Community Feedback

This blog post comes from the Myotonic Dystrophy Foundation.

Pharmacodynamic Biomarkers and DM
There is now strong support for the concept that a panel of splicing events may serve as a pharmacodynamic biomarker for go/no go decisions in drug development for myotonic dystrophy type 1 (DM1) and Duchenne muscular dystrophy (DMD). Data establishing splicing event sensitivity to free MBNL levels has converged with the natural history of alternative splicing patterns in DM patients to yield a subset of splicing events with the sensitivity and reproducibility to evaluate candidate therapeutics in early stage clinical trials. Quantitative pharmacodynamic biomarkers are invaluable in de-risking industry drug discovery and development, as they facilitate early stage assessment of molecular target engagement and modulation and may inform dose ranging studies. The only caveat is the dependence of these measures upon repeated muscle biopsies (a risk reduced, but not eliminated, by more tolerable needle biopsies). The identification and validation of a non-invasive assay of patient splicing status would be a valuable step forward for clinical trials in DM.

Early Support for a Non-Invasive Biomarker for DM1
Dr. Thurman Wheeler and colleagues at Massachusetts General, Harvard Medical, and Boston Children’s have explored the concept that a subset of extracellular RNAs (exRNAs) released into blood or urine may: (a) reflect alternative splicing status in DM-affected tissues and (b) thereby serve as an easily accessible pharmacodynamic biomarker platform for DM1 (Antoury et al., 2018). These studies were supported in part by a grant to facilitate “Development of Biomarkers for Myotonic Studies” from Myotonic Dystrophy Foundation/Wyck Foundation.

The research team initially found that > 30 transcripts that are alternatively spliced in DM1 muscle biopsies were detectable in human blood and urine samples; follow-up studies confirmed the presence of RNAs in extracellular fluids/exosomal particles. Normalized DMPK expression levels in urine from DM1 patients, by droplet digital PCR, were ~50% of unaffected controls. Assessments of DM1-established alternative splicing events showed that a subset (10/33) also occurred in urine exRNA, including being conserved in longitudinal (6-26 month) studies of the same patients. Assessments of alternative splicing events in blood exRNA did not yield the same value.

Using principal component analysis of 10 alternative splicing events observed in urine exRNA, the research team then generated a putative composite biomarker panel for DM1. The ensuing predictive model of alternative splicing in DM1 proved to be 100% accurate in comparisons of training and independent validation data sets to distinguish DM1 from unaffected controls and in distinguishing disease status of subsequently enrolled subjects. The research team also linked alternative splicing patterns in urine exRNA to variation in DM1 clinical phenotypes, suggesting that modeling of urine exRNA alternative splicing may allow both the tracking of disease progression and the impact of candidate therapeutics.

Finally, to address questions as to the source of urine exRNA, the team assessed alternative splicing in urinary tract cells of DM1 mouse models (the ubiquitous Mbnl1 ko and the tissue-specific HSALR). While kidney and bladder cells of the Mbnl1 ko reflected patterns in skeletal muscle, assessments of the same tissues in the HSALR showed no differences from control mice. These data strongly suggested that the exRNAs assessed in urine reflect exosomes released from urinary tract cells. Some of the alternatively spliced transcripts in urine exRNA also were shown to be altered by antisense oligonucleotide drugs previously shown to correct splicing patterns in DM1 mouse models. The research team’s parallel studies of Duchenne muscular dystrophy also supported the concept that urine exRNA has utility as a pharmacodynamic biomarker in drug intervention studies.

Towards a Non-Invasive Biomarker for DM1
Taken together, these data provide compelling proof of concept that a panel of alternative splicing events assessed in urine may serve as a robust composite biomarker of DM1 progression and as a tool for assessment of candidate therapeutics. A non-invasive biomarker such as this would greatly extend the ability to perform repeated measurements in longitudinal natural history studies (as a disease progression biomarker) and in interventional clinical trials (as patient stratification and pharmacodynamic biomarkers), including making assessment of pediatric DM1 patient cohorts feasible. Although it is not essential to formally qualify a biomarker, existing regulatory agency guidance documents (see References below) provide a valuable evidentiary framework for moving non-invasive biomarker work towards an accepted clinical tool for DM1.

