The Role Of LRP4 In Modulation Of Excitatory Transmitter Release And Exploration Of The Underlying Mechanism

Neurons are the subjects that mediate information transmission in the nervous system. Nowadays, more and more studies suggested an important role of astrocytes in modulating information processing and thus considered a part of ’tripartite synapse’. Astrocytes can sense synaptic activity by expressing a variety of neurotransmitter receptors which are mainly G proteins associated, and in turn regulating neuronal transmission by releasing "gliotransmitters", such as D-serine, Glutamate and ATP. As a co-agonist binding to glycine site of NMDA receptor, D-serine released from astrocytes was demonstrated to enhance NMDA receptor activiation and enabled LTP induction in vitro and in vivo. Raising [Ca2+]; in astrocytes pharmacologically or by electrical neuronal stimulation, leads to Glutamate release from astrocytes, which evokes slow inward currents in nearby neurons by activating extrasynaptic NMDA receptors, or promote presynaptic transmitter release either through activating extrasynaptic NR2B subunit containing NMDA receptors or group I mGluRs. ATP, acting as a signaling molecule, nowadays is drawing more and more attention in neuroscience. ATP mediates Ca2+ wave in astrocytes and suppress firing rate of neurons through adenosine A1 receptor by converting to adenosine by ectoenzymes. Glutamate released from neurons activated ATP releasing from astrocyte and in turn causes heterosynaptic suppression through presynaptic P2Y receptors directly in vitro and presynaptic A1 receptors activated by adenosine, a degradation product of ATP in vivo. Blocking the release of ATP from astrocytes by expressing a dominant-negative SNARE domain specifically in astrocytes enhanced field excitatory synaptic potential, mediated through presynaptic A1 receptors activated by lower level of adenosine converted from ATP, suggesting a critical role of ATP from astrocytes on neuronal activity and an exocytotic way of releasing ATP from astrocyte. Furthermore, ATP release from astrocytes was also suggested through Ca2+-dependent exocytosis of lysosome. However, how ATP release was regulated is not well known. Pathologically, purinergicsignaling is involved in many disease, such as epilepsy. Adenosine could reduce seizures in a rat model of temporal lobe epilepsy induced by pilocarpine. A ketogenic diet which has been used in the treatment of refractory epilepsy has been reported that the anti-epileptic effect was mediated by increasing activation of adenosine A1 receptors, which was caused by reduced adenosine kinase.LRP4, a member of the low-density lipoprotein receptor gene family, is a single-transmembrane protein with a large extracellular domain with multiple LDLR repeats, EGF-like and β-propeller repeats. Mice die at birth in the absence of LRP4, and do not form the NMJ, it was then identified that LRP4 is a co-receptor for agrin and is necessary and sufficient to enable agrin signaling. Further studies on LRP4 suggested LRP4 from muscle also serves as a retrograde signaling for presynaptic differentiation at NMJ while motoneuron LRP4 is dispensable for NMJ formation although it is important for postsynaptic differentiation. In light of NMJ serving as an informative model of synaptogenesis and LRP4 also expresses in the brain, it is attempting to know whether LRP4 has a function in central nervous system (CNS). To date, despite a few studies showing that LRP4 is located in the postsynaptic region and bind to PSD95 via its C-terminal PDZ domain and affects neuronal viability in vitro serving as a receptor for apolipoprotein E (apoE), however, exact expression and function of LRP4 in the CNS are largely unclear.To investigate the role of LRP4 in the CNS, we generated various LRP4 conditional knockout mice, in which GFAP-Cre; LRP4 f/f mouse is whole-brain knockout one. Western blot showed LRP4 band was totally deleted from GFAP-Cre; LRP4 f/f hippocampus. GFAP-Cre; LRP4f/f mouse had nomal