What are the cellular mechanisms that dictate and coordinate the interactions between synapses allowing information to be associated, segregated, and dismissed as appropriate for a given experience? Consider a neuron with synapses being activated by different learning-associated information streams. During initial experience such activation elicits synaptic plasticity at a subset of the synapses, a phenomenon thought to be the cellular substrate of memory 
. While learning prompts the storage of different bits of information in multiple synapses of the cell by means of synaptic plasticity, the interactions between these plastic synapses, a process that we call “synaptic plasticity interactions
”, are hypothesized to be key elements for the association and segregation of synaptic plasticity-encoded information induced in different synapses of the same cell 
. We speculate that induction of synaptic plasticity can form particular domains of plasticity-associated metabolic activity within dendrites; a process that we called “functional compartmentalization”. Thus, induction of synaptic plasticity may form functional groups of plastic synapses within morphologically defined dendritic compartments. And it is within and between these “functional compartments” that the proper association and segregation of learning-induced plastic changes takes place.
Neurons in the hippocampus receive inputs from different brain areas that need to be integrated for proper encoding of information 
. The encoding properties of a hippocampal neuron may conceivably comprise the integration of multiple forms of synaptic plasticity elicited at separate synapses of the same neuron 
In this study we investigated the interaction between LTP and LTD elicited in separated synaptic inputs to a CA1 pyramidal neuron of the mouse hippocampus and found that it is temporally and spatially regulated. In particular, there appear to be functionally separate dendritic compartments corresponding to the anatomical domains we used for inducing LTP and LTD. Our findings further highlight the integrative capability of CA1 neurons of the mouse hippocampus, particularly, in regard to the induction and expression of opposing forms of synaptic plasticity within the same cell 
. Our main findings are summarized as follows: 1) that intracompartmental interactions are stronger in magnitude than transcompartmental ones, 2) that the magnitude of the interaction depends on the time separation between LTP and LTD inductions, 3) that during intracompartmental interference between LTP and LTD, only the subsequent form of synaptic plasticity is affected, 4) that cooperation and interference between LTP and LTD can not occur at the same time intervals, and 5) that the intracompartmental interference between LTP and LTD depends on new protein synthesis.
Intracompartmental protein synthesis dependent LTP/LTD interference
Our finding that the interference between LTP and LTD depends on the synthesis of new proteins suggests that proteins generated in response to either strong LTP or strong LTD –inducing stimuli might restrict the expression of an opposite form of synaptic plasticity. We suggest that cooperative interactions could only take place when this interfering activity is reduced. Thus, in addition to previously described mechanisms contributing to the interference between LTP and LTD 
, our study shows de novo
synthesis of protein as a novel mechanism for this interference. The requirement of protein synthesis would represent a mechanistic switch that implies a higher activity threshold for a long lasting interference whose functional compartmentalization could rely on mechanisms for homeostatic regulation 
. The absence of dependency on transcription blockage for the LTP/LTD interference supports a specific role of protein synthesis on this phenomenon.
What could be the signaling pathways underlying the protein synthesis dependent interference? Two possible candidate pathways emerge, the protein kinase A (PKA) pathway for LTP and the metabotropic glutamate receptor (mGluR) pathway for LTD. Each of these signaling pathways is known to elicit the protein synthesis dependent phase of the strong forms of LTP and LTD used in this study, respectively 
If synthesis of protein factors is required for the LTP/LTD interference, where does this protein synthesis occur? The location could be somatic, synaptic or both. Somatic protein synthesis would allow the seeding of protein along the somatodendritic axis. Somatic protein synthesis could follow nuclear activation by the signal generated at the stimulated synapses in response to strong synaptic plasticity induction that travels back to the nucleus and triggers gene transcription 
. This scenario would require target-directed trafficking of new proteins towards the active (functional) dendritic compartment 
(conjectured in 
). Synaptic protein synthesis 
appears less likely because it would require synaptically produced proteins to journey away from the active synapse. Furthermore, both possibilities would require a large amount of protein to be seeded nearby potentially active synapses. A third alternative rests on the notion that the interactive properties between synapses expressing synaptic plasticity rely on dendritic protein synthesis 
(conjectured in 
). In this scenario, activation of protein synthesis via plastic mechanisms could stretch to a single branch or a larger portion of a dendritic tree depending on the strength of the propagating signal originated in the active synapse(s). Hence, synaptic activation in a given dendritic region would set off the synthesis of specific protein factor(s) that would favor the expression of a particular form of synaptic plasticity within that region (functional compartment) 
Transcompartmental LTP/LTD interference
Our data suggest that signals from each form of synaptic plasticity could meet and interfere with each other when generated at separate dendritic compartments, but such interference does not appear to require the synthesis of mRNA or protein. Evidence of pre-transcriptional mechanisms regulating gene expression at the level of activation of transcriptional activators and repressors could probably underlie transcompartmental LTP/LTD interference 
Additionally, a mechanistic dissociation between intra and transcompartmental interactions might rely on changes in cellular excitability originated by strong synaptic activation. Changes in excitability can affect plastic mechanisms 
, and synaptically-driven dendritic depolarization can generate somatic spiking and spike back-propagation that can invade the opposing dendritic compartment 
potentially modifying the expression properties of synaptic plasticity in that compartment. It is yet to be examined, however, whether such phenomenon occurs during LTP/LTD interference or whether synaptically-driven dendritic depolarization is restricted to just one dendritic compartment.
