Immunoprecipitation and mass spectrometry identify non-cell autonomous Otx2 homeoprotein in the granular and supragranular layers of mouse visual cortex

Plasticity in the visual cerebral cortex is regulated by the internalization of Otx2 homeoprotein into parvalbumin neurons in cortical layers II/III and IV. However the Otx2 locus is not active in these neurons and the protein is imported from external sources, including the choroid plexus. Because Otx1 and Otx2 may have redundant functions, we wanted to verify if part of the staining in parvalbumin neurons corresponds to Otx1 transported from cortical layer V neurons. It is demonstrated here that Otx staining in layer IV cells is maintained in Otx1-null mice. The immunoprecipitation of extracts from finely dissected granular and supragranular cortex (layers I-IV) gave immunoblots with a band corresponding to Otx2 and not Otx1. Moreover, high-resolution mass spectrometry analysis after immunoprecipitation identifies two peptides within the Otx2 homeodomain. One of these peptides is specific for Otx2 and is not found in Otx1. These results unambiguously establish that the staining in parvalbumin neurons revealed with the anti-Otx2 antibodies used in our previous studies identifies non-cell autonomous Otx2.

Neural circuits generated during embryonic development are remodeled by environmental inputs during periods of heightened plasticity in postnatal development 1 . These critical periods are limited to specific windows of time that are different for each sensory system. In the visual system, the primary visual cortex is subjected to a critical period for ocular dominance plasticity during which connections from a weak or absent eye can be permanently overtaken by the strong eye 2 . The integrated action between inhibitory and excitatory circuits determines critical period onset, with a major role being played by the maturation of fast-spiking parvalbumin (FSPV) interneurons 3 . We have shown that Otx2 homeoprotein helps determine critical period timing by signaling FSPV cells in mice 4 . Conditional knock-down in heterozygous floxed mice just prior to normal critical period timing is sufficient to delay onset, while cortical infusion of recombinant Otx2 protein accelerates both onset and closure 4 .
Remarkably, Otx2 protein in the cortex is non-cell autonomous. The Otx2 locus is silent, as shown by PCR, in situ hybridization and Otx2 +/GFP mice, while Otx2 protein is detectable by immunohistochemistry and immunoblot 4,5 . We have shown that cortical infusion of recombinant Otx2 protein results in specific uptake by FSPV cells, while injection in the retina results in its transport along the visual pathway and into these same cells 4 . Blocking extracellular Otx2 through infusion of antibodies or specific peptides reduces uptake of endogenous Otx2 in FSPV cells 4,6 . Furthermore, we recently showed that the choroid plexus expresses Otx2 and secretes it into the cerebrospinal fluid 5  An outstanding question is whether cortical Otx1 homeoprotein also plays a role in the critical period. Indeed, Otx1 is expressed in the cerebral cortex during development and continues to be expressed by layer V neurons in the adult 8 . It is thus possible that it is secreted by layer V cells and transferred into above granular and supragranular layers where Otx2 protein is detected. Unfortunately, most antibodies (commercial and academic) for Otx1 and Otx2 are pan-Otx thereby ruling out immunohistochemical approaches for conclusive evidence. Since genetic manipulation of the Otx2 locus has resulted in reduced protein in layer IV visual cortex 7 , we sought confirmation whether Otx1 is also present in these layers in the adult by using Otx1 knockout mice and proteomic approaches. GAPDH-reverse CCGGCATCGAAGGTGGAAGAG.

Animals
The experiment was performed in duplicate. For calculating fold expression, gene-to-GAPDH ratios were determined by using the 2 -∆∆Ct method referenced to Otx1 expression in the lateral geniculate nucleus (LGN).

Mass spectrometry analysis
After immunoprecipitation, proteins were separated on SDS-PAGE gels (Invitrogen) and stained with colloidal blue staining (LabSafe GEL BlueTM GBiosciences). Gel slices were excised and proteins were reduced with 10 mM DTT prior to alkylation with 55 mM iodoacetamide. After washing and shrinking the gel pieces with 100% MeCN, in-gel digestion was performed using trypsin (Promega) overnight in 25 mM NH4HCO3 at 30°C.
Peptides were extracted and analyzed by nano-LC-MS/MS using an . We set carbamidomethyle cysteine, oxidation of methionine and N-terminal acetylation as variable modifications. We set specificity of trypsin digestion and allowed 2 missed cleavage sites and we set the mass tolerances in MS and MS/MS to 2 ppm and 0.5 Da, respectively. The resulting Mascot files were further processed by using myProMS (v 3.0) 15 and the estimated false discovery rate (FDR) by automatically filtering the Mascot score of all peptide identifications was less than 0.5%. Mass spectroscoy data: This Mascot .dat file ("F022130_part.dat") contains the experiment description and the data pertaining to the 2 peptides that match Otx2 protein.

Otx1 locus but not Otx2 locus is active in the mouse visual cortex
We compared the expression of the Otx1 and Otx2 loci of various brain regions by performing quantitative PCR ( Figure 1A). While both mRNA were detected in thalamic and cerebellar structures, only Otx1 mRNA was found in the visual cortex. This result confirms the previously reported absence of GFP expression in the visual cortex of Otx2 +/GFP knockout mice and lack of signal in the visual cortex of wild type mice after Otx2 in situ hybridization, even though Otx2 antibodies label FSPV cells 4 . These cells are enwrapped by a dense extracellular matrix called perineuronal nets (PNNs) when localized to layer IV of the cortex ( Figure 1B). Otx1 locus is active in layer V 8 . In immunohistochemical analysis of Otx1 knockout mouse, almost all Otx signal is lost in layer V while signal from Otx2 protein continues in PNN-labeled cells ( Figure 1B). However, we cannot preclude that some Otx1 is also present in layer IV cells in wild type mice.

