Therapeutic antidepressant potential of a conjugated siRNA silencing the serotonin transporter after intranasal administration

Major depression brings about a heavy socio-economic burden worldwide due to its high prevalence and the low efficacy of antidepressant drugs, mostly inhibiting the serotonin transporter (SERT). As a result, ~80% of patients show recurrent or chronic depression, resulting in a poor quality of life and increased suicide risk. RNA interference (RNAi) strategies have been preliminarily used to evoke antidepressant-like responses in experimental animals. However, the main limitation for the medical use of RNAi is the extreme difficulty to deliver oligonucleotides to selected neurons/systems in the mammalian brain. Here we show that the intranasal administration of a sertraline-conjugated small interfering RNA (C-SERT-siRNA) silenced SERT expression/function and evoked fast antidepressant-like responses in mice. After crossing the permeable olfactory epithelium, the sertraline-conjugated-siRNA was internalized and transported to serotonin cell bodies by deep Rab-7-associated endomembrane vesicles. Seven-day C-SERT-siRNA evoked similar or more marked responses than 28-day fluoxetine treatment. Hence, C-SERT-siRNA (i) downregulated 5-HT1A-autoreceptors and facilitated forebrain serotonin neurotransmission, (ii) accelerated the proliferation of neuronal precursors and (iii) increased hippocampal complexity and plasticity. Further, short-term C-SERT-siRNA reversed depressive-like behaviors in corticosterone-treated mice. The present results show the feasibility of evoking antidepressant-like responses by selectively targeting neuronal populations with appropriate siRNA strategies, opening a way for further translational studies.


INTRODUCTION
Major depressive disorder (MDD) is a severe, chronic and lifethreatening disease with a high incidence; affecting ca. 120 million people worldwide. [1][2][3] The midbrain serotonin (5-hydroxytryptamine (5-HT)) system has a critical role in many brain functions, including mood control. Derangements of serotonin pathway are involved in MDD, and most antidepressant drugs aim to increase serotonergic function. 4 Serotonin transporter (SERT) is a key player in MDD, by controlling the active 5-HT fraction and, being the target of most prescribed antidepressant drugs, the selective serotonin reuptake inhibitors (SSRI) and the selective serotonin and norepinephrine reuptake inhibitors (SNRI). 5,6 These drugs need to be administered for long time before clinical improvement emerges, and they fully remit depressive symptoms in only onethird of patients leaving a large proportion of people with partial or incomplete clinical responses. 7,8 For these reasons, there is an urgent need to improve antidepressant treatments.
Chronic-but not acute-SSRI treatments elicit a series of neurobiological changes relevant for antidepressant activity. Hence, chronic SSRI treatments downregulates SERT, increasing forebrain serotonergic neurotransmission and neuronal plasticity in the hippocampus, [9][10][11][12] although the precise mechanisms involved remain uncertain. Likewise, chronic SSRI treatments internalize SERT and reduce SERT-binding sites without affecting SERT mRNA levels. 9,10,13,14 In particular, fluoxetine (FLX) promotes the biogenesis of microRNA-16, resulting in a downstream repression of SERT levels in mouse 5-HT neurons, accompanied by antidepressant-like effects in the chronic mild stress and forced-swim animal models. 15 Altogether, these data uncover the functional significance of SERT downregulation in mediating antidepressant responses. The identification of intracellular networks underlying SERT downregulation may be a new target for the development of fast-acting antidepressants. Hence, exogenous small interfering RNAs (siRNAs) have been preliminarily investigated as potential therapeutic tools to silence the expression of critical genes in 5-HT neurons. [16][17][18] Intracerebral treatments with siRNA against SERT-or their related antisense oligonucleotides-significantly decreased SERT expression and function in the rodent brain and evoked cellular and behavioral responses predictive of clinical antidepressant activity. 16,17,19 Despite these exciting prospects, the utility of RNA interference (RNAi)-based silencing strategies for MDD treatment is severely compromised by the extreme difficulty to deliver oligonucleotide sequences to their neuronal functional sites, due to the need to cross several biological barriers after administration and the evident complexity of the mammalian brain. 20,21 Here we have used targeted delivery of a sertraline ligand-conjugated siRNA directed against SERT (C-SERT-siRNA) to downregulate SERT expression selectively in raphe 5-HT neurons. We show that C-SERT-siRNA silenced SERT expression/function and evoked fast and robust antidepressant-like responses after intranasal (i.n.) administration in mice. Moreover, it reversed the depressive-like behavior in corticosterone-treated mice due to the increased 5-HT signaling and synaptic plasticity. These results highlight the potential of RNAi-based antidepressant therapies targeting genes linked to antidepressant responses, such as SERT or the 5-HT 1A -autoreceptor 18 through a clinically feasible (i.n.) administration route.

