Cell Conversions Transformed From Black Box Guesswork to Systematic Discovery
The ability to reliably convert the phenotype of one cell to another would be a game-changer for regenerative medicine. However, predicting the reprogramming factors necessary to induce cell conversion has largely relied on trial and error, revealing the need for a more systematic approach.
In 2016, Rackham et al. presented a predictive system called Mogrify in a paper titled: "A predictive computational framework for direct reprogramming between human cell types”, published in Nature Genetics.
Since then, the UK company has continued to work towards its goal of transforming the future development of cell therapies. This year, Mogrify has grown its team, appointed Dr. Jane Osbourn (former Vice President for Research and Development and Site Leader at MedImmune (AstraZeneca)) as Chair of the Board, relocated to the Bio-Innovation Centre in Cambridge, and won Business Weekly's "Disruptive Technology" award in March.
To learn more about Mogrify's approach to achieving cell conversion through transdifferentiation, we interviewed Pierre-Louis Joffrin, Corporate Development Executive at Mogrify.
Michele Wilson (MW): Transdifferentiation is the process of converting one cell type to another without going through a pluripotent state. Can you elaborate on our current ability to achieve cell conversion in this way?
Pierre-Louis Joffrin (PJ): Transdifferentiation is defined as the conversion of one cell type into another without going through pluripotency by the forced expression of cDNA encoding for transcription factors. The first example of transdifferentiation was reported in 1987 by Davis, Weintraub and Lassar, in a seminal study titled “Expression of a single transfected cDNA converts fibroblasts to myoblasts”.
Despite this discovery, the field of transdifferentiation was relatively on standby until 2006, when Yamanaka reported that pluripotent stem cells could be induced from mouse fibroblasts. Since then, many laboratories around the world have embarked on a journey to revisit the transdifferentiation concept published 20 years earlier. Soon after, combinations of transcription factors to convert human fibroblasts into neurons, cardiomyocytes and hepatocytes started to emerge.
Since 2010, over 400 scientific reports have been published reporting or validating transdifferentiation protocols, using different cell types as the source material. However, the success of a conversion depends on the exact identification of the combination of transcription factors amidst an infinitely large search space of >1022 possible combinations. Consequently, conversions have largely relied on a combination of educated guess and experimental trial and error, with little room for optimization. As a result, few conversions achieve the desired outcome of a functional mature cell.
MW: Mastering the technique of transdifferentiation would create many possibilities for regenerative medicine therapies. Can you paint a picture of what you hope the future will hold for this technique?
PJ: Transdifferentiation has huge potential in regenerative medicine in two distinct ways. Firstly, conversions can be used to produce cells for autologous or allogeneic implantation in patients with diseases where the number of functional cells are diminished, such as chondrocytes in osteoarthritis. Autologous chondrocyte implantation (ACI) is an approved therapy for cartilage defects that could be enhanced by transdifferentiating a more scalable cell type, such as fibroblasts, to increase the number of chondrocytes for re-implantation.
Secondly, transdifferentiation could be used to reprogram cells in vivo, to convert an undesired cell type into a desired cell type at the affected site in the body. A potential example of this would be converting white adipose tissue into brown adipose tissue, to reduce obesity and to help maintain glucose homeostasis for type-2 diabetes patients. So far, there are no approved in vivo reprogramming therapies, but as an increasing number of cell conversions are discovered and delivery systems are developed, these therapies should draw more attention as they bypass the problem of immunogenicity associated with the allogeneic implantation of foreign cells.
MW: What are the main limitations that are holding back progress in reprogramming techniques and applications?
PJ: Many of the currently identified transdifferentiation protocols are inefficient, leading to the conversion of only a small subset of cells, sometimes <1%. As a result, several transdifferentiation protocols are not reproducible. This could be due to factors such as:
(1) the optimal transcription factor combination is yet to be identified
(2) the delivery method for the transcription factors is not ideal, or
(3) the culture conditions to enhance the conversion and maintenance of the desired cell type have not been discovered.