Can this compartmental restriction be overridden? It seems plausible to think that using a much stronger induction protocol for the induction of the prior (priming) form of synaptic plasticity would facilitate transcompartmental interaction. However, we have demonstrated that increasing the strength of the priming stimulation did not facilitate the cooperative interaction between a weak and a strong form of LTP across separate dendritic compartments 
Simultaneous induction of LTP and LTD
When both forms of synaptic plasticity are induced simultaneously, LTD overpowers LTP within the same dendritic compartment, while LTP overpowers LTD across dendritic compartments. The precise time-interval dependent behavior of the LTP and LTD interactions breaks downs when both forms of plasticity are induced simultaneously. The nature of induction can regulate the outcome of the interaction between LTP and LTD 
, however, the nature of synaptic plasticity appears to be independent of LTP or LTD dominance during the interference between LTP and LTD. We are uncertain of the cause of these phenomena. If for instance, mechanisms for LTD induction “kick in” before LTP induction mechanisms, even if both inducing stimuli are delivered simultaneously, we would have seen always LTD overpowering LTP, irrespective of the location of the inputs. This is not the case in our study. To the best of our knowledge, we are unaware of any mechanism that could shed light into why one form of plasticity overpowers the other differently depending on the intra or transcompartmental location of the stimulated synaptic paths.
Correlative behavior between interfering and cooperative LTP/LTD interactions
Our finding that the interactions between LTP and LTD within the same dendritic compartment are both spatially and temporally restricted suggests a common pathway between these two processes. Our study identifies a temporal restriction for the expression of interfering and cooperative interactions between LTP and LTD ( and ). Interference within the same dendritic compartment is observed with time intervals between inductions of 0, 15 and 45 min, but not with a 90 min interval (0 min and 15 min vs. 45 min in the case of transcompartmental interference). Remarkably, the cooperative interaction can be observed (only intracompartmentally) at the time interval which interference is no longer observed.
Correlated behavior of the interfering and cooperative interactions between LTP and LTD.
We propose a model () based on the compartment-specific capture of L-LTP 
and on current notions for the integration of distinct forms of synaptic plasticity 
. Activity dependent mechanisms elicited by either strong LTP or LTD induction would prime a given compartment by modifying the availability of specific LTP or LTD factors (LTP or LTD mRNA and protein) to neighboring synapses. The synthesis of these protein factors would prevent synapses localized within the same functional dendritic compartment from consolidating an opposite form of synaptic plasticity. However, once the activity of these putative factors has subsided, cooperative interactions could be allowed by the productive use of all-purpose plasticity proteins, which are common for the expression of any form of synaptic plasticity.
