Forgetting dyskinesia: a case for levodopa-induced motor memories in Parkinson disease
Since 1970, the best treatment for Parkinson disease (PD) has been to replace the loss of dopamine due to the neurodegenerative process, by providing a drug that is a precursor to dopamine, called levodopa. While levodopa is very helpful for improving PD symptoms in the short-term, long-term treatment induces development of uncontrollable movements, known as levodopa-induced dyskinesia (LID), in 50-70% of PD patients within 5-10 years after treatment commencement. Once established, these involuntary movements progressively become worse, suggesting a “priming effect” or the acquisition of an aberrant form of motor memory.
Viewing motor dysfunction as aberrant motor memory opens a new line of discussion towards the characteristics of LID development in PD. When we think about motor learning and memory, at a circuit level, we are speaking of adaptation within different brain regions that undergo structural and functional plasticity to adapt their neural connections. These adaptations reflect on the control of motor skills that both essential for survival and involve fine behavior, such as driving a car or playing an instrument. Altering neural connections requires novel synthesis of proteins. In the context of motor memories, novel synthesis of proteins is facilitated by dopamine. Specifically, in the context of PD, the loss of dopamine receptor stimulation is thought to lead to a loss of the ability to modulate neural connections to form new motor memories. When levodopa treatment replaces lost dopamine, it reverses this effect, allowing for normal movement and motor memories to occur. However, as the loss of dopaminergic neurons continues, larger doses of levodopa are necessary to overcome the deficit and these are often delivered in a cyclical fashion: A dose of levodopa is given when parkinsonism is prevalent and dopamine is low, the levels of dopamine quickly increase and a benefit is observed, then dopamine levels fall and parkinsonism returns, necessitating another levodopa dose. This cyclical release of dopamine is thought to be associated with LID onset, though precisely how remains a mystery.
Perhaps the large swings in dopamine leads to an exacerbation of the normal role of dopamine in the formation of motor memories, whereby the peaks and troughs of dopamine release induce the development of motor memories that become excessive and non-specific, creating an overabundance of formerly useful movements disassociated from their normal behavioral cues. Due to these overlaps between motor and behavioral memory, we wondered: Does striatal dysfunction in LID mimic the processes observed in normal learning and memory processes? If we approach LID like a bad memory, can we find ways to cause the brain to “forget” its previous treatment history to prolong the usefulness of levodopa for PD?
Based on work by researchers in the learning and memory field, the physical basis for memories appears to stem from a specific subset of cells that are activated by specific stimuli and influence each other to elicit a specific response. These cells are known as “engram cells”. Though originally investigated in the hippocampus, engram cells have been observed in other brain regions and may be activated and elicit specific responses to a variety of stimuli, including fear, reward, and drugs of abuse. We now know a lot about these cells: We know how they are initially recruited, how they alter their structure and function, and how they alter their connections with other cells as a stimulus-response relationship is solidified. Our line of research is an extrapolation from what we know about engram cells in the context of hippocampal learning and memory, and asking whether the same principles could apply in the context of LID.
Since that initial brainstorm, our short-term goal was to lay the foundation for thinking about LID as a form of bad motor memory by trying to figure out what cells could be storing this memory. Could there be “engram cells” for LID? Past research has shown that the striatum is pivotally involved in both motor memory and the development of LID. But the striatum is made up of many different cells, including neurons, supporting cells, and immune cells. Which ones could be storing the memory for levodopa? The primary goal of our research was to create a map of all the changes in gene expression of striatal cells in a mouse model of PD from the first exposure to levodopa and how that response evolves with repeated levodopa treatments.
We identified gene expression changes that were occurring in over 100,000 individual cells during dyskinesia development using single nuclei RNA sequencing. This data revealed that many of the most significant changes were observed in dopamine D1 receptor-expressing medium spiny neurons (D1-MSNs), one of the major components of the striatal output circuit. We found that some of these D1-MSNs were expressing genes indicating that they were activated by levodopa, as well those necessary for creating new neural connections. These genes were nearly identical to those expressed by hippocampal engram cells. Furthermore, we noticed that initially many D1-MSNs were activated by levodopa treatment; however, after repeated exposures, the number of these activated D1-MSNs was reduced. Although this seems counterintuitive, this also occurs when you learn something new: many cells are required to initially form a memory and produce the correct response; however, as you get better at recalling the memory, fewer cells are necessary to quickly retrieve it. Perhaps these D1-MSNs serve as “engram”-like cells for LID, encoding the deleterious response to levodopa treatment.
We hope this work leads to a change in mindset in how the research community approaches LID, and perhaps other types of movement disorders, as the result of aberrant “motor memories” and use what we know about how the hippocampus functions in learning and memory to inform our research into movement disorders. Our future goals are to investigate what is driving these differences in gene expression. In addition to providing the instructions for how to make a protein, there are also multiple mechanisms that regulate when to make those proteins. For example, our RNA map indicated that the gene Inhba was enriched in activated D1-MSNs when we compared acute levodopa treatment to repeated treatment. This gene encodes for a subunit of Activin A, a protein that has been linked to hippocampal learning and memory. To test whether blocking this protein’s function could be therapeutically relevant, we administered an Activin A receptor antagonist alongside levodopa in our animal model. We found that this strategy blocked development of LID, indicating that this may be a viable therapeutic target for impeding the dyskinesia. In ongoing studies, we will identify which regulatory regions of DNA become active following levodopa treatment, and how these regions contribute to the development of the memory for treatment. By building on our RNA map, we want to create a corresponding atlas of the DNA regulatory regions active in individual cells across the striatum. We hope this will allow us to understand what molecules are turning these DNA regulatory regions “on”, with the hope that we can block these molecules and erase the motor memories that are formed in specific engram-like cells following levodopa treatment. If our work is fruitful, we expect that we can prolong the treatment of PD with levodopa, as other researchers search earnestly for a cure to the neurodegenerative aspects of the disease.
Karen L. Eskow Jaunarajs, PhD is an Assistant Professor at the Center for Neurodegeneration and Experimental Therapeutics, Department of Neurology at the University of Alabama at Birmingham
Note: Dr. Jaunarajs will present with Dr. David Standaert on this blog topic as part of the Research Spotlight 2025 series on December 2, 2025. Register here.
Ideas and opinions expressed in this post reflect that of the author(s) solely. They do not reflect the opinions or positions of the World Parkinson Coalition®