Structural Determination of StAhpC2 and MpGlpO

By: Callia K. PaliocaMentor: Dr. P. Andrew KarplusDepartment of Biochemistry & Biophysics, OSU HHMI Summer Program 2010

Structural Studies of StAhpC2 and MpGlpO

Thank you Kevin. Good morning everyone. I am excited at the opportunity to share with you all what I have been researching throughout the summer. Thanks to the HHMI summer program and Dr. P. Andrew Karplus here at Oregon State, I have been working to determine the structures of two enzymes, ones that are found in human pathogens and that are a target for the development of new drugs against diseases.1Add collection of crystal photosProteins are IMPORTANTEnzymes are vital in all of lifes functionsMost drugs interact with proteins

ActinMyosinThese proteins, as are all proteins, are very important. It is an array of enzymes that allow you to do everyday functions. The proteins actin and myosin, for example interact by sliding past one another to produce the movement in muscles, movements we depend on. Drugs are also part of the use in proteins. One of the most common drugs is aspirin. This works by blocking the actions of proteins, called cyclooxygenases, that produce hormones that cause inflammation, headaches, and other aches. It is proteins that the drugs interact with.2Protein Structure MattersFunction flows from structureStructure-based drug design

HIV protease drug complexSo these functions we see in all these important proteins, the functions are directly related to the structure. Thus, if you can determine the structure of a protein, it allows you to understand the function better. Therefore, the study of the structure (involving the active site, the conformations, the catalytic cycle) is essential in studying pathogens to create a drug against them. An important example of structure-based drug design was in HIV. By understanding a critical viral enzyme in the HIV life cycle, HIV protease, scientists were able to model the active site and design inhibitors specific to the target region. Seeing how the substrates bind allow us to see how the structure is affected. Using this information helps us to understand how the organism can survive and how we can manipulate its defenses.3Leave in Salmonella poisioning etc?X-Ray Crystallography MethodsGoal: Obtain single, high-quality diffracting crystals and use the data to solve the structureGrow CrystalDiffraction & Data ProcessingSolve the Phase ProblemElectron DensityMapRefinement of StructureThe main way a structure is solved is by x-ray crystallography, which basically involves making crystals of proteins, freezing them, and then shooting them with x-rays. We can take the information from our detectors and model a structure. The process can be broken down into 5 steps which I will be reiterating later: obtaining a single and high-quality diffracting crystal, exposing it to an x-ray source to get a diffraction pattern and then processing the data, solving the so-called phase problem, getting an electron density map, and finally building and refining the structure.4Food Poisoning

Macrophage attacking bacteria

The first drug target I am working to solve the structure for is a protein involved in food poisoning. The organism responsible for all aches and uncomfort is Salmonella typhimurium. This little bacterium has seen its share of the headlines, causing many recalls, including the recent one with eggs. When this organism infects the body, our immune systems sends out its armies in the form of macrophages. These incredible machines shoot out pseudopods that capture the bacteria and destroy it with peroxides. This is the same defense you use when you pour hydrogen peroxide on a cut to kill microbes. Some of the bacteria can evade our macrophages because they too have defenses, in the form of proteins called the peroxiredoxins.5

StAhpC2 and the PeroxiredoxinsCatalytic cycle of this family of proteinsFunction: uses cysteine residues to breakdown peroxides

Functionally, these proteins breakdown peroxides with the use of cysteine residues, one of the many building blocks of a protein. The catalytic cycle of a typical Prx is shown here. This cysteine, known as the peroxidatic cysteine, is oxidized to sulfenic acid by the reduction of peroxide. Now for the protein to handle another molecule of peroxide, it needs to recharge. As water is released, the two sulfur atoms form a disulfide bridge. The cycle comes around with the help of a disulfide reductase. In the presence of peroxide from the human immune system, the enzyme continues in this catalytic cycle. However, if there is an increase in peroxide, some enzymes undergo what is called the inactivation shunt. Here the sulfenic acid is further oxidized to sulfinic acid with the reduction of a second molecule of peroxide. In a peroxide-rich environment, sensitive enzymes will go into this shunt and become inactivated, that is they will not be able to further reduce peroxide. Robust enzymes, on the other hand, will continue in this cycle even though there is an abundant source of peroxide. 6if inactivation shunt is reversible question

