Results and discussion

A miRNA that impacts larval movement

Movement is the main output of the nervous system allowing animals to walk, fly, crawl, swim and maintain their posture, so that they can find prey, mate partners, escape predators and relocate within habitats. Despite the key biological and adaptive roles of movement across animal systems, how developing embryos manage to organise the necessary molecular, cellular, and physiological processes to initiate patterned movement is still unknown. Although it is clear that the genetic system plays a role, how genes control the formation, maturation and function of the cellular networks underlying the emergence of motor control remains poorly understood.

Recent work in our laboratory has shown that miRNAs – which are short regulatory non-coding RNAs that repress the expression of target genes [1, 2] – have pervasive roles in the articulation of complex movement sequences such as those involved in body posture control in the young Drosophila larva [3–6]; these observations, as well as those from others in Drosophila and other systems [7–11] hinted at the possibility that these non-coding RNA molecules might also be involved in the control of more fundamental aspects of motor development and control. To explore this question, we first searched for miRNAs that might affect the simple locomotor patterns of the Drosophila first instar larva. The L1 larva which is a convenient model to investigate the genetics of movement given that: (i) the assembly of the machinery for movement must be fully completed by this developmental stage so as to satisfactorily propel the animal into the external world, and (ii) if analysed sufficiently early, for example during the first few minutes after hatching, the animal had no real chance of compensating or learning ways around a putative defect, increasing the probability of detection by means of a suitable motor test. To extract a signature of larval movement, we applied a whole animal imaging method based on frustrated total internal reflection (FTIR) [12, 13] which renders high resolution and high contrast movies to both normal and miRNA mutant first instar larvae. This led us to discover that a single miRNA, miR-2b-1, had a significant impact on larval movement (Figure 1B-1C). miR-2b-1 belongs to the miR-2 family [14] and ΔmiR-2b-1 mutant larvae show a substantial decrease in larval speed (Figure 1C) suggesting that absence of this specific genetic component compromises the ability of the larva to move normally. Given that these larval tests were conducted within a 30-min period post-hatching, we considered the possibility that the defects observed in the larvae stemmed from changes in earlier ontogenetic processes that occur in the embryo. More specifically, we decided to test the possibility that early embryonic movement patterns might be affected by the lack of normal expression of miR-2b-1.

Figure 1

Previous work had provided an excellent first characterisation of the onset of embryonic movement patterns in wild type embryos [15, 16]; these early studies were based on the manual annotation of representative videotaped muscle contractions [16] or GFP-labelled muscle Z-lines analysed under spinning disc confocal microscopy [15]. Despite their attributes, these approaches were highly labour intensive and lacked the necessary throughput required to simultaneously analyse large numbers of embryos enabling a sensitive genetic screen. In consequence, we developed a new automated approach capable of quantifying movement in large populations of embryos.

A high throughput approach to quantify movement in Drosophila embryos

To monitor the onset of embryonic movement, which, in normal embryos, occurs during the final third of embryogenesis (i.e. 14-16h after egg laying (AEL) [15, 16] [Figure 1D-1E] we developed a 3D printed chamber system capable of hosting multiple embryos submerged under a thin layer of halocarbon oil to ensure adequate oxygenation and hydration (Figure 1F-1G) compatible with digital imaging by a charge coupled device (CCD) camera (Figure 1H). The camera captures the reflected light after its physical contact with the embryo; in this setting, even a subtle movement performed by the embryo results in a change in the path of reflected light, leading to variations in signal intensity detected by individual pixels in the CCD sensor, allowing for an accurate measurement of embryonic movement (Figure 1I-1J). To extract quantitative movement information from individual embryos we applied an image segmentation protocol to define regions of interest (ROIs) corresponding to each embryo and collected pixel intensity values from all ROIs at 4 frames/sec (Figure 1K-1L). The data allow us to plot variations in average grey pixel intensity over embryonic time, which provide a quantitative signature of the ontogeny of movement in the individual embryo (Figure 1M). From this, we were able to observe distinct phases of movement that are consistent with previous data: namely, the onset of a phase of disorganised movements ∼16 hours after egg laying (hAEL) and its transition into a phase characterised by rhythmic bursts of activity and inactivity ∼18 hAEL [15, 16]. These phases have been termed as ‘myogenic’ and ‘neurogenic’ based on their respective dependence on neural input (Figure 1N) [15, 17–19], and have been observed in Drosophila, as well as in other systems, including vertebrates [20–24] strongly suggesting that this is a general feature of motor development. See Movie S1 for a high resolution recording of Drosophila embryo movement.

