A Mouse Model for Conditional Secretion of Specific Single-Chain Antibodies Provides Genetic Evidence for Regulation of Cortical Plasticity by a Non-cell Autonomous Homeoprotein Transcription Factor

During postnatal life the cerebral cortex passes through critical periods of plasticity allowing its physiological adaptation to the environment. In the visual cortex, critical period onset and closure are influenced by the non-cell autonomous activity of the Otx2 homeoprotein transcription factor, which regulates the maturation of parvalbumin-expressing inhibitory interneurons (PV cells). In adult mice, the maintenance of a non-plastic adult state requires continuous Otx2 import by PV cells. An important source of extra-cortical Otx2 is the choroid plexus, which secretes Otx2 into the cerebrospinal fluid. Otx2 secretion and internalization requires two small peptidic domains that are part of the DNA-binding domain. Thus, mutating these “transfer” sequences also modifies cell autonomous transcription, precluding this approach to obtain a cell autonomous-only mouse. Here, we develop a mouse model with inducible secretion of an anti-Otx2 single-chain antibody to trap Otx2 in the extracellular milieu. Postnatal secretion of this single-chain antibody by PV cells delays PV maturation and reduces plasticity gene expression. Induced adult expression of this single-chain antibody in cerebrospinal fluid decreases Otx2 internalization by PV cells, strongly induces plasticity gene expression and reopens physiological plasticity. We provide the first mammalian genetic evidence for a signaling mechanism involving intercellular transfer of a homeoprotein transcription factor. Our single-chain antibody mouse model is a valid strategy for extracellular neutralization that could be applied to other homeoproteins and signaling molecules within and beyond the nervous system.


Introduction
During postnatal life, the cerebral cortex passes through critical periods (CPs) of plasticity allowing the neuronal circuitry to shape in response to environmental stimuli. CPs are driven by the maturation of a subset of inhibitory interneurons, the fast-spiking parvalbumin-expressing GABAergic neurons (PV cells), present in layers III and IV of the cerebral cortex [1]. Plasticity terminates with the full maturation of PV cells and the consolidation of the Excitation/ Inhibition (E/I) cortical balance [2]. CPs for different sensory, motor or cognitive behaviors are spread out during postnatal development and thus open and close at different times [3]. In the mouse, plasticity for the establishment of binocular vision opens at post-natal day 20 (P20) and closes at P40 [4]. Closing one eye during this period (but not before P20 or after P40) leads to an irreversible loss of visual acuity for the temporarily closed eye, a state known as amblyopia, which affects 3% of the human population.
The transfer of the homeoprotein (HP) transcription factor Otx2 from extra-cortical sources into PV cells of the primary visual cortex (V1) regulates CP timing [5]. Cortical infusion of recombinant Otx2 accelerates both CP onset and closure. Recently, the choroid plexus was identified as one of these sources, as Otx2 knock-down specifically in the choroid plexus reopens a window of plasticity in the adult V1 [6]. Reopening plasticity in the adult was also achieved by blocking Otx2 binding to complex sugars embedded in perineuronal nets (PNNs) that surround PV cells, thus reducing its internalization [7,8]. Together, these results led to the important concept that Otx2 accumulation by PV cells leads to a first concentration threshold that opens plasticity at P20 and to a second threshold that closes plasticity at P40; the maintenance of a non-plastic adult state requires the continuous internalization of this HP [9,10].
In this report we aimed at developing genetic evidence for direct non-cell autonomous regulation of plasticity by Otx2 in the mouse. An obvious approach would be the production of a "cell autonomous-only" mouse through the mutation of Otx2 import and/or export sequences. This strategy is precluded by the fact that the latter sequences are part of the DNA-binding domain and cannot be modified without modifying the cell autonomous transcription of Otx2 targets [11,12]. Based on previous studies on non-mammalian embryonic development [13][14][15][16], we turned to a strategy that employs the extracellular expression of neutralizing single-chain variable fragment (scFv) antibodies. We generated conditional knock-in mouse lines (scFvOtx2 tg/o and scFvPax6 tg/o ) expressing secreted scFv antibodies directed against Otx2 or Pax6 (as a control). Targeted expression of anti-Otx2 scFv in juvenile or adult mice resulted in the delay or reactivation of plasticity genes, respectively. This strategy provides the first genetic evidence for HP transfer in mammals and will allow for precise genetic manipulation of noncell autonomous homeoprotein activities to study their functions in embryonic and postnatal development.

