1 2 Title : 3 An active role for astrocytes in synaptic plasticity ? 4 5

28 29 Recently, Henneberger et al. (2010) block hippocampal long-term synaptic 30 potentiation (LTP) induction by “clamping” intracellular calcium concentration of individual 31 CA1 astrocytes, suggesting calcium-dependant gliotransmitter release from astocytes plays 32 a role in hippocampal LTP induction. However, using transgenic mice to manipulate 33 astrocytic calcium Agulhon et al. (2010) demonstrate no effect on LTP induction. Until the 34 question of how intracellular calcium causes gliotransmitter release is answered, the role of 35 astrocytes in synaptic plasticity will be incompletely understood. 36 37 38 BODY TEXT 39 While glial cells were discovered over 100 years ago, an understanding of their role 40 in brain physiology has come about relatively recently. From their original name “glia” it 41 was apparent that they were thought to simply glue the brain together. Later, it was found 42 that glia comprise a heterogeneous population of cells that include oligodendrocytes, 43 astrocytes and microglia, each performing unique physiological functions in the nervous 44 system (Somjen 1988). In particular, astrocytes are important for neuronal metabolism, 45 synapse formation, transmitter reuptake and potassium buffering (Kimelberg 2010). As 46 these roles for astrocytes became accepted amongst the scientific community newer, more 47 active roles for astrocytes have been proposed, involving astrocytic release of ATP, 48 glutamate and D-serine (Zhang and Haydon 2005). While there have been some detailed 49 studies of “gliotransmitter” release from astrocytes in culture, there is little data supporting 50 gliotransmission in intact brain tissue. 51 52 Henneberger et al (2010) test the hypothesis that astrocytic release of D-serine is 53 essential for generation of long-term synaptic potentiation (LTP) at the Schaffer collateral 54 (SC) – CA1 pyramidal cell (CA1) synapse of the hippocampus (Figure 1, left panel). This is a 55 reasonable hypothesis based on three previously established findings. First, D-serine is an 56 agonist of the glycine-binding site of N-methyl D-aspartate receptors (NMDARs) (Mothet 57 2000). Second, astrocytes in the CA1 are known to contain D-serine in their cytoplasm 58 (Schell 1997). Third, astrocytes in culture are known to release D-serine in a calcium59 dependant manner (Mothet 2005). Also, it was recently shown that D-serine is the co60 agonist of NMDARs during LTP in the supraoptic nucleus of the hypothalamus, and strong 61 evidence for astrocytic D-serine release was demonstrated (Panatier et al. 2006). 62 Henneberger et al (2010) aim to clarify the role of astrocytic D-serine release in 63 hippocampal synaptic LTP. 64 65 The use of “calcium-clamp” on individual astrocytes is certainly the strongest aspect 66 of the study by Henneberger et al (2010). Previous studies have used high concentrations of 67 the calcium buffer EGTA (e.g. 10 mM) to suppress calcium transients in astrocytes. 68 However, the use of exogenous calcium buffers can only inhibit rapid calcium transients but 69 likely does not affect slow changes in free calcium due to homeostatic mechanisms 70 (Henneberger et al. 2010). By making whole-cell patches with an internal solution that 71 contained low concentration of calcium and EGTA Henneberger et al (2010) are able to 72 eliminate slow and transient calcium fluctuations throughout individual astrocytes by 73 effectively “clamping” internal calcium at 50 to 80 nM. 74 75 Previous evidence indicates that D-serine release from astrocytes is dependent on 76 increased intracellular calcium (Mothet 2005). Thus, the authors clamp calcium to eliminate 77 all calcium-dependant transmitter release from the patched astrocyte while simultaneously 78 using standard protocols to induce and record LTP at the SC – CA1 synapse of the 79 hippocampus (Figure 1, left panel). Utilizing this approach the authors demonstrate that 80 LTP is induced in the region of an astrocyte that is patched using control internal solution 81 (i.e. little calcium buffering capacity), but not when the astrocyte is calcium-clamped. 