MATERIALS AND METHODS Animals
Male C57BL/6J mice (10-14 weeks; Charles River, Lyon, France) were housed under controlled conditions (22 ± 1°C; 12-h light/dark cycle) with food and water available ad libitum. Animal procedures were conducted in accordance with standard ethical guidelines (EU regulations L35/118/ 12/1986) and approved by the local ethical committee.
To study in vivo intracellular distribution and incorporation of conjugated siRNA into 5-HT neurons, C-NS-siRNA was additionally labeled with alexa488 in the antisense strand (A488-C-NS-siRNA). We used C-NS-siRNA instead of C-SERT-siRNA to examine the brain distribution after i.n. administration because C-SERT-siRNA reduces SERT expression (see Results section), this compromising the penetration of new doses into 5-HT neurons through SERT. Along these lines, we assumed that the main factor conferring the neuronal target selectivity was the presence of covalently bound sertraline rather than the oligonucleotide sequence. Stock solutions of all siRNAs were prepared in RNAse-free water and stored at − 20°C until use. Sequences are shown in Supplementary Table S1.

Treatments
For i.n. administration, mice were slightly anesthetized by 2% isoflurane inhalation and placed in a supine position. 18 A 5-μl drop of phosphatebuffered saline (PBS) or conjugated siRNA (C-NS-siRNA and C-SERT-siRNA) was applied alternatively to each nostril once daily. A total of 10 μl of solution containing 30 μg (2.1 nmol day − 1 ) of conjugated siRNA was delivered for 1, 4 or 7 days, and mice were killed at 1, 3, 7 or 15 days after last administration. To evaluate the C-SERT-siRNA efficacy on SERT knockdown, mice were i.n. treated with the C-SERT-siRNA at 10, 30 or 100 μg day − 1 (0.7, 2.1 or 7 nmol day − 1 , respectively) during 7 days and were killed 24 h after last administration.
3 days and 7.5 μg ml − 1 (1.1 mg kg − 1 day − 1 ) during 10 or 31 days to allow a gradual recovery of endogenous corticosterone plasma level. 22,23 Cortico solutions were no more than 3 days old and mantained in opaque bottles to protect it from light. From day 21, one group of animals treated with Cortico received daily i.n. PBS, C-NS-siRNA or C-SERT-siRNA for 7 days. Another group of mice treated with Cortico received i.p. injections of saline or FLX for 7 or 28 days (Supplementary Figure S1).

Plasma corticosterone levels
Posttreatment corticosterone was measured in plasma obtained from blood samples after cardiac puncture at 1500-1600 hours using trisodium citrate as anticoagulant. After blood centrifugation at 1000 g for 15 min, all plasma samples were stored at − 80°C before assay by using a commercially available kit by radioimmunoassay of 125 I-labeled rat corticosterone (Coat-A-Count, Siemens, Healthcare Diagnostics, Berkeley, CA, USA). A gamma counter (Perkin Elmer Wallac Wizard 1470, Turku, Finland) was used to measure radioactivity of the samples. 24 In situ hybridization Mice were killed by pentobarbital overdose, and the brains were rapidly removed, frozen on dry ice and stored at − 80 ºC. Coronal tissue sections (14μm thick) were cut using a microtome-cryostat (HM500-OM, Microm, Walldorf, Germany), thaw-mounted onto 3-aminopropyltriethoxysilane Quantitative image analysis of film autoradiograms Autoradiograms were analyzed and relative optical densities (ROD) were obtained using a computer-assisted image analyzer (MCID, Mering, Germany). The system was calibrated with 3 H-or 14 C-microscales standards to obtain fmol mg − 1 protein equivalents from ROD data. The slide background and non-specific densities were subtracted. ROD were evaluated in two or three adjacent sections by duplicate of each mouse and averaged to obtain individual values. MCID system was also used to acquire pseudocolor images. Black and white photographs were taken from autoradiograms using a Wild 420 microscope (Leica, Heerbrugg, Germany) equipped with Nikon DXM1200F digital camera and ACT-1 Nikon software (Soft Imaging System Gmbh, Münster, Germany). Images were processed with Photoshop (Adobe Systems, Mountain View, CA, USA) by using identical values for contrast and brightness.

