Self-organizing multicellular structures designed using synthetic biology

Synthetic genetic circuits can induce cells to form simple 3D structures reminiscent of those generated during early embryonic development. This advance will help engineers build tissues that have desirable structures.
Jesse Tordoff is in the Computational and Systems Biology Program, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA.

Search for this author in:

Ron Weiss is in the Departments of Biological Engineering and of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA.

Search for this author in:

The structures of living organisms have properties that any engineer might hope to recreate. They can self-heal, grow and adapt, and they can have an astonishing range of material properties, from the strength of bone to the lightweight flexibility of an insect wing. To make these structures, a fertilized egg follows a developmental program — a set of instructions for cell behaviour encoded in its DNA. If we could understand and control the development of biological shapes, then we could harness the properties of living structures to build better organs in vitro and to generate designer materials that could mimic some of the abilities of living organisms. Writing in Science, Toda et al.1 present a method for creating synthetic, designable developmental instructions, paving the way for researchers to engineer customizable biological shapes.

It has been proposed that all that is needed to make the diverse structures of the animal kingdom is a small set of fundamental tools — about ten shape-changing operations, including cell death, adhesion and movement2. To decide which of these actions to use, cells can communicate with each other to establish their relative positions.

Toda et al. used an engineered cell-communication system called synNotch3 to mirror this biological set-up. SynNotch is adapted from Delta–Notch signalling — a signalling pathway found in nature, in which cells that have membrane-spanning Notch receptors sense Delta proteins on the surface of neighbouring cells. An intracellular effector domain is cleaved from Notch following ligand binding, and moves to the nucleus to regulate gene expression. In synNotch, the natural core of the Notch protein is used, but the ligand that is sensed and the effector domain that responds are customizable. In this way, it is possible to create multiple channels of modifiable cell–cell communication. With the appropriate choice of ligand and effector, the system can act independently of native Delta–Notch signalling to drive cell behaviour in customizable ways.

The authors engineered the cells so that the synNotch sensors regulated the expression of genes that encode cadherin proteins, which have long been known for their ability to create spatial organization in tissues. Cadherin proteins mediate cell–cell adhesion, and so are essential for holding cells together and creating tissue boundaries during development4. Much like oil and water, cell populations that have different patterns or levels of cadherins can sort themselves into separate groups after being mixed together, and can self-assemble into a range of structures in vitro5,6.

To create a synthetic program to guide shape formation, Toda et al. built several genetic circuits composed of different synNotch sensors that, when activated by a neighbouring cell, drive the expression not only of different levels or types of cadherin, but also of different ligands to bind to other sensors. In addition, each sensor drives the expression of a gene that encodes a fluorescent protein (green, red or blue), so changes in cell organization can be easily visualized. The authors mixed together cell populations harbouring these different circuits and allowed them to communicate and move freely. They found that engineered communication between the cells led to cadherin-driven cell rearrangement, which in turn led to different cell–cell interactions, producing cycles of communication and shape change (Fig. 1).

Figure 1 | Synthetic genetic circuits generate self-assembling structures. Toda et al.1 have built circuits that, when expressed in different combinations in a population of cells, caused the cells to self-organize into various structures. In this example, the authors used two circuits. Initially, one group of cells (labelled type 1) expressed a blue fluorescent protein, and the other (type 2) did not fluoresce. When the populations were mixed, a ligand (designated A) produced by type 1 cells activated a receptor (sensor A) on type 2 cells, leading to cleavage of an intracellular effector domain from the sensor. This domain moved to the nucleus to trigger the expression of genes encoding ligand B, the adhesion protein E-cadherin and the protein GFP, which made the cells fluoresce green. E-cadherin caused type 2 cells to adhere to one another, rearranging the cell population. Ligand B then signalled to nearby type 1 cells, activating a different sensor (B). This led to the expression of a protein (RFP) that caused red fluorescence and low levels of E-cadherins. Because low E-cadherin production made the cells somewhat adhesive, they formed a second ring of cells. In this way, the circuits produced cycles of cell–cell communication and self-organization.

Toda and colleagues observed remarkably complex cell behaviours. Cells self-organized to generate 3D structures, including a bullseye pattern and a sphere surrounded by multiple smaller nodes of different colours. The researchers could design instructions to produce specific structures, such as asymmetric forms — a key part of embryonic development. Furthermore, they showed that a structure of nested spheres could regenerate after being cut in two, as is often the case for self-organized tissues in living organisms.

The researchers next built a circuit to generate differential gene expression in a population of cells that was initially identical — a process that mimics cell differentiation. To do this, they designed synNotch circuits to emulate one feature of the native Delta–Notch system known as lateral inhibition, in which Notch, when activated by Delta from a neighbouring cell, inhibits the expression of Delta in the receiving cell. This signalling produces a chequerboard pattern of two distinct cell populations, one expressing Notch, the other Delta, from an initially uniform population.

In the authors’ lateral-inhibition circuit, one of these cell populations produced a green fluorescent protein, the other red. In addition, the two effector domains also promoted the production of different levels of the protein E-cadherin. In this way, the group was able to generate a structure that had rings of colour starting from a single uniform cell population.

With this work, Toda et al. have shown how we can design developmental programs to make new living shapes. Of course, there are limits to this approach. The authors’ biggest structures are only a few hundred micrometres across, and adhesion-driven self-organization alone is unlikely to generate structures of the size or complexity of organs. But advances in other types of synthetic-biology shape control could help to fill in some of the gaps. For instance, cells have been generated that can be artificially polarized such that asymmetric cell–cell contacts can be made7, and synthetic circuits have been designed to modify the behaviour of bacteria so that, across a whole population, arrangements are formed that resemble Turing patterns8. These patterns — such as stripes, spirals or the spots on a giraffe — arise during development as a result of biological signalling programs.

In the future, the toolkit established by Toda et al. could be expanded to generate short- and long-distance cell–cell communication alongside a synthetic system that controls all of the shape-changing operations involved in making biological structures. This could eventually give engineers total control when designing shapes that have some of the properties of living multicellular organisms. Such a development would be a huge advance. Not only could we map the rules of developmental biology by establishing the limits and constraints of shape-changing biological operations, but we could also grow replacement organs and make adaptive living materials — for example, buildings that could construct and heal themselves.

Nature 559, 184-185 (2018)

doi: 10.1038/d41586-018-05564-5


  1. 1.

    Toda, S., Blauch, L. R., Tang, S. K. Y., Morsut, L. & Lim, W. A. Science 27, eaat0271 (2018).

  2. 2.

    Davies, J. A. J. Anat. 212, 707–719 (2008).

  3. 3.

    Morsut, L. et al. Cell 164, 780–791 (2016).

  4. 4.

    Halbleib, J. M. & Nelson, W. J. Genes Dev. 20, 3199–3214 (2006).

  5. 5.

    Nose, A., Nagafuchi, A. & Takeichi, M. Cell 54, 993–1001 (1988).

  6. 6.

    Cachat, E. et al. Sci. Rep. 6, 20664 (2016).

  7. 7.

    Loza, O. et al. eLife 6, e24820 (2017).

  8. 8.

    Karig, D. et al. Proc. Natl Acad. Sci. USA (2018).

Download references

Nature Briefing

An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday.