- Email: [email protected]
The Many Worlds of Plasticity Rules
TINS 1367 No. of Pages 4
Spotlight
The Many Worlds of Plasticity Rules Jackie Schiller,1,* Shai Berlin,1 and Dori Derdikman1 Two recent papers have tackled the fundamental questions of how place fields are formed in a new environment and what plasticity mechanisms contribute to this process. Bittner et al., in their recent publication, discovered a novel plasticity rule that, in contrast to previous rules, spans the behavioral, seconds-long, timescale. Sheffield et al. have monitored, for the first time, dendritic activity during place field formation, and show the emergence of spatially tuned local NMDA spikes in basal dendrites of CA1 neurons. Together, these papers suggest that multiple complementary dendritic plasticity mechanisms may contribute to place field formation in changing environmental contexts. Plasticity rules are considered to be part of the fundamental building blocks of brain wiring. One of the most influential plasticity rules explored in the past decades is Hebbian spike[127_TD$IF]-timing-dependent plasticity (STDP), where synapses are either potentiated or depressed according to the causal order of pre- and postsynaptic activation within a narrow, milliseconds-long, time-window (Figure 1A) [1]. Potentiation is expressed when the postsynaptic potentials repeatedly contribute, typically dozens of times, to postsynaptic firing. Deviations from this directive have emerged, the most prominent being local Hebbian, anti-Hebbian, and non-Hebbian rules, with some of these processes invoking local dendritic spikes as postsynaptic mediators of
plasticity [2,3]. In a recent study, Sheffield et al. [4] directly monitored dendritic activity during place field formation and observed the emergence of local dendritic NMDA spikes that preceded place field formation and were tuned to its future spatial location. They suggest that these local NMDA spikes serve to potentiate the spatially clustered inputs which participated in their generation. In another recent study, Bittner et al. demonstrated a new form of non-Hebbian plasticity in the hippocampus, which notably – and in contrast to most previous rules – extends over several seconds [5]. The authors coined the term ‘behavioral timescale synaptic plasticity’ (BTSP), distinguishing it from the traditional milliseconds-long STDP rule, as well as from other nonHebbian rules. They suggest that BTSP may underlie place field formation.
Mechanisms for Place Field formation One of the brain systems where plasticitymediated changes can be best studied is the navigation system of the hippocampus and related brain regions. The two recent papers mentioned above [4,5] shed new light on the cellular events leading to the formation of emerging place fields in CA1 pyramidal neurons. Both reports show that, in a subset of neurons, the emergence of place fields requires potentiation of inputs. However, it seems that the underlying plasticity mechanisms proposed in these studies differ, arguing that multiple mechanisms [128_TD$IF]may contribute to the formation of place fields. This notion is further strengthened by the observation that 50% of place fields emerge as soon as the first lap in a novel environment, precluding the need for synaptic potentiation for their formation [4]. Strikingly, in some instances[129_TD$IF], place fields can form abruptly in a novel environment. In such cases place fields can suddenly emerge during multiple traversals of the
environment, without any particular or indicative somatic activity [4–6]. This abrupt formation suggests a plasticity mechanism different from STDP, most likely mediated by active dendritic mechanisms such as dendritic spikes [2]. Sheffield et al. suggest a self-organizing mechanism, based on local synaptic plasticity, that is initiated by the inputs and is mediated by local, likely NMDA, dendritic spikes (Figure 1B–E). Indeed, local calcium transients in basal dendrites, reminiscent of the dendritic hallmark of NMDA spikes, appeared several laps before somatic place field formation. Collectively, these local dendritic NMDA spikes can summate to provide the prolonged time-window needed for plasticity of inputs which are stretched in time [3]. Bittner et al. propose an additional intriguing mechanism. Inputs that did not necessarily participate in the generation of a plateau potential could be potentiated over an extended time-period by the sudden appearance of a plateau potential, leading to the appearance of a place field. The authors [130_TD$IF]go on to propose a mechanism that, rather than being self-organizing, requires