Regulation of PDF receptor signaling controlling daily locomotor rhythms in Drosophila

Each day and in conjunction with ambient daylight conditions, neuropeptide PDF regulates the phase and amplitude of locomotor activity rhythms in Drosophila through its receptor, PDFR, a Family B G protein-coupled receptor (GPCR). We studied the in vivo process by which PDFR signaling turns off, by converting as many as half of the 28 potential sites of phosphorylation in its C terminal tail to a non-phosphorylatable residue (alanine). We report that many such sites are conserved evolutionarily, and their conversion creates a specific behavioral syndrome opposite to loss-of-function phenotypes previously described for pdfr. That syndrome includes increases in the amplitudes of both Morning and Evening behavioral peaks, as well as multi-hour delays of the Evening phase. The precise behavioral effects were dependent on day-length, and most effects mapped to conversion of only a few, specific serine residues near the very end of the protein and specific to its A isoform. Behavioral phase delays of the Evening activity under entraining conditions predicted the phase of activity cycles under constant darkness. The behavioral phenotypes produced by the most severe PDFR variant were ligand-dependent in vivo, and not a consequence of changes to their pharmacological properties, nor of changes in their surface expression, as measured in vitro. The mechanisms underlying termination of PDFR signaling are complex, subject to regulation that is modified by season, and central to a better understanding of the peptidergic modulation of behavior.


INTRODUCTION
The properties of the persistent circadian rhythmic behavior are provided in Table 1   was not obviously or systematically effected (Suppl. Fig. 2L), while that of the Evening peak was 211 significantly delayed by several (Suppl. Fig. 2M). We note two features -first the progressive addition 212 of more Ala substitutions in the series 1-4A, 1-5A, 1-6A (Suppl. Fig. 2I-K) tended to produce large and 213 delayed Evening peaks in the time period ZT11.5-12.5 (3-4 h after lights-off), with the variant 1-6A 214 producing the most pronounced delay. The delayed Evening peak activity at ZT12 sometimes appeared 215 at the expense of, the normal Evening peak activity that occurred prior to Lights-off (e.g., Suppl. Fig.2E-1 216 and F-1), although when averaged across all days in LD, that effect for specific variants was not 217 significant (Suppl. Fig. 3E-2 and F-2). Second, a delayed Evening phase was also produced by two of the 218 Simple PDFR variants (Suppl. Fig. 2M), the 6A and 7A variants, each of which contain only a pair of Ser-219 to-Ala substitutions. The amplitude of the Morning peak (ZT18-20) was increased modestly by variants 220 2-3A, 7A and 1-5A (Suppl . Fig 3C-1, G-1 and J-1, red arrows) or strongly by 6A (Suppl. Fig. 3F-1, red 221 arrow). The amplitude of the Evening peak was also increased by others in the PDFR variant series 222 (Suppl. Fig. 3, blue arrows), of which the 6A, 7A, 1-5A and 1-6A variants had effect sizes most similar to 223 that of 1-7A (Supp. Fig. 3F-3. G-3. J-3 and K-3). We also note that combining the Single 5A, 6A and 7A 224 variants into a Multiple Variant (5-7A) did not produce the anticipated additive effects on increasing
12:12 conditions, han mutant flies (lacking pdfr function) typically display elevated nocturnal activity, a 232 lack of morning anticipation prior to Lights-on, and a pronounced advance in the peak of the Evening 233 behavior [11]. Although we note that some reports have observed remnants or a full bout of Morning

238
Restoration of pdfr function by the 1-7A PDFR variant under these conditions ( Fig. 3C-1) increased the 239 amplitude of the Morning activity peak but not its phase ( Fig. 3C-2, red arrow) but not its phase ( Fig.   240   3D). 1-7A expression in equinox conditions rescued the advanced phase of the Evening activity peak, 241 but did not further delay it past Lights-OFF (Fig. 3C-4 and 3E), unlike what we observed under short-day 242 conditions. The amplitude of the Evening peak was only modestly elevated following expression of WT 243 or 1-7A PDFRs. The period (tau) of the 1-7A variant was considerably lengthened, but see discussion 244 below about the influence of the GFP fusion. The phase of the major DD rhythm was delayed relative to 245 the evening peak under entraining conditions (Fig. 3H, green line). In sum under equinox-like 246 photoperiods, the PDFR 1-7A variant significantly increased the amplitude of the morning activity peak 247 compared to WT PDFR, but did not differentially affect its phase, or the amplitude or phase of the 248 Evening peak; under subsequent DD, the rhythmic activity displayed a delayed phase starting within the 249 very first cycle.

