Interplay of hippocampal long-term potentiation and long-term depression in enabling memory representations

Abstract

Hippocampal long-term potentiation (LTP) and long-term depression (LTD) are Hebbian forms of synaptic plasticity that are widely believed to comprise the physiological correlates of associative learning. They comprise a persistent, input-specific increase or decrease, respectively, in synaptic efficacy that, in rodents, can be followed for days and weeks in vivo. Persistent (>24 h) LTP and LTD exhibit distinct frequency-dependencies and molecular profiles in the hippocampal subfields. Moreover, causal and genetic studies in behaving rodents indicate that both LTP and LTD fulfil specific and complementary roles in the acquisition and retention of spatial memory. LTP is likely to be responsible for the generation of a record of spatial experience, which may serve as an associative schema that can be re-used to expedite or facilitate subsequent learning. In contrast, LTD may enable modification and dynamic updating of this representation, such that detailed spatial content information is included and the schema is rendered unique and distinguishable from other similar representations. Together, LTP and LTD engage in a dynamic interplay that supports the generation of complex associative memories that are resistant to generalization. This article is part of a discussion meeting issue 'Long-term potentiation: 50 years on'.

Keywords: LTD; LTP; hippocampus; learning and memory; rodent cognition; synaptic plasticity.

Figure 1.

LTP profiles in the cornu ammonis (CA) and dentate gyrus (DG) subfields of the hippocampus of freely behaving young adult rats and mice. (a) LTP evoked with high-frequency stimulation (HFS, 100 Hz, four trains of 100 pulses each, 5 min intertrain interval) results in different LTP profiles in associational/commissural (AC)–CA3, mossy fibre (MF)–CA3 and Schaffer collateral (SC)–CA1 synapses of the cornu ammonis of dorsal rat hippocampus. Vertical scale bar: 2 mV, horizontal scale bar: 10 ms. (b) Persistent (>24h) LTP can be evoked with a single train of stimuli delivered at 100 Hz in the intermediate DG (closed triangles), but not the dorsal rat DG (open squares). Vertical scale bar: 5 mV, horizontal scale bar: 5 ms. (c) Persistent LTP in medial perforant path (MPP) inputs to the dorsal rat DG can be induced with 10 trains of stimuli at 100 Hz and 10 s intertrain intervals. The same stimulation pattern fails to induce LTP at lateral perforant path (LPP)-DG inputs. Vertical scale bar: 4 mV, horizontal scale bar: 3 ms. (d) HFS (four trains of 50 pulses each at 100 Hz, 5 min intertrain interval) results in LTP of different magnitudes in CaOlaHsd (open circles), CBA/J (grey triangles) and C57BL/6 (closed circles) mice in SC–CA1 synapses in vivo. To note: C57BL/6 mice develop progressive deafness beginning in the 4th postnatal week; CBA/J mice become blind within 4 weeks after birth; CaOlaHsd mice have no appreciable sensory deficits in their first 12 months of life (see citations below). Vertical scale bar: 2 mV, horizontal scale bar: 10 ms. Black arrows depict the time-points of stimulation. Insets show the analogue examples of field excitatory postsynaptic potentials (fEPSPs) recorded at the time points indicated by the numbers in the graphs. Line breaks indicate change in time scale. Figures modified from [,–29].

Figure 2.

LTD profiles in the cornu ammonis (CA) and dentate gyrus (DG) subfields of the rat hippocampus. (a) LTD evoked with low-frequency stimulation (LFS, 900 pulses at 1 Hz) results in different LTD profiles at AC–CA3, MF–CA3 and SC–CA1 synapses in the hippocampus in freely behaving rats. Vertical scale bar: 2 mV, horizontal scale bar: 10 ms. (b,c) Result of the same stimulation protocol (900 pulses at 1 Hz) at lateral perforant path (LPP, closed black circles) and medial perforant path (MPP, open circles)–DG synapses. Vertical scale bar: 5 mV, horizontal scale bar: 5 ms. Black arrow depicts the time-point of stimulation. Insets show analogue examples of fEPSPs recorded at the time points indicated by the numbers in the graphs. Line breaks indicate changes in time scale. Figures modified from: (a) [69,70]. (b,c) [39,71].

