In vivo recording of the circadian calcium rhythm in Prokineticin 2 neurons of the suprachiasmatic nucleus

Prokineticin 2 (Prok2) is a small protein expressed in a subpopulation of neurons in the suprachiasmatic nucleus (SCN), the primary circadian pacemaker in mammals. Prok2 has been implicated as a candidate output molecule from the SCN to control multiple circadian rhythms. Genetic manipulation specific to Prok2-producing neurons would be a powerful approach to understanding their function. Here, we report the generation of Prok2-tTA knock-in mice expressing the tetracycline transactivator (tTA) specifically in Prok2 neurons and an application of these mice to in vivo recording of Ca2+ rhythms in these neurons. First, the specific and efficient expression of tTA in Prok2 neurons was verified by crossing the mice with EGFP reporter mice. Prok2-tTA mice were then used to express a fluorescent Ca2+ sensor protein to record the circadian Ca2+ rhythm in SCN Prok2 neurons in vivo. Ca2+ in these cells showed clear circadian rhythms in both light–dark and constant dark conditions, with their peaks around midday. Notably, the hours of high Ca2+ nearly coincided with the rest period of the behavioral rhythm. These observations fit well with the predicted function of Prok2 neurons as a candidate output pathway of the SCN by suppressing locomotor activity during both daytime and subjective daytime.

robust circadian rhythm in the mouse SCN, peaking around zeitgeber time (ZT)/circadian time (CT) 4 27 .Its expression is also dependent on neuronal activity 29 .In addition, the receptor for Prok2 (Prokr2) is abundantly expressed in primary target nuclei of the SCN output pathway 27,30 .Furthermore, central administration of Prok2 suppresses nocturnal locomotor activity in rats 27 .Thus, Prok2 is likely released during the (subjective) day and inhibits locomotor activity to generate the circadian behavioral rhythm in nocturnal rodents.Correspondingly, Prok2-and Prokr2-deficient mice have attenuated circadian rhythms of behavior and other physiological parameters without affecting TTFL in their SCN 31,32 .
To elucidate how Prok2 neurons regulate circadian rhythms, genetic manipulations specific to these neurons for the recording and artificial manipulation of their activity would be a powerful approach.Here, we have established a Prok2-tTA knock-in mouse line that allows genetic manipulation of Prok2 neurons via the Tet system.The Tet system is a widely used genetic tool in which the tetracycline transactivator (tTA) binds the tetracycline-responsive element (TRE) and activates the gene downstream of the TRE 33,34 .In addition, we used this mouse tool to record the circadian rhythm of intracellular Ca 2+ concentration ([Ca 2+ ] i ) in SCN Prok2 neurons in vivo and revealed its temporal relationship to the behavioral rhythm.In neonatal explants, SCN cells show robust daily [Ca 2+ ] i rhythms that depend on the TTFL, and the TTFL is also regulated by cytosolic Ca 2+35,36 .Moreover, differential [Ca 2+ ] i rhythms in different SCN neuron subtypes have been reported both in slices and in vivo 5,19,[37][38][39] .Therefore, recording [Ca 2+ ] i rhythm from Prok2 neurons would be a first step towards understanding their precise role in circadian time-keeping.

Generation of Prok2-tTA knock-in mice
To elucidate the function of Prok2 neurons, genetic manipulations specific to these neurons would be useful.For this purpose, we generated knock-in mice expressing tTA2 40 specifically in Prok2 neurons.To do so, we employed the CRISPR/Cas9-mediated homologous recombination to target the Prok2 gene of the mouse genome and inserted a tTA2-WPRE-polyA cassette near the Prok2 gene start codon in its exon 1 in C57BL/6 J mice (Prok2-tTA) (Fig. 1a).To localize tTA2 activity, we crossed Prok2-tTA mice with Actb-tetO-EGFP reporter mice, which express EGFP in the presence of tTA 41,42 .EGFP + cells were observed in several brain regions that were reported to express Prok2 28,30,43 , including the olfactory bulb, nucleus accumbens, lateral septum, islands of Calleja, medial preoptic area, SCN, paraventricular hypothalamic nucleus, arcuate nucleus, and the Edinger-Westphal nucleus (Fig. 1b).In addition, we found EGFP + cells in the ventromedial hypothalamic nucleus and the pedunculopontine tegmental nucleus.Furthermore, EGFP + cells were distributed sparsely in the cerebral cortex, striatum, and hippocampus.EGFP + cells in the regions where Prok2 expression has not been reported might result simply from ectopic expression or possibly from better sensitivity due to the use of the Tet system and WPRE (woodchuck hepatitis virus posttranscriptional regulatory element).

