Behind the paper: Tumors produce factors that promote extramedullary hematopoiesis

In this ‘behind the paper’ post, Derek Barisas discusses how analyzing some old data shaped a hypothesis and led them to find the factors derived from tumors that promote extramedullary hematopoiesis.

My name is Derek Barisas, and I am an MD/PhD student at Washington University School of Medicine in St. Louis. Our new paper in PLOS Biology focuses on the mechanisms underlying the expansion of hematopoiesis into the spleen in the context of a mouse model of breast cancer. Our project is part of a growing field interested in extramedullary hematopoiesis in adult animals and contributes to a wider appreciation of the relevance of extramedullary hematopoiesis to a broad spectrum of clinically significant diseases. For those unfamiliar, hematopoiesis primarily occurs within the marrow of the bones in adult animals. However, in times of stress, the body is able to expand the hematopoiesis to secondary sites. This expansion of hematopoiesis to sites outside of the bone marrow, extramedullary sites, can be helpful to produce enough immune cells to fight an infection but is thought to be pathologic when occurring for a prolonged period of time, like in malignancies.

Though I must admit, I was a bit of a late comer to this particular party. Extramedullary hematopoiesis was something that was only memorable in my pre-clinical medical education as a factoid about the nature of “blueberry muffin” rashes that accompany prenatal TORCH infections, a group of pre-natal infections that share a similar disease phenotype that includes a rash from a form of cutaneous hematopoiesis. So, if you had told me that I was going to spend my PhD studying the process, I would be a little surprised to say the least. But, as I came to learn more about it, I felt like I had stumbled upon an underappreciated gold mine.

The project started out, as most good projects do, with a strong phenotype where we were trying to understand the profound bias towards neutrophils in the peripheral blood in our genetic model of breast cancer. Our lab is interested in emergency myelopoiesis, and some work had been done on this model before I had joined. The idea for my project was to take my background in single cell RNA-sequencing data analysis and apply it to this question. One unexpected problem made this plan much more difficult: the COVID-19 pandemic shut down our lab within 3 months of me starting. So, for the next few months, I had to move the project along with only the previous data collected by the lab, my series of single cell RNA-sequencing data, and the scientific literature.

Image credit

Derek Barisas, CC BY 4.0

Two pieces of data and a pair of 30-year-old papers jump started the project. The first was an old serum cytokine screen. Within it, Leukemia Inhibitory Factor (LIF) was a stand-out cytokine upregulated in the tumor model. While looking through the single cell RNA-sequencing data, I noticed a small population at the very end of the cluster table. This population intrigued me because it expressed KITL, the key hematopoietic stem cell growth factor. But the most exciting part was that this population was the only one that expressed the cognate receptor for LIF. It felt like a great hook, but I needed something to tie it together. This is where the two papers came into play. Thirty years ago, they discovered that LIF affected hematopoiesis, but not in the bone marrow where I had been looking with my single cell RNA-sequencing data. Instead, the paper suggested that the spleen was the location where hematopoiesis expanded in response to LIF.

Here, I had to take my first jump. These papers did not use modern methods for identifying hematopoietic stem and progenitor cells. So, to determine whether LIF had any function in expanding hematopoiesis, I had to do mouse experiments. Luckily, COVID-19 restrictions had been lessened by this time, so I could finally test my hypothesis: LIF injections would increase hematopoietic stem and progenitor cells in the spleen. I was nervous. I had the backing data and the support of some high-quality publications, but I needed some experimental data to know that I had a working hypothesis. When the positive data came off the flow cytometer, I was ecstatic.

But there was a problem, the genetic model we started with was cumbersome to use, so I needed a new platform. There I was lucky because our lab worked directly above an expert in this particular model: Dr. David DeNardo. With his help, we transitioned to a transplantable cell line model his lab had developed. At the end of my first experiment, I was nervous, but that feeling dissipated quickly. Without measuring, I could tell from the first spleen that they were much bigger than usual and that probably meant my new model was inducing extramedullary hematopoiesis. With this tool in hand, we began to fill out important supporting data for our hypothesis around LIF and extramedullary hematopoiesis.

As data around splenic hematopoiesis and LIF accumulated, a question started to percolate into our discussions: “What was going on with the hematopoietic stem and progenitor cells in the spleen during this process?”. Here we turned again to single cell RNA-sequencing but decided to incorporate splenic cells in comparison to the bone marrow cells. At the end of the analysis, we were shocked to see TNFα expression specifically in the splenic hematopoietic stem and progenitor cells. Now we had a new question: how was the tumor driving this expression change? So, we thought about upstream regulators of TNFα that are known to be important in our mouse model of breast cancer. This line of thinking led us to the cytokine IL-1α which can be released upon cell death as epithelial tumors grow beyond their blood supply and is an initiator of inflammation in the context of sterile tissue injury. To our excitement, IL-1α proved to be an important player for extramedullary hematopoiesis in our model too.

The last major piece of data to tie up the project was looking up the two cytokines that appeared important to extramedullary hematopoiesis, IL-1α and LIF, on the The Cancer Genome Atlas RNA sequencing dataset. Even on just a preliminary analysis, it was clear: IL-1α and LIF were upregulated in many human cancer types as well as our model. That felt great to me because, as an MD/PhD student, I wanted my work to be as relevant as possible to human disease.

Image credit

Derek Barisas, CC BY 4.0

To top off this journey, publishing in PLOS Biology was a great experience. The editorial staff and the peer reviewers were instrumental in making our paper as powerful as it is now. It was also great to be on the forefront of changes to make science more accessible to other researchers and to interested members of the general public.

These anecdotes are just the beginning of the twists, turns, excitements, and setbacks that lead to our paper. We hope everyone reading it enjoys it as much as we did in making it!

About the author

Derek Barisas is an MD/PhD student in the lab of Kyunghee Choi at Washington University School of Medicine in St. Louis. 0000-0003-0707-2659