Otx2 protein but not Otx1 protein is found in granular and supragranular layers of visual cortex
In order to analyze Otx homeoprotein distribution in the visual cortex, we turned to immunoprecipitation (IP) experiments. We first performed IP on whole visual cortex extracts, which showed both Otx1 and Otx2 protein (Figure 2A). This result is confirmed by IP of choroid plexus, which strongly expresses Otx2 but only very weakly expresses Otx1. To analyze granular and supragranular content, we dissected and extracted the superior layers of posterior adult mouse cortex ( Figure 2B) and performed IP by using cross-linked magnetic beads. Immunoblot analysis detected only Otx2 but not Otx1 protein ( Figure 2C). This result was confirmed by mass spectrometry analysis on these extracts, which identified 2 Otx2 peptides (6.6% coverage with 100% specificity, Figure 2D). While this number of peptides is low, it was expected given that Otx2 is predicted to be poorly ionizable. Furthermore, identification has an error of 1 protein in 20,000 (19,999 are true) with a FDR of less than 0.5% for 2 peptides in our database of 16,671 sequences, thus the chance of misinterpretation is very low. These peptides are 100% specific for Otx2 and they have the same ions as the reference spectrum from a sample of purified Otx2 protein. While the second peptide is also specific for Otx1 and Crx homeoproteins, the first peptide is unique for Otx2 ( Figure 2D). These results confirm that the homeoprotein localized in granular and supragranular FSPV cells is indeed Otx2 and not Otx1.

Discussion
The non-cell autonomous activity of homeoprotein transcription factors is now well established. There are clear phenotypes with recently developed in vivo single-chain secreted antibodies that neutralize extracellular homeoproteins yet leave intact cell autonomous activities 9-11 . Non-autonomy can also be demonstrated by comparing mRNA and protein expression. Indeed, the absence of mRNA in presence of the protein argues in favor of non-cell autonomy. However, when the receiving territory is a short distance from the producing territory, one could invoke the possibility of cell migration or mRNA instability to bring into question the reality of homeoprotein transfer. In the visual system, Otx2 protein is found in the visual cortex far from two potential sources of Otx2 (where the Otx2 locus is active), namely the eye and the choroid plexus 5,12 . Indeed, the Otx2 locus is not active in the adult cerebral cortex as verified by using the Otx2 +/GFP mouse, quantitative RT-PCR and in situ hybridization 4 . In addition, conditional Otx2 ablation in the choroid plexus reduces its content in FSPV cells, further supporting non-cell autonomy 5 .
However, since most Otx1 and Otx2 antibodies are pan-Otx antibodies, it was still conceivable that some of the protein seen in FSPV cells by immunohistochemistry could correspond to Otx1 expressed in layer V of the cerebral cortex and transferred into PV cells. The present study shows that the staining in layer IV is maintained in the Otx1 knockout mouse and that IP experiments of layers I-IV give immunoblot bands with expected Otx2 size and can be used to identified Otx2 by mass spectrometry. These results confirm that FSPV cells in granular and supragranular layers of the cerebral cortex only contain non-cell autonomous Otx2 and do not contain Otx1.
It may seem surprising that Otx1 expressed in layer V is not secreted and internalized by FSPV cells. Indeed the protein presents a homeodomain nearly identical to that of Otx2 and thus contains the two sequences necessary for internalization and secretion (for review see 7 ). However, previous studies have demonstrated that homeoproteins are transported from the basolateral to the apical side of polarized cells and thus into the axon 7,13 . Given their polarity and orientation, the pyramidal cells of layer V that express Otx1 are thus very unlikely to release it at the level of FSPV cells. In contrast, the choroid plexus epithelial cells present their apical surface toward the ventricles allowing Otx2 secretion into the cerebral spinal fluid.
In conclusion, this study demonstrates that Otx2 is the only non-cell autonomous Otx family protein in the granular and supragranular FSPV cells.  Studies, mostly from A. Prochiantz's laboratory, have demonstrated that the transcription factor (TF) Otx2, produced by external sources such as the choroid plexus, is internalized into parvalbumin neurons of layers II/III and IV of the visual cortex, thereby controlling its plasticity. So far these studies had not ruled out the possibility that part of this effect could be mediated by Otx1, a closely related TF, which may act redundantly with Otx2. In this study, the authors combined the use of mice with Otx1 immunoprecipitation from layers I-IV extracts and high-resolution mass spectrometry analysis to demonstrate the sole presence of Otx2 peptides in these layers. The study thus demonstrates that Otx2 is the only non-cell autonomous Otx family member functional in fast-spiking parvalbumin cells of layers I-IV of the visual cortex. This is a clear and well-executed study that solves an existing concern.

Data availability
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. The authors address an issue that is unresolved in the literature. Is the Otx found in the cortex, and responsible in part of cortical plasticity, cortical derived Otx1 or choroid plexus derived Otx2? The problem exists because the antibodies used previously can not distinguish between the two. In the present study using KO mice and a variety of measures, the authors prove Otx2 is the protein in the Otx1 cortical cells and that this therefore is a cell non-autonomous phenomenon. The one question I have relates to figure 1B, which suggests there is greater Otx2 staining in the KO mouse cortex. If this is a Otx1 reliable observation the authors should comment. The low peptide coverage in the MASS SPEC could be a concern but the authors address that and in addition the other supporting data decrease the concern. -/-