Initial Characterization of the Voltage-Dependent Behavior of NMDARs

In considering these early MacDonald publications retrospectively, the reader is reminded that this work originated before the molecular identification of glutamate ionotropic receptors (iGluRs) via cloning between the years 1989 and 1992.Reference Hollmann and Heinemann ¹ In fact, in the late 1970s/early 1980s, some doubt remained concerning the role of glutamate as the main excitatory transmitter in the central nervous system. Though proof of glutamate’s role as an excitatory amino acid (EAA) was lacking, substantive progress was being made in establishing the existence of multiple receptors for EAAs. The identification of EAA receptor (EAAR) subtypes was made possible through the identification of the agonists NMDA,Reference Curtis and Watkins ^{2 ,} Reference Watkins ³ kainate,Reference Shinozaki and Konishi ⁴ and quisqualate.Reference Shinozaki and Shibuya ⁵ Quisqualate was later superseded by α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) as an iGluR subtype defining agonist.Reference Krogsgaard-Larsen, Honoré, Hansen, Curtis and Lodge ⁶ Particularly convincing at this time was the distinction between NMDA versus non-NMDA receptors (quisqualate/AMPA and kainate). This was first based on the observation that responses to NMDA, but not to quisqualate or kainate, could be blocked by elevated extracellular concentrations of Mg²⁺.Reference Davies and Watkins ^{7 ,} Reference Evans, Francis and Watkins ⁸ Although the differential sensitivity of EAARs to Mg²⁺ was recognized, the mechanisms underlying the selective inhibition of NMDA responses by Mg²⁺ was not. Suggested mechanisms included that Mg²⁺ (1) antagonizes NMDA responses at the level of the receptor, ionophore, or other site or (2) exerts a stabilizing effect on the neuronal membrane or (3) associates with selected agonists and reduces their availability for receptor binding.Reference Davies and Watkins ^{7 ,} Reference MacDonald and Wojtowicz ^{9 ,} Reference Nistri and Constanti ¹⁰

Although progress had been made in characterizing the pharmacological properties of EAARs, progress in characterizing the conductance associated with NMDA and non-NMDA receptors lagged behind. One of the contentious issues of the day was to define the ionic species that permeate each channel when opened. More puzzling, was evidence that the application of selected EAAs, including DL-homocysteic acid and NMDA, is associated with a membrane conductance decrease rather than increase,Reference Engberg, Flatman and Lambert ¹¹ anticipated on the basis of agonist-induced opening of a receptor associated ionophore. The majority of the work undertaken at this time, including the work by Engberg et al (1978),Reference Engberg, Flatman and Lambert ¹¹ was performed using microiontophoresis to administer EAAs during in vivo recordings. These approaches made progress in advancing our understanding of conductance mechanisms associated with EAARs difficult for a number of reasons: (1) stable long-term recordings were more difficult to maintain because of cardiac and respiratory movements; (2) use of anesthetics could have a confounding influence on the responses generated by EAAs; (3) uptake mechanisms, which limit the response to applied EAAs, are more prominent in intact preparations maintained at body temperature; (4) the ionic composition of the extracellular milieu could not be precisely determined; and (5) the concentration of EAAs applied through microiontophoresis could not be estimated. These and other limitations were increasingly appreciated and noted in high-profile reviews of the era.Reference Nistri and Constanti ^{10 ,} Reference Mayer and Westbrook ¹²

It is within this state of knowledge of EAARs and the technical approaches used for their study that John launched his own laboratory in 1979 with a focus on identifying and characterizing the pharmacological and biophysical properties of EAARs. He initiated his independent career investigating current and voltage responses generated in neurons following the application of excitatory and/or inhibitory amino acids, their synthetically derived or naturally occurring analogues. John’s prescient early adoption of primary neuronal cultures paired with rapid drug delivery by pressure application, allowed his group to make headway in understanding conductance mechanisms associated with EAARs.

In his first publications as an independent investigator,Reference MacDonald and Wojtowicz ^{9 ,} Reference MacDonald and Porietis ^{13 ,} Reference MacDonald and Wojtowicz ¹⁴ the MacDonald laboratory confirmed the initial report of a paradoxical conductance decrease in response to selected EAAs, now recognized as having in common the ability to strongly activate NMDARs (Figure 2A). Conversely, a strict conductance increase was observed in response to application of the non-NMDAR agonists, kainate and quisqualate. These findings were extended by demonstrating that the conductance decrease observed in response to NMDAR agonists was converted to an increase upon membrane depolarization. The conductance decrease observed, but not the increase, was selectively susceptible to elevated Mg²⁺ concentrations and coapplication of DL-α-aminoadipate, one of the first relatively specific NMDAR antagonists. Importantly, conductance increases observed in responses to kainate were spared from the actions of both Mg²⁺ and DL-α-aminoadipate. Although their significance was not yet fully appreciated at the time, these observations provided the first clue to the unique voltage-dependent behavior of NMDARs. They also strengthened accumulating evidence supporting the existence of multiple receptors for EAAs—in particular, that the receptor for kainate represents a distinct receptor class with distinct conductance properties.

