Impaired TrkB receptor signaling underlies corticostriatal dysfunction in Huntington’s disease

Summary

Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder. The debilitating choreic movements that plague HD patients have been attributed to striatal degeneration induced by the loss of cortically supplied brain-derived neurotrophic factor (BDNF). Here we show that in mouse models of early symptomatic HD, BDNF delivery to the striatum and its activation of tyrosine-related kinase B (TrkB) receptors were normal. However, in striatal neurons responsible for movement suppression, TrkB receptors failed to properly engage postsynaptic signaling mechanisms controlling the induction of potentiation at corticostriatal synapses. Plasticity was rescued by inhibiting p75 neurotrophin receptor (p75_NTR) signaling or its downstream target phosphatase-and-tensin-homolog-deleted-on-chromosome-10 (PTEN). Thus, corticostriatal synaptic dysfunction early in HD is attributable to a correctable defect in the response to BDNF, not its delivery.

Introduction

Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder resulting from a CAG expansion in the huntingtin gene (The Huntington’s Disease Collaborative Research Group., 1993). The earliest signs of pathology in HD are in the striatum (Menalled and Chesselet, 2002; Raymond et al., 2011), a subcortical structure involved in the control of movement and action selection (Gerfen and Surmeier, 2011). The striatum exerts this control by transforming excitatory synaptic input from the cerebral cortex into patterned activity in two parallel projection systems: so-called direct and indirect pathways. Activity in direct pathway spiny projection neurons (dSPNs) promotes action, whereas activity in indirect pathway spiny projection neurons (iSPNs) suppresses action (Gerfen and Surmeier, 2011). A deficit in the ability of cortical circuits to drive iSPNs has long been hypothesized to underlie unwanted movement in the early stages of HD (Zuccato et al., 2010). In recent years, this loss has been attributed to impaired expression and release of BDNF by corticostriatal terminals (Gauthier et al., 2004; Zuccato and Cattaneo, 2007).

To gain a better mechanistic grasp of how alterations in BDNF signaling might selectively affect how iSPNs translate cortical inputs, the corticostriatal network was studied in brain slices from hemizygous BACHD mice crossed with reporter lines for dSPNs and iSPNs (André et al., 2011; Gerfen and Surmeier, 2011; Gray et al., 2008). The BACHD mouse is a transgenic model of HD in which the full-length human mutant huntingtin (mHtt) gene has been inserted using a bacterial artificial chromosome (BAC) (Gray et al., 2008). These mice display progressive motor and physiological deficits that are pronounced by 6 months of age (André et al., 2011; Gray et al., 2008). To our surprise, striatal levels of BDNF and mRNA for its receptor – the TrkB receptor (TrkBR) – appeared normal in symptomatic BACHD mice. This also was true in the Q175 knockin mouse model of HD, which displays a similar progressive motor and physiological phenotype (Menalled et al., 2012; Heikkinen et al., 2012). Moreover, activity dependent phosphorylation of TrkBRs – the first step in postsynaptic BDNF signaling – was normal in striata from 6 month old BACHD mice. However, downstream TrkBR signaling through Akt was significantly impaired. This deficit was attributable to up-regulation in the expression of phosphatase-and-tensin-homolog-deleted-on-chromosome-10 (PTEN) and amplification of BDNF signaling through p75 neurotrophic receptor (p75_NTR) – a well known inhibitor of TrkBR signaling (Song et al., 2010).

Results

BDNF expression and delivery to the striatum was normal in HD mice

At the outset of our study, the expression of BDNF mRNA in the cortex of BACHD and Q175 heterozygous knock-in mice was assessed using qPCR. In contrast to the original description of these mice (Gray et al., 2008), we found no evidence of reduced abundance of cortical BDNF mRNA in BACHD mice at either 2 or 6 months of age, nor did we find any evidence of reduced BDNF protein levels in either the cortex or striatum (Figure 1a–c; Figure S1a). To determine if this was peculiar to this HD model, 6 month old heterozygous Q175 knock in mice were examined, but again, no reduction in cortical BDNF mRNA expression was found (Figure 1d). Several previously published qPCR primer sets were examined to ensure that our results were not simply a consequence of primer choice or poor amplification efficiency. All of them yielded similar results. One possible explanation for the discrepancy is that previous work relied upon a single, highly variable ‘reference gene’ for normalization of transcript abundance, rather than a weighted average of several more stable transcripts (Pfister et al., 2011) (Figure 1; Figure S1d, Table S1).

Figure 1.

BDNF mRNA and protein levels are unaltered in BACHD mutant mice. (A) Diagram depicting cortical production and expression of BDNF, which is delivered to the striatum (top). Map of 6 primer sets used spanning the mouse BDNF gene (accession number NM_001048141.1; bottom). 1= Bdnf_Yang_IV; 2= Bdnf_Zuccato;; 3=Bdnf_Yang_CDS; 4= Bdnf_H03 ; 5= Bdnf_A03; 6=Bdnf_Conforti. (B) Boxplots showing relative cortical BDNF mRNA expression in 5–6 month old BACHD mice (WT: N=6; mutant: N=5), as measured with qPCR (right). (C) Western blots showing relative BDNF protein levels (normalized to tubulin) in the cortex and dorsal striatum of 6–8 month old WT and BACHD mice (WT: N=3; HD: N=3).(D) Boxplots showing relative cortical BDNF mRNA expression in 5–6 month old heterozygous Q175 knock-in mice (WT: N=5–6; Q175: N=5–6). *p<0.05, Mann-Whitney nonparametric test.

Although its cortical and striatal levels were not reduced in the HD models, BDNF release from cortical terminals and activation of postsynaptic TrkBRs could be impaired (Gauthier et al., 2004). To test this idea, two assays were performed. First, qPCR was used to assess the expression of full length or truncated TrkBR mRNA in iSPNs and dSPNs isolated by fluorescence activated cell sorting (FACS). No change in TrkBR expression was found in either type of SPN in symptomatic (5–6 month old) BACHD mice (Figure S1c). Next, depolarization-induced phosphorylation of striatal TrkBRs was used as an assay of BDNF release and receptor binding. Modest depolarization of wildtype brain slices by incubation in elevated extracellular KCl (17 mM) for 10 minutes significantly increased phophorylation of striatal TrkBRs (Figure 2a). However, TrkBR phosphorylation induced by depolarization was identical in 6–9 month old WT and BACHD mice (Figure 2b).