82 Interestingly, the effect of LTP inhibition is spatially restricted to the region of the 83 individual calcium-clamped astrocyte. The ability to induce LTP persisted when the pipette 84 monitoring dendritic field potentials is moved away from the calcium clamped astrocyte. 85 This coincides with the observation that individual astrocytes occupy domains that do not 86 overlap with neighboring astrocytes. The reader is left with the impression that each 87 astrocyte regulates synapses within its own local region of influence, with little overlap 88 from neighboring astrocytes. 89 90 The authors next attempt to address the mechanism by which astrocytes facilitate 91 LTP, i.e. whether astrocytes release D-serine that co-agonizes the CA1 NMDARs and leads to 92 LTP induction. To test this the authors first step is logical; rescue the loss of LTP with 93 exogenous D-serine. Exogenous D-serine does rescue LTP near the calcium-clamped 94 astrocyte, suggesting a lack of glycine-binding site activation blocks LTP induction during 95 calcium-clamp experiments. The next step, in this reader’s mind, is to demonstrate that D96 serine is the co-agonist involved in LTP inductions. In my opinion, this is where the authors’ 97 experimental approach becomes problematic. 98 99 In an effort to suppress gliotransmission the authors use Fluoroacetate because it is 100 considered a selective “gliotoxin”, and in so doing are able to block LTP. However, 101 fluoroacetate is selective to astrocytes only because they take it up across their membranes 102 more readily than their neuronal counterparts (Hassel et al. 2002). Possible effects on 103 neurons aside, the effect of fluoroacetate on astrocytes will not be D-serine-specific. 104 Fluoroacetate is a metabolic poison, and will certainly cause inhibition of membrane 105 transport processes, most notably glutamate. Experiments blocking astrocytic glutamate 106 transport may clarify the effect seen with fluoroacetate. As discussed below, relatively little 107 is known about gliotransmitter release in vivo, so the authors use the most astrocyte108 specific inhibitor available, however, with so many possible non-specific effects it is difficult 109 to interpret the results of these experiments. 110 111 In more compelling experiments Henneberger et al (2010) use HOAsp, an inhibitor 112 of D-serine production, loaded into the intracellular astrocyte pipette to block LTP 113 induction. However, HOAsp only effectively blocks LTP after high frequency stimulation in 114 the presence of APV, an NMDAR antagonist, which they claim expels previously produced D115 serine. In addition, HOAsp is not specific to D-serine production and can affect pyruvate 116 metabolism, hence the general metabolism of the patched astrocyte (Strísovský et al. 2005). 117 Both experiments utilizing fluoroacetate and HOAsp are not as informative as simply 118 inhibiting D-serine activation of NMDARs using the antagonist DAOO, a serine-degrading 119 enzyme, as was done in similar studies in hypothalamus slice preparation and hippocampal 120 cultured neurons (Panatier et al. 2006, Yang et al. 2006). Thus, the authors leave the reader 121 to wonder whether D-serine is the NMDAR co-agonist involved in LTP induction. 122 123 A different test of the hypothesis that calcium-induced release of gliotransmitter by 124 astrocytes is required for LTP at the SC – CA1 synapse of the hippocampus comes from 125 Agulhon et al (2010) who report contradictory results when utilizing a genetic approach to 126 control astrocytic intracellular calcium. This study uses two transgenic mouse models. One 127 line is an astrocyte-specific knockin of the MrgA1 receptor (MrgA1+); an exogenous 128 receptor that acutely increases intracellular calcium, via a Gq GPCR, when activated by its 129 peptide agonist, which is also exogenous to the CNS. Using MrgA1+ mice they demonstrate 130 that increased intracellular calcium in astrocytes does not affect LTP in the hippocampus. 131 While this experiment does not actually contradict the results of Henneberger et al (2010), 132 as they never tested the consequence of increasing astrocytic intracellular calcium, it 133 certainly is not very supportive of an astrocytic role in LTP induction. 