As usual, some background may be helpful here: You’ve heard of genomics, the analysis of large numbers of gene variants as a clue to disease causation. But genomics doesn’t deal with the actual proteins produced by those genes. That led to proteomics, the analysis of proteins in a tissue or fluid sample, as a more relevant window into disease mechanisms and possibly as a path to diagnostic tests. But proteins in samples can be influenced by many things such as breakdown of proteins by enzymes or by the cells’ normal garbage disposal mechanisms. A solution is “transcriptomics,” which identifies and quantifies messenger RNA (mRNA) on its way from being encoded (“transcribed”) by DNA to being translated into protein. Although every cell in our bodies has our entire genetic endowment in the form of DNA, only a fraction of those genes is actually transcribed into mRNA. The types and amounts of mRNA produced depend on the job of the specific cell and its protein needs at the moment. Click here if you’d like to chew on a highly technical but excellent, current review of the topic.
In the current issue of the prestigious Journal of Clinical Investigation, a team mostly from the Mayo Clinic Jacksonville has compared the transcriptomes of tissue from cerebrum and cerebellum with autopsy-confirmed PSP to those with Alzheimer’s disease and to controls with no autopsy evidence of neurodegenerative disease. I’ll cut to the chase: To the researchers’ surprise, they found the four types of tissue to be quite similar in their transcriptomic profiles. This result suggests that a treatment or a diagnostic test directed at one or more of the protein (or mRNA!) abnormalities in either disease could work for the other disease as well. We already suspected that, but without a lot of supporting evidence beyond the superficial observation that both diseases involve tau aggregation. This new paper provides some better evidence.
I’ll return to the paper’s practical implications in a minute. But first, back to biology class you go: “O-omics” results (Wikipedia lists 45 kinds of “omics” in biomedicine to date) are by their nature shotgun approaches, typically yielding a long, inscrutable list of statistically significant but questionably meaningful differences between subject groups. So, seeking some sort of useful pattern in the data, researchers divide the genes, proteins or mRNAs yielding statistically significant “hits” into categories based on their known functions. These categories are called “networks” because they’re based on interactions or on commonalities of function. The process of creating and using such standardized, defined categories in genomics is called “gene ontology.”
The authors of the new paper point out that previous transcriptomic work in AD has revealed differences from controls in many such networks, most prominently “immune function, myelination, synaptic transmission and lipid metabolism.” They also point out that it’s usually difficult to know if these mRNA differences are the cause of the cellular damage in AD or merely the brain’s reactions to that damage.
Now back to this project: As I said above, the transcriptomic work analyzed not only the cerebral cortex, where the researchers knew ample pathology exists in both AD and PSP, but also, as a kind of control group, the cerebellar cortex, where the standard autopsy shows little or no damage in AD or PSP. Another feature of the study design was to use the temporal lobe as the source of its cerebral samples. That part of the brain is heavily involved in AD by standard methods but little or not at all in PSP.
Surprisingly, the results were that the transcriptome abnormalities (relative to non-neurological controls) were quite similar in all four types of tissue – AD temporal cortex, PSP temporal cortex, AD cerebellar cortex and PSP cerebellar cortex. In their words, “The DEG [differentially expressed gene] changes between AD and PSP in two regions of the brain demonstrate a striking conservation [consistency] of transcriptomic changes across these different neurodegenerative diseases.”
Because the degree of traditionally measurable cell damage differed markedly across those four sets of samples, they infer that the changes are “upstream,” meaning at an early step in the disease process, rather than “downstream,” in reaction to damage. That would mean that even in the early stages of AD and PSP, the disease process is already at work in areas that have not started to show physical signs of damage as assessed by microscope, MRI or PET scan.
What were the networks most affected by the transcriptomic changes? In the authors’ words: “Up-regulated in both AD and PSP were gene networks serving chromatin modification, gene expression, chromosome organization and metabolism of nucleotides. In the cerebellum the shared upregulated genes link to biological processes relating to RNA and RNA transcription, cell-cell junctions, and heart, kidney, gland, and circulatory system development. Shared down-regulated genes in AD and PSP are associated with gene ontology cell compartment terms related to mitochondrial and ribosomal functions in both the temporal cortex and the cerebellum.”
The paper’s first author is Xue Wang, PhD, a bioinformatician at Mayo. The last two authors, sharing credit as senior authors, are Todd Golde, MD PhD of the University of Florida and Nilufer Ertekin-Taner, MD PhD. I emailed Dr. Taner for her take on the results. She replied in part, “These findings can be leveraged to develop multifaceted therapies and biomarkers that address these common, complex and ubiquitous molecular alterations in neurodegenerative diseases.” I’d agree.
So, this unexpected discovery suggests that it may prove fruitless to look for causes of specific neurodegenerative diseases in their gene expression profiles. Abnormal gene expression may not be the true origin of the disease, but only the cell’s reaction – not to downstream physical damage visible with standard tools, but to some other, far more upstream causation. Yet, interrupting that reaction at the level of its mRNA might be the key to halting the progression of multiple, of not all, neurodegenerative diseases. Adding to the appeal of that approach is that the treatment targets are available at a very early stage of the disease process.
Testing that suggestion will have to start with analyzing more than just AD and PSP, and I’m guessing that this effort is already in progress thanks to the excellent collection at the Mayo Brain Bank. I’ll keep you informed.