Heterozygous Prok2-tTA mice show normal circadian behavior rhythm
The Prok2-tTA allele should be equivalent to a Prok2 knockout allele because the Prok2 coding sequence was interrupted by a tTA2-WPRE-polyA sequence.In addition, previous studies reported attenuated circadian rhythms in Prok2-and Prokr2-deficient mice.Therefore, we tried to obtain homozygous Prok2-tTA mice by intercrossing heterozygous mice.However, no homozygous mice grew up to weaning, whereas 5 wildtype and 18 heterozygous littermates did, suggesting postnatal lethality of homozygous mice.This result was consistent with previous observations of Prok2-and Prokr2-deficient mice that their postnatal survival rates drastically dropped after backcrossing to C57BL/6 for 6 ~ 7 generations 31,32 .
To confirm that the Prok2-tTA allele causes no overt effect on circadian behavior, we next recorded the daily rhythms of spontaneous locomotor activity of heterozygous Prok2-tTA mice (Fig. S1).These mice demonstrated clear circadian behavioral rhythm in both light-dark (LD) and constant dark (DD) conditions, comparable to those reported for control mice in similar genetic backgrounds and recording conditions 15,16,39 .

In vivo recording of the circadian Ca 2+ rhythm in SCN Prok2 neurons
Prok2 has been implicated as an output molecule of the central circadian clock of the SCN, which is released during the (subjective) day to suppress locomotor activity in mice 27 .For Prok2 to play this role, Prok2 neurons should also be active during the (subjective) day.To directly test this possibility, we next recorded the [Ca 2+ ] i rhythm in SCN Prok2 neurons in vivo by fiber photometry 39,48 while monitoring the locomotor activity rhythm.To do this, the fluorescent Ca 2+ indicator jGCaMP7s 49 was expressed specifically in these neurons by focal injection of a tTA-dependent AAV vector (Fig. 3a).www.nature.com/scientificreports/ In LD, when plotted on the actogram, a daily [Ca 2+ ] i rhythm was observed in SCN Prok2 neurons, higher during the light phase and lower during the dark phase (Fig. 3b,c).Such [Ca 2+ ] i rhythms persisted in DD, confirming that the observed rhythms were truly circadian and not driven by the external LD cycle.Importantly, our fiber photometry method did not detect a significant circadian oscillation of fluorescence when control EGFP was expressed in the SCN Prok2 neurons (Fig. 3d-f).Thus, our measurements of jGCaMP7s fluorescence were likely to reflect [Ca 2+ ] i in Prok2 neurons correctly.
For quantitative analyses, we defined the peak phase and period of the [Ca 2+ ] i rhythms, as well as the onset and offset of the hours of high [Ca 2+ ] i (Fig. 4a top) 19 .For this purpose, the data were detrended to remove the gradual signal decrease over the recording days and then smoothened to remove fast signal fluctuations within hours.Daily Ca 2+ onset and offset were defined as the times when the value crossed 0 upward (i.e., [Ca 2+ ] i rising) and downward (i.e., [Ca 2+ ] i falling), respectively.The midpoints of Ca 2+ onset and offset were defined as the peak phases, and the intervals between two adjacent peaks were defined as the periods.We considered these definitions are more appropriate than other methods, such as sine curve fitting, because the waveforms of [Ca 2+ ] i rhythms appeared to deviate from the typical sinusoidal curve and to be noisy with multiple small peaks within the hours of high [Ca 2+ ] i (Fig. 3c).
The peak of the [Ca 2+ ] i rhythm in SCN Prok2 neurons was in the middle of the day in LD (ZT5.8 ± 0.0) and the subjective day in DD (CT5.9 ± 0.1) (Fig. 4b).Its period in LD (24.0 ± 0.1 h) was equal to the 24 h LD cycle, and that in DD (23.9 ± 0.0 h) was comparable to the behavioral free-running period (23.8 ± 0.1 h).Intriguingly, the daily onset and offset of Ca 2+ almost coincided with the offset and onset of the behavioral activity period, respectively (Fig. 4a bottom and Fig. S2).