To establish whether the readings of motor activity at the neurogenic phase detected by our approach were indeed dependent on neural activity, we expressed the inwardly rectifying potassium channel (Kir) [25] in all embryonic neurons using the pan-neuronal driver elav-Gal4 seeking to suppress action potentials across all embryonic neuronal types. The results of this experiment show that whilst movement patterns during the early chaotic phase remain unchanged by this treatment (Figure 2C, 2E), motor activity at the rhythmic phase is almost completely eliminated (Figure 2C, 2F) strongly indicating that the emergence of this latter phase depends on normal neural activity. This is in agreement with previous observations of the effects of embryonic neural activity inhibition by other methods [15, 18]. In addition, spectral analysis demonstrates that the movement frequencies that characterise the rhythmic phase (Figure 2B, 2G) do not emerge in Kir embryos (Figure 2D, 2G). Altogether these observations suggest that miR-2b-1 might exert its roles on embryonic movement, at least in part, due to action within the developing embryonic nervous system.

Figure 2

The miRNA miR-2b-1 affects embryonic movement patterns

Given that the machinery for larval movement is assembled during embryogenesis [26–29] we considered the possibility that miR-2b-1 might have an impact on the emergence of movement in the fly embryo. To explore this, we applied the approach described above to normal and ΔmiR-2b-1 mutant embryos (Figure 2H, 2J). These experiments showed that ΔmiR-2b-1 mutant embryos displayed a different pattern of embryonic movement when compared with their wild type counterparts. Although the distinct phases of embryonic movement are clearly recognisable in mutant embryos, the overall amount of movement appeared greatly reduced (Figure 2J).

Indeed, comparison of quantity of movement in wild type and ΔmiR-2b-1 mutant embryos either during the earlier myogenic phase (Figure 2L) or during the neurogenic phase (Figure 2M) shows significantly lower levels of movement in mutants. Furthermore, spectral analyses of embryonic movement traces reveal that miRNA mutant embryos shift to a higher frequency of movement bouts during the rhythmic phase when compared to normal embryos (Figure 2I, 2K) and that average bout length is also shortened (Figure S1).

miR-2b-1 expression and roles in the embryonic nervous system

The fact that removal of miR-2b-1 impacts the neurogenic phase of embryonic movement suggests that this miRNA might be expressed in the nervous system and exert a functional role there. To further explore this model, we first conducted a spatial expression analysis in the developing embryo using fluorescence RNA in situ hybridisation (FISH). Taking advantage from the fact that miR-2b-1 is a ‘mirtron’ [30] we prepared an in situ probe against miR-2b-1’s host gene Btk, which encodes the Bruton tyrosine kinase [31, 32] (Figure 3A). These FISH experiments show that the miR-2b-1 precursor miRNA transcript (pre-miRNA) [2] is expressed in multiple embryonic tissues, including the CNS (Figure 3B). To further confirm the expression of miR-2b-1 in the nervous system, we conducted fluorescence-activated cell sorting (FACS, Figure 3C) to isolate embryonic neuronal samples (labelled by means of elav>GFP) followed by RT-PCR for the mature miRNA transcript and observed miR-2b-1 specific signal (Figure 3D). Therefore, the results of these two distinct and complementary methods provide strong evidence of neural expression of the miRNA.

