Exploring miR-9 Involvement in Ciona intestinalis Neural Development Using Peptide Nucleic Acids

The microRNAs are small RNAs that regulate gene expression at the post-transcriptional level and can be involved in the onset of neurodegenerative diseases and cancer. They are emerging as possible targets for antisense-based therapy, even though the in vivo stability of miRNA analogues is still questioned. We tested the ability of peptide nucleic acids, a novel class of nucleic acid mimics, to downregulate miR-9 in vivo in an invertebrate model organism, the ascidian Ciona intestinalis, by microinjection of antisense molecules in the eggs. It is known that miR-9 is a well-conserved microRNA in bilaterians and we found that it is expressed in epidermal sensory neurons of the tail in the larva of C. intestinalis. Larvae developed from injected eggs showed a reduced differentiation of tail neurons, confirming the possibility to use peptide nucleic acid PNA to downregulate miRNA in a whole organism. By identifying putative targets of miR-9, we discuss the role of this miRNA in the development of the peripheral nervous system of ascidians.


Introduction
The microRNAs (miRNAs) are small (19−25 nucleotides in length) non-coding RNAs, which have emerged as a fundamental class of potent regulatory molecules [1]. They are involved in many developmental and physiological processes [2], as well as in pathological conditions. Dysregulation of miRNAs is indeed associated with a variety of pathologies [3], and the possibility to regulate gene expression by interfering with miRNA networks is one of the most intriguing fields of novel drug development (miRNA-based therapeutics); miRNAs replacement or inhibition could represent an effective therapeutic approach in many diseases [4]. Although huge advances have been made in miRNA-based therapy, the in vivo stability of miRNA analogues is still one of the main challenges in this field [5,6].

miR-9 Expression during C. intestinalis Development
In order to determine miR-9 expression in developing C. intestinalis embryos, we performed whole mount in situ hybridization with a specific locked nucleic acid (LNA) probe for mature Cin-miR-9 (accession: MIMAT0016396). In swimming larvae, the hybridization signal was localized in the epidermal sensory neurons of dorsal and ventral midlines of the tail ( Figure 2D). These results were confirmed by in situ hybridization with a probe specific for pri-miR-9 (Figure 2A-C). In juveniles, the miR-9 signal was detectable in several cells scattered in the epidermis near the oral syphon ( Figure  2E,F).

miR-9 Downregulation by PNAs
To investigate the role of miR-9 in development, microinjections of PNA-a9 and PNA-sc9 (used as controls) were performed in C. intestinalis eggs. No difference in the developing rates was observed between embryos microinjected with PNA-a9 (65%) and controls injected with the medium (72%) and with PNA-sc9 (66%). Most of the embryos injected with PNA-a9 displayed a rounded and shorter trunk than controls (70%). The morphology of the embryos injected with PNA-sc9 was not affected by the treatment (78%), being comparable with the controls injected with the medium.
When we analyzed the expression of neural marker genes, PNA-a9-injected embryos showed anomalies at the level of the peripheral nervous system. Indeed, in situ hybridization with a pan neural marker, Ci-Pans [37], revealed a reduced expression in the epidermal sensory neurons of both the trunk and tail (80%; Figure 3). Particularly, in the tail, transcripts were absent in the proximal dorsal and ventral epidermal neurons, however some neurons were eventually marked in the ventral tip of the tail ( Figure 3C,D,F,G,I). Similarly, anterior sensory neurons of the trunk, including the papillary sensory neurons, were reduced in PNA-a9-injected embryos ( Figure 3D,F).

