Humanized animal models to quickly address clinical relevance of disease gene alleles
RediMODEL - Humanized C. elegans strains for phenotype discovery of clinical variants
RediMODEL System is a fast system for robust phenotyping of genotype differences. View the link below to see how RediMODEL kits can speed your research program.
Showcase: Zebrafish Verified-Clutch Service - Two phased project for getting GFP and loxP insertions at native loci
Zebrafish are a powerful organism for drug discovery and are starting to secure a role in personalized medicine discovery.1 Understanding Knudra Transgenics uses a CRISPR approach using Cas9 complexes and long-homology arms 2-4 to create clutches of embryos testing positive for a high degree of mutagenesis.
Verified Clutch Service. The Verified Clutch Service is a zebrafish project done in two phases. In the first phase, we screen a set of CRISPR sgRNAs to identify efficient cutters of the genome. Next, in the second phase, we use the most efficient cutter along with a donor homology specific to the cut site. By utilizing a verified cutter and a donor homology built specifically for that sgRNA we see our success for germline integrations increase significantly and GFP insertions in zebrafish are now becoming routine in the clutches supplied to our researcher customers. The collaboration of having us design the edit, make the injection mix, inject, and verify editing in a portion of the injected clutch is working well for most of our researchers. They can then use their facilities (at a lower cost) to identify germline edits.
Floxed Conditionals. We are currently exploring the creation of floxed conditional alleles in fish. We are using the two-phased approach. In the standard first phase approach, we screen a set of CRISPR sgRNAs to find out which ones are efficient cutters of the genome. Next, in a modified second phase approach, we use a pair of good cutters, each with high efficiency for cutting pre and post of the start and stop codons of a gene. We are currently screening this new type of projects to see if this can get the bi-allele conversion on the same strand. It may be that getting double loxP insertion on one chromosome is a rare and elusive event. Nevertheless, we will know that for whichever line has a single loxP site inserted, the other site is just a simple phase II repeat effort to floxed allele transgenesis.
For ordering information click here: http://www.knudra.com/fish-crispr
2. Kawahara A, Hisano Y, Ota S, Taimatsu K. Site-Specific Integration of Exogenous Genes Using Genome Editing Technologies in Zebrafish. Int J Mol Sci. 2016 May 13;17(5). pii: E727. doi: 10.3390/ijms17050727.
4. Hisano Y, Sakuma T, Nakade S, Ohga R, Ota S, Okamoto H, Yamamoto T, Kawahara A. Precise in-frame integration of exogenous DNA mediated by CRISPR/Cas9 system in zebrafish. Sci Rep. 2015 Mar 5;5:8841. doi: 10.1038/srep08841.
Showcase: ODN-mediated Precise Deletions in the C. elegans Genome
Now Removing More Than 12,000bp! Knock-outs of genes are a commonly used tool for biologists to understand gene function 1 . Examination of phenotypes when the gene is deleted can reveal insights into what role that gene plays in the organism. Knudra Transgenics has found that a CRISPR approach using Cas9 complexes and short-homology arms 2 can be very effective at creating genomic deletions.
Knudra is pushing the boundary of precision deletion size. The following is an example of a 12.8kb deletion of all of the coding of the mod-5 locus. Using our proprietary algorithms, we selected two genome-targeting sgRNA sites flanking the genomic edit site. The first sgRNA targeted a sequence 100bp upstream from the ATG start codon. The second sgRNA was in the 3’ UTR at 100bp downstream from the TAG stop codon (Figure 1A). This deletion eliminated 100% of the mod-5 coding region. Our donor homology is a Oligo Deoxyribo-Nucleotide (ODN) designed to have 35bp homology arms with 55bp of content between the two arms (Figure 1B). The sgRNAs and ODN were ordered synthetically. We combined these with a co-CRISPR marker 3 (sgRNA and ODN) and Cas9 protein. Eighteen N2 animals were injected. Five days later, we screened for the phenotype caused by the insertion of the co-CRISPR marker mutation. Twenty-four F1 animals were then isolated and allowed to lay progeny. The adults were harvested for PCR to identify deletion mutants. Our three-primer PCR approach would only generate a band of 813bp if the deletion occurred and 614bp in wild-type (Figure 1C). We saw 3 lines that contained the deletion band. We sequenced the deletion and saw that repair had occurred correctly.