an external instructive signal to cause the emergence of the plateau [13_TD$IF]potential that plays a key role in potentiating synaptic inputs (Figure 1B–E). Unlike the previously described gradual mechanisms such as STDP, BTSP expresses abruptly – within one to two laps of the animal exploring a new environment. A combination of the two mechanisms proposed by Bittner et al. (2017) and by Sheffield et al. (2017) is also possible. For instance, we envisage that local dendritic NMDA spikes may potentiate local inputs that, upon crossing a threshold, will eventually participate in plateau-potential generation, which in turn potentiates additional inputs over an extended timewindow. If true, one can expect to observe subthreshold correlates of the local dendritic spikes at the soma before
Trends in Neurosciences, Month Year, Vol. xx, No. yy
1
TINS 1367 No. of Pages 4
(A)
(B)
(C)
Δt bAP
−50
50 100 Time (ms)
(D)
NMDA spike
1s
−150 −100 −50 0 50 Time (ms)
−4000
−2000
0 2000 Time (ms)
4000
(E)
Figure 1. Plasticity Rules and Their Behavioral Relevance. (A) Hebbian STDP with both LTP and LTD occurring in a short milliseconds-long time-window. (B) Non-Hebbian LTP with local NMDA spikes and a time-window extending over 100–150 ms. (C) Non-Hebbian LTP with plateau potentials and a time-window extending over several seconds. (D) Encoding an episodic memory sequence with STDP. For example, an animal traversing a known territory (depicted by a complete trajectory in the map), with various events occurring on its way. (E) Non-Hebbian plasticity (as in B,C) is more adapted to building the internal cognitive map in an unknown terrain or context (depicted by an incomplete map, [126_TD$IF]broken grey lines) where there is propensity for the formation of new place fields. Plots shown in (A), (B), and (C) are adapted from [1], [3], and [5], respectively. Abbreviations: LTD, long-term depression; LTP, long-term potentiation; STDP, spike[127_TD$IF]-timing-dependent plasticity.
2
Trends in Neurosciences, Month Year, Vol. xx, No. yy
TINS 1367 No. of Pages 4
the appearance of the plateau potential. However, parvalbumin interneuron activation in the axosomatic region could conceal this type of activity at the soma, as suggested by the findings in Sheffield et al.
temporal integration, owing to the large amounts of ‘spare’ G proteins at the membrane and the slow GTPase activity of Ga subunits. Future experiments will be necessary to uncover the cellular and molecular mechanisms that underlie this novel form of plasticity.
Taken together, thereby, place fields could emerge via multiple plasticity mechBehavioral Relevance of Multiple anisms that are contingent on the Plasticity Rules strength of the inputs, the state of the Why is BTSP needed? Scientists have animal, external conditions, and more. been busy, for some time, trying to understand how events at behavioral timescales could be encoded by synapses Possible Mechanisms for BTSP A prolonged, non-Hebbian, time-window that operate at substantially faster timefor plasticity, extending some 150 ms scales. When considering the functions of backwards and several tens of millisec- the hippocampus, timescales of different onds forwards (Figure 1B), was previously orders of magnitude should be described for distal basal dendritic synap- accounted for. The hippocampus was ses of layer 2–3 neocortical pyramidal suggested to encode episodic memories, neurons, operating by an NMDA dendritic as well as mapping of environments, in spike and a BDNF induction mechanism different contexts. Typical episodic events [3]. Strikingly, BTSP shows related char- contain conjunctive information about acteristics, but with a fivefold increase of where, when, and what we are doing. the time-window (Figure 1C). BTSP This requires registration at accurate requires both NMDA receptor (NMDAR) timescales. [134_TD$IF]By contrast, identifying differand L-type calcium channels, similarly to ent contexts does not necessarily dendritic spike plasticity, as previously demand the same temporal accuracy: a reported for CA1 pyramidal neurons [7]. red room is a red room, two seconds before and two seconds later. Therefore, How can a plateau-potential event, it is possible that different plasticity rules occurring hundreds of milliseconds to can underlie episodic versus contextual several seconds before or after activation encoding mechanisms. Specifically, of inputs, cause synaptic plasticity of BTSP and local NMDA spikes can be these inputs? A possible mechanism associated with contextual encoding, may implicate the prolonged, seconds- whereas traditional STDP may be