Results
We generated knock-in mouse lines expressing inducible secreted Myc-tagged anti-Otx2 or anti-Pax6 single-chain antibodies (scFvOtx2 tg/o and scFvPax6 tg/o ) designed to neutralize HPs in the extracellular space without affecting cell autonomous functions. Tagged anti-Otx2 and anti-Pax6 scFv constructs were generated from hybridoma, cloned behind a floxed STOP cassette, and knocked into the ROSA26a locus ( Fig 1A).
To validate scFv expression, fibroblasts cultured from adult scFvOtx2 tg/o ear biopsies were treated with the cell-permeable Cre bacterial recombinase Cre-TAT [11]. Cells and media were collected 48h later and analyzed by Western blot (Fig 1B). Anti-Otx2 scFv was only present in cells treated with Cre-TAT and in the medium collected from these cells. Intracellular control RhoA confirms cell integrity. The anti-Myc antibody revealed two bands in the cell extract and in the medium. The lower band corresponds to the scFv recombinant protein and the upper band to the scFv-GFP fusion protein, establishing that the P2A skipping peptide was not 100% efficient. The slightly lower mobility of the antibodies retrieved from the extracellular milieu is due to differences in glycosylation [17].
The anti-Pax6 scFv functional activity has been demonstrated earlier [13,16]. To examine whether secreted anti-Otx2 scFv binds Otx2 specifically, the media of scFvOtx2 tg/o fibroblasts treated or not with Cre-TAT were used as a source of primary antibodies. Otx2 is recognized by the medium of scFvOtx2 tg/o fibroblasts treated with Cre-TAT ( Fig 1C). To confirm scFv in vivo activity, we crossed scFv mice with PV::Cre mice, where the Cre recombinase is driven by the PV promoter [18], and extracted adult cerebellum, which expresses both HPs [19] as well as PV. The anti-Otx2 scFv co-immunoprecipitated with Otx2 and, conversely, Otx2 co-immunoprecipitated with the scFv (Fig 1D). Association with Otx2 protein was not found in PV::Cre or PV::Cre;scFvPax6 tg/o cerebellar extracts.
Two models were developed to disrupt non-cell autonomous Otx2 in V1. A first model was based on local scFv secretion by PV cells. A second model consisted in inducing scFv secretion in the cerebrospinal fluid (CSF) from the choroid plexus, which is an endogenous source of extracellular Otx2 [6]. Pax6 is poorly expressed in the postnatal cortex and choroid plexus [20,21], making scFvPax6 tg/o mice an appropriate negative control. To induce scFv expression by PV cells, scFv mice were crossed with PV::Cre mice, where recombination is driven by the PV promoter starting at around P10 in V1 [1]. At P30, PV cells of PV::Cre;scFvOtx2 tg/o and PV:: Cre;scFvPax6 tg/o mice but not of PV::Cre mice express scFvs in the V1 binocular zone (Fig 2A  and S1 Fig). While most PV cells express the scFv, curiously some cells not stained with the PV antibody also express the scFv. This may result from poor PV expression at P30 or by the capture of secreted scFv by PV-negative cells. To induce scFv expression in the choroid plexus and subsequent secretion into the CSF, adult scFvPax6 tg/o and scFvOtx2 tg/o mice were injected with Cre-TAT in the lateral ventricles [6]. Western blot analysis of the CSF confirmed the induction and secretion of both the scFv and scFv-GFP proteins 5 d post-injection ( Fig 2B).
The accumulation of Otx2 in V1 PV cells controls CP timing through the maturation of PV cells [10]. CP kinetics are paralleled by increasing PV expression and assembly of extracellular matrix perineuronal nets (PNNs) revealed by the WFA lectin [22]. To follow PV cell maturation, P30 V1 sections were stained with antibodies against Otx2 and PV and with fluorescent WFA (Fig 3A). PV::Cre and PV::Cre;scFvPax6 tg/o mice were undistinguishable for all criteria used. In contrast, PV::Cre;scFvOtx2 tg/o mice showed a decrease in the number of PV cells containing detectable amounts of Otx2 ( Fig 3B). In addition scFv expression reduced by~20% the number of cells decorated with Otx2 and PV antibodies or with WFA ( Fig 3C). Intensity of Otx2 staining was also reduced by 20%. The number of GABAergic neurons and Calretinin (CR) neurons were not modified by scFv induction (Fig 3C and S2 Fig). This possibly reflects limited antibody diffusion and is in agreement with the fact that Otx2 is primarily captured by PV cells and does not influence the maturation of other interneurons [5]. Thus, the anti-Otx2 Due to the existence of thresholds, small changes in Otx2 concentrations within PV cells can have dramatic effects on V1 plasticity [6,7]. Plasticity is paralleled by the expression of plasticity genes in V1 layer IV such as Arc, Fos, Nr4a1 and Egr4 [23][24][25][26][27][28]. Quantitative mRNA analysis of V1 layer IV extracts at P30 demonstrated that the expression of the anti-Otx2 scFv, in addition to delaying PV cell maturation, significantly decreased the up-regulation of Fos, Egr4 and Nr4a1 transcripts (Fig 3D). Arc and c-Fos are expressed throughout V1 layers, and   two markers of PV cell maturation were reduced by 20 to 30%, with no change in the percentage of CR neurons or of the entire population of GABAergic neurons (Fig 4A and 4B and S2  Fig). In parallel, plasticity transcript expression was dramatically enhanced in V1 layer IV ( Fig  4C), suggesting induced brain plasticity similar to the effect obtained by recombination of Otx2 in the adult choroid plexus [6].
To verify if the reactivation of plasticity genes was reflected at a physiological level, we used optical imaging of intrinsic signals to measure visual acuity [29,30]. Two weeks after intra-ventricular injection of Cre-TAT or NaCl, adult (P105) scFvOtx2 tg/o mice underwent 4 d of monocular deprivation (MD) prior to optical imaging in the opposite hemisphere ( Fig 5A). The responses to different spatial frequencies were measured and quantified, as illustrated in Fig 5B  and 5C. The expression of anti-Otx2 scFv enhances visual acuity of the ipsilateral eye ( Fig 5D) but does not modify that of contralateral eye (Fig 5E), leading to a shift in ocular dominance for the highest spatial frequencies (above 0.14 cpd, Fig 5F).

