Project

Self-assembly of oligomeric proteins into fractal supramolecular assemblies as directed by phosphorylation.

Introduction

Figure 1. Three examples of protein assemblies, from left to right: triple peptide helices in collagen (N. Dilmen, Wikimedia Commons); icosahedron in adenovirus capsid (D. Goodsell, PDB-101); icosahedron in carboxysome (T. Yeates).

Protein supramolecular assemblies are formed when individual protein subunits bind to one another and form larger agglomerates. Protein assemblies have varied functions in nature (figure 1): collagen, a assembly formed from triple peptide helices, forms the connective tissue of animals [1]; the adenovirus capsid, a icosahedron formed from hexagon and pentagon shaped proteins, packages the viral genome and enzymes. [2] Protein assemblies are also used to optimize metabolic processes. The carboxysome, which is a protein icosahedron that contains carbonic anhydrase and ribulose-1,5-bisphosphate carboxylase oxygenase (RuBisCO), improves the carbon-fixing activity of RuBisCO by increasing the local concentration of dissolved carbon dioxide. [3]

Figure 2. Rationale for protein assemblies. In the above 2-step metabolic pathway, the substrate (orange square) is converted to the intermediate (circle) by the first enzyme (hexagon); the intermediate is converted into the product (star) by the second enzyme (teal square). In an free enzyme system, product conversion experiences lag since the intermediate needs to find the second enzyme. Forming an assembly between the two enzymes eliminates this lag. Data is simulated in graph.

A biotechnological application of protein assemblies is to optimize the catalytic efficiency of proteins belonging to a metabolic pathway. [4] Metabolic pathways are composed of several enzymes that process substrates through a series of steps until they reach their final product form. These pathways can be used to synthesize drugs, produce biofuels, and degrade harmful pollutants. Lag can occur in the pathway, since intermediates need to find the next enzyme in order to become processed (figure 2). This lag can be eliminated if the enzymes are assembled together into an assembly. As soon as the substrate enters the assembly, the intermediate can find its enzyme quickly, thus improving the overall efficiency of the pathway.

We intend to use protein assemblies to improve the efficiency of the atrazine degradation pathway.

Atrazine degradation by Pseudomonas

Atrazine is the most widely used herbicide in the United States, covering nearly 25% of total US crop acreage in 2000. Approximately 76 million pounds of atrazine is applied on US soil every year. [5] Since atrazine is stable and slightly soluble in water, atrazine contamination has been detected in about 75% of rural and urban stream water near agricultural areas. [6] Concern over the effects of atrazine on human health began in 2002, when Hayes et al. [7] demonstrated that male African clawed frogs (Xenopus laevis) larvae exposed to atrazine at low (≥1.0 ppb) doses formed hermaphroditic adult frogs. That study resulted in a class action lawsuit between Midwestern municipalities, who have been exposed to atrazine, and Syngenta, the manufacturer of atrazine. The suit was settled in 2012, when Syngenta paid $105 million to the municipalities. [8]

Figure 3. Atrazine degradation in Pseudomonas sp. ADP. Figure by N. Shiomi, InTech, A Novel Bioremediation Method for Shallow Layers of Soil Polluted by Pesticides.

Atrazine is degraded by the soil bacterium Pseudomonas sp. ADP in a six step metabolic pathway (figure 3). In the first step, AtzA dechlorinates atrazine to form hydroxyatrazine. Second, hydroxyatrazine is deaminated by AtzB to form N-isopropylammelide. Third, N-isopropylammelide is again deaminated by AtzC to form cyanuric acid. Cyanuric acid then enters a pathway catalyzed by AtzD, AtzE, AtzF where it is converted into ammonia and carbon dioxide. It is important to note that, depending on the triazine ring substituents, the first three metabolites in the pathway (atrazine, hydroxyatrazine, and N-isopropyl ammelide, for atrazine degradation) may still be harmful, while cyanuric acid is relatively benign and frequently added to swimming pools as a disinfectant. [9]

Our design

See experiments for more details.

