Young DAPK1 knockout mice have altered presynaptic function

Abstract

The death-associated protein kinase 1 (DAPK1) has recently been shown to have a physiological function in long-term depression (LTD) of glutamatergic synapses: acute inhibition of DAPK1 blocked the LTD that is normally seen at the hippocampal CA1 synapse in young mice, and a pharmacogenetic combination approach showed that this specifically required DAPK1-mediated suppression of postsynaptic Ca²⁺/calmodulin-dependent protein kinase II binding to the NMDA-type glutamate receptor (NMDAR) subunit GluN2B during LTD stimuli. Surprisingly, we found here that genetic deletion of DAPK1 (in DAPK1^−/− mice) did not reduce LTD. Paired pulse facilitation experiments indicated a presynaptic compensation mechanism: in contrast to wild-type mice, LTD stimuli in DAPK1^−/− mice decreased presynaptic release probability. Basal synaptic strength was normal in young DAPK1^−/− mice, but basal glutamate release probability was reduced, an effect that normalized with maturation.

NEW & NOTEWORTHY Young death-associated protein kinase 1 (DAPK1) knockout mice have reduced basal glutamate release probability, an effect that normalized with maturation. This provided a compensatory mechanism that may have prevented a reduction of long-term depression in the young DAPK1 knockout mice.

INTRODUCTION

Long-term potentiation (LTP) and depression (LTD) bidirectionally regulate synaptic strength and are thought to underlie learning, memory, and cognition (1–4). Both LTP and LTD prominently require the Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) (Refs. 5–7, for review see Ref. 8), and we have recently shown that the directionality of CaMKII signaling is regulated by another member of the CaM kinase family during these opposing mechanisms of plasticity, the death-associated protein kinase 1 (DAPK1) (9). DAPK1 is highly localized in dendritic spines that contain postsynaptic excitatory synapses (9, 10) and is well known for its roles in inducing neuronal cell death (10–15). During both LTD and ischemia, DAPK1 is activated by calcineurin (CaN), which in turn induces DAPK1-mediated phosphorylation of the GluN2B subunit of the NMDA type glutamate receptor (NMDAR) at S1303 (9, 10). DAPK1 binding to GluN2B or phosphorylation of S1303 of GluN2B is sufficient to inhibit the well-studied CaMKII binding to GluN2B (9, 16–18); this CaMKII/GluN2B binding is necessary for normal LTP but not for LTD (19, 20). Thus DAPK1 makes CaMKII binding to GluN2B LTP specific by inhibiting this binding during LTD. This function of DAPK1 is essential for normal LTD, as acute pharmacological DAPK1 inhibition blocked LTD in wild-type (WT) mice but not in mice that have a mutation in GluN2B that prevents CaMKII binding (and thus no longer require suppression of this binding by DAPK1-signaling) (9). Here we sought to further investigate this essential function of DAPK1 in LTD using DAPK1 knockout (DAPK1^−/−) mice (21).

We hypothesized that the DAPK1^−/− mice would have decreased levels of LTD, similar as observed after acute DAPK1 inhibition (9). Surprisingly, we found instead that DAPK1^−/− mice had enhanced LTD. However, the LTD assessed in young DAPK1^−/− mice showed an unusual presynaptic component that was not seen in WT mice. This indicated altered presynaptic function in the young DAPK1^−/− mice. Indeed, the young DAPK1^−/− mice showed reduced glutamate release probably, an effect that normalized with maturation.

MATERIALS AND METHODS

Mice

WT and DAPK1^−/− mice (21) (on a C57BL/6 background) from heterozygous or homozygous breeder pairs within established colonies were used for electrophysiology and protein expression experiments. Experimenter was not blinded to genotype. All animal procedures were approved by the University of Colorado Institutional Animal Care and Use Committee and carried out in accordance with National Institutes of Health best practices for animal use. All animals were housed in ventilated cages on a 12:12-h h light-dark cycle and were provided ad libitum access to food and water.

Materials

Chemicals were obtained from Sigma unless otherwise noted.

