Epigenome editing of TGF-BETA receptors using DCAS9-KRAB system to alter mesenchymal stem cell differentiation potential

Degenerative disc disease is a major cause of workplace disability and costs the US $100 billion annually. A tissue donation of an intervertebral disc (IVD) is able to restore disc height, function, and mobility in a human patient. However, this solution relies on tissue donations, which are scarce, and has drawbacks like rejection by the host. This has led to significant interest in generating a tissue-engineered IVD. Despite significant progress, current tissue-engineered solutions have been unable to generate the tissue gradients in the IVD that contribute to its unique mechanical properties. We seek to explore techniques to generate these gradients using the lentiviral dCas9-KRAB system, a genetic engineering tool that modifies a cell's genome to express dCas9-KRAB and a single guide RNA to epigenetically downregulate target genes. We aim to utilize the dCas9-KRAB system to alter the differentiation potential of human adipose-derived mesenchymal stem cells by downregulating the transforming growth factor β receptors TGFBR1 and TGFBR2. We were able to attain up to 55% downregulation of target genes. However, past results on other projects were able to attain higher downregulation. Here, we diagnose possible causes for this project's suboptimal downregulation as well as how to conduct future work.

EPIGENOME EDITING OF TGF-BETA RECEPTORS USING DCAS9-KRAB SYSTEM TO ALTER MESENCHYMAL STEM CELL DIFFERENTIATION POTENTIAL by Paul J. Wissler A Senior Honors Thesis Submitted to the Faculty of The University of Utah In Partial Fulfillment of the Requirements for the Honors Degree in Bachelor of Science In Biomedical Engineering Approved: ______________________________ Robert D. Bowles, PhD Thesis Faculty Supervisor _____________________________ David Grainger, PhD Chair, Department of Bioengineering _______________________________ Kelly Broadhead, PhD Honors Faculty Advisor _____________________________ Sylvia D. Torti, PhD Dean, Honors College December 2017 Copyright © 2017 All Rights Reserved ii ABSTRACT Degenerative disc disease is a major cause of workplace disability and costs the US $100 billion annually. A tissue donation of an intervertebral disc (IVD) is able to restore disc height, function, and mobility in a human patient. However, this solution relies on tissue donations, which are scarce, and has drawbacks like rejection by the host. This has led to significant interest in generating a tissue-engineered IVD. Despite significant progress, current tissue-engineered solutions have been unable to generate the tissue gradients in the IVD that contribute to its unique mechanical properties. We seek to explore techniques to generate these gradients using the lentiviral dCas9-KRAB system, a genetic engineering tool that modifies a cell’s genome to express dCas9-KRAB and a single guide RNA to epigenetically downregulate target genes. We aim to utilize the dCas9-KRAB system to alter the differentiation potential of human adipose-derived mesenchymal stem cells by downregulating the transforming growth factor β receptors TGFBR1 and TGFBR2. We were able to attain up to 55% downregulation of target genes. However, past results on other projects were able to attain higher downregulation. Here, we diagnose possible causes for this project’s suboptimal downregulation as well as how to conduct future work. iii TABLE OF CONTENTS ABSTRACT .................................................................................................................... ii INTRODUCTION .......................................................................................................... 1 BACKGROUND ............................................................................................................ 5 METHODS ................................................................................................................... 10 RESULTS ..................................................................................................................... 15 DISCUSSION ............................................................................................................... 21 ACKNOWLEDGMENT............................................................................................... 27 REFERENCES ............................................................................................................. 28 1 INTRODUCTION About 90% of spinal surgeries are due to degenerative disc disease (DDD), a disorder affecting the intervertebral disc (IVD) [1]. DDD is characterized by several symptoms, including herniation, spinal stenosis, and degenerative spondylolisthesis [1]. The IVD comprises three distinct tissues: the annulus fibrosus, the nucleus pulposus, and the cartilaginous endplate. Current orthopedic interventions currently cannot address DDD with solutions for IVD correction and healing directly. Instead, DDD is treated through spinal fusion that literally fuses the adjacent vertebrae together with a bony union to inhibit motion via bone grafting or metal plates, or artificial disc replacements (ADRs) which are orthopedic implants usually composed of two articulating metal or polymeric discs designed to restore motion to the spine. Spinal fusion suffers from complications such as adjacent vertebral segment degeneration and adjacent vertebral disease. ADRs, on the other hand, have other drawbacks such as the traditional complications of artificial implants, e.g. osteolysis, inflammation, and device degradation, as well as spontaneous vertebral fusion of the adjacent segments [2] [3] [4]. Additionally, both interventions do not restore the natural mechanical function of the IVD. These issues have prompted researchers to seek alternative biological solutions. One successful biological solution is an allograft, or tissue donation, of an IVD into a patient that is able to restore disc function and mobility [5]. This prompted global investigation