A synthetic multicellular system for programmed pattern formation

(Mitja Crček)

A synthetic multicellular system for programmed pattern formation

Subhayu Basu, Yoram Gerchman, Cynthia H. Collins, Frances H. Arnold & Ron Weiss

Nature 434, 1130-1134 (28 April 2005)

INTRODUCTION

Pattern formation is a frequently observed behavior in the living world. Many years ago scientists started to examine a pattern formation in prokaryotes and later also in eukaryotes. Here we focus on bacterial patterns because their formation has lower complexity and is therefore easier to understand. Microbiologists found that in nature, bacteria have to deal with many different environmental conditions. To increase the viability of the single bacteria, bacterial colonies are formed. To do that, bacteria had to develop complex communication pathways. Those networks are nowadays well studied and therefore synthetic biologists could start researching synthetic multicellular systems for programmed pattern formation.

Here we first explain the basics of cell-cell communication using acyl-homoserine lactone (AHL) as signal molecule so the pattern formation process will be easier to understand. We describe Tet regulatory system, lux operon, lac operon and lac repressor to explain the molecular basis and show the role of AHL.

In the second section we explain how the signaling process works in engineering cells and how the output protein is produced. We describe the molecular network in sender cells and receiver cells. The correlation between various AHL concentrations (high, medium or low) and output protein production (e.g. GFP) is shown as well as the role of repressors and activators. We show why only the receiver cells that are at appropriate distances from the sender cells can express GFP.

In the third part we focus on pattern formation. We first describe the construction of plasmids for sender and for receiver cells. We explain how bacterial strains for experiments were prepared and show simulations and experimental result of their GFP production. Illustrated with photographs we describe circular pattern formation and explain why the system works like that. Last but not least we describe the basics of ring formation dynamics and show the design of other patterns.

AHL signaling

Cell-cell communication is the basis to explore the pattern formation. To understand the process of communication between bacterial cells using acyl-homoserine lactone (AHL), we will first discuss the Tet regulatory system, lux operon, formation of AHL and the lac repressor.

Tet regulatory system

Tetracycline-controlled transcriptional activation is a method of inducible gene expression. Antibiotic tetracycline or its derivates (e.g. doxycycline) are used to reversibly turn the transcription of the target gene on or off. Transcription is regulated by tetracycline transactivator (tTA) protein which is formed by fusing TetR (tetracycline repressor from E.coli) with activation domain of VP16 protein (from Herpes Simplex Virus).

There are two very similar Tet regulatory systems, Tet-OFF and Tet-ON system. Both systems are based on tetracycline transactivator which activates expression. The tTA in Tet-OFF system is active in absence of tetracycline (or analogs) and is therefore capable of binding to Tet promoter which leads to transcription of target gene. When concentration of tetracycline is high, it binds to tTA and tTA is no longer capable of binding to Tet promoter. The Tet-ON system works similarly, but in the opposite way. 4 amino acid mutations of tTA allow binding of transactivator to the Tet promoter only in presence of tetracycline (or analogs). In absence of tetracycline, the tTA in Tet-ON system is unable to bind to the promoter. The simplicity of this mechanism is one of the main reasons why Tet regulatory system is commonly used in both prokaryotic and eukaryotic systems [1].

Lux operon

The lux operon encodes genes for self-regulation and was first found in Vibrio fischeri. Regulation is based on the presence/absence of AHL and crucial protein regulator is transcription activator LuxR; see Figure 1 here [2]. Figure description: Quorum-sensing model in P. fischeri using lux operon. LuxI synthesizes AHL which causes dimerization and activation of LuxR (transcriptional activator whose gene luxR is under the constitutive promoter). Active LuxR can bind to pLuxR promoter which leads to transcription of other genes in lux operon, also luxI (positive feedback). SAM = S-adenozilmetionin, ACP = acyl transfer protein, CRP = repressor protein for cAMP [2].

LuxR gene is under the constitutive promoter. Its protein product LuxR has a binding site for AHL and is inactive in absence or at very low concentrations of AHL (<10 nM). As seen in Figure 1 LuxR is activated when AHL binds and causes dimerization. Active LuxR can bind to pLuxR promoter and expression of other genes in lux operon can begin.

LuxI is one of the generated proteins in this process and functions as an AHL synthase. It converts S-adenozilmetionin to AHL which leads to higher concentration of AHL in the cell causing more LuxR molecules to become active. This is how positive feedback is formed. Expression of the luxR gene is regulated by several factors such as heat shock, catabolite repression (CRP), and even LuxR itself (at very high concentration of AHL) [2].