Using this method, we have been able to identify deletion mutants from seven of seven client projects and three of four internal research projects (Table 1). The only failed project had a genomic region considerably larger than the others (31.2kb). Failure was likely due to a small screen size deployed for a very rare event. Only 20 to 35 F1 animals are typically screened, so it may be that screening more animals would allow us to isolate very large deletions, but it may be more effective to make a large deletion sequentially in small segments. On average, 46% of the F1s we screened contained a deletion. Intriguingly two projects had oligo-guided precision deletion in all of the the F1 co-CRISPR animals; however, the efficiency of deletion generation does not always directly correlate with deletion size (see Table 1). This may be an indication that sgRNA cutting efficiency is a larger determinant in biallelic deletions. Our average deletion size for the projects that worked is 5002.5bp. The largest deletion we have made is 12,775bp.
We are excited by these results and what it means for the C. elegans research community. Large genomic deletions can now become commonplace and will allow for scientist to gain a greater understanding of biology by being able to make complete loss-of-function alleles.
For ordering information click here: http://www.knudra.com/crispr-worm
RediMODEL - Finding Pathogenic Phenotypes in GWAS Genotypes
Synopsis: Many human diseases are caused by mutations in the genome. Recent progress in DNA sequencing has identified massive amounts of rare alleles that appear to be clinically associated with a disease. Yet, definitive proof of pathogenicity remains unknown for many clinical variants. Knudra Transgenics has developed the RediMODEL as a fast system for robust phenotyping of genotype differences. In the RediMODEL platform, individual human mutations are inserted into the animal model genome.
The rapid advance of genome biology reveals a wide variety of genetic differences occur between individuals in a population. With this advance, an alarming gap in knowledge is occurring - the consequence of these genetic differences remains largely unknown. The problem, as restated by a review article author:
"One of the biggest challenges facing us in this new age of genomic medicine is the functional validation of variants identified in exome/whole genome sequencing approaches." - Timothy C. Cox1
How do we determine functional consequences (phenotype) from the array of genetic differences (genotype)? Genome-Wide Association Studies (GWAS) are the data engine used to reveal the genetic differences occurring within human populations. For instance, Mallick et al. performed sequence variant analysis on 300 individuals from 142 diverse backgrounds. They observed 38.2 million Single Nucleotide Polymorphisms (SNPs), short insertion/deletion polymorphisms (indels), and small tandem repeats (STRs) in this small population subset.2 This high level of diversity in the genome complicates the clinician’s ability to find the right therapeutic approach. Because multiple sequence variants can co-segregate through a patient family, the clinician’s challenge is to find ways to accurately determine which genetic factors are causative of the disease. Assigning risk factor status to a co-segregating allele can be problematic. In some cases, an allele association seen in one study can not be replicated in another study on a different set of patients.3 The bottom-line impact of GWAS information is a revelation that a great deal of diversity exist in genotype and there is still much mystery about the phenotypic consequence of suspect alleles. As a result, new technologies are needed to determine the pathogenic consequence of disease-associated gene variation.
Animal models can be revealing of disease conditions, yet careful planning is needed in order to optimize outcomes.1 The recent advent of CRISPR technology is allowing precise animal models to be constructed at a significantly faster pace. Now the basics of life-cycle biology are becoming the main limiting factor. For mice and zebrafish, the rate of model generation can take six months to a year. As a result, establishing a line is done at a significant animal husbandry expense. Model generation in C. elegans is faster due to its faster life cycle. The C. elegans nematode can reach the reproductive adult stage in three days, which is more than 20 times faster than the three and six month maturation times occurring for fish and mice respectively. In just a few weeks, a germline integration line can be established in C. elegans, which is in significant contrast to the year or more it takes to establish a homozygote in mice or fish. This fast life cycle leads to a fast production cycle, so the cost of animal model generation is more than an order of magnitude lower than fish or mice. When an appropriate animal model can be generated to mimic the human disease state in a small model organism such as C. elegans, there will be significant cost savings in identifying pathogenic alleles.
A Homology-driven System. The C. elegans animal model is useful for determining the pathogenic consequence of human disease gene polymorphisms. The C. elegans nematode exhibits a surprisingly high degree of conservation of biology to humans for disease genes. For instance, a set of 30 human disease categories were queried on DisGeNET (a discovery platform integrating information on gene-disease associations)4 for the 10 most associated genes for each disease category. A set of 245 unique human genes were uncovered in diseases ranging from ALS, Adenocarcinoma, and Autism Spectrum Disorder to Bipolar-Schizophrenia, Charcot-Marie-Tooth Disease, and Pemphigus Vulgaris. The 245 genes were queried on the database DRSC Integrative Ortholog Prediction Tool5 for finding gene homologs between species. 187 sequence worm homologs to the human disease genes were uncovered. As a result, in this subset sampling, the homology similarity between human disease genes and C. elegans orthologs calculates to 76.3%. From this finding, It can be inferred that the capacity to create high-throughput screening platforms for pathogenic allele discovery should be applicable for 75% of all the disease genes in the human genome.