Figure 2.

Signaling downstream of TrkBRs is attenuated in the BACHD striatum. (A) Relative expression of phosphorylated TrkBR (pTrkB) protein (normalized to actin) in the dorsal striatum after incubation in 3 mM (left) or 17 mM (right) KCl. 17 mM KCl elevated phosphorylation of TrkBRs in WT dorsal striatum (3 mM: N=5; 17 mM: N=5). (B) Incubation in 17 mM KCl elevates pTrkBR protein expression (normalized to actin) similarly in WT and BACHD dorsal striatum (WT: N=4; HD: N=4). Data is normalized to WT median. (C) Incubation in 17 mM KCl led to significantly reduced phosphorylation of Akt in the dorsal striatum of BACHD mice (WT: N=5; HD: N=5). Phosphorylated Akt (pAkt) is expressed as the ratio of pAkt over total Akt. All data is from 6–9 month old mice. Representative Western blot lanes are shown below. * p<0.05, Mann-Whitney nonparametric test.

Taken together, our data suggest that in relatively young but symptomatic BACHD and heterozygous Q175 mice cortical production of BDNF, its delivery to the striatum, as well as its ability to bind and activate striatal TrkBRs were normal. So, if there is indeed a BDNF signaling defect in HD, it must be downstream of TrkBRs at this age.

TrkBR signaling can be assessed with cellular resolution

To test the hypothesis that TrkBR signaling was impaired, one of its targets – Akt – was examined following induction of BDNF release by depolarizing ex vivo brain slices. Indeed, the ratio of phosphorylated Akt to total striatal Akt was significantly lower in brain slices from BACHD mice than wild-type mice following circuit-level depolarization (Figure 2c), suggesting there was an HD-related deficit in TrkBR signaling. To help pinpoint the cellular site of this signaling deficit, RNAseq was performed on mRNA from 5–6 month old BACHD dSPNs and iSPNs that had been isolated by FACS (Figure S2; Table S2; Table S3). These data were then used to predict alterations in signaling networks in HD iSPNs and dSPNs. The top signaling network identified by this analysis was the huntingtin network in both iSPNs and dSPNs (data not shown). The next most prominently altered network was that of BDNF. Interestingly, the BDNF signaling network was diminished in BACHD iSPNs but elevated in dSPNs (Figure S2a–b). Not only does this analysis suggest that the BDNF/TrkBR signaling deficit is specific to iSPNs at this age, but it confirms the inference that there is not a general deficit in BDNF delivery to the striatum.

To unequivocally ascertain whether TrkBR signaling was impaired in SPNs, another assay was needed. Previous work had shown that corticostriatal long-term potentiation (LTP) in SPNs required BDNF activation of postsynaptic TrkBRs (Jia et al., 2010). This form of synaptic plasticity is postsynaptically induced and expressed, making it an ideal assay of postsynaptic TrkBR signaling.

Although necessary, TrkBRs are not sufficient to induce LTP at corticostriatal synapses. Postsynaptic NMDA receptors (NMDARs) and G-protein coupled receptors (GPCRs) linked to activation of protein kinase A (PKA) also are necessary for LTP induction (Calabresi et al., 2007; Kreitzer and Malenka, 2008; Shen et al., 2008). In iSPNs, the obligate GPCRs are adenosine A2a receptors (A2aRs). Co-activating these three signaling pathways –TrkBR, NMDAR and A2aR – should lead to postsynaptically induced LTP in SPNs which could be measured using two photon uncaging of glutamate on dendritic spines, creating a bioassay with cellular resolution. To this end, SPNs were voltage clamped at −90 mV, in the presence of pharmacological activators of TrkBRs (BDNF, 50 ng/ml), A2aRs (CGS21680, 200 nM) and NMDARs (NMDA, 5 µM). To prevent active propagation of activity and indirect effects on the striatal network, the Na⁺ channel blocker tetrodotoxin (1 µM) was bath applied. At −90 mV, NMDARs are blocked by Mg²⁺. To transiently unblock them and trigger LTP induction, SPNs were depolarized 4 times to +20 mV for 1 second, resting 10 seconds between depolarizations (Figure S3a). This cell-specific induction paradigm produced a significant, long-lasting enhancement in postsynaptic glutamate receptor currents at about one-third of the spines examined using two photon laser uncaging of glutamate (Figure 3a–f; Figure S3b–c). Co-activation of all three postsynaptic signaling pathways (TrkBR, NMDAR, A2aR) was necessary to induce potentiation of uncaging-evoked excitatory postsynaptic currents (uEPSCs), as disrupting any one signaling pathway (blocking BNDF signaling through TrkBRs with 150 nM K252a or scavenging BDNF with 4 µg/ml TrkB-Fc; blocking NMDARs with 50 µM AP5; blocking A2aRs with 200 nM SCH58261) resulted in the attenuation or loss of potentiation (Figure 3f). In dSPNs, this protocol also worked when the A2aR agonist was replaced with a D1 receptor agonist (Figure 3g). Both extracellular signal-regulated kinase (ERK) and phosphoinositide-3 kinase (PI3K), which are known to be activated by TrkBRs, were necessary to induce potentiation, as it was blocked by antagonizing either one (Figure 3f). Most importantly, this provided a bioassay for TrkBR signaling with the level of resolution needed.

Figure 3.

