Periodic Reporting for period 2 - UPRmt (The Mitochondrial Unfolded Protein Response)
Reporting period: 2020-05-01 to 2021-10-31
We proposed to genetically map and mechanistically characterize UPRmt pathways across three different animal species, with the goal to link the UPRmt with metabolic homeostasis, health- and lifespan and to translate benefits of modulating the UPRmt into targeted human therapies.
Mapping the full complexity of UPRmt signaling, which involves communication between nuclear and mitochondrial genomes as well as environmental factors, poses a challenge for classical reverse genetic approaches. We have assembled a new mouse genetic reference panel (GRP), the hybrid diversity panel (HDP), from 30 BXD (a 2-parent recombinant inbred panel), 21 Collaborative Cross (CC; an 8-way recombinant inbred panel), and 34 inbred strains (including BXD & CC founders). This panel combines high genetic diversity (~2x the variation in humans) with the capacity to resample “isogenic” individuals to characterize genetic and environmental contributions to phenotypes. Our plan is centered on achieving 4 Specific Aims (see Figure 1):
Aim 1: Map mammalian UPRmt genes and molecular and mitochondrial networks in vivo after the induction of the UPRmt in the HDP at 3 different time-points throughout life with 2 different inducers – referred to as the tissue collection pipeline.
Aim 2: Integrate these UPRmt networks with a large set of clinical, mitochondrial, and molecular phenotypes collected throughout life to establish links between the UPRmt and homeostasis, health- and lifespan. Referred to as the phenotyping pipeline.
Aim 3: Mechanistically validate new UPRmt genes and networks, using loss-of-function studies in cells, worms and mice.
Aim 4: Clinically translate the most promising UPRmt regulators and systems using joint genetic association studies with human cohorts.
Mapping the full complexity of UPRmt signaling, which involves communication between nuclear and mitochondrial genomes as well as environmental factors, poses a challenge for classical reverse genetic approaches. We have assembled a new mouse genetic reference panel (GRP), the hybrid diversity panel (HDP), from 30 BXD (a 2-parent recombinant inbred panel), 21 Collaborative Cross (CC; an 8-way recombinant inbred panel), and 34 inbred strains (including BXD & CC founders). This panel combines high genetic diversity (~2x the variation in humans) with the capacity to resample “isogenic” individuals to characterize genetic and environmental contributions to phenotypes. Our plan is centered on achieving 4 Specific Aims (see Figure 1):
Aim 1: Map mammalian UPRmt genes and molecular and mitochondrial networks in vivo after the induction of the UPRmt in the HDP at 3 different time-points throughout life with 2 different inducers – referred to as the tissue collection pipeline.
Aim 2: Integrate these UPRmt networks with a large set of clinical, mitochondrial, and molecular phenotypes collected throughout life to establish links between the UPRmt and homeostasis, health- and lifespan. Referred to as the phenotyping pipeline.
Aim 3: Mechanistically validate new UPRmt genes and networks, using loss-of-function studies in cells, worms and mice.
Aim 4: Clinically translate the most promising UPRmt regulators and systems using joint genetic association studies with human cohorts.
Aim 1: Map mammalian UPRmt genes and molecular and mitochondrial networks in vivo after the induction of the UPRmt in the HDP at 3 different time-points throughout life with 2 different inducers – referred to as the tissue collection pipeline (see Figure 2).
Mice in the tissue collection pipeline are used to analyze how the UPRmt is controlled by genetic and environmental effects. We are also collecting a vast set of mitochondrial phenotypes and multi-omic traits—transcriptome, epigenome, proteome, and metabolome—in livers of the HDP strains in basal conditions and after induction of the UPRmt at 3 different points throughout life, i.e. young (2 mo), adult (12 mo) and old (24 mo), by the administration of Doxycycline (Dox) and the NAD+ precursor, nicotinamide riboside (NR). So far, the 2, 12, and 24 mo tissue collections were respectively finished in 56, 51 and 27 strains. Tissues are biobanked and all samples are tracked using SLIMS. To avoid batch effects, we will start molecular trait measurements only after completion of the tissue collection in Q3 2023.
Aim 2: Integrate these UPRmt networks with a large set of clinical, mitochondrial, and molecular phenotypes collected throughout life to establish links between the UPRmt and homeostasis, health- and lifespan. Referred to as the phenotyping pipeline.
