Research history of Eric L. Hargreaves
My graduate work in Peter Cain's lab examined synaptic plasticity, scrutinizing all the evidence that then linked learning to long-term potentiation (LTP). LTP is an enduring increase in synaptic efficacy commonly induced by brief electrical stimuli and remains the reigning neural phenomenon used to model learning and memory.
A typical "chronic" LTP experiment would consist of implanting stimulating and recording electrodes in one of the several monosynaptic hippocampal pathways. Brief test pulses are then applied at the stimulating electrode and an evoked potential (EP) is registered at the recording electrode, reflecting the monosynaptic pathway's degree of activation, or the pathway's "efficacy".
Subsequently, if a series of brief tetanizing test-pulse trains are delivered through the stimulating electrode and then normal EP recording is once again resumed, the EP will be greatly enhanced or "potentiated". Thus, the EP now reflects a much greater through-put or efficacy of the monosynaptic pathway after tetanization. The potentiation, dependent upon the tetanization parameters can be long-lasting, up to weeks and months. Hence the phenomenon of Long-term Potentiation or LTP.
I was the first to setup and induce LTP in Peter Cain's lab. As it turns out I was also the first to setup and induce LTP at the University of Western Ontario, beating out Stan Leung by only a few months, which Stan was happy to confirm many years later.
LTP was then and still is now the reigning phenomenon with which to model memory processes and its potential underlying mechanisms. However, evidence directly linking LTP to learning was scant.
Although Peter Cain's lab, at that time, largely dealt with Kindling, another plasticity phenomnon used to model focal epilepsy, I wanted to change that, and arrived with a specfic "behavioral-LTP" project in mind.
Behavioral-LTP is a paradigm based on the premise that if LTP is like learning and learning is like LTP then Evoked Potentials (EPs) should exhibit an LTP-like increase during learning.
More advanced than earlier work, we contrasted a hippocampal dependent task (David Olton's radial-arm maze) to a non-hippocampal dependent task (oneway active avoidance in a shuttle-box) and, on the insistence of Case Vanderwolf, further employed controls for ongoing behavior, a known modulator of evoked potentials. Mimicking the best of previous work we further employed a counter-balanced within-animal design, such that different groups of animals ran through the tasks in a different order, and then finally, all were "LTP'd" for real to see if our system and implants could induce, sustain and detect changes in neuroplasticity.
Although the subtle effects of ongoing behavior were observed, once behavior was held constant, no LTP-like effects were expressed. Not surprisingly, we were also able to easily induce LTP in the same animals using a standard 400Hz 20ms burst LTP protocol of electrical stimulation.
Still believing in the viability of the behavioral-LTP phenomenon, I turned to the dominant hippocampal-dependent learning paradigm of the time, the Morris water-maze, hoping that the one day learning protocol would "mass" the amount of change into a detectable difference in the EPs.
One of the other students in the lab had earlier modified a circular open-field into a water-maze, but she manually traced the path of the swimming rats on a clipboard as they searched for the hidden platform. I could not stand the idea of having such a subjective and ethereal measure, so I installed an overhead video camera and thus the water-maze era in Peter Cain's lab began.
Regardless, as before, with the one-way active avoidance and the Radial-arm maze, the results from the Morris water-maze were negative. The animals clearly learned the location of the hidden platform, and retained this information the next day when given a probe trial, in which the hidden platform was removed causing the animals to swim circles around where the platform had been. Yet, as before, no detectable differences in the EPs were discerned, even though we recorded across an abbreviated input/output curve enhancing our chances of finding differences (Hippocampus, Download PDF 889KB).
Not wishing to wrestle with reviewers over a second negative finding, I thought we could contrast the failure of the behavioral-LTP paradigm with the success of the "LTP-saturation" paradigm.
LTP-saturation is based on the idea that LTP and learning will compete with each other for the finite amount of plasticity available within the hippocampus. Given that LTP is artificial and not subtle, it was hypothesized that when pitted against each other LTP would win out over learning. Thus, if all the available plasticity were used by repeatedly LTP'ing the pathway until it would LTP no more, and thus become staurated, then there would be no available plasticity with which to learn, resulting in deficits in water-maze acquition.
We modeled our attempt after the highly toted Castro et al. (1989) Nature paper out of the McNaughton and Barnes' lab, which had found that sturating levels of LTP prevented water maze acquisition, but when the LTP was allowed to decay back down to baseline levels, some 30 days later, the animals easily learned the location of the hidden platform during a second attempt.