Induction of synaptic potentiation at iSPN dendritic spines. (A) Maximum intensity projection (MIP) of an iSPN from a 1 month-old D₂BAC mouse. (B) Somatic uEPSCs triggered at the two labeled spines in (A) before (pre) and 15 minutes after (post) the induction protocol. (C) Long lasting heterogeneous potentiation of four neighboring iSPN spines. (D) The induction protocol potentiated approximately one third of iSPN dendritic spines only when it included the 4 depolarizing steps to +20 mV (depolarizations: N=52 spines, 7 cells, 5 animals; no depolarizations: N=30 spines, 5 cells, 3 animals). (E) Spines potentiated similarly in juvenile (1 month, N=13, 5 animals) and aged (6–8 month) iSPNs (N=13, 2 animals) and dSPNs (N=7, 1animal). (F) Boxplots showing the percentage of spines per iSPN that potentiated (with (depol) and without (No depol) the 4 depolarizing steps to +20 mV, from (D)). No potentiation was seen without the depolarizing steps. Potentiation was attenuated by AP5 (50 µM; NMDA and glycine omitted; N=19 spines, 3 cells, 2 animals), the TrkB inhibitor K252a (150 nM; N=31 spines, 4 cells, 2 animals) the BDNF scavenger TrkB-Fc (4 µg/ml; BDNF omitted; N=23 spines, 3 cells, 2 animals), the PI3K inhibitor LY294002 (50 µM; N=32 spines, 4 cells, 2 animals), the MEK/ERK inhibitor U0126 (30 µM, N=34 spines, 5 cells, 3 animals) or the A2a receptor antagonist SCH58261 (200 nM, CGS21680 omitted; N=47 spines, 6 cells, 3 animals). (G) Photomicrographs showing cortical (left) and thalamic (right) injection sites of AAV-ChR2 in 1–2 month-old D₂BAC mice. Optogenetically evoked EPSCs were more readily potentiated in iSPN dendritic spines receiving cortical (N=48 spines, 5 cells, 3 animals) vs thalamic (N=32 spines, 9 cells, 3 animals) inputs. (H) Optically evoked EPSCs were potentiated in a mixed population of SPNs from mice expressing ChR2 under the control of the Thy1 promoter (Ind: N=42 spines, 7 cells; No Ind: N=27 spines, 4 animals, 5 cells; both the A2a adenosine (200 µM CGS21680) and D₁ dopamine (5 µM 6-Chloro-PB hydrobromide) receptor agonists included in bath). * p<0.05, Fisher’s Exact test (compared to depol). # p<0.05 Fisher’s Exact test (compared to No depol).

As cortical pyramidal neurons that project to the striatum release BDNF, but thalamic neurons that project to the striatum do not (Altar et al., 1997), it was our working hypothesis that BDNF-dependent synaptic potentiation was limited to corticostriatal synapses. To test this hypothesis, a channelrhodopsin 2 (ChR2) expression construct was virally delivered to sensorimotor cortex pyramidal neurons or to intralaminar thalamic neurons that project to the striatum. Two weeks after infection, brain slices were prepared and the responses of individual SPN spines to adjacent ChR2 stimulation with a fixed minimal diameter (approximately 1 µm) 473 nm blue laser were assayed in the presence of 1 µM TTX and 100 µM 4-AP (Petreanu et al., 2009). In the region of the striatum most closely linked to sensorimotor cortex, about 70% of iSPN spines on distal dendrites (>100 µm from the soma) were responsive to ChR2 stimulation in the cortically transduced mice (48 out of 68 spines). In this same region, about 30% (32 out of 115) of iSPN spines in distal dendrites were responsive to ChR2 stimulation in thalamically transduced mice. At corticostriatal synapses, the probability of inducing synaptic potentiation was similar to that seen previously (Figure 3h). In contrast, at thalamostriatal synapses the median probability of potentiation was zero (Figure 3h). Plasticity also was examined in SPNs from transgenic mice expressing ChR2 in cortical, but not in thalamic, neurons (Arenkiel et al., 2007). In these mice, potentiation was induced in roughly half the connected spines (Figure 3h), consistent with it being a corticostriatal phenomenon.

BDNF-dependent postsynaptic plasticity was lost in BACHD and Q175 iSPNs

To appraise the integrity of TrkBR signaling in SPNs expressing mHtt, synaptic potentiation was examined in brain slices from BACHD mice crossed into the SPN reporter lines (Figure 4a). In slices from the brains of mice that were 6–8 months of age – a time before frank cell loss (Gray et al., 2008) or gross dendritic atrophy (Figure S4), but a time when there are clear motor symptoms – uEPSC potentiation in dSPNs was normal (Figure 4b). However, in iSPNs, potentiation was effectively lost (Figure 4c). In younger (2–3 month old) BACHD mice, uEPSC potentiation in iSPNs was reduced, but not lost (Figure S5a), suggesting that the impairment was progressive. Furthermore, the impairment in synaptic plasticity was not unique to this HD model, as iSPNs from 6 month-old heterozygous Q175 knock-in mice had the same deficit (Figure 4d).

Figure 4.

LTP is lost in SPNs in two mouse models of HD. (A) MIPs of iSPNs from a 7 month-old WT and BACHD mouse. (B–D) Boxplots showing the percentage of spines per SPN that potentiate in response to the induction protocol in dSPNs (B; WT: N=40spines, 5 cells, 3 animals; HD: N=38 spines, 5 cells, 3 animals) and iSPNs (C; WT: N=40spines, 5 cells, 2 animals; HD: N=38 spines, 5 cells, 3 animals) from 6–8 month-old BACHD mice and iSPNs from 6–8 month-old heterozygous Q175 mice (D; WT: N=28 spines, 5 cells, 4 animals; HD: N=37 spines, 5 cells, 4 animals). uEPSC potentiation is significantly attenuated in both mutant lines. * p<0.05, Fisher’s Exact test. (E) Electrical induction of LTP in 6–8 month old WT and BACHD iSPNs (WT: N=6 cells, 3 animals; HD: N=5 cells, 3 animals). Percent of EPSC baseline after (25–30 min post induction) is shown to the right. (F) Miniature EPSCs in response to strontium-mediated asynchronous cortical glutamate release in 6 month-old WT and BACHD iSPNs. Stimulation artifact is truncated for presentation. (G) (Left) High magnification MIPs of distal iSPN dendrites from a 6 month old WT and BACHD mouse. (Right, Top) Boxplots showing that amplitudes of strontium-mediated cortical miniature EPSCs are significantly reduced in BACHD iSPNs (WT: N=8, 4 animals; HD: N=11, 3 animals) but not dSPN (WT: N=13, 3 animals; HD: N=13, 2 animals). (Right, Bottom) Boxplots showing that dendritic spine density is decreased in 6–8 month old BACHD iSPNs, but not dSPNs (WT iSPNs: N=6, 2 animals; BACHD iSPNs: N=7 3 animals; dSPN WT: N=6, 2 animals; dSPN BACHD: N=7, 2 animals). * p<0.05, Mann-Whitney nonparametric test.