With new tools of synthetic biology we can create devices for protein expression using parts of lux operon. We can replace luxI, luxC, luxD, luxA, luxB and luxE genes with our gene of interest but leave the LuxR as transcriptional activator; see scheme here. Scheme description: Schematic LuxR receiver. Expression of luxR gene is under constitutive or inducible promoter. AHL activates LuxR which is luxR gene product. Active LuxR then causes a transcription of our gene of interest. RBS = ribosome binding site, T = terminator, LuxpR = LuxR promoter, PoPS-out = signal that comes out (e.g. GFP) and can be measured.

Therefore AHL can be used as signaling molecule which is, as you will see in the continuation of this page, the basis of programmed pattern formation.

Lac operon and Lac repressor

The lac operon or lactose operon is an operon with the task for transport and metabolism of lactose in E.coli and some other enteric bacteria. It has three structural genes, lacZ, lacY and lacA which encode β-galactosidase, lactose permease, and galactoside O-acetyltransferase, respectively; see figure here [3]. The lac operon uses a well-regulated control mechanism so that lac genes are expressed only when necessary (i.e. at high concentration of lactose or analogs). Transcription of the lac operon genes is suppressed when LacI repressor coded by lacI gene binds to lac operator.

The lac operon actually uses a two-part control mechanism of expression. In the presence of glucose, the catabolite activator protein remains inactive so the enzymes cannot be produced. That shuts down lactose permease to prevent transport of lactose into the cell because it does not need the lactose from the media.

The second control mechanism is the regulatory response to lactose which is regulated by LacI. In the absence of lactose, the lac repressor binds to lac operator and therefore halts production of the enzymes encoded by the lac operon. LacI gene is under a constitutive promoter so the regulation of lac operon is constant. When there is a high concentration of lactose in the cell, lactose is converted into allolactose which inhibits the lac repressor's DNA binding ability. This leads to transcriptional activation of the lac operon; see figure here [3]. Figure description: Regulation of the lac operon. In absence of lactose, LacI binds to lac operator which obstructs the RNA polymerase from binding to the promoter (top). This results in very low levels of mRNA encoding LacZ and LacY. Once present in the cell, lactose can convert to allolactose which binds to LacI. An allosteric change in shape of LacI result as inability of binding of repressor to operator region (bottom). 1: RNA polymerase, 2: repressor, 3: promoter, 4: operator, 5: allolactose, 6: lacZ, 7: lacY, 8: lacA [3].

All the structural genes can be replaced with our gene of interest so we obtain a mechanism which is easy to regulate and can produce the desired protein (e.g. GFP). Expression of lacI gene can be regulated with activators or repressors such as lambda repressor (cI). If we place the gene encoding for cI under the control of constitutive promoter (like pLuxR) we can control the expression of LacI and therefore the production of desired protein.

Signaling process and output protein production

Figure 1a in the origin article shows us the design of the synthetic bacterial multicellular system for a signaling process that can lead to pattern formation [4]. Figure description: A signaling process and GFP production. a, the correlation between various AHL concentrations (high, medium or low) and expression of cI, LacIM1, LacI and GFP. b, approximation of the AHL gradient as a function of the distance from the sender cells. c, signaling process. Only the receiver cells that are at appropriate distances from the sender cells can express the GFP [4].

We can see that only receiver cells at intermediate distances from sender cells express the output protein. The signaling process begins when tetracycline (or analog) is added to the sender cells which use Tet-ON regulatory system. Tetracycline binds to TetR (which is a positive regulator) and that has an impact on the expression of LuxI. LuxI is an enzyme that synthesizes AHL, which diffuses through the cell membrane and forms a chemical gradient around the senders; see Figure 1a.

AHL in media can easily diffuse into nearby receiver cells which use the Lux receiver, a part of lux operon. AHL binds to LuxR, a positive AHL-dependent transcriptional regulator, which activates the expression of two regulatory proteins, lambda repressor (cI) and modified Lac repressor - LacIM1. cI is a strong repressor from bacteriophage lambda and represses the expression of wild type LacI, which is under the constitutive promoter. LacIM1 is a product of a codon-modified lacI and has (in the receiver cells) the same role as LacI - to repress the expression of the output protein (in this case the green fluorescent protein - GFP).

Receiver cells that are close to the sender cells receive high concentrations of AHL, resulting in high cytoplasmic levels of LacIM1 and cI. Although cI disables the expression of LacI, the level of LacIM1 is high enough to repress the expression of GFP, resulting in absence of the mentioned protein. Receivers that are far from the sender cells have very low concentration of AHL or no AHL at all, resulting in inactive protein LuxR. That means there is no expression of LacIM1 and also cI, which is unable to repress the expression of wild type LacI, resulting again in GFP repression. In both cases we cannot detect the fluorescence of the cells (see Figure 1a).

At intermediate distances from the sender cells, intermediate AHL concentrations result in moderate levels of LacIM1 and cI. Because the repression efficiency of LacIM1 is significantly lower than that of cI, cI effectively shuts of the expression of LacI while the threshold concentration of LacIM1 required to repress GFP production is not achieved. This difference between repression efficiencies of LacIM1 and cI affords bacteria the non-monotonic response to AHL dosages.

Pattern formation

Plasmids for sender and receiver cells

To achieve the cell behavior described previously, vectors that contain a certain bio-bricks should be designed. Scientist from Princeton University and California Institute of Technology engineered vectors for sender and receiver cells and used them to reach a pattern formation in bacteria [4]. As you can see in Figure 1b, sender plasmid (pSND) contains luxI gene, which is under the control of Tet promoter. For the purpose of the experiment, cells containing Tet regulatory system were transformed with pSND, so they could express LuxI. To reach a specific expression of GFP they engineered two different vectors for receiver cells and named them high-detect (pHD{x}) and low-detect (pLD) plasmids (we will explain later why there are two plasmids).

High-detect plasmid (pHD{x} - see Figure 1c) contains lacIM1, luxR and gfp genes. They engineered three different high-detect strains (HD1, HD2 and HD3) to achieve the best pattern formation process. The HD1 strain contains a hypersensitive LuxR mutant, HD2 incorporates the wild-type LuxR, and HD3 strain expresses LuxR from a reduced-copy-number plasmid. Low-detect plasmid (pLD - see Figure 1d) contains cI and lacI genes and is crucial for non-monotonic response to AHL as you will see later.

Figure 1b, 1c and 1d description: Plasmids for pattern formation. a, plasmid map for sender cells. b, high-detect plasmid. c, low-detect plasmid. Three versions of the high-detect plasmid with different sensitivities to AHL were constructed (regions of mutation are underlined: pHD1, LuxR; pHD2, wild-type; pHD3, ColE1) [4].

Simulated and experimental behavior of engineered bacterial strains

Simulations and experimental results show how prepared bacterial strains behave in different AHL concentration (see Figure 2) [4]. Figure description: Simulated and experimental liquid-phase behavior of transformed cells. a, b, simulations (a) and experimental results (b) of the AHL dosage response for HD1 strain (hypersensitive LuxR, blue), HD2 strain (wild-type LuxR, red) and HD3 (a reduced-copy-number plasmid, black). c, d, simulations (c) and experimental results (d) of three strains BD1 (blue), BD2 (red) and BD3 (black), all three consist of the same pLD plasmid and different pHD{x} plasmids (pHD1, pHD2, pHD3, respectively).

In agreement with model predictions (Figure 2a) the dosage responses of all three HD strains (containing only pHD{x}) showed reverse correlation to AHL concentrations (see Figure 2b). The only but crucial difference between them is in their sensitivities. Note that all experiments were done with transformed E.coli strain DH5α. For more details of experiment conditions see Methods in [4].

To reach the non-monotonic response to AHL, it is required to use both vectors, high-detect and low-detect plasmid (Figure 2c and 2d). By combining the pLD with each of the pHD{x}, they obtained three different strains (named BD1, BD2 and BD3 in [4]). Those BD (band-detect) strains showed predicted non-monotonic response to different AHL concentrations with different thresholds (see Figure 2d). Results also correlated well with model predictions (see Figure 2c).

Circular pattern formation

Spatiotemporal simulations of a system predicted that by placing sender cells next to receiver cells, the described network could direct pattern formation on solid media. If we place sender cells in the middle of the plate, a circular pattern is expected to be formed. Furthermore, a bullseye pattern could be achieved by mixing different variants such as BD1, BD2 and/or BD3.

Experiments shows the formation of such pattern on a plate with BD3 or BD1 cells (with gfp gene) and BD2-Red cells (similar to BD2 with dsRed-Express replacing gfp gene). See Figure 3 [4]. Figure description: Experimental solid-phase behavior of described networks. a, photo of agar plate with the sender disk in the middle. b, bullseye pattern with sender cells in the middle, inner green fluorescent ring (made by BD3 cells) and outer red fluorescent ring (made by BD2-Red cells). c, another bullseye pattern made by mixture of BD1 and BD2-Red cells. Scale bar; 5 mm.

First, two different bacterial strains (BD3/BD1 and BD2-Red) were spread evenly on top of an agarose plate. Then, a disc containing sender cells, which express CFP (cyan fluorescent protein; GFP-like protein from jellyfish Aequorea macrodactyla [5]), was placed in the middle of the plate (Figure 3a). Those plates were incubated at 37 °C overnight. Bullseye patterns were captured with a fluorescence microscope. As seen in Figure 3b (BD3/BD2-Red experiment), BD3 cells formed a green fluorescent ring near the sender disk, whereas BD2-Red cells formed a ring further from sender cells. In the middle we can see a cyan fluorescent ring, created by sender cells, but no red or green fluorescent color. This phenomenon is the consequence of different sensitivities of BD strains to AHL (see the text above and Figure 2).

Similarly, when BD2-Red cells and BD1 cells were mixed and plated with a sender disk, green fluorescent disk was detected around the red fluorescent ring (see Figure 3c). Different diameters of the two BD2-Red rings (30 mm versus 22 mm) can be explained by variations of the AHL gradients due to differences in population of receivers and senders as well as in the growth rates. Subtle environmental differences between experiments such as agar densities and nutrient conditions can also affect different ring formation.

Ring formation dynamics

Understanding the dynamic behavior of the system is essential for predicting pattern formation. The ring formation activity of BD2 cells was measured over the course of 36 hours to study the system dynamics. Fluorescence was recorded every 90 minutes for a rectangular region protruding from the sender disk (see Figure 4a) [4]. Figure description: Results show the time-evolution of fluorescence (of GFP) for cells as a function of the distance from the sender cells. Dark blue: low fluorescence, yellow and green: intermediate fluorescence, red: high fluorescence.

For the first 15 hours no fluorescence was observed. Later, as we can see in Figure 4a, low levels of fluorescence (brighter blue and green color) emerged about 10 mm from the senders. Fluorescence was then significantly increased between 5 and 18 mm from the sender cells reaching a steady-state maximum at 10 mm.

Other experiments show that stability of LacI has the strongest correlation with fluorescence response times and positional shifts. As it was shown ([4]), the stability of LacI affects GFP expression so the LacI decay has an impact on fluorescence. Although dynamics is very important to predict pattern formation, it is not an issue of this topic; for further reading see [4].

Formation of other patterns

The final pattern and dynamics depend not only on the sender-receiver network but also on the arrangement of sender cells. It is possible to create many different patterns by placing multiple sender disks in various configurations. The number of sender cells (sender disks), their density and the distance between them lead to different AHL gradients reflected in different intricate patterns formation (see Figure 5). Figure description: Formation of different patterns. a, simulation of cell behavior on solid media with two sender cells on determined distance; formation of an ellipse. b-d, experimental results showing various patterns formed. Sender cells were expressing DsRed-Express and receiver cells GFP. Patterns are based on the placement and different concentration of sender cells. b, two sender disks; ellipse. c, three sender disks; heart. d, four sender disks; clover [4].

Summary of pattern formation

Pattern formation is a frequently observed behavior in a prokaryote world. To form a living pattern, cells had to develop communication systems. One of the well described is the AHL signaling system and that is why it was first used to create a synthetic multicellular system for programmed pattern formation.

As we have seen, sender cells use LacI to synthesize acyl-homoserine lactone (AHL). AHL is a small molecule which can diffuse through the cell membrane so it can behave as signaling molecule. Receiver cells uptake the AHL molecule which, in those cells, binds to LuxR. Activated LuxR causes the expression of cI and LacIM1 and those two repressors affect expression of the output protein (e.g. GFP). Sender and receiver cells can be prepared by molecular engineering using different vectors for each type of cells. Those cells are then used to create a pattern on solid media where the driving force for formation is AHL gradient. Time evolution of fluorescence can explain the dynamics of pattern formation, which shows the strongest correlation with stability of LacI. Formation of non-circular patterns is based on the placement and initial concentrations of sender cells and the color of patterns depends on used fluorescent protein.

References

[1] Z. Zhu, T. Zheng, C. G. Lee, R. J. Homer, in J. A. Elias, „Tetracycline-controlled transcriptional regulation systems: advances and application in transgenic animal modeling“, Semin. Cell Dev. Biol., Vol. 13, No. 2, pp. 121–128, apr 2002.

[2] J. E. González in M. M. Marketon, „Quorum Sensing in Nitrogen-Fixing Rhizobia“, Microbiol. Mol. Biol. Rev., Vol. 67, No. 4, pp. 574–592, jan 2003.

[3] „lac operon“, Wikipedia, the free encyclopedia. 20-nov-2014.

[4] S. Basu, Y. Gerchman, C. H. Collins, F. H. Arnold, in R. Weiss, „A synthetic multicellular system for programmed pattern formation“, Nature, Vol. 434, No. 7037, pp. 1130–1134, apr 2005.

[5] N.-S. Xia, W.-X. Luo, J. Zhang, X.-Y. Xie, H.-J. Yang, S.-W. Li, M. Chen, in M.-H. Ng, „Bioluminescence of Aequorea macrodactyla, a Common Jellyfish Species in the East China Sea“, Mar. Biotechnol., Vol. 4, No. 2, pp. 155–162, mar 2002.