Team:Johns Hopkins-Wetware/lightproject

Today’s increasingly complex research and manipulation of biological pathways poses a demand for rapid, controllable gain and loss of biomolecular function. Bioengineering of pathways for industrial processes is hindered due to lack of understanding of non-native proteins in the yeast chassis. Optimizing pathways and adjusting expression of relevant proteins is a tedious task. The 2012 JHU iGEM team set about developing a tool to facilitate optimization and controlling flux of pathways in order to maximize efficiency in manufacture.

Focusing on Optogenetics

Light-inducible proteins are a very modern focus of research, and even more recently a component in engineered systems. We considered PhyB, Cry2, CcaS and the TULIPs system (Strickland, et al. 2012), and any combination of the aforementioned. Finally, we decided on just the TULIPs system because of its relative advantages:

• - Immediate response to stimulus and lack thereof, much like an on- and off-switch,

• - Required only a single wavelength of light, whereas most others required two (on and off),

• - Proven to work in yeast (Strickland, et al. 2012),

• - Tunable - small mutations in the light-protein constructs can toggle sensitivity,

• - The proteins are small, compared to others. Increases likelihood that proteins can pass through nuclear envelope, or other organelles,

• - Does not require addition of any exogenous chemicals.

PhyB and CcaS required an exogenous cofactor called phytochromobilin, which does not eliminate the need to add chemicals and contradicts one of our goals. Alternatively, we could have transformed yeast with the genes required to make this chromophore, but we would not like for the success of our overall project to be dependent upon this. Also, a previous and unsuccessful attempt to synthesize PCB in yeast was discouraging. CcaS also could not be used because we could not find a yeast intracellular signaling pathway that would turn on only the gene we needed; its signaling mechanism also did not allow for precise temporal and spatial control. Cry2 was rather large, and did not have an immediate “off-switch”; it reverts to its inactive form at a late and imprecise time.

Theoretically, the system can be adapted to control loss or gain of just about any function in the yeast chassis. In manufacturing, this allows for light to precisely optimize production throughout an entire bioreactor. For example, if yeast produces a toxic biproduct, but the mechanism to degrade the biproduct is costly, the adapted TULIPs system can be used to turn on the degradation pathway. The functionality of the pathway might be optimized so that the toxin is removed with as little drag to efficency as possible.

TULIPs stands for TUnable, Light-controlled Interacting Protein Tags. The first of the “tags” is the photoswitchable LOV2 domain of AsLOV2, which is a phototropin from Avena sativa. When hit with 470 nm light, the J-alpha helical domain loosens from the rest of the protein, leaving it available for interaction with other molecules.

The engineered PDZ (ePDZ) protein is the other major component of this system. It will bind to a short peptide epitope (–SSADTWV–COOH) or certain truncations of this epitope when epitope is available. By fusing this short epitope to the C-terminal of the J-alpha domain, we can take advantage of LOV2’s active and inactive states to expose and hide (respectively) the epitope from the ePDZ domain. When the light shines, the epitope is immediately exposed to bind with its partner, ePDZ. When the 470 nm light turns off, the epitope, which is literally an extension of the J-alpha helix, snaps back to the LOV2 domain, dissociating from ePDZ.

One of the advantages of TULIPs that makes it favorable to other systems is that it is tunable. In Strickland (2012), various ePDZ mutants and truncations of the short peptide epitope were tested. Caging and binding ratios varied pre- and post-excitation. Although we only chose to use one ePDZ variant, we made our choice by reviewing characterizations of each variant. We based our choice on a) proportion of change between caging ratios pre- and post-excitation, b) how little binding there is in the absence of photoexcitation (closer to control is best), and c) how high the incidence of binding is. We chose ePDZb1, and decided to try out other mutants when we had found some of our experimental constructs successful.

Strategies

We employ three main strategies to photo-switchably turn the functionality of a given protein on and off:

1. 1. Splitting the protein and fusing each portion to either LOVpep or ePDZ, and dimerizing the protein with photoexcitation, thus becoming a whole functional protein.

2. 2. Fusing each “tag” to different proteins. When the tags are dimerized, function and intracellular transport of at least one protein is inhibited.

