Mmp1 processing of the PDF neuropeptide regulates circadian structural plasticity of pacemaker neurons

Abstract

In the Drosophila brain, the neuropeptide PIGMENT DISPERSING FACTOR (PDF) is expressed in the small and large Lateral ventral neurons (LNvs) and regulates circadian locomotor behavior. Interestingly, PDF immunoreactivity at the dorsal terminals changes across the day as synaptic contacts do as a result of a remarkable remodeling of sLNv projections. Despite the relevance of this phenomenon to circuit plasticity and behavior, the underlying mechanisms remain poorly understood. In this work we provide evidence that PDF along with matrix metalloproteinases (Mmp1 and 2) are key in the control of circadian structural remodeling. Adult-specific downregulation of PDF levels per se hampers circadian axonal remodeling, as it does altering Mmp1 or Mmp2 levels within PDF neurons post-developmentally. However, only Mmp1 affects PDF immunoreactivity at the dorsal terminals and exerts a clear effect on overt behavior. In vitro analysis demonstrated that PDF is hydrolyzed by Mmp1, thereby suggesting that Mmp1 could directly terminate its biological activity. These data demonstrate that Mmp1 modulates PDF processing, which leads to daily structural remodeling and circadian behavior.

Figure 1. Mmps are key players of the structural plasticity of PDF neurons.

A. Schematic diagram illustrating the standard protocol and method for the analysis of the complexity of the PDF axonal arbor on confocal images. In all figures “VEH” and “RU” stand for “vehicle”- and “RU486”-containing fly food. B. Adult-specific Mmp overexpression triggers structural phenotypes. Left panel. Representative confocal images of GFP immunoreactivity at the dorsal protocerebrum at the early subjective day (CT2) and early subjective night (CT14) during the 4^th day of constant darkness (DD4). Right panel. Quantitation of total axonal crosses. Wild type flies display circadian structural remodeling of axonal terminals while animals overexpressing Mmp1 or Mmp2 show reduced and constant complexity. Throughout the manuscript all experimental groups include CD8GFP, so the control group “+” refers to a single copy of CD8GFP;pdf-GS. Throughout the manuscript the average ± standard error of the mean is shown. C. Adult-specific Mmp downregulation also affects dorsal axonal branches. Silencing either Mmp1 or Mmp2 abolished circadian structural plasticity leading to a more complex structure clamped at the daytime configuration. Data represents the average of 4 to 5 experiments and a minimum of 27 brains were analyzed per CT/Genotype. Different letters indicate statistically significant differences with a p<0.05 (Two-way ANOVA with a Duncan post-hoc test). For more details, see the Statistics section in Materials and Methods. “+” refers to a single copy of the pdf-GS/CD8GFP;Dcr2 transgenes. In both experiments all the experimental groups include RU to induce expression. Scale: 10 µm.

Figure 2. Mmp1 modulates behavioral rhythmicity.

A. Representative actograms (left panel) and quantitation of percentage of rhythmicity (right panel) from overexpression experiments. Locomotor activity of individual flies was recorded for 4 days under light-dark cycles and then transferred to constant darkness (gray area) for 9 additional days. Overexpression of Mmp1 or Mmp2 with one UAS copy does not affect circadian locomotor activity. “+” in the x axis refers to a single copy of CD8GFP; pdf-GS. NS, non significant. B. Adult-specific Mmp downregulation trigger opposite effects on locomotor rhythmicity. Silencing Mmp1 but not Mmp2 alters normal circadian locomotor activity. “+” in the x axis refers to a single copy of CD8GFP; pdf-GS. Data represents at least 3 independent experiments and a minimum of 32 flies per Genotype/Condition were analyzed. Different letters indicate statistically significant differences with a p<0.05 (Two-way ANOVA with a Duncan post-hoc test). For other controls and measurements of endogenous period see Table S1.

Figure 3. Cell autonomous Mmp1 expression regulates PDF levels.

A. Overexpression experiments Left panel. Representative confocal images of PDF immunoreactivity at the dorsal protocerebrum taken during CT2 and CT14 on DD4. Right panel. PDF levels at the dorsal protocerebrum. Control flies exhibit circadian oscillation of PDF levels, while Mmp1 overexpression reduces PDF to levels lower than those observed at nighttime in controls. In contrast, Mmp2 overexpression leads to intermediate levels. “+” in the x axis refers to a single copy of CD8GFP; pdf-GS. B. Downregulation experiments. Reducing Mmp1 but not Mmp2 levels abolishes circadian oscillations in PDF immunoreactivity to constant daytime levels. “+” in the x axis refers to a single copy of CD8GFP, Dcr2; pdf-GS. Data represents the average (± standard error of the mean) of at least 3 independent experiments and a minimum of 23 flies per Genotype/CT were analyzed. Different letters indicate statistically significant differences with a p<0.05 (Two-way ANOVA with a Duncan post-hoc test). In overexpression experiments logarithmic transformation was applied to fulfill ANOVA requirements. In both experiments all the experimental groups include RU to induce expression. Scale: 10 µm.

Figure 4. PDF defines the axonal remodeling of its own neurons.

A. Quantitation of total axonal crosses from UAS-PDF rescue experiments. Overexpression of PDF rescues the structural plasticity defects caused by Mmp1 overexpression. “+” in the x axis refers to a single copy of CD8GFP; pdf-GS. Data represents the average (± standard error of the mean) between 3–5 independent experiments and a minimum of 21 flies were analyzed per Genotype/CT. B. PDF downregulation prevents circadian axonal remodeling of sLNv terminals and reduces daytime complexity to nighttime levels. “+” in the x axis refers to a single copy of CD8GFP, Dcr2; pdf-GS. Data represents the average (± standard error of the mean) between 3 independent experiments and a minimum of 25 flies were analyzed per Genotype/CT. In both experiments different letters indicate statistical differences with a p<0.05 (Two-way ANOVA with a Duncan post-hoc test) and all the experimental groups include RU to induce expression.

Figure 5. Mmp1 processes the PDF neuropeptide in vitro.

A–D. Reverse-phase HPLC profiles of Mmp1 alone (A), PDF alone (B), PDF+Mmp1 (C) or PDF+Mmp1+Batimastat (D) incubated for 1 h at 37°C. C. Peaks 1 through 4 contained PDF fragments and the peak 5 was full-length PDF as determined by MS/MS shown in Table 1. D. Note the absence of PDF degradation products when Mmp1 was preincubated with the MMP inhibitor Batimastat. Fractions 6 and 7 included PDF 1–19 as identified by MS/MS shown in Table 1. E. Schematic representation of Mmp1 preferred cleavage sites within PDF. Arrows indicate the peptide bonds hydrolyzed by Mmp1 as determined by MS/MS analysis. In bold and italics, P1' residues.

Figure 6. A model for the regulation of circadian axonal remodeling of sLNv neurons.

The bidirectional arrow between electrical activity and Mmp1 suggests a possible coordination of both processes. Mmp1 effects on structural plasticity are dependent on the modulation of PDF levels at the sLNv terminals, via direct proteolysis, while Mmp2 appears to act downstream of the neuropeptide. Electrical activity regulates the overall level of complexity but it is not required to determine the circadian aspect of this remodeling. Given our current understanding Fas2 and EcR could act either upstream or downstream of PDF; however, the well-known Fas2 function points to a more direct modulation of circuit structure. Changes in the size of “PDF” and “Mmp1” molecules illustrate oscillations in abundance along the day.