Electrophysiology

Hippocampal dissections and slice electrophysiology were performed as described previously (9, 22). Mixed sex postnatal day (p)13–17 mice, which have robust NMDAR-dependent LTD (23), were used for LTD studies. For studies in mature slices, 8- to 12 wk male mice were used. Dissection solution containing the following (in mM): 220 sucrose, 12 MgSO₄, 10 glucose, 0.2 CaCl₂, 0.5 KCl, 0.65 NaH₂PO₄, 13 NaHCO₃, and 1.8 ascorbate was used for hippocampal dissection and slicing. Slice recovery (minimum of 1.5 h) and recording was conducted in 32°C artificial cerebral spinal fluid containing the following (in mM): 124 NaCl, 2 KCl, 1.3 NaH₂PO₄, 26 NaHCO₃, 10 glucose, 2 CaCl₂, 1 MgSO₄, and 1.8 ascorbate saturated with 95% O₂-5% CO₂. Recordings were made in the CA1 dendritic layer in response to 0.1-ms constant-current bipolar electrode stimulation in the Schaffer collaterals at the CA2 to CA1 interface. After achieving a stable response, an input/output (I/O) curve was generated by increasing the amplitude of stimulus intensity at a constant interval until a maximal response or population spike was noted. Response slope (mV/ms) was plotted against stimulus intensity (mA) over a range encompassing the majority of the I/O data. However, not all slices fit in this range, and not all slices had data points at all stimulus intensities due to actual start values varying between slices and end points being determined by the occurrence of a population spike. Thus the linear slope of the I/O curve for each individual slice was also calculated by linear regression [response slope (mV/ms) by the stimulus intensity (mA)], and the maximum prespike slope (mV/ms) was also noted. These additional analyses are less subject to alterations in electrode resistance and placement and allowed for complete data sets from all slices to be used in the analysis. Stimulus intensity was set to 40% or 70% of the maximum response slope for LTP and LTD experiments, respectively. Paired pulse recordings were done post-I/O from 40% of maximal response before baseline acquisition. LTD was induced with 900 pulses at 1 Hz [low-frequency stimulation (LFS)]. LTP was induced with 1 train of 100 pulses delivered at 100 Hz [ high-frequency stimulation (HFS)]. If necessary, stimulus intensity was reduced during paired pulse experiments taken poststimulation to avoid contamination with population spiking in the second pulse. Responses were recorded and analyzed using WIN LTP software (24).

Subcellular Fractionation

Age- and sex-matched littermate pups (p13–16) were used for fractionation experiments. Euthanasia and hippocampal dissection were conducted as in electrophysiological experiments except hippocampi were flash frozen in liquid nitrogen. Cellular fractionation was conducted as previously described with modification (25). Frozen hippocampi were homogenized with a pestle and drill homogenizer in sucrose buffer (SB) containing the following (in mM): 10 Tris, pH 7.4, 320 sucrose, 1 Na₃VO₄, 2.5 NaF, 1 EDTA, 1 EGTA, cOmplete protease inhibitor cocktail (EDTA free, Roche), and PhosSTOP phosphatase inhibitor cocktail (EDTA free, Roche). A sample was taken for the homogenate fraction. The remaining homogenate was then spun at 1,000 g for 10 min. Supernatant (S1) was decanted to a fresh tube and spun at 10,000 g for 15 min. The pellet (P2) containing the synaptosomal plasma membrane was rinsed with SB and resuspended in 900 µl Triton buffer containing 0.5% Triton X-100 and the following (in mM): 10 Tris, pH 7.4, 1 Na₃VO₄, 2.5 NaF, 1 EDTA, 1 EGTA, cOmplete protease inhibitor cocktail (EDTA free, Roche) and PhosSTOP phosphatase inhibitor cocktail (EDTA free, Roche). The resuspension was subsequently spun at 32,000 g for 20 min. The supernatant (TxS, synaptic fraction) was removed, and subjected to acetone precipitated at 4°C overnight, dried, and resuspended in 60 µl storage buffer containing the following (in mM): 1 EDTA and 10 Tris pH 8 in 1% SDS. The pellet [TxP-postsynaptic density (PSD) fraction] was resuspended in 60 µl storage buffer. Protein content was determined using the BCA method (Bradford) and equal protein was subjected to SDS-PAGE and western analysis as described previously (9, 22, 26, 27). Cellular fractions were subjected to SDS-PAGE on polyacrylamide gels, transferred to PVDF membranes for 1–2 h at 4°C, and blocked in either 5% milk or BSA (corresponding to antibody dilutions) in Tris-buffered saline with Tween-20 containing (TBST) the following (in mM): 20 Tris HCl pH 7.55, 150 NaCl, and 1% Tween-20. Membranes were then blotted for (in 1% milk or BSA in TBST) GluN2B (NeuroMab 75-097, 1:1,000, BSA), pS1303 GluN2B (Millipore, 07-398, 1:1,000, milk), CaMKIIα (BD Biosciences, 611292 1:2,000, milk), pT286 CaMKIIα (Phospho-Solutions p1005-286, 1:2,000, milk), CaMKIIβ (BD Biosciences, 611292, 1:2,000, milk), pT287 CaMKIIβ (Phospho-Solutions p1005-286, 1:2,000, milk), GluA1 (Millipore AB1504, 1:2,000, BSA), or DAPK3 (Cell Signaling 29285, 1:1,000, BSA) followed by HRP-conjugate secondary antibodies (in 1% milk or BSA in TBST, anti-rabbit, GE Healthcare NA934V, 1:4,000–1:6,000, or Bio-Rad 170-6515, 1:3,000–1:6,000, Anti-mouse, GE Healthcare NA931V, 1:5,000–1:10,000). Synaptosome and PSD fractions were incubated in the same antibody dilution and developed and imaged simultaneously. Blots were developed using chemiluminescence [Super Signal West Femto (Thermo-Fisher) or Western Lighting Plus ECL (Perkins Elmer)] and imaged using the ChemiImager 4400 system (Alpha-Innotech).

