PSD-95 KO and PSD-93/95 DKO mice display improved posthypoxic recovery

More obvious KO effects were seen during synaptic recovery after reoxygenation as reflected by a return of fEPSP amplitudes. When oxygen was resubmitted 1 minute after the onset of HSD, PSD-95 KO as well as PSD-93/95 DKO slices recovered completely within 9 to 10 minutes, whereas fEPSP amplitudes of WT and PSD-93 KO samples stagnated at about 70%

(WT) to 90% (PSD-93 KO) below baseline, reaching a plateau thereafter (Figure 12A). Using two-way ANOVA to test for a genotype effect, I detected significant differences comparing PSD-95 KO and PSD-93/95 DKO to control samples, respectively [F3,63 = 4.204, p = 0.009;

PSD-95 KO vs. WT, p = 0.004; PSD-93/95 DKO vs. WT, p = 0.003; all other p values > 0.05].

In case of reoxygenation after 2 minutes, synaptic recovery of PSD-95 KO and DKO mice was again higher compared to WT, and in addition to PSD-93 KO slices [Figure 12B; F3,61 = 8.458, p < 0.001; PSD-95 KO vs. WT, p = 0.003; PSD-95 KO vs. PSD-93 KO, p = 0.008; PSD-93/95 DKO vs. WT, p < 0.001; PSD-93/95 DKO vs. PSD-93 KO, p < 0.001; all other p values > 0.05].

Full recovery was achieved slightly later, i.e. at minute 10 to 11, compared to earlier reoxygenation. In contrast, again WT and PSD-93 KO mice did not recover completely, this time even less by reaching 60 to 70% of baseline levels.

When oxygen was deprived for 4 minutes upon HSD detection, the acute brain slices were not capable anymore to recover from metabolic stress – at least within the 20 minutes of recovery where recordings have been performed. In fact, synaptic responses of all genotypes remained under 50% and where hardly distinguishably from noise peaks in the raw traces (Figure 12C).

Nonetheless, I obtained significantly stronger recovery for PSD-95 KO and DKO neuronal network [F3,52 = 18.391, p < 0.001; PSD-95 KO vs. WT, p < 0.001; PSD-95 KO vs. PSD-93 KO, p < 0.001; PSD-93/95 DKO vs. WT, p < 0.001; PSD-93/95 DKO vs. PSD-93 KO, p < 0.001; all other p values > 0.05] which likely resulted from the higher noise impact on smaller EPSP amplitudes in these mice as specified before.

The here detected impaired synaptic recovery after the longest hypoxic period was implied before by tissue light reflectance, which remained on very high levels even upon reoxygenation after 4 minutes (Figure 8). Accordingly, the recovery of both, synaptic function and physical properties of the brain tissue, was impaired after the extensive hypoxic period.

Highlighted throughout Figure 12A–C is a marked and abrupt peak in synaptic responses shortly after oxygen resupply in DKO and to a smaller extent also in PSD-95 KO mice (indicated by arrowheads). In addition, PSD-93/95 DKO slices showed posthypoxic potentiation with fEPSP amplitudes transiently exceeding the baseline by 10% for reoxygenation after 1 minute, and 20% when reoxygenated after 2 minutes, followed by return to prehypoxic levels. These two observations were virtually absent in WT and PSD-93 KO

Results

65

mice. Previous studies reported a similar phenomenon, which is referred to as “anoxic LTP”

(Crépel et al., 1993; Gozlan et al., 1994). Anoxic LTP delineates hypoxia-induced selective potentiations of the NMDAR-mediated component of synaptic transmission. This persistent form of potentiation is assumed to partially account for delayed neuronal cell death (Crépel et al., 1993; Gozlan et al., 1994). In contrast, the potentiation in PSD-93/95 DKO mice seen here was transient and characterized by increased fEPSP amplitudes, representing both AMPAR- and, to a lesser extent, NMDAR-mediated transmission. A similar transient potentiation has been reported by Frenguelli (1997) and linked to strong activity of AMPARs and NMDARs during the hypoxic episode. Strong receptor activation is most likely caused by the excess of extracellular glutamate, since its uptake is impaired or even reversed by hypoxia (Szatkowski

& Attwell, 1994). Indeed, the resulting high intracellular Ca²⁺ levels have been shown to provoke transient potentiations of synaptic transmission (Kauer et al., 1988). Furthermore, HSD-accompanied cell swelling might contribute to the observed potentiation as reported for hyptonia-induced reduction in the interstitial space (Ballyk et al., 1991; Chebabo et al., 1995).

Thus, several mechanisms possibly account for transient potentiation in mice lacking both PSD-95 and PSD-93. They all imply a rather unstable neuronal network, in which synaptic activity and cell volume may easily under- and overshoot due to metabolic changes.

Collectively, these results show improved synaptic recovery in PSD-95 KO mice for reoxygenation within less than 4 minutes, demonstrated by faster and more complete reinstatement of normoxic fEPSP amplitudes. This result further strengthens the observed enhanced hypoxia tolerance due to loss of PSD-95, as clearly seen by a delayed onset of HSD (Figure 7) and attenuated change of the HSD-accompanying tissue light reflectance (Figure 8). Additional evidence was given by a trend for decelerated HSD wave propagation (Figure 9) and slightly delayed hypoxia-induced loss of synaptic function in PSD-95 KO mice (Figure 11). Interestingly, all these effects were consistently abolished by additional knockout of the paralog PSD-93, but with the only exception of synaptic recovery. Indeed, mice with a loss of both proteins, PSD-93 and PSD-95, as well showed improved ability for posthypoxic recovery as seen in Figure 11. However, synaptic function in DKO mice seemed to be more unstable and susceptible for changes in oxygen levels as indicated by the abrupt and prominent increase in fEPSP amplitudes shortly after reoxygenation and by transient posthypoxic potentiation during recovery (Figure 12).

66

Figure 12: Brain slices from PSD-95 KO and PSD-93/95 DKO mice demonstrated earlier and increased posthypoxic recovery. (A-C) Displayed are time courses for fEPSP amplitudes normalized to prehypoxic baseline after reoxygenation at t=0. (A, B) PSD-95 KO and DKO slices showed complete synaptic recovery, whereas WT and PSD-93 KO slices recovered markedly slower and did not return to normoxic baseline levels. (C) When oxygen was resubmitted 4 min after HSD onset, acute brain slices failed to recover within the analyzed time period of 20 min.

(A–C) Arrowheads indicate an abrupt amplitude increase detected in DKO and partially in PSD-95 KO mice. Please note the before mentioned influence of noise peaks on the calculation of amplitude values, wherefore zero-amplitudes at t=0 are not quiet reached. (A) WT, n/m = 16/9;

PSD-95 KO, n/m = 19/11; PSD-93 KO, n/m = 16/9; PSD-93/95 DKO, n/m = 16/11. (B) WT, n/m = 15/9; PSD-95 KO, n/m = 17/10; PSD-93 KO, n/m = 18/11; PSD-93/95 DKO, n/m = 15/9. (C) WT, n/m = 13/9; PSD-95 KO, n/m = 13/10; PSD-93 KO, n/m = 15/8; PSD-93/95 DKO, n/m = 15/9. Two-way ANOVA followed by post-hoc LSD multiple comparison tests. ***p < 0.001; **p < 0.01. All data were given as means ± SEM.

Results

67

3.2. Synaptic composition of excitatory cortical neurons