5-Bromo-2'-deoxyuridine (BrdU) administration
BrdU was purchased from Sigma-Aldrich and dissolved in saline solution. The last day of antidepressant treatment, mice were injected with 4 × 75mg kg − 1 BrdU, i.p., every 2h and were killed 24h later in according to Santarelli et al. 11 Immunohistochemistry Mice were anesthetized with pentobarbital and transcardially perfused with 4% paraformaldehyde in sodium-phosphate buffer (pH 7.4). Brains were collected, postfixed 24 h at 4°C in the same solution and then placed in gradient sucrose 10-30% for 3 days at 4°C. After cryopreservation, serial 30-μm thick sections were cut through the olfactory bulbs, the hippocampal formation, amygdala and the midbrain raphe nuclei. Immunohistochemical procedure was performed for SERT, BrdU, Ki-67, NeuroD, NeuN, Doublecortin (DCX), glial fibrillary acidic protein (GFAP) and Iba-1 using biotin-labeled antibody procedure. 17   Behavioral studies All mice were tested at 24h after treatments. All tests were performed between 1000 and 1500 hours. Behavioral assessments were examined with at least an interval of 1-2 days between tests. They were conducted in the following order: (1) open field test, (2) sucrose preference test, (3) novelty suppressed-feeding test, and (4) tail suspension test. On test days, animals were transported to a dimly illuminated behavioral room and were left undisturbed for at least 1 h before testing. Behavioral tests were conducted by an experimenter blind to mouse treatments. Details are shown in Supplementary Information. Intranasal delivery of conjugated SERT-siRNA A Ferrés-Coy et al

Statistical analyses
All results are given as mean ± s.e.m. Data were analyzed using GraphPad Prism 6.0 (GraphPad, San Diego, CA, USA). Statistical analyses were performed by two-tailed Student's t-test and one-way or two-way analysis of variance followed by Tukey's post-hoc test as appropriate. In novelty suppressed-feeding test, we used the Kaplan-Meier survival analysis due to the lack of normal distribution of the data. Animals that did not eat during the 10-min testing period were discarded. Mantel-Cox log-rank test was used to evaluate differences between experimental groups, as described by Samuels and Hen. 27 Differences were considered significant when Po0.05.
Completed statistical analyses are summarized in Supplementary Table S3.

RESULTS
Sertraline-conjugated siRNA is internalized into 5-HT neurons by endocytosis after i.n. administration We used a previously developed strategy of transporter-mediated neuronal delivery of siRNA, in which the SSRI sertralinewhich selectively binds to SERT-was chemically conjugated to the oligonucleotide. 18 The working hypothesis was that the presence of sertraline would allow the selective enrichment of this conjugated siRNA in 5-HT neurons, where the targeted transporter (SERT) is differentially expressed. 28 As first evidence in support of this mechanism, we previously showed that sertraline pretreatment (20 mg kg − 1 , i.p.) prevented the effects of sertraline-conjugated siRNA, indicating that conjugated siRNA molecules enter 5-HT neurons via SERT. 18 For localization purposes, we synthesized an alexa488-labeled sertraline-conjugated nonsense-siRNA (A488-C-NS-siRNA). Confocal fluorescence microscopy revealed that A488-C-NS-siRNA was intracellularly detected in TPH 2 -positive midbrain 5-HT neurons after i.n. administration (Figure 1a). Confocal analysis showed that A488-C-NS-siRNA molecules were more efficiently uptaken by TPH 2positive neurons in the DR than in median raphe nucleus (MnR) (Figures 1b and d and Supplementary Figure S2). In addition, A488-C-NS-siRNA was absent in cells of brain areas close to the application site (olfactory bulbs) or to brain ventricles (hippocampus) (Supplementary Figure S3), supporting that surface SERT expression is a requirement for oligonucleotide uptake and internalization. Sertraline-conjugated siRNA was possibly accumulated in 5-HT cells perhaps by endocytosis and entered in a complex network of trafficking pathways as vesicles containing A488-C-NS-siRNA co-localized with Rab5 (early endosome marker) and Rab7 (late endosome marker) (Supplementary Figure S4). Further studies are needed to fully characterize the route used by sertralineconjugated siRNA molecules to reach raphe 5-HT neurons.