weight, cortical lamina structure as well as numbers of neurons and astrocytes (body weight:P= 0.518; density of neurons:P= 0.405; density of astrocytes:P= 0.402). Interestingly, when we tried to induce seizure by pilocarpine, GFAP-Cre; LRP4f/f mice always showed slower seizure development (P= 0.011). They did so when we used another seizure inducing drug (PTZ) (P= 0.023), indicating that abnormal neuronal activity in the GFAP-Cre; LRP4f/f mice. Electrophysiologically, we found lower sponetanous firing rate in CA pyramidal neurons of GFAP-Cre; LRP4f/f mouse (P= 0.002), even so when we blocked inhibitory activity (P= 0.038). Along with normal intrinsic excitability (P= 0.94), it suggests reduced excitatory input strength in GFAP-Cre; LRP4f/f mice. In addition, recording of sEPSC showed lower frequency with normal amplitude (frequency:P= 0.017; amplitude:P= 0.358), whereas sIPSC was normal (frequency:P= 0.381; amplitude:P= 0.563), these further demonstrated reduced excitatory strength. In agreement, in muscle rescued LRP4 ko mice, which is another brain LRP4 ko line, we also found decreased frequency of sEPSC with normal sIPSC (sEPSC frequency:P= 0.015; sEPSC amplitude:P= 0.545; sIPSC frequency:P= 0.848; sIPSC amplitude: 0.754).Reduced excitatory input strength could be caused by several reasons as below:1. Spine numbers were decreasd; 2. Dysfunction of post synapse, such as abnormal AMPA receptor function or decreased number of AMPA receptors; 3. Decreased presynaptic vesicle release. To address this question, first we did golgi staining to check density of spines in CA1 neuronal apical dendrites, and found no difference of density between wildtype and GFAP-Cre; LRP4f/f mice (P= 0.784). Comparation property of AMPA and NMDA receptor and their ratio showed no significant difference (AMPA:P= 0.204; NMDA:P= 0.883; ratio:P= 0.652), indicating normal function of post synapse. To test the probability of presynaptic vesicle release, we first test paired-pulse ratios, and found increased ratios in CA1 neurons of GFAP-Cre; LRP4f/f mice (P= 0.000), indicating altered presynaptic vesicle release. Then we utilized NMDA receiptor blocker MK-801 to test decay rate of NMDA mediated current. Significant lower decay rate was found in GFAP-Cre; LRP4f/f mice (P= 0.004). Similarly, by minimal stimulation approach, decreased probability of presynaptic vesicle release was also suggested (P= 0.013). Along with normal synaptic potency (P= 0.901), these results strongly suggested that reduced excitatory input strength was caused by decreased presynaptic vesicle release other than dysfunction of post synapse.Decreased presynaptic vesicle release could be caused by two possible reasons:1. Abnormal development of axon terminal; 2. regulating signaling that suppressed presynaptic release. To address this question, first we took EM images from Schaffer collateral (SC) region in hippocampus and found unchanged density of synapses, length of PSD, density of vesicles as well as diameter (synapse:P= 0.707; PSD:P= 0.203; Vesicle:P= 0.737; Diameter:P= 0.439), suggesting normal axon terminal development. Vesicles in terminal could be divided into three pools:Reserve pool; Recycling pool; Readily releasable pool (RRP), in which RRP is tightly related with transmitter release. To test whether vesicles in RRP is normal, we puffed sucrose (0.5 M) to SC region which could evoked all the vesicles in RRP releasing simultaneously. In agreement with EM study, size of RRP was not changed in GFAP-Cre; LRP4f/f mice (Charge:P= 0.624), suggesting altered release of presynaptic vesicle is due to unknown regulation mechanism rather than structural problem.Because of the lack of working antibody for staining, we generated other several conditional ko mice. NEX-Cre and Camk2a-Cre are known that ere activity is specifically in excitatory neurons. When crossed with LRP4f/f mice, supposedly LRP4 will be deleted from excitatory neurons in the NEX-Cre; LRP4f/f and Camk2a-Cre; LRP4f/f mice. Interestingly, Western blots showed no change