into tissue engineered strategies for replicating the IVD. Bowles et al. successfully replicated an IVD composed of both the annulus fibrosus and nucleus pulposus in a rat model by taking specific biomaterials and arraying them in such a way as to imitate the IVD [6]. However, this IVD had unanatomical absolute tissue 2 demarcations between the annulus fibrosus and nucleus pulposus boundaries instead of the natural gradients found in the native IVD. As these tissues gradients are essential contributors to the unique mechanical properties of the IVD, this tissue structure-function relationship must be preserved [7]. The next step in tissue engineering IVDs is to produce methods for better replicating the characteristic anatomical gradients of the native IVD. The primary barrier toward generating tissue gradients is the lack of laboratory understanding of how to generate these cell gradients in engineered tissue [8]. In vitro fabrication is not yet capable of generating the aforementioned gradients. One technique would be to physically place stem cells into gradients on a biomaterials scaffold and dose them with various bioactive agents (e.g., cytokines and chemokines) to prompt differentiation to cartilage-like tissues with these preserved cell gradients [9]. However, because selective chondrogenesis (differentiation of stem cells into cartilage) and selective osteogenesis (differentiation of stem cells into bone) each require interfering factors, it is practically impossible to treat a common stem cell area with a single growth factor mixture and elicit two different tissue types as a gradient (e.g., both cartilage and bone co-localized) [10]. Spatially selective cell manipulations must be developed to achieve such a goal. This goal could be facilitated by genetic engineering techniques to alter spatially selective differentiation potential of stem cells selectively via down-regulation and/or upregulation of specific cell surface receptors and then sorting them onto a biomaterial scaffold in specific gradients by using visual markers such as fluorescent proteins (e.g., GFP). The dCas9-KRAB system is a lentiviral–based genetic engineering technique that utilizes CRISPR (clustered regularly interspaced short palindromic repeats) for epigenetic 3 modifications [11]. CRISPR is able to precisely target specific portions of DNA within cells to modify their genome. CRISPR functions via single guide RNAs (sgRNAs) that bind to CRISPR-associated protein 9 (Cas9) to cut out portions of DNA that complement the sgRNA. This produces an elegant tool to select and modify small pieces of a cell’s genome. Alternatively, the dCas9-KRAB system utilizes deactivated Cas9 (dCas9) instead of Cas9, that instead of cutting out DNA at the genomic level, epigenetically down-regulates the gene by blocking that gene via an attached KRAB protein (Figure 1). The dCas9-KRAB system can be used to alter stem cell differentiation potential by introducing genes for dCas9, KRAB, and the sgRNA to epigenetically modify the gene expression levels or their respective encoded proteins within stem cells [11]. GFP can also be genetically introduced to check for successful integration of the lentiviral plasmid Lentivirus with dCas9KRAB Plasmid and sgRNA dCas9 Cell membrane Polybrene KRAB sgRNA Promoter Figure 1. Diagram of the dCas9-KRAB system. Lentivirus penetrates cell membrane and nucleus, resulting in the cell producing dCas9-KRAB, which epigenetically blocks adjacent DNA from being expressed. Adapted from www.fshsociety.com [40] and www.the-scientist.com [38]. 4 into the cell’s genome. This system should theoretically be able to knock down specific genes associated with chondrogenesis. The targets for gene down-regulation in this project are the two cellular TGF-β receptors which are powerful signal partners often utilized in chondrogenesis by mesenchymal stem cells (specific stem cells found in fat, bone, etc.) via dosing with cytokine TGF-β [12]. Chondrogenesis in human adipose (fat)-derived stem cells (hASCs) has been shown to occur when exposed to TGF-β and BMP-6 while the absence of TGFβ leads to osteogenesis [13]. Down-regulation of the TGF-β receptors via the dCas9KRAB system should therefore alter their differentiation potential such that when they are exposed to BMP-6 and TGF-β, they undergo osteogenesis instead of chondrogenesis. In turn, hASCs could be genetically edited using the dCas9-KRAB system. Then, unedited and edited cells placed together on a scaffold with specific spatial gradients between the desired cell types that are then dosed with a cocktail of TGF-beta and BMP6 should lead to generation of a tissue mimic with specific gradients between bone and cartilage. This achievement would be important for the creation of a new tissueengineered IVD as an alternative to orthopedic implants for patients with DDD, providing a more functional, personalized implant. 5 BACKGROUND1 Back pain is a leading cause of disability in the United States that cost $100 billion in 2006 [14]. A major cause of back pain is degenerative disc disease (DDD). Early symptoms of DDD can be found in up to 40% of people under 30, and up to 90% of people over 50 [15]. DDD affects the intervertebral discs (IVDs) by literally degenerating them. The IVD is a cartilaginous organ with the gelatinous nucleus pulposus at the center and the stiff annulus fibrosus on the periphery, both sandwiched between cartilaginous vertebral endplates that connect the IVD to the adjacent bony vertebrae. The IVD is mechanically similar to a tire, where the annulus fibrosus is the rubber and the nucleus pulposus is the air. The nucleus pulposus absorbs the IVD’s axial load due to its composition of proteoglycans and type II collagen that allow it to retain water (i.e. keeps air in the tire). The annulus fibrosus is composed of type I collagen, making it stiff so that it provides structure to the IVD. The vertebral endplates allow the IVD to integrate into the surrounding vertebrae due to its tissue gradients that transition from the cartilage of the IVD into the bone of the vertebrae [15]. Typically, DDD is observed when the nucleus pulposus can no longer retain water, meaning the IVD can no longer handle normal loads. This leads to fissures in the nucleus pulposus that can extend to the annulus fibrosus that eventually leads to loss of mechanical function that can lead to destruction of the endplates and vertebrae, causing intense pain [15]. 