Although we were able to demonstrate that the induction of a small bout of epileptiform activity or "after dischrage" (AD) generated learning deficits in the water-maze, bi-lateral saturating levels of LTP did not.
At the time we were in good company, as our findings matched failures to replicate from several other groups including Kate Jeffery in Richard Morris' lab and Donna Korol of the McNaughton and Barne's lab, from where the original finding had come. As the originator of the saturation paradigm, Bruce McNaughton did the smart, and honorable thing, and gathered all the "re-saturation" attempts together, inviting them to publish in a special issue of the journal Hippocampus including us (Download PDF 889KB).
As a side note it is worth mentioning that many of these published results from the LTP-saturation paradigm also lent themselves to supporting the failure of behavioral-LTP paradigm. To explain... ...a standard control for the saturation paradigm was to run a low frequency test pulse group, which received LTP implants and exihibited stable EPs throughout the learning process to show that there was no impairment due to the implant and recording procedures. The control groups doubled as a behavioral-LTP experiment, exhibiting stable EPs throughout the task, indicating that there are no LTP-like effects as a consequence water-maze learning.
As a consequence of these two results, we in Peter Can's lab became intent upon testing all the major lines of evidence that linked LTP to learning, looking for alternate explanations and other underlying causes for the published relationships between learning and LTP.
Brain temperature modulation of EPs, had recently been shown by Edvard Moser under Per Andersen to account for much of the effects of short term environmentally modulated LTP or STEM. STEM was an LTP-like shift in the evoked potential expressed after the exploration of novel environments, but not expressed after repeated home cage recording sessions.
STEM was a variant of behavioral-LTP published by Ed Green, while in the McNaughton and Barnes' lab. Ed had collegially contacted me earlier, and through an e-mail exchange had further directed me to the Mosers' brain temperature work.
For our own purposes, we sought to explain an enduring effect of sleep on the EP, which extended into the awakened state, which may have explained some of the complex-environment enhanced EPs reported by Pat Sharp, while she was in the McNaughton and Barnes' lab. We increased and decreased body and brain temperature through the direct manipulations of radiant heat from a heat lamp, prolonged water-immersion (at room temperature), and urethane anesthetic. These conditions were run in addition to our psuedo-sleep phenomenon of interest. Recording and stimulating electrodes for capturing hippocampal EPs were implanted unilaterally in the hippocampus of one hemisphere, while a "thermistor", for simultaneously monitoring brain temperature, was implanted in a homotopic position of the contralateral hippocampus in the other hemisphere.
Our results indicated that our pseudo-sleep phenomenona was accompanied by a decrease in brain temperature of 2 to 3.7 degrees C. Further, at the higher of two stimulation intensities, the gradual shift in brain temnperature accounted for over 50% of the variance in some of the hippocampal EP measures (Brain Research, Download PDF 822KB).
NMDA antagonism of LTP and water-maze acquisition: Richard Morris, in the early eighties, had made a "big splash" with his water-maze. However, it was not the maze itself, but showing that lesions to the hippocampus prevented rats from learning the location of a submerged platform, even though the rats could still navigate to the same platform in the milky pool of water if the location of the platform was "cued", identified with a small but visible flag.
Around the same time, in the early eighties, Graham Collingridge and his colleagues discovered that the successful induction of LTP in the hippocampus was dependent upon the activation of n-methyl-d-aspartate NMDA, a subreceptor of excitatory amino acid (EAA) neurotransmitter system. Consequently, pharmacologically blocking NMDA subreceptor activation during tetanization prevented the induction of LTP without greatly affecting normal synaptic transmission. Thus, if LTP mechanisms and learning mechanisms were one in the same, then blocking LTP should also block learning, and function like a transient pharmacological lesion to the hippocampus.
Consequently, in a landmark Nature paper, Richard Morris, collabourating with Gary Lynch's lab, implanted cannulae invtraventricularly in rats and infused them with aminophosphonovaleric acid (APV) a potent NMDA antagonist. The NMDA antagonist blocked the normal learning of the hidden platform's location in the water-maze, but did not prevent learning the discrimination between two visible platforms, which were moved around the maze, from trial to trial, only one of which was sufficiently stable enough to climb on and escape from the water. Another set of animals were implanted and infused with the same levels of APV, and after a number of days underwent an acute LTP experiment. Predictably, the APV infused animals did not exhibit LTP after tetanization, while saline infused controls did.
