Personalizing Cancer Recovery Through Reprogrammed Cells

Indika Rajapakse, professor of computational medicine & bioinformatics and mathematics at the University of Michigan, understands the devastating impact cancer has on lives. His wife and collaborator, Lindsey Muir, research assistant professor of computational medicine and bioinformatics, lost both of her parents to the disease. Driven by a commitment to advance treatment, his research with the (Genome + Cell) Reprogramming Lab is dedicated to “reprogramming” skin cells to regenerate bone marrow after chemotherapy.

Rajapakse calls this “the treatment for the treatment.” When a cancer patient undergoes high-dose chemotherapy, some healthy cells, including those in the bone marrow, can be damaged. The formation of blood cells takes place in bone marrow, and damage can decrease white blood cells needed to fight infection, red blood cells carrying oxygen throughout the body and platelets critical for blood clotting and wound healing.

To address this issue, E. Donnall Thomas, professor emeritus at the University of Washington and director emeritus of the clinical research division at the Fred Hutchinson Cancer Research Center pioneered methods of providing new bone marrow cells to patients through bone marrow transplantation, which won him the 1990 Nobel Prize in Medicine. Bone marrow transplantation involves extracting healthy hematopoietic stem cells (HSC), specialized stem cells that give rise to all different types of blood cells in the body, from a donor’s bone marrow. These healthy HSCs are then injected into a chemotherapy patient where they can repopulate blood cells.

Thomas conducted much of his pioneering work at the Fred Hutchinson Cancer Research Center in Seattle where Rajapakse was a postdoctoral fellow. Rajapakse had the honor of meeting Thomas, a key inspiration for his work. The Fred Hutchinson Cancer Research Center was also the site of another breakthrough that would impact Rajapakse’s work. The 1987 discovery of MyoD, a protein that tells cells to become muscle, demonstrated for the first time that the fate of a cell can be dramatically altered by a small set of genes.

Rajapakse, with collaborator Lindsey Muir and a team that includes Walter Meixner, Cooper Stansbury, Joshua Pickard and Sarah Lee, now aim to build on these previous discoveries by using a patient’s own DNA, focusing on directly reprogramming a patient’s plentiful and accessible skin cells to bone marrow cells. This is significant because 60-70% of bone marrow transplant recipients experience Graft v Host Disease (GVHD). This occurs when the transplanted donor immune cells (the graft) recognize and attack the recipient’s tissue (the host), causing inflammation and potentially damaging organs.

UMAP plot showing clustering of fibroblast, hematopoietic cell types, and
iHSCs

Moreover, African Americans and people of mixed race find it difficult to match for a bone marrow transplant. Human Leukocyte Antigen (HLA) matching is based on genetic markers inherited from ancestors. Finding someone genetically similar, with a good HLA match, decreases the likelihood of complications such as GVHD after transplantation. White Americans, despite having diverse European ancestries, have relatively lower genetic diversity and often share similar HLA profiles. For patients of mixed race however, it is more difficult to find a donor with the same genetic makeup. People with African ancestry generally have high genetic diversity, resulting in only about a 30% chance for an African American to find a donor, compared to 80% for a white person. This problem is compounded by the high cancer burden faced by African Americans.

“I want to humanize science,” Rajapakse explained. “This research is about helping people and not just advancing knowledge. I am from Sri Lanka, a developing country and I understand the burden expensive treatments place on those who cannot afford them. Our research is about making a real difference in people’s lives and eventually reducing treatment costs.”

At the heart of his work lies the concept of gene expression and cellular identity. Every cell in the human body contains the same genetic information—over three billion base pairs organized into 23 chromosome pairs, encoding more than 20,000 genes. However, not all genes are active in every cell. The specific combination of active, or expressed genes and inactive, or silenced genes determines a cell’s identity and function. For example, skin cells express a different set of genes than bone marrow cells, even though both contain identical DNA. Precise regulation of gene activity is key to manipulating cellular identity. By controlling which genes are active or inactive, Rajapakse and his team can guide the reprogramming process and manipulate gene expression to transform one cell type into another.

To achieve this, Rajapakse and his team are using computational science. They have developed an innovative algorithm that optimizes the use of transcription factors, proteins that regulate gene expression. This algorithm uses a data-guided approach to model cell dynamics based on gene expression data. By analyzing how genes are expressed in skin cells, it identifies the precise steps and optimal combination of transcription factors needed to convert them into bone marrow cells or any other cell type.

This process focuses on identifying the key regulatory factors that can influence a cell’s identity. Central to this approach is chromatin, a complex of DNA and proteins that form chromosomes. Chromatin organization plays a crucial role in controlling gene expression and is found in two forms: heterochromatin and euchromatin. In heterochromatin, the chromatin is tightly packed and condensed, making the DNA inaccessible. Euchromatin, in contrast, is loosely packed, making the DNA more accessible. The algorithm must identify how to remodel the chromatin structure, loosening it where bone marrow genes need to be activated and tightening it where skin cell genes need to be silenced, to fully reprogram the cell.

