The Story Behind the Paper: How calmodulin became efficient

by John Lisman

Story behind: Nat Neurosci. 2011 Mar;14(3):301-4. Calmodulin as a direct detector of Ca²⁺ signals. Faas GC, Raghavachari S, Lisman JE, Mody I.

Long-term potentiation, a model for memory, is triggered by the activation of the calmodulin-dependent protein, CaMKII in dendritic spines. Sri Raghavachari, my former postdoc, and I were interested in how exactly CaMKII gets activated during LTP. It seemed that it should be straightforward to account for this in a computational model. Our confidence was based on the fact that a lot of groundwork had been done—the elevation of Ca²⁺ that triggered this process had been measured by Ryohei Yasuda and the interactions of Ca²⁺, calmodulin and CaMKII had all been determined in test tube experiments. But when Sri put this all together in a standard biochemical model, the simulations indicated that there would be virtually no CaMKII activation. Clearly something was wrong because Ryohei Yasuda had shown that under the same conditions in which he measured Ca²⁺ elevation in spines, he could also measure strong CaMKII activation.

During the summer I work at the Marine Biological Laboratory (MBL) in Woods Hole, Massachusetts and Sri came down to visit. We spent hours trying to figure out what was wrong with the simulations. Sri had carefully checked the simulations and determined that the program could accurately account for other data. Thus, the problem was not a bug in the program, but rather in an assumption we had put into the program. Every day, we awoke, convinced we had found the erroneous assumption; by nightfall we had rejected that idea.

One of the great things about working at MBL is the number of other neuroscientists there and the collegial atmosphere. We went to talk to Bill Ross, an expert in the measurement of Ca²⁺ in neurons. It had long been known that when Ca²⁺ enters neurons, very little of it stays free because most gets bound to “buffer” molecules. These are like the pH buffers all biochemists use; it’s just that Ca²⁺ buffers bind Ca²⁺ instead of protons. Bill asked us a lot about the particular Ca²⁺ buffers that were in spines—what was known about their molecular identity and their Ca²⁺ binding properties. Moreover, he wanted to know what assumptions about the buffers we had built into our simulations.

The answer to this was simple: we had followed the “standard” dogma based on the work of the Nobel Prize winner, Ernst Neher. He had determined that neurons contained a “fast buffer” that very rapidly binds the entering Ca²⁺. He had not been able to determine what type of molecule this was. Every model of Ca²⁺ dynamics that had since been developed had incorporated this fast buffer into the scheme and we had followed this convention. Thus, when Ca²⁺ entered the cytoplasm, 95% got bound to the fast buffer; only the remaining 5% was free and could activate calmodulin.

Perhaps there was something wrong with this assumption, but to evaluate this issue we had to learn about how Ca²⁺ buffers work, something we knew little about. Fortunately, Isabelle Llano was working as instructor in the MBL Neurobiology course. Because of her expertise in the small proteins that buffer Ca²⁺ in neurons, we went to chat with her. We learned a lot from her, but she also pointed us to Guido Faas, who was working with Istvan Mody at UCLA and measuring the kinetics of how Ca²⁺ binds to protein buffers. Previous work had measured the equilibrium properties of Ca²⁺ binding to proteins—the binding rate was then inferred by calculation. In contrast, Guido was using advanced methods to rapidly jump the Ca²⁺ concentration and then actually measure how fast Ca²⁺ would bind.

As we learned more about Ca²⁺ buffers from Guido and read the literature more carefully about calmodulin, we finally came up with a radical but intellectually satisfying new model—-perhaps the fast buffer that Neher had measured was none other than calmodulin itself. This would certainly radically change our computer simulations—-instead of calmodulin responding to only the 5% of Ca²⁺ that remained free after the bulk of Ca²⁺ was soaked up the unknown “fast buffer”, calmodulin would be activated by all the Ca²⁺ ions that entered, making the process of calmodulin activation and CaMKII activation much more efficient.

But for this to be true, calmodulin would have to bind to Ca²⁺ fast, faster than to the other major Ca²⁺ binding proteins in neurons (e.g. calbindin). When we talked to Guido about this possibility, he was excited to test it. His previous work had dealt only with calbindin, but he could now extend the work to calmodulin. Indeed, he could reconstitute the buffering in spines, putting both calmodulin and calbindin into his cuvettes. When the results came in, they were stunning; calmodulin has extraordinarily fast Ca²⁺ binding kinetics, much faster than that of calbindin.

With the new binding parameters provided by Guido, Sri reformulated his computer model. He took out the unknown fast buffer and replaced it with only calmodulin (which is at surprisingly high concentration) and calbindin. The simulations now showed that enough calmodulin was activated to account for the measured activation of CaMKII, our holy grail.

The new view we propose makes sense: Calmodulin is the transducer that couples Ca²⁺ entry to enzyme activation. It would make sense for calmodulin to be as efficient a detector of Ca²⁺ as possible and thus to directly intercept the entered Ca²⁺. Our results indicate that this is the case. Ca²⁺ triggered reactions are implicated in hundreds of forms of biological signaling. We therefore believe that this new view of Ca²⁺ signaling will have broad applicability.