Neuronal Culture and Imaging

Neurons from early postnatal (p1) DAPK1^−/− mice were cultured on glass coverslips for imaging experiments. Preparation of dissociated cultures was conducted as previously described (28, 29). At 18 days in vitro, cultures were transfected using lipofectamine with mCherry-DAPK1 and green fluorescent protein (GFP). After 1 day of transfection, cultures were fixed and stained for VGluT1 using an anti-mouse Alexa 647 dye. Neurons were imaged using a Axio Observer microscope (Carl Zeiss) fitted with a ×63 Plan-Apo/1.4 numerical aperture objective, using 405-, 488-, 561-, and 638-nm laser excitation and a CSU-XI spinning disk confocal scan head (Yokogawa) coupled to an Evolve 512 EM-CCD camera (Photometrics) and controlled using Slidebook 6.0 software (Intelligent Imaging Innovations [3i]), as described previously (30). Hippocampal neurons were selected based on pyramidal shaped soma and presence of spiny apical dendrites. Images of axons and dendrites of transfected neurons were acquired with a ×63 objective. Images at ×40 images were taken for the whole neuron stitched image.

Regions of axons and dendrites were analyzed using Pearson’s correlation. A threshold-based mask was generated from the GFP channel to mask only the transfected cell. Pearson’s correlation coefficient for VGluT1 versus DAPK1 and VGluT1 versus GFP was then assessed within the mask.

Statistics

All data are expressed as means ± SE. Statistical analysis was conducted in GraphPad Prism 7 (GraphPad Software Inc.). Data were subject to Shapiro-Wilk test of normality and a Brown-Forsythe test (3 or more groups) or an F test (2 groups) to determine equal variance to determine qualification for parametric tests. Specific tests and P values are indicated in the text and figure legends. Alpha was set at 0.05 to determine statistical significance.

RESULTS

Young DAPK1^−/− Mice Have Enhanced Levels of LTD and Normal Levels of LTP

We have previously demonstrated that acutely blocking DAPK1 activity pharmacologically inhibits the induction of hippocampal CA1 LTD by low-frequency stimulation (LFS; 1 Hz for 15 min) and that an essential function of DAPK1 during LTD is inhibiting CaMKIIα accumulation at synapses (9). Thus we expected that DAPK1^−/− mice would have impaired LTD but normal LTP (9). We first tested LFS-induced LTD in the hippocampal CA1 region of young (p13–16) DAPK1^−/− mice. Surprisingly, and in contrast to our previous findings with acute DAPK1 inhibition, DAPK1^−/− mice and their WT littermates both had robust LFS-induced LTD (as a change from baseline in each slice) (DAPK1^−/− = 52.54 ± 3.61%, P < 0.0001, WT = 65.58 ± 3.39%, P < 0.0001, one sample t test from 100%) (Fig. 1A). Instead, LTD was actually enhanced in DAPK1^−/− mice compared with WT mice (P = 0.018, unpaired two-tailed t test) (Fig. 1A).

Figure 1.

