul D Robbins. 2010. “Identification of a Cardiac Specific Protein Transduction Domain by in Vivo Biopanning Using a M13 Phage Peptide Display Library in Mice.” PloS One 5 (8): e12252. doi:10.1371/journal.pone.0012252. Zeng, Hui, Claudia Pappas, Jessica A Belser, Katherine V Houser, Weiming Zhong, Debra A Wadford, Troy Stevens, Ron Balczon, Jacqueline M Katz, and Terrence M Tumpey. 2012. “Human Pulmonary Microvascular Endothelial Cells Support Productive Replication of Highly Pathogenic Avian Influenza Viruses: Possible Involvement in the Pathogenesis of Human H5N1 Virus Infection.” Journal of Virology 86 (2): 667–78. doi:10.1128/JVI.06348-11. Zhu, Chun-Hong, Vincent Mouly, Racquel N Cooper, Kamel Mamchaoui, Anne Bigot, Jerry W Shay, James P Di Santo, Gillian S Butler-Browne, and Woodring E Wright. 2007. “Cellular Senescence in Human Myoblasts Is Overcome by Human Telomerase Reverse Transcriptase and Cyclin-Dependent Kinase 4: Consequences in Aging Muscle and Therapeutic Strategies for Muscular Dystrophies.” Aging Cell 6 (4): 515–23. doi:10.1111/j.1474-9726.2007.00306.x. Zimmermann, Stefan, and Uwe M Martens. 2008. “Telomeres, Senescence, and Hematopoietic Stem Cells.” Cell and Tissue Research 331 (1): 79–90. doi:10.1007/s00441-007-0469-4. Zvereva, M I, D M Shcherbakova, and O A Dontsova. 2010. “Telomerase: Structure, Functions, and Activity Regulation.” 75 (13): 1563–83.

144