A PARP1-ERK2 synergism is required for the induction of LTP

Unexpectedly, a post-translational modification of DNA-binding proteins, initiating the cell response to single-strand DNA damage, was also required for long-term memory acquisition in a variety of learning paradigms. Our findings disclose a molecular mechanism based on PARP1-Erk synergism, which may underlie this phenomenon. A stimulation induced PARP1 binding to phosphorylated Erk2 in the chromatin of cerebral neurons caused Erk-induced PARP1 activation, rendering transcription factors and promoters of immediate early genes (IEG) accessible to PARP1-bound phosphorylated Erk2. Thus, Erk-induced PARP1 activation mediated IEG expression implicated in long-term memory. PARP1 inhibition, silencing, or genetic deletion abrogated stimulation-induced Erk-recruitment to IEG promoters, gene expression and LTP generation in hippocampal CA3-CA1-connections. Moreover, a predominant binding of PARP1 to single-strand DNA breaks, occluding its Erk binding sites, suppressed IEG expression and prevented the generation of LTP. These findings outline a PARP1-dependent mechanism required for LTP generation, which may be implicated in long-term memory acquisition and in its deterioration in senescence.


Supplemental Results
To examine possible effect of PARP1 inhibitors on excitatory postsynaptic NMDA current, evoking LTP in hippocampal CA3-CA1 connections 23 , NMDA currents recorded from depolarized cells in the CA1 pyramidal cell layer were measured, before and after application of PJ-34 and ABT-888 at concentrations affecting LTP (Fig 1, e and f). The currents were measured in response to low frequency stimulation, because high frequency typically induces rundown of NMDAR-mediated EPSC (Fig S1).

MEK inhibition prevented LTP induction without impairing already established LTP
Field excitatory postsynaptic potentials (fEPSPs) were recorded from hippocampal slices (n=4, prepared from 2-month-old C57BL/6 male mice (n=3)). High frequency stimulation (100 Hz, 1 sec) was induced using two sets of bipolar electrodes placed on both sides of the slide and equidistant from the recording pipette, such that two independent stimulation-recording channels were used for each slice (Methods). After 20 minutes of baseline recording, stimulation was applied to one of the pathways, which resulted in a stable LTP of a magnitude of 1.52 ± 0.01, recorded for 120 min. In contrast, same stimulation delivered to the second pathway after 50 min perfusion of the MEK inhibitor PD98059 (50 µM, AdipoGen) failed to generate LTP (average values measured for 60 min after stimulation, 1.15 ± 0.01 above baseline; Fig S2). LTP generated 10 min before application of the MEK inhibitor PD98059 was not impaired. A sample illustration of individual records sampled at the indicated time intervals is presented (Top). A train of high-frequency stimulation (100 Hz, 1sec, denoted with arrows) was 3 delivered to each of 2 independent stimulation-recording pathways. The first stimulation delivered to one of the pathways induced a response of long-term potentiation (LTP). Same stimulation delivered to the second pathway, 50 minutes after application of the MEK inhibitor PD98059 (50 µM), failed to produce LTP, but did not impair the already generated LTP.
Three repeats of high frequency stimulation inducing LTP in hippocampal slices (Fig 1) induced long-term synaptic potentiation, indicated by pre-synaptic vesicle recycling in cultured cerebral neurons ( Figure S3).    Two methods for measuring PARP1 activation were compared. We found that PARP1 activation inducing its polyADP-ribosylation can be measured by the shift in its isoelectric point (pI) towards acidic pH. The pI of PARP1 in un-stimulated neurons (about pH 9.5 ; Fig 3) was shifted towards lower pH, due to negatively charged ADP-ribosyl moieties forming polymers on activated PARP1 in the nuclei of NGF treated neurons.  site. This may underlie polyADP-ribosylation of Erk-bound PARP1 9 (Figs 3, 6C, 7d). The NAD binding site is not similarly exposed by the same calculated movement in DNA-bound PARP1.

Measurement of synaptic long-term potentiation based on FM dye staining 1
Synaptic vesicle release at single synapses in primary neuronal cultures was determined using the activity-dependent FM1-43 dye as described before 1  homodimer: PatchDock 2 , followed by FireDock 3 for docking refinement, and pydock 4 . The electrostatic potentials of PARP1 (aa531-1014) and ERK2 monomer were calculated using APBS and projected on their molecular surface using UCSF Chimera 6 . Positively charged patches that are predicted to bind Erk2 on PARP1 (residues aa633-637 and aa747-752) were selected for in-silico molecular docking (by a computational method that predicts the preferred orientation of two molecules forming a stable complex).
Calculated intra-molecular dynamics (directions of motion) in PARP1 bound to phosphorylated Erk2. The conformational changes of PARP1 and Erk2 were predicted using the anisotropic network model, a normal mode analysis tool available online: