Distinct target-specific mechanisms homeostatically stabilize transmission at pre-and post-synaptic compartments

Neurons must establish and stabilize connections made with diverse targets, each with distinct demands and functional characteristics. At Drosophila neuromuscular junctions, synaptic strength remains stable in a manipulation that simultaneously induces hypo-innervation on one target and hyper-innervation on the other. However, the expression mechanisms that achieve this exquisite target-specific homeostatic control remain enigmatic. Here, we identify the distinct target-specific homeostatic expression mechanisms. On the hypo-innervated target, an increase in postsynaptic glutamate receptor (GluR) abundance is sufficient to compensate for reduced innervation, without any apparent presynaptic adaptations. In contrast, a target-specific reduction in presynaptic neurotransmitter release probability is reflected by a decrease in active zone components restricted to terminals of hyper-innervated targets. Finally, loss of postsynaptic GluRs on one target induces a compartmentalized, homeostatic enhancement of presynaptic neurotransmitter release called presynaptic homeostatic potentiation that can be precisely balanced with the adaptations required for both hypo- and hyper-innervation to maintain stable synaptic strength. Thus, distinct anterograde and retrograde signaling systems operate at pre- and post-synaptic compartments to enable target-specific, homeostatic control of neurotransmission.


83
Finally, at the Drosophila NMJ, presynaptic homeostatic plasticity can be expressed at a subset 84 of terminals on a single motor neuron depending on GluR functionality at particular targets (Li et 85 al., 2018a), demonstrating that this form of homeostatic plasticity is target-specific and strongly 86 suggesting it is also synapse-specific. Together, these studies and others have demonstrated 87 that the physiologic, metabolic, and/or structural properties at terminals of a single neuron can 88 be selectively modulated according to the identity and needs of the targets and synapses they 89 innervate. However, the nature of the trans-synaptic dialogue and the molecular mechanisms 90 that achieve target-specific plasticity are not well understood.

91
A seminal study published over 20 years ago found that distinct target-specific 92 modulations in synaptic activity maintain stable neurotransmission following biased innervation 93 at terminals of motor neurons at the Drosophila NMJ (Davis and Goodman, 1998). In this 94 manipulation, biased innervation is achieved by overexpression of the trans-synaptic cell 95 adhesion factor fasciculin II (fasII) on one of the two muscle targets innervated by motor 96 neurons (Davis and Goodman, 1998). This leads to hyper-innervation of the target 97 overexpressing fasII at the expense of the adjacent target, which is hypo-innervated.

98
Remarkably, synaptic strength, as assessed by electrophysiological recordings, was maintained 99 at levels similar in amplitude to normally innervated NMJ targets. Since this pioneering study, 100 however, the molecular and cellular expression mechanisms that achieve this target-specific 101 homeostatic modulation have remained enigmatic.

102
We have investigated how terminals of an individual neuron adapt to simultaneous hypo-103 and hyper-innervation to maintain stable synaptic strength on two adjacent targets. Our analysis 5 reveals that a novel homeostatic signaling system operates in the hypo-innervated target to 105 precisely enhance the abundance of postsynaptic GluRs, offsetting reduced presynaptic 106 neurotransmitter release and stabilizing synaptic strength. In contrast, no apparent adaptations 107 are observed in the hyper-innervated target. Rather, presynaptic release probability is 108 homeostatically reduced, accompanied by a target-specific decrease in the abundance and 109 density of active zone components. Finally, we find that presynaptic homeostatic potentiation  Table S2). To bias innervation on 124 these targets, we used the H94-Gal4 driver to drive expression of the cell adhesion molecule 125 fasciculin II (fasII) early in development selectively on muscle 6 [(M6>FasII; (Davis and 126 Goodman, 1998)]. Immunostaining of M6>FasII NMJs confirmed biased innervation with ~150% 127 of boutons above controls on muscle 6 (hyper-innervated), and a parallel reduction of ~50% in 128 boutons on muscle 7 (hypo-innervated) (Fig. 1A-E), consistent with the previous study (Davis 129 and Goodman, 1998). However, despite these opposing changes in bouton numbers, 6 electrophysiological recordings of M6>FasII found that synaptic strength, measured by the 131 excitatory postsynaptic potential (EPSP) amplitude, was similar on both targets and unchanged 132 from their respective controls (Fig. 1C,D,E). This implies target-specific mechanisms modulate 133 neurotransmission on hypo-and hyper-innervated terminals to maintain stable NMJ strength.

134
To gain insight into how EPSP amplitudes remain similar to baseline values at NMJs 135 with biased innervation, we next examined miniature neurotransmission. On hypo-innervated 136 muscle 7, mEPSP amplitudes were significantly increased by ~40% compared to baseline 137 values (Fig. 1C,D), as previously observed (Davis and Goodman, 1998). Quantal content (QC) 138 was thus decreased by ~40%, a value similar in magnitude to the reduction in bouton number 139 ( Fig. 1D). In contrast, mEPSP amplitude was not significantly different on the hyper-innervated 140 muscle 6 NMJ compared to baseline (Fig. 1C,E), with no apparent change in quantal content 141 ( Fig. 1E), as previously observed (Davis and Goodman, 1998). Finally, analysis of quantal 142 content normalized per bouton on muscle 6 NMJs revealed an ~30% reduction (Fig. 1E), 143 suggesting a target-specific, homeostatic decrease in presynaptic neurotransmitter release.

144
Together, this data indicates that distinct target-specific mechanisms operate to stabilize 145 neurotransmission at hypo-vs. hyper-innervated NMJs. It was previously reported that at hypo-innervated NMJs following M6>FasII, levels of the 150 postsynaptic GluR subunit GluRIIA were increased (Goel and Dickman, 2018 Goel et al., 2019a). Second, multivesicular release has been observed in some systems 7 (Rudolph et al., 2015) and was raised as a possibility in the original study to potentially explain 157 the increased quantal size (Davis and Goodman, 1998), although there remains no evidence for 158 multi-vesicular release at the fly NMJ. In addition to these two presynaptic mechanisms, other 159 postsynaptic mechanisms are possible that could explain the increased mEPSP amplitude on

285
PhTx application also induced robust PHP at muscle 7 NMJs in M6>FasII+ GluRIIA RNAi , with a 12 significant reduction in mEPSP amplitude but normal EPSP amplitude due to enhanced quantal 287 content (Fig. S3A,C,E). These results demonstrate that presynaptic release sites at terminals of 288 the same neuron can be selectively modulated with exquisite target specificity to compensate 289 for GluR loss and can be superimposed with the homeostatic plasticity induced by biased

329
The induction mechanism involved in how reduced innervation is sensed to ultimately         Table S1.

416
To quantify sum puncta intensity, the total fluorescence intensity signal of individual 417 puncta were calculated without regard to area as described (Goel et al., 2019a). For each 418 particular sample set, thresholds were optimized to capture the dynamic range of intensity levels

451
Recordings were rejected if the V rest was more depolarized than -60 mV, if the R in was less than

456
The readily releasable pool (RRP) size was estimated by analyzing cumulative EPSC 457 amplitudes while recording using a two-electrode voltage clamp (TEVC) configuration as    Table S2.

563
Quantification of BRP and CAC puncta number (D) and density (E) on muscle 6 in M6>FasII 564 normalized as a percentage of wild type muscle 6 values reveals a small but significant increase 565 in BRP puncta number, while BRP and CAC puncta density is significantly reduced on muscle 6 566 in M6>FasII. Quantification of BRP and CAC intensity (F) shows a significant reduction on 567 muscle 6 in M6>FasII, while the total fluorescence intensity of all BRP and CAC puncta 568 summed across the entire muscle 6 NMJ (G) is unchanged compared to wild type muscle 6.