Sertraline-conjugated SERT-siRNA induces selective and safe suppression of SERT expression We first examined the effect of i.n. C-SERT-siRNA administration (30 μg day − 1 ) on SERT expression. SERT mRNA and binding site levels were significantly lower in the raphe nuclei of C-SERT-siRNAtreated mice compared with control groups (PBS-and C-NS-siRNAtreated mice), with a maximal reduction after 7-day treatment. Immunohistochemistry analysis confirmed these results at the protein level (Figures 2a and d and Supplementary Figure S5). C-SERT-siRNA suppressed SERT expression more markedly in the DR than in MnR, as supported by the higher intracellular density of oligonucleotide in DR 5-HT neurons (Figure 1). We then evaluated the dose-related effects of C-SERT-siRNA (10-30-100 μg day − 1 during 7 days, i.n.) on SERT expression in the DR. All doses significantly decreased SERT mRNA expression 24 h after last administration, with no significant differences between doses, despite a slightly greater reduction at 30 and 100 μg day − 1 (Supplementary Figure S6). Next we assessed the temporal pattern of SERT decrease. For this purpose, mice were i.n. administered with C-SERT-siRNA at 30 μg day − 1 during 7 days and killed at different times after last administration. SERT mRNA level in the DR was significantly lower in C-SERT-siRNA-treated mice that in the control group at 1 and 3 days postadministration, with a recovery of SERT expression to control values at 7 and 15 days postadministration (Supplementary Figure S7). These values agree with data in the literature indicating a short half-life of SERT. 19 The reduced SERT expression was accompanied by a decreased function, as assessed by intracerebral microdialysis in the CPu, a DR-innervated area. Hence, i.n. C-SERT-siRNA treatment (30 μg day − 1 , 7-day) doubled basal extracellular 5-HT levels in the CPu versus PBS-treated mice (9.9 ± 1.6 and 4.7 ± 0.7 fmol fraction − 1 , respectively; n = 7-8; Po0.01) and provoked a lesser response of the SSRI citalopram to increase extracellular 5-HT (Figure 2e).
5-HT neuronal activity and 5-HT release, thus counteracting the facilitation of 5-HT transmission induced by SERT blockade. [29][30][31] This inhibitory feedback is an essential component of the delayed therapeutic action of antidepressant drugs. Long-term FLX treatment (10 mg kg − 1 , 28-day, but not 7-day, i.p.) attenuated 5-HT 1A -autoreceptor-mediated activation of G-proteins as expected 32 and prevented the effect of 8-OH-DPAT (selective 5-HT 1A receptor agonist) to reduce hippocampal 5-HT release. In contrast, 7-day C-SERT-siRNA administration (30 μg day − 1 ) was sufficient to downregulate 5-HT 1A -autoreceptors and to avoid the 8-OH-DPAT effect on 5-HT release (Figures 3c and d). Neither treatment modified the hippocampal [ 35 S]GTP-γ-S binding in the presence of 8-OH-DPAT (Supplementary Figure S10c). The reduction in SERT expression/function, together with the 5-HT 1Aautoreceptor downregulation, increased extracellular 5-HT concentration in the CPu and hippocampus more rapidly and markedly than after FLX (Figure 3e). The present results indicate that cellular changes-classically associated to long-term SSRI treatment-can be achieved after only 1 week of i.n. C-SERT-siRNA administration. These observations suggest that SSRI and C-SERT-siRNA distinctly regulate the molecular mechanisms controlling SERT function, resulting in a more rapid and efficient increase of presynaptic 5-Ht function with the RNAi strategy.
