The circadian clock of mammals and the fruit fly Drosophila melanogaster relies on homologous components. The positive regulators CLOCK and BMAL/CYCLE belong to the basic helix-loop-helix (bHLH) PER-ARNT-SIM (PAS) transcription factors (King et al., 1997; Darlington et al., 1998; Hogenesch et al., 1998; Rutila et al., 1998). PERIOD (PER), which also belongs to the PAS protein family, is a negative regulator shared among vertebrates and insects (Hardin et al., 1990; Zylka et al., 1998).

In D. melanogaster, PER interacts with Drosophila-type TIMELESS protein (dTIM) (despite the general conservation of the molecular mechanisms and genetic components among various vertebrates and insects, some important differences exist. Furthermore, the gene/PROTEIN names vary in the literature as they were historically evolving. Here, we use the prefix m-for the so-called mammalian type and the prefix d-for the Drosophila-type proteins. See the supplementary text for more detailed notes on circadian clock gene terminology).

PER:dTIM dimerization in the cytoplasm is necessary for subsequent nuclear localization of PER and dTIM (Saez and Young, 1996; Meyer et al., 2006). dTIM is an essential component of the fruit fly circadian clock, because tim null mutations result in complete arrhythmicity (Sehgal et al., 1994), whereas missense mutations affect τ (Rothenfluh et al., 2000; Wulbeck et al., 2005), and certain d-tim mutations affect the temperature compensation of the circadian clock (Matsumoto et al., 1999; Singh et al., 2019). Furthermore, dTIM is a key component of the light-mediated synchronization in Drosophila (Hunter-Ensor et al., 1996; Myers et al., 1996; Zeng et al., 1996), which involves light-dependent interaction with Drosophila-type CRYPTOCHROME (dCRY) (Ceriani et al., 1999; Peschel et al., 2009). dCRY serves as a deep brain circadian photoreceptor (Emery et al., 2000) with no impact on the behavioral rhythmicity in constant-dark conditions (DD) at ambient temperature (Stanewsky et al., 1998), although d-cry depletion reduced rhythmicity at 18°C (Dolezelova et al., 2007). Interestingly, d-cry mutations abolish transcriptional oscillations in peripheral clocks, which allowed the identification of this mutant in a luciferase reporter-based screen (Stanewsky et al., 1998). In mice, mammalian-type CRYPTOCHROME (mCRY) is present as two paralogous and closely related proteins, that are essential for (light-independent) clock function, while dimerizing with one of the three mammalian PER proteins (Zylka et al., 1998; Kume et al., 1999; Putker et al., 2021).