Figure 2

Voltage-dependent behavior of NMDA receptor responses. (A) Current-clamp recording of voltage response of cultured neurons to DL-Kainate (100 µM; top trace) and L-aspartate (100 µM; bottom trace), applied by pressure microperfusion. Application of kainate and aspartate evoked rapid membrane depolarizations (v, voltage traces). Timing of agonist application shown on pressure trace (p). Vertical deflections represent membrane responses to superimposed negative current injections (each of 50 ms duration). Reduced membrane response to negative current injections during the kainate-evoked response is consistent with an increase in membrane conductance. Increased membrane response to negative current injections during the aspartate evoked was anomalous in suggesting a decrease in membrane conductance. From MacDonald and Wojtowicz (1980) with permission. (B) Current-voltage (I-V) relations in the presence (open circle) or absence (closed circle) of L-aspartate (500 µM), applied by pressure microperfusion. I-V relations were generated from voltage-clamped cultured neurons. L-aspartic acid induces a region of negative slope conductance from −50 to −26 mV. We now know that the region of negative slope corresponds to increased block of NMDARs by magnesium as the membrane potential is made more negative. From MacDonald et al (1982) with permission.

All of the findings described so far were derived from experiments using current-clamp recordings whereby membrane depolarizations were recorded in response to the application of selected EAAs. By applying Ohm’s law, changes in membrane conductance could be estimated by monitoring the change in voltage deflections evoked by repeated current injections superimposed upon the depolarizing response generated by applied EAAs. Several theories were advanced to explain the decrease in membrane conductance observed with applied NMDAR agonists, perhaps chief among them being the suggestion that this was due to the closing of voltage-dependent K⁺ channels in response to membrane depolarization.Reference Engberg, Flatman and Lambert ¹⁵ Implicit with this stated mechanism is an acknowledgment of the inherent limitations of current-clamp recordings; that neuronal membrane conductance is highly nonlinear and varies with membrane potential resulting from changes in underlying voltage-dependent conductances. Recognizing this, MacDonald and Wojtowicz (1982) acknowledged that “A misinterpretation of the sign of ΔG (i.e. whether conductance increased or decreased) arises when voltage deflections pass through the range of membrane potential where a region of negative slope conductance is present, or in other words, that range where a voltage-dependent current is activated”; and moreover, that “a highly voltage-dependent increase of GNa or Gca (i.e. Na⁺ or Ca²⁺ conductance, respectively) could provide an explanation for an apparent decrease in Gm (i.e. membrane conductance) equally as well as a decrease in GK (i.e. K⁺ conductance).”Reference MacDonald and Wojtowicz ⁹ In keeping with this, they acknowledged that more definitive evidence concerning the nature of the conductance changes elicited by NMDAR agonists would require conductance measurements using voltage-clamp techniques and then proceeded to undertake such measurements.Reference MacDonald, Porietis and Wojtowicz ¹⁶ Taking advantage of the ease of access and stability afforded by the use of primary cultured neurons, they performed voltage-clamp recordings and evaluated the current-voltage (I-V) characteristics of the conductance evoked by L-aspartate. They observed that L-aspartate, now recognized as a specific NMDAR agonist, induces a region of negative slope conductance over a range of potentials from −50 to −26 mV and thus demonstrated conclusively the voltage-dependent behavior of NMDARs (Figure 2B). Influenced by these reports of MacDonald and colleagues, the mechanistic basis of voltage-dependent NMDAR channel behavior was established shortly thereafter when it was shown that NMDARs are subject to voltage-dependent block by Mg²⁺.Reference Mayer, Westbrook and Guthrie ^{17 ,} Reference Nowak, Bregestovski, Ascher, Herbet and Prochiantz ¹⁸