To verify that the plasticity deficit in HD neurons also was present when corticostriatal terminals were engaged, a conventional high frequency stimulation (HFS) LTP induction protocol was tested (Calabresi et al., 2007; Jia et al., 2010). In this protocol, BDNF release was induced by electrical stimulation of corticostriatal axons. In wild-type iSPNs, this protocol induced a robust LTP (Figure 4e); however, the same induction protocol failed to induce LTP in iSPNs from 6 month old BACHD mice (Figure 4e; baseline EPSP amplitude was 3.01±0.72 in WT iSPNs and 4.06±1.11 mV in BACHD iSPNs, p=0.4286, Mann Whitney nonparametric test). These results are consistent with a previous study suggesting that LTP, as measured with field potential recordings, is reduced in the striatum of rapidly progressing R6/2 HD mice (Kung et al., 2007). Although these experiments do not unequivocally demonstrate a synaptic locus for the HD deficit, they are consistent with the impairment seen in isolated, voltage-clamped spines.

A deficit in the ability to induce LTP should lead to both a reduction in the number of glutamate receptors at corticostriatal synapses and a decrease in dendritic spine density. To test these predictions, corticostriatal miniature excitatory postsynaptic currents (mEPSCs) were measured following electrical stimulation of cortical afferent fibers in the presence of Sr²⁺, rather than Ca²⁺. Sr²⁺ substitution results in the asynchronous release of quanta from stimulated terminals (Xu-Friedman and Regehr, 2000). As predicted, the amplitude of mEPSCs was reduced at corticostriatal synapses on iSPNs (but not dSPNs) in brain slices from 6 month-old BACHD mice (Figure 4f). Although there was no change in total dendritic length (Figure S4c), dendritic spine density was reduced in iSPNs (but not dSPNs) from 6 month-old BACHD mice (Figure 4g). Thus, the deficit in LTP induction was paralleled by structural adaptations in HD iSPNs.

What was responsible for the LTP deficit in iSPNs expressing mHtt? Previous work has demonstrated that ifenprodil-sensitive extrasynaptic NMDARs increase in SPNs in the YAC128 mouse model of HD (Milnerwood et al., 2010). The situation was similar in BACHD iSPNs. As in the YAC128 iSPNs, the ratio of corticostriatal synaptic NMDAR to AMPAR currents was not altered in BACHD iSPNs; however, in the presence of the glutamate transport blocker DL-threo-β-Benzyloxyaspartic acid (allowing the engagement of extrasynaptic receptors) the NMDAR component was greater in BACHD than wild-type iSPNs (Figure S5b–c). 2PLUG experiments on distal spines of iSPNs corroborated this result (Figure S5d). Acute ifenprodil application did not restore uEPSC potentiation in iSPNs from BACHD mice, nor did it eliminate uEPSC potentiation in WT iSPNs (Figure S5e–f). However, our data does not exclude the possibility that sustained activation of extrasynaptic NMDARs- over days or weeks- triggers adaptations that underlie the synaptic deficit in HD models.

P75_NTR signaling was responsible for the loss of BDNF-dependent synaptic potentiation

If BDNF and TrkBRs are not altered, why is BDNF-dependent potentiation lost in BACHD iSPNs? Although down-regulation of the TrkBR linker proteins p52/p46 Shc has been reported in an in vitro model of HD (Ginés et al., 2010), these linkers are not necessary for TrkBR-dependent synaptic potentiation (Minichiello et al., 2002). In addition to activating TrkBRs, BDNF also can robustly activate p75_NTRs, albeit less potently (Bothwell, 1995; Zhang et al., 2008). The concentration of BDNF used to mimic synaptic release in our experiments was near the IC₅₀ of p75_NTRs (Bothwell, 1995). Furthermore, although developmentally down-regulated (Dechant and Barde, 2002), p75_NTRs continue to be expressed at low levels in adult SPNs from both wild-type and BACHD mice (Figure S5g–h), and have recently been implicated in HD pathology (Brito et al., 2013). To test for their involvement, p75_NTR expression was knocked down by viral delivery of a shRNA (Figure S5i). Indeed, knockdown of p75_NTR expression fully restored BDNF-dependent synaptic potentiation in BACHD iSPNs (Figure 5a).

Figure 5.

LTP is rescued in BACHD iSPNs by inhibition of p75_NTR signaling through PTEN or stimulation of FGF receptors. (A) The percentage of potentiating spines per cell (in response to the induction protocol) in 5–8 month old BACHD iSPNs in control conditions (no drug, N=46 spines, 6 cells, 2 animals; viral injection of a control shRNA, N=25 spines, 4 cells, 1 animal), after p75_NTR knockdown by viral infection of p75_NTR shRNA constructs (N=32 spines, 5 cells, 2 animals), in the presence of the p75_NTR inhibitor peptide pep5 (1 µM, N=7 spines, 5 cells, 4 animals), the ROCK inhibitor Y-27632 (10 µM, N=28 spines, 4 cells, 2 animals), the PTEN inhibitor bpV(pic) (5 µM; N=48 spines, 6 cells, 4 animals) or PTEN knockdown by viral infection with a PTEN shRNA construct (N=29 spines, 4 cells, 2 animals). (B) Relative change in PTEN mRNA expression in iSPNs of 5–6 month-old BACHD mice (WT: N=6, HD: N=6). (C) Electrical induction of LTP in 5–6 month old BACHD iSPNs (N=5 cells, 3 animals). Percent of EPSC baseline after (25–30 min post induction) is shown in the inset; BACHD iSPN data in the absence of bpV(pic) from Fig. 2e is shown for comparison. (D) The percentage of potentiating spines per cell in 6–8 month old BACHD iSPNs in the presence of aFGF (20 ng/ml, N=38 spines, 4 cells, 3 animals) or aFGF plus the FGFR antagonist PLX052 (10 µM; N=32 spines, 4 cells, 2 animals), with BDNF excluded. (E) Relative change in FGFR1 expression in iSPNs of 5–6 month old BACHD mice (WT: N=7; HD: N=6). ** p<0.05, Mann-Whitney nonparametric test; * p<0.05, Fisher’s Exact test. (F) Proposed model of LTP induction in an iSPN spine.