Figure 3.

Functional differentiation of the facilitation of LTD by spatial content learning. (a) Novel exploration by rats of objects placed in holeboard (HB) holes, during weak low-frequency stimulation (wLFS, 1 Hz, 900 pulses) results in LTD (>24 h) in SC–CA1 synapses. Afferent stimulation with wLFS only results in STD that lasts for ca 30 min. Arrows in (a–c) indicate when wLFS was applied (in the presence or absence of item–place exploration). (b) Re-exposure to the same objects in the same holeboard positions in conjunction with wLFS fails to result in LTD. However, exposure of the animals to a spatial re-configuration of the same objects results in LTD (>24 h). (c) Stimulation of PP inputs to the DG with wLFS results in STD that persists for <60 min. Novel exposure of rats to objects placed in holeboard holes in conjunction with wLFS fails to induce LTD. (d,e) Analogue examples of evoked potentials recorded from the DG (d) or the CA1 region (e) prior to wLFS (i), 5 min post-wLFS (ii) and 24 h after wLFS (iii) in an animal that received wLFS only (top row) and in an animal that received wLFS during the exploration of objects in the holeboard holes (bottom row). (f) In SC–CA1 synapses, LTD is expressed when wLFS is applied during novel exploration of novel objects placed in holeboard holes (i) [68], or novel objects concealed under sand inside holeboard holes [11] (ii). LTD is also expressed when wLFS is applied in conjunction with the novel exploration of spatially discriminable auditory frequencies that emanate from loudspeakers placed under holes in the floor (iii) [124] , or spatially distributed odours that diffuse through holes in the floor (iv) [125]. The DG does not respond with a change in synaptic strength to any of these conditions: rather, it expresses LTD when exploration of novel constellations of large landmark features of the environment occurs in conjunction with wLFS (v) [1]. (g) Summary of responses of different hippocampal synapses to the abovementioned conditions. PP–DG synapses and MF–CA3 synapses express LTD following exploration of novel configurations of landmark items in space. The AC–CA3 and SC–CA1 synapses do not respond to this kind of information. In contrast, LTD is expressed in these synapses following exploration of novel constellations of items concealed in holeboard holes. PP–DG and MF–CA3 synapses do not respond to subtle item–place information. n.k.: not known. Panels (a–f) are modified from [11].

Figure 4.

In this concept, a rat is introduced to a completely novel spatial environment where high-contrast items (tree, park bench and letterbox) can be perceived from their initial positions without the need for visual acuity. Movement in the environment allows the animal to acquire metric information about its position relative to these landmark features, as well as to locate salient local features that can only be discovered when the animal is close to them (such as crates that could be used for shelter, food remnants that are under the park bench or water that is beside the letterbox). (a) Initial exposure to a novel spatial environment results in immediate induction of LTP in an input-specific manner through hippocampal subfields (red neurons within the top hippocampus schema). By this means a neuronal ensemble is selected in a distributed hippocampal network that serves to retain a schematic representation of the spatial environment. (i–iv) Input-specific LTP (>24 h) is induced in PP–DG (i), MF–CA3 (ii), SC–CA1 (iii) and AC–CA3 (iv) synapses when exposure to entirely novel space is coupled with weak afferent stimulation (weak high-frequency stimulation, wHFS). (b) If time is spent moving within the environment, allocentric representations are cumulatively created that permit the acquisition of dimensional, orientational and directional (e.g. landmark) information that allows the integration of allocentrically relevant details into the initially acquired spatial schema. (v-vi) This information is encoded by means of LTD (>24h) in PP–DG (v) and MF–CA3 (vi) synapses (blue neurons within the middle hippocampus schema). (c) Salient local details of the environment (i.e. information that can only be found by means of proximal exploration and which is integrated into the allocentric reference frame) are acquired by means of LTD (> 24h) that is expressed in SC–CA1 (vii) and AC–CA3 synapses (viii) (blue neurons within bottom hippocampus schema). By this means the spatial representation is modified and refined, such that it can be discriminated from other similar representations. Graphs are modified from [1,11,24].