134 135 The second mouse line utilized by Agulhon et al (2010) is a knockout of inositol 136 triphosphate receptor 2 (IP3R2). IP3R2 is the receptor thought to be responsible for 137 increasing intracellular calcium in astrocytes, whereas IP3R1 and IP3R3 are expressed in 138 neurons. The authors claim total elimination of astrocyte calcium transients, both 139 spontaneous and those induced by synaptic activation. However, unlike the results obtained 140 using pipette solution to calcium-clamp astrocytes (Henneberger et al. 2010), the IP3R2 141 knockout mice demonstrate both shortand long-term potentiation that is identical in 142 amplitude to the wild type population. 143 144 When faced with contradictory results one often looks for explanation in the 145 different methods used. Both models boast abolishment of astrocytic calcium transients. 146 Henneberger et al (2010) use an internal solution that effectively buffers free calcium 147 fluctuations, whereas the IP3R2 knockout model blocks calcium from entering the 148 cytoplasm from intracellular stores. Could these two approaches have differential ability to 149 block calcium dependant transmitter release? The answer is probably yes. Henneberger et 150 al (2010) are able to clamp calcium, whereas Agulhon et al. (2010) block all calcium 151 fluctuations due to IP3R2 activation. Unfortunately, exactly how increased intracellular 152 calcium causes release of gliotransmitters is not certain (Figure 1, right panel). A recent 153 review (Hamilton and Attwell 2010) points out that activation of PAR1, but not P2Y1, 154 receptors causes transmitter release in astrocytes, although both result in increased 155 intracellular calcium. Equally perplexing, the calcium influx generated by activation of 156 MrgA1 in astrocytes of MrgA1+ mice (described above) does not mediate glutamate157 induced inward currents in local neurons, while uncaging IP3 to increase intracellular 158 calcium in the same mice did result in an increased frequency of neuronal mEPSCs. 159 160 It would appear that simply increasing intracellular calcium is not sufficient for 161 gliotransmitter release. If calcium is truly required for transmitter release, then it may need 162 to occur in specific nanodomains, where it is difficult to experimentally manipulate calcium 163 fluctuations. In this case, knocking out IP3R2 or clamping intracellular calcium may block164 generalized astrocytic calcium oscillations, but may not effectively control calcium in165 certain nanodomains. Calcium entry in these locations may occur via mechanisms other166 than the IP3 pathway, including TRP or voltage gated calcium channels. Or perhaps the167 knockout mice have elevated levels of D-serine present in the synaptic cleft, due to168 membrane mechanisms altered by lack of IP3R2 function. Yet, it is also possible that169 buffering intracellular calcium could have other unanticipated effects, besides eliminating170 calcium fluctuations. However, the latter possibility still strongly implies individual171 astrocytes play a role in the synaptic plasticity of the hippocampus.172173The past two decades have seen a dramatic advance in the field of astrocyte174 physiology, with evidence of astrocytes playing an active role appearing in many facets of175 synaptic function. The present study suggests a new role for astrocytes in memory176 processing. Indeed, altering astrocyte physiology affects LTP induction at the SC – CA1177 synapse of the hippocampus, but because of lack of evidence that D-serine is required for178 LTP induction and contradictory results when using a transgenic approach (Agulhon et al.179 2010) exactly how this occurs remains unanswered. To further understand how astrocytes180 actively regulate synaptic transmission a few questions must be addressed (Figure 1, right181 panel). Is D-serine the co-transmitter involved in LTP? How does elevated astrocytic182 intracellular calcium lead to transmitter release? Is calcium increase necessary? If so, what183 proteins does it act on to cause transmitter release, and what is their cellular location? It is184 important that the mechanism(s) of gliotransmitter release are reconciled in order to form185 a more complete understanding of how astrocytes actively participate in hippocampal LTP.186187188 REFERENCES189190Agulhon C, Fiacco TA, McCarthy KD. Hippocampal shortand long-term plasticity are not191 modulated by astrocyte Ca2+ signaling. Science. 327: 1250-4, 2010.192193Hamilton NB, Attwell D. Do astrocytes really exocytose neurotransmitters? Nat Rev194 Neurosci. 