3. 3. Tethering LOVpep to the plasma membrane by fusing it to a transmembrane domain at the N-terminus, as was done in Strickland (2012), and fusing the whole target protein to ePDZ. Upon photoexcitation, the cytosolic fusion protein is sequestered by the membrane-bound LOVpep, rendering it functionally silent.

See sections titled "Activation of Protein Function" and "Deactivation of Protein Function" to understand how we target proteins in further detail.

In the case of first strategy, we obviously hope to dimerize and turn on the protein upon photoexcitation, but there are two options to consider within this strategy. When the cell already expresses the protein, we can choose to keep that protein and modulate between normal expression and overexpression, or we can knock down the protein and so that the pathway is only induced by photoexcitation.

Controlling the cell cycle

As proof of concept, we apply these strategies to proteins that regulate the cell cycle in yeast. Many studies report cell cycle arrest when the gene is overexpressed or knocked down, so by copying these expression patterns with our constructs, we expect to arrest the cell cycle at the expected stage.

The proteins whose functionality we are modulating are listed with the method we use to induce arrest. Click here to see the proteins.

LED light tool

In order to activate our light proteins, we put together a circuit that turns on LED light. It emits wavelengths of about 470 nm. We intend to use the tool in rooms with otherwise low levels of light, so that the light components will not be photoexcited without turning on the LED light tool.

To test the protein activation system, we selected endogenous yeast cell cycle checkpoint proteins whose overexpression is known to arrest the cell cycle with the idea that blue light stimulation should activate the function of these proteins and cause the cell cycle to arrest. Although there is no guarantee that two separately-transcribed parts of a protein will be a functional unit when dimerized, we were encouraged by a similar and successful experiment that used CRY2 and CIB1 as the light-inducible components and Cre recombinase as the modulatable protein, whose halves were each fused to one of CRY2 and CIB1 (Tucker CL, et al. 2010). We better our chances of success by attempting multiple splitting points, and avoiding functional domains of the protein. We also try to split away from the ends of the protein, so that one part might not be functional without its partner.

We designed and constructed two 'acceptor vectors' which can be used to easily express the N or C-terminal halves of a protein as a fusion with LOVpep or ePDZ and a fluorescent tag for localization tracking. We have co-transformed a total of 9 unique split protein expression vector pairs into yeast and have confirmed expression through fluorescent microscopy. Currently, we are beginning the first light-induction experiments to see if cell-cycle arrest is achieved.

We envision that optogenetic protein activation could be used to rapidly and reversibly change the flow through a biosynthetic pathway. For example, one could activate a protein to break down a toxic intermediate at a certain level, balancing the flow through the pathway and the stress on the cell.

To use the cell cycle to test light-inducible loss of function, we selected two sets of proteins endogenous to yeast: one set of 'cell-cycle' proteins essential for the progression of the cell cycle, and one set of 'tether' proteins with specific cellular localization patterns. By fusing ePDZ to a cell-cycle protein and LOVpep to a tether protein, we expect that upon blue light stimulation, binding between ePDZ and LOVpep will relocalize the cell cycle protein to the tether protein's location, such that the cell-cycle protein cannot perform its function and the cell cycle will be arrested.

Alternatively, we may relocalize a protein by fusing one protein to each of the "tags," and upon dimerization, one protein's functionality will be inhibited. Either the other protein directly blocks it from functioning, or the other protein prevents it from transporting to other parts of the cell. For example, a protein may be prevented from leaving the nucleus when it is paired up with its partner.

We have constructed 7 yeast strains with unique combinations of cell-cycle and tether genes tagged with LOVpep/ePDZ and a fluorescent protein for tracking cellular localization. We have confirmed expression of both fusion proteins and observed the correct sub-cellular localization of our tether proteins through fluorescent microscopy. Currently, we are beginning light-induction experiments to see if protein re-localization is observed and if cell-cycle arrest occurs.

Here we have presented this optogenetic system as a tool to improve the manufacture of important chemical compounds. In an industrial setting, a fermentation vat could be programmed to optimize flow through a biosynthetic pathway simply by turning a blue light on or off. However, this system has many other important applications. This system allows quick, reversible, spatio-temporal control of protein function, and is orthogonal in yeast.