Our goal is to apply protein assemblies to the problem of improving atrazine degradation. We plan to make the enzymes of the atrazine degradation pathway connect to one another to form a protein assembly. By co-localizing members of a metabolic pathway, we hope to optimize the overall efficiency of the pathway.

Figure 4. The symmetry of AtzA (hexagon) and AtzC (square) enable two connections.

First, AtzA and AtzC were chosen, since crystal structure data of both enzymes are available (PDB 4V1X and 2QT3, respectively). Since structural data of AtzB is limited, AtzB was not included in the assembly. AtzA forms a hexamer, and AtzC forms a tetramer. Their symmetries allow for divalent connections between each subunit (figure 4) in the resulting assembly.

Second, we chose an engineered super binder src homology 2 (SH2) binding domain as the means of connecting AtzA to AtzC. Binding domains are small modular proteins that bind to specific binding peptides. They are found in many cell signaling pathways. Binding domains and binding peptides can be genetically fused to proteins (AtzC and AtzA, respectively, in our application) in order to create fusion proteins that bind to each other.

The SH2 domain is a binding domain that binds to its respective binding peptide when a tyrosine residue on the peptide is phosphorylated. Kaneko et al. [10] performed directed evolution and discovered that three mutations (Thr8Val, Ser10Ala, Lys15Leu) in wild type Fyn SH2 domain (PDB 1FMK) resulted in a 75 fold improvement in binding affinity (\(K_d =\) 5.2nM compared to 0.39μM). For this reason, we decided to use this ‘super binder’ triple-mutant SH2 domain in our protein assembly. Additionally, phosphorylation dependent binding enables us to design protein assemblies that are triggered by the presence of phosphorylation.

Based on computational modeling, the binding peptide was fused onto the N-terminus of AtzA, while the binding domain was fused onto the C-terminus of AtzC. Modeling using the protein design suite Rosetta was used to select for stabilizing mutations. Several non-clashing mutations were made on the fusion protein genes. Resulting designs were expressed, purified and analyzed for phosphorylation by kinase, binding to each other, assembly formation activity, maintenance of enzymatic activity, disassembly induced by dephosphorylation, and structure.

Additionally, we developed computational simulations to model possible modes of assembly formation and enzymatic kinetics in assemblies.

Summary of key results

Figure 5. Overview of phosphorylation-dependent assembly. AtzA with SH2 peptide (hexagon) is phosphorylated by a kinase. It forms a assembly with AtzC with SH2 domain (box), which can propagate to form a fractal agglomerate. Disassembly by dephosphorylation can occur with incubation with phosphatase.

See the results page and the discussion page for more details.

We succeeded in developing a phosphorylation-dependant protein assembly of the atrazine degradation pathway.

1. Assembly formation is phosphorylation dependant.

Figure 6. DLS between phosphorylated AtzA and AtzC (green) and non-phosphorylated AtzA and AtzC (red).

Dynamic light scattering (DLS) estimates the hydrodynamic radius of particles in solution. The peak at 10nm represents the individual subunits, while the peak at 1000nm represents the formed assembly. Assembly formation only occurs when the AtzA subunit is phosphorylated, suggesting that the peak at 1000nm does not represent non-specific protein precipitation upon mixing the two components.

2. Assembly formation can be triggered by antagonistic kinase / phosphatase activity.

Figure 7. DLS for assembly before phosphatase treatment (green), and after (blue).

As the assembly uses a phosphorylation-dependant SH2 domain, we reasoned that dephosphorylation should result in disassembly. Incubation of the assembly overnight in YopH tyrosine phosphatase at 4U/μL results in complete disassembly, as detected by DLS.

3. Molecular imaging suggests assembled enzymes form a hyperbranched fractal-like structure

Figure 8. Helium-ion microscopy of assembly (left) and diffusion limited fractal assembly simulation (right; rendered in PyMOL).