How Motifs Relate to SensitivityCorrelation between YF and a helical structureLeads to sensitive enzymesSensitiveRobustThe structural basis to the difference between robust and sensitive enzymes is currently being researched. Sensitivity has been linked to certain motifs, or specific sequences of amino acids. There is a correlation between a YF motif, consisting of the amino acids tyrosine and phenylalanine and a helical structure (shown in top left of picture). Those without these two pieces lead to a more looped region. This region is key in the formation of the disulfide bond in the catalytic cycle. If there is a helix with the YF motif, there is less room in which the cysteine residues can interact. However, a looped region gives plenty of opportunity for these two to come together, especially since this area (pointing to the loop) is more disordered than the helix. Therefore, it is more energetically favorable for the enzyme with the loop region, and without the YF motif, to be able to form a disulfide bond than an enzyme with the YF. This motif has thus been linked to sensitive enzymes as it is less likely to go around the cycle and to shoot off into the inactivation shunt where no disulfide bridge is formed. 7PUT IN LOCALLY UNFOLDED STUFFAhpC2 C50SYL vs. YFSeen in significant human pathogens that cause African sleeping sickness, ulcers, malaria, and Beaver Fever

Our protein, AhpC2,however, has similar, yet different motifs that have not been well studied. AhpC2, if you are curious, stands for the second alkyl hydroperoxide reductase that uses subunit C in salmonella typhimurium. The first of which, AhpC has the two motifs that have been well studied. AhpC2 has a YL motif instead of the YF motif. This is a substitution of a benzene ring for the rest of the side chain. The reason this unique motif is important is that it is seen in various pathogens such as those that cause diseases like African sleeping sickness, ulcers, malaria, and giardia, otherwise known as BEAVER FEVER! 8HypothesisStAhpC2 is a representative of the proteins in deadly human pathogensStructure gives insights into pathogen survival in humans and forms basis for drug designYF motif significance can also be studied

AhpC2 is a representative of the protein in these various pathogens, and that by studying the structure, I can gain insights into how these and Salmonella typhimurium survive in the body. I will also be able to examine the effects of the YL motif. 9Crystals!Pure protein from Dr. Leslie Pooles lab, WFUOptimizing conditions(NH4)2HPO4pH

The first step in determining the structure is to grow a crystal. I got pure protein from our collaborators in Dr. Leslie Pooles laboratory at Wake Forest University. Based off of previous work done in the lab, combining a precipitant of ammonium phosphate and a neutral to basic pH with the protein gives reproducible crystals. So I worked with this information to try to make a single crystal form.10Crystals!

These are three representatives of crystals that have grown for AhpC2. Some (picture 1 and 2) are visibly not single. Others seem to be a single diamond in shape, but have etchings or what look like multiple thin layers stacked upon each other. I exposed these to x-ray beams while being rotated, giving a diffraction image of spots. 11Add in cp-bi diffraction imageCrystals!Predictions do not line up

Stalled progress

Here is a piece of one diffraction image. I was able to determine that they are in fact protein crystals, but although they are macroscopically singular, they in fact have multiple lattices and the data is not able to be properly processed. These spots, known as reflections, and their relative intensities, or brightness, give information of the arrangement of the atoms. Therefore, it is important that the computer program can recognize the individual reflections. Shown here is the predictions of the program I used. The blue and yellow boxes indicate where it thinks there is one reflection and ideally they would rest with the spot in the middle of the box, as you will see later on. In some places, the computer is offset. In others, it predicts one spot where there is visibly two. Sometimes, it even misses entire sections. All of our crystal samples produce similar diffraction patterns that is complex enough to mess up the computers predictions. Until I can grow single crystals or get the computer to identify the reflections better, I am stalled in solving the structure of this protein. While working with this protein, I had another drug target project that was giving much more success, one involving the mycoplasma organisms.12The Mycoplasma DiseasesMycoplasma pneumoniaeWalking pneumoniaMycoplasma mycoidesContagious bovine pleuropnemonia