1 This section is intended for readers unfamiliar with back pain, tissue engineering, and CRISPR 6 The standard surgical intervention is spinal fusion, where the vertebrae surrounding the affected IVD are fused together with a bony graft, reducing pain and degeneration by stopping motion [2] [1]. While this is effective at reducing pain and slowing degeneration, it has the obvious drawback of reducing spinal mobility. This reduction in motion also shifts the load normally handled by the degenerated IVD to the neighboring discs possibly leading to adjacent segment disease. This means the neighboring discs will start to exhibit similar symptoms as the degenerated disc [2]. The alternative to spinal fusion is replacing the degenerated disc with an artificial disc replacement (ADR) that retains the mobility of the replaced IVD [3]. However, ADRs have not overtaken spinal fusion as the gold standard for many reasons, one of those being that they can cause the adjacent vertebral segments to fuse spontaneously together, negating the primary benefit of ADRs [3]. Other drawbacks include the traditional disadvantages of orthopedics such as device degradation, inflammation, etc. [4]. Ideally, these interventions would more perfectly replicate the biomechanical environment of the IVD, including hardness and mobility, while also integrating into the native environment. While these techniques have shown promise, new strategies that overcome these shortcomings are necessary. The ideal solution is essentially biological because what the patient really needs is a cure that completely replicates a native human IVD. This suggests techniques from tissue engineering. This is because orthopedic implants have difficulty imitating the native biological environment, but theoretically tissue engineering can. Tissue engineering applies the principles of engineering and the life sciences towards developing biological substitutes that restore, maintain, or improve tissue/organ function [16]. In 7 2015, there were several exciting instances of tissue engineering that ranged from drug testing to whole organ engineering. Specifically for DDD, there was a clinical trial from Berlemann et al. that used a product called NuCore® that created a biomimetic protein polymer that mimics the nucleus pulposus that successfully reduced back pain in patients [15]. Another tissue engineering technique is to modulate cellular expression to create desired tissues. Scientists do this by manipulating the genome, the genes for any given organism that can be found in each of its cells. However, part of cell differentiation results from selective gene expression that is part of the epigenome. This selective expression can be used to distinguish one cell from the next. The epigenome refers to the chemical modifications to DNA that determine which sequence of genes are turned on or off for any given cell type. The set of genes expressed determine a cell’s phenotype, which is essentially the qualities of any given cell from neurons to bone. Histone modifications play a critical role in determining which genes are activated [17]. DNA is wrapped around histones, which are disc-like proteins, to make gene regions inaccessible to expression. Acetylation (addition of an acetyl group) or methylation (addition of a methyl group) determines whether a piece of DNA wraps itself around a histone. Acetylation of the positively charged ε-lysine residue on histones neutralizes its charge, weakening histone/DNA interactions [18]. Histone lysine and arginine residues can also be methylated, resulting in either expression or suppression of the associated DNA [19]. DNA itself can be methylated at the promoter, wrapping it around histones and turning off the associated gene [20] [21]. 8 Cis-regulatory elements are typically found near genes and play a major role in controlling genetic expression [22]. These elements are the promoters, enhancers, silencers, and insulators. Promoters encourage transcription of the gene into RNA, enhancers increase the likelihood of transcription, silencers bind transcription factors called repressors that prevent transcription, and insulators stop the interaction between enhancers and promoters [22]. The location of cis-regulatory elements is determined by DNase I-seq [23]. This is accomplished by exposing DNA to DNase I, a protein that degrades DNA. If the DNA degrades quickly in a particular location, then it might contain a cis-regulatory element. All of these factors have a significant impact on genetic engineering. There have been many tools that modulate genetic expression, such as viral transduction where a virus delivers a plasmid (circular piece of human-made DNA) into a cell’s DNA. There are also zinc-finger nucleases and TALENS, proteins engineers can use to modulate genetic expression [24]. These proteins are capable of binding to DNA and are normally attached to effector proteins that confer some sort of functionality such as cutting the DNA (nucleases) or making epigenetic edits through acetylation or methylation of DNA [24]. These tools have been deployed to treat diseases like hemophilia and sickle-cell disease [24]. One of the most exciting developments in genetic engineering is called CRISPR, a system that is more easily programmed than ZFNs and TALENs [25]. All of these techniques show how important genetic engineering is becoming, and thus how important it is for them to be developed further. CRISPR, which stands for Clustered Regularly Interspaced Palindromic Repeats, is a relatively recent technology first formalized as a genetic