An important feature of the negative TTFL is the temporal regulation of subcellular localization of participating proteins. In addition to nuclear localization signals (NLS), some circadian clock proteins also contain nuclear export signals (NES) (Saez and Young, 1996; Vielhaber et al., 2001; Ashmore et al., 2003; Yildiz et al., 2005; Hara et al., 2011; Saez et al., 2011; Jang et al., 2015; Singh et al., 2019; Giesecke et al., 2021). Thus, the resulting nuclear import/export strongly affects the suppression potential of the negative feedback loop and thereby τ. The stability and subcellular localization of the negative complex, such as PER and dTIM, is regulated by posttranslational modifications (Li et al., 2019; Crosby and Partch, 2020), including phosphorylation and dephosphorylation by several kinases and phosphatases (Sathyanarayanan et al., 2004; Leloup 2009; Reischl and Kramer, 2011; Agrawal and Hardin, 2016; Narasimamurthy and Virshup., 2021). One of the most explored circadian clock kinases is DBT which was first identified as a clock component in a Drosophila screen when the short- (DBT^S) and long- (DBT^L) free-running period mutants were identified (Price et al., 1998; Kloss et al., 1998, Figure 8). However, as it turned out later, dbt is also known as discs overgrown, a gene which had been discovered for its role during development (Jursnich et al., 1990; Zilian et al., 1999). Mammalian homologs of DBT are CKIδ/ε, which were shown to be essential for the clock in the hamster, human, and mice (Lowrey et al., 2000; Xu et al., 2005; Meng et al., 2008). The interaction between this kinase and PER is remarkably stable (Kloss et al., 2001; Lee et al., 2001; Aryal et al., 2017). Overexpression of either DBT^L or DBT^S variants in Drosophila resulted in the same τ as was produced by the corresponding alleles of the endogenous gene (Muskus et al., 2007), whereas in vitro studies using non-physiological substrates implied, surprisingly, that both mutants have reduced kinase activity (Kivimaë et al., 2008; Venkatesan et al., 2019). Furthermore, unlike most enzymes, CKIδ/ε activity is temperature insensitive (Isojima et al., 2009), but paradoxically, the hamster CK1ε^tau mutant is a temperature compensation mutant (Tosini and Menaker, 1998). The conundrum started to unravel in the context of the PER phosphorylation pattern elicited upon the action of multi-kinase hierarchical activities identified in several model organisms (Xu et al., 2007; Ko et al., 2010; Chiu et al., 2011; Lam et al., 2018). The current phosphoswitch model involves two competing phosphorylation sites on mouse (Mus musculus) PER2, the phosphodegron and the FASP (familial advanced sleep phase, Toh et al., 2001) sites, which regulate PER2 stability in opposing ways (Zhou et al., 2015; Masuda et al., 2020). Thus, the temperature-sensitive phosphoswitch slows down PER2 degradation at higher temperatures, resulting in a global temperature-compensated system. Somewhat similar, multiple phospho-clusters are detected on Drosophila PER, which cumulatively contribute to PER stability and transcriptional repressor activity (Chiu et al., 2008; Kivimaë et al., 2008; Garbe et al., 2013; Top et al., 2018). Therefore, the phosphoswitch mechanism might be conserved across species, even though the details differ, as a phosphodegron with functionally heterogeneous sites was recently reported for Drosophila (Joshi et al., 2022).

Both in mammals and Drosophila, the PER phosphorylation pattern is defined by the synergistic action of multiple kinases (see the text above), phosphatases (Sathyanarayanan et al., 2004; Fang et al., 2007; Reischl and Kramer, 2011), and some additional post-translational modifications, such as O-GlcNAcylation and acetylation (Kaasik et al., 2013; Li et al., 2019). The PER phosphorylation dynamics is regulated by yet another level of complexity, as is indicated by the distinct capacity of CKIδ splice isoforms. CKIδ1 and CKIε (both similar in the last 16 amino acids of their carboxy-terminal tails, here abbreviated as “C-terminal tails”) are more active in priming kinase activity at the FASP site, whereas CKIδ2 is more potent in priming the degron site (Fustin et al., 2018; Narasimamurthy et al., 2018). The CKIδ/ε C-terminal tail autophosphorylation inhibits its kinase activity (Graves and Roach, 1995; for review see Narasimamurthy and Virshup, 2021). As was shown for Drosophila DBT, the C-terminal tail stabilizes interactions between the kinase and the substrate, while the C-terminal tail autophosphorylation inhibits substrate binding (Dahlberg et al., 2009; Fan et al., 2015). Furthermore, two residues on the DBT kinase domain influence its affinity to PER (Dahlberg et al., 2009). However, no splicing isoforms of DBT exist in Drosophila as dbt is an intronless gene in this species.

The temperature-independent activity of CK1δ/ε was connected to sequence motifs close to the active site of the kinase, where Lysine 224 was identified as key for the temperature-compensated primed phosphorylation (Shinohara et al., 2017). Importantly, the K224D mutation in CK1δ shortens τ and affects temperature compensation in the mammalian system in vitro. Notably, the corresponding region of Drosophila DBT was systematically explored by Venkatesan et al. (2019) who identified a second NLS in positions 220–224. This region in DBT is further important for its interaction with BRIDE of DOUBLETIME (BDBT) (Venkatesan et al., 2015), a non-canonical FK506-binding protein with tetratricopeptide repeats that might promote the assembly of larger protein complexes (Fan et al., 2013).