Figure 1. RediMODEL: Animal-Model-in-a-Kit
RediMODEL kit. Knudra Transgenics is making humanized C. elegans animal models available in a fast and easy format that any researcher can use (Figure 1). The humanized worm is shipped to the client as a RediMODEL kit. The end user opens the box and follows the simple protocol. Two days later, after incubation and growth, mature animals are ready for experimental analysis. As a key feature, the preparation and handling of C. elegans requires only basic pipetting skills. Mastery of preparation and analysis protocols is quickly achieved by researchers from a wide variety of backgrounds and skill sets. The worms and reagents are transferable to and from the incubation plate with standard pipetting instrumentation. By optimizing the RediMODEL kit to enable liquid handling of an animal model, a significant ease-of-use is established.
Assays on a humanized C. elegans. Liquid handling allows simple integration with a variety of instrumentation. C. elegans is now established for routine analysis in 96 and 384 formats.6 A variety of plate reader formats and high content imaging systems have also been optimized for use on C. elegans.7 Similarly, a variety of microfluidics platforms has been developed to immobilize and manipulate C. elegans.8 One of the most promising microfluidic platforms is now commercially available from NemaMetrix, Inc.9 Using their ScreenChip cartridge system, NemaMetrix has demonstrated that electrophysiology data of pharynx pumping can be recorded with passage of worms through an analysis channel. As each worm enters and is paused in the channel, nearby electrodes pick up the electrical signature of the rhythmic pumping of the pharynx. Multiple cardiac-like parameters are recorded, such as beat frequency, amplitude, component interval changes, and more, resulting in rapid electrophysiological recordings. Additional phenotypic parameters, like internal morphology and fluorescence indicators, can be recorded simultaneously. A solid body of data can be acquired in one afternoon, instead of the many weeks needed for older electrophysiological recording systems.
SLC6A4 example. Easy to perform electrophysiology is very practical for analyzing phenotypes of ion channels. For instance, in the 139 member solute carrier family, the homolog (mod-5) of human serotonin receptor (SLC6A4) was analyzed for its effects on pharynx pumping behavior. Two loss-of-function point mutations (“LOF” alleles) were examined. In prior reports and upon visual examination, the LOF alleles have no gross phenotypic difference from wild-type animals. Yet, consistent with prior results,10-12 each exhibit a strong defect in pumping behavior in the absence of food stimulus when measured on the ScreenChip system. The presence of serotonin in the synapse of the neuromuscular junctions of the pharynx leads to rhythmic pumping behavior. An average of 7-fold increase in pumping rate is observed in two LOF alleles (Figure 2A). The LOF in the native serotonin transporter is resulting in a build up of serotonin at the synapse. Exogenous addition of fluoxetine (a serotonin transporter antagonist) behaves like the null and leads to higher pumping rates (Figure 2B). As a result, serotonin builds up at the synapse, by either genetic loss of transporter, or exogenous addition of serotonin, provides a significant LOF phenotype that is easily detected on a ScreenChip system as increased pumping frequency.
Figure 2. Knock-out and Knock-in data on SLC6A4. Using a ScreenChip system, electrophysiology recording of pharyngeal pumping was performed. A) Comparison of wild-type animals to two null alleles obtained from a C. elegans genetic stock center reveals a significant increase in pumping frequency. B) The addition of 100 ug/ml fluoxetine antagonist (PROZAC) to N2 animals results in increased pumping frequency. C) A gain-of-function “humanized” worm (SLC6A4 line) shows significant loss of pumping. D) The addition of 10 mM serotonin (5-HT) causes restoration of pumping towards wild type in SLC6A4 line. The statistical confidence denotation of ++ is p =0.05 and +++ is p= 0.005.