Discussion
The possibility of the anti-Otx2 scFv neutralizing cell autonomous Otx2 cannot be ignored even though the amount of antibody that might escape the secretion pathway is low and the disulfide bond necessary for antibody activity is reduced in the intracellular milieu [31,32]. This risk can be excluded when PV cells express the antibody, given that the Otx2 locus is not active in these cells. In contrast, the issue can be raised for the choroid plexus, which strongly expresses Otx2. However, we have verified that potential Otx2 target genes in the choroid plexus are not quantitatively modified in the recombined scFvOtx2 tg/o plexus. For example an identified target gene is C1qL3 which, compared to GAPDH, is up-regulated from 1.01 ± 0.11 (n = 3) to 5.21 ± 0.62 (n = 3) in the plexus of Otx2 +/mice and unchanged 1.40 ± 0.07 (n = 3) in the scFv mouse choroid plexus.
The consequences of expressing the anti-Otx2 scFv under the control of the PV promoter are only a modest reduction of Otx2 content, PV expression and PNN assembly at P30. One likely reason is that PV expression starts at around P10 [1] and that it might take some time before the concentration of extracellular antibody is high enough to completely block Otx2 internalization. Another possibility is that the high affinity (in the nM range) of Otx2 for PNN complex sugars [7] partially protects Otx2 from being neutralized by the antibody. It is nevertheless clear that the delayed time course in PV cell maturation is enough to delay the expression of plasticity genes.
A similarly modest reduction in cortical Otx2 content occurred after anti-Otx2 scFv expression was induced in the adult choroid plexus. Nonetheless, the level of reduction is comparable to what was previously obtained when Otx2 expression was knocked down in the adult choroid plexus, which resulted in the reopening of V1 binocular plasticity [6]. This suggests that the antibody complexes most of the extracellular transcription factor in the CSF. The remaining Otx2 still present in PV cells after 2 weeks either reflects the stability of protein present before scFv expression or the existence of sources other than the choroid plexus, in particular the eye [5,33] and the pineal gland, another established site of Otx2 expression [34]. Regardless, this modest reduction is enough to induce a dramatic re-expression of plasticity genes in the adult and an enhancement of visual cortex plasticity. The plasticity observed after 4 d of MD in P105 mice corresponds to an increase in ipsilateral visual acuity with no change to the deprived contralateral eye. Indeed, Sato & Stryker [30] previously showed that adult plasticity mainly impacts the response of the non-deprived eye. Furthermore, they showed that 7 d of MD were necessary to induce moderate plasticity in the cortex of adult (P90) mice, while only 4 d were required to induce strong plasticity in juvenile (P30) mice. Thus, Otx2 capture by the scFv antibody in adult (P105) mice accelerates ocular dominance plasticity.
Intercellular transfer has been observed in vivo and/or in vitro for many HPs and the conservation of the transfer sequences in the DNA-binding domain suggests that this is a general property of this transcription factor family. Our single-chain antibody strategy brings clear genetic evidence for Otx2 signaling and will be generally useful to verify HP transport in vivo and to analyze the roles of this novel mode of signal transduction. This same strategy could indeed be applied to the neutralization of any signaling entity within and beyond the nervous system.
This report demonstrates that the extracellular neutralization of Otx2 by the inducible expression of an anti-Otx2 single-chain antibody leads to a reduction in Otx2 capture by PV cells and either delays their maturation or "rejuvenates" them depending on when the antibody is expressed. In both cases, several genes associated with a plastic state are regulated in the expected direction. This provides the genetic evidence for a novel signaling mechanism operating through homeoprotein transfer in the mouse.