miR-9 Downregulation by PNAs
To investigate the role of miR-9 in development, microinjections of PNA-a9 and PNA-sc9 (used as controls) were performed in C. intestinalis eggs. No difference in the developing rates was observed between embryos microinjected with PNA-a9 (65%) and controls injected with the medium (72%) and with PNA-sc9 (66%). Most of the embryos injected with PNA-a9 displayed a rounded and shorter trunk than controls (70%). The morphology of the embryos injected with PNA-sc9 was not affected by the treatment (78%), being comparable with the controls injected with the medium.
When we analyzed the expression of neural marker genes, PNA-a9-injected embryos showed anomalies at the level of the peripheral nervous system. Indeed, in situ hybridization with a pan neural marker, Ci-Pans [37], revealed a reduced expression in the epidermal sensory neurons of both the trunk and tail (80%; Figure 3). Particularly, in the tail, transcripts were absent in the proximal dorsal and ventral epidermal neurons, however some neurons were eventually marked in the ventral tip of the tail ( Figure 3C,D,F,G,I). Similarly, anterior sensory neurons of the trunk, including the papillary sensory neurons, were reduced in PNA-a9-injected embryos ( Figure 3D,F).  We further characterized the effects of PNA-a9 injection; samples were processed for in situ hybridization with Ci-POU-IV and Ci-V-Glut probes. Ci-POU-IV is a gene involved in neural precursor specification and is a marker of the epidermal sensory neurons [38] (Figure 4A,B,D,E). Ci-POU-IV expression was completely absent in the tail of 80% of the larvae developed from injected eggs with PNA-a9 and only a few isolated neurons were faintly marked in the ventral midline ( Figure 4C,F), confirming previous results. Ci-V-Glut [39] marks the sensory neurons of the papillae and the tail, the posterior visceral ganglion, and the posterior neural tube in the controls ( Figure 4G). In injected larvae, the expression of this gene was evident only in a few neurons of the proximal neural tube, very close to the trunk (90% of tested embryos; Figure 4H We further characterized the effects of PNA-a9 injection; samples were processed for in situ hybridization with Ci-POU-IV and Ci-V-Glut probes. Ci-POU-IV is a gene involved in neural precursor specification and is a marker of the epidermal sensory neurons [38] (Figure 4A,B,D,E). Ci-POU-IV expression was completely absent in the tail of 80% of the larvae developed from injected eggs with PNA-a9 and only a few isolated neurons were faintly marked in the ventral midline ( Figure  4C,F), confirming previous results. Ci-V-Glut [39] marks the sensory neurons of the papillae and the tail, the posterior visceral ganglion, and the posterior neural tube in the controls ( Figure 4G). In injected larvae, the expression of this gene was evident only in a few neurons of the proximal neural tube, very close to the trunk (90% of tested embryos; Figure 4H)
Then, we tested these selected transcripts by bioinformatics analysis. At present, accurate in silico identification of miRNA targets is a great challenge. Numerous programs have been developed to predict putative miRNA targets in vertebrates, Drosophila, and worms (e.g., TargetScan, DIANA, miRANDA, etc.), whereas only a few can be employed in non-standard animal models. In this study, we used 3 free available programs: RNAhybrid [42], RNA22 version 2.0 [43], and PITA executable [44].
Then, we tested these selected transcripts by bioinformatics analysis. At present, accurate in silico identification of miRNA targets is a great challenge. Numerous programs have been developed to predict putative miRNA targets in vertebrates, Drosophila, and worms (e.g., TargetScan, DIANA, miRANDA, etc.), whereas only a few can be employed in non-standard animal models. In this study, we used 3 free available programs: RNAhybrid [42], RNA22 version 2.0 [43], and PITA executable [44].
We found one or more target sites in all analyzed transcripts. Only considering sites within 3 UTR, all 3 programs identified the same target site in Ci-FoxG, whereas 2 programs predicted the same target site in Ci-Hes-b, Ci-caderin, and Ci-SoxB1/2 mRNAs (Table 1).