better long, binding of glutamate to [132_TD$IF]NMDARs suited to handling episodic memories (‘priming’) [8]. These bound but non-con- (Figure 1D,E). ducting [13_TD$IF]NMDARs may be opened once the plateau potential appears, suggesting STDP can be used to resolve episodic a conditional bistability mechanism. Alter- encoding via neural mechanisms of involving theta natively, the time-gap could be bridged by time-compression temporal integration through activation of phase-precession, which allows longer biochemical cascades which could sup- sequences in time to be compressed into port potentiation once plateau potentials shorter timescales, making the comare evoked. For example, metabotropic pressed sequences amenable to tradireceptors, coupled to G proteins, could tional Hebbian STDP [10]. The easily bridge a seconds-long time-interval requirement for multiple pairings in time, [9]. They can implement chemical that are necessary for STDP expression,
can be realized by the fact that the same event will appear in multiple theta cycles during compression, only to be replayed, at a later time, during sharp-wave ripple events. BTSP, [135_TD$IF]by contrast, is well suited for contextual encoding because it extends over several seconds (forward and backwards) in time. It can associate between events, such as being in a given position in a particular room, without any concern about the time-difference between the two. Furthermore, the single-pairing nature of BTSP is highly suited for processing novel contexts efficiently, such as the formation of new place fields with changing context in the environment. Going from biology to robotics, it is interesting to note that the above discussion is intimately related to the issue of simultaneous localization and mapping (SLAM), a known problem in robotic navigation systems. The task of the navigating system is to chart the environment (‘mapping’) and, while doing so, to determine the position of the robot within that environment (‘localization’). The challenge in mapping is that one needs to know where one is, whereas for localization one needs a mapped environment – an inherent circular problem. In conclusion, it is possible that multiple plasticity mechanisms have evolved in the hippocampus to handle two very different components of spatial memory: mapping of the environment in diverse contexts (BTSP; local NMDA spike plasticity) versus localization and the formation of temporal episodes in a specific context (STDP). Acknowledgments We thank Y. Schiller, S. Marom, and Guy Major for [136_TD$IF]24commenting on the manuscript. We thank Allen and Jewel Prince Center for Neurodegenerative Disorders of the Brain (J.S. and D.D.), the Rappaport Foundation (J.S. and D.D.), the The Adelis Foundation - joint Technion and Weizmann program (J.S. and D.D.), and the Israel Science Foundation (ISF-1096/17; S.B.).
Trends in Neurosciences, Month Year, Vol. xx, No. yy
3
TINS 1367 No. of Pages 4
1
Department of Neuroscience, The Rappaport Faculty of Medicine and Research Institute, Technion-Israel Institute of Technology, 1 Efron Street, Haifa 31096, Israel
3. Gordon, U. et al. (2006) Plasticity compartments in basal dendrites of neocortical pyramidal neurons. J. Neurosci. 26, 12717–12726
7. Remy, S. and Spruston, N. (2007) Dendritic spikes induce single-burst long-term potentiation. Proc. Natl. Acad. Sci. U. S. A. 104, 17192–17197
*Correspondence: [email protected] (J. Schiller).
4. Sheffield, M.E.J. et al. (2017) Increased prevalence of calcium transients across the dendritic arbor during place field formation. Neuron 96, 490–504
8. Paoletti, P. et al. (2013) NMDA receptor subunit diversity: impact on receptor properties, synaptic plasticity and disease. Nat. Rev. Neurosci. 14, 383–400
5. Bittner, K.C. et al. (2017) Behavioral time scale synaptic plasticity underlies CA1 place fields. Science 357, 1033– 1036
9. Pavlos, N.J. and Friedman, P.A. (2017) GPCR signaling and trafficking: the long and short of it. Trends Endocrinol. Metab. 28, 213–226
6. Lee, D. et al. (2012) Hippocampal place fields emerge upon single-cell manipulation of excitability during behavior. Science 337, 849–853
10. Jensen, O. and Lisman, J.E. (2005) Hippocampal sequence-encoding driven by a cortical multi-item working memory buffer. Trends Neurosci. 28, 67–72
https://doi.org/10.1016/j.tins.2018.01.006 References 1. Feldman, D.E. (2012) The spike-timing dependence of plasticity. Neuron 75, 556–571 2. Lisman, J. and Spruston, N. (2005) Postsynaptic depolarization requirements for LTP and LTD: a critique of spike timing-dependent plasticity. Nat. Neurosci. 8, 839–841
4