Genome-wide patterning of multi-way contacts. Incidence matrix visualization of the top 10
most common multi-way contacts per chromosome. Used with permission.

The potential of this technology is immeasurable. With the ability to reprogram cells into any type, these methods could be used far beyond bone marrow regeneration. It could be possible to regenerate heart muscle cells after a heart attack, create insulin-producing pancreatic cells for diabetes treatment, repair macular degeneration in the eye, accelerate wound healing or conduct any other necessary cell replacement due to aged or non-functioning cells.

The Rajapakse lab is equipped with a high-powered microscope capable of viewing the genes of a skin cell to identify cells capable of reprogramming. Those cells are then transferred to a DNA sequencer to determine the precise order of nucleotides (A, T, C and G) in the DNA molecule. However, this process destroys the original cells, and sometimes a large number of experiments must be run to gather enough data for conclusive outcomes. That is why the lab is equipped with the BioAssemblyBot 400 (BAB) robotic arm, known as “BAB in the lab.” BAB automates caring for and imaging of live cells and is capable of automated DNA sequencing and assembly of a bone marrow tissue model, including live cells and structures, using its biological 3D printer.

The potential uses of BAB are vast. Its biological 3D printer presents numerous possibilities for research ranging from creating printed human tissue for studying disease mechanisms and testing new therapies to the potential of printing functional tissue or even entire organs for transplantation. Equipped with live streaming cameras, data streaming of the sequencer, live cell imaging and BAB’s robotics-assisted automation, manual human management of experiments may become unnecessary. This capability not only enhances efficiency and precision but also creates the possibility for researchers around the world to use BAB.

“I would like to see labs at institutions with limited funding have the opportunity to use BAB,” said Lindsey Muir. “This would empower more researchers, including those with fewer resources, to utilize BAB’s advanced technology. With its capability for remote operation, BAB could create more equitable access to cutting-edge research tools, enabling important discoveries at institutions that lack extensive funding.”

In 2024 Rajapakse and his collaborators were awarded a grant from the Defense Advanced Research Projects Agency (DARPA) to develop an AI-assisted foundation model for direct cell reprogramming, known as the TwinCell Blueprint. This initiative aims to revolutionize the field by enabling rapid, AI-guided exploration of new methods for directly reprogramming one cell type into another. This research will radically accelerate discovery in understanding, predicting and modulating genome function through the development of predictive models. The team includes co-PIs Lindsey Muir and Alex Gorodetsky, associate professor of aerospace engineering as well as industry partners NVIDIA and Oxford Nanopore Technologies.

Front Left to Right: Indika Rajapakse, Lindsey Muir, “BAB”
Back Left to Right: Joshua Pickard, Walter Meixner, Jillian Cwycyshyn

The TwinCell Blueprint is designed as a comprehensive system that integrates BAB’s robotics-assisted wet lab automation, advanced sequencing and imaging technologies to build extensive datasets for cells undergoing reprogramming. This data is then integrated with a novel AI modeling approach to demonstrate direct reprogramming of human skin cells to blood stem cells. Their method pairs a “biological twin,” comprised of the actual cells with a “digital twin,” an AI model, in an adaptive learning loop. By simulating experiments on a computer, the digital twin allows researchers to explore new possibilities without the need for costly and time-consuming physical trials. The digital twin proposes new experiments and protocols, while data from the biological twin refines the model’s predictions. The model’s creativity will be assessed based on the range of viable solutions it generates.

This approach offers tremendous potential to benefit scientific disciplines that traditionally lack access to the large-scale datasets that have propelled advancements in text and vision-based AI-models. It paves the way for predictive dynamic models in biology and the medical fields, with the goal of significantly improving cellular reprogramming efficiency applications in regenerative medicine, cancer research, tissue repair, drug discovery and personalized medicine.

“Cellular reprogramming exemplifies the transformative potential of computational science and AI in medicine. Indika’s team is advancing cancer treatment while democratizing access to leading-edge therapeutic approaches,” said Karthik Duraisamy, Samir & Puja Kaul Director of MICDE. “This work represents exactly the kind of interdisciplinary innovation that will define the future of personalized medicine, bringing together expertise in computational mathematics, biology, engineering and computer science to solve critical healthcare challenges.”

The (GENOME + CELL) Reprogramming Lab is redefining cancer treatment by tackling disparities in healthcare. With 60-70% of patients suffering post-chemotherapy effects, Rajapakse’s approach offers solutions beyond traditional bone marrow transplants. By reprogramming a patient’s own cells, the lab offers a personalized approach, moving beyond one-size-fits-all solutions and addressing the diverse genetic landscapes across populations. With advanced AI and robotics, they aim to make personalized therapies accessible and affordable, providing hope for patients when they need it most.