Antoury L, Hu N, Balaj L, Das S, Georghiou S, Darras B, Clark T, Breakefield XO, Wheeler TM. Analysis of extracellular mRNA in human urine reveals splice variant biomarkers of muscular dystrophies. Nat Commun. (2018) 9: 3906. doi: 10.1038/s41467-018-06206-0. PMID: 30254196

Framework for Defining Evidentiary Criteria for Biomarker Qualification. Foundation for the National Institutes of Health (FNIH) Evidentiary Criteria Writing Group. October 2016. (announcement) (pdf)

Guidance for Industry and FDA Staff: Qualification Process for Drug Development Tools. (pdf)

This commentary originally appeared as an Editor’s Choice in Science Translational Medicine. Thanks to STM and Steven Jay for permission to reprint here.

The sci-fi thriller I, Robot tells the story of robots attempting to take over the world based on their interpretation of the three governing laws of their programming. This plan is thwarted with the help of Sonny, a unique robot who can ignore the three laws due to being programmed differently. This movie illustrates how selective programming can be a powerful tool that can be used to turn a subset of a population against the rest. This same concept underlies the strategy of gene-directed enzyme prodrug therapy (GDEPT) for cancer, which involves specific delivery of a gene to cancer cells that allows for subsequent activation of a systemically administered prodrug into a toxic form only in cells where an enzyme encoded by the delivered gene is present. Several GDEPT strategies have advanced to clinical trials; however, the specificity and fidelity of gene delivery are still limiting factors to successful translation.

Toward addressing these limitations, Wang et al. describe the use of modified extracellular vesicles (EVs) for targeted delivery of mRNA to cancer cells overexpressing the HER2 receptor. EVs are nanoscale vesicles secreted by many cell types that have been co-opted for a variety of therapeutic applications. However, targeted delivery using EVs has been challenging, as has encapsulation of large nucleic acid cargo. To address cargo encapsulation, the authors applied a transfection-based approach to successfully load exogenous mRNA encoding for the enzyme HChrR6 into EVs. To address targeting, the authors created a novel chimeric protein consisting of a HER2 antibody fragment to target the receptor on cancer cells and the C1C2 domain of lactadherin, which interacts with the EV membrane. By mixing mRNA-loaded EVs with purified chimeric protein, the EVs were endowed with targeting capability for HER2-overexpressing cancer cells. Delivery of these EVs followed by systemic administration of the prodrug 6-chloro-9-nitro-5-oxo-5H-benzo-(a)-phenoxazine (CNOB) resulted in near complete growth arrest of orthotopically implanted HER2-overexpressing breast tumors in mice.

This report establishes a new and versatile approach for improving GDEPT that could be applied to a wide variety of cancers and other diseases. Significant barriers to translation of this approach remain, most notably the problem of scalability of EV-based approaches. However, the methods and strategy described are likely to have broad utility in further developing both GDEPT and therapeutic EVs.

Highlighted Article
J.-H. Wang, A. V. Forterre, J. Zhao, D. O. Frimannsson, A. Delcayre, T. J. Antes, B. Efron, S. S. Jeffrey, M. D. Pegram, A. C. Matin, Anti-HER2 scFv-directed extracellular vesicle-mediated mRNA-based gene delivery inhibits growth of HER2-positive human breast tumor xenografts by prodrug activation. Mol. Cancer Ther. (2018) 17:1133-1142. PMID:29483213 doi:10.1158/1535-7163.MCT-17-0827

The extracellular RNA communication program (ERCP) has established an ontology group to work towards standardization and classification of the terms used to describe extracellular vesicles (EVs). As part of this ontology subcommittee effort, we have worked with domain experts and ontology experts to propose new terms and relationships that can be added to Gene Ontology (GO) for describing EVs. Thus far, we have created three versions shown in Figures 1, 2 and 3 with different levels of detail (click on each figure for a higher resolution image).

