IBM Research: How to build a better cell through engineering

There’s nothing better on a hot day than a cold beer. Hefeweizen? Lager? Maybe a pilsner? It’s a matter of yeast that gives all beer its flavor. Yeast is the single-celled organism that makes an otherwise simple list of ingredients, complex. And it’s not just for common brewing or baking. Because different strains perform in different ways – some fermenting at high temperatures, other at cold temperatures – yeast is the one of the most-studied organisms in the world.

If we were to take a look at the microscopy, we would see that these yeast strains are morphologically different. Their internal organization differs, sometimes dramatically. The effect of that is reflected in the way the organism performs specific functions. In this case fermentation. A paradigm of cell biology is that the alterations of cell morphology are connected to alterations of chemical reactions inside the various cell structures, called organelles. Changes in the architecture of the cell will correspond to a difference in cell behavior, with either suppression or modification of existing cellular capabilities, or even the emergence of new functionalities.

Engineering a cell

There is, however, little mechanistic understanding of the links between cell morphology and cell function. Evolution has provided all living organisms with a genetic code that contains all the information necessary for each and every cell to function properly, from single cell organisms, like yeast, to multi-cellular, incredibly complex higher organisms like us humans.

Yet, decoding all the information contained in the genetic code is still an impossible task, today. Tens of thousands of genes are activated at specific times during the existence of just one yeast cell, and even more chemical reactions will occur during its lifespan. Hundreds of protein species will inhabit the cell, providing the building blocks of life. Monitoring this activity is a task that cannot be tackled with today's technology. Fundamental understanding of such an important part of the life of our cells needs a paradigm shift.

We need a new discipline: cellular engineering.

It is evident that such a task cannot be accomplished by a single research institution. Exceptional experimental capabilities need to go hand in hand with top-of-the-line computational and analytical tools. This is why the University of California, San Francisco, the University of California, Berkeley, Stanford University, San Francisco State University, the San Francisco Exploratorium, and IBM Research have joined forces. The collaboration wants to design, build, and test models of cellular organization. And ultimately develop cells that can perform specific functions.

Think of cell morphology as a proxy for cellular state and function. We can study it through microscopy methods, which are orders of magnitude cheaper than genetic screens, and provide enough accuracy to be used routinely in almost any laboratory. But building relationships between morphology and functionality requires extensively targeted experimental efforts, where cells, like beer yeasts, need to be subject to specific stimuli, and then imaged.

At the same time, it requires computational capabilities that need supercomputing-level power, and need new analysis tools to extract meaningful information, and learn from the wealth of data that will be produced. Furthermore, informed mathematical models are needed to quickly generate novel hypotheses from principles of cellular organization that can be tested in the laboratory. Finally, molecular information of a cell – its chemistry, genetics, proteomics -- will be needed in order to generate the ontology necessary to engineer cells that perform specific functions.

Mapping of the entire cell structure can be broken down to useable information using techniques like PCA and LDA, then machine learning is applied to understand the parsed data.

Modeling the morphology of cancer

The model example of the usefulness of fundamental understanding of the connection between cell morphology and cell behavior is cancer pathology.

When a doctor suspects that a patient has a tumor, the doctor will perform a biopsy: the removal of tissue samples of that are imaged and reviewed by a pathologist. The doctor then relies on an extensive set of image correlations to existing cases to compare the shape of the patient’s organelles (a cell’s wall, nucleus, mitochondria, and other internal parts) to those imaged from tumor tissues at different stages of progression. The correlation in the morphology becomes a correlation in functionality: the patient’s cells, if indeed cancerous, will replicate indefinitely, providing the hallmark behavior of any tumor.

This is, in many cases, an art, and it relies on invasive techniques, the expertise of the doctor, the state of the tumor, and different levels of understanding of the involved cellular mechanisms. A general mechanistic understanding of the morphological transformation processes does not yet exist, but it could dramatically improve the efficiency of diagnosis, and even reduce the need for invasive tests and further genetic screens. Moreover, cellular engineering could help introduce innovative therapeutic measures, as protection of other cells by aggregating and dividing labor between cells, or creation of new complex organizations for novel bio-materials to protect against cancerous cells.

The perfect cell for the perfect job

What is our goal with this research? If we know how cells adjust their morphology, we could control these mechanisms, predict their performance, and increase their range of capabilities. Engineered cells could be used to monitor the activity of bioreactors, devices to carry out specific chemical processes in a lab environment. NASA recently used advanced bioreactors to grow tissue samples of liver, muscle, cartilage and bone. Scientists could soon have an endless reservoir of tissues to test their research hypothesis, with a clear impact on the future of bioengineering and therapeutic sciences.

The disposal of organic pollutants are increasingly done by bioremediation, or using specific enzymes to convert waste into an output that is non-toxic and can serve another purpose within its environment. Specific detectors monitor the production of the product to generate useful information by proper modulation of the enzymes’ internal morphology. We can also build novel organic toxicity sensors by sensing cells’ morphology, and understanding their modifications.

This isn’t science fiction. The spirotox test estimates the toxicity of a volatile compound, like a gas, by monitoring cell organelles’ deformation after toxic exposure.

These sorts of sentinels cells, engineered to respond to changes in composition or concentration, could also be used to prevent food adulteration and protect people from food fraud. Or, more efficient biofuels can be engineered through reprogramming cell morphologies to maximize the efficiency of energy production.