Richard Morris followed this work with a number of refinements and controls, and other research groups began to follow with similar NMDA-antagonist work, as the approach gained momentum and popularity.
However, due to our previous work we found ourselves part of a small, almost "underground movement" of researchers, who were a little more skeptical of LTP's relation to learning then was becoming generally accepted. As a consequence of our status, at the 1990 Neuroscience conference in St. Louis, I had a long discussion with Julian Keith, who along with his supervisor, Jerry Rudy had just published a strong critique of the NMDA/LTP/learning strategy. In their Psychobiology review Keith and Rudy (1990) argued two main points, first that the NMDA-antognist rats did exhibit some degree of learning or savings, and second that the animals exhbited sensori-motor impairments. The review was a target article, in which the authors of a number of critiqued studies were allowed to respond. This resulted in a lively exchange, and one response in particular, we noted, was a response concerning a non-competitive NMDA antagonist, MK801, which we had lying around the lab.
Instead of getting directly into the water-maze/NMDA-antagonist fray, I thought we could implement a behavioral assessment of the effects of MK801 paying keen attention to its sensori-motor effects. Thus, I generated what amounted to a "Neuropsychological Assessment Test Battery" for rats, evaluating a number of spontaneous and simple motor behaviors, independent of any learning.
In the animals' suspended wire-mesh home cages, I assessed their tactile reaction to cotton swabs probing their hind and forepaws unseen from beneath the cages. I further borrowed a task from Ian Whishaw, the "chocolate-chip-cookie-slurry" test, in which cookies were crushed and mixed with milk to poduce a "slurry". The slurry was then smeared onto a ruler and pressed up against the rats' cage. The amount of the chocolate chip cookie slurry that the rats were able to lick off revealed how far they could stick out their tongues, a very sensitive measure of motor dysfunction.
I examined the rats swimming ability in a very simple swim to the end of a clear aquarium and climb out on a vertical wire mesh, looking at latency, and angle of the body in relation to the water surface while swimming (ie closer to horizontal or closer to vertical). Finally, I evaluated the rats spontaneous activity in an automated open-field.
Essentially the results showed deficits on all tests at some doses, and even on the lowest dose of 0.05mg/kg MK801 rats showed deficits on some tests. We did not aim very high with our publishing aspirations, with hopes of flying under the radar of most of those involved the ongoing fray. Unfortunately, we picked up Richard Morris as a reviewer, who was kind enough to identify himself as someone who was not completely unbiased in these matters. Initially he agreed that most individuals would be crazy to use MK801, due to its non-competitive activity-dependent nature, and he himself had published a sole abstract using the drug, thereafter returning to the competitive APV family AP5, AP7 etc... Yet, he thought that most of our findings had already reported and fairly considered and therefore were redundant, and unessary to publish again. Luckily I had crafted our study to use the exceptionally low dose of 0.05mg/kg and we were able to battle back with the editors that deficits at this extremely low dose had not been reported. The result was the fastest turn over of an experimental study I have ever had from conception (November 1990) to execution (December/January/February 1990/91) to analysis (March 1991) to submission (July 1991) to revisions and acceptance (October 1991) to publication (March 1992) (Behavioural Brain Research, Download PDF 931KB). This was followed a couple of years later by the MK801 activity study that I had run first, in order to establish the optimal time window of behavioral assessment, but that had been placed on the back burner due to other projects (Pharmacology Biochemistry and Behavior, Download PDF 500KB).
However, it was these projects that instigated the NMDA-antagonist water-maze program in the "Cain lab". After lobbying for and getting a new water-maze we began at the beginning, running hippocampal lesioned animals, APV infused animals, and any new NMDA competitive antagonists we could get our hands on for which Nova Pharmaceutical donated two, soon to be discontinued, compounds... NPC12626 and NPC17442. Additionally we examined other lesion sites reputed to impair learning, such as the anterior thalamus, and the non-NMDA glutamate subreceptor system with CNQX. The approach was essentially the same as I had initiated, do the water-maze using standard protocols, and then do a number of other sensori-motor assessments. I helped setup the program, the new maze, and ran some of the animals and their analyses, but as I was getting ready to leave the bulk of the running rested with an honours' undergrad Jeff Hall and another graduate student Deborah Saucier. The work was published after I had left, and so I was "grandfathered" in the two back-to-back Behavioral Neuroscience publications (Download PDF 1.63MB, Download PDF 1.31MB).