Young death-associated protein kinase 1 (DAPK1) mice have enhanced levels of long-term depression (LTD) and normal levels of long-term potentiation (LTP). A: LTD was induced in wild-type (WT) and DAPK1^-/- [knockout (KO)] mice with 900 pulses at 1 Hz (low-frequency stimulation). Left: plot of the normalized field postsynaptic excitatory potential (fEPSP) over time. Middle: plot of individual slices and statistical comparison between groups (average of last 10 min of LTD, “B”) Right: representative traces (average of 3 time points, 1 min) from baseline “A” and LTD “B” for WT and DAPK1^-/- mice. Scale bars = 5 ms (horizontal) and 0.5 mV (vertical). WT: n = 9 slices from 6 mice (4 litters); DAPK1^-/-: n = 11 slices from 7 mice (5 litters). *P < 0.05, unpaired two-tailed t test. B: LTP was induced in WT and DAPK1^-/- mice with 100 pulses at 100 Hz (high-frequency stimulation). Left: plot of the normalized fEPSP over time. Middle: plot of individual slices and statistical comparison between groups (average of last 10 min of LTP, “B”). Right: representative traces (average of 3 time points, 1 min) from baseline “A” and LTD “B” for WT and DAPK1^-/- mice. Scale bars = 5 ms (horizontal) and 0.5 mV (vertical). WT: n = 9 slices from 5 mice (5 litters); DAPK1^-/-: n = 8 slices from 5 mice (5 litters). NS, not significant, unpaired two-tailed t test.

We next tested LTP in the hippocampal CA1 region of the young DAPK1^−/− mice. In mature DAPK1^−/− mice, CA1 LTP has been described to be normal (10). To rule out that appearance of normal LTP is due to a ceiling effect, we used a mild LTP induction protocol with a single train of high-frequency stimulation (HFS; 100 Hz for 1 s). Both WT and DAPK1^−/− mice demonstrated HFS induced LTP (WT = 132.6 ± 7.05%, P = 0.002, DAPK1^−/− = 130.1 ± 6.46%, P < 0.002, one sample t test from 100%) that did not differ between the genotypes (P = 0.801 unpaired two-tailed t test) (Fig. 1B). Thus young DAPK1^−/− mice have a directional impairment in synaptic plasticity with enhanced LTD but normal LTP.

Normal Strength and Composition of Synapses in Young DAPK1^−/− Mice

Looking for an explanation for the difference in results seen with acute inhibition of DAPK1 versus chronic deletion, we reasoned that the prolonged developmental absence of DAPK1 might induce synaptic strengthening by allowing the accumulation of CaMKII not only during LTP but also during LTD. Such an increase in the basal synaptic strength of the slice could thereby occlude blockade of LTD and/or enhance the amount of inducible LTD. To test this hypothesis, hippocampal slices from WT and DAPK1^−/− mice were subjected to increasing stimulus intensity until the CA1 response plateaued or a population spike was noted. A linear regression was calculated for each slice individually by plotting the rise in response against the stimulus intensity (Fig. 2, A and B). Neither the slope of response (mV/ms response/mA input) nor maximum achievable slope (mV/ms) varied between genotypes, indicating similar synaptic strength in young DAPK1^−/− mice as in WT mice of the same age (WT linear slope = 27.36 ± 2.97, DAPK1^−/− = 22.57 ± 1.90, P = 0.166, unpaired two-tailed t test; WT maximum slope = 0.86 ± 0.09, DAPK1^−/− = 0.72 ± 0.03, P = 0.142, unpaired Mann–Whitney U test) (Fig. 2, A–C).

Figure 2.

Normal synaptic strength and composition in young death-associated protein kinase 1 knockout (DAPK1^-/-) mice. A: input output slope (mV/ms) at varied stimulus intensity (mA) over a determined range in slices from young wild-type (WT) and DAPK1^-/- [knockout (KO)] mice. Only the first point differed significantly, *P < 0.05, unpaired two-tailed t test. B: input output slope (m) for all individual slices in young WT and DAPK1^-/- mice. WT: n = 19; DAPK1^-/-: n = 24 (from mice in young long-term potentiation and long-term depression experiments). NS, not significant, unpaired two-tailed t test. C: maximum prepopulation spike response (mV/ms). NS, not significant, unpaired Mann–Whitney U test. D: Western blots of hippocampal cellular fractionation from young WT and DAPK1^-/- mice (2 mice each genotype). PSD, postsynaptic density.