To provide an additional test of p75_NTR involvement, the effects of inhibiting its signaling partners were examined. p75_NTR interacts directly with the Rho GDP dissociation inhibitor (Rho-GDI), which controls activation of the small GTPase RhoA (Yamashita and Tohyama, 2003). RhoA, in turn, activates Rho-associated kinase (ROCK), which phosphorylates and activates phosphatase-and-tensin-homologdeleted- on-chromosome-10 (PTEN) (Yang and Kim, 2012). PTEN dephosphorylates phosphoinositide products of PI3K, blunting its signaling (Song et al., 2010). As shown above, PI3K signaling is necessary for TrkBR-induced uEPSC potentiation. To test for the involvement of Rho-GDI, slices were incubated with a TAT-peptide targeting its interaction domain with p75_NTR (referred to as “Pep5” by Yamashita et al.) (Ilag et al., 1999; Yamashita and Tohyama, 2003); Pep5 rescued BDNF potentiation in BACHD iSPNs (Figure 5a). Next, ROCK, the target of RhoA, was inhibited with Y-27632 (10 µM); this also rescued potentiation (Figure 5a). Lastly, PTEN was targeted; either pharmacologically inhibiting PTEN with dipotassium bisperoxo(picolinato)oxovanadate (PTP Inhibitor XV bpV(pic), 5 µM) or knocking down its expression with a virally delivered shRNA rescued potentiation measured by both 2PLUG (Figure 5a) and electrical stimulation (Figure 5c).

These results clearly point to p75_NTR signaling through PTEN as responsible for disrupting TrkBR signaling in HD iSPNs. This signaling pathway is also present in wild-type iSPNs, however pharmacological inhibition of PTEN had no effect on synaptic potentiation in wild-type iSPNs or dSPNs (Figure S5j). What was different in HD iSPNs? The expression of p75_NTR mRNA was not measurably different in iSPNs isolated by FACS from HD striata (Figure S5g–h). However, the expression of PTEN was significantly elevated in BACHD iSPNs, but not dSPNs (Figure 5b), providing a mechanistic foundation for the enhancement of p75_NTR signaling.

If the HD deficit in TrkBR signaling is simply a consequence of BDNF co-activation of p75_NTR, then the signaling of growth factors that do not bind to p75_NTR should be unaffected. In SPNs, fibroblast growth factor (FGF) binds to receptors that, like TrkBRs, are tyrosine kinase receptors that promote LTP induction (Eswarakumar et al., 2005; Flajolet et al., 2008). Replacing BDNF with FGF (20 ng/ml) led to normal synaptic potentiation in 6–8 month-old BACHD iSPNs (Figure 5d). Quantitative profiling of iSPN mRNA from 6 month-old BACHD mice revealed that FGFR 1 mRNA expression was elevated compared to controls (Figure 5e; Figure S5k), suggesting that FGFRs were part of an attempt to compensate for the TrkBR signaling deficit.

Discussion

For the past decade, the striatal pathology in HD has been attributed to a decrease in cortical BDNF delivery (Gauthier et al., 2004; Zhang et al., 2003; Zuccato et al., 2001; Zuccato and Cattaneo, 2007). As a consequence, BDNF replacement has been viewed as a promising approach for slowing disease progression (Canals et al., 2004; Zuccato and Cattaneo, 2007). Although it does not exclude the possibility that later in the progression of the disease such trafficking deficits manifest themselves, our work argues that deficits in TrkBR and p75_NTR signaling precede them.

A cornerstone of the BDNF delivery hypothesis has been the apparent drop in cortical BDNF mRNA expression in young HD mouse models. In our hands, there was no evidence of decreased cortical BDNF expression in either BACHD or heterozygous Q175 mice at 6 months of age – a time point at which behavioral and physiological deficits are evident (André et al., 2011; Gray et al., 2008). The discrepancy between our results and previous ones did not stem from primer design, as a variety of published and unpublished primer sets yielded the same result. However, the discrepancy could be due to differences in the way transcript abundance was estimated. For qPCR, an internal standard is necessary to correct the threshold cycle number for variation in the amount of starting material. Commonly, a ‘reference gene’ is used for this purpose that is assumed to be unaffected by the experimental intervention being examined. Betaactin or glyceraldehyde 3-phosphate dehydrogenase (GAPDH) transcripts often are used for this purpose, as in the previous studies of BDNF abundance (see Table S1). However, these two transcripts are not ideal standards because of variability in their expression (Figure S1). A more appropriate strategy is to use a weighted average of several stable transcripts to derive a standard. Doing this led to the conclusion that there was no change in BDNF mRNA levels in cerebral cortex of BACHD and heterozygous Q175 HD models. In agreement with this conclusion, the phosphorylation of striatal TrkBRs in response to depolarization-induced release of BDNF by cortical terminals was normal in tissue from BACHD mice.

It is worth noting, however, that in 6 month old homozygous Q175 knock-in mice, there was a significant decrease in cortical BDNF expression (Figure S1b), in agreement with the assertion that mHtt can regulate BDNF expression when expressed at high enough levels (Canals et al., 2004). However, because of the more rapid disease progression in these mice (Menalled et al., 2012; Heikkinen et al., 2012) comparisons to the BACHD model are problematic. Moreover, the relevance of this gene-dose effect to the human condition is uncertain (Wexler et al., 1987).

In spite of there being no detectable change in cortical BDNF expression or striatal delivery in early stage HD models, there was a clear deficit in the postsynaptic response of iSPNs to BDNF. To unequivocally assay postsynaptic TrkBR signaling in a single, phenotypically defined neuron, a strategy for inducing synaptic potentiation at visualized spines that did not depend upon activation of presynaptic fibers was developed. Unlike synaptic potentiation at other synapses (Harvey and Svoboda, 2007; Tanaka et al., 2008), corticostriatal long-term synaptic potentiation requires co-activation of three synaptic or peri-synaptic proteins: TrkBRs, NMDARs and a G_olf coupled receptor (A2aRs or D1Rs, depending upon SPN type) (Calabresi et al., 2007; Jia et al., 2010; Shen et al., 2008). There is no requirement for presynaptic activity other than to release the agonists for these receptors. The only postsynaptic induction requirement (beyond intracellular signaling) is membrane depolarization to relieve the Mg²⁺ block of NMDARs. In the presence of agonists for these three receptors, four brief postsynaptic depolarizations (to open NMDARs) were all that was required to produce lasting potentiation of axospinous, corticostriatal synapses, as judged by focal uncaging of glutamate on spine heads or optogenetic stimulation of corticostriatal terminals. These studies demonstrates not only the factors necessary for induction of corticostriatal potentiation but, for the first time, what is sufficient for induction.

The ability to induce corticostriatal synaptic potentiation using this protocol or with a conventional HFS protocol was lost specifically in iSPNs from BACHD and heterozygous Q175 mice. Potentiation was normal in dSPNs. In BACHD iSPNs, there were two structural correlates of the deficit. One was a reduction in the average number of postsynaptic AMPARs, as estimated by the amplitude of corticostriatal mEPSCs at membrane potentials that were not permissive for NMDAR currents. The other was a reduction in spine density. Both observations are consistent with previous work suggesting that potentiated axospinous synapses not only have more synaptic AMPARs but are more stable (Kasai et al., 2010; Kopec et al., 2006; Mendez et al., 2010).

This cell-specific deficit in potentiation was attributable to up-regulation of p75_NTR signaling through PTEN, resulting in impaired TrkBR activation of PI3K and Akt. Striatal p75_NTR signaling is elevated in other HD models (Brito et al., 2013), buttressing this conclusion. Both pharmacological and molecular approaches supported this conclusion. Either decreasing p75_NTR expression or blocking its coupling to RhoA normalized TrkBR signaling. Antagonizing the target of RhoA, ROCK, also corrected TrkBR signaling. Either decreasing PTEN expression with a shRNA or pharmacologically diminishing its activity restored TrkBR signaling. The alteration in p75_NTR signaling in HD iSPNs was attributable to the up-regulation of PTEN, not p75_NTR itself. Why PTEN was up-regulated is unclear. It is possible that PTEN trafficking is altered by mHtt, leading to alterations in expression and function (Gericke et al., 2006).

One additional piece of evidence in support of the involvement of p75_NTR signaling was the finding that synaptic potentiation was restored in HD iSPNs by FGF. FGFRs expressed by iSPNs have previously been shown to promote the induction of corticostriatal synaptic potentiation through mechanisms similar to those of TrkBRs (Flajolet et al., 2008). However, unlike BDNF, FGF does not bind to p75_NTR. Thus, astrocyte release of FGF (Tooyama et al., 1993) and FGFR signaling might partially compensate for the loss of trophic support derived from BDNF.

Our results also are concordant with work implicating Ras Homolog Enriched in Striatum (Rhes) in HD pathogenesis (Subramaniam et al., 2009). Rhes inhibits Akt signaling (Harrison et al., 2013) and the elimination of Rhes increases Akt phosphorylation and diminishes the impact of mHtt (Harrison et al., 2013; Harrison and Lahoste, 2013). Moreover, extrasynaptic NMDAR signaling – which is up-regulated in SPNs of HD models (Milnerwood et al., 2010) – increases Rhes expression (Okamoto et al., 2009). Thus, extrasynaptic NMDAR signaling through Rhes and p75_NTR signaling through PTEN might synergize to suppress Akt signaling, which is known to be necessary for neuronal vitality (Humbert et al., 2002).

As the anchors of the basal ganglia circuitry responsible for motor suppression, diminished inervation of iSPNs can provide a straightforward explanation for early hyperkinetic HD motor symptoms (Gerfen and Surmeier, 2011; Menalled and Chesselet, 2002; Raymond et al., 2011). Although this behavior may not be fully manifest in rodent models at this age, complete removal of TrkBR signaling in iSPNs leads to hyper-locomotion in otherwise normal mice (Besusso et al., 2013). This deficit could in turn trigger adaptations throughout the neural network linking the striatum and cortex (Kozorovitskiy et al., 2012), resulting in cortical pathology associated with HD (Cepeda et al., 2007). It also is possible that enhanced p75_NTR signaling in the cerebral cortex induces pathology, as deficits in TrkBR signaling, whether in the striatum or cortex, have profound effects on neuronal function, as well as mHtt toxicity (Humbert et al., 2002; Warby et al., 2005). While in our slice preparation p75_NTR activation is likely achieved by BDNF, it should be noted that other ligands (such as proNGF or proBDNF) may contribute to p75_NTR engagement in vivo (Dechant and Barde, 2002; Fayard et al., 2005; Song et al., 2010). Although it is unlikely that all of the pathology in HD can traced to a single cause, our studies strongly suggest that in the early stages of the disease, p75_NTR antagonism should be part of an effective therapeutic strategy for restoring normal BDNF and Akt signaling. Moreover, given that its tissue distribution narrows and expression level falls with development (Dechant and Barde, 2002), p75_NTR antagonism should be less likely to have unacceptable side-effects in adult HD patients than strategies targeting TrkBRs, which continue to be robustly and widely expressed in the adult brain.

Abbreviated Experimental Procedures

Electrophysiology

Parasagittal brain slices (275 µm) were prepared from D₂BACEGFP (FVB), Thy1-ChR2 (C57BL/6), BACHD (FVB) × D₂BAC-EGFP (FVB), BACHD (FVB) × D₁BAC-EGFP (FVB) or Q175 (C57BL/6) transgenic mice following procedures approved by the Northwestern University Animal Care and Use Committee, as previously described (Day et al., 2008). Patch pipettes were loaded with internal recording solution containing (in mM): 120 CsMeSO₃, 5 NaCl, 10 TEA-Cl (tetraethylammonium-Cl), 10 HEPES, 5 Qx-314, 4 Mg-ATP, 0.3 Na-GTP, 0.2 Fluo 4 pentapotassium salt and 0.05 Alexa Fluo 568 hydrazide Na salt (Invitrogen), pH 7.25, 270–280 mOsm^–1. Slices were transferred to a submersion-style recording chamber on a fixed stage Olympus BX51 upright microscope and perfused with 95%O₂/5%CO₂-bubbled ACSF. containing (in mM) 124 NaCl, 3 KCl, 2 CaCl₂, 1 MgCl₂, 26 NaHCO₃, 1 NaH₂PO₄, and 16.66 glucose. Electrophysiological recordings were made using a Multiclamp 700B amplifier; protocols were delivered and recorded using the custom-written freeware package WinFluor (John Dempster, Strathclyde University, Glasgow, Scotland, UK), as previously described (Day et al., 2008; Plotkin et al., 2011). Recordings were performed at room temperature and filtered at 1 kHz. Strontium-mediated asynchronous miniature EPSC amplitudes (2 mM Sr²⁺ in external recording solution) were measured as previously described using an internal recording solution composed of (in mM) 120 CsMeSO₃, 15 CsCl, 8 NaCl, 10 TEA-Cl, 10 HEPES, 2–5 Qx-314, 0.2 EGTA, 2 Mg-ATP and 0.3 Na-GTP (Ding et al., 2008).