Sertraline-conjugated SERT-siRNA rapidly increases and facilitates maturation of newborn cells The clinical antidepressant action is also associated with neural stem cell proliferation, neurogenesis and the establishment of new synaptic contacts in brain circuits controlling motivation, emotion and cognition. 33 Table S4). Dendritic morphology of newborn cells was comparable after 7-day C-SERT-siRNA and 28-day FLX treatments, as indicated by Sholl analysis on DCX-positive neurons with tertiary dendrites, the number of intersections and dendrite length (Figures 4f and g). These data support that RNAi-induced SERT knockdown rapidly enhances neuronal plasticity in hippocampus as a result of the increased 5-HT signaling. These cellular changes were accompanied by a higher activation of neuroplasticity-associated genes in the hippocampus. Seven-day C-SERT-siRNA or 28-day FLX treatments increased comparably the expression of BDNF and its TRKB receptor as well as VEGF and ARC essentially in the dentate gyrus (Figure 4h and Supplementary Figure S12). Likewise, mRNA neuritin and PSD95 levels, two critical downstream mediators of antidepressant/BDNFinduced plasticity [35][36][37] were selectively increased in the dentate gyrus of both the mice groups. However, a 7-day FLX regime was without effect on these variables (Figure 4).
Short-term sertraline-conjugated SERT-siRNA efficiently attenuates the behavioral alterations in the corticosterone depression model Finally, we confirmed the potential therapeutic benefit of the RNAi-induced downregulation of SERT expression in a stressrelated model of depression: the corticosterone model-a wellestablished stress inducer. 12,22 Despite displaying plasma corticosterone levels and open field behavior similar to controls, mice exposed to a low oral dosage of corticosterone for 28 or 49 days showed a persistent depressive-like behavior, characterized by reduced sucrose preference, increased latency in the novelty suppressed feeding paradigm and increased immobility time in the tail suspension test (Supplementary Figure S13). These depressive-like behaviors were reversed to the same extent by 7-day i.n. C-SERT-siRNA (30 μg day − 1 ) and by 28-day FLX (10 mg kg − 1 ) treatments ( Figure 5). In contrast, 7-day FLX did not evoke any antidepressant-like effects.

DISCUSSION
Here we show that C-SERT-siRNA evokes very fast (7-day) and robust antidepressant-like responses in control and corticosterone-treated mice, comparable than those evoked by a 28-day FLX treatment. Prior studies using RNAi strategies to elicit antidepressant-like responses in rodents used unmodified siRNA sequences directed against SERT or 5-HT 1A receptors. 16,17,25 The study confirms and extends our previous observations on the use of sertralineconjugated siRNA sequences to silence genes in 5-HT neurons. 18 The design of the conjugated siRNA allows its selective enrichment in 5-HT neurons after i.n. administration, opening new ways for the therapeutic use of RNAi strategies to treat mood disorders, which may overcome the limitations of standard antidepressant treatments, that is, slow clinical action and low efficacy.
As a consequence of the fast and effective SERT downregulation, 7-day C-SERT-siRNA treatment (i) reduced 5-HT 1Aautoreceptor expression/function, (ii) facilitated forebrain serotonin neurotransmission, (iii) accelerated the proliferation of neuronal precursors, (iv) increased the expression of growth factors (for example, BDNF, VEGF) and genes promoting neurite outgrowth (for example, neuritin, PSD95) and (v) increased hippocampal dendritic complexity and synaptic plasticity. All these variables are predictive of clinical antidepressant action. In addition, short-term C-SERT-siRNA treatment normalized stress-induced depressive-like behaviors, which were only sensitive to 28-day FLX treatment. Interestingly, the above actions were produced by the i.n. administration of very small doses (2.1 nmol day − 1 ) of C-SERT-siRNA illustrating the effectiveness of the present RNAi strategy.
The precise mechanism(s) used by C-SERT-siRNA reach 5-HT neurons are not fully understood. Unlike other strategies used for oligonucleotide delivery to target neuronal populations, 38,39 here we used a transporter (SERT)-mediated process to target a selective gene (SERT) expressed in 5-HT neurons, as previously used to silence 5-HT 1A -autoreceptors. 18 Paradoxically, this approach could limit the extent of the intracellular C-SERT-siRNA accumulation, as the conjugated siRNA enters 5-HT neurons via SERT. 18 However, the present results indicate that the remaining expression of SERT-even after 7-day C-SERT-siRNA treatment-is sufficient to evoke a very large increase of presynaptic serotonergic function.