In parallel, we are evaluating a wide set of physiological traits collected under basal, non-stressed conditions in the HDP strains in adult and old mice; this constitutes the phenotyping pipeline. So far 61 cohorts of 8 chow fed females/strain have entered in the phenotyping pipeline that scores for cardio-metabolic and neuro-behavioural traits. The projected end date of the phenotyping pipeline is foreseen in Q3 2023. In addition to these classical phenotypes, we are continuously monitoring in-cage activity and collecting environmental measurements throughout life using Tecniplast’s Digital Ventilated Cages (DVC).
Aim 3: Mechanistically validate new UPRmt genes and networks, using loss-of-function studies in cells, worms and mice.
We have completed a large phenotyping and multi-omic experiment on a panel of 100 strains from a C. elegans RIAIL GRP derived from the N2 and the CB4856 strains under control and Dox conditions. This experiment will allow us to identify conserved UPRmt networks and thus validate the mouse findings.
Aim 4: Clinically translate the most promising UPRmt regulators and systems using joint genetic association studies with human cohorts.
Mapping of QTLs, the identification of genes/networks involved in UPRmt signaling (Aim 1), and the multiscalar UPRmt models (Aim 2) will result in candidate genes for future mechanistic validation and clinical translation in Aims 3-4.
The prospectively collected data of the HDP pipeline enable us to evaluate the extent of coupling between the UPRmt induced with NR and Dox and mitochondrial and multi-omics molecular traits (Aim 1), on the one hand, and with metabolic homeostasis, and a broad array of healthspan traits (Aim 2), on the other hand, while at the same time allowing to map the genetic determinants that drive these molecular and clinical circuits. To do so, we will utilize genetic, bioinformatic and systems genetics frameworks to integrate multi-omics molecular data, mitochondrial traits, and the largest extant phenome dataset. Particularly informative will be the combination of PheWAS with mediation analysis to causally link genes with molecular and clinical phenotypes that we have already implemented (PMID: 29199021, 31754022). Finally, we are developing methodology to validate our findings in human population data from the UK Biobank and GTEx projects using Mendelian Randomization.
Mice in the tissue collection pipeline are used to analyze how the UPRmt is controlled by genetic and environmental effects. We are also collecting a vast set of mitochondrial phenotypes and multi-omic traits—transcriptome, epigenome, proteome, and metabolome—in livers of the HDP strains in basal conditions and after induction of the UPRmt at 3 different points throughout life, i.e. young (2 mo), adult (12 mo) and old (24 mo), by the administration of Doxycycline (Dox) and the NAD+ precursor, nicotinamide riboside (NR). So far, the 2, 12, and 24 mo tissue collections were respectively finished in 56, 51 and 27 strains. Tissues are biobanked and all samples are tracked using SLIMS. To avoid batch effects, we will start molecular trait measurements only after completion of the tissue collection in Q3 2023.
Aim 2: Integrate these UPRmt networks with a large set of clinical, mitochondrial, and molecular phenotypes collected throughout life to establish links between the UPRmt and homeostasis, health- and lifespan. Referred to as the phenotyping pipeline.
In parallel, we are evaluating a wide set of physiological traits collected under basal, non-stressed conditions in the HDP strains in adult and old mice; this constitutes the phenotyping pipeline. So far 61 cohorts of 8 chow fed females/strain have entered in the phenotyping pipeline that scores for cardio-metabolic and neuro-behavioural traits. The projected end date of the phenotyping pipeline is foreseen in Q3 2023. In addition to these classical phenotypes, we are continuously monitoring in-cage activity and collecting environmental measurements throughout life using Tecniplast’s Digital Ventilated Cages (DVC).
Aim 3: Mechanistically validate new UPRmt genes and networks, using loss-of-function studies in cells, worms and mice.
We have completed a large phenotyping and multi-omic experiment on a panel of 100 strains from a C. elegans RIAIL GRP derived from the N2 and the CB4856 strains under control and Dox conditions. This experiment will allow us to identify conserved UPRmt networks and thus validate the mouse findings.
Aim 4: Clinically translate the most promising UPRmt regulators and systems using joint genetic association studies with human cohorts.
Mapping of QTLs, the identification of genes/networks involved in UPRmt signaling (Aim 1), and the multiscalar UPRmt models (Aim 2) will result in candidate genes for future mechanistic validation and clinical translation in Aims 3-4.