Impact of Graduate work The total of my graduate work assisted in slowing down the popular research that rushed to embrace LTP as learning. As such my work helped set standards and controls that went into later research more clearly drawing the lines between LTP and learning. I still believe that there are valid connections between the mechanisms underlying learning and those underlying LTP, and later co-authoring, with Matthew Shapiro a response to an LTP target article. As such, I guess I remain an optimist, and do not wish to prevent good LTP research until a better and more versatile plasticity phenomenon appears. At fault then, as it is now, it is not so much the LTP phenomenon or its underlying mechanisms, but the way LTP was measured. In particular, the amount of gross activity assessed by hippocampal EPs submerges any subtle distinctions that may occur in a morass of forced simultaneous unit discharge. Although I had set out upon a course to validate the relationship between LTP and learning, the course quite soon became a crusade of how to do certain kinds of research. As such, our lab, and those of us in it, became quickly known as "L'enfants terrible". I also knew that a career correcting other peoples work, was not what I wanted to embark upon. Apart from being a research pariah, one would continually have to wait for others to come up with original work to follow and improve upon. Luckily in my dissatisfaction with the hippocampal EP as an appropriate method for assessing learning, I had already fallen in love with another phenomenen, requiring learning a whole 'nother new recording technique, "place cells", and single unit recording. Thus, for my postdoc I set out to master single unit recording and understand the phenomenon of hippocampal place cells.
My Postdoctoral work
For my Postdoctoral work I sought a position where I could dually pursue single unit recording and NMDA research. My search ended at Matthew Shapiro's lab, when he was at McGIll University. Being in Montreal was also a benefit for apart from being a beautiful and amazing city, it had fantastic historical implications for neuroscience, by housing the Montreal Neurological Institute or MNI and McGill University. Hebb, Kindling, H.M., Peter Milner, Brenda Milner, Ron Melzack... ...the list and opportunitues seemed endless.
Primed-burst Potentiation I thought I had escaped the clutches of that "dirty little" drug called MK801, but Matthew already had a series of papers dealing with the drug and learning impairments on the radial-maze. Thus, to complete the set I initially was to find an NMDA-antagonist disruption of plasticity caused by Mk801 using extremely low doses thought to impair learning without impairing sensori-motor performance.
We suspected that standard LTP protocols would be too powereful, driving past the low levels of antagonism, so I implemented a program of the more subtle primed-burst potentiation (PBP), in which a leading pulse is followed at a "theta" interval by a sole burst of stimulation.
Prior to my work, PBP had been executed only in the ventral hippocampal commissure to CA1 pathway by David Diamond, when working with Greg Rose. Thus, shifting PBP to the perforant-path/dentate-gyrus required working out new parameters. Of course, I insisted on carrying over the critical behavioral controls I had employed during my graduate work. Finally, as a historical quirk/artifact, Matthew had run his radial-maze work using female Wistar rats, which required jiggling my usual coordinates, and learning to take estrous swabs without upsetting the rats. Thus, final work assessed the effects of low doses of MK801 on the estrous cycle, on the LFP pattern of theta, on normal synaptic transmission, and finally on PBP synaptic plasticity itself.
Results, not surprisingly confirmed what Matthew wanted to know, that the same low doses of Mk801 that had disrupted learning in the radial-maze could disrupt plasticity. However, the drug also altered the LFP at theta frequency and attenuated the normal synaptic transmission. Yet, the latter could be compensated for by increasing the stimulation intensity, such that the EPs were recorded at pre-drug levels, and thus, the plasticity, after compensation was appropiately assessed. Another advantage of the PBP plasticity was that it was fairly short acting, even though it was still NMDA-dependent. Therefore, each animal served as its own non-drugged control, and was perfectly matched for EP amplitude and stimulation intensities.
Overall, it was one of the most careful and well thought out pieces of research I have done. Even Greg Rose, who was one of the reviewers, related to me at a later meeting how impressed he had been by our thoroughness.
However, our finding essentially replicated work that had been first published some 10 years earlier, and had been published a number of times therafter in various forms with various drugs, of which ours was now one of many. As such, for all the thought and care that went into the PBP NMDA article, it was redundant. As a result, a little more than a decade later it has garnered a measly 9 citations. An effort that maybe could have been better spent on research that nobody had yet to break into. Hardly worth the effort at all, Current work