Another possibility for the enhanced LTD noted in the DAPK1^−/− mice could be compensatory differences in postsynaptic composition, such as a compensatory increase in the phosphorylation (p) of S1303 of GluN2B, which could directly prevent CaMKII accumulation at the synapse (17, 18), i.e., even in the absence of DAPK1. To test this and examine the relative concentration of other synaptic proteins, we performed fractionation to isolate the synaptic and PSD fraction from young WT and DAPK1^−/− mouse hippocampi (2 mice each genotype). We found no obvious differences in total GluN2B or pS1303 GluN2B in the synaptic or PSD fraction (Fig. 2D). Additionally, no obvious differences in CaMKIIα or β or their autonomous pT286/287 forms were noted (Fig. 2D). Furthermore, consistent with basal synaptic strength being normal, levels of the AMPA receptor subunit GluA1 were also unchanged in the synaptic fractions of DAPK1^−/− mice (Fig. 2D). Another possibility for the enhanced LTD could be a compensatory increase in the closely related DAPK1 family member DAPK3 at synaptic sites. However, we found no differences in DAPK3 expression, and DAPK3 was virtually undetectable in synaptic and PSD fractions (Fig. 2D). Thus the enhanced LTD in DAPK1^−/− mice cannot be easily explained by changes in basal synaptic strength or postsynaptic composition.

Young DAPK1^−/− Mice Have Reduced Glutamate Release Probability

DAPK1 has been identified in a large-scale kinase screen to be necessary for clathrin-mediated endocytosis (31) and has been shown to phosphorylate syntaxin-1A to disrupt its interaction with munc18 (32). Disrupting this interaction frees syntaxin-1A to participate in formation of the SNARE complex and vesicular docking (32). However, a functional role for DAPK1 in endocytosis, vesicular docking, or vesicular release has not yet been demonstrated. Thus we hypothesized that the DAPK1^−/− mice could have altered presynaptic function, which could explain the noted enhancement of LTD. To test this possibility, we first performed paired pulse experiments on our hippocampal slices, varying the interpulse interval from 10 to 750 ms. At the 25-ms interpulse interval, the amount of paired pulse facilitation in the CA1 became significantly enhanced in the DAPK1^−/− mice compared with WT mice (WT = 1.64 ± 0.06, DAPK1^−/− = 1.88 ± 0.08, P = 0.029, unpaired two-tailed t test) and remained enhanced at 50 (WT = 1.66 ± 0.08, DAPK1^−/− = 1.91 ± 0.05, P = 0.009)-, 75 (WT = 1.57 ± 0.06, DAPK1^−/− = 1.81 ± 0.08, P = 0.037)-, and 100-ms intervals (WT = 1.52 ± 0.06, DAPK1^−/− = 1.70 ± 0.06, P = 0.038) (Fig. 3A). At the 250-ms interval, the paired pulse ratio of the DAPK1^−/− mice returned to WT levels and also did not differ at 500 or 750 ms. Enhanced facilitation at the shorter intervals indicates that the DAPK1^−/− mice have a decreased glutamate release probability.

Figure 3.

Young death-associated protein kinase 1 knockout (DAPK1^-/-) mice have altered glutamate release probability and presynaptic long-term depression (LTD). A: plot of the paired pulse ratio (pulse 2/pulse 1) at varying interstimulus intervals in young wild-type (WT) and DAPK1^-/- [knockout (KO)] mice. WT: n = 9; DAPK1^-/-: n = 11 (from mice in young LTD experiments). *P < 0.05, **P < 0.01, unpaired two-tailed t test of each interval. B: analysis of the response facilitation and decay during the first 50 pulses of the HFS used to induce long-term potentiation (LTP). WT: n = 9; DAPK1^-/-: n = 10. *P < 0.05, **P < 0.01, unpaired two-tailed t test of each pulse. C: paired pulse ratio (pulse 2/pulse 1) before (Base) and 1 h after stimulus (LTD) in young WT and DAPK1^-/- mice. WT: n = 9; DAPK1^-/-: n = 11. NS, not significant; *P < 0.05, paired two-tailed t test. D: paired pulse ratio (pulse 2/pulse 1) before (Base) and 1 h after stimulus (LTP) in young WT and DAPK1^-/- mice. WT: n = 7; DAPK1^-/-: n = 8. NS, not significant, paired two-tailed t test.