Discussion
In this study, we generated Prok2-tTA knock-in mice in which the tTA2-coding sequence was introduced into the endogenous Prok2 locus.The expression of tTA was primarily restricted to the brain regions reported to express Prok2 mRNA.As expected, the SCN contained many tTA-positive cells in both the shell and the core.Furthermore, the expression of tTA was highly specific for Prok2 neurons in the SCN.Therefore, in combination with transgenic mice or viral vectors with TRE-mediated transgene expression, Prok2-tTA mice can express any protein specifically in Prok2 neurons.In addition, many Cre driver mice specific for a particular type of SCN neurons are currently available, such as Avp-ires-Cre 50 , Avp-Cre 16 , Vip-ires-Cre 51 , Nms-Cre 52 , Grp-Cre 53 , Drd1a-Cre 54 , and Vipr2-Cre 38 .Therefore, by crossing one of them with Prok2-tTA mice, the double transgenic mice would allow us to simultaneously apply different genetic manipulations to Prok2 and another type of SCN neurons via the Cre/loxP and Tet systems.Such a dual-targeting strategy would allow us to directly study the interactions between Prok2 neurons and other SCN neurons, which would be a powerful tool for studying the SCN, a small but complex neuronal network composed of many types of neurons, including Prok2 neurons 1 .
A previous study reported co-localization of Prok2 and Avp or Vip mRNA in the rat SCN: approximately 50.2% of Avp + cells expressed Prok2 whereas 29.4% of Prok2 + cells expressed Avp; approximately 41.6% of Vip + cells expressed Prok2 whereas 21.8% of Prok2 + cells expressed Vip 26 .We also observed a similar partial overlap of www.nature.com/scientificreports/Prok2-tTA expression with AVP or VIP peptide.The slight difference in the proportions of double-positive cells may be due to differences in species and methods used to detect expression (i.e., in situ hybridization vs. immunostaining).Although we pretreated the mice with colchicine to reduce neuropeptide transport to the nerve terminals, thereby facilitating cell type identification, we sometimes encountered difficulties distinguishing immunoreactive cell bodies from nerve terminals, which could result in somewhat ambiguous cell counts.Intriguingly, Prok2 neurons were predominantly distributed in the central part of the coronal sections of the middle SCN, surrounded ventrally by VIP neurons and dorsally, medially, and laterally by AVP neurons.Since AVP and VIP neurons have been suggested to play different roles in the circadian pacemaking of the SCN network, it would be interesting to investigate in the future whether Prok2 also has different functions between Prok2 + /AVP + , Prok2 + /VIP + , and Prok2 + /AVP − /VIP − neurons.Using Prok2-tTA mice, we successfully recorded the [Ca 2+ ] i rhythm in SCN Prok2 neurons in vivo.Its peak phase was around the midday (CT5.9 ± 0.1) and almost the same as that of VIP neurons (CT5.6 ± 0.2), but later than that of AVP neurons (CT3.2 ± 0.7) 19 .It has been reported that the rhythm of Prok2 mRNA expression in the SCN peaks around CT4 26,27 .Therefore, the [Ca 2+ ] i rhythm may be slightly delayed compared to the mRNA rhythm.Because protein synthesis and maturation often lag minutes to hours behind mRNA transcription, the Prok2-neuronal [Ca 2+ ] i rhythm seems temporally organized according to Prok2 expression.Notably, the rise and fall of Prok2-neuronal [Ca 2+ ] i mostly delineated the rest period of the behavioral rhythm.These observations fit well with the function of Prok2 as a candidate SCN output molecule released during the (subjective) day to suppress locomotor activity 27 .

