Part II: An In-depth Description of The Study For Readers With a Biomedical Background
A guest post by Jingcheng Wu
More on c-src: what it is and its relations to cancer
C-src is short for c-src tyrosine kinase. Kinase is a type of enzyme that removes a part (phosphate group) of the molecule (ATP) that is required for every energy-expending process in the body, and attaches it to a specific amino acid (tyrosine, threonine or serine) of a protein (substrate). C-src belongs to a family of kinases called the Src tyrosine kinase.
C-src stimulates the pathways that induce cell growth, generate new blood vessels, prevent cell suicide, and give cells ability to migrate1,2 – all necessary to give rise to proliferation of invasive cancer cells. When there is a mutation to the gene that encodes c-src, mutant c-srcs produced could mimic the functions of the normal signal transduction c-srcs.3 When there is over or mis-expression of the said gene, too many normal c-srcs would be produced. In both cases, it is like stepping on a gas pedal of a car. Once the aforementioned abnormality is coupled with the loss of tumor suppressor gene functions,4 it is like additional loss of the brake of a car, and the car takes off and wreaks havoc.
What is the mechanism of action of existing drugs on the market?
C-src holds the ATP and substrate to close proximity at a small, pocket-like region in itself called the active site. To do so, the c-src active site needs certain shape and atomic arrangement to attract or repulse certain atoms of the ATP and substrate, so that the ATP and substrate fit into the active site of c-src.
Existing small molecule inhibitor drugs like Gleevec (for treating a chronic blood cancer, CML) contain substances structurally similar to ATP.5 Such substances fit into the c-src active site, and disrupt its interaction with actual ATPs.5
Why are existing drugs on the market toxic and ineffective?
The strategy mentioned above is quite effective but has the following flaws:
1: C-src is just one member of one family of kinases, and the active sites of kinases have similar structures because they perform similar functions. Moreover, many kinases do not involve in cancer development and just carry out their benign and necessary tasks in the cells. As a result, such drugs not only inhibit c-src in cancer cells, but also the activities of other friendly kinases in normal cells.
2: Frequent mutations take place in c-src, and some of the mutations can vary the structure of the active site.6 When that happens, ATP might still fit, but the structurally resembling drug substance might not fit.
To tackle this low selectivity and drug resistance problem, we looked at finding a novel drug-binding site on c-src.
What novel drug-binding site on c-src did we find and how does it work?
Proteins including c-src are chains of amino acids. The chains fold, twist, vibrate and change shapes (conformations) all the time due to their internal interactions among the amino acids7 and interactions with surrounding environment.
Inactive c-src undergoes a series of conformational changes to open up its active site allowing substrate binding, and subsequently becomes active. This dynamic process pauses at two metastable conformations before reaching the end active state.16 These two metastable conformations are the intermediates I1 and I2.16 We were able to find a drug-binding site to trap c-src in the I2 conformation so it cannot move on to adopt the fully active conformation. Such task can be achieved by using any drug that binds to a particular region of c-src that is not the active site (hence allosteric site).16 This approach has lower toxicity, fewer side effects and longer lasting action compare to existing treatments on the market.
For the drug-bond region to inhibit c-src conformational change from I2 to active state, it has to communicate with the active site. It doe so via interactions within and between those clustered components of c-src (domain) that behave relatively independently of the rest of the protein. Some of the domains can be many amino acids away from another (long range) yet they communicate with each other and act cooperatively.16 The close-neighboring amino acids also communicate with each other extensively, forming a local network to act in concert to unfold during activation.16
What have we done that had not been done before with c-src?
Before, researchers could tell the static structural differences between active and inactive c-srcs (the two end states). They also identified the metastable intermediates I1 and I2. Yet they did not know how long it took and through what sequence of events and conformational changes to get from one end state to the other. On the other hand, we captured the entire dynamic process. We found out for the first time that the time (106 s) it takes to tansit from inactive state to active state (activation) is about five times longer than the time (21 s) it takes for the reverse process (deactivation). We also discovered the thermodynamic and kinetic features of c-src activation for the first time. Moreover, we found out more details about the two intermediates. In addition, this is the first study that used Markov State Model (MSM) on large-scale complex conformational changes of enzymes, whereas previously MSM had been only used on simulating protein-folding re-arrangements.16 Last but not least, we found all the conformations of the “A loop” (a loop structure critical for c-src activation) of the protein chain, whereas other research groups found only one (which is the open conformation).8
What approaches did we use to conduct our research?