These advances in understanding the mechanisms underlying the voltage-dependent properties of NMDARs coincided with parallel progress in understanding their key contribution to the induction of synaptic plasticity (see CollingridgeReference Collingridge ¹⁹ for review). Making use of the recently introduced specific NMDAR antagonist, (D,L)-2-amino-5-phosphonopentanoate (AP5), Collingridge et al (1983)Reference Collingridge ¹⁹ demonstrated that NMDARs are responsible for the induction of long-term potentiation (LTP) at the Schaffer collateral-commissural pathway.Reference Collingridge, Kehl and McLennan ²⁰ Intriguingly, they also noted that AP5 had no effect on baseline excitatory synaptic transmission or pre-established LTP. In contrast, γ-D-glutamylglycine, a weak antagonist of AMPA and kainate receptors, depressed synaptic transmission. This suggested that AMPA/kainate receptor subtypes mediate fast excitatory synaptic transmission, whereas NMDARs serve as a trigger for the induction of synaptic plasticity. But the mechanisms permitting the differential contribution of each receptor subtype to baseline transmission versus induction of plasticity were not recognized. The discovery of the voltage-dependent block of NMDARs by Mg²⁺ allowed the now-established role of the NMDAR as a coincident detector and gate for the induction of synaptic plasticity to be proposed on theoretical grounds,Reference Collingridge ²¹ supporting Hebb’s postulate for how memories are formed.

Additional noteworthy discoveries from the MacDonald laboratory during this era include results from a series of detailed and elegant studies characterizing the mechanisms of block of NMDARs by the dissociative anesthetics ketamine and phencyclidine,Reference Honey, Miljkovic and MacDonald ^{22 -} Reference Orser, Pennefather and MacDonald ²⁵ in which the voltage-dependent block of NMDARs by ketamine and phencyclidine was first described. This was the first example of John’s interest in addressing a clinical problem and of forging strong ties with a physician trainee, namely Beverley Orser, a well-recognized clinician-scientist (departments of Physiology and Anesthesia, University of Toronto).

Posttranslational Regulation of iGluRs via Serine/Threonine and Tyrosine Phosphorylation

The characterization of the voltage-dependent properties of NMDARs and identification of its dependence on Mg²⁺, coupled with knowledge regarding the contribution of NMDARs to the induction of synaptic plasticity, led to an explosion of interest in understanding the regulation and function of NMDARs. The elucidation of the important role of protein phosphorylation in regulating iGluRs, and in particular NMDARs, and the functional consequence of changes in NMDAR phosphorylation on the induction of synaptic plasticity would come to represent an area to which John made important contributions over the ensuing years. Highlighting the flurry of interest and intensive research activity in this area, many of the MacDonald laboratory’s groundbreaking findings were reported alongside analogous findings by competing laboratories, often in the very same journal issue.

Initial Characterization of the Requirement for Protein Phosphorylation in Maintaining iGluR Function

Among the earliest evidence that protein kinase activity can regulate ion channel function stemmed from studies in which purified catalytic fragments of protein kinase A (PKA), or its specific inhibitor protein kinase inhibitor, were injected into neurons from the mollusc Aplysia.Reference Castellucci, Kandel, Schwartz, Wilson, Nairn and Greengard ^{26 ,} Reference Kaczmarek, Jennings and Strumwasser ²⁷ The advent and subsequent rapid adoption of patch-clamp techniques introduced by Neher and SakmannReference Hamill, Marty, Neher, Sakmann and Sigworth ²⁸ would greatly facilitate the elucidation of the role of protein phosphorylation in regulating ion channel function in mammalian neurons. The whole-cell patch-clamp configuration, which allows for the exchange of soluble factors between the intracellular milieu and pipette solution, would prove especially beneficial in this respect. Using large-tipped electrodes (tip diameter >1 µm, 3-4 MΩ resistance) routinely used for whole-cell recordings, the MacDonald laboratory would provide the first evidence that NMDARs can be regulated by phosphorylation. In their studies, they noted that NMDARs’ currents would progressively rundown over time unless a means of preserving intracellular adenosine triphosphate (ATP) levels was included in their patch pipette solutions.Reference MacDonald, Mody and Salter ^{29 ,} Reference Mody, Salter and MacDonald ³⁰ Especially effective was an ATP regenerating solutionReference Forscher and Oxford ³¹ containing ATP, phosphocreatine, and creatine phosphokinase (Figure 3). In contrast, inclusion of a nonhydrolysable analogue of ATP (β,γ-methylene ATP) was ineffective, implying that hydrolysis of ATP, necessary for phosphorylation reactions, rather than its mere presence was required. The findings led Mody, Salter, and MacDonaldReference Mody, Salter and MacDonald ³⁰ to state: “On the basis that high-energy phosphates promote phosphorylation we suggest that protein phosphorylation regulates ion flow through the NMDA channel.” And: “Intracellular phosphorylation could be a means of functional control of NMDA receptor channels, as important as the effects of extracellular regulatory factors like Mg²⁺,Reference Mayer, Westbrook and Guthrie ^{17 ,} Reference Nowak, Bregestovski, Ascher, Herbet and Prochiantz ¹⁸ Zn²⁺,Reference Peters, Koh and Choi ^{32 ,} Reference Westbrook and Mayer ³³ and glycine.”Reference Ascher and Nowak ^{34 ,} Reference Johnson and Ascher ³⁵