Single spine synaptic potentiation

SPNs were held at −90 mV in the presence of a “LTP cocktail” containing 50 ng/ml BDNF, 5 µM NMDA, 5 µM glycine, 1 µM TTX, 200 nM CGS21680, 1 µM sulpiride, 2 µM muscarine, 50 µM CPCCOEt, 1 µM MPEP and 0.1% BSA, dissolved in HEPES buffered ACSF, unless otherwise stated. In experiments performed in dSPNs, CGS21680 was replaced with 5 µM 6-Chloro-PB hydrobromide and 0.1% sodium metabisulfite. Muscarine had no detectable effect on LTP induction, and was thus omitted in most cases (in all compared groups). This LTP cocktail was superfused (0.4 ml/hr) over the slice using a syringe pump and multi-barreled perfusion manifold (Cell MicroControls, Norfolk, VA). Baseline uEPSCs were recorded from approximately 8 coplanar distal (greater than 90 µm from soma) spines, separated by 200 ms, and averaged 3–8 times (5 s inter-trial interval). The cell was then stepped from −90 to +20 mV for 1 s, 4 times total, 10 s between steps. uEPSCs were then measured from the same spines in the same manner 10–30 minutes after the depolarizing voltage steps. Unless otherwise indicated, represented time points are 10–15 minutes post-induction. Potentiation was defined as an increase in EPSC amplitude of at least 150% of baseline (based on Figure 1d; Figure S5). Series resistance was monitored, and recordings with a change of more than 20% were rejected from analysis.

2-photon laser scanning microscopy, 2-photon laser uncaging and ChR2 stimulation

dSPNs and iSPNs were identified by somatic EGFP expression using 2-photon excited fluorescence induced and detected using a Prairie Ultima laser scanning microscope system (Prairie Technologies). Fluorescent and bright-field images were viewed in register using a Dodt contrast detector system. Cells were patched using video microscopy with a Hitachi CCD camera and an Olympus 60×/1.0 NA lens. Green (EGFP) and red (Alexa 568) signals were acquired using 810 nm excitation (Verdi/Mira laser). Following patch rupture the internal solution was allowed to equilibrate for 10–15 minutes before imaging. High magnification z-series were acquired with 0.072 µm² pixels with 4 µs dwell time and 0.3 µm z-steps. At the end of each experiment whole cell z-series were acquired with 0.36 µm² pixels with 10 µs dwell times and 1 µm z-steps. Glutamate uncaging was achieved using a Chameleon-XR laser system (Coherent Laser Group, Santa Clara, CA). 5 mM MNI-glutamate (Tocris, Cookson, Ellisville, MO) was superfused over the slice through the same syringe used to deliver the “LTP cocktail.” Glutamate was uncaged adjacent to individual spines using 1 ms pulses of 720 nm light typically 10–20 mW in power at the sample plane. Photolysis power was tuned via a third Pockels cell modulator (Con Optics, Danbury, CT) to achieve uncaged- EPSCs (uEPSCs) averaging 5–15 pA. ChR2 stimulation was achieved with a Prairie Instruments blue laser (473 nm) and shuttered to a minimal fixed diameter (approximately 1 µm). Presynaptic terminals adjacent to individual distal spines were targeted. All ChR2 studies were performed in the presence of 1 µM TTX, 100 µM 4-AP and the “LTP cocktail”.

Electrical LTP induction

iSPNs from parassagital 6–7 month old WT and BACHD mice were patched with electrodes containing (in mM): 135 KMeSO₄, 5 KCl, 0.5 CaCl₂, 10 HEPES, 2 ATP-Mg, 0.5 GTP-Na, 5 phosphocreatine-Tris, 5 phosphocreatine-Na and 0.3 EGTA. Recordings were made in continuously perfused oxygenated Mg²⁺-free ACSF containing (in mM): 135 NaCl, 2.5 KCl, 25 NaHCO₃, 1.25 NaH₂PO₄, 10 glucose and 2 CaCl₂. Electrical stimulation was performed with a parallel bipolar electrode placed in the overlying cortex. The high frequency stimulation (HFS) LTP induction protocol was composed of 1 sec 100 Hz trains, repeated 4 times at a 10 sec interval, and performed as previously described by others (Jia et al., 2010). Series resistance was monitored, and recordings with a change of more than 20% were rejected from analysis.

Stereotaxic ChR2 injections

Stereotaxic injections of adeno-associated viral vector (AAV) carrying genes for channelrhodopsin (ChR2) and mCherry (Addgene 20938) were made in either the sensorimotor cortex or thalamus (encompassing the intralaminar nuclei and parts of the ventral thalamus) of isoflurane-anesthetized 6 week old D₂BAC-EGFP mice. The titer was 2×10⁹ genomic copies per microliter. Mice were allowed to recover for at least 2 weeks post injection. Post hoc visualization of injection sites was performed in acute slices after recordings.

Gene expression profiling

Quantitative polymerase chain reaction (qPCR) was used to determine the abundance of transcripts of interest with procedures similar to that described previously (Chan et al., 2012). To increase accuracy of gene expression analysis, a panel of reference genes (Atp5b, Cyc1, Gapdh, H2afz, Hmbs, Uchl1) were included in order to identify appropriate reference genes with stable expression across genotypes and mouse strains. Weighted PCR cycle thresholds (CTs) based on the stability of each housekeeping gene were calculated. Experiments for each gene of interest were run in triplicates. Desalted primers were custom synthesized (Invitrogen) and intron-spanning whenever possible. The mRNA levels in each subgroup of samples were characterized by their median values. Results were presented as fold difference relative to the median of their respective wild-type controls.