Trends in Neurosciences, Month Year, Vol. xx, No. yy
Spotlight
The Many Worlds of Plasticity Rules Jackie Schiller,1,* Shai Berlin,1 and Dori Derdikman1 Two recent papers have tackled the fundamental questions of how place fields are formed in a new environment and what plasticity mechanisms contribute to this process. Bittner et al., in their recent publication, discovered a novel plasticity rule that, in contrast to previous rules, spans the behavioral, seconds-long, timescale. Sheffield et al. have monitored, for the first time, dendritic activity during place field formation, and show the emergence of spatially tuned local NMDA spikes in basal dendrites of CA1 neurons. Together, these papers suggest that multiple complementary dendritic plasticity mechanisms may contribute to place field formation in changing environmental contexts. Plasticity rules are considered to be part of the fundamental building blocks of brain wiring. One of the most influential plasticity rules explored in the past decades is Hebbian spike[127_TD$IF]-timing-dependent plasticity (STDP), where synapses are either potentiated or depressed according to the causal order of pre- and postsynaptic activation within a narrow, milliseconds-long, time-window (Figure 1A) [1]. Potentiation is expressed when the postsynaptic potentials repeatedly contribute, typically dozens of times, to postsynaptic firing. Deviations from this directive have emerged, the most prominent being local Hebbian, anti-Hebbian, and non-Hebbian rules, with some of these processes invoking local dendritic spikes as postsynaptic mediators of
plasticity [2,3]. In a recent study, Sheffield et al. [4] directly monitored dendritic activity during place field formation and observed the emergence of local dendritic NMDA spikes that preceded place field formation and were tuned to its future spatial location. They suggest that these local NMDA spikes serve to potentiate the spatially clustered inputs which participated in their generation. In another recent study, Bittner et al. demonstrated a new form of non-Hebbian plasticity in the hippocampus, which notably – and in contrast to most previous rules – extends over several seconds [5]. The authors coined the term ‘behavioral timescale synaptic plasticity’ (BTSP), distinguishing it from the traditional milliseconds-long STDP rule, as well as from other nonHebbian rules. They suggest that BTSP may underlie place field formation.
Mechanisms for Place Field formation One of the brain systems where plasticitymediated changes can be best studied is the navigation system of the hippocampus and related brain regions. The two recent papers mentioned above [4,5] shed new light on the cellular events leading to the formation of emerging place fields in CA1 pyramidal neurons. Both reports show that, in a subset of neurons, the emergence of place fields requires potentiation of inputs. However, it seems that the underlying plasticity mechanisms proposed in these studies differ, arguing that multiple mechanisms [128_TD$IF]may contribute to the formation of place fields. This notion is further strengthened by the observation that 50% of place fields emerge as soon as the first lap in a novel environment, precluding the need for synaptic potentiation for their formation [4]. Strikingly, in some instances[129_TD$IF], place fields can form abruptly in a novel environment. In such cases place fields can suddenly emerge during multiple traversals of the
environment, without any particular or indicative somatic activity [4–6]. This abrupt formation suggests a plasticity mechanism different from STDP, most likely mediated by active dendritic mechanisms such as dendritic spikes [2]. Sheffield et al. suggest a self-organizing mechanism, based on local synaptic plasticity, that is initiated by the inputs and is mediated by local, likely NMDA, dendritic spikes (Figure 1B–E). Indeed, local calcium transients in basal dendrites, reminiscent of the dendritic hallmark of NMDA spikes, appeared several laps before somatic place field formation. Collectively, these local dendritic NMDA spikes can summate to provide the prolonged time-window needed for plasticity of inputs which are stretched in time [3]. Bittner et al. propose an additional intriguing mechanism. Inputs that did not necessarily participate in the generation of a plateau potential could be potentiated over an extended time-period by the sudden appearance of a plateau potential, leading to the appearance of a place field. The authors [130_TD$IF]go on to propose a mechanism that, rather than being self-organizing, requires an external instructive signal to