To further our understanding of vesicular release and decay, we next analyzed the first 50 pulses of a 100 Hz stimulation as we have done previously (22). Consistent with enhanced paired pulse facilitation at several intervals, the second pulse of the stimulus train (100 Hz = 10-ms interpulse interval) was significantly higher in DAPK1^−/− mice (WT = 1.62 ± 0.06, DAPK1^−/− = 1.94 ± 0.07, P = 0.002, unpaired two-tailed t test) (Fig. 3B). The rest of the rising phase of the HFS did not differ significantly between genotypes (Fig. 3B). However, consistent with a lower initial release probability (or larger back up vesicular pool), the late decay phase of the HFS train was significantly higher in the DAPK1^−/− mice (unpaired two-tailed t test of each time point) (Fig. 3B). This result was somewhat surprising, as despite the altered stimulus created by presynaptic differences, DAPK1^−/− mice had the same LTP level as WT mice (see Fig. 1B). Together, these results indicate that the apparent effects of DAPK1 knockout on the levels of postsynaptic plasticity could be convoluted by abnormal basal presynaptic function or by additional presynaptic changes induced during stimulation.

DAPK1^−/− Mice Have a Presynaptic Component to LFS-Induced LTD That Is Not Seen in WT Mice

With altered presynaptic function established we next sought to test the possibility of a presynaptic component to the expressed LTD/LTP in the DAPK1^−/− mice. To accomplish this, we performed a paired pulse stimulus (50-ms interpulse interval) during the initial baseline acquisition and at 1 h poststimulus. In addition to the increase in basal paired pulse ratio with a 50-ms interval (Fig. 3A), the DAPK1^−/− mice had an additional increase in this ratio after LTD (base = 1.94 ± 0.07, LTD = 2.08 ± 0.10, P = 0.02, paired two-tailed t test) (Fig. 3C). This additional increase was not noted in WT mice (base = 1.74 ± 0.06, LTD = 1.67 ± 0.08, P = 0.285, paired two-tailed t test) (Fig. 3C). This result is consistent with an additional decrease in glutamate release probability after LTD in the DAPK1^−/− mice and strongly indicates a presynaptic component to LTD in the DAPK1^−/− mice.

LTP had no effect on paired pulse ratio in either WT or DAPK1^−/− mice (WT base = 1.69 ± 0.09, LTD = 1.63 ± 0.07, P = 0.175; DAPK1^−/− base = 1.90 ± 0.06, LTD = 1.87 ± 0.06, P = 0.627, paired two-tailed t test) (Fig. 3D). Therefore, these experiments indicate that the young DAPK1^−/− mice have a presynaptic component to hippocampal CA1 LTD, while LTP expression remains postsynaptic.

Glutamate Release Probability Is Normalized in Mature DAPK1^−/− Mice

Previous work has indicated normal synaptic transmission and levels of LTP in older DAPK1^−/− mice (10). Thus we decided to examine if the reduced glutamate release probability seen in here in young DAPK^−/− mice is normalized with maturation (at 8–12 wk of age). Basal synaptic strength was determined to be the same in mature DAPK1^−/− mice as in WT mice of the same age, as neither the slope of response (mV/ms response/mA input), or maximum achievable slope (mV/ms) varied between genotypes (WT linear slope = 39.04 ± 9.02, DAPK1^−/− = 33.48 ± 4.87, P = 0.583, unpaired two-tailed t test; WT maximum slope = 0.82 ± 0.08, DAPK1^−/− = 0.88 ± 0.06, P = 0.593, unpaired two-tailed t test) (Fig. 4, A–C). Thus, consistent with previous reports (10), we found that basal synaptic strength had normalized in the mature DAPK1^−/− mice.

Figure 4.

Mature death-associated protein kinase 1 knockout (DAPK1^-/-) mice have normal presynaptic release probability A: input output slope (mV/ms) at varied stimulus intensity (mA) over a determined range in slices from mature wild-type (WT) and DAPK1^-/- [knockout (KO)] mice. B: input output slope (m) for all individual slices in mature WT and DAPK1^-/- mice. WT: n = 8 slices from 5 mice (3 litters); DAPK1^-/-: n = 9 slices from 5 mice (3 litters). NS, not significant, unpaired two-tailed t test. C: maximum prepopulation spike response (mV/ms). NS, not significant, unpaired two-tailed t test. D: paired pulse ratio (pulse 2/pulse 1) at varying interstimulus intervals in mature WT and DAPK1^-/- mice. WT: n = 8 slices from 5 mice (3 litters); DAPK1^-/-: n = 7 slices from 5 mice (3 litters). No significant differences, unpaired two-tailed t test of each interval.