Although the circadian clock is in general conserved among bilaterian species, some notable variations in the PER/dTIM/dCRY/mCRY feedback exist with some functional implications (Kotwica-Rolinska et al., 2022a). Therefore, we decided to explore and define the variability in insect DBT proteins. As a reference, we analyzed deuterostomian homologs of DBT and dated the origin and separation of CKIδ from CKIε. Furthermore, we have identified four novel casein kinase I genes specific to the Drosophila genus. We identified conserved regions in DBT specific to Diptera, functionally tested some of them in D. melanogaster, and analyzed their impact on temperature compensation of the circadian clock.

Details of Oxford Nanopore Technology (ONT) transcriptome sequencing will be described elsewhere. Briefly, P. apterus brains and other tissues were dissected and poly A+ mRNA was isolated using Dynabeads mRNA DIRECT Kit (Life Technologies) according to the manufacturer’s instructions and 100 ng of the polyA+ mRNA was then reverse-transcribed, turned to double-stranded DNA, and the sequencing adaptors were added using PCR-free Direct cDNA Sequencing kit (SQK-DCS109; Oxford Nanopore Technology) according to the manufacturer’s instructions. The library was immediately sequenced on a MinION device (Oxford Nanopore Technology). Base calling was performed after the run using Guppy 3.6.0 at a high-accuracy setting. Obtained tissue-specific transcriptomes were used in exhaustive searches using P. apterus dbt mRNA sequence as a query. All dbt transcripts were retrieved, manually inspected, and mapped to the in-house P. apterus genome (hybrid assembly of Illumina and ONT data, which will also be published separately), and a dbt gene model was built. All dbt transcripts were sequentially mapped to four defined individual dbt isoforms and only reads unequivocally distinguishing specific dbt isoform (protein)-coding sequences were counted.

The putative phosphorylation sites were predicted in silico using NetPhos 3.1 server at http://www.cbs.dtu.dk/services/NetPhos/ and scores higher than 0.5 were plotted in alignments.

The target site was designed to induce a double-strand strand brake in the C-terminal tail coding part of dbt gene. Two gRNA sequences (PAM, which is not part of the gRNA, is shown in square brackets) targeting GCG​ATG​CTG​GGC​GGC​AAT​GG[AGG] and GTC​GGC​CTT​CGA​TAC​GGA​TG[CGG], respectively, were prepared from custom-synthesized oligonucleotides and cloned into pBFv-U6.2 (Kondo and Ueda, 2013) obtained from fly stocks of National Institute of Genetics, Japan (NIG-FLY). Plasmids were injected into y ¹ v ¹ P{nos-phiC31\int.NLS}X; attP40 (II) (NIG-FLY#: TBX-0002) flies with docking site on the second chromosome, transformants identified by eye color rescue, and balanced by y ² cho ² v ¹ /Y ^hs-hid ; Sp/CyO (NIG-FLY#: TBX-0008).

Flies expressing Cas9 specifically in germ cells (nos-Cas9) from the second chromosome insertion (NIG-FLY#: CAS-0001; y ² cho ² v ¹ ; attP40{nos-Cas9}/CyO) were crossed with U6gRNA-encoded transgenic strains (also located on the second chromosome). Resulting F1 offspring thus expressed both gRNA and CAS9 on second chromosomes, which potentially targeted the dbt gene located on the third chromosome and induce insertions and deletions as a result of the non-homologous-end-joining (NHEJ) mechanism. The resulting F1 offspring were crossed to y ² cho ² v ¹ ; Pr Dr/TM6C, Sb Tb (NIG-FLY#: TBX-0010) to balance the modified third chromosomes with TM6C. Individual F1 flies were used in heteroduplex mobility shift assay (Kotwica-Rolinska et al., 2019) to identify flies with the highest degree of mosaicism in the targeted dbt locus, thus, the crosses with the highest frequency of NHEJ-induced mutants were identified. From these selected crosses, F2 males and females with third chromosome balancer were individually crossed back to y ² cho ² v ¹ ; Pr Dr/TM6C, Sb Tb flies (NIG-FLY#: TBX-0010) to establish lines with identically modified third chromosomes. Mutated region was identified by polymerase chain reaction (PCR) and sequencing.