Western blot analysis

Coronal brain slices (350 µM) were prepared from isofluraneanesthetized mice as described above. After recovering in normal ACSF for 30–60 minutes (35°C), slices were incubated in normal ACSF or ACSF containing 17 mM KCl (in mM: 110 NaCl, 17 KCl, 26 NaHCO₃, 1 NaH₂PO₄, 2 CaCl₂, 1 MgCl₂, 12.5 glucose, 300–305 mOsm^–1) for 10 minutes. Assuming intracellular K⁺ concentrations of 120 mM, 17 mM external KCl depolarizes neurons to the new K⁺ equilibrium potential of −50.2 mV, evoking both pre- and post- synaptic activity. After incubation, dorsal striata and overlying cortex were dissected from each slice, homogenized in tissue extraction reagent I (Invitrogen) premixed with Halt protease and phosphatase inhibitor cocktail (Thermo), sonicated and centrifuged. Supernatants were electrophoresed on a 4–12% SDS-polyacrylamide gel, transferred onto a poly-vinylidene difluoride membrane (Invitrogen), blocked with 5% bovine serum albumin in TBST (tris buffer saline tween), incubated overnight at 4 C with primary antibodies, then washed and incubation with appropriate anti-rabbit or anti-mouse peroxidase-conjugated secondary antibodies. Antibodies were BDNF (1:200, Sigma), phospho-Akt (Ser473, 1:1000, Cell Signaling), Phospho-TrkB (Y515, 1:500, Abcam), Akt (1:1000, Millipore), α-Tubulin (1: 500, Sigma), β-Actin (1:1000, Cell Signaling), p75_NTR (1:1000, Millipore), Gapdh (1:2000, Abcam). Immunoreactivity was detected with an enhanced chemiluminescent reagent (Advansta) and visualized by a ChemiDoc XRS imager (Bio-Rad). The integrated optical density of the target band was quantified in ImageJ (NIH) after background subtraction. The levels of the protein of interest were expressed as a ratio to the relevant loading controls. Dorsal striata of both hemispheres of each animal were combined to equal one sample.

shRNA knockdown

To silence PTEN expression an AAV serotype 2 vector encoding for rat PTEN shRNA (siPTEN) was created. The sense, hinge and siPTEN sequence used to make the PTEN shRNA was 5’- GATCCCCGAGTTCTTCCACAAACAGAACTTCCTGTCATTCTGTTTGTGGAAGAACTCTTTTTTGGAAT-3’. The integrity of construct was verified by sequencing. Vector stocks were prepared by packaging the plasmids into AAV serotype 2 particles using helper-free plasmid transfection system (Morgenstern et al., 2011). To silence p75_NTR expression, an shRNA oligomers targeting the mouse p75_NTR (NM_033217.3) was created using shRNA design tools. The uniqueness of selected target sequences was confirmed by NCBI blast search. Oligos were synthesized from IDT. Two shRNA and one control constructs were created: p75–233-shRNA (CAAGGAGACATGTTCCACA),-p75–890 (CGCTGACAA CCTCATTCCT) and (AGGATCAAATTGATAGTAAACC) expressing tdTomato, and were packaged into AAV9 (Virovek). Knockdown of p75_NTR expression was confirmed at the mRNA level by qPCR in PC12 cells. AAVs were stereotaxically injected into the dorsal striatum as described above. For p75_NTR injections, both 233 and 890 constructs were co-injected. Mice were allowed to recover for at least two months before slice experiments were performed. Infected SPNs were identified by expression of a red fluorescent reporter.

NMDA/AMPA receptor ratio measurements

SPNs were voltage clamped with a cesium-based internal recording solution (described above in Sr²⁺ experiments). Electrically evoked EPSCs were induced as previously described (Ding et al., 2008). AMPA receptor mediated responses were measured as the peak inward current recorded at a holding potential of −70 mV. NMDA receptor mediated responses were measured as the amplitude of the outward current 50 ms following the stimulus artifact, at a holding potential of +40 mV. Measurement of EPSC areas in the presence of the glutamate uptake blocker TBOA was performed as described by others (Milnerwood et al., 2010). Uncaging evoked EPSCs were evoked as described in the main text. AMPA and NMDA receptor components of uEPSCs were measured as described above, from holding potentials of −80 mV and +40 mV, respectively.

Dendrite morphology and spine density analysis

Dendrite morphology was calculated from low magnification z-series of Alexa Fluor 568 filled SPNs described above. Acquired image stacks were deconvolved using a 60–65 iteration adaptive blind deconvolution algorithm with AutoQuant X software (MediaCybernetics). The deconvolved image stack was then imported into the Imaris data visualization package (Bitplane Scientific Software). Dendrite morphology was analyzed using the filament tracer module and the incorporated Sholl analysis. For spine density measurements, high magnification images of Alexa Fluor 568 dendrites (approximately 90–120 µm from soma) were obtained as described above (pixel size = 0.072 µm²). MIPs were created from 0.5 µm z-steps, and spines were manually counted along a dendritic region of known distance.

Immunohistochemistry

WT mice were perfused with 4% paraformaldehyde (PFA) and 50 µm coronal slices were made through the dorsolateral striatum. Immunohistochemistry was performed on free-floating brain slices using a rabbit polyclonal antisera made against the intracellular domain of the mouse p75 receptor (1:200 dilution; Millipore) (Huber and Chao, 1995). Detection was achieved using an Alexa-488-conjugated anti-rabbit secondary antibody. Slices were mounted on slides using Vectashield mounting medium and visualized on a Fluoview FV10i confocal microscope (Olympus).

RNAseq

FACS BACHD iSPNs and dSPNs were pooled from four animals (150,000 cells/animal) to create iSPN and dSPN pools for sequencing. Sequencing was performed by Expression Analysis on an Illumina Hi-seq 2000. Paired-end sequencing was performed, 4-plexed across lanes for a total of around 38 million 50mer paired reads per sample. Alignment and QC were conducted in Omnicsoft using the OSA algorithm (Hu et al., 2012); FPKMs were then calculated following standard formulas. QC assessment found all samples of high quality both at the RNA quality and alignment mapping levels. For each comparison (BACHD vs WT), significance was assessed using DESeq (Anders and Huber, 2010) using the single replicate protocol of DESeq. The resulting RNAseq data, p-values and fold changes were submitted to Ingenuity’s upstream regulator analysis, and the confidence of identified matches determined with a Fisher’s exact test and connected using Ingenuity’s knowledge base.

Statistical analysis

Differences in volume, uEPSC amplitudes and mRNA expression were examined using the Mann-Whitney U non-parametric test of significance, unless otherwise stated. Normality was not assumed for comparisons. Data points greater than 1.5 standard deviations from the mean were considered outliers and removed. The percent of spines that potentiated vs did not potentiate in different experimental groups was examined using the Fisher’s Exact test (or chi-square test if n>100), with potentiation defined as 150% of baseline (Figure 1d). Variance is graphically presented as error bars or box-whisker plots. Control and experimental groups were interleaved where possible. Differences were considered statistically significant if p<0.05.