of LRP4 level in hippocampus of these two ko mice, indicating no or very little expression of LRP4 in excitatory neurons. Functionally, we test the response to seizure induction by PTZ, sEPSC and paired-pulse ratios, all of which were not significantly different (PTZ:0.858; sEPSC frequency:P= 0.521; ratio:0.74). These results further demonstrated an dispensable role of LRP4 in excitatory neurons. Similarly, we generated PV-Cre; LRP4f/f mice in which LRP4 was deleted from PV-positive inhibitory neurons. Western blots and functional test all showed negative results (PTZ:P= 0.650; sEPSC frequency:0.589; ratio:P= 0.342). Combining all the results together suggested that LRP4 in astrocytes may be critical for neuronal activity.To test the hypothesis, we designed co-culture experiment. Pure Neurons and astrocytes from wildtype and GFAP-Cre; LRP4Cf/f genotypes were cultured separately. On DIV11, neurons were cultured together with astrocyte non-contactly. One day later we recorded sEPSC from neurons and found frequency of sEPSC was decreased when wildtype neurons were put together with ko astrocytes (P= 0.007). However, frequency of sEPSC from group in which koneuorns were with wildtype astrocytes was not changed (P= 0.516), indicating a dispensable role of LRP4 in neurons. Unexpectedly, frequency of sEPSC from group in which ko neurons were with ko astrocytes was decreased (P= 0.004). Results above suggested that LRP4 in astrocytes but not in neurons is critical for neuronal activity.To study mechanism of how neuronal activity was regulated by LRP4 in astrocytes, we tested whether gliotransmitters were involoved. Glutamate and D-serine from astrocytes has been studied that they promoted presynaptic transmitter release through presynaptic NMDA receptors. We blocked activity of NMDA receptors by AP-5 and then recorded sEPSC which haven’t shown significant difference, suggesting these two molecules were not involved. Astrocytes could release ATP, which had been reported regulating presynaptic vesicle release. To study this, we treated slices with exogenous ATP and found frequency of sEPSC recorded from wildtype mice showed significant decrease whereas no change in ko mice, indicating loss of LRP4 in astrocytes may promote releasing ATP from astrocytes. To test the hypothesis, we utilized luciferin kit to measure the ATP concentration of medium from cultured astrocytes and neurons, showing dramatic increased concentration of ATP in ko astrocytes group with no change in neuorns, which is in agreement with our hypothesis. To search the downstream of ATP signaling which mediates decreased frequency of sEPSC. We test the effect of P2 receptor antagonist suramin, A2A receptor antagonist SCH58261 and A1 receptor antagonist DPCPX, and found only DPCPX showed different effect on wildtype and ko mice, suggesting A1 receptor was the mediater.LRP4 has been known as a co-receptor in NMJ, and studies also suggested a role of LRP4 in regulation of wnt signaling. To test whether Agrin or wnt plays a role in ATP release from astrocytes by binding to LRP4, we treated Agrin or wnt5a on cultured astrocytes for 24 hr and measured the ATP concentration of medium. Results showed that Agrin but not wnt5a dramatically decreased the concentration of ATP in wildtype astrocytes culture, however it had no effect on GFAP-Cre; LRP4f/f astrocytes. These results suggested that Agrin-LRP4 signaling in CNS regulates ATP release from astrocyte through which it may affect neuronal activity. Indeed, frequency of sEPSC was significantly increased when Agrin was treated on slice from wildtype mice but not ko ones.In conclusion, our study demonstrated that Agrin increases excitatory neuronal activity by binding to LRP4 in astrocytes which leads to less ATP release and finally less adenosine. To date, it is the first time that we report specific LRP4 expression pattern and the function of agrin-LRP4 signaling in central nervous system, which will shed light on the mechanism on the interaction between neurons and astrocytes.