Target gRNA sites were selected so that Cas9-mediated cleavage was directed to a target locus of 100 bp upstream and downstream of the dbt K244 site. To avoid off-target cleavage optimal target sites were identified using CRISPR target finder (http://flycrispr.molbio.wisc.edu/tools). One gRNA target was chosen that was close to the target locus. Complementary target site oligos also contained a 5′ guanine for transcription from the U6 promoter and a 3 bp overhang compatible with BbsI sites. Oligos were annealed using standard primer annealing reactions and cloned into BbsI linearized pCFD3 plasmid (Port et al., 2014) via T4 DNA ligation.

Donor plasmids that contain the desired dbt mutations and all elements necessary for homologous recombination were constructed in 3 subsequent cloning steps. In each round of cloning the 1.5 kb 5′ homology arm and the 1.5 kb 3′ homology arm were individually PCR amplified from nos-Cas9 flies (Port et al., 2014) using outside primers dbtBMHRF and dbtBMHRR in combination with respective internal primers. Outside primers dbtBMHRF and dbtBMHRR contain a 15 bp overhang for In-Fusion cloning that is homologous to linearized vector ends. Inside primers have 5′ 15–20 bp extensions that are complementary to each other in addition to one defined mutation for each round of cloning. In the initial round of cloning a silent SalI site was introduced that can be used to screen for transformants. The two fragments (5′ homology arm and 3′ homology arm) were assembled into plasmid pBS-KS-attB1-2-PT-SA-SD-0-2xTY1-V5 (Addgene) that was linearized with XbaI and HindIII using In-Fusion cloning. In a second round of cloning the homology arms were amplified again using the pBS donor plasmid from the previous round as a template. Outside primers were as described above while the inside primers introduced either the K224D or the K224E mutation, respectively. In-Fusion cloning was used to assemble the fragments as described above. The resulting plasmid was then used in a final round of PCR to introduce PAM site mutations to avoid unwanted Cas9 cleavage within the donor plasmid. See Supplementary table S3 for a detailed list of all primers.

Donor plasmids containing the desired mutation along with gRNA plasmids were verified by sequence analysis and scaled up for injections using Qiagen plasmid midiprep. 6 µg of each plasmid were precipitated and eluted in injection buffer. gRNA construct and donor plasmids were mixed prior to injection and the mix was injected into freshly laid embryos of nos-Cas9 flies (Port et al., 2014). Surviving adults were backcrossed in batch crosses to y w; +; Dr/TM3 flies to balance 3rd chromosome modifications. Individual male and female flies from this cross were crossed again to y w; +; Dr/TM3. After letting the females lay eggs for 3–5 days, adult transformant flies were used for molecular screening.

In general, a total of 95 flies for each mutation were screened using PCR in combination with restriction digests. A ∼800 bp target locus was amplified by PCR using genomic DNA from individual flies. 20 units of SalI were then added to half of the PCR reaction and incubated for 2 h at 37°C. The resulting products were analyzed on agarose gels. The remaining PCR product of samples that showed digested products of the correct size were then used for sequencing to verify the presence of the desired mutations.

No homozygous flies could be obtained for both K224 dbt mutations and both stocks are balanced over TM3. Thus, for behavior experiments, flies harboring the dbt mutations were crossed against y w controls and only flies without TM3 were tested. For dbt C-terminal tail mutants, homozygous flies were tested.

Two to four-day old males were loaded into glass tubes containing 5% sucrose in 2% agar and loaded into the DAM2 TriKinetics system (Waltham, MA) and locomotor activity was recorded as previously described (Pfeiffenberger et al., 2010). K224 mutant flies were exposed to 12 h Light: 12 h Dark regime (LD) for 3 days, followed by 5–7 days in constant darkness (DD) to assess their free-running periods at constant temperatures of 18°C, 25°C, or 29°C. Period length and their significance (RS values) were determined using autocorrelation and Chi-square periodogram analysis functions of the fly toolbox implemented in MATLAB (MathWorks) (Levine et al., 2002). Period values with associated RS values ≥1.5 were considered rhythmic (Levine et al., 2002).