We next performed paired pulse experiments, varying the interpulse interval from 10 to 750 ms. In contrast to observations in the younger mice, paired pulse facilitation in the mature DAPK1^−/− mice did not differ from their WT counterparts at any interpulse interval (unpaired two-tailed t test of each interval) (Fig. 4D). Thus the reduced glutamate release probability seen in the young DAPK1^−/− mice also normalized with maturation.

Indication for Presynaptic Localization of DAPK1

Postsynaptic DAPK1 enrichment in dendritic spines has been established previously (9, 10, 16). Here, we decided to explore potential additional presynaptic DAPK1 localization, because our current study suggests additional presynaptic functions of DAPK1 and because a previous study indicated that DAPK1 can phosphorylate the presynaptic substrate protein syntaxin-1A (32). Biochemical fractionation is not suitable, as synaptosomal fractions (such as in Fig. 2D) contain both pre- and postsynaptic components. Similarly, immunocytochemical localization to presynaptic boutons versus postsynaptic spines is also difficult, due to the limited spatial resolution. Thus we decided to instead determine the localization of mCherry-DAPK1 expressed in hippocampal neurons from DAPK1^−/− mice. GFP was coexpressed as a cell fill, to make it easier to follow the axons and dendrites; the vesicular glutamate transporter 1 (VGluT1) was immunostained as marker for presynaptic boutons (Fig. 5, A–C). The mCherry-DAPK1 was detected both in axons (Fig. 5, A and B) and in dendrites (Fig. 5, A and C), in both cases with apparent enrichment in the synaptic locations that were marked by the VGluT1 staining (Fig. 5, B and C). Notably, the level of co-localization with the presynaptic marker VGluT1 was indistinguishable between the presynaptic DAPK1 in the axonal compartments and the postsynaptic DAPK1 in the dendritic compartment (Fig. 5D). This highlights the necessity of using mCherry-DAPK1, as it indicates that costaining of endogenous DAPK1 and a presynaptic marker could not differentiate between pre- versus postsynaptic localization. The mCherry-DAPK1 appeared to be enriched presynaptically in boutons and postsynaptically in spines (see Fig. 5, B–D). However, on the presynaptic side, a similar apparent enrichment was seen also for the GFP cell fill (Fig. 5, B and D) and is therefore likely mainly driven by the relatively large bouton volume compared with the narrow caliber of axons. Consistent with the alterations of presynaptic functions in the DAPK1^−/− mice, these results indicate that DAPK1 can indeed localize to presynaptic compartments, even if it is not enriched in boutons over axons.

Figure 5.

Indication for both pre- and postsynaptic localization of death-associated protein kinase 1 (DAPK1). A: green fluorescent protein (GFP) cell fill (green) outlines a hippocampal neuron cultured from DAPK1^-/- mice. The neuron coexpressed mCherry-DAPK1 (red) and was immunostained for the presynaptic marker protein VGluT1 (blue). Boxed axon and dendrite areas are shown magnified in B and C. B: a presynaptic axonal region of the neuron shown in A, indicating localization of mCherry-DAPK1 in presynaptic compartments including axons and boutons. C: a postsynaptic dendritic region of the neuron shown in A. The micrograph indicates that resolution limitation causes significant apparent overlap of the presynaptic marker VGluT1 with the postsynaptically localized DAPK1. D: quantification of colocalization of VGluT1 with presynaptic axonal DAPK1 and GFP cell fill as compared with postsynaptic dendritic DAPK1 and GFP cell fill. Similar VGluT1 colocalization with both presynaptic and postsynaptic DAPK1 illustrates the limited resolution. Similar VGluT1 colocalization of both presynaptic DAPK1 and GFP cell fill illustrates that DAPK1 is likely not enriched in presynaptic boutons over axons. Axon segments n = 9 from 5 neurons; dendrite segments n = 11 from 5 neurons. NS, not significant. *P < 0.05, two-way ANOVA with Bonferroni post hoc test.