Gain-of-function mutant. With the loss of function easily and significantly observed, restoration of activity as gain-of-function (GOF) mutagenesis was explored. If restoration of activity can be achieved by adding the human homolog to the genome, the synonymous function between the worm and human homolog becomes verified. Applied to the serotonin transporter, a codon-optimized sequence of the human SLC6A4 gene was added to the genome under a promoter for strong expression in the pharynx (myo-2p::SLC6A4::tbb-2utr). The result is strong overexpression of human serotonin transporter. If active in C. elegans, the transgene is predicted to have a significant GOF phenotype leading to suppression of pharynx pumping. Indeed, when the construct is inserted at single copy using transposon-mediated gene insertion (MosSCI), the resulting animals show a severe pumping defect (Figure 2C). This pumping defect is restored to near wild-type levels by exogenous addition of 5-HT (Figure 2D). With strong GOF phenotype established, it now becomes important to build the proper controls and configurations that will allow demonstration of the extent to which the human gene can rescue for mod-5 function. Specifically, will full restoration of function to wild type levels occur when human SLC6A4 is expressed in replacement of mod-5 at its native locus? So far, these preliminary results show a humanized gene can have significant functional activity when it is inserted at single copy into the genome and driven to high expression in the appropriate tissue.
Disease modeling. With at least 75% of human disease genes as addressable with RediMODEL platforms, a study of ion channel disease genes was performed. 120 human disease genes in neurotransmission were queried for the presence of homologs in C. elegans. 97 genes C. elegans were detected as homologous to the human disease genes (Figure 3). Many of these homologs are already available as LOF alleles (frequently as missense, frameshift or stop). Although over 90% of the genes listed have loss of function, the true null character of the homolog function may be lacking and will require full deletion of the worm homologs coding sequence (“trueDEL” alleles). Alternatively, for about 20% of the genes, the alleles are only maintainable as a heterozygote, which implies full null status is achieved with current alleles. As a result, a range of loss-of-function phenotypes will be available for production as RediMODEL Knock-Out kits.
Figure 3. Genes in Neurological Disease. 97 of 120 neurological genes found to have homologs in worms and distribute to 15 disease categories. Green indicates knock-out (KO) alleles that are currently available for use in RediMODEL kit format.
Human gene polymorphisms. The results with SLC6A4 indicate gene replacement with human sequence can bring restoration of function to the animal model. Functional replacement can be explored for all of the 97 alleles. The SLC6A4 gene has only one main splice variant. Yet, in dbVar database,13 there are 59 polymorphic differences. In the more complex dystrophin gene (DMD), there are 17 splice variants and 3126 sequence variants in dbVar. When all observed sequence differences in the 97 genes are tabulated, the polymorphism complexity reaches 49,669 sequence variants (Figure 4). As a result, there are an average of three different wild-type knock-in splice variant configurations per human disease gene, and each wild-type knock-in may need further knock-in of about 500 types of clinical variants per gene so that clinically-relevant phenotyping can be acquired.
Figure 4. Increased complexity occurring in the Polymorphism Landscape. 97 genes were revealed as genes with homologs to human disease genes. On average, 3 splice variants are needed to cover the splicing variability occurring human genome. For the 97 genes, the numbers of clinical variants are approximately 500 per human disease homolog.
Clinical Variant SNPs. Many polymorphisms observed in the genome have not been correlated for their impact on disease. The ClinVar database14 is a subset of dbVar that attempts to correlate genomic variation with human health. For the 59 dbVar sequence variants in the SLC6A4 gene, there are 48 polymorphisms as SNPs in the ClinVar database. Only one of these is correlated as a risk factor for SLC6A4-associated diseases, which places the remaining 47 alleles in an undetermined category for being neither benign nor pathogenic in status. The Ensembl database 15 provides an even deeper look at the polymorphism space with over 487 variants occurring in the longest isoform of SLC6A4. Bioinformatic algorithms of SIFT16 and PolyPhen17 report 24% of the alleles as suspect. Yet, for the 11 known pathogenic alleles, the bioinformatics approaches are detecting only two alleles as being suspect. Applied to other genes, bioinformatics approaches consistently detect pathogenicity in only a subset of known pathogenic alleles. In the DMD gene, potentially hundreds of alleles fail bioinformatics detection of pathogenicity. The longest isoform of DMD has 8242 SNPs. ClinVar database reveals 384 of DMD alleles are assigned as pathogenic and 6 as benign. This covers 44% of the DMD variants in ClinVar. The remaining 56% of the DMD changes are in the undetermined category, of which, a minority are flagged by bioinformatics as being suspect. Applied to other genes, a set of 15 genes were examined for the percentage of Single Nucleotide Polymorphisms (SNPs) that remain undetermined (Figure 5). An average of 70% of SNP’s observed in the 15 genes remains ambiguous for their benign or pathogenic assignment.
Figure 5. 15 disease genes of broad impact. A representative set of human disease genes were selected for use in the construction of RediMODEL kits containing gene specific KOs, KIs and clinical variants (CV). * indicates null of C. elegans ortholog that are non-viable.
Disease types. The 15 disease genes of Figure 5 range in human relevance. Disease fields such as cancer, neurology, cardiology, and early aging are well represented. It has become clear that a given disease gene will have multiple associations to various disease types.18 The average disease type association for the 15 disease genes are 345 diseases/gene. As a result, the RediMODEL kits are expected to have a broad impact across a variety of NIH agencies, such as cancer (NCI), cardiomyopathies (NHLBI, NINDS) aging (NIA), psychiatric (NIAAA, NIDA, NIMH, NINDS), muscular degeneration (NIAMS), metabolic disorders (NIDDK), toxicity (NIEHS), and other disease (NIGMS).
Rare Disease. A focus on the discovery of pathogenic variants in disease genes inherently addresses a form of personalized medicine for diseases that are rare. Therapeutics developed to restore normal function (ie. CRISPR/Cas9 or antisense DNA) are therapies that serve a rare disease population when prevalence threshold does not exceed 200,000 persons.19 The variants in disease genes contribute to the expanding number of rare diseases, which, under the European system, are now approaching 20,000 disease categories.20 Therapeutics developed on rare disease genes are eligible at the FDA for expedited development activities via the Orphan Drug Act.21 The speed of animal model development in C. elegans, and its ease-of-use brings to many researchers an affordable platform for fast phenotyping of rare disease alleles.
Competing technologies. Various technologies ranging from bacterial-yeast expression systems to mammalian animal models are used to gain insight into the pathological effects of clinical variants. For instance, expressing disease genes in bacteria is a classic way to detect pathological phenotypes when the disease gene is an enzyme whose catalytic capacity can be measured.22 Yet, for disease genes with complex interaction phenotype, expression in bacteria removes the gene from native context and prevents the full exploration of the pathological profile of a genetic variant. A mouse or rat animal model is the “gold standard” for finding clear phenotypes from a clinical variant, yet the cost and amount of time spent are high, and results can be challenging to interpret from a drug development standpoint.23 Disease modeling with Induced Pluripotent Stem Cells (iPSC) offers an exciting platform to study clinical variants,24 but the removal of the cells from their native context of the intact animal regrettably removes the important effect of a tissue-based environment. 3D cell culturing techniques and organs-on-a-chip can be useful in restoring proper microenvironment context,25 but the ease of use for routine analysis clinical variant phenotypes has yet to evolve.Simple Model Organisms. The depth of genetic understanding in Drosophila and C. elegans have made these model organisms powerful surrogates to deploy in clinical variant understanding and drug discovery. For instance, insertion of human p53 in replacement of the Drosophila homolog was recently done for a variety of clinical variants of the TP53 gene.26 Foci defect phenotypes were observed that were consistent with similar defects in pathogenic p53 cancer tissues. In C. elegans, similar humanization of the dopamine transporter (DAT) was performed.27 The authors demonstrate two pathogenic alleles of DAT (399delG and 941C>T) gave significant defect phenotypes. The 399delG, which is known to be more pathogenic in humans, also gave a more significant LOF phenotype, when inserted as a homolog replacement into the C. elegans genome. These results demonstrate the humanization of small animal models can be used as a powerful system for detecting pathogenic phenotypes in the clinically-relevant SNPs that are associated with human disease.
RediMODEL kits bring the power of a simple model organism into the hands of researchers worldwide. With hundreds planned, soon a RediMODEL kit will be available for your favorite gene. Explore the current RediMODEL kits at the following link.
2. Mallick, Swapan, Heng Li, Mark Lipson, Iain Mathieson, Melissa Gymrek, Fernando Racimo, Mengyao Zhao, et al. “The Simons Genome Diversity Project: 300 Genomes from 142 Diverse Populations.” Nature 538, no. 7624 (October 13, 2016): 201–6. doi:10.1038/nature18964.
3. Morgan, Thomas M., Harlan M. Krumholz, Richard P. Lifton, and John A. Spertus. “Nonvalidation of Reported Genetic Risk Factors for Acute Coronary Syndrome in a Large-Scale Replication Study.” JAMA 297, no. 14 (April 11, 2007): 1551–61. doi:10.1001/jama.297.14.1551.