engineering tool in 2012 9 [25]. CRISPR technically refers to a series of genetic motifs that Ishino et al. observed in E. Coli in 1987 and were eventually discovered in many different species of both bacteria and archaea [25]. These motifs, or CRISPRs, were investigated for many years before the mechanism by which they were introduced was fully discovered [25]. These sequences were introduced by the bacteria itself by proteins called CRISPR Associated Proteins, or Cas proteins, that are able to cut DNA at the genomic level. This was done in order to remove DNA introduced by viruses that forced the bacteria to produce more virus [25]. If the bacteria survived, however, it could “remember” the virus’s genetic code through a single guide RNA that binds to Cas proteins, directing them to the viral insertion site. Essentially, CRISPR originated as the bacterial immune system. Scientists expanded CRISPR’s functionality to epigenome engineering by taking Cas9 (one of several Cas proteins) and deactivating its nuclease activity, creating deactivated Cas9 (dCas9). This dCas9 has the ability to bind to DNA, but is unable to cut. Scientists then attached a protein called Krüppel associated box, or KRAB, to the dCas9 [11]. KRAB methylates DNA which then wraps around the histone, closing it off. This means that scientists and engineers can now “turn off” specific genes if they target the promoter of those genes with the dCas9-KRAB system. 10 METHODS A. Overall Experimental Design The overall design of this project sought to create sgRNAs to target the cellular promoter of either TGFBR1 or TGFBR2 that were cloned into the pLV hU6-sgRNA hUbC-dCas9-KRAB-T2a-GFP plasmid. The plasmid pLV hU6-sgRNA hUbC-dCas9KRAB-T2a-GFP was a gift from Charles Gersbach (Addgene plasmid # 71237) [26], herein referred to as dCas9-KRAB. This plasmid along with the packaging plasmids psPAX2 and pMD2.G were transfected into human embryonic kidney (HEK 293T) cells in culture to produce lentivirus. We then transduced hASCs with said lentivirus. Real time quantitative PCR (RT-qPCR) measured mRNA concentrations of TGFBR1 and TGFBR2. The results were plotted against DNase hypersensitivity data from the UCSC (University of Southern California, Santa Cruz) Genome Browser (https://genome.ucsc.edu/). B. sgRNAs Design The sgRNAs were designed by taking the DNA transcripts of TGFBR1 and TGFBR2 from the UCSC genome browser and plugging them into the CRISPR design website from MIT (crispr.mit.edu). The promoter for each gene was screened by taking 1000 base pairs upstream of the start codon in 250 base pair segments. Guides were selected based on their off-target effects score which is given on a scale from 0 to 100, where 100 means the lowest number of off-target effects and 0 means the largest number. One guide per 250 base pair segment was chosen. Forward and reverse primers for DNA annealing of each guide were designed and submitted to the University of Utah Health Sciences (HSC) DNA/Peptide Synthesis Core (Table 1). 11 Table I: TGF-β Receptor sgRNAs Target Name TGFBR1 g1 TGFBR1 g2 TGFBR1 g3 TGFBR1 g4 TGFBR2 g1 TGFBR2 g2 TGFBR2 g3 TGFBR2 g4 Target Sequence GGAGCGTCT CGCAGTAAAT T TCAGGATCTG GGTCTGACG C ACCAGATCTT CCGCGCCTA G CAGCCGCGA GCGCCGGTT TC TACCCTCTCA TGTGCAAACG CACCACTATC ACTTCGTGAT GTGCTCGCG ACTCAATAGA T AGTCGGCCA AAGCTCTCG GA Length Forward Primer Reverse Primer 20 CACCGggagcgtctcgcag taaatt AAACaatttactgcgagacgc tccC 20 CACCGtcaggatctgggtct gacgc AAACgcgtcagacccagatc ctgaC 20 CACCGaccagatcttccgc gcctag AAACctaggcgcggaagatc tggtC 20 CACCGcagccgcgagcgc cggtttc AAACgaaaccggcgctcgcg gctC CACCGtaccctctcatgtgc aaacg CACCGcaccactatcacttc gtgat AAACcgtttgcacatgagagg gtaC AAACatcacgaagtgatagtg gtgC 20 CACCGgtgctcgcgactca atagat AAACatctattgagtcgcgag cacC 20 CACCGagtcggccaaagct ctcga AAACtccgagagctttggccg actC 20 20 Target sequences for sgRNAs and primers ordered from HSC cores that were annealed and cloned into the dCas9-KRAB plasmid. C. sgRNA Cloning into pLV hU6-sgRNA-dCas9-KRAB-T2a-GFP plasmid (Figure 2) The primers from the DNA/Peptide Synthesis Core for each guide were annealed together and phosphorylated. The dCas9-KRAB plasmid was digested at the BsmBI sites, and the product was purified in a Qiagen PCR Purification Kit. The annealed primers and digested dCas9-KRAB plasmid were ligated together in one final PCR reaction before being transformed into Stbl3 bacteria for plasmid production. The Stbl3 bacteria were grown up overnight before having the plasmids extracted and two samples of each guide concentrated with a Qiagen Miniprep Kit and sent to the HSC DNA Sequencing Core. Half of the bacteria were aliquoted in 25% glycerol stocks and stored in a -80 ̊ C freezer. 12 Figure 2. The pLV hU6-sgRNA hUbC-dCas9-KRAB-T2a-GFP (Addgene plasmid # 71237) plasmid. It is 14,981 base pairs long and has BsmBI sites for cloning in the desired sgRNA. The plasmid functions off of the hUbC promoter when inserted into the genome, and expresses dCas9, KRAB, and the sgRNA that epigenetically blocks the gene of interest. GFP fluoresces to indicate successful transduction. Plasmids positive for the sgRNAs were multiplied with Qiagen Midiprep or Maxiprep Kits using the Stbl3 bacteria made from glycerol stocks. D. Lentivirus Production One six-well plate (Thermo-Fisher Scientific) was plated with 3.6x106 total HEK293T cells (plated at 600,000 cells per well) and then transfected with each sgRNA in its associated dCas9-KRAB plasmid. One plate per guide was transfected plus two plates for the non-target plasmid. This totaled nine cell-lines transfected. Media was 13 replaced every day at the same time. The first media change was not collected, but the second and third media changes were. This media was concentrated to 100X by adding one volume of Lenti-X™ Concentrator to three volumes of the collected media and incubated for at least 30 minutes and up to 3 days at 4°C. This was centrifuged at 1500 x g for 45 minutes at 4°C. The supernatant was removed and the virus pellet was resuspended in 240 µL of DMEM