DISCUSSION

While acute DAPK1 inhibition blocks the expression of NMDAR-dependent LTD (9), we found here that DAPK1^−/− mice (in which DAPK1 has been chronically deleted) instead had enhanced LTD. This enhancement was not due to increased basal postsynaptic strength that could be explained by our mechanism for the effects of acute DAPK1 inhibition on LTD (allowing increased CaMKII synaptic localization; Ref. 9) or by obvious changes in synaptic composition related to this mechanism. However, we found that the DAPK1^−/− mice had a presynaptic LTD mechanism that was not seen in WT mice: the young DAPK1^−/− mice had increased paired pulse facilitation (indicating reduced glutamate release probability) that was even further increased after the LTD stimulation (indicating even further decreased glutamate release probability). By contrast, in WT mice, paired pulse facilitation was unaffected by the LTD stimuli, demonstrating that the presynaptic manifestation of LTD is specific to the DAPK1^−/− mice. Notably, LTP in the young DAPK1^−/− mice appeared normal, despite of the basally reduced release probability, and despite the resulting apparent alterations in the normalized amplitude of the stimulus train response. This result further highlights that synaptic communication may be altered in the DAPK1^−/− mice, even if the end result (amount of LTP) remains the same between genotypes. In older DAPK1^−/− mice, normal LTP has been reported previously (10), but at this age, we found that release probability had normalized.

After decades of debate, it is commonly accepted that LTP and LTD are manifested largely at the postsynapse (Refs. 33–35, but see Ref. 36 for lasting argument). Paired pulse facilitation is a widely accepted means to determine pre- versus postsynaptic locus of expression in plasticity (36–38). In paired pulse experiments, two pulses separated by a short (10 s to 100 s of ms) interval are delivered. Residual Ca²⁺ in the presynaptic terminal from the first pulse is thought to be additive with Ca²⁺ influx from the second pulse increasing release probability of the second pulse, thereby creating facilitation (39, 40). Paired pulse facilitation remains unchanged in WT mice after long-term plasticity stimuli in the hippocampal CA1 region (22, 37, 41), and this was also observed in our current study. In contrast to short-term facilitation/depression, which is characterized by short lived alterations to active sites, vesicle pools, and Ca²⁺ entry and buffering after stimulus trains and which normalizes over a few minutes (42), DAPK1^−/− mice had changes in paired pulse facilitation that were still apparent 1 h after LTD stimulation. Thus our conclusion is that this facilitation is attributed to a form of long-term plasticity that occurs at the presynapse in the DAPK1^−/− mice.

Our results demonstrate that the presynaptic physiology of young DAPK1^−/− mice is abnormal, indicating that DAPK1 may play an additional role in presynaptic development and/or regulation of transmitter release. In addition to the known roles of DAPK1 in clathrin-mediated endocytosis (31, 43), one potential previously identified mechanism for DAPK1^−/− to alter presynaptic (and thereby overall synaptic) function would be through alterations in SNARE machinery formation through DAPK1-mediated phosphorylation of syntaxin 1 A at S188 (32). Syntaxin 1 is a necessary component of the SNARE machinery, with syntaxin 1 deletion affecting multiple modes of vesicular release and titrated rescue resulting in a dose-dependent increase in readily releasable vesicles and increased release probability (44). The phosphorylation of S188 of syntaxin 1 A alters its association with Munc 18-1 (32), possibly by moving it from an open to a closed conformation, an event necessary to allow syntaxin 1 A binding with its additional partners in the SNARE complex and full vesicular docking (45, 46). However, as paired pulse facilitation changes were not noted with acute pharmacological DAPK1 inhibition (9), our results may instead indicate an as of yet undefined role for DAPK1 at the presynapse or may point to compensatory homeostatic mechanisms related to DAPK1 loss at the postsynapse.

Together, our results here demonstrate profoundly altered synaptic function in young DAPK1^−/− mice. This is manifested in altered glutamate release probability and a nonphysiological presynaptic manifestation of LTD. Thus the results here have implications for the interpretation of experiments examining DAPK1 function in young mice through genetic knockout, as the underlying synaptic physiology is likely to be altered in ways that go beyond acute DAPK1 function. Notably, however, the altered glutamate release probability in the young DAPK1^−/− mice appeared to normalize with maturation.

GRANTS

The research was funded by National Institute of Neurological Disorders and Stroke Grants F31-NS-092265 (to D.J.G.) and R01-NS-081248, R01-NS-110383, and R01-NS-118786 (to K.U.B.).

DISCLOSURES

K.U.B. is a cofounder and board member of Neurexis Therapeutics. The other authors declare that they have no competing interests.

AUTHOR CONTRIBUTIONS

D.J.G. and K.U.B. conceived and designed research; D.J.G. and J.E.T. performed experiments; D.J.G. analyzed data; D.J.G. and K.U.B. interpreted results of experiments; D.J.G. prepared figures; D.J.G. drafted manuscript; J.E.T. and K.U.B. edited and revised manuscript; D.J.G., J.E.T., and K.U.B. approved final version of manuscript.