6. Gosai, Sager J., Joon Hyeok Kwak, Cliff J. Luke, Olivia S. Long, Dale E. King, Kevin J. Kovatch, Paul A. Johnston, et al. “Automated High-Content Live Animal Drug Screening Using C. Elegans Expressing the Aggregation Prone Serpin α1-Antitrypsin Z.” PloS One 5, no. 11 (November 12, 2010): e15460. doi:10.1371/journal.pone.0015460.
7. O’Reilly, Linda P., Cliff J. Luke, David H. Perlmutter, Gary A. Silverman, and Stephen C. Pak. “C. Elegans in High-Throughput Drug Discovery.” Advanced Drug Delivery Reviews 69–70 (April 2014): 247–53. doi:10.1016/j.addr.2013.12.001.
8. Mondal, Sudip, Evan Hegarty, Chris Martin, Sertan Kutal Gökçe, Navid Ghorashian, and Adela Ben-Yakar. “Large-Scale Microfluidics Providing High-Resolution and High-Throughput Screening of Caenorhabditis Elegans Poly-Glutamine Aggregation Model.” Nature Communications 7 (October 11, 2016): 13023. doi:10.1038/ncomms13023.
10. Horvitz, H. R., M. Chalfie, C. Trent, J. E. Sulston, and P. D. Evans. “Serotonin and Octopamine in the Nematode Caenorhabditis Elegans.” Science (New York, N.Y.) 216, no. 4549 (May 28, 1982): 1012–14.
11. Sze, J. Y., M. Victor, C. Loer, Y. Shi, and G. Ruvkun. “Food and Metabolic Signalling Defects in a Caenorhabditis Elegans Serotonin-Synthesis Mutant.” Nature 403, no. 6769 (February 3, 2000): 560–64. doi:10.1038/35000609.
12. Song, Bo-mi, and Leon Avery. “Serotonin Activates Overall Feeding by Activating Two Separate Neural Pathways in Caenorhabditis Elegans.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 32, no. 6 (February 8, 2012): 1920–31. doi:10.1523/JNEUROSCI.2064-11.2012.
18. Piñero, Janet, Àlex Bravo, Núria Queralt-Rosinach, Alba Gutiérrez-Sacristán, Jordi Deu-Pons, Emilio Centeno, Javier García-García, Ferran Sanz, and Laura I. Furlong. “DisGeNET: A Comprehensive Platform Integrating Information on Human Disease-Associated Genes and Variants.” Nucleic Acids Research, October 19, 2016. doi:10.1093/nar/gkw943.
24. Csöbönyeiová, Mária, Ľuboš Danišovič, and Štefan Polák. “Recent Advances in iPSC Technologies Involving Cardiovascular and Neurodegenerative Disease Modeling.” General Physiology and Biophysics 35, no. 1 (January 2016): 1–12. doi:10.4149/gpb_2015023.
25. Breslin, Susan, and Lorraine O’Driscoll. “Three-Dimensional Cell Culture: The Missing Link in Drug Discovery.” Drug Discovery Today 18, no. 5–6 (March 2013): 240–49. doi:10.1016/j.drudis.2012.10.003.
27. Illiano P, Lanzo A, Leo D, Paglione M, Zampi G, Gainetdinov RR,4, Di Schiavi E. "A Caenorhabditis elegans model to study dopamine transporter deficiency syndrome." Eur J Neurosci. 2017 Jan;45(1):207-214. doi: 10.1111/ejn.13366. Epub 2016 Sep 2.
Exploring Genetic Complexity in Proteins with a High Citation Rate for Disease Association
Synopsis: Many highly cited genes in neurological disease are reported to have a variety of disease associations per gene. Animal model studies are useful for helping gain a deeper understanding of the phenotypic consequence of observed disease alleles. Knudra uses a variety of approaches (cis and trans) to create platforms for diseases biology discovery. One particular implementation is RediMODEL kits combined with microfluidics systems to rapidly explore ion channel pathologies.