Mycoplasma is the genus of a variety of species. The Mycoplasma pneumoniae bacterium is the cause of walking pneumonia. Although a less severe form of pneumonia, its symptoms reflect those of the flu and many spread the disease unknowingly, causing community outbreaks. Members of the genus are also associated with contagious bovine pleuropnemonia, a disease infecting as much as 80% of Africas cattle.13Fill Slide

GlpO (Glycerolphosphate oxidase)

Our protein, called GlpO (or glycerolphosphate oxidase) from the Mycoplasma pneumoniae organisms, the one that causes walking pneumonia, follows this catalytic cycle. First glycerol is extracted from the host (in this case human) blood stream where it is converted to 3-phosphoglycerol from ATP. GlpO works to oxidize 3-phosphoglycerol to DHAP, or dihydroxyacetone phosphate. This molecule participates in glycerol metabolism and provides energy to the bacterium. In the process, GlpO generates peroxide, which comes back and kills the host cells, causing pneumonia. If we understand the structure of this protein, the idea is that it will aid in the development of a drug against walking pneumonia.14where to put the fact that it has FADWhat is knownStructure of GlpONo substrates able to be bound DHAP,tartaric acid, 2-phosphoglycerate, menadione, G3P, phosphoenolpyruvate

So a separate structure of GlpO has already been solved. A cartoon version can be seen here. This molecule bound in yellow is called FAD. It is a cofactor that binds into the protein, but is not the substrate. The FAD molecule works to transfer the electrons in the redox reaction carried out by GlpO. This molecule, acquired in the body from the vitamin riboflavin, actually produces a yellow light visible in the protein. Although there is already a structure of GlpO available, this protein was unable to bind substrates and inhibitors. If I can get a picture of how these work and bind in GlpO, I can understand the way in which a drug might bind and how the structure is changing as the protein works, getting small screenshots of a protein in action. A list of the substrates we will be working with is shown here. 15HypothesisDrug targetSubstrates, products and inhibitors in GlpOInsight into Mycoplasma and other organisms cause diseaseBasis for structure-based drug design

I am using MpGlpO because it serves as a drug target. By getting substrates to bind in this protein and the solving the structure, we can better understand how inhibitors bind. I can also gain insight into Mycoplasma pneumonia and other organism that use the glycerol metabolism mechanism.16CrystallizationPure protein from Dr. Al Claibornes lab, WFUNaCl and neutral-basic pH buffer

I acquired pure protein from Al Claibornes lab also at Wake Forest University and was able to successfully grow crystals. Out of more than 500 conditions I tried and about 30 leads, two crystal forms emerged. Beautiful and sharp, yellow GlpO crystals grew with NaCl and a neutral to basic pH buffer. I saw both trigonal and tetrahedral pyramidal crystals. I took these crystals exposed them to x-rays.17Diffraction and Data ProcessingResolution: enough to define atomic structure (3.0 )

I took complete data sets using both home and synchrotron sources. I could identity that these were single crystals and that at 3A, they were good enough quality to define the atomic structure. This diffraction image shows that the computer was better able to recognize and predict the spots for the boxes line up with the reflections. After successfully processing this data, I could proceed to dealing with the phase problem.18Molecular ReplacementThe Phase ProblemStructure of related protein (3DME)