cause the emergence of the plateau [13_TD$IF]potential that plays a key role in potentiating synaptic inputs (Figure 1B–E). Unlike the previously described gradual mechanisms such as STDP, BTSP expresses abruptly – within one to two laps of the animal exploring a new environment. A combination of the two mechanisms proposed by Bittner et al. (2017) and by Sheffield et al. (2017) is also possible. For instance, we envisage that local dendritic NMDA spikes may potentiate local inputs that, upon crossing a threshold, will eventually participate in plateau-potential generation, which in turn potentiates additional inputs over an extended timewindow. If true, one can expect to observe subthreshold correlates of the local dendritic spikes at the soma before
Trends in Neurosciences, Month Year, Vol. xx, No. yy
1
TINS 1367 No. of Pages 4
(A)
(B)
(C)
Δt bAP
−50
50 100 Time (ms)
(D)
NMDA spike
1s
−150 −100 −50 0 50 Time (ms)
−4000
−2000
0 2000 Time (ms)
4000
(E)
Figure 1. Plasticity Rules and Their Behavioral Relevance. (A) Hebbian STDP with both LTP and LTD occurring in a short milliseconds-long time-window. (B) Non-Hebbian LTP with local NMDA spikes and a time-window extending over 100–150 ms. (C) Non-Hebbian LTP with plateau potentials and a time-window extending over several seconds. (D) Encoding an episodic memory sequence with STDP. For example, an animal traversing a known territory (depicted by a complete trajectory in the map), with various events occurring on its way. (E) Non-Hebbian plasticity (as in B,C) is more adapted to building the internal cognitive map in an unknown terrain or context (depicted by an incomplete map, [126_TD$IF]broken grey lines) where there is propensity for the formation of new place fields. Plots shown in (A), (B), and (C) are adapted from [1], [3], and [5], respectively. Abbreviations: LTD, long-term depression; LTP, long-term potentiation; STDP, spike[127_TD$IF]-timing-dependent plasticity.
2
Trends in Neurosciences, Month Year, Vol. xx, No. yy
TINS 1367 No. of Pages 4
the appearance of the plateau potential. However, parvalbumin interneuron activation in the axosomatic region could conceal this type of activity at the soma, as suggested by the findings in Sheffield et al.
temporal integration, owing to the large amounts of ‘spare’ G proteins at the membrane and the slow GTPase activity of Ga subunits. Future experiments will be necessary to uncover the cellular and molecular mechanisms that underlie this novel form of plasticity.
Taken together, thereby, place fields could emerge via multiple plasticity mechBehavioral Relevance of Multiple anisms that are contingent on the Plasticity Rules strength of the inputs, the state of the Why is BTSP needed? Scientists have animal, external conditions, and more. been busy, for some time, trying to understand how events at behavioral timescales could be encoded by synapses Possible Mechanisms for BTSP A prolonged, non-Hebbian, time-window that operate at substantially faster timefor plasticity, extending some 150 ms scales. When considering the functions of backwards and several tens of millisec- the hippocampus, timescales of different onds forwards (Figure 1B), was previously orders of magnitude should be described for distal basal dendritic synap- accounted for. The hippocampus was ses of layer 2–3 neocortical pyramidal suggested to encode episodic memories, neurons, operating by an NMDA dendritic as well as mapping of environments, in spike and a BDNF induction mechanism different contexts. Typical episodic events [3]. Strikingly, BTSP shows related char- contain conjunctive information about acteristics, but with a fivefold increase of where, when, and what we are doing. the time-window (Figure 1C). BTSP This requires registration at accurate requires both NMDA receptor (NMDAR) timescales. [134_TD$IF]By contrast, identifying differand L-type calcium channels, similarly to ent contexts does not necessarily dendritic spike plasticity, as previously demand the same temporal accuracy: a reported for CA1 pyramidal neurons [7]. red room is a red room, two seconds before and two seconds later. Therefore, How can a plateau-potential event, it is possible that different plasticity rules occurring hundreds of milliseconds to can underlie episodic versus contextual several seconds before or after activation encoding mechanisms. Specifically, of inputs, cause synaptic plasticity of BTSP and local NMDA spikes can be these inputs? A possible mechanism associated with contextual encoding, may implicate the prolonged, seconds- whereas traditional STDP may be better long, binding of glutamate to [132_TD$IF]NMDARs suited to handling episodic memories (‘priming’) [8]. These bound but non-con- (Figure 1D,E). ducting [13_TD$IF]NMDARs may be opened once the plateau potential appears, suggesting STDP can be used to resolve episodic a conditional bistability mechanism. Alter- encoding via neural mechanisms of involving theta natively, the time-gap could be bridged by time-compression temporal integration through activation of phase-precession, which allows longer biochemical cascades which could sup- sequences in time to be compressed into port potentiation once plateau potentials shorter timescales, making the comare evoked. For example, metabotropic pressed sequences amenable to tradireceptors, coupled to G proteins, could tional Hebbian STDP [10]. The easily bridge a seconds-long time-interval requirement for multiple pairings in time, [9]. They can implement chemical that are necessary for STDP expression,
can be realized by the fact that the same event will appear in multiple theta cycles during compression, only to be replayed, at a later time, during sharp-wave ripple events. BTSP, [135_TD$IF]by contrast, is well suited for contextual encoding because it extends over several seconds (forward and backwards) in time. It can associate between events, such as being in a given position in a particular room, without any concern about the time-difference between the two. Furthermore, the single-pairing nature of BTSP is highly suited for processing novel contexts efficiently, such as the formation of new place fields with changing context in the environment. Going from biology to robotics, it is interesting to note that the above discussion is intimately related to the issue of simultaneous localization and mapping (SLAM), a known problem in robotic navigation systems. The task of the navigating system is to chart the environment (‘mapping’) and, while doing so, to determine the position of the robot within that environment (‘localization’). The challenge in mapping is that one needs to know where one is, whereas for localization one needs a mapped environment – an inherent circular problem. In conclusion, it is possible that multiple plasticity mechanisms have evolved in the hippocampus to handle two very different components of spatial memory: mapping of the environment in diverse contexts (BTSP; local NMDA spike plasticity) versus localization and the formation of temporal episodes in a specific context (STDP). Acknowledgments We thank Y. Schiller, S. Marom, and Guy Major for [136_TD$IF]24commenting on the manuscript. We thank Allen and Jewel Prince Center for Neurodegenerative Disorders of the Brain (J.S. and D.D.), the Rappaport Foundation (J.S. and D.D.), the The Adelis Foundation - joint Technion and Weizmann program (J.S. and D.D.), and the Israel Science Foundation (ISF-1096/17; S.B.).
Trends in Neurosciences, Month Year, Vol. xx, No. yy
3
TINS 1367 No. of Pages 4
1
Department of Neuroscience, The Rappaport Faculty of Medicine and Research Institute, Technion-Israel Institute of Technology, 1 Efron Street, Haifa 31096, Israel
3. Gordon, U. et al. (2006) Plasticity compartments in basal dendrites of neocortical pyramidal neurons. J. Neurosci. 26, 12717–12726
7. Remy, S. and Spruston, N. (2007) Dendritic spikes induce single-burst long-term potentiation. Proc. Natl. Acad. Sci. U. S. A. 104, 17192–17197
*Correspondence: [email protected] (J. Schiller).
4. Sheffield, M.E.J. et al. (2017) Increased prevalence of calcium transients across the dendritic arbor during place field formation. Neuron 96, 490–504
8. Paoletti, P. et al. (2013) NMDA receptor subunit diversity: impact on receptor properties, synaptic plasticity and disease. Nat. Rev. Neurosci. 14, 383–400
5. Bittner, K.C. et al. (2017) Behavioral time scale synaptic plasticity underlies CA1 place fields. Science 357, 1033– 1036
9. Pavlos, N.J. and Friedman, P.A. (2017) GPCR signaling and trafficking: the long and short of it. Trends Endocrinol. Metab. 28, 213–226
6. Lee, D. et al. (2012) Hippocampal place fields emerge upon single-cell manipulation of excitability during behavior. Science 337, 849–853
10. Jensen, O. and Lisman, J.E. (2005) Hippocampal sequence-encoding driven by a cortical multi-item working memory buffer. Trends Neurosci. 28, 67–72
https://doi.org/10.1016/j.tins.2018.01.006 References 1. Feldman, D.E. (2012) The spike-timing dependence of plasticity. Neuron 75, 556–571 2. Lisman, J. and Spruston, N. (2005) Postsynaptic depolarization requirements for LTP and LTD: a critique of spike timing-dependent plasticity. Nat. Neurosci. 8, 839–841
4
Trends in Neurosciences, Month Year, Vol. xx, No. yy