Several lines of evidence suggest that integrin receptors play a pivotal role in consolidation of long-term potentiation (LTP), but which of the many integrin dimers are involved remains to be discovered. The present study used an LTP reversal paradigm to test if alpha3 integrins make an important contribution. Function blocking alpha3 monoclonal antibodies or vehicle were locally infused into recording sites in field CA1 of rat hippocampal slices and LTP induced with theta burst stimulation. Low frequency trains of pulses were applied 30 min after the theta bursts. Previous work indicates that low frequency stimulation reverses LTP when applied immediately after induction but is largely ineffective after 30-45-min delays. If the antibodies were to block consolidation, then they should extend the period over which potentiation is vulnerable to disruption. There was no detectable difference between the two groups in the initial degree of LTP or within slice decay of potentiation 1-10 min after induction; a small but reliable decay occurred from 10 to 30 min with antibody treatment (P<0.01) but not in control slices. Percent potentiation was not statistically different for vehicle (55 +/- 19%, mean +/- S.D.) and anti-alpha3 (43 +/- 21%) slices at 30 min post-theta bursts. Five-Hz stimulation ("theta pulse" stimulation) 30 min after induction caused a reduction of LTP. The percent loss of potentiation after the 1-min trains was greater in the antibody-treated slices than in controls (98 +/- 4% vs. 62 +/- 28%, P<0.01, U-test) and correlated (r=0.84, alpha3 slices) with the percent LTP present prior to low frequency stimulation, as expected if the stimulation reversed potentiation. Recovery occurred in both groups but percent LTP was significantly smaller in experimental slices at 10 min post-theta pulses (5 +/- 11% vs. 36 +/- 15%, P<0.01). Recovery continued for 20 min after theta pulses and, in accordance with earlier work, was nearly complete for the control slices (50 +/- 19% vs 55 +/- 15%, 40 min post- vs. immediately pre-theta pulses). LTP remained depressed after 40 min of recovery in the anti-alpha3 slices (23 +/- 19% vs. 43 +/- 21%) at which point it was substantially less than that found in controls (P<0.01). Western blots with anti-alpha3 antibodies identified a polypeptide with the molecular mass (155 kDa) expected for the alpha3 subunit and further showed that it is broadly distributed in brain. Subcellular fractionation experiments demonstrated that alpha3 is concentrated in synaptic membranes over homogenates to about the same degree as the GluR1 subunit of the alpha-amino-3-hydroxy-5-methyl-4-isoxazole propionate-type glutamate receptor. From these results we suggest that alpha3-containing integrins are localized to synapses and are needed to stabilize a slowly decaying form of LTP. The findings also show that vulnerability to reversal can be used in place of extended recording sessions in studying consolidation.