Figure 3

Rundown of NMDAR prevented by ATP regenerating solution. (A) Rundown of currents evoked by L-aspartate applied to cultured neurons when the intracellular recording solution does not contain high energy phosphates. Note the progressive amplitude decline of L-aspartate–evoked currents over time. Expanded traces show representative responses from the times indicated. (B) Summary graph from a series of comparable recordings in which current responses were normalized to the initial response recorded at t=0. (C) Rundown of currents evoked by L-aspartate is prevented when the intracellular recording solution contains an ATP regenerating solution containing ATP, phosphocreatine, and creatine phosphokinase. (D) Summary graph from a series of comparable recordings in which current responses were normalized to the initial response recorded at t=0. From MacDonald et al (1989) with permission.

The MacDonald laboratory would soon provide more definitive functional evidence supporting that iGluRs, in this case AMPA/kainate receptors, can be regulated by phosphorylation.Reference Wang, Salter and MacDonald ^{36 ,} Reference Wang, Taverna, Huang, MacDonald and Hampson ³⁷ These experiments were spearheaded by Lu-Yang Wang (then a PhD candidate and now at the SickKids Research Institute, University of Toronto) who would remain a lifelong friend of the MacDonalds. They first demonstrated that AMPA/kainate receptors rundown during intracellular dialysis unless an ATP-regenerating system was included in patch solutions, supporting that the maintenance of intracellular phosphorylation determines the magnitude of AMPA/kainate receptor-mediated currents. In an experimental tour-de-force, the authors then exploited the ability to alter the contents of the pipette solution through internal perfusion during a whole-cell recording. They demonstrated that intracellular perfusion with a solution containing cAMP and the catalytic subunit of PKA could reverse the rundown of AMPA/kainate receptor currents observed during an initial recording period in which the same neuron was perfused with a control recording solution. The reported ability of PKA to functionally regulate AMPA/kainate receptors was published in notable company as comparable findings were reported by Paul Greengard and associatesReference Greengard, Jen, Nairn and Stevens ³⁸ in the same issue of Science. The significance of their findings were immediately appreciated by Wang, Salter, and MacDonald,Reference Wang, Salter and MacDonald ³⁶ who stated: “…persistent protein kinase activity is associated with the early postsynaptic events leading to induction of long-term potentiation in the hippocampus. Regulation of AMPA-kainate receptors by such activity may thus contribute to some aspects of postsynaptic plasticity.”Reference Malenka ^{39 ,} Reference Malinow, Schulman and Tsien ⁴⁰

At this time, it was recognized that the rise in current amplitude observed in response to increased protein kinase activity could proceed via phosphorylation of an AMPA/kainate receptor subunit or of a regulatory protein associated with the channel. This was addressed by Wang and colleagues (1993) in the very first study to exploit knowledge of the primary sequence of a recently cloned iGluR (GluR6)Reference Egebjerg, Bettler, Hermans-Borgmeyer and Heinemann ⁴¹ and site-directed mutagenesis to abrogate consensus phosphorylation sites for PKA. The ability of PKA to augment kainate responses was abolished by alanine substitution at presumed serine phosphorylation sites of the GluR6 subunit when expressed in HEK293 cells. Corroborating findings were published the very same week by the Huganir laboratory.Reference Raymond, Blackstone and Huganir ⁴²

Particularly compelling evidence that NMDARs are regulated by phosphorylation/dephosphorylation would come from the groups of MacDonald,Reference Wang, Orser, Brautigan and MacDonald ⁴³ Salter,Reference Wang and Salter ⁴⁴ and Mody,Reference Lieberman and Mody ⁴⁵ who independently reported in the same issue of Nature the important role of serine/threonine and tyrosine phosphatases in regulating the function of NMDARs. The coinciding publications were notable for the fact that both Mody and Salter had trained with John as postdoctoral fellows. By now a recurring theme with former trainees, Mike Salter and John would continue to collaborate and maintain a close friendship to the very end. In the studies reported in Nature, the MacDonald group used the phosphatase inhibitors okadaic acid and calyculin A, at concentrations selective for the serine/threonine phosphatases PP1 and PP2A, and demonstrated a resulting enhancement of NMDAR channel activity. They exposed inside-out membrane patches from hippocampal neurons to purified PP1 or PP2A and recorded the effect upon NMDAR single-channel activity. Exposure to either PP1 or PP2A reduced NMDAR open probability; in each case, this effect could be reversed by subsequent application of okadaic acid. In concert with evidence from Lieberman and Mody showing that NMDAR are also subject to regulation by the Ca²⁺-dependent serine/threonine phosphatase PP2B (or calcineurin), these findings demonstrated the ability of phosphatases to constrain the activity of NMDARs. These results provided definitive evidence that endogenous serine/threonine protein kinases are constitutively active in maintaining the function of NMDARs. The work was important in providing a functional correlate to emerging biochemical evidence demonstrating that NMDARs are indeed subject to serine/threonine phosphorylation.Reference Tingley, Roche, Thompson and Huganir ⁴⁶ In a similar vein, Wang and Salter (1994) used comparable methodologies to provide the first evidence that NMDARs are regulated by protein tyrosine kinases and phosphatases. Collectively, these studies nicely complemented other evidence that NMDARs could be regulated by protein kinases.Reference Aniksztejn, Bregestovski and Ben-Ari ^{47 -} Reference Chen and Huang ⁵⁰ This effectively established the now well-recognized concept that the level of NMDAR activity is governed by the balance of kinase and phosphatase activity.

As is the hallmark of ground breaking discoveries, the findings from the MacDonald laboratory, proving that ionotropic glutamate receptors are regulated via protein phosphorylation, helped establish an entirely new area of study focused on elucidating how cell signaling events are orchestrated to regulate glutamate receptors, synaptic transmission, and plasticity. John and colleagues would go on to make major contribution in each of these areas, most notably in studies examining the regulation of NMDARs and consequent induction of synaptic plasticity by G-protein coupled receptors for glutamate (mGluR5Reference Kotecha, Jackson, Al Mahrouki, Roder, Orser and MacDonald ⁵¹ and mGluR2/3Reference Trepanier, Lei, Xie and MacDonald ⁵² ), acetylcholine (mAchRReference Lu, Xiong and Lei ^{53 ,} Reference Martyn, De Jaeger and Magalhães ⁵⁴ ), pituitary adenylate cyclase-activating polypeptide (PAC1R),Reference Macdonald, Weerapura and Beazely ^{55 ,} Reference Yang, Trepanier and Sidhu ⁵⁶ vasoactive intestinal peptide (VPACR),Reference Yang, Trepanier and Li ⁵⁷ and dopamine (D1RReference Yang, Trepanier and Sidhu ⁵⁶ and D2RReference Beazely, Tong, Wei, Van Tol, Sidhu and MacDonald ^{58 ,} Reference Kotecha, Oak and Jackson ⁵⁹ ).

Aberrant Ca²⁺ Permeable Nonselective Channel Activation in Models of Stroke and Neurodegenerative Disease

The introduction of AP5, the first NMDAR selective antagonist,Reference Davies, Francis, Jones and Watkins ⁶⁰ was followed by the first evidence implicating NMDARs as a causative agent underlying excitotoxicity.Reference Simon, Swan, Griffiths and Meldrum ⁶¹ The effectiveness of NMDAR antagonists in cell culture and animal models of stroke firmly established NMDAR-initiated neuronal injury as the dominant conceptual model not only for stroke-related neuronal cell death, but more broadly for neurodegenerative diseases. These and other findings led to great enthusiasm and optimism that pharmacological antagonism of NMDAR-mediated neurotoxicity could protect against brain ischemiaReference Albers, Goldberg and Choi ⁶² and on this basis clinical stroke trials of agents targeting the NMDAR were initiated. However, NMDAR antagonists failed to provide neuroprotection. Moreover, in many patients treatment had to be terminated because of side effects, which included psychosis. Accordingly, the field moved from great optimism to great pessimism about the prospects for NMDAR antagonists in treating brain ischemia.Reference Birmingham ^{63 ,} Reference Ikonomidou and Turski ⁶⁴