The prospectively collected data of the HDP pipeline enable us to evaluate the extent of coupling between the UPRmt induced with NR and Dox and mitochondrial and multi-omics molecular traits (Aim 1), on the one hand, and with metabolic homeostasis, and a broad array of healthspan traits (Aim 2), on the other hand, while at the same time allowing to map the genetic determinants that drive these molecular and clinical circuits. To do so, we will utilize genetic, bioinformatic and systems genetics frameworks to integrate multi-omics molecular data, mitochondrial traits, and the largest extant phenome dataset. Particularly informative will be the combination of PheWAS with mediation analysis to causally link genes with molecular and clinical phenotypes that we have already implemented (PMID: 29199021, 31754022). Finally, we are developing methodology to validate our findings in human population data from the UK Biobank and GTEx projects using Mendelian Randomization.
We implemented and validated a number of these data collection and analysis methods already in other studies, so the analytical pipelines will be established and ready at the time the data from Aim 1 and 2 are collected. Certain of these data management strategies extended beyond the strict bio-medical field, but were essential given the mere size and ambition of our project (see Figure 3). As such we:
- Adopted the Celoxis professional project management tool: In addition to existing mouse management and breeding software (pyRAT), we needed to find solutions to facilitate logistics, planning of human resources and long experimental pipelines. We have an on-premise instance running and have integrated it in our existing infrastructure.
- Developed an app for real-time phenotype collection: In collaboration with Genohm, the company behind our Laboratory Information Management System, SLIMS, we piloted a novel application for mouse data monitoring and phenotyping using handheld devices with barcode readers. While the system works, we have not implemented it in the HDP, but plan on using this for other projects.
- Established data management practices: Phenotyping data and metadata are regularly collected and formatted in a structured and standardized form in a semi-automated way with a dedicated informatic pipeline. Data quality is assessed in terms of data type and structure and by the application of filtering criteria.
- Defined a knowledge base that ensures that data and metadata collected with different methods (such as RNAseq and behavioral tests) are linked and consistently described. It is organized in a hierarchy of tables that store structured information about data file location and format and describes variables, samples and animals following domain standards (i.e. ontologies and controlled vocabularies). This is essential for the analysis, integration and sharing of the large and diverse dataset.
- Adopted a data analysis framework that supports open science has been established using the Renku platform (https://datascience.ch/renku/renku-platform) for which a private instance has been deployed at EPFL. Data, code and computing environments are kept under version control and data analysis is performed in highly compartmentalized workflows that enable the full reproducibility of results while facilitating the sharing and reuse of data.
- Engaged in the creation of a biobank. The tissue collection and phenotyping pipeline will generate an excess of 100,000 primary biosamples. Along with the central services of EPFL, we have been in talks over the last two years to construct an ISO or other standard-compliant institutional biobank. The HDP samples are projected to be among the first samples in this biobank. This will make the samples discoverable by the wider scientific community and will facilitate sharing of biosamples within adequate ethical and legal frameworks.
- Adopted the Celoxis professional project management tool: In addition to existing mouse management and breeding software (pyRAT), we needed to find solutions to facilitate logistics, planning of human resources and long experimental pipelines. We have an on-premise instance running and have integrated it in our existing infrastructure.
- Developed an app for real-time phenotype collection: In collaboration with Genohm, the company behind our Laboratory Information Management System, SLIMS, we piloted a novel application for mouse data monitoring and phenotyping using handheld devices with barcode readers. While the system works, we have not implemented it in the HDP, but plan on using this for other projects.
- Established data management practices: Phenotyping data and metadata are regularly collected and formatted in a structured and standardized form in a semi-automated way with a dedicated informatic pipeline. Data quality is assessed in terms of data type and structure and by the application of filtering criteria.
- Defined a knowledge base that ensures that data and metadata collected with different methods (such as RNAseq and behavioral tests) are linked and consistently described. It is organized in a hierarchy of tables that store structured information about data file location and format and describes variables, samples and animals following domain standards (i.e. ontologies and controlled vocabularies). This is essential for the analysis, integration and sharing of the large and diverse dataset.
- Adopted a data analysis framework that supports open science has been established using the Renku platform (https://datascience.ch/renku/renku-platform) for which a private instance has been deployed at EPFL. Data, code and computing environments are kept under version control and data analysis is performed in highly compartmentalized workflows that enable the full reproducibility of results while facilitating the sharing and reuse of data.
- Engaged in the creation of a biobank. The tissue collection and phenotyping pipeline will generate an excess of 100,000 primary biosamples. Along with the central services of EPFL, we have been in talks over the last two years to construct an ISO or other standard-compliant institutional biobank. The HDP samples are projected to be among the first samples in this biobank. This will make the samples discoverable by the wider scientific community and will facilitate sharing of biosamples within adequate ethical and legal frameworks.