to make a 100X stock of the original solution that was stored at -80°C. E. Lentiviral Transduction Immortalized hASCs (ATCC) were plated at 40,000 cells per well for each guide plus the non-target. Cells were transduced by 20X lentivirus stock for 24 hours. Cells were run through fluorescence activated cell sorting (FACS) by the HSC Flow Cytometry core to sort positive expression based on GFP expression. GFP positive cells from each cell line were run through RT-qPCR by the HSC Genomics Core with TaqMan® primers for TGFBR1 or TGFBR2 (ThermoFisher Scientific). RT-qPCR was run on three samples from each cell line. The results were processed by calculating the mean and standard deviation of down-regulation for each cell line. Each guide was then compared against the up-regulation of either TGFBR1 or TGFBR2. A second batch of hASCs were targeted with TGFBR1 downregulation and then dosed with BMP-6 for three days before analyzed by RT-qPCR. This was done to verify results from Hennig et al. that TGFBR1 mRNA was not produced in hASCs unless exposed to BMP-6 [10]. F. DNase Hypersensitivity Trendline DNase hypersensitivity, a measure of how open DNA is based on how quickly it is degraded by DNase I, was compared against the RT-qPCR results because DNase 14 hypersensitivity is linked to the presence of promoters [27], which the dCas9-KRAB system can accurately target [26]. RT-qPCR results were compared to others run by our lab that analyzed guides for hASC cell lines targeting TNFR1, IL1R1, GREM1, GREM2, and NOG. Based on a Spearman’s rho corellation calculated from these experiments, we were able to predict the correlation between DNase hypersensitivity and the likelihood of down-regulation of a gene using a specific sgRNA. Data for DNase Hypersensitivity was attained from Replicates 1 and 2 on the UCSC Genome browser. This was done over a 1020 base pair range, with the 20 base pair sgRNA in the middle and 500 base pairs on either side. 15 RESULTS The sgRNAs (Table 1) were successfully cloned into the dCas9-KRAB plasmid (Figure 2). This was checked by genomic sequencing which confirmed that at least one sample per guide was successfully integrated into the dCas9-KRAB plasmid. Lentiviral production and subsequent transductions were also successful as indicated by the expression of GFP verified with a fluorescent microscopy (Figure 3). Once successful transduction was confirmed, the cells were sorted via FACS to create a population of cells that had been successfully transduced. All of the guides together had an average transduction efficiency of 32.2%, but also had a standard deviation of 17.5%, when not taking gating efficiency into account (Figure 4). When Figure 3. hASCs fluorescing with GFP post-transduction with dCas9-KRAB system. 20x magnification. 16 considering gating efficiency which ensured a monoclonal population, the average sorting efficiency becomes 17.1% with a standard deviation of 13.6%. This variability in gating efficiencies ultimately resulted in a wide range of harvested hASCs ranging from 48,000 to 200,000 cells from 25 cm2 flasks. Despite this wide range, we were still able to plate these cells into 25 cm2 flasks (ThermoFisher Scientific) and 9.6 cm2 wells (ThermoFisher Scientific). The cells were then grown up to 75 cm2 flasks (ThermoFisher Scientific) and frozen down then plated for RT-qPCR. Since it has been shown by Hennig et al. [10] that TGFBR1 upregulation is dependent upon BMP-6 dosing, we made two sets of all the TGFBR1 targeted cell lines for RT-qPCR: one that would be dosed with 10 ng/mL BMP-6 and one that would not. Transduction Efficiency (%) Transduction Efficiency 80 70 60 50 40 30 20 10 0 Gene/Guide Figure 4. Viral transduction efficiency measured using two genes at four sgRNAs per gene and one non-target stem cell line (i.e. model dCas9-KRAB that does not downregulate any gene), with each bar having n = 1. All of the guides together had an average stem cell transduction efficiency of 32.2% (standard deviation of 17.5%). 17 There was no statistically significant difference between the two groups after three days of exposure with and without BMP-6 (Figure 5). For all groups not dosed with BMP-6, there were three guides that had significant downregulation (Figure 6). Two guides (g1, g3) for TGFBR1 and one guide for TGFBR2 (g4) were significantly downregulated, but two guides (g1, g3) for TGFBR2 were significantly upregulated instead of downregulated. We attained up to 55% downregulation in TGFBR1 and 40% in TGFBR2. However, we also got up to 50% upregulation in TGFBR2. We were also unable to perform RT-qPCR on TGFBR2 g2. Based on DNase hypersensitivity values obtained from the UCSC Genome Browser, it would be predicted that the guides with the best downregulation are TGFBR1 Comparison of TGFBR1 Downregulation With and Without BMP-6 Dosing Fold Change 1.5 1 0.5 0 Naïve NT 1 2 3 4 Guide - + Figure 5. The y-axis is fold change of each guide represented in the x-axis. Naïve were unmodified hASCs. Non-targets were controls that used dCas9-KRAB system but did not target any specific portion of DNA. Blue bars (-) represent cells not dosed with BMP-6, orange bars (+) represent cells that were. * = p < 0.05 of guides to non-target 18 g3, g4 and TGFBR2 g3, g4, which were approximately 50 for TGFBR1 and 150 for TGFBR2 (Table 2)2. For this project combined with other projects using the dCas9-KRAB system in our lab, the correlation between DNase hypersensitivity and fold downregulation based on Spearman’s rho was -0.4377 (p=0.0367), indicating a relatively weak correlation between a decrease in DNase hypersensitivity as fold expression increases (Figure 7). Without TGFBR1 and TGFBR2, Spearman’s rho was -0.7382 (p=0.0011), which is a considerably stronger correlation. With only TGFBR1 and TGFBR2, Spearman’s rho was 0.1429 (p=0.7599), which is very weak, but if the major outlier (one of the TGFBR2 guides) is removed, the Spearman’s rho becomes -0.3714 (p=0.4685), which is 1.8 RT-qPCR Analysis of TGFBR1 and TGFBR2 * * 1.6 Fold