Two to four-day old C-terminal tail mutant males were loaded into the DAM2 TriKinetics system as described above, exposed to LD for 5 days, followed by 10 days in DD to assess their free-running periods at constant temperatures of 17°C, 20°C, 25°C, or 28°C. To determine τ during the first 10 days in DD, Lombe-Scargle periodogram analysis was performed using ActogramJ (Schmid et al., 2011) and double-plotted actograms were eye inspected in parallel.

Although mammalian CKIε, CKIδ, and Drosophila DBT are conserved components of the circadian clock, the mouse CKIε sequence did not rescue either the lethality or the rhythmicity of dbt-deficient Drosophila (Sekine et al., 2008). Therefore, to identify the key differences, we compared the protein sequences of mouse CKIε and CKIδ, Drosophila melanogaster DBT, and DBT of the most basal insect, the firebrat Thermobia domestica (Figure 1). All four proteins consist of the conserved casein kinase domain with a substantial C-terminal tail, whereas the N-terminal extension is minimal. However, a detailed inspection revealed two major differences. Within the kinase domain, the region 210–244 of D. melanogaster DBT differs from all three sequences. Notably, D. melanogaster DBT contains Asparagine (N) instead of Threonine (T) in position 220, which contrasts with DBT homologs from Thermobia, mouse, (Figure 1A), and more distant kinases. A recent study indicates that autophosphorylation of T220 influences substrate specificity (Cullati et al., 2022), thus the N at 220 prevents this posttranslational regulatory modification. Remarkable sequence divergence is observed in the C-terminal tail. Although the tail contains positions conserved even among both mouse CKI sequences and T. domestica DBT, surprisingly, the tail sequence is quite different in Drosophila with major insertions and deletions compared to the mammalian and firebrat sequences. These data indicate that during insect evolution DBT acquired substantial changes present in recent D. melanogaster DBT.

To be able to perform a comprehensive analysis of DBT/CKIε/CKIδ evolution, we first explored the phylogeny of casein kinases I in insects and used representative deuterostomian species as a reference. Tau-tubulin kinase served as an outgroup. CKI formed five distinct clusters (Figure 2A and Supplementary Figure S1): DBT/CKIε/CKIδ, CKIα, CKIγ, and two additional clusters not assigned to a specific CKI-type. These two clusters, provisionally labeled as CG9962 and CG2577, seem to be specific to the Drosophila genus, as no representative was found even in the dipteran genera Musca or Ceratitis. Similarly, two Drosophila-specific clusters are branching at the base of CKIα; thus, we use the provisional terms CKIα-like I and CKIα-like II (see Supplementary Table S1 for all non-DBT acc. numbers and Supplementary Table S2 for DBT, CKIε, and CKIδ acc. numbers). In line with the observed phylogenetic clustering of CKI, differences were identified in the N- and C-terminal tail lengths (Figure 2B) and in the sequence motifs within the kinase domain, including the activation loop (Philpott et al., 2020) and residues N275 and R279, which are responsible for enhanced substrate-specific binding to DBT (Dahlberg et al., 2009) (Figure 2C).

Having unambiguously identified CKI types, we performed a systematic audit of the DBT sequences across insects with three goals: explore possible patterns in the C-terminal tail variability, determine when the NKRQK motif arose, and identify whether we may correlate these changes in DBT with additional changes in the circadian clock setup.

In our comprehensive analysis, we identified and further analyzed DBT sequences from 55 species representing 20 insect orders and 9 deuterostomian classes (Figure 3, Supplementary Figures S2–S7, and Supplementary Table S2). Whereas only one dbt gene was found in all analyzed insects, up to as many as three dbt paralogs were identified in the zebrafish Danio and four paralogs in the clawed frog Xenopus. In mammals, reptiles, and birds, two paralogous genes, CKIε and CKIδ, are known. Similarly, in the sea lamprey Petromyzon marinus, two dbt-like genes were discovered; however, a detailed sequence comparison indicates that these dbt-like genes result from lamprey-specific gene duplication (Figure 4). The CKIε/CKIδ separation is observed in sharks, rays, and fishes, and thus seems to be a result of gene duplication specific to Gnathostomata. The second duplication of CKIδ led to two CKIδ genes present in the zebrafish Danio rerio and probably a similar but independent (Xenopus-specific) duplication happened in the ancestor of the clawed frog Xenopus laevis. In addition to gene duplications and quadruplications, a various number of protein isoforms can be produced in some organisms from individual genes as a result of alternative splicing (see the C-terminal tail analysis below).