HeightThe phase problem occurs specifically with trying to detect light. During x-ray diffraction, the protein absorbs the x-ray light, and then diffracts rays of light. Light acts as a wave and we can detect the heights of the wave. We cannot, however, detect the phase, in this case the shift of the x-axis. This information is lost. So to piece the puzzle together I used molecular replacement. Molecular replacement involves using another protein of similar structure as a crude model for the unknown protein. To do this, I searched the Protein Data Bank for a protein with a known structure that had a similar amino acid sequence to GlpO, for a similar amino acid sequence relates to a similar structure with which I can base the phases off of. The related protein I used, Protein Data Bank entry 3DME, was 27% identical in its sequence. 19A Good SolutionDifferences between model and map

Once I performed molecular replacement, I had to be sure my result gave real information rather than just what is called model bias. This is the bias that comes from the related protein model influencing too much how the electron density map looks. To evaluate this, a useful tool is to have part of the structure I know is present in our protein, but was not included in the search model, acting as an internal control. Both proteins have an FAD molecule in their structure. To create the internal control, this molecule was left out to begin with. Then, any density that we see for the FAD must be real information and not model bias. As you can see here was a big blob of density, saying something should be there that isnt. Ta dah, we had found our FAD with the 3 rings fitting into this big blob and another electron-rich region of FAD, the pyrophosphate group corresponded with this density blob. I could now conclude this was a good molecular replacement solution and could start to refine our structure.

20Structure RefinementProcess of manually changing sequenceAvoiding model bias

During refinement to make a better structure, I began the process of manually mutating residues to match the amino acid sequence of GlpO. Because of the dangers of model bias, my strategy was to only model GlpO sequence in where there was evidence for the residue. For instance, here there was an alanine residue in the 3DME model, quite short. The alignment of the 3DME sequence to GlpO showed that there should be a glutamic acid residue in its place, a much longer amino acid. There is more density above this alanine indicating that there should be something longer there, and thus I had excellent evidence to model a glutamic acid. One thing I learned was that the alignments cannot always be trusted, therefore, it was important to walk through the structure, all 370 residues, and manually build the sequence only when there was good evidence. The refinement process is broken up into rounds, where I model as much as I can see and then let the map improve so that I can see more in the next round. The measure of the quality of the model is based on what is called the R-free factor. The best value is the lowest value and as you can see, the quality of the model has improved over the many rounds of refinement. So where do we go to next?21Future WorkContinue refinement of structureSoaking of substratesOptimization of cryo-protectant15% glycerol

0% glycerol15% glycerol10% glycerolI am continuing the refinement process, making the model better and hoping to see our R-free decrease until the structure is optimum. This is for the native GlpO structure, without any substrates. The next goal is to soak substrates into the protein to get a view of how the inhibitors bind, giving insight into this mighty drug target. Yesterday, we got diffraction patterns of many crystals soaked in substrates from a synchotron x-ray source at UC Berkeley to see if the substrates had bound and if we could get a clear picture of its binding site. Also, in my original diffraction patterns, ice rings were present, which are these sharp concentric circles. These can skew the data and can indicate that the crystal may have been harmed during freezing, limiting the detail I can see for the structure. During freezing, I originally used an oil as a protectant in order to avoid degradation and ice formation, but as you can see here, ice rings formed. So now I have optimized the conditions. At ten percent glycerol, the ice rings are diminished and at 15%, they are essentially gone. I am now using 15% glycerol as a cryo-protectant, hoping to get better resolution and quality data for solving the native and substrate bound GlpO structures.

It is these structures that I hope to solve in order to provide a basis for designing a drug against walking pneumonia and other related diseases. Although AhpC2 is presently stuck, perhaps it too can be solved. Even though learning x-ray crystallography and going from a small and beautiful crystal to a complicated structure seemed like a complex task, I have learned that with hard work and a little luck, I will be contributing to the vast amount of information on the incredibly important proteins of this world.

22AcknowledgementsDr. P. Andrew KarplusThe rest of the Karplus labDr. Andrea Hall, Camden Driggers, Ian WinterHHMI, CrippsURISCOur collaborators at Wake Forest UniversityDr. Leslie PooleDr. Al ClaiborneDr. Kevin Ahern

23Citations of research + articles, images?