change 1.4 1.2 1 0.8 0.6 0.4 * * * 0.2 0 Gene/Guide Figure 6. The y-axis is fold change of each guide represented in the x-axis. Naïve were unmodified hASCs. Non-targets were controls that used dCas9-KRAB system but did not target any specific portion of DNA. * = p < 0.05 of guides to non-target 2 Values are the relative hypersensitivity of the 1020 base pair range to the sensitivity of the entire genome 19 considerably stronger. With all of the values together except the outlier, Spearman’s rho was -0.5596 (p=0.0068), which is much stronger than with the outlier included. This indicates a correlation between DNase hypersensitivity without TGFBR1 and TGFBR2 that decreases significantly when experiments for this project are factored in. Table 2: sgRNA DNase Hypersensitivity Values TGFBR1 g1 TGFBR1 g2 TGFBR1 g3 TGFBR1 g4 TGFBR2 g1 TGFBR2 g2 TGFBR2 g3 TGFBR2 g4 Average/ Rep 1 Std Dev/ Rep 1 Average/ Rep 2 14.451 42.308 77.712 80.404 2.324 23.460 181.295 170.735 14.764 40.952 74.639 72.243 4.014 51.980 210.63 202.239 10.191 19.462 34.885 36.860 2.556 22.450 126.081 97.68 Std Dev/ Rep 2 9.405 17.009 35.798 35.287 2.874 37.451 130.589 122.463 FINAL AVG 12.321 30.885 56.299 58.632 2.440 22.955 153.688 134.2075 Guide DNase hypersensitivity for all guides from UCSC Genome Browser using replicates 1 and 2. Values are the relative hypersensitivity of the 1020 base pair range to the sensitivity of the entire genome 20 DNase correlation with hASCs DNase Hypersensitivity Signal 250 200 TNFR1 IL1R1 150 TGFBR1 100 TGFBR2 GREM1 50 GREM2 NOG 0 0 0.5 1 1.5 2 Fold Change in Gene Expression Figure 7. Dot plot showing correlation between DNase hypersensitivity signal and fold change in gene expression obtained from RT-qPCR. White dots are from previous experiments in our lab that also used the dCas9-KRAB system, red dots are TGFBR1, and blue dots are TGFBR2. Red arrow points to the major outlier in TGFBR2. Spearman’s rho: -0.4377, p<0.05 21 DISCUSSION DDD is a major cause of back pain in the U.S. and a tissue engineered IVD would be a preferable alternative to current orthopedic and surgical solutions that can cause significant discomfort or pain to patients such as limited segment mobility, osteolysis, and inflammation. The formation of a tissue engineered IVD may be accomplished in part by directing cell differentiation via epigenetic engineering tools like the dCas9KRAB system that downregulates target genes via methylation of target sites. We have clearly demonstrated that modulation of the TGF-β receptors using the dCas9-KRAB system is possible, but that it is also subject to certain classes of errors that highlight the subtle complexity of genetic engineering. RT-qPCR results showed both upregulation and downregulation. Fortunately, most of the guides that showed significant change in regulation were downregulated and not upregulated, which indicates that we can modulate the response of TGFBR1 and TGFBR2 towards BMP-6, altering their differentiation potential. Interestingly, despite the results from Hennig et al., which claimed that TGFBR1 gene expression in hASCs is practically nonexistent if not dosed with BMP-6, we obtained no statistically significant difference in gene expression between dosed and undosed groups. Hennig et al. evaluated TGFBR1 expression qualitatively via real time PCR (not RT-qPCR), where they ran PCR on TGFBR1 cDNAs that were loaded onto a 1.5% agarose gel containing ethidium bromide and visualized under UV light. We, however, used a quantitative technique, and since RT-qPCR evaluates relative expression levels and not absolute expression, it may be that trace amounts of TGFBR1 were amplified. It could also be a matter of difference 22 between cell population, where Hennig et al. happened to have a cell population that keeps TGFBR1 constitutively downregulated but our hASCs did not. The upregulation of TGFBR2 g1 and g3 was unexpected based on our lab’s experience with the dCas9-KRAB system as well as available literature. It is not uncommon for dCas9 to be associated with upregulation of genes if connected to specific proteins such as p300 [28], but has not been observed with the dCas9-KRAB system. Experimental error may be to blame; however, there may be a reason for this upregulation that is tangentially related to the system itself. The most likely cause of this is experimental error. This is demonstrated by the consistency of other experiments in our lab (Fig.6), that had a very strong correlation between DNase hypersensitivity and downregulation of target genes. The most likely source of experimental error likely arose post-transduction, since we know that transduction was successful and the guides should have been successfully integrated into the plasmid based on sequencing data. We say this because the cells took several weeks to grow up, during which time the cell phenotype could have changed as expression levels of the lentiviral products may have decreased, as is known to happen with lentiviral transduction [29]. In other words, if we had a higher transduction efficiency, and if the cells had grown up faster (as cell replication rates fall post-transduction due to toxicity of transduction conditions [30]), then this error may not have occurred. However, there is a second possibility, albeit less likely, that could explain this. When plotting DNase hypersensitivity against fold regulation of each guide (Figure 7), it is clear that for most guides our lab has created for other projects as well as for this one, that DNase hypersensitivity is correlated with downregulation, not 23 upregulation. However, DNase hypersensitivity merely refers to open portions of DNA, not necessarily to whether or not those open portions of DNA contain a promoter. It is also known that such regions can also contain enhancers, insulators, and silencers [27] [22] [31] [32]. It is possible (although unlikely) that instead of hitting the promoter or an enhancer, g1 and g3 for TGFBR2 hit a silencer or perhaps an insulator tied to an enhancer. If so, this