In This Issue
December 1, 1997 Volume 27 Issue 4
Fall Issue of The Bridge on Bioengineering.

Engineering Cell Function

Monday, December 1, 1997

Author: Douglas A. Lauffenburger

Combining the capabilities from molecular and cell biology with the engineering analysis of complex systems permits cellular-level problem solving.

When one looks at the macroscopic structure of living systems, what one is really seeing is the cooperative activity of cells of different types communicating by different molecules. What sorts of cell functions are going on? A myriad. Metabolism, producing energy, proteins, lipids, and so forth; adhesion, causing attachment among cells and between cells and the polymers they produce for structural integrity; proliferation, which is cells replacing themselves; migration, or movement from place to place; mechanical contraction, for things like the beating heart; and signaling, or communication within and among cells. And then there is differentiation, which determines the set of properties a cell takes on at any point in time.

There are three essential aspects of a cell. One is the genes, which determine the cell's potential. Human cells encode more than 100,000 genes. At any one point in time, about 90 percent of them might be silent, and only about 10 percent of them actually might be expressed as proteins. So, something is telling the genes to turn on or to turn off. These signals come from the functional proteins - enzymes and receptors - which represent the two other essential aspects of the cell, in consort with the structural proteins and lipids.

Enzymes, which carry out metabolism, motor function, and so forth, are regulated by the receptors and expressed by the genes. The receptors, which sit on the cell surface, communicate with the cell interior through chemical ligands, cytokines, growth factors, cellular matrix proteins, and the environment. Not only are the cells a chemically responsive system, they're mechanically responsive - you can cause a signal to be sent by putting mechanical stresses on the cells. All of these responses work through the same biochemical pathways. Thus stimulated, receptors send signals to the genes, shutting some off and turning others on, and send signals to the enzymes, telling them to turn up or down their activities.

What I have described is a triangular, interactive regulatory system and, from an engineering point of view, there are three ways to modulate the system's behavior. You can do genetic engineering to alter the genes; you can use pharmacology to potentially alter the enzymes; and you can use "biomaterials" to present different types of ligands to the receptors. All three of these approaches are unified through cooperative regulation.

To show how rapidly our understanding of these regulatory networks is growing, let me point out one important class of signaling enzyme, the kinases. Kinases add phosphate groups to other proteins and by so doing alter their activities. Over the last 20 years, we have gone from essentially not knowing what any of these were to identifying almost 100 of them. In the next 10 years, with the human genome being sequenced, we will likely characterize many more. Having identified these enzymes involved in function, we will want to know how they work together. A single cell has feedback interaction with all the molecules inside and outside it, with all the other cells in the same tissue, and with cells in the other tissues.

As scientists, we would like to understand how cell functions are governed by molecular properties. Can we take an integrated systems approach to these molecular mechanisms? Can we quantify some of these activities? Can we model them? This is what we have to do if we are to do a better job of the science. As engineers, if we understand some of the science in an integrated, quantitative way, can we make cells do what we want via molecular manipulations? The molecular manipulations might be pharmacological, they might involve biomaterials, or they might entail genetic engineering. These manipulations have a tremendous number of applications, from functional genomics, diagnosing diseases at the molecular and cellular levels, therapeutics involving cytokines and growth factors, to gene therapy and cell-based therapies, including tissue engineering.

In functional genomics, we have to figure out what proteins the genes turn on and what the proteins do, how they affect cell function. The idea with gene therapy is that if something is wrong, you can fix it by putting in a new gene and having the cells in the body express that gene. There are many diseases that potentially can be treated this way, but the science and technology are challenging. In tissue engineering, there are many different approaches to creating vascularized tissue in vitro. The general idea is to put cells in a polymer matrix, where they will organize into functioning tissue - endothelial cells, liver cells, hepatocytes. But, what are the rules by which you do that? How do you bring that about?

The central question is, can we determine how to alter a molecular property - through a gene modification, a biomaterial, or a drug - in order to obtain a desired change in cell function? This is not easy. Or, rather, it's easy to change molecules, but it's very difficult to predict what's going to happen when you do.