The NKRQK motif (positions 220–224 in Drosophila DBT) within the catalytic domain distinguishing D. melanogaster DBT from the mouse homologs (Figure 1) was identified in all analyzed dipteran insects (Figure 3). Apart from the desert locust Locusta migratoria (the sequence was confirmed by Sanger sequencing and only one dbt gene was identified in the genome), all non-dipteran species contain the TKRQK motif. Therefore, we compared the presence of NKRQK to known changes in the circadian clock setup, such as the presence of mCRY, loss of dTIM (Kotwica-Rolinska et al., 2022a), and transformation of BMAL to CYC (Meireles-Filho et al., 2006). The transition of BMAL, a transcription factor with a transactivation domain, to CYC, a transcription factor which lacks the transactivation domain, was identified exclusively in Cyclorrhapha, a subset of Diptera (Supplementary Figure S8). However, this change perfectly corresponds to the loss of mCRY and does not agree with the presence of NKRQK (Figure 3).

Since none of the known changes in the clock setup correlated with the presence of the NKRQK motif in DBT, we analyzed the evolution of BDBT, a non-canonical FK506-binding protein interacting with DBT in Drosophila (Fan et al., 2013). First, we performed a phylogenetic analysis of available known FK506-binding proteins. The unrooted phylogenetic tree in Figure 5 (and the full tree version in Supplementary Figure S9) represents how various FK506-binding proteins evolved over time. Notably, dipteran BDBTs form a clear cluster that is separated from all remaining proteins. Furthermore, FK-506 binding proteins from Mecoptera (Scorpionflies) and Siphonaptera (Fleas), the closest relatives of Diptera, do not cluster with BDBT. Therefore, BDBT has been so heavily modified in Diptera that we cannot reliably identify the corresponding BDBT in any non-dipteran insect. Interestingly, in Diptera, the rise of the BDBT gene correlates with the transition from TKRQK to NKRQK motifs. The only other occurrence of the NKRQK motif was observed in Locusta (Orthoptera), whereas the sister polyneopteran lineages (termites, roaches, phasmids, Mantophasmatodea) contain TKRQK. However, FK-506 binding proteins of all polyneopteran lineages (including Orthoptera) branch together independently of TKRQK to NKRQK motifs.

The well-established impact of the C-terminal tail on the priming capacity of CKIδ splice isoforms in mice (Narasimamurthy and Virshup, 2021) prompted us to explore the C-terminal tail variability in the identified DBT/CKI dataset. The in silico predicted phosphorylation pattern was depicted for probabilities >0.5 (Figure 3 and Supplementary Figures S2–S7). When compared to the kinase domain, the C-terminal tails are the most variable parts of the proteins. Somewhat similar phosphorylation patterns and comparable lengths are found among C-terminal tails in vertebrates. In insects, however, the length and sequence of the C-terminal tails are remarkably variable. The shortest tails were identified in Lepidoptera (in several species only around 50 amino acids), whereas the longest tail in the Anopheles mosquito exceeded 250 amino acids (Figure 3). Putative phosphorylation was more prevalent in the C part of the C-terminal tail; however, the predicted phosphorylation patterns were quite variable in insects.