would explain the upregulation. If this is true, then a new method for singling out possible targets for sgRNAs needs to be created that takes into account offtarget effects, DNase hypersensitivity, and a knowledge of the locations of cis-regulatory elements. This highlights our lack of knowledge of the human genome. Essentially, the only methods currently available to determine cis-regulatory element identity is through computational analysis, largely with ChIP-seq combined with DNase hypersensitivity analysis. Using this, researchers across the biological sciences have been able to find particular motifs capable of distinguishing different cis-regulatory elements, but this approach often fails due to the similarity of motifs, not to mention the gap between different fields (e.g. plants versus human genomes) [23]. Essentially, the only methods currently available to avoid targeting cis-regulatory elements are expensive and labor intensive. The only real way to identify these elements is to wait for other researchers to identify them and add them to the ENCODE project which has been characterizing the genome since 2003 [33]. As the ENCODE project expands and more ChIP-seq data is obtained, cisregulatory factors will be targeted more reliably as their locations are revealed. For this specific project, however, redesigning the guides and conducting RT-qPCR again may 24 improve downregulation. If so, then there is a greater chance that differentiation experiments will be successful. However, we could move forward with differentiation experiments with the sgRNAs we have already designed. This would involve using a minimal common differentiation medium, similar to that used by Li et al. [34] (although Li et al. used bone marrow stromal cells instead of hASCs), to allow for both chondrogenic and osteogenic phenotypes from hASCs. These cells will be put into a 3D culture, such as pellet or agarose bead, and dosed with 10 ng/mL of TGF-β and BMP-6 [12] [13]. The 3D culture is necessary for chondrogenic differentiation [10] [35]. Assuming we obtain optimal differentiation results, the cells would be candidates for laminar flow sorting, a microfluidic technique that can sort cells into gradients onto a scaffold based on cell type (Figure 8). These cell lines could then be put into our experiment with laminar flow sorting, a project underway in our lab. If successful, this could lead to the creation of a tissue-engineered IVD that is seeded with a variety of Figure 8. Laminar flow printing of stem cells onto a collagen scaffold demonstrates the ability to produce cellular gradients based on red-green transition (left to right). Courtesy: David Ede, University of Utah, Back Pain and Engineered Therapeutics Laboratory. 25 genetically engineered cell lines that restores IVD function. This approach could then extend to other tissue engineering applications that require tissue gradients, such as the pancreas, the knee, and tendon-bone interfaces. The dCas9-KRAB system allows us to quickly, accurately, and relatively cheaply downregulate any gene with a 20 base pair sgRNA. Lentivirus allows us effectively to introduce dCas9, KRAB, the sgRNA, and GFP with the hU6 promoter consistently at a single site. The dCas9-KRAB system is highly accurate with limited off-target effects [26]. When compared with other systems like TALENs and ZFNs, this system is less expensive and rapidly conducted [36] [37]. Two major drawbacks to our approach are the time and monetary investments. Cell culture, especially of stem cells, is a lengthy process. The hASCs had a doubling time of approximately one week compared to HEK293T cells that double within one or two days. The hASCs are also expensive, as are their media components, not to mention the lentivirus production. If these cells are taken forward to laminar flow printing, it would likely take one week for the cells to expand to confluency, and at least another month for them to differentiate into appropriate cell types. It is unfeasible to use these techniques for treatments in case of emergency, which means that organ donors and artificial implants will still be necessary in the future. The dCas9-KRAB system shows considerable promise for tissue engineering, particularly for the creation of whole organ implants. Our lab has already had considerable success with the system, and the inferior results of this paper are an exception to that success. Although there is a chance that utilization of the system can benefit from development of the resources available for guide design, such as the location 26 of cis-regulatory elements, it is certainly robust enough in its current state for developing novel therapeutic techniques. 27 ACKNOWLEDGMENT P. J. Wissler thanks Dr. Robert Bowles, Niloofar Farhang, Joshua Stover, Bryton Davis, and David Ede for their help and advice on this project. Oligonucleotides were synthesized by the DNA/Peptide Facility, part of the Health Sciences Center Cores at the University of Utah. Sequencing was performed at the DNA Sequencing Core Facility, University of Utah. This work was supported by the University of Utah Flow Cytometry Facility in addition to the National Cancer Institute through Award Number 5P30CA042014-24. RT-qPCR was performed by the Genomics Core Facility, a part of the Health Sciences Cores at the University of Utah. This work was supported by funding from the Undergraduate Research Opportunities Program at the University of Utah awarded to Paul J. Wissler. 