In light of their effectiveness in animal models, the failure of drugs targeting NMDARs was especially perplexing, and many proposals have been advanced to explain their failure. These include that the therapeutic window for drugs targeting NMDARs is quite narrow and that block of NMDAR was associated with dose-limiting side effects, including severe psychosis. These facts were acknowledged by John and colleagues in a commissioned opinion article in which they lament; “So we are left in a quandary. Blocking NMDA receptors is potentially neuroprotective in stroke but ideally the blocker should be present during the stroke and the patient must put up with being psychotic. An alternative approach is to abandon targeting NMDA receptors entirely, in favor of potential downstream ‘cell death’ targets.”Reference MacDonald, Xiong and Jackson ⁶⁵ At the time of this opinion article, this approach was already established in the MacDonald laboratory and would represent an area of focus for years to come. Key strategies used for identifying/targeting downstream cell death targets included interfering with the assembly of protein signaling complexes that tether components responsible for initiating cell death to NMDARs and preventing the activation of additional sources of Ca²⁺ entry recruited downstream of the NMDAR. Much of this work was undertaken in close collaboration with the laboratory of Michael Tymianski, another clinician-scientist who trained in the MacDonald laboratory and would remain a friend and longstanding collaborator of John’s.

Uncoupling of NMDARs From the Ca²⁺-Dependent Production of Cell Damaging Oxidative/Nitrosative Stress

Despite its recognized importance to neurological function in health and in disease, the weight of evidence to date suggests that NMDARs are not a viable therapeutic target, at least not for agents that indiscriminately reduce channel function. There are exceptions to this rule because a few select drugs are used clinically in a limited capacity (e.g. ketamine, memantine), but certainly not in the context of neuroprotection in stroke. And yet from the quagmire of more than 30 years of research seeking to identify neuroprotective agents based on the excitotoxic theory of ischemic stroke, a promising neuroprotective agent targeting the NMDAR emerged through research headed by Michael Tymianski in close collaboration with John and others, including Mike Salter and Yu-Tian Wang. Counterintuitively, the drug in question does not inhibit NMDAR function at all. Rather, it uncouples the NMDAR from a scaffolding complex largely responsible for excitotoxicity initiated following aberrant NMDAR-mediated Ca²⁺ entry. Neuroprotection by TAT-N2B9c, now renamed NA-1, in rodent stroke modelsReference Aarts, Iihara and Wei ^{66 -} Reference Sun, Doucette and Liu ⁶⁸ has been extended to non-human primatesReference Cook, Teves and Tymianski ^{69 ,} Reference Cook, Teves and Tymianski ⁷⁰ and more recently humans.Reference Hill, Martin and Mikulis ⁷¹ Of note, a phase I dose escalation trial in healthy volunteers showed that NA-1 administration was well-tolerated, without evidence of drug-related adverse events. NA-1 is currently in a trial called FRONTIER (Field Randomization of NA-1 in Early Responders), which aims to have paramedics administer NA-1 within 3 hours of symptom onset to patients with suspected acute stroke. If successful, the validation of such an approach may not only provide protection, but also extend the narrow window (currently only 3-4.5 hours) for treating patients with clot busting agents. NA-1 represents the lead therapeutic compound of NoNo Inc, a company headed by Michael Tymianski, of which John was a founding member and served on the Scientific Advisory Board.

Contribution of Ca²⁺ Permeable Nonselective Cation Channels to Excitotoxicity

The failure of human stroke trials forced a critical reevaluation of the excitotoxic model of stroke with an eye toward uncovering NMDAR-independent mechanisms that contribute to cell death. With this in mind, and in conjunction with Michael Tymianski’s group, the MacDonald laboratory probed for previously unrecognized contributors to anoxic cell death. They began by reexamining the effects of NMDARs blockers on anoxic cell death in cultured neurons. A key initial finding was that an antiexcitotoxic cocktail of blockers targeting glutamate receptors (i.e. AMPARs and NMDARs) and voltage-gated Ca²⁺ channels could prevent cell death resulting from brief (1 hour) but not extended episodes of oxygen-glucose deprivation (OGD; 1.5 hours or more). Cell death associated with extended OGD remained Ca²⁺ dependent and involved the entry of Ca²⁺ from the extracellular space. This suggested that additional routes of Ca²⁺ entry, distinct from those gated by NMDARs and voltage-gated Ca²⁺ channels, were recruited with delay during extended OGD. Conventional pathways for Ca²⁺ uptake, including Na⁺/Ca²⁺ exchanger, were ruled out, implying that a previously unrecognized Ca²⁺ permeable conductance was responsible. How then could they identify the conductance involved? Electrophysiological evidence from recordings conducted in cultured neurons exposed to chemical anoxia, combined with knowledge from past studies conducted by the MacDonald laboratory would provide key pieces to resolving this puzzle.