Figure 3

To gain more insight on the neural roles of miR-2b-1 in regard to embryonic movement, we conducted a genetic reconstitution experiment in which we analysed the consequences of restoring miR-2b-1 expression in the nervous system in an otherwise ΔmiR-2b-1 null mutant (Figure 3E). Results in Figure 3F show that elav-driven expression of miR-2b-1 in a ΔmiR-2b-1 mutant background leads to a phenotypic rescue, producing embryos that display statistically indistinguishable movement patterns to those recorded in control embryos, indicating that neural expression of miR-2b-1 is sufficient to restore a normal onset of embryonic movement. To further examine the biological roles of neural miR-2b-1 expression, we assessed the impact of restoring miRNA expression in embryonic neurons on first instar larval locomotor patterns using the FIM approach described above (Figure 1B-C) and observed that when ΔmiR-2b-1 mutant larvae are developmentally provided with pan-neuronal miR-2b-1 expression, the characteristic miRNA larval mutant phenotype is rescued, with specimens moving at natural speed (Figure 3G, S2A). The experiments described above strongly indicate that expression of miR-2b-1 in the nervous system is biologically relevant and sufficient to rescue the embryonic and larval movement defects observed in ΔmiR-2b-1 mutants. In turn, this suggests that absence of miR-2b-1 must impinge a morphological and/or a functional deficit in the developing nervous system of the embryo. To tease apart these potential biological effects, we examined the structure of the nervous system in normal and ΔmiR-2b-1 mutant embryos by means of immunohistochemistry and confocal microscopy and observed no detectable differences (Figure S2B). In contrast, analysis of neural activity patterns in the embryo by means of GCaMP6 functional imaging using a movement distortion correction approach (i.e. tdTomato, Figure 3H-3J [17]), shows that miRNA mutant embryos have a reduced level of calcium dynamics when compared with their control counter parts (Figure 3K-3M); notably, this occurs during a previously identified ‘critical period’ when neural activity levels are of crucial importance to the development of stable neural circuits [33–35].

A genetic link between miR-2b-1 and embryonic movement

Our gene expression, genetic reconstitution, morphological and functional imaging data support a model in which miR-2b-1 plays a physiological role in the developing embryonic nervous system. This raises the question of how might this regulatory miRNA system interact with the physiological control of the neuron during embryogenesis. To explore the genetic elements that link miR-2b-1 to its role in embryonic movement we searched for candidate miR-2b-1 target genes using the ComiR bioinformatic platform (Figure 4B) [36, 37]. A common issue with bioinformatic predictions of miRNA targets is the generation of false positives; in this regard, ComiR integrates multiple miRNA target prediction algorithms – each one with its intrinsic strengths and weaknesses[38–41] – seeking to identify a set of consistent bona fide miRNA targets that satisfy the filters of multiple algorithms, thus reducing the generation of false positives[36, 37]. Applying ComiR to miR-2b-1 produced a list of high probability targets organised in the form of an ascending ranking (Figure 4B). At the very top of the list was CG3638, an uncharacterised Drosophila gene predicted to encode a chloride channel protein; this highly ranked target was of interest to us because of its potential role in the physiological control of anionic conductances, and its broad evolutionary conservation across insects and mammals[42], including humans (Figure 4E-I).

Figure 4

A genuine genetic target for a given miRNA is predicted to: (i) be de-repressed (up-regulated) in the absence of the miRNA; and (ii) be down-regulated under miRNA ectopic expression. Analysis of CG3638 expression shows that this target meets the predictions of a genuine miR-2b-1 target in full: expression of CG3638 is upregulated in ΔmiR-2b-1 mutants (Figure 4B) and reduced under neural over-expression of miR-2b-1 (Figure 4D). As mentioned above, phylogenetic analysis of CG3638 reveals that it belongs to an evolutionarily conserved family of chloride channel genes[42, 43], with representatives across distantly related lineages of insects and vertebrates including mammals, strongly indicating a functional role (Figure 4E). Comparison of the properties of CG3638 and its human orthologue reveal the characteristic seven trans-membrane domains with an external carboxyterminal topology (Figure 4F, 4H) further supporting orthology, and applying AlphaFold – an artificial intelligence computational method able to predict protein structures with atomic accuracy [44] – to the proteins encoded by the Drosophila CG3638 gene and the human TTYH1 gene reveals the marked similarities between these two polypeptides (Figure 4G, 4I).