Ethics statement
All animal procedures, including housing, were carried out in accordance with the recommendations of the European Economic Community (86/609/EEC), the French National Committee (87/848) and French bylaws (AGRG1240332A / AGRG1238724A / AGRG1238767A / AGRG1238729A / AGRG1238753A). For surgical procedures, animals were anesthetized with Xylazine (Rompun 2%, 5mg/kg) and Ketamine (Imalgene 500, 80mg/kg). For biochemical analysis, mice were sacrificed by cervical elongation. This research (project no. 00704.02) was approved by Ethics committee n°59 of the French Ministry for Research and Higher Education.

scFv construction
Hybridoma expressing monoclonal antibodies (CD4 and CD7) against Otx2 were generated, and CD7 hybridoma mRNAs were used to construct scFv minigenes as described in [35]. CD7-scFv coding sequence was subcloned as described in [16], with an N-terminal signal sequence and C-terminal 6xMyc and His tags followed by a skipping peptide (P2A)-eGFP cassette for expression of either a fusion protein (scFv-myc-his-P2A-GFP) or two separate proteins, depending on P2A peptide efficiency. Anti-Pax6 scFv [13] was transferred into the same final vector.

Animals
scFvOtx2 and scFvPax6 knock-in mouse lines were generated by the Institut Clinique de la Souris (Strasbourg, France) by transgenesis with targeted insertion in the Rosa26 allele. PV-Cre mice were obtained from The Jackson Laboratory (stock 8069).

Cell culture of ear fibroblasts
Fibroblasts were generated from ear biopsies from adult scFvPax6 tg/o and scFvOtx2 tg/o mice and treated with 10 μM cell-permeable Cre (Cre-TAT, production described in [6]) in DMEM [11]. Cells and culture media were collected 48 h later and processed for Western blot analysis. Conditioned media were also used as primary antibodies on Western blots.

Cre-TAT protein injection
Stereotaxic injection (bregma: x = -0.58 mm, y = ±1.28 mm, z = 2 mm) of Cre-TAT in 15% DMSO into the lateral ventricles of anesthetized mice was performed with a Hamilton syringe at a rate of 0.2 μl/min. Five or fifteen days after injection, mice were processed for cerebrospinal fluid sampling (described in [6]), cortical layer dissection or immunohistochemistry.