Take something seemingly simple, such as trying to produce a therapeutic protein by inserting the protein's gene into cells that lack it. One would think that if we put 10 copies of that gene in, we should get 10 times as much protein produced. But, it has been found that often the more copies of the gene you put in, the less protein you get out, which is not what one would intuitively expect (Wittrup et al., 1994). This illustrates well the notion that biology is rarely a single-step process.

Cellular Process Not Single Step
Here's another example. Suppose you want to make growth factors to use for cytokine therapy to stimulate cells to proliferate or migrate. You might think that if you could make a growth factor that would bind even tighter to the cell-surface receptors than the natural protein, it would give a stronger signal. It turns out that this can be exactly the wrong answer. If you do protein engineering and actually decrease how well growth factor binds to its receptor, the result can be an increased rate of cell proliferation (Reddy et al, 1996). Why is that? Again, the cellular response is not a single-step process. But, if you look at it as an integrated system of many, many processes occurring simultaneously, you can start to understand it.

So, to make a change in a molecule is relatively easy; to understand what the cells are going to do once you've made the change is hard. And this is exactly where engineering thinking is extremely valuable. We can ask the right questions. For instance, if you wish to alter a cell's behavior, what are the heuristic rules and correlations? What are the design principles? Can we develop a useful model for this process? If so, what are the variables and what are the parameters? In the end, if you can come up with this information and create phase diagrams, or phase plots, it should be possible to predict how you would alter a cell's function by altering its molecular properties. This is what many of us are defining as "cell engineering" (Nerem, 1991).

The goal is to change cell function by changing molecules, which also is the goal of biological scientists. The trick is that the engineering way of thinking adds an extra dimension to the approach. What we want to know is, as for any engineering system, can the output of the system, in this case cell function, be predicted based on the inputs? The inputs are not now flow rates, temperature, and pressure, but molecular properties such as rate constants, binding affinities, and signaling rates. What we are doing is combining the capabilities from molecular and cell biology with the engineering analysis of complex systems.

The parameters that characterize molecular properties should be very familiar to engineers. They include thermodynamic, kinetic, mechanical, and transport measures. One can characterize the behaviors of molecules in terms of concentrations, equilibrium affinities, binding and coupling rate constants, signaling rate constants, and mechanical strengths and compliances. If you decide that these things are important and you can measure them, you can then change them rationally. This is the beauty of modern molecular biology. It serves as the experimental methodology for altering parameters in engineering models for cell function.

Not only does systems integration matter, not only are there multiple steps, but it turns out that numbers matter, concentrations and rates matter. It is not just a matter of determining what a single molecule is going to do. Quantitative changes in molecular properties can lead to qualitative changes in cell functions, or phenotype as it is known in biology. Just because a cell now looks like it's behaving differently than it was before doesn't mean that something fundamentally structural has changed. Rather, it may be a quantitative difference in some of the key molecular properties.

I would like to illustrate this point by considering the phenomenon of cell movement. The phenotype of a cell may appear to be motile or immotile, migratory or nonmigratory. Take a vascular endothelial cell, a cell that sits in your blood vessels. If you wanted to colonize a biomaterial for an artificial vessel or heart, these are the cells you need to mobilize to crawl onto the material surface and cover it up. In contrast to nonmigratory endothelial cells, the migratory phenotype puts out new membrane extensions, attaches its receptors to proteins on the surface, makes mechanical connections intracellularly, and pulls itself forward. The forward movement requires detachment of the adhesion receptors on the cell surface.

In our lab, we've tried to manipulate cell migration systematically by changing the number of copies of adhesion receptors, which sit in the cell membrane and bind extracellular proteins such as fibronectin and fibrinogen. Our thinking was that if you change the number of receptors, you might change the balance of mechanical forces, thereby affecting cell speed. We have also varied the levels of the extracellular matrix proteins fibronectin and fibrinogen as well as their thermodynamic parameter of interaction (i.e., their binding affinity). What we have found is that numbers matter, not just the number of receptors but also the number of proteins in the environment.

