Aromatase inhibition affects brain gene expression in a cohort of common marmosets
Letrozole is an adjuvant therapy, meaning that this oral drug is a secondary treatment offered to post-menopausal breast cancer patients after a primary form of treatment (e.g., chemotherapy) in order to avoid the recurrence of cancer. This drug works by blocking the enzyme aromatase, which converts androgens (like testosterone) to estrogens (like estradiol) in the brain and body. By blocking aromatase and decreasing peripheral levels of estrogens, it can prevent or slow the progression of estrogen-sensitive cancer cells.
Image description: a cartoon graphic of the typical aromatization of androgens into estrogens like estradiol (E2).
However, aromatase inhibitors (let’s call them AIs, for short) are accompanied by severe side effects in people with ovaries (a gender inclusive term instead of implying that only women have ovaries when any gender can have ovaries.) These symptoms can often be severe enough to deter patients from taking this potentially life saving treatment. In addition, most postmenopausal people who take Letrozole will take the drug once per day for several years or until the cancer returns.
Image description: a cartoon graphic depicting a dark-skinned Black person surrounded by the most common side effects of longterm Letrozole treatment.
Researchers have previously shown that AIs like Letrozole recapitulate some of the same side effects that we see in humans (e.g., thermoregulation issues and cognitive impairment) in primates. Gervais et al (2019) showed how common marmosets treated with Letrozole for 4 weeks show impairments in spatial working memory, sex-specific reductions in thermoregulation, and even changes in the intrinsic excitability of hippocampal neurons!
What’s more, the animals treated with Letrozole showed an unusual region-specific increase in estrogen levels—namely the hippocampus, which is involved in learning and memory. We wondered if these behavioral changes were accompanied by gene expression level changes in the brain and, if so, would we find sex differences that partially explain why female animals were uniquely affected? Would we also be able to understand the mechanism(s) by which this AI increased estrogen in the hippocampus?
Image description: gene set enrichment analyses (GSEAs) highlight that female transcripts are unique enriched for steroid hormone biosynthesis. This is in alignment with Gervais et al.’s (2019) results, which found E2 synthesis in the brain to be an effect driven by female marmosets.
Image description: gene set enrichment analyses (GSEAs) highlight that Letrozole-treated female animals show unique enrichment for genes involved in Alzheimer’s disease when compared to Letrozole-treated males (Edwards et al., 2022).
Image description: a cartoon graphic of the mechanism of “aromatase inhibition”, where letrozole—the aromatase inhibitor—blocks the enzyme from synthesizing more estrogens.
Image description: A prototype of a marmoset brain matrix. We 3D printed this matrix in order to carefully dissect frozen brain tissue and extract the hippocampus for RNA-sequencing—a technique which allows researchers to look at the expressed genes (or transcripts, mRNA) in a given tissue or cell type of interest. For the SolidWorks design of this matrix, please see the supplementary materials section of the manuscript or email me for the file(s). It costs about $20 to 3D print.
After isolating our brain region of interest—the hippocampus—we set out to extract quality RNA, measure our concentrations and quality of that RNA, prepare cDNA libraries, and sequence our samples on the NovaSeq 6000. This is considered the “wet lab” portion of the work because it necessitates working in a laboratory with appropriate personal protective equipment or PPE. Once we get our data back, we transition to a “dry lab” workflow which involves programming in R, python, and bash scripting while using a high performance computing cluster (UNITY; https://unity.rc.umass.edu/).
While this may sound challenging, please remember that I did not have any experience in coding or RNA work before my PhD. These skills were learned during graduate school and it’s true that anyone can learn them and understand them.
Image description: A graphic from the manuscript which outlines the workflow for this project. A) Schematic featuring the species and brain region of interest for bulk RNA-sequencing (RNA-seq); B) the “dry lab” portion of the work, which includes taking our fastq files from the sequencing facility and aligning those to a reference genome, counting our transcripts or “reads”, testing for differentially expressed genes (DEGs), and running further functional analyses of those DEGs. C) DEGs among Letrozole vs Control groups. The dotted lines delineate different quadrants of interest. Any transcripts which fall to the upper left quadrant are significantly downregulated in Letrozole-treated animals. Any transcripts located in the upper right quadrant are significantly upregulated in Letrozole-treated animals.