In the most detailed version (Figure 1), each term (node) and relationship (edge) is required to have a minimal amount of evidence, potentially consisting of a single publication. As the knowledge evolves, some terms/relationships may be tentative, as in cases where the population of vesicles is heterogeneous and/or the biogenesis of the vesicles is uncertain. It is anticipated that there will be future revisions of the ontology structure (e.g., terms and relationships can be added, removed, and rearranged based on new knowledge). The main advantage of this approach is that it reflects the current state of the literature. Since it has more comprehensive coverage, including specific subtypes of EVs identified by subgroups of researchers, it can support more detailed annotation. The disadvantage, however, is that it is prone to changes. Some of the terms and relationships are likely to be revised in future iterations.

In the simplified/incremental approach (Figure 2 and Figure 3), each term and relationship must have strong evidence with a high level of community consensus to be included into the ontology. Because of the early stage of this field, using this approach, the current iteration of the ontology will include only terms that are quite general. It is anticipated that future iterations of the ontology will include additional terms for more specific types of extracellular vesicles/particles as they are better defined in terms of biophysical properties, functional characteristics, and macromolecular composition. The main advantage of this approach is the less anticipated need to remove or correct terms/relationships in the future. The disadvantage, however, is the lack of specificity in terminology limiting the ability to annotate datasets with precise terms.

We are in the process of building community consensus. We submitted our proposals to several consortiums and societies including the External RNA Controls Consortium (ERCC), International Society for Extracellular Vesicles (ISEV), and American Society for Exosomes and Microvesicles (ASEMV) for their review and approval. In addition, we will engage the broader community in the discussion.

The proposed versions of EV classification correspond to what we perceive as the lumper vs. splitter approaches. In the context of biological taxonomy, lumpers are those who prefer to classify organisms into larger groups based on substantial differences, resulting in fewer species overall, while the splitters prefer to use minute differences to classify living organisms into a greater number of separate groups. These two different approaches may not create significant conflicts for a well-understood domain. However, for a new or partially understood domain like extracellular RNA (exRNA) communications, it is difficult to decide which approach is better. Domain experts often contend vigorously about where the boundary ought to lie in terms of classification. For example, should there be separate terms for vesicles that are formed in endosomes/multivesicular bodies (sometimes termed exosomes) and vesicles that bud directly from the cell surface (sometimes termed microvesicles or shedding microvesicles)?

Finally, there are two major goals of the ontology subcommittee: i) publish a manuscript on the current considerations regarding the ontology of exRNA-containing particles, and ii) develop a recommended vesicle ontology to be submitted to Gene Ontology. We would like these documents to reflect the opinions of the exRNA community, so we need your input.  Please post comments on the following questions, or send your thoughts to

What we have here is not a new debate. It is just another example of a classic argument between lumpers and splitters, one that may not be resolved any time soon. That lack of resolution is fine, because understanding grows from these debates. We should not be frustrated when scientists cannot agree with each other. Instead we should encourage their debates. New ideas and innovations are forged from debates, and are stronger because of them.

Which version of the three draft ontologies do you favor, and why? Specifically:

  • Should there be separate terms for vesicles that are formed in endosomes/multivesicular bodies (sometimes termed exosomes) and vesicles that bud directly from the cell surface (sometimes termed microvesicles or shedding microvesicles)?
  • How much literature support does a specific particle type need to be included in the ontology?
  • Are there use cases associated with each version?

Figure 1. Most detailed version. Note that we have qualified the term Exosome, as Extracellular Exosome (as “Exosome” is already defined in GO for the intracellular protein complex)

Figure 2. Less detailed version.

Figure 3. Least detailed version.