Ethics statements
All experiments were performed in accordance with the Japanese Neuroscience Society and Kanazawa University guidelines for laboratory animal care and use.Experimental protocols were approved by the Animal Care and Use Committee and Gene Recombination Experiment Safety Committee of Kanazawa University and Tokyo Medical and Dental University.The study was carried out in compliance with the ARRIVE guidelines.

Animals
To generate Prok2-tTA mice, we inserted a tTA2-WPRE-polyA cassette 25 bp downstream to the start codon of Prok2 gene in its first exon by the CRISPR/Cas9-mediated targeting strategy as described previously 41 (Fig. 1a).The donor DNA was synthesized, containing tTA2 cDNA 40 , woodchuck hepatitis virus posttranscriptional regulatory element (WPRE), polyA signal derived from human growth hormone gene, and 1.5 kb sequences of the mouse Prok2 gene (NCBI Gene: 50501) 5' and 3' to the insertion site.Endogenous initiation codon was inactivated by an A to T mutation.One-cell stage zygotes were obtained by mating C57BL6/J males and females (CLEA Japan).Prok2-crRNA (5′-CAG CAG AAG UAG CAG UAG CGguuuuagagcuaugcuguuuug-3′) and tracrRNA (5′-AAA CAG CAU AGC AAG UUA AAA UAA GGC UAG UCC GUU AUC AAC UUG AAA AAG UGG CAC CGA GUC GGU GCU-3′) were chemically synthesized and purified by high performance liquid chromatography (Fasmac).A mixture of recombinant Cas9 proteins (NEB), Prok2-crRNA, tracrRNA, and pProk2-tTA2-WPRE-polyA targeting vector were injected into pronuclei of one-cell stage zygotes using a micromanipulator/microscope (Leica) and injector (Eppendorf).Embryos were then washed and cultured for over an hour in KSOM medium (ARK resource) and transferred into pseudopregnant ICR female mice (CLEA Japan).Presence of the knock-in allele was verified by PCR using tail genomic DNA.One F0 founder mouse was obtained, backcrossed at least twice with C57BL/6 J, and then used for the experiments in heterozygous condition.To evaluate the specific expression of tTA2, Prok2-tTA mice were crossed to Actb-tetO-EGFP reporter mice 41,42 .All mice were maintained under a strict 12 h light/12 h dark cycle in a temperature-and humidity-controlled room and fed ad libitum.

Histological study
Prok2-tTA; Actb-tetO-EGFP mice were sacrificed around ZT4 ~ 5 by transcardial perfusion of PBS followed by 4% paraformaldehyde fixative.Serial coronal brain slices (30 µm thick) were prepared using a cryostat (CM1860, Leica) and collected in four series.One of these was further subjected to in situ hybridization chain reaction (HCR) or immunostaining.
In situ HCR for Prok2 mRNA was performed using in HCR v3.0 55 (Molecular Instruments).Prior to prehybridization, sections were pretreated as previously described for in situ hybridization, with proteinase K treatment replaced by 1% sodium borohydride to avoid digestion of the EGFP protein 56 .Pre-hybridization and subsequent procedures were performed essentially according to the protocol for fixed frozen tissue sections provided by Molecular Instruments (http:// molec ulari nstru ments.org), except that the sections were floated in solution in a 2 mL microcentrifuge tube.The probe set for Prok2 was designed and synthesized by Molecular Instruments (lot number: PRJ347) and used in combination with HCR Amplifier B1 labeled with Alexa Fluor594 (Molecular Instruments).EGFP expression was detected by its native fluorescence.

Figure 4 .
Figure 4.The phase relationship of the [Ca 2+ ] i rhythm in SCN Prok2 neurons and behavior rhythm.(a) Plots of locomotor activity onset (black), activity offset (gray), GCaMP onset (green), GCaMP offset (light green), and GCaMP peak (magenta) of mean ± SEM (left column) and individual mice data (right column).Identical marker shapes indicate data from the same animal.(b) Peak phases of the GCaMP fluorescence rhythm in LD (top) or DD (bottom) were shown as Rayleigh plots.Individual dots indicate the peak phases of each mouse.Values are mean ± SEM. n = 6.