Discussion
We explored miR-9 involvement in the C. intestinalis development, performing knockdown experiments with antisense peptide nucleic acids (PNAs). PNAs are nucleic acid analogues characterized by a synthetic peptide backbone. These oligomers specifically bind to complementary DNA or RNA sequences, obeying Watson-Crick base paring [11]. In recent years, different studies have demonstrated PNA specificity for their complementary miRNAs in vitro [4,7,8].
In ascidians, recent comparisons between PNAs and the commercial AntagomiRs (Dharmacon, USA) antisense molecules have demonstrated the ability of PNAs to efficiently downregulate miRNAs. The effects induced by injections of these molecules (PNAs and AntagomiRs) into C. intestinalis eggs were comparable, confirming the reliability of PNA-based knockdown [22].
Modified PNAs were also used for sequence-based knockdown mRNA translation through the same morpholino phosphorodiamidate (MO) mechanism of action in zebrafish. Studies comparing the strength of these molecules demonstrated that they were equally effective in downregulating target expression. In addition, PNA oligomers proved to be more specific than conventional MO: even a one-base mismatch in PNA was sufficient to prevent PNA binding with target sequences, while three mismatches were necessary to prevent MO inhibitory effects [45].
Furthermore, miR-9 is one of the most conserved miRNAs in bilaterian animals, characterized by specific expression in the nervous systems of both vertebrates and invertebrates [27,46,47]. According to our analyses, in the developing nervous system of the ascidian C. intestinalis, mature miR-9 is present in epidermal sensory neurons of the tail and in the papillary sensory neurons. This expression was confirmed using a probe for pri-miRNA. It could be possible that miR-9 is also expressed at low levels in other regions of the developing embryos and cannot be detected by in situ hybridization. We found that downregulation of miR-9 by PNA-a9 microinjections in eggs specifically affected the development of the peripheral nervous system (PNS), as demonstrated by in situ hybridization with specific markers. Although quantitative analysis would be necessary to completely confirm PNA-a9 efficiency, the results appear to be highly reliable. Moreover, we never observed the expression of Ci-POU IV, which is required for PNS specification, in the epidermis of the tail and the trunk of injected embryos.
The PNS of ascidian larvae is composed of a limited number of epidermal sensory neurons (ESNs) located in the trunk and in the tail midline. These ciliated neurons are thought to have mechanosensory functions. In the tail, dorsal and ventral ESNs arise from two separate populations of cells with partially different gene circuits [40,48,49]. The tail epidermis midline is a neurogenic region and the Notch pathway negatively controls the number of cells adopting the ESN fate [48]. Subsequently, the sequential activation of MyT1, POU IV, atonal, and NeuroD induces ESN development. Larvae injected with PNA-a9 showed few or no differentiated ESNs. It could be hypothesized that downregulation of miR-9 repressed neural fate determination, as it inhibited neural differentiation by ectopic expression of some of its targets. Two inhibitory mechanisms have been proposed for preventing ESN formation; one involving Notch effector genes, such as those belonging to Hes transcriptional repressor family, and the other through SoxB2 gene activation [40]. Our bioinformatics analysis indicated that Hes is a putative target of miR-9 in C. intestinalis, differing from the results of Spina et al. (2011), who did not identify binding sites for miR-9 in any members of the Notch pathway. Interestingly, a set of miR-9 targets are members of the Hes gene family in vertebrates, where they inhibit neural differentiation by repressing proneural genes [27]. The C. intestinalis genome contains three Hes genes (Hes-a, Hes-b, Hes-c) with different patterns of expression [40,50]. Particularly, Hes-b is expressed outside the epidermis midlines, probably preventing ESN formation [40]. Downregulation of miR-9 could induce Hes-b ectopic expression in the midline, thus inhibiting PNS differentiation in this region. SoxB2 is also a putative target identified by our analysis, and it is expressed throughout the epidermis but not in the midlines. Although few studies are available about this gene in ascidians, it has been demonstrated that MsxB knockdown caused SoxB2 upregulation in the dorsal midline [51]. Moreover, in C. intestinalis, SoxB2 transcripts have been found to be a target of miR-124, another essential neural miRNA, suggesting that its downregulation is necessary to allow nervous system differentiation [40,52].
It is known that miR-9 is the only predicted miRNA to target E-cadherin transcripts in vertebrates [35]. This transmembrane glycoprotein forms the core of cellular adherens junctions and sequesters β-catenin from the cytoplasm. In C. intestinalis, cadherin superfamily genes have been characterized and include 2 classical cadherins: Cadherin and Cadherin II [53]. Our in silico investigations revealed that Cadherin, but not Cadherin II, could also be a possible target of miR-9 in ascidians. In addition, it has been reported that Cadherin overexpression decreases the cytoplasmatic and nuclear levels of β-catenin, a transcription factor involved in endoderm specification. Disturbance in endoderm differentiation has been demonstrated to lead to differentiation of excess epidermis cells [54]. Thus, miR-9 could also act through Wnt pathways, participating in cell fate determination during ESN development.
Both bioinformatics programs used in this study predicted the same miR-9 target site in the FoxG (the forkhead transcription factor) transcript. Among tetrapods, the miR-9 target sequence at FoxG1 3 UTR (previously called BF-1) is highly conserved [41]. This gene is present in the C. intestinalis genome, however no data are available about FoxG expression and functions. In vertebrates, it is expressed in the telencephalic region of the neural plate and its expression persists in the adult telencephalon. During embryonic development, it has been shown to be essential in driving ventral telencephalic fate downstream of Hh/Fgf pathways and controlling the extent of the dorsal compartment through direct transcriptional repression of Wnt ligands [55]. In the cephalochordate, amphioxus AmphiBF-1 expression has been characterized, and notably, at 3 days of development, it is expressed in the anterior-ventral part of the cerebral vesicle [56]. Conservation of the FoxG1 expression pattern in cephalochordates and vertebrates suggests that this gene and its regulatory network could be evolutionarily conserved. However, the C. intestinalis FoxG target sequence shows only partial similarities with vertebrates, and further specific analyses are needed to solve this hypothesis.
Overall, our results showed miR-9 involvement in ascidian neural development, further underlining the evolutionary conservation of both miRNA sequences and miRNA-mediated post-transcriptional pathways between vertebrates and ascidians, which have already been suggested by other authors [40,52]. In addition, we described in vivo biological activity of a PNA oligomer directed against the miRNA-9 in C. intestinalis embryos. Our results suggest that anti-miR PNA-a9 is able to reach its specific target in the developing ascidian embryos with high efficiency, as underlined by the lack of effect induced by the scrambled sequence. This is further evidence that an unmodified PNA can be successfully used in knockdown strategies in the multicellular organisms of ascidian larvae.