To explore the relationship between CG3638 and the embryonic movement phenotype displayed by ΔmiR-2b-1 mutants, we tested the effects of an artificial reduction of CG3638 in the genetic background of the miRNA mutant (Figure 4K). In this scenario, should the levels of expression of CG3638 be relevant to the triggering of the embryonic movement phenotype, we predict that a reduction in CG3638 expression levels should compensate its cellular effects, and, accordingly, reduce or even rescue the embryonic phenotype. The results of this experiment show that this is indeed the case, with embryonic movement of the ΔmiR-2b-1 mutant effectively rescued by a reduction in CG3638 (Figure 4J). Based on this role of embryonic movement we termed CG3638 as Janus, making reference to the Roman god of movement and the symbol of change and transitions.

Mapping the focus of action of miR-2b-1 within the known networks underlying embryonic movement

Having observed the effects of miR-2b-1 on embryonic movement we wondered about the site of action of this miRNA in regard to circuit components previously linked to embryonic movement. In this respect, previous work has identified embryonic motor neuronal components, as well as interneurons and elements of the sensory system as playing key roles in the control of motor development. These include a motor component that includes all embryonic motor neurons which command the stereotypic array of muscles in the embryonic body wall [45, 46], as well as specific elements of the sensory system – in particular the chordotonal system – which detect early myogenic movements in the embryo and transmit the information to the pattern generators thus modulating motorpatterns (Figure 4L) [17, 19, 47, 48]. To determine which one of these known circuitry elements might be the principal focus of action of miR-2b-1 in connection to embryonic movement control we artificially expressed miR-2b-1 in each motor and sensory circuit elements – using drivers OK371-Gal4 [49] and iav-Gal4 [50], respectively – in an otherwise null mutant background for miR-2b-1, asking whether these genetic restorations were sufficient to improve or perhaps even rescue normal movement patterns. To ensure that circuit-specific Gal-4 drivers were active and UAS-driven miRNA levels achieved the necessary cumulative values for biological activity we chose to measure effects on early larval movement patterns tested in 30-min old first instar larvae (L1s). The results of these experiments are shown in Figure 4M-N M-O. Here we observe that restoring expression of miR-2b-1 in the motor neuronal domain defined by the OK371 driver is insufficient to affect the defective movement patterns observed in miR-2b-1 null mutants (Figure 1B-C and Figure 4M). In contrast, re-establishing expression of the miRNA in all eight chordotonal organs leads to a full rescue of the motor phenotype (Figure 4N) suggesting that this aspect of the sensory system might be the one where miR-2b-1 exerts relevant actions during the normal development of movement. In line with this, we observe that the mature miR-2b1 transcript is indeed expressed in FACS-isolated embryonic chordotonal organs prepared from wild type embryos (Figure 4P) and, that the genetic target of miR-2b-1, Janus, is also expressed in these cells in normal embryos (Figure 4Q). In addition, expression of Janus in chordotonal organs prepared from miR-2b1 null mutants is upregulated (Figure 4Q) lending further support to a model in which Janus is de-repressed in these specific sensory elements. Furthermore, artificial reduction of Janus implemented as a stratagem to decrease the effects of de-repression specifically within the chordotonal system is sufficient to rescue normal movement patterns (Figure 4R). Altogether, these findings strongly suggest that miR-2b-1 impacts the emergence of embryonic movement, at least in part, via effects on the sensory circuit components that underlie motor development, rather than affecting the actual generation of motor patterns.

Our work identifies a miRNA system that plays a role in the emergence of embryonic movement in the fly embryo, and offers a new approach to analyse the roles of non-coding RNAs and protein coding genes at the critical period when patterned movement develops. Understanding the molecular elements controlling the onset of motor development in Drosophila will put us one step closer to understanding the molecular basis of embryonic movement in other species, including vertebrates, whose embryos seem to undergo remarkably similar transition phases to those reported here [20].

Acknowledgements

We wish to thank Tom Baden for their contributions to the development of the embryonic movement pipeline and computational analysis of movement patterns, respectively. We also thank all members of the Alonso Lab for helpful discussions and feedback on this work. We are very grateful to Marta Zlatic for sharing fly lines, and to the Bloomington Stock Center and the Vienna Drosophila Resource Center for providing fly stocks. This research was funded by a Wellcome Trust Investigator Award (098410/Z/12/Z) made to C.R.A., and a UK Medical Research Council Project Grant (Ref: MR/S011609/1) given to C.R.A.

Competing interests

The authors declare no competing interests.