Scientific research has revealed a core set of proteins common to a wide variety of neurological diseases. (Alzheimer's, brain tumors, Parkinson’s, Muscular Dystrophies, Multiple Sclerosis, Glioblastoma, Charcot-Marie-Tooth Disease, Fragile X Syndrome, Huntington's Disease, Intellectual Disability, Ataxia Telagiectasia, von Hippel-Lindau Disease, and Deafness).1
Image credit: word cloud generated with www.jasondavies.com/wordcloud
Highest among these is GFAP, an intermediate filament with surprising connectivity to a wide variety of published disease associations. In the filament-like class we also have DMD, the dystrophin protein of Duchenne Muscular Dystrophy disease.2 The mutation landscape in dystrophin is becoming well understood. In a recent comprehensive study, researchers found and characterized the mutations in 576 families with DMD disease.3 Most of these mutations (82%) are large genomic rearrangements in the gene. The remaining portion of mutants are single nucleotide polymorphisms (SNPs) whose effects can be easy to model in the homologs of other organisms as frameshift and nonsense mutations. Only a small portion (0.9%) are the often difficult-to-interpret missense SNPs that lead to the change of only one amino acid in the protein. According to the dbSNB database,4 the dystrophin gene has 1562 discovered polymorphisms in its sequence. Of these, 280 are defined as being pathogenic. For the remaining, 34 mutations are defined as being benign, which is frequently seen as a DNA base change that codes for a synonymous silent mutation.5 What remains are 80% of the mutations which can be lumped in a class of "unknown significance."
Another gene with profound disease impact is HTT. A majority of HTT's SNPs (99%) are in the "unknown significance" class. For the 1078 SNPs reported in the HTT gene, there are over 1000 disease-associated alleles in the "unknown significance" category. Obviously there is a big need in the medical field to rapidly define a gene mutation as either "pathogenic" or "benign." Making assignment of allele status based on statistical correlation of gene mutation to the presence of disease in a family is a common approach, but it can be troublesome. For instance, genetic risk factors for acute coronary disease were examined a second time by researchers and, for 70 mutant variants assigned as risk factors in a previous study, the reexamination found only 1 mutant retained "nominally statistical significance.” 6
Animal Model Studies. To help assign and define pathogenicity in suspected mutant alleles of a disease gene, researchers are turning to animal models for helping get some answers. In the zebrafish model organism, gene silencing has been used to verify candidate genes for contribution to ciliary diseases.7 Since the discovery of gene silencing in C. elegans,8 researchers have used gene silencing screens on disease-gene homologs to further characterize the diseased state.9 C. elegans offers a readily tractable animal model platform for the general use by any researcher because it is reported to have homologous genes for nearly 70% of human disease-related genes.10 This conservation of gene function among the genomes of humans and C. elegans gives the researcher a unique tool for human disease study. Applied to the set of genes important in neurological disease, the C. elegans animal model has 11 genes of homologous function
Loss-of-function alleles. Clinical variants in disease are frequently loss-of-function alleles. RNAi can be used to model loss of function, but challenges in neuronal expression,11 strength of gene knock-down,12 and promiscuity to off-target effects 13 cause mimicry of clinical variant effects to be a challenging endeavor in RNAi applications. A more straightforward approach to understanding clinical variants is to directly install the polymorphic difference into the animal model then observe if it has effects that can be linked to pathogenicity.
There are many caveats to creating useful animal models, one of which is lamented by a review paper author:
"In my own experience, I have lost count of the number of mouse models that were originally reported as not having a phenotype in heterozygotes but in which my lab or others has subsequently found significant (i.e., clinically relevant) phenotypes." 14
Although animal models can be very powerful for understanding a clinical variant's capacity for pathogenicity, the high cost and long time-line, leave the researcher with a high degree of uncertainty in proceeding with a mouse animal model. An alternative is to turn to more simple models such as C. elegans to provide either the preliminary data for giving the researcher confidence in proceeding into an animal model, or, alternatively, give the researcher the critical animal model data that is need for high-profile publication of their findings. There are two ways to proceed with animal model generation. Prior to the advent of targeting nucleases, one could bring a gene and its control elements in-trans at a genomic locus free from local expression effects.
Manipulation on an in-trans platform allows the researcher to directly get data from rare, but sometimes very important, missense mutations. Alternatively, the researcher can use the new targeting nuclease tools (ex. CRISPR/Cas9) to directly install in-cis at the native locus of a homologous gene.
The installation of a clinical variant change at a native locus has an advantage of allowing the researcher to assess the clinical variant effect in the optimized context of natural physiology. This approach is recommended when the worm homolog is close in sequence to the disease gene candidate. However, some homologs are distantly related in primary sequence. As a result, installation of a clinical variant may miss a clinically-important phenotype. When low or absent homology occurs in a worm functional ortholog, a prudent approach can be to design hypermophic expression in appropriate tissues that would be likely to lead to strong phenotype. For neurological disease involving ion-channel signaling, animal models with strong phenotypes can be obtained by a high level of transgene expression in neuronal and muscular tissues. For instance, dopaminergic expression of A53T, A30P and wild-type of alpha-synuclein in C. elegans caused:
"worms expressing A30P or A53T mutant α-synuclein show failure in modulation of locomotory rate in response to food, which has been attributed to the function of dopamine neurons. This behavioral abnormality was accompanied by a reduction in neuronal dopamine content and was treatable by administration of dopamine.” 15
This clinically-relevant finding helps define the pathogenicity of the clinical variants observed in patients. To help researchers achieve a better understanding of clinical variant impact in disease, Knudra is developing a series of animal models in C. elegans with the Knudra RediModel Kit.
