aeffen | Cambridge, MA | August 25, 2019
It's Day −9 of graduate school, and for the past two weeks, I've been rotating in the lab of Myriam Heiman, who uses cell type-specific molecular profiling techniques to study mechanisms of enhanced vulnerability in neurodegenerative diseases like Huntington's disease (HD). In shorter words, it means that the people in her lab use things like translating ribosome affinity purification (TRAP) and RNA sequencing to figure out what specific cells are doing. I'm investigating something I'll call the glycolytic switch in oligodendrocytes (OLs), which may be altered in HD.
Why glycolysis? The story starts with this simple idea: in addition to ensheathing neurons with myelin, OLs play a continual active role in nourishing neurons. They do this by producing lactate via glycolysis, which is then shuttled across the OL membrane to diffuse over to the axon under its care. The neuron can then consume its neighbor's lactate via oxidative phosphorylation (oxphos) to regenerate ATP. The general concept of moving metabolites intercellularly has been dubbed the lactate shuttle hypothesis and has been studied across many tissues and organ systems. In the context of the brain, this hypothesis has enjoyed the limelight in recent years as a way of integrating many intriguing clues linking blood glucose, astrocytes, neurons, and (finally!) oligodendrocytes.
Sidebar: why are neurons so needy? Actually, they do a lot of extremely costly work. For example, a projection neuron, like one of the medium spiny neurons (MSNs) of the basal ganglia, may have an axon that stretches many millimeters or even centimeters away from its cell body – for a cell, that's an incredible distance! Those axons are constantly pumping ions across their membranes, maintaining complex synaptic relationships at their termini, and responding dynamically to stimuli despite their spatial estrangement from their nuclei. The metabolic cost of maintaining such a large and active cell is simply staggering.
Just to give you a vague sense of scale, for the statistically-minded: using the NEURON simulation environment to model a dopaminergic neuron, one may find that "the energy cost of [action potential] propagation increases according to a power law function with respect to the size/complexity and/or surface area of the axonal field" (Pissadaki and Bolam, 2013; emphasis mine). Very long neurons need much more energy than the sphere of cytoplasmic soup you might imagine as the stereotypical cell.
Neurons live on a precipice of oxidative stress, balancing between producing more energy and generating harmful byproducts like reactive oxygen species (ROS). Cells have invented ways to combat this. For example, keeping H2O2 in check is the glutathione antioxidant system:
2 glutathione (reduced) + R2O2 → GSSG (oxidized) + 2 ROH
Glutathione levels are in turn sustained by the reducing agent [1] NADPH, which can be regenerated by the pentose phosphate pathway (PPP). The PPP is intimately related to glycolysis, because both processes utilize glucose as their substrate. (I'll get to this in a bit.) Thus, regulating how much glucose is shunted to the PPP in the neuron is one way the neuron strikes a balance between energy production and antioxidant defense.
The lactate shuttle is how OLs make the metabolic calculus a little easier. By supplying the neuron with extra lactate, glia allow neurons to put more glucose into the PPP while still regenerating enough ATP via oxphos.
The bioenergetics of the brain are far more complex and nuanced than I can summarize here, but I've chosen to focus on how OLs regulate their glucose metabolism in the context of the lactate shuttle. While OLs are capable of performing glycolysis and oxphos (1) to meet their internal metabolic demands like any other cell, they typically perform aerobic glycolysis (2) to feed the neurons they're coupled to.
glucose → glucose-6P → pyruvate → TCA cycle → oxphos (1)
glucose → glucose-6P → pyruvate ↔ lactate → shuttle to axon (2)
It is hypothesized that OLs undergo a shift in metabolic regulation from (1) to (2) during their development, perhaps during or after myelination. This is how they become metabolically coupled to neurons via the lactate shuttle. It has been found that adult OLs can subsist essentially entirely on aerobic glycolysis if their mitochondria are knocked out after myelination (Funfschilling et al. 2012)! In the absence of further perturbation, adult OLs reach a nice steady state of glycolysis and lactate export that nourishes their axonal partners.
However, there are other good uses for glucose in the cell besides producing lactate, which ultimately may be relevant to the progression of HD. Some contextual clues:
Whether fatty acid metabolism dysregulation is the chicken or the egg – or one of the many chickens in the multi-step process of HD pathogenesis – is a complicated question. However, by comparing the logical outcome of these ideas with actual transcriptional data from the R6/2 mouse model of HD, it is probable that heightened fatty acid metabolism plays a part in HD degeneration by increasing stress on vulnerable mitochondria and elevating ROS levels in the cell.
As a consequence of this, OLs may compensate by increasing PPP throughput, which supports the anti-ROS glutathione system. This is achieved by diverting glycolysis at an early step:
glucose → glucose-6P → ribulose-5P + NADPH (3)
The way that the cell achieves the switch from (1) to (3) is by downregulating or inactivating certain rate-limiting enzymes or allosteric activators of glycolysis.
Why do we care what OLs are doing with their glucose in the diseased brain again? Because OLs and axons are metabolically coupled, these changes don't happen in a vacuum. Perturbing the flux of carbon energy sources through glycolysis in OLs directly affects the amount of lactate that gets shuttled to axons. OLs may be dealing with disease-related mitotoxicity and shunting glucose-6P to the PPP, but they may also be inadvertently starving the neurons they aim to serve. If we confirm that this switch occurs at significant levels in HD OLs, we can more confidently continue to pursue the hypothesis that OL-axon metabolic coupling is a relevant component of HD pathogenesis.
I'm now looking for the protein expression fingerprints of this switch in the mouse brain. So far, I've been establishing what "normal" expression looks like, as well as confirming that our immunostaining markers of OLs actually mark OLs. I've also been playing around with the results of a differentially expressed gene (DEG) analysis performed on snRNA-seq data from R6/2 and wild-type mice. That rudimentary attempt at synthesizing knowledge from data has proven surprisingly fun and fruitful, and you might hear more about it in a later blog post. All in all, it's been great to put a hypothesis to the experimental test again – I really missed being in the lab, and I can't wait to see where these ideas go!
[1] Refresher on reduction and oxidation (redox) terminology.
A reducing agent, reducer, or reductant is a compound that "donates" an electron to another compound in a redox reaction. In a redox reaction, the reducer loses electrons and becomes oxidized. The oxidized form is now an oxidizer. An oxidizing agent, oxidizer, or oxidant is a compound that "accepts" an electron from another compound. In a redox reaction, the oxidizer gains electrons and becomes reduced. The reduced form is now a reducer.
For example, a cell may synthesize NADP+ de novo. NADP+ is an oxidizer and can be reduced to form NADPH. NADPH is a reducer. It can be oxidized to form NADP+ and turn an oxidized compound (GSSG) into a reduced compound (GSH). Redox reactions always happen in pairs. It's often easiest to check logic by identifying which compound is the electron donor or acceptor, then assigning labels based on the reaction direction of interest.
NADP+ (oxidized form, is oxidizing/gets reduced) ↔ NADPH (reduced form, is reducing/gets oxidized)