Electrophysiological recordings demonstrated that extended exposure of cultured neurons to chemical anoxia evoked a current that could be blocked by Zn²⁺ and the trivalent cation, Gd³⁺. I-V curves were outwardly rectifying and reversed at a membrane potential near 0 mV, supporting the involvement of a nonselective cation channel. Proof of Ca²⁺ permeability followed from the finding that the reversal potential was predictably altered by varying the extracellular concentration of Ca²⁺. Interestingly, reducing the extracellular concentration of Ca²⁺ dramatically increased the amplitude of the current induced by chemical anoxia and reduced current rectification. This suggested that the monovalent cation permeability is reduced by Ca²⁺ in a voltage-dependent manner. On the basis of comparable sensitivity of OGD-evoked Reference Lieberman and Mody ⁴⁵ Ca²⁺ uptake to Gd³⁺ during extended OGD, the conductance characterized biophysically during chemical anoxia was termed I_OGD.Reference Aarts, Iihara and Wei ⁶⁶

Many of the attributes of I_OGD matched those described by the MacDonald laboratory in studies exploring a calcium-sensing nonselective cation channel (I_csNSC).Reference Xiong, Lu and MacDonald ⁷² The identification of I_csNSC stemmed serendipitously from experiments undertaken by the MacDonald laboratory while exploring mechanisms contributing to Ca²⁺-dependent inactivation of NMDARs. The MacDonald laboratory and others had shown that Ca²⁺ entry via NMDARs recruits a negative feedback mechanism that limits the pathological entry of Ca²⁺. Ca²⁺-dependent inactivation of NMDARs, initiated when Ca²⁺/CaM associate with the C-term tail of the GluN1 subunit,Reference MacDonald, Xiong, Lu, Raouf and Orser ⁷³ is routinely studied by comparing the NMDA-evoked responses recorded in the presence or absence of extracellular Ca²⁺. Characteristically, when conducting these experiments, Ca²⁺ is reduced or entirely eliminated only from the NMDA-containing solution under the assumption that the application of a low Ca²⁺ solution would not by itself initiate a response. Quite surprisingly, no one had tested this explicitly by conducting a simple and necessary control experiment, namely, to apply a low Ca²⁺ containing solution and determine whether a membrane response is elicited. The MacDonald laboratory noted that graded decreases in the extracellular concentration of Ca²⁺ (by as little as 100 µM) evoked graded increases in the amplitude of a nonselective cation current in hippocampal neurons.Reference Xiong, Lu and MacDonald ⁷² Like I_OGD, the current could be inhibited by a variety of divalent cations, including Ca²⁺, Mg²⁺, Ba²⁺, and Cd²⁺, and could be blocked by Gd³⁺.Reference Xiong and MacDonald ⁷⁴ This revealed a close correspondence between the properties of I_OGD and I_csNSC and suggested that the same channel was likely responsible for both conductances. However, the identity of the channel responsible was not known and, on this basis, the MacDonald laboratory initiated a project to identify the channel underlying I_csNSC through expression cloning in collaboration with Dr. Hubert Van Tol (a close friend and collaborator, deceased 2006). Notoriously labor-intensive, it is fortunate that the identification of the channel underlying I_csNSC did not have to await completion of expression cloning. Instead, the answer would come from efforts aimed at cloning transient receptor potential (TRP) channels.Reference Clapham, Runnels and Strubing ⁷⁵

The identification of TRPM7 and characterization of key properties,Reference Runnels, Yue and Clapham ⁷⁶ matching those reported for I_OGD and I_csNSC, would provide the final piece of the puzzle, allowing the MacDonald and Tymianski laboratories to move forward and demonstrate that TRPM7 contributes to the conductance underlying I_OGD Reference Aarts, Iihara and Wei ⁶⁶ and I_csNSC.Reference Wei, Sun and Olah ⁷⁷ Because specific pharmacological blockers for TRPM7 were not available, and in the absence of a TRPM7 knockout mouse, the critical evidence allowing this to be conclusively demonstrated was the use of RNA interference by small interfering RNA (siRNA), which up until that point had proven difficult to implement in cultured neurons.Reference Krichevsky and Kosik ⁷⁸ To suppress TRPM7 expression, primary cultured neurons were transfected with 21-nt siRNA duplex targeting TRPM7 (siRNA_TRPM7-1). Knockdown of TRPM7 by siRNA_TRPM7-1 suppressed I_OGD, reduced Reference Lieberman and Mody ⁴⁵ Ca²⁺ uptake, attenuated overall reactive oxygen specied production and extended the survival of cultured neurons exposed to OGD for up to 3 hours. In a follow-up articled published in 2009, the MacDonald group demonstrated that suppressing the expression of TRPM7, by viral mediated delivery of a specific short hairpin RNA sequence, conferred resistance to neuronal death after brain ischemia and preserved neuronal morphology and function.Reference Sun, Jackson and Martin ⁷⁹ Also, it prevented ischemia-induced deficits in LTP and fear-associated and spatial navigational memory tasks. This study established that regional suppression of TRPM7 is feasible and well-tolerated, a critically important point if pharmacological inhibition of TRPM7 is to be exploited for therapeutic benefit. Collectively, these studies revealed a critical role for TRPM7 in mediating anoxic neuronal cell death.