Helium-ion microscopy (HIM) of the assembly has resulted in a ~2μm diameter particle. We developed a computational simulation of the assembly, which forms a globular agglomerate. This is similar to the globular agglomeration seen in the HIM image.

4. Activity assays indicate assembly is composed of folded, active protein

Figure 9. Activity of assembled (red) and unassembled (green) AtzA (above) and assembled (red) and unassembled (green) AtzC (below).

AtzA catalyzes the degradation of atrazine, which is measured in the first assay. AtzC degrades N-isopropylammelide, which is measured in the second assay. As these assays suggest, both AtzA and AtzC retain activity while they are assembled. This indicates that the assemblies are not misfolded protein aggregates, but active, folded protein catalysts.

Future work

There are several directions we can take this project.

First, we wish to investigate whether the size of the assembly can be controlled, either by partial phosphorylation of the AtzA hexamer or by doping in free SH2 domain. The result may be a mid-sized assembly that can be detected using DLS, and structurally characterized further.

Second, we wish to complete the pathway by including AtzB into the assembly. Since structural data on AtzB is limited, incorporating AtzB into the assembly will require optimization. We may choose to use different binding domain-peptide interactions (e.g., SH3, PDZ) between AtzB and AtzA/C in order further our control over the assembly.

Third, further structural characterization can be performed on the assembly. Fluorescent microscopy can reveal the phase conditions of the assembly, and cyro-electron microscopy can resolve the overall three-dimensional structure of the assembly.

This project can be extended to provide us with a deeper understanding of the design principles for creating and controlling protein assemblies using chemical stimuli such as phosphorylation, and can be used to optimize atrazine degradation in vivo.

A plant or a soil microbe could be engineered to express the assembly in vivo, along with the kinase. The kinase could be activated by a G-protein coupled receptor that is activated by the presence of environmental atrazine. The result is a atrazine degrading complex that only forms when the organism is exposed to atrazine. This is beneficial, since constitutive recombinant protein production is a metabolic burden and recombinant proteins and their assemblies are often harmful to the cell.

References

1. Bella J, Eaton M, Berman HM. 1996. Crystal and molecular structure of a collagen-like peptide at 1.9A resolution. Science 266: 75-81.

2. Reddy VS, Natchiar SK, Stewart PL, Nemerow GR. 2010. Crystal structure of human adenovirus at 3.5 A resolution. Science 329: 1071-5.

3. Kerfeld CA, Sawaya MR, Tanaka S, Nguyen CV, Phillips M, Beeby M, Yeates TO. 2005. Protein structures forming the shell of primitive bacterial organelles. Science 309: 936-8.

4. Chen AH, Silver PA. 2012. Designing biological compartmentalization. Trends Cell Biol 22: 622-70.

5. LeBaron HM, MacFarland J, Burside OC eds. 2008. The triazine herbicides: 50 years revolutionizing agriculture. Elsevier: Oxford, UK.

6. Gilliom RJ et al. 2006. The quality of our nation’s waters: Pesticides in the nation’s streams and ground water, 1992-2001: U.S. Geological Survey Circular 1291.

7. Hayes TB, Collins A, Lee M, Mendoza M, Noriega N, Stuart AA, Vonk A. 2002. Hermaphroditic, demasculinized frogs after exposure to the herbicide atrazine at low ecologically relevant doses. Proc Natl Acad Sci USA 99: 5476-80.

8. Berry I. 2012, May 25. Syngenta settles weedkiller lawsuit. The Wall Street Journal.

9. De Souza ML, Wackett LP, Sadowsky MJ. 1998. The atzABC genes encoding atrazine catabolism are located on a self-transmissible plasmid in Pseudomonas sp. strain ADP. Appl Environ Microbiol 64: 2323-6.

10. Kaneko T, Huang H, Cao X, Li X, Li C, Voss C, Sidhu SS, Li SS. 2012. Superbinder SH2 domains act as antagonists of cell signaling. Sci Signal 5: ra68.