What we have come up with is a model and design principle that makes a prediction for the effects of all those parameters. Our model takes into account that a cell has on its surface adhesion receptors that bind to the proteins in the environment. These, in turn, are connected to the cytoskeleton, where the actin/myosin motors are located that generate an internal force. The mechanical properties of the cell, in terms of viscosity, elasticity, and so forth, depend a lot on the molecular arrangements. Some of the force exerted by the cell is dissipated in the cell mechanics, but some of it is transmitted to the substructure, where it creates traction, allowing the cell to move.

Mechanics-Chemistry Connections
There is an intimate connection between mechanics and chemistry that is going to show up over and over again in biology. In the example of cell motility, the rate constants for how the receptors bind to the proteins depend on the strength of the bond. As this force is increased, the bonds will dissociate faster. With simple models like this, you can come up with design principles that explain the relationship between the molecular property and the cell function. From our model of cell movement, a simple parameter ratio arises. If you plot cell speed as a function of that parameter ratio, everything collapses to a single curve. This ratio is analogous to classical engineering dimensionless groups like the Reynolds number, except that it happens to have in it some key molecular properties of the cells. Using this model, you can predict how fast the cell is going to crawl, no matter how many receptors you put in it, no matter how many matrix proteins you put on the surface, no matter what the binding affinity (Palecek et al., 1997).

The same numbers game applies in another case I'd like to describe, this time related to cell proliferation. As is true in the case of cell motility, it is the signal transduction dynamics, essentially the chemical pathways affecting and affected by key regulatory genes, that determine whether a cell will proliferate or not. Jay Bailey, a pioneer in this field, has developed a model for this cellular process that takes into account the signaling going on, enzymatic cascades turning genes on or off, and the resulting proteins that tell the cell it's time to move into and through the proliferation cycle (Hatzimanikatis et al., 1995).

Of course, Bailey has not included all of the hundreds of reactions involved in this process. We don't know them all. And, as engineers, we are used to dealing with incomplete information. What he and others have done very effectively is to parse out of the system some of the key elements. The result is very interesting: a phase diagram, or phase plot. By plotting the levels of one of the regulatory proteins versus a rate constant for how fast something is generated in a signaling pathway, he can predict when a cell should proliferate and when it shouldn't.

Engineers solve complicated problems like this. It's our great strength. We think in terms of inputs and outputs, parameters, models, variables, predictions, and phase diagrams. But now we are trying to apply our methods to biology at the molecular and cellular levels. Twenty years ago, we did not have the technology to access biological systems, to change them, to design them at the molecular level. Today, with modern molecular biology and cell biology, we have the tools to carry out these modifications. We can apply engineering thinking to the fascinating systems presented by biology, and by so doing benefit both technology and science.


  • Hatzimanikatis, V., K. Lee, W. Renner, and J. E. Bailey. 1995. Biotech. Letters 17:669.
  • Lauffenburger, D. A., and J. J. Linderman. 1993. Receptors: Models of Binding, Trafficking, and Signaling. New York: Oxford University Press.
  • Nerem, R. 1991. Cellular engineering. Ann. Biomed. Eng. 19:529.
  • Palecek, S., J. Loftus, M. Ginsberg, D. A. Lauffenburger, and A. F. Horwitz. 1997. Nature 385:537.
  • Reddy, C., S. Niyogi, A. Wells, H. S. Wiley, and D. A. Lauffenburger. 1996. Nature Biotech. 14:1696.
  • Wittrup, K. D., A. S. Robinson, R. Parekh, and K. Forrester. 1994. Annals of the New York Academy of Sciences 745:321.
  • About the Author:Douglas A. Lauffenburger is professor and director of the Center for Biomedical Engineering at the Massachusetts Institute of Technology. This paper is based on remarks he made at the 1997 NAE Annual Meeting Technical Symposium, held 8 October.