251
Behavioral effects of the other PDFR Variants under equinox conditions (Suppl. Fig. 4 & 5). We 252 tested each of the nine other PDFR Ala-variants by comparing their effects on locomotor behavior as we 253 had done for the PDFR 1-7A variant. The phases of the Morning and Evening peaks were not obviously 1, red arrows). Relative to WT PDFR expression, the Evening Activity peak amplitude was increased by 257 expression of the 4A, 6A, 7A, 5-7A and 1-5A variants (Suppl. Fig. 5D-3, F-3, G-3, H-3 and J-3, blue 258 arrows). We also note that combining the Single 5A, 6A and 7A variants into a Multiple Variant (5-7A) 259 did not produce the anticipated additive effects on increasing Morning peak amplitude at ~ZT21 (Suppl.

291
Effects of PDFR variants on period and phase of rhythmic activity in DD. In DD, following Short Day 292 conditions, the PDFR variants that produced 3-4 h delays in the evening peak often generated periods ~ 293 1-2 hr longer than the controls and also lowered % arrhythmicity (Table 1). This was especially true for 294 the Multiple Variant series (e.g., PDFR 1-5A, 1-6A and 1-7A (Table 1). However the correlation between 295 delayed evening peaks and a longer tau in DD was not absolute. For example, the 6A, 7A and 5-7A       properties. However, we wished to test this assumption explicitly to define the extent to which a bulky 355 GFP fused to the C Terminal tail might confer some of the behavioral properties we might otherwise 356 ascribe to PDFR C terminal sequences. We therefore compared a 1-7A version of PDFR to a wild type 357 PDFR, both of which lacked GFP fusions. We found that the additional Evening phase delays in short day 358 and long day produced by 1-7A expression occurred despite the absence of a GFP fusion (Supp. despite the absence of a GFP fusion (Suppl. Fig. 8J-3). Finally, the phase of the main DD rhythmic day (Suppl. Fig. 8F, green bar) and was delayed relative to that of the LD Evening activity peak in long 363 days (Suppl. Fig. 8N, green bar). We note however that the longer period produced by the 1-7A variant  Fig. 9, 10 and 11). When 375 noted, they were poorly correlated with behavioral effects (Suppl.

377
PDFR variants tended to display longer surface lifetimes. The 1-4 and 1-7 variants displayed higher basal 378 levels, but no others were different from the WT levels (Suppl. Fig. 12). Following 20 min exposure to 379 PDF, neither the WT not any of the variants displayed a change in surface expression levels (Suppl. opposite to that of pdfr loss of function phenotypes: these would potentially include (for example) a constructs tested slightly advanced the evening peak (Suppl. Fig. 15). Over-expressing GPRK1 cDNAs did 420 not affect morning or evening phase; over-expressing GPRK2 broadened the evening peak. Under DD,

421
krz RNAi flies were uniformly arrhythmic, while GPRK RNAi's were normal or slightly lengthened the 422 circadian period (Suppl.   Table 4) demonstrates such post-translational modifications can occur.

510
However, we caution that these phospho-peptide measurements derive from whole head extracts and

525
This same mis-match of expected behavioral effects for 5-7A was also seen in other photoperiodic 526 conditions (Suppl. Fig. 6M). Likewise, Langlet et al. [36] reported that effects on mutating Serine 527 residues in the carboxy terminus of VPAC1 did not produce additive effects. We propose that depending on which other neighboring residues are also modified. Thus we speculate that CL5 takes on 530 outsize importance in determining desensitization rates for PDFR and so may itself be subject to      positions Ser/ Thr/ and Tyr residues as described in Fig. 1 and Suppl. Fig. 1: white asterisks are the subset of Ser/Thr/ and Tyr residues that were targets of mutational analysis in this study. Mutation of CL6 and CL7 consistently demonstrated the greatest delaying effects on the phases of Evening locomotor peaks; they were also among the sites with greatest influence on the amplitudes of the Morning and Evening peaks. The results suggest their phosphorylation normally will have the greatest effect to slow or terminate PDFR signaling. Other sites are also effective, although to lesser degrees (as indicated by their font sizes), including CL2-3, CL4 and CL5. Primarily, such phosphorylation will decrease the duration of PDFR signaling (red bars) and so reduce (for example) the delay that PDF imposes on the period of neuronal activation displayed by PDFR-responsive pacemaker groups (like the Evening cells and the DN3 [10, 19]). The model also predicts that the effects of phosphorylating CL5 will depend on the phosphorylation status of neighboring sites. In some contexts (e.g., PDFR 1-5A), it will help terminate PDFR signaling, but in others (e.g., PDFR 567A) it may promote the duration or extent of PDFR signaling, perhaps by blocking the effects of phosphorylating CL6 and CL7 (green bar).