Alternative splicing of mouse CKIδ transcripts affects the biochemical properties of resulting CKIδ protein isoforms. In all analyzed vertebrate species, CKIδ was alternatively spliced with impact on the predicted phosphorylation pattern in the terminal part of the C-terminal tail. In contrast, CKIε was alternatively spliced only in a few vertebrate species (Supplementary Figure S7). Alternative splicing was detected in dbt of many insects, including the mosquito Anopheles albimanus (Diptera), all analyzed beetles (Coleoptera), all analyzed butterflies/moths (Lepidoptera), all analyzed hymenopteran species (the honey bee Apis mellifera, the jewel wasp Nasonia vitripennis, and the turnip sawfly Athalia rosae), the pea aphid (Acyrthosiphon pisum), both analyzed species of Thysanoptera (the western flower thrips Frankliniella occidentalis and the melon thrips Thrips palmi), the silverleaf whitefly Bemisia tabaci, true bugs (Heteroptera: the water strider Rhagovelia antilleana, the kissing bug Rhodnius prolixus, the common bed bug Cimex lectularius, the brown marmorated stink bug Halyomorpha halys, and the linden bug Pyrrhocoris apterus), and the drywood termite Cryptotermes secundus (Supplementary Figures S3–S5). As a representative of the true bugs, having access to the linden bug P. apterus brain transcriptome obtained by Oxford Nanopore Technology, we analyzed the presence and abundance of all four identified dbt isoforms (Figure 6). The three most abundant isoforms comprising 99% of dbt transcripts encode proteins with predicted altered phosphorylation patterns. Thus, alternative splicing of dbt might serve as a regulatory step influencing and modulating the properties of DBT in some insects. In flies including Drosophila, however, dbt is an intronless gene. A similar intronless gene organization might exist in other insect species. However, the identification of only one transcript variant should be interpreted with caution. The reliable decision whether dbt gene contains introns requires a good non-fragmented genome assembly, ideally with well-annotated gene models. On the other hand, the identification of only one dbt isoform in the transcriptome of a particular species may reflect only shallow sequencing or might be affected by transcript assembly and post-sequencing processing.

The absence of introns in dbt genes of flies and the remarkable diversity of the C-terminal tail among flies and mosquitoes motivated a detailed comparison of this part of DBT in Diptera. Protein alignment of 15 species representing major dipteran lineages revealed a conserved region in the C-terminal tail, where a short 12 amino acid motif is identified in all dipteran species, and an even longer motif is shared among Cyclorrhapha (Figure 7). This conserved motif contains residues with high scores predicting their phosphorylation.

To test whether the identified sequence motifs in DBT have an impact on the function of the circadian clock in vivo, we applied reverse genetic tools to the fruit fly D. melanogaster and focused on two regions: 1) residue K224, which was established as important for temperature compensation in mammalian CKIδ (Shinohara et al., 2017), and which is also part of NKRQK motif (Figure 3), and 2) the C-terminal tail (Figure 7).

Three different C-terminal tail mutants were created, encompassing or bordering the conserved C-terminal tail domain (Figures 8A, B). All of them are homozygous viable and displayed only very mild, yet significant circadian phenotypes. The deletion of amino acids downstream of position 370, that is the part comprising conserved cyclorrhaphan and dipteran motifs, slightly extended τ when compared to controls (Figures 8C, D and Supplementary Table S4). Deletions of the very end of the C-terminal tail and frameshift (∆411–440) and a 15-bp in-frame deletion upstream of conserved motifs (∆366–370) mildly shortened τ (Figures 8E, F and Supplementary Table S4).

The basic lysine residue K224 was replaced by the acidic residues aspartic acid (D) or glutamic acid (E), respectively. In both cases, homozygous mutants were not viable, thus, heterozygous flies were analyzed. In both mutants, τ was significantly shorter compared to control flies. Moreover, these two mutants differ in their temperature compensation (Figures 8G–J). The K224D heterozygotes displayed a τ of ∼20 h at 18°C and lengthening of τ up to 22 h at higher temperatures (Q₁₀ = 0.94) (Figures 8H, J). The K224E heterozygotes displayed an approximately 5-h faster clock compared to the wild type (τ ∼ 19) at all three tested temperatures and showed reduced rhythmicity at 18°C (Figures 8I, J). The shorter τ corresponded to earlier evening activity peaks in LD regimes in both mutant lines, K224D (Supplementary Figure S11) and K224E (Supplementary Figure S12), when compared to control y w flies (Supplementary Figure S10).