Layer IV dissection
The dissection of tissue enriched in cortical layer IV was performed in ice cold PBS with micro scalpels. Visual cortex areas were first excised from both brain hemispheres and then cut coronal to produce two thin rectangular strips per hemisphere. These strips were laid on their sides to expose the cortical layers. A lengthwise cut in two equal parts gave the supragranular layer, which was then cut lengthwise in two equal parts to provide enriched layer IV.

In situ hybridization
In situ hybridization on cryosections (20 μm) were performed as previously described [36]. Briefly, hybridization was carried out overnight at 70°C with DIG-labeled probes. Sections were washed and incubated with alkaline phosphatase-conjugated anti-DIG (1/2000, Roche) overnight at 4°C. Sections were then washed and the color reaction was carried out at room temperature. Reaction was stopped by PBS washes and sections were mounted with Fluoromount (Southern Biotech). Mosaïc images (10x magnification) were acquired with a Nikon 90i microscope.

Quantitative PCR
Layer IV of visual cortex was manually dissected and RNA was extracted using the RNeasy Lipid Tissue Mini (Qiagen) according to manufacturer's instructions. cDNA was synthesized from 500 ng of total RNA with the QuantiTect Reverse Transcription kit (Qiagen). For realtime PCR, triplicates were analyzed with a LightCycler 480 II (Roche). Gene-to-GAPDH ratios were determined using the 2 -ΔΔCt method.

Monocular deprivation and surgery for optical imaging
For monocular deprivation, mice were anesthetized prior to suturing of the left eye as described in [37]. Animals were checked daily to ensure sutures remained intact. The eye was opened immediately before recording.
For optical imaging, mice were anesthetized with urethane (1.2 g/kg, i.p.) and a sedative (chlorprothixene, 8 mg/kg i.m.). Atropine (0.1 mg/kg) and dexamethasone (2 mg/kg) were injected subcutaneously and body temperature was maintained at 37°C. Mice were placed in a stereotaxic instrument and craniotomy was made over the primary visual cortex in the hemisphere contralateral to the deprived eye. The exposed area was covered by 2.5% agarose and a glass coverslip.

Visual stimulation
Visual stimuli were randomly displayed on a 21" LCD monitor located 20 cm in front of the animal. Coordinates of the monitor were reprogrammed in order to maintain spatial and temporal frequencies constant across the visual field's eccentricities (See examples in Fig 5C, [38]). Each stimulus consisted of oriented sine-wave gratings drifting in one direction perpendicular to the orientation at a temporal frequency of 1.5 Hz. Ten spatial frequencies (SFs) were tested, ranging from 0.0303 to 1 cycle per degree (cpd) in a logarithmic scale. Monocular stimuli were displayed for 5s. A grey screen was presented between trials for 10s. Each SF was displayed 16 times with a different orientation/direction.

Intrinsic signal optical imaging
Optical images of intrinsic signals were recorded with a Dalsa 1M60 CCD camera controlled with an Imager 3001 system (Optical Imaging Ltd). Images were acquired after a 2x2 pixels spatial binning with a resolution of * 10.5 μm/pixel using a 135x50 mm tandem lens (Nikon) configuration. The focal plane was set 400 μm below the cortical surface, and intrinsic signals were acquired with a 700 nm illumination wavelength. The size of the recorded region was * 2.5 x 3.6 mm. Five frames of 1s duration each were stored right after the stimulus onset.
For data analysis, the last three recorded frames, which correspond to the strongest intrinsic signal, were averaged and divided by the first recorded frame. The generalized indicator function method was applied to identify the stimulus-related activity patterns [38]. An additional low-pass filter with a Gaussian kernel of 3 pixels SD was applied to smooth the data. The ipsilateral eye was stimulated, the area of interest was delineated over the primary visual cortex and the SF tuning curve in the binocular zone was calculated by averaging the response for each SF stimulus. This curve was least-square fitted to extract visual acuity. For each eye the response integral (trapezoidal approximation) elicited by SFs was calculated. Ocular dominance values ranged between -1 and 1 (ipsilateral and contralateral eye response, respectively) and calculated for each SF threshold with the contrast formula from the response integral of each eye.

Statistical analysis
Statistical analysis was performed with Prism software (GraphPad). Pairwise comparison was done using Student t-test.