After isolating our brain region of interest—the hippocampus—we set out to extract quality RNA, test our concentrations and quality of that RNA, prepare libraries, and sequence our material on a NovaSeq 6000. This is considered the “wet lab” portion of the work because it necessitates working in a laboratory with appropriate personal protective equipment or PPE. Once we get our data back, we transition to a “dry lab” workflow which involves programming in R, python, and bash scripting while using a high performance computing cluster (UNITY; https://unity.rc.umass.edu/).
While this may sound challenging, please remember that I did not have any experience in coding or RNA work before my PhD. These skills were learned and it’s true that anyone can learn them and understand them.
We showed for the first time that the AI Letrozole results in sex- and treatment-specific changes at the level of gene expression in the brain. This work is highly preliminary, which means that we cannot make any blanket statements or conclusively say what exactly is happening! As hard as that is to hear, this work still has value in that we can now generate new hypotheses and test them to advance this area of research. Science can affirm—but not definitively prove—phenomenon in the natural world. We test hypotheses for greater understanding, but our understanding will always be limited as there is always so much more to learn and understand about any given topic. What’s important is that we remember our science is here to improve our collective understanding and serve others.
Some additional questions that remain are:
Which specific cell types in the brain are most impacted by Letrozole treatment?
Potential tools to address this question: scRNA-sequencing to look at DEGs by cell type!Which regions of the genome are most accessible after chronic Letrozole treatment?
Potential tool to address this question: ATAC-sequencing to look at regions of chromatin accessibility!How does Letrozole treatment affect the function of cellular subtypes in the brain? Especially microglia (reflected in some of our top GO terms from our bulk RNA-seq data?)
Potential tools to address this question: (1) functional assays in cultured or primarily cells, which are much easier for isolating specific cell types (e.g., using FACS) and testing aspects of their function (e.g., microglial migration, phagocytosis, etc) in vitro using conjugated fluorescent markers for cell types and genes of interest; (2) electrophysiological recordings of neurons (primary or cultured neurons, brain imaging or live recordings, etc)What is the mechanism by which AIs increase estradiol (E2) in the hippocampus?
Potential tools to address this question: (1) Metabolomics (or smaller scale, ELISA) to measure estrogen metabolites and other estrogens and steroid hormones in the periphery and brain. (2) Measure if Letrozole at specific doses actually crosses the blood brain barrier in your model of choice (e.g., rodents, primates, humans). (3) Knock out models which eliminate specific receptors or proteins (e.g., the STS pathway, which is a pathway for making estrogen metabolites and is frequently upregulated in response to cancer drugs in peripheral tissues.) (4) Organoids (from donors with different sex chromosomal organization) that resemble the hippocampus; (5) so many more potential models and/or experiment ideas! Feel free to reach out for more ideas or guidance if you are a trainee studying this topic in the future!Understanding the landscape of CYP450 superfamily proteins—the enzymes which metabolize nearly 80% of all drugs we take and xenobiotics which enter the body—in humans and other species’ brains. We have such modest understanding of these CYP genes and their expression in the brain and peripheral tissues and how they have been conserved and/or diverged across time.
Potential tools to address this question: Immunofluorescence of CYP genes involved in drug metabolism and steroid synthesis (like CYP2A6, which metabolizes Letrozole, or CYP19a1 (aromatase) which synthesizes estrogens like estradiol (E2). It would be great to identify where they are in the brain, in which cell types, and under what conditions or states is the expression most robust (e.g., stress, disease, age, sex, etc).; It would also be incredible to see more genomic analyses comparing CYP genes and their conservation/function across time.It is also poorly understood whether or not the absence of estrogen in the periphery (due to menopause, estropause (rodents), gonadectomy, or AI-treatment) influences brain levels of estrogen and on what timescale (across aging trajectory, menopause transition, etc). It is possible that AIs indirectly (peripheral mechanisms) or directly (cross BBB) impact the brain. This is an exciting area of study and I hope future trainees will be passionate about better understanding the relationship between aromatase inhibition, immune function, estrogen synthesis, and drug metabolism in different species, sexes, brain regions, tissue types, and cell types. It is cool from a basic biology perspective, and it has a real impact for patients suffering from AI side effects!
Image description: cytoscape pathway analyses show a slight increase in cytochrome p450 superfamily genes which are involved in estrone (E1) metabolism. This may reflect compensatory mechanisms by which AIs like Letrozole increase estrogen in the brains of marmosets.