Moreover, as trans-nasal oligonucleotide delivery to brain is mediated by extracellular mechanisms, [40][41][42] it is also possible that C-SERT-siRNA can use this extracellular pathway before being taken up by serotonergic terminals and transported back to cell bodies in the midbrain. This view is showed by the association of the conjugated siRNA to Rab7, supporting traffic via late endomembrane compartments.
The present results indicate that C-SERT-siRNA is preferentially accumulated into DR (versus MnR) 5-HT neurons. Anatomical and functional differences between the two 5-HT subsystems have been reported. [43][44][45] In particular, SERT expression is greater in DR than in MnR and DR-innervated areas are more sensitive to the action of SSRI, 40 which suggests a preferential uptake of C-SERT-siRNA by DR axons.
Consistent with previous findings using intra-raphe SERT-siRNA infusion, 17 i.n. C-SERT-siRNA treatment triggered a complex cascade of signaling events that ultimately results in downregulation of SERT and also 5-HT 1A -autoreceptors on midbrain serotonin neurons. As both mechanisms tightly control the active 5-HT fraction, their downregulation by C-SERT-siRNA dramatically increased the extracellular 5-HT levels in the forebrain in a faster way than that produced by the pharmacological SERT blockade with FLX. The intracellular mechanisms by which short-term C-SERT-siRNA or chronic SSRI treatments downregulate SERT and 5-HT 1A receptor levels to increase the serotonergic neurotransmission are still poorly known. Recent reports suggest a role for altered patterns of gene expression in mediating the long-term therapeutic effects of SSRIs, focusing on the potential involvement of microRNAs as fine-tuners and on-off switches of gene expression. 15,46 Hence, miR-135 levels were upregulated after single and chronic FLX administrations, 47 suggesting that this microRNA may act as an endogenous homeostatic mechanism to maintain the physiological balance between SERT and 5-HT 1Aautoreceptors.
Similarly to intra-raphe SERT-siRNA application, 17 the short-term i.n. administration of C-SERT-siRNA-but not FLX-evoked hippocampal postsynaptic responses to the enhanced serotonergic signaling, which are predictive of clinical antidepressant effects. These responses included the proliferation of progenitor cells, stimulation of dendritic branching, acceleration of the maturation of immature DCX neurons and increased expression of spine/ synapse markers. Moreover, these postsynaptic changes were accompanied by the reversal of the behavioral deficits caused by prolonged corticosterone exposure. The C-SERT-siRNA ability to increase synaptic plasticity may be mediated by an enhancement of BDNF/TRKB signaling, among others that were upregulated after 7-day C-SERT-siRNA treatment. Activation of their downstream signaling pathways was shown to enhance maturation and dendritic development. 48 Induction of neuritin and PSD95 are also indicative of an increased synapse formation and function. 37 However, additional antidepressant mechanisms may be involved, resulting from the enhanced 5-HT neurotransmission in other brain areas implicated in MDD, such as the ventromedial prefrontal cortex.
In conclusion, our findings indicate that RNAi-based SERT repression elicits faster and more effective antidepressant-like actions than persistent SERT blockade with FLX. Enhancing serotonin signaling through downregulation of SERT expression evokes standard antidepressant responses, promotes the generation of new hippocampal neurons, increases synaptic plasticity and counteracts behavioral deficits in a stress-induced depression model. Furthermore, the study has a high translational value due to the use of a clinically feasible (i.n.) administration route. [40][41][42] This opens new therapeutic perspectives for the treatment of mood disorders, including MDD.

CONFLICT OF INTEREST
FA has received consulting and educational honoraria on antidepressant drugs from Lundbeck and he is PI of grants from Lundbeck. He is also a member of the advisory board of Neurolixis Inc. AB and FA are authors of the patent WO/2011/131693 for the siRNA and ASO (antisense oligonucleotides) molecules and the targeting approach related to this work. GA and AM are board members of nLife Therapeutics S.L. The rest of authors declare no competing financial interest.