28 REFERENCES [1] V. Palepu et al., "Biomechanics of Disc Degeneration," Adv Orthop, vol. 2012, no. 17, 2012. [2] A. S. Hilibrand et al., "Adjacent segment degeneration and adjacent segment disease: the consequences of spinal fusion?," Spine J., vol. 4, no. 6, pp. S190-194, 2004. [3] Y. Robinson et al., "Spine imaging after lumbar disc replacement: pitfalls and current recommendations," Patient Saf Surg, vol. 3, no. 15, 2009. [4] J. Reeks et al., "Materials and Their Failure Mechanisms in Total Disc Replacement," Lubricants, vol. 3, pp. 346-364, 2015. [5] D. Ruan, "Intervertebral disc transplantation in the treatment of degenerative spine disease: A preliminary study," Lancet, vol. 369, pp. 993-999, 2007. [6] R. D. Bowles et al., "Tissue Engineered intervertebral discs produce new matrix, maintain disc height, and restore biomechanical function to the rodent spine," PNDAS, vol. 108, no. 32, pp. 13106-13111, 2011. [7] K. Migacz et al., "Gradient composite materials for artifical intervertebral discs," Acta Bioeng Biomech, vol. 16, no. 3, 2014. [8] N. H. Dormer et al., "Emerging Techniques in Stratified Designs and Continuous Gradients for Tissue Engineering of Interfaces," Ann Biomed Eng., vol. 38, no. 6, pp. 2121-2141, 2010. 29 [9] H. J. Kim, "Chondrogenesis using mesenchymal stem cells and PCL scaffolds," J Biomed Mater Res A., vol. 92, no. 2, pp. 659-66, 2010. [10] T. Hennig et al., "Reduced Chondrogenic Potential of Adipose Tissue Derived Stromal Cells Correlates With an Altered TGFβ Receptor and BMP Profile and Is Overcome by BMP-6," J. Cell. Physiol., vol. 211, no. 3, pp. 682-691, 2006. [11] L. A. Gilbert et al., "CRISPR-Mediated Modular RNA-Guided Regulation of Transcription in Eukaryotes," Cell, vol. 154, no. 2, pp. 442-451, 2013. [12] S. Maeda et al., "Endogenous TGF-Beta signaling suppresses maturation of osteoblastic mesenchymal cells," EMBO J., vol. 23, no. 3, pp. 552-563, 2004. [13] X. Zhang et al., "The Roles of Bone Morphogenetic Proteins and Their Signaling in the Osteogenesis of Adipose-Derived Stem Cells," Tissue Eng Part B Rev., vol. 20, no. 1, pp. 84-92, 2014. [14] J. N. Katz, "Lumbar disc disorders and low-back pain: socioeconomic factors and consequences [review]," J Bone Joint Surg Am., vol. 88, no. suppl 2, pp. 21-24, 2006. [15] B. Pennicooke et al., "Biological Treatment Approaches for Degenerative Disc Disease: A Review of Clinical Trials and Future Directions," Cureus, vol. 8, no. 11, p. e892, 2016. [16] H. Wobma et al., "Tissue Engineering and Regenerative Medicine 2015: A Year in Review," Tissue Eng Part B Rev, vol. 22, no. 2, pp. 101-113, 2016. [17] M. Park et al., "The epigenome: the next substrate for engineering," Genome Biol., vol. 17, no. 183, 2016. 30 [18] A. J. Bannister et al., "Regulation of chromatin by histone modifications," Cell Res., vol. 21, pp. 381-395, 2011. [19] P. Volkel et al., "The control of histone lysine methylation in epigenetic regulation.," Biochimie, vol. 89, no. 1, pp. 1-20, 2007. [20] A. Ma et al., "Targeted gene suppression by inducing de novo DNA methylation in the gene promoter," Epigenetics Chromatin, vol. 7, no. 20, 2014. [21] D. H. K. Lim et al., "DNA methylation: a form of epigenetic control of gene expression," TOG, vol. 12, no. 1, pp. 37-42, 2010. [22] P. Kolovos et al., "Enhancers and silencers: an integrated and simple model for their function," Epigenetics Chromatin, vol. 5, no. 1, 2012. [23] A. M. Sullivan et al., "DNase I hypersensitivity mapping, genomic footprinting, and transcription factor networks in plants," Curr Opin Plant Biol, Vols. 3-4, pp. 40-47, 2015. [24] T. Gaj et al., "ZFN, TALEN, and CRISPR/Cas-based methods for genome engineering," Trends Biotechnol, vol. 31, no. 7, pp. 397-405, 2013. [25] J. A. Doudna et al., "The new frontier of genome engineering with CRISPR-Cas9," Science, vol. 346, no. 6213, pp. 1258096-1 - 1258096-9, 2014. [26] P. I. Thakore et al., "Highly Specific Epigenome Editing by CRISPR/Cas9 Repressors for Silencing of Distal Regulatory Elements," Nat Methods., vol. 12, no. 12, pp. 1143-1149, 2015. 31 [27] T. S. Furey, "ChIP-seq and Beyond: new and improved methodologies to detect and characterize protein-DNA interactions," Nat Rev Genet., vol. 13, no. 12, pp. 840-852, 2012. [28] I. B. Hilton et al., "Epigenome editing by a CRISPR–Cas9-based acetyltransferase activates genes from promoters and enhancers," Nat. Biotechnol., vol. 33, pp. 510517, 2015. [29] Y. Mao et al., "Lentiviral Vectors Mediate Long-Term and High Efficiency Transgene Expression in HEK 293T cells," Int J Med Sci., vol. 12, no. 5, pp. 407415, 2015. [30] P. Lin et al., "Polybrene Inhibits Human Mesenchymal Stem Cell Proliferation during Lentiviral Transduction," PLoS ONE, vol. 6, no. 8, 2011. [31] P. N. Cockerill, "Structure and function of active chromatin and DNAse I hypersensitive sites," FEBS J., vol. 278, no. 13, pp. 2182-2210, 2011. [32] A. Baniahmad, "Modular structure of a chicken lysozome silencer: Involvement of an unusual thyroid hormone receptor binding site," Cell, vol. 61, no. 3, pp. 505514, 1990. [33] H. Qu et al., "A Brief Review on the Human Encyclopedia of DNA Elements (ENCODE) Project," Genomics Proteomics Bioinformatics, vol. 11, no. 3, pp. 135141, 2013. [34] J. Li et al., "A Minimal Common Osteochondrocytic Differentiation Medium for the Osteogenic and Chondrogenic Differentiation of Bone Marrow Stromal Cells in 32 the Construction of Osteochondral Graft," Tissue Eng Part A, vol. 15, no. 9, pp. 2481-2490, 2009. [35] A. Winter et al., "Cartilage-like gene expression in differentiated human stem cell spheroids: A comparison of bone marrow-derived and adipose tissue-derived stromal cells," Arthritis Rheumatol, vol. 48, no. 2, pp. 418-429, 2003. [36] L. Yi et al., "CRISPR-Cas9 therapeutics in cancer: promising strategies and present challenges," Biochim. Biophys. Acta, vol. 1866, no. 2, pp. 197-207, 2016. [37] T. Sato et al., "Genome Editing in Mouse Spermatogonial Stem Cell Lines Using TALEN and Double-Nicking CRISPR/Cas9," Stem Cell Reports, vol. 5, no. 1, pp. 75-82, 2015. [38] C. Storrs, "A CRISPR Fore-Cas-t," The Scientist, 1 March 2014. [Online]. Available: http:/www.the-scientist.com/?articles.view/articleNo/39239/title/ACRISPR-Fore-Cas-t/. [Accessed 8 December 2016]. [39] L. D. Moore et al., "DNA Methylation and its Basic Function," Neuropsychopharmacology, vol. 38, no. 1, pp. 23-38, 2013. [40] P. Jones et al., "Biology 101 for understanding the CRISPR research," FSH Society, [Online]. Available: https://www.fshsociety.org/2015/12/qa-on-crisprresearch/. [Accessed 8 December 2016].