Synthesis and Characterization of PNAs
PNA-a9 and PNA-sc9 were synthesized by manual solid-phase synthesis using Boc/Z chemistry [57]. The commercially available Boc/Z-protected PNA monomers were purchased from ASM Research Chemicals GmbH (Hannover, Germany). The MBHA resin was purchased from VWR International (Milan, Italy), and it was loaded manually to 0.2 mmol/g with Boc/Z-adenine PNA monomer for PNA-a9 and with Boc-thymine PNA monomer for PNA-sc9, following the previous procedure [58]. The PNA purification was performed by reverse-phase RP-HPLC with an Agilent 1200 Series system, equipped with a diode array detector (UV detection at 260 nm). The purity of PNA-a9 and PNA-sc9 was checked by RP-HPLC analyses, and their identities were confirmed by MALDI-TOF mass spectra, which were recorded with a Bruker Daltonics Omniflex (Milan, Italy), equipped with a 337 nm pulsed nitrogen laser (sinapinic acid as matrix

Animals and Embryos Microinjections
Ciona intestinalis adults were collected along the coasts of Roscoff by the fishing service of the Roscoff Biological Station (Roscoff, France). Animals were maintained in aquaria filled with artificial sea water (Instant Ocean; salinity about 32% ) and provided with a circulation system, as well as mechanical, chemical, and biological filters. Constant light condition was preferred to promote gamete production. Gametes from three adults were obtained surgically from the gonoducts and reared in artificial sea water buffered with 1M HEPES (ASWH; pH 8.0) at 18 ± 1 • C until the stages of interest. Before being fixed in 4% paraformaldehyde, 0.5 M NaCl, and 0.1 M 3-(N-morpholino)propanesulfonic acid, embryos at gastrula, neurula, and tailbud stages were dechorionated in ASW containing 1% sodium thioglycolate and 0.05% protease.
For PNA microinjections, only batches in which 90% or more of the eggs developed normally were used. Dechorionated eggs were microinjected with solutions of 0.7 mM and 1 mM PNAs (PNA-a9 and PNA-sc9) in distilled water, plus 5 µg/µL Fast Green as the vital dye, as described previously [59]. In vitro cross-fertilization was performed and embryos were reared at 16 • C in a 1% agarose-coated petri dish in ASW until they reached late tailbud and larva stages (16 and 22 h post fertilization) [60].
To confirm the result, a DIG riboprobe specific for pri-miR-9 was also synthetized. First, a PCR product of 578 bp was amplified from C. intestinalis genomic DNA using 5 -TACTGGAGCGTGTTAGGTTTATTG-3 and 5 -AGGGATGCCATGATTAGTAGTGAC-3 as forward and reverse primers, respectively. Then, a fragment of 357 bp was obtained by performing a nested PCR with 5 -AACAACGGCCGTATTGCTTT-3 (forward) and 5 -ACCCCAAATGCTGTTTCGTG-3 (reverse). After cloning with TOPO TA Cloning Kit (Thermo Fisher Scientific, Milan, Italy), DIG-labelled antisense and sense riboprobes were transcribed with Sp6 and T7 RNA polymerase, using a DIG RNA labelling kit (Roche, Monza, Italy).
Standard in situ hybridizations with riboprobes were performed [61]. To characterize the effects of PNA injections, the following neural marker genes were employed: Ci-Pans [37], Ci-POU IV [38], and Ci-V-Glut [39]. For each probe, at least 25 injected and control embryos were analyzed. For each experimental group, the percentages of samples with abnormal expression were calculated as (number of samples with altered expression / total number of analyzed samples) × 100.

Conflicts of Interest:
The authors declare no conflict of interest.