**3. Discussion**

PNAs are synthetic mimics of natural oligonucleotides, which bind complementary DNA/RNA strands with high sequence specificity [3]. In comparison with DNA/DNA or DNA/RNA duplexes, PNA/DNA and PNA/RNA duplexes show excellent thermal stability due to the neutral PNA backbone that lacks a repulsion charge when binding with DNA or RNA, providing more stable complexes [5]. Although they display numerous advantages, in vivo applications are still rare due to their poor cellular uptake. The latter, however, could be overcome by using experimental models, such as the ascidian

*C. intestinalis*. In this model organism, microinjections allow direct delivery of antisense molecules into eggs, perturbing gene activity during embryonic development [16]. To verify this hypothesis, we chose miR-7 as PNAs target. miRNAs are potent endogenous regulators of gene expression with fundamental roles in development [21].

In *C. intestinalis*, hundreds of miRNAs have been identified (www.mirbase.org) but expression data are reported only for miR-124 [30]. In fact, characterization of miRNAs expression is extremely challenging due to their tiny size and low level of expression. When performing in situ hybridization, the possibility to get nonspecific signals is rather high. Thus, we first optimized the hybridization protocol with DIG-labeled LNA (Exiqon) probes to obtain specific staining even when miRNA levels are particularly low, as in miR-7 s case. In particular, modifications in hybridization temperature and incubation time were found to highly improve miRNAs detection.

Employing this protocol, we described for the first time miR-7 expression during ascidian development. In *C. intestinalis*, miR-7 mature transcripts were detected at late tailbud and larva stages in the central nervous system, particularly, in the ventral posterior part of the sensory vesicle (Figure 2C–E). miR-7 neural expression has been reported in different animal models, specifically in photoreceptors and/or neurosecretory tissues [24,26,33]. miR-7 is considered part of the evolutionary conserved fingerprint of neurosecretory cells [24]. In *C. intestinalis*, the ventral region of the larval sensory vesicle has been proposed to be homologous to vertebrate hypothalamus and retinal amacrine cells [34–36]. Thus, miR-7 expression also appears highly conserved in ascidians. Its expression domain, together with that already reported for Ci-Rx [37], Ci-Nk2, and Ci-Otp, [35] further supports the homology of this region to vertebrate hypothalamus [24,33,35]. Moreover, miR-7 was reported to be expressed in mammalian retina; and in ascidians, miR-7 was identified in the photoreceptive neuroepithelium [36], further indicating the striking evolutionary conservation of this miRNA among chordates.

In addition, miR-7 genomic position seems extremely conserved. In C. *intestinalis* genome, miR-7 was found within the last intron of the heterogeneous nuclear ribonucleoprotein K (hnRNP K) gene, similarly to those already reported for hsa-miR-7-1 in *Homo sapiens* [38] and in *Drosophila* [39].

hnRNP K is referred as a ubiquitously expressed gene involved in different aspects of RNA functions: transcription, editing, processing, and translation [40]. During vertebrate embryogenesis hnRNP K is uniformly expressed in all blastomeres; and only at later stages, transcripts appear more concentrated in specific tissues, such as central nervous system and mesodermal derivatives [41]. In *C. intestinalis*, Ci-hnRNP K displays a similar expression during early developmental stages; while at the late tailbud and larva stages, transcripts accumulate in the epidermal sensory neurons (Figure 2F–H), i.e., ascidian peripheral nervous system. Like in vertebrates, Ci-hnRNP K expression does not overlap with that of its host miR-7. Since genes are oriented in the same direction (Figure 1D), post-transcriptional regulation of miR-7 biogenesis is likely to occur, as demonstrated in both mice and humans [38].

Then, we designed a 22-mer PNA complementary to *C. intestinalis* miR-7 (PNA-a7, Figure 1C) as well as a PNA scrambled sequence (PNA-sc7, Figure 1C), to verify the specificity of the interaction between PNA-a7 and miR-7. Our hybridization analysis confirmed that PNA-a7 efficiently downregulates miR-7 and that PNAs interaction occurs in a sequence-specific manner, as samples injected with PNA-sc7 never showed a similar signal reduction (Figure 3). To further verify PNAs effectiveness, we compared the effects induced by miR-7 downregulation by injecting the commercial AntagomiR molecules, commonly used in miRNAs knockdown studies (AmiR-7) [28], and our PNAs. We checked the expression of two pan-neural genes, Ci-ETR [31] and Ci-Syn [32]. Ci-ETR signal was not affected by neither the PNAs nor AmiR-7 injection, suggesting that perturbation of miR-7 expression does not affect nervous system differentiation. On the contrary, Ci-Syn signal was reduced in the posterior neural tube of embryos injected with PNA-a7 and AmiR-7 (Figure 4F,H), but not in those injected with PNA-sc7. Synapsins are neuronal phosphoproteins that constitute a small family of synaptic molecules, specifically associated with synaptic vesicles. They exert a key role in neurite outgrowth and synapse formation [32]. In human neural embryonic stem cells, miR-7 overexpression during their neuronal differentiation increased synapsin expression. Synapsin mRNA

is not a direct target of miR-7 but miR-7 levels are positively correlated with synapsin expression, suggesting that this miRNA could act upstream of the synapsin pathway, and play an important role in synaptic development [39]. Similarly, in *C. intestinalis* embryos, Ci-Syn expression decreased after miR-7 knockdown, indicating that miR-7 has a functional role in synaptic plasticity and neurite elongation in ascidians.

Comparable results were obtained with both PNA-a7 and AmiR-7, confirming the reliability of our results and the specificity of PNAs. These are in agreement with different in vitro research, which demonstrated PNAs specificity for their complementary miRNAs [42]. In vivo experiments using modified PNAs have been performed in mouse: miR-155 inhibition by PNAs was demonstrated to be sequence specific, not affecting levels of unrelated miRNAs and mainly recapitulating the effects of genetic deletion of miR-155 [8].