11: 227-38, 2010.195196Hassel B, Paulsen RE, Johnsen A and Fonnum F. Selective inhibition of glial cell197 metabolism in vivo by fluorocitrate. Brain Research. 576: 120-124, 1992.198199Henneberger C, Papouin T, Oliet SH, Rusakov DA. Long-term potentiation depends on200 release of D-serine from astrocytes. Nature. 463: 232-6, 2010.201202Kimelberg HK. Functions of mature mammalian astrocytes: a current view. Neuroscientist.203 16: 79-106, 2010.204205Malenka RC and Nicoll RA. Long-Term Potentiation--A Decade of Progress? Science. 285:206 1870-4, 1999.207208Mothet JP, Parent AT, Wolosker H, Brady RO Jr, Linden DJ, Ferris CD, Rogawski MA,209Snyder SH. D-serine is an endogenous ligand for the glycine site of the N-methyl-D-210 aspartate receptor. Proc Natl Acad Sci U S A. 97: 4926-31, 2000.211212Mothet JP, Pollegioni L, Ouanounou G, Martineau M, Fossier P, Baux G. Glutamate213 receptor activation triggers a calcium-dependent and SNARE protein-dependent release of214 the gliotransmitter D-serine. Proc Natl Acad Sci U S A. 102: 5606-11, 2005.215216Panatier A, Theodosis DT, Mothet JP, Touquet B, Pollegioni L, Poulain DA, Oliet SH.217 Glia-derived D-serine controls NMDA receptor activity and synaptic memory.218 Cell. 125: 775-84, 2006.219220Schell MJ, Brady RO Jr, Molliver ME, Snyder SH. D-serine as a neuromodulator: regional221 and developmental localizations in rat brain glia resemble NMDA receptors. J Neurosci. 17:222 1604-15, 1997.223224Somjen GG. Nervenkitt: notes on the history of the concept of neuroglia. Glia. 1: 2-9, 1988.225226Strísovský K, Jirásková J, Mikulová A, Rulísek L, Konvalinka J. Dual substrate and227 reaction specificity in mouse serine racemase: identification of high-affinity dicarboxylate228 substrate and inhibitors and analysis of the beta-eliminase activity. Biochemistry. 44:229 13091-100, 2005.230231Yang Y, Ge W, Chen Y, Zhang Z, Shen W, Wu C, Poo M, and Duan S. Contribution of232 astrocytes to hippocampal long-term potentiation through release of D-serine. Proc Natl233Acad Sci U S A. 100: 15194-9, 2003.234235Zhang Q, Haydon PG. Roles for gliotransmission in the nervous system. J Neural Transm.236 112: 121-5, 2005.237238 FIGURE 1 LEGEND239240Model of astrocyte regulation of long-term synaptic potentiation (LTP) in the241hippocampus. The left panel depicts the experimental design utilized by Henneberger et al242 (2010) to clamp astrocyte intracellular calcium and record dendritic field potentials at the243 Schaffer collateral (SC) – CA1 pyramidal cell (CA1) synapse in the hippocampus. The right244 panel depicts a model for LTP. The currently accepted model of this hippocampal synaptic245 LTP involves glutamate being released from the presynaptic neuron and diffusing across the246 synaptic cleft to activate postsynaptic receptors. N-methyl D-aspartate (NMDA) receptor247 activation is necessary for LTP induction and requires both glutamate binding and248 activation of the glycine-binding site. Classically, the latter is thought to be accomplished by249 tonic levels of glycine but Henneberger et al (2010) propose that D-serine, released from250 astrocytes, activates the glycine-binding site. In this model glutamate activates mGluRs on251 astrocytes to increase intracellular calcium and cause D-serine release, which in turn binds252 the glycine-binding site of the NMDA receptor. By clamping internal calcium fluctuations of253 an astrocyte via a patch pipette Henneberger et al (2010) demonstrate elimination of LTP254 that supports this hypothesis, however another study (Agulhon et al. 2010) that uses a255 transgenic approach to block intracellular calcium transients finds opposite results. In order256 to more completely understand the role of astrocytes in hippocampal synaptic LTP three257 questions must be answer in vivo. What mechanism does increased intracellular calcium act258 on to promote D-serine release? How is D-serine released by astrocytes? And finally, is D-259 serine the endogenous co-transmitter involved in synaptic LTP of the hippocampus?260261 ??[Ca]imGluRD-serine?PresynapticNeuronPostsynapticNeuron NMDARsGlutamateGlycine-binding site AstrocytePresynapticNeuronPostsynapticNeuron AstrocyteCa2+ lampWhole-cellPatch RecordingField-potentialRecordingNormalHFSCa2+ clampHFS