We combined MSM based, massively distributed computational method, statistical method and other algorithms and techniques to simulate c-src dynamic conformational changes at the atomic level.16 There are tens of thousands of atoms in the protein itself and in the surrounding water molecules to simulate. Atoms change their energetic, vibrational and kinetic properties within less than a trillionth of a second. The time it takes for c-src to transit from inactive state to active state is around one tenth of a thousandth of a second,16 which is quite a long time on the atomic scale. The conformational transition also could follow numerous different paths to reach the end states.16 To simulate the entire c-src conformational landscape with surrounding water molecules at the atomic scale, while considering all the possible paths it could take over such a large timescale, it requires enormous computing power and vast amount of resources to carry out.
Luckily, we can break down the entirety of the computation into millions of small parts, and have donors from all over the world to each take one part and complete the computation on their personal electronic devices such as laptops, computers and Playstation3s. The previous studies had to employ simplification strategies that omitted key fine details on the kinetics of conformational transitions due to lack of computing power.
How would the methods outlined in this study potentially increase drug selectivity?
Some kinases, especially the ones in the same family with c-src, have amino acid sequences and structures highly resemble c-src. Such similarities pose difficulties of increasing drug selectivity. Fortunately, our detailed conformational landscapes help to distinguish subtle structural differences among proteins. For example, the Hck kinase (another member of the Src tyrosine kinase family), like c-src, also has two metastable intermediates I1 and I2.9 I1s of Hck and c-src are similar, but the “A loop” of Hck I2 is partially unfolded whereas the “A loop” of c-src I2 is fully unfolded.16 For drugs or fluorescent probes to bind to the novel allosteric binding site, the “A loop” has to be fully unfolded.16 Thus, the drugs or probes would only bind to c-src I2, not Hck I2. In other words, the subtle structural difference detected by our method between c-src I2 and Hck I2 allows drugs to selectively only inhibit c-src.
Other findings from the study
Diversity within intermediate states of c-src: When c-src goes through two intermediates (I1&I2) to switch between active and inactive states, the I1 does not always maintain exactly the same structure, and the same applies to I2. C-src I1 has a partially unfolded “A loop,”16 which is like a loose ribbon flapping around in the wind. In addition, there are a few amino acids of the “A loop” that fluctuate and adopt not two, but multiple intermediate conformations along the activation pathway.16 When taking snapshots of the intermediate states at different times, the “A loop” would look different.
Slow rate of autophosphorylation: As mentioned earlier, c-src is a kinase that transfers a phosphate group (phosphorylation) from ATP to its substrate. It turns out that the substrate of c-src can be another c-src.16 The c-src can phosphorylate a member of its own kind, so this is called trans-autophosphorylation. Activating c-src is like turning on an old-fashioned light bulb. When you first turn on the switch, the light bulb flashes quickly alternating between on (active) and off (inactive) states. The way to lock the light bulb at the “on” state so that you get stable light source is for the c-src to either bind to its substrate (another c-src) or get trans-autophosphorylated by another c-src.10,16 It is important to note that for the substrate to be able to receive a phosphate group from ATP, it has to expose the site that the phosphate group attaches to.9 Initially when there are scarce active c-srcs floating around, the chance of encountering one and subsequently getting trans-autophosphorylated is small.16 As time goes on, more and more active c-srcs are available and the chance of encountering one is much greater, which speeds up the process exponentially.16 Our model predicts that the time evolution of active c-src population is sensitive to changes of concentrations of inactive c-src and briefly active c-src (not phosphorylated so not locked in the active state).
“A loop” has to unfold before C helix can change conformation: The different parts of c-src undergo conformational changes in specific orders. Also, certain parts of the protein like the C helix remains folded during activation,16 although this structure as a whole can rotate inward or outward.