RediModel Kit. Knudra is creating a series of animal models to provide better understanding of how genetic diversity has impact on human diseases. Transgenic C. elegans are humanized with disease genes and then shipped to client in an easy-to-use kit. The end user pipets in a drug or exposes animals to apropirate eniviromental condition then monitors the outcome.
ScreenChip System. To aid in monitoring outcome, various high thoughput appoaches can be used. Systems that monitor multiple variable simulaneouosly provide some of the most eluminating data. In a collaboration with NemaMetrix, the animals of a RediModel kit are being optimized for use on can ScreenChip System. End users of the RediModel kit can apply their drug or enviroment exposed transgenics to the ScreenChip System and measure multiple morphological properties - the end user gets simultaneous data aquisition on 10+ different parameters ranging from electrophysiology to body morphometrics (length, width and size).
Combined Plafrom. By using RediModel Kits with ScreenChip System, a clinical researcher gains access to an easy-to-use platform for rapid profiling of clinical variants for pathological propensity. The platform makes it easier for the researcher to tackle the growing challenge of genetic diversity. For instance, in a study of 300 individuals from 142 diverse populations, researchers observed 38.2 million Single nucleotide Polymorphisms (SNPs), insertion/deletion polymorphisms (indels), and short tandem repeats (STRs) between the individual genomes.16 As genome wide studies are applied to various disease state and their associated genes, the number of variants with "unknown significance" will undoubtedly be on the rise.
Fast and Efficient Platform for Quantifying Phenotypes of Genetic Polymorphisms
Synopsis: Humanized animal models show significant promise for addressing personalized medicine questions. A key feature of the use of C. elegans is a wide variety of quantitative phenotyping strategies can be deployed. One of the most promising quantitative techniques is the ScreenChip microfluidics system.
The explosive growth of whole exome sequencing is identifying genetic causes of disease at a rapid pace. In many instances, exome sequencing is allowing rapid discovery of the genome variants in disease-causing genes.1 Yet it remains challenging to obtain clear evidence that a clinically-discovered variant is pathogenic. In Whole Exome Sequence (WES) studies, the noise in the background is immense. For instance, it has been observed that:
"WES generates a high number of variants per individual, a large proportion of which are still of unknown significance. The extreme difficulty of interpreting these variants has created a bottleneck in the clinical application of the technology." 2
Use of Animal Models: In an effort to overcome the "unknown significance" problem, many researchers are turning to animal models for the answer. By humanizing an animal model with a clinically-interesting sequence variant, the researcher frequently obtains an unequivocal demonstration of causality when a predicted phenotype is observed. Knudra is developing a series of important disease-related genes in C. elegans to be used as high throughput kits for rapid dissection of pathogenicity for the exploding landscape of Clinical Variants (ClinVar). For instance, consider the ryanodine receptor (RYR1). Currently, Pubmed's ClinVar database has accumulated 851 polymorphic alleles within the RYR1 gene. Of these, only 11% have been demonstrated to be of pathogenic significance. For the remaining alleles, 15% have been deemed to be benign. That leaves the remaining 74% (630 alleles!) as being polymorphisms of "Unknown Significance." This is an important consideration for the anesthesiologist worried about the occurrence of Malignant Hyperthermia Syndrome4 on the patient about ready to undergo the knife. Obviously there is a big need for a new platform that can rapidly indicate propensity for pathogenesis in a polymorphic variant of "Unknown Significance."
The Platform: In collaboration with our partner at NemaMetrix 5, the kits of transgenic C. elegans are ready-to-use animal models that are combined with a rapid platform for measuring ion channel conductance in intact animals. The result is a system for use in rapidly profiling pathophysiology of a wide vary of lines humanized with neurological disease genes.
Future Work: A focus on neurological diseases has a potential for high impact. The combination of Knudra's humanized animal model RediModel and the NemaMetrix ScreenChip System provides unprecedented speed in phenotype discovery for disease states that are challenging to model in single-cell culture systems. The medical researcher now has affordable access to a reliable animal model platform for speedy discovery of clinical variant pathogenicity.