What is the size of a functional compartment? To gain insight into an estimation of the size of compartmentalization, we can address three issues, 1) the location of the active inputs, 2) the morphological distinction between each dendritic tree, and 3) the functional distinction between each dendritic tree. We know that the stimulated synaptic paths are input specific as paired-pulse facilitation (PPF) analysis shows no heterosynaptic cross-activation (see Methods, also demonstrated in 
). This could ensure that at the synaptic level, we are activating two separate set of synapses. Each set of synapses belongs to either the same or different morphologically defined dendritic trees. In Alarcon et al. (2006) we demonstrated that synaptic tagging was restricted to each dendritic tree, suggesting a functional compartmentalization of the tag mechanism. Another independent study corroborated these findings and further characterized the molecular mechanisms of the tag in basilar and apical dendrites 
. Altogether, we suggest that the functional size of the compartment would depend on the strength of the stimuli used for synaptic activation. Importantly, the key or operational word here is “functional”. Weak stimuli would generate a smaller functional compartment, while stronger stimuli a bigger one; having a single spine and the whole cell the lower and upper limit, respectively. Considering only the cable properties of a dendritic tree, the size of a functional compartment would not be defined by a metric function, but an activity one. The size of a functional compartment would be defined by a Gaussian-like activity function. It follows, therefore, that there would be no definitive boundaries for a given functional compartment, as plasticity-associated activity would peak at the site of the stimulated synapses and exponentially decay towards each side of the dendritic tree. This seemingly tidy depiction of a functional compartment changes if one considers mechanical barriers (e.g. soma, organelles) that would constrain the biochemical diffusion or transport of plasticity factors generated in response to synaptic activation. Then, how big would a functional compartment be in our studies? Here we argue that production of new protein is relevant for the intracompartmental interaction between LTP and LTD, and elaborate that the source of these proteins may be somatic or dendritic (nearby the site of synaptic activation). On the other hand, transcompartmental interaction is weaker than the intracompartmental one but not absent and appears independent of the synthesis of new protein. We think that the size of basilar and apical functional compartments extends and overlaps one to another at the level of electrotonic properties given that strong stimulus protocols as the ones used in this study would depolarize an entire dendritic tree and possibly invade the opposite tree 
. But at the level of biochemical signals, namely protein factors, the size narrows down to match a morphologically defined dendritic tree. The size of the functional compartment might even narrow down to a dendritic domain (a fraction of the dendritic tree) if one considers the source of protein factors after synaptic activation to be only local (dendritic translation).
Synaptic plasticity interactions and the encoding of information
Hippocampal neurons receive a large number of synaptic inputs potentially capable of inducing long-term synaptic plasticity. The cellular mechanisms regulating synaptic plasticity interactions (i.e. the interaction between synaptic inputs expressing synaptic plasticity in a single neuron) seem to be crucial for understanding the cellular basis of encoding multiplex information.
Encoding of information at specific compartments is mainly defined by the anatomy of hippocampal circuits 
. Changes in neural activity that modulate hippocampal oscillations (e.g. theta, gamma) 
are suitable candidates to modulate the induction of synaptic plasticity in these synaptic paths 
. Indeed, changes in hippocampal oscillations occur with learning 
. As induction of synaptic plasticity develops in various temporal fashions in multiple synapses, neurons will utilize synaptic plasticity interaction mechanisms to integrate these plastic events 
. Conceivably, the relationship between changes in input activity, hippocampal oscillations and synaptic plasticity interactions could impact a subset of the neuron population which could specifically encode multiplex information related to a given behavioral experience 
. Neurons within a particular population ensemble could therefore generate particular output spike activity stamps 
that will impact the decoding of information in extra-hippocampal areas in order to produce behaviorally relevant outputs 
What could be the function of synaptic plasticity interactions? Particularly, how could the temporal and the spatial restrictions of the interaction between LTP and LTD lead to the proper encoding of information? Our study indicates that interactions between LTP and LTD in the CA1 area of the mouse hippocampus undergo spatial and temporal regulation. CA1 pyramidal neurons initially disfavor the coexistence of two opposite forms of synaptic plasticity. Dual existence (i.e. cooperative interactions) is only allowed after a given period of time. Interference between two inputs received at short time intervals provides a possible mechanism for disruption of unwanted information. Activity dependent disruption of unwanted information seems to be necessary for the stabilization of a memory trace 
. In time, once a trace has been consolidated, another one can be associated to it.
Spatial and temporal interactions amongst plastic synapses of a neuron might enable the processing of information arriving into its distinct functional compartments from different brain areas, and associate or segregate such information. We propose that this information processing arises from the changes in synaptic weights due to synaptic plasticity interactions like those we have described.