Myristate-binding pocket in c-src could serve as another allosteric drug-binding site: A tyrosine kinase called c-ABL, which is a key component of a mutant fusion protein that causes a chronic blood cancer (CML), has similar fold as c-src.11 C-ABL and c-src both have an ATP binding pocket in the active site, a myristate-binding pocket not in the active site (thus allosteric) and an “A loop.” For c-ABL, binding of myristate to the myristate-binding pocket can be “felt” by the ATP binding pocket and the “A loop.” The ATP binding pocket and the “A loop” respond by changing their conformations, which lead to c-ABL activity supression.12 Due to the structural similarity of c-src with c-ABL, binding of a drug to the c-src myristate-binding site could produce a similar effect as observed in c-ABL.16
Future outlooks based on this study
Future studies on c-src can include its two regulatory domains (SH2, SH3) and its locked active state (trans-autophosphorylated state) in the simulations.16
Furthermore, the discovery of the new allosteric inhibitor drug-binding site can be potentially used simultaneously with the ATP binding pocket of the active site13 as binding sites for a group of drugs called “fragment-based inhibitors.”14 Such a drug has two tightly (covalently) linked fragments that bind to two different sites of the same target kinase.14 It is equivalent to a “super drug” that combines the effect of an existing small molecule inhibitor drug with the effect of a drug that traps c-src in I2.
In addition, a recent crystal structure of CDK2 (a serine/threonine kinase critical for G1 to S phase transition in the cell cycle.17 Inhibitors of it arrest cell cycle and prevent cancer development) is found to be similar to c-src I2.15 A fragment-based inhibitor (AT7519) has been designed to inhibit CDK2.14 Perhaps AT7519 can be slightly modified to inhibit c-src as well, since CDK2 and c-src I2 are structural analogues.
The same methodology and techniques used in this study can be applied to the other members of the Src tyrosine kinase family besides Hck and c-src to find out their subtle structural differences.16 Then these differences can be harnessed for future design of selective drugs that target each individual member like what we did for Hck and c-src.
For more technical details, please refer to the original paper.
- Blume-Jensen, P. et al. Oncogenic kinase signaling. Nature. 411, 355-365 (2001).
- Wheeler, D., Lida, M. & Dunn, E. The role of Src in solid tumors. Oncologist. 14 (7): 667-678 (2009).
- Chial, H. Proto-oncogenes to oncogenes to cancer. Nature Education 1(1):33 (2008).
- Knudson, A. Mutations and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci USA. 68(4): 821-823 (1971).
- Fausel, C. Targeted chronic myeloid leukemia therapy: Seeking a cure. Am J Health Syst Pharm 64, S9-15 (2007).
- Daub, H., Specht, K. & Ullrich, A. Strategies to overcome resistance to targeted protein kinase inhibitors. Nat. Rev. Drug Discov. 3, 1001-1010 (2004).
- Bruce, A., Johnson, A., Lewis, J., Raff, M., Roberts, K., & Walters, P. The shape and structures of proteins. Molecular biology of the cell 4th edition (2002).
- Meng, Y. & Roux, B. Locking the active conformation of c-Src kinase through the phosphorylation of the activation loop. J. Mol. Biol. 426, 423-435 (2014).
- Yang, S., Banavali, N. & Roux, B. Mapping the conformational transition in src activation by cumulating the information from multiple molecular dynamics trajectories. Proc. Natl Acad. Sci. USA. 106, 3776-3781 (2009).
- Roskoski, R. Src protein – tyrosine kinase structure and regulation. Biochem. Biophysics. Res. Commun. 324, 1155-1164 (2004).
- Cowan-Jacob, S. et al. The crystal structure of a c-src complex in an active conformation suggests possible steps in c-src activation. Structure. 13, 861-871 (2005).
- Zhang, J., Yang, P. & Gray, N. Targeting cancer with small molecule kinase inhibitors. Nat. Rev. Cancer. 9, 28-39 (2009).
- Martin, M. P. et al. A novel approach to the discovery of small-molecule of cdk2. Chem biochem 13, 2128-2136 (2012).
- Gill, A. New lead generation strategies for protein kinase inhibitors-fragment based screening approaches. Mini Rev. Med. Chem. 4, 301-311 (2004).
- Betzi, S. et al. Discovery of a potential allosteric ligand binding site in cdk2. ACS Chem. Biol. 6, 492 (2011).
- Shukla, D., Meng, Y., Roux, B. & Pande, V. Activation pathway of Src kinase reveals intermediate states as targets for drug design. Nat. Commun. 5:3397 doi: 10.1038/ncomms4397 (2014).
- Tsai, L., Lees, E., Faha, B., Harlow, E. & Riabowol, K. The cdk2 kinase is required for the G1-to-S transition in mammalian cells. Oncogene. 8(6): 1593-