Beyond TRPM7, the identification of additional Ca²⁺ permeable nonselective cation channels contributing to anoxic cell death would prove to be a very fertile area for John and his collaborators with important contribution of the MacDonald laboratory to the discovery of ASIC channels,Reference Xiong, Zhu and Chu ⁸⁰ pannexin channels,Reference Thompson, Jackson and Olah ⁸¹ and TRPM2.Reference Belrose, Xie, Gierszewski, MacDonald and Jackson ^{82 -} Reference Ostapchenko, Chen and Guzman ⁸⁵ This includes work completed in the last few days of John’s life and just recently published, in which absence of TRPM2 was shown to decrease signs of neuronal toxicity and improve cognition in aged Alzheimer’s model mice.Reference Ostapchenko, Chen and Guzman ⁸⁵ The rationale for these studies, albeit speculative, derived in large part from work published by John and colleagues.Reference Belrose, Xie, Gierszewski, MacDonald and Jackson ^{82 -} Reference Xie, Belrose and Lei ⁸⁴ These studies linked TRPM2 activation to cellular pathways associated with Alzheimer’s disease pathology (e.g. increased oxidative stress,Reference Olah, Jackson and Li ⁸³ reduced neuronal oxidant defense,Reference Belrose, Xie, Gierszewski, MacDonald and Jackson ⁸² and increased GSK3β activityReference Xie, Belrose and Lei ⁸⁴ ). In the absence of experimental evidence directly linking TRPM2 to Alzheimer’s disease pathology, a leap of faith was needed to initiate longitudinal behavioral studies in which TRPM2 knockout was crossed with Alzheimer’s mouse model (APP_Swe/PS1∆E9). John had the conviction to initiate these studies when he approached us to undertake these experiments collaboratively. The positive outcome of these experiments is particularly rewarding to all of us.

Mentoring and Administrative Achievements

In addition to his remarkable scientific achievements, John assumed a number of leadership roles and made important contributions in this capacity as well. John had a significant role in the recent expansion of the Department of Physiology at the University of Toronto. This is one of the largest basic science departments in North America, with more than 120 faculty members. John was appointed initially at the Department of Pharmacology at the University of Toronto; however, he moved to the Department of Physiology in 1990 to become a founding member of the Medical Research Council (MRC) group in nerve cells and synapses, with Milton Charlton and several other colleagues. John served initially as graduate coordinator before assuming the Chairmanship of the Department (2001-2008). During his tenure, the department obtained 11 Canada Research Chairs, 3 Canadian Institutes of Health Research new investigator awards, a heart and stroke investigator award, and the Michael Smith Chair in neurosciences and mental health. John was well-known for his strategic partnerships with other departments and research institutes. It is no surprise that John is dearly missed by his departmental colleagues and friends (see http://www.physiology.utoronto.ca/history-department for a detailed account of John’s leadership).

In 2008, John moved his laboratory to the University of Western Ontario, where he would work until his retirement in 2013. He assumed the position of director of the Robarts Research Institute for 3 years, a turbulent time because of the market crash of 2008. However, John continued to deliver outstanding advice and support for new investigators. The three of us are particularly grateful for his support, leadership, and insightful collaborations. His presence at Robarts had a huge impact on collaborative grants and manuscripts and our careers. We particularly miss our afternoon teas together in which ideas would flow freely, leading to several publications in collaboration despite the short period of time.Reference Martyn, De Jaeger and Magalhães ^{54 ,} Reference Ostapchenko, Chen and Guzman ^{85 ,} Reference Ostapchenko, Beraldo and Mohammad ⁸⁶ Science was an important guide for John and even during his short time at Western he managed to publish outstanding work that has had a lasting impact on all around him. We are not just proud of having had time with him as a colleague, friend, and a mentor, we and all his collaborators and trainees are also extremely grateful to have had John impact our lives.