Cellular validation and bioequivalence are vital, especially in the context of regenerative medicine. The cellular phenotype of the generated cells must be assessed to determine bioequivalence with native cells. Yet, scientific data showing this is poor, in some cases, or absent altogether.
In several cases, cells resembling the desired target cell type are transiently generated, but fail to be maintained in culture. This could be because the combination of transcription factors fails to induce a self-sustaining endogenous gene expression change, or because the optimal culture conditions to capture this cellular state have not been identified.
MW: Can you tell us about Mogrify’s system that predicts the reprogramming factors necessary to induce cell conversion?
PJ: Mogrify has developed a proprietary cellular conversion technology, which makes it possible to transform any mature human cell type into any other, without having to go through a pluripotent stem cell- or progenitor-cell state. The platform takes a systematic big-data approach to identify the optimal transcription factors and/or small molecules needed to convert and culture any human cell type. This is achieved in three distinct steps.
Firstly, the gene expression levels of the source and target cell types are compared, using next-generation sequencing, gene regulatory and epigenetic network data, to determine the change in gene expression required to achieve the conversion. Secondly, all transcription factors are ranked according to their potential effect (both direct and indirect) on the differential gene expression identified previously. Lastly, the optimal combination of transcription factors is determined by obtaining maximal coverage of the differential gene expression (minimum of 98%), whilst avoiding overlap in effect from the different factors.
Mogrify’s results have been experimentally validated, and can also predict the transcription factors used in known transdifferentiation experiments, serving as a directory of defined factors for any direct cell reprogramming.
MW: Mogrify’s predictive algorithm considers gene expression data, as well as regulatory network information. Can you provide examples of “regulatory network information” that have been incorporated here?
PJ: Regulatory networks are a way of representing causal interactions between transcription factors and their downstream gene targets, in order to calculate both direct and indirect effects on gene expression levels. These types of relationships can be inferred from different kinds of data such as DNA sequence, chromatin structure, gene or protein expression.
Mogrify incorporates data from two main databases: MARA, which models protein-DNA interactions and is representative of any potential direct effects on gene expression by transcription factor binding to promoter regions, and STRING, which models protein-protein interactions and is therefore representative of any potential indirect effects on gene expression through pathway cross-talk.
By capturing both of these regulatory network databases, Mogrify can rank the transcription factors, taking into account transcription factor-DNA binding motifs, gene promoter regions, gene expression and protein-protein interactions.
MW: To what extent do you expect Mogrify predictions to aid progress in the field of reprogramming?
PJ: The process of transdifferentiation is still a black box. Thus far, all the combinations of factors required to convert one cell type into another have been identified using a trial and error approach and some educated guesses based on expert knowledge of a specific cell type. The Mogrify algorithm is designed to turn this black box into a predictable variable, identifying sets of transcription factors for direct cell conversion regardless of prior knowledge.
This will streamline the discovery of new transdifferentiation protocols, that can be used to generate cell conversions which exhibit safety, efficacy and scalable manufacturing profiles suitable for development as cell therapies for regenerative medicine and oncology. Mogrify’s algorithm can also identify which genes are repressing cell-state conversions—a factor that has proven successful in enhancing conversion protocols in the past. Moreover, newer versions of the algorithm now include epigenetic data, which can be leveraged to predict growth factors and cytokines that are necessary for the maintenance of a specific cell type in culture.
MW: Can you tell us about any exciting upcoming or ongoing projects?
PJ: The lead program at Mogrify is the development of chondrocyte conversions (via its subsidiary, Chondrogenix). As part of this program we are creating a scalable supply of chondrocytes from fibroblasts, which can be allogeneically implanted into the patient (without the need for gene editing, since cartilage is immune-privileged). This would effectively democratize the already approved ACI therapy, and make it considerably cheaper by transforming it into an off-the-shelf product. This initial program is currently entering pre-clinical testing. We’re also working on converting osteoarthritic chondrocytes to healthy ones in vivo, in order to create the first disease-modifying therapy for osteoarthritis whereby the course of the disease is deterministically reversed.
Pierre-Louis Joffrin was speaking to Michele Wilson, Science Writer for Technology Networks.