Smad anchor for receptor activation nuclear localization during development identifies Layers V and VI of the neocortex

Smad anchor for receptor activation (SARA, zfyve9) has been classically observed in early endosomes of different cells types where it regulates vesicular transport of proteins and membrane components. Very few other members of the zinc finger FYVE domain‐containing family (zfyve) have different functions other than controlling membrane trafficking. By analyzing SARA localization throughout mouse embryonic brain development, we detected that besides the endosomal localization it also targets neuronal nuclei, specifically of the cortical layers V/VI. These findings were confirmed in human brain organoids. When evaluating neuronal cell lines, we found that SARA accumulates in nuclei of PC‐12 cells, but not Neuro‐2a, highlighting its specificity. SARA functions as a specific marker of the deep cortical layers until the first postnatal week. This temporal regulation corresponds with the final phases of neuron differentiation, such as soma ventral translocation and axonal targeting. In sum, here we report that SARA localization during brain development is temporarily regulated, and layer specific. This defined pattern helps in the identification of early born cortical neurons. We further show that other zfyve family members (FYCO1, WDFY3, Hrs) also distribute to nuclei of different cells in the brain cortex, which raises the possibility that this might be an extended feature within the protein family.

In cell lines and cultured hippocampal neurons, SARA presents a vesicular labeling partially colocalizing with the early endosome (EE) marker Rab5 (Arias, Siri, & Conde, 2015;Hu et al., 2002). Accordingly, we have previously reported a punctate vesicle-like distribution of endogenous SARA in the mice embryonic neocortex (Mestres, Chuang, Calegari, Conde, & Sung, 2016). We have also addressed its role in controlling membrane and membrane proteins trafficking and how this function is important for normal development of the nervous system (Mestres & Sung, 2017). Particularly, during neocortical development, SARA fine-tunes the amount of the cell adhesion protein L1 that distributes to the cell surface of migrating neurons (Mestres et al., 2016). Also, in the eye, SARA tethers vesicles with the outer segment of rod photoreceptors, thereby participating in photoreceptor renewal (Chuang et al., 2007). Now, we inform that SARA might have an additional role regarding its novel localization into deep-layer nuclei of the brain cortex. Importantly, we also show that the nuclear localization aspect is shared with other members of the zfyve family. Although in a different spatiotemporal pattern, FYCO1 (zfyve7), Hrs (zfyve8), and WDFY3 (zfyve25), also localize to nuclei of different cell types in the cortex during development.

| Animals
Mouse (C57BL/6; Janvier Labs, Saint Berthevin, France) brains from E13 to P2 were collected after decapitation, and immediately fixed in 4% PFA in 0.1 M phosphate buffer (pH 7.4), overnight at 4 C. For postnatal brains (P5-P30), isoflurane anesthetized animals were transcardially perfused through the ascending aorta with 10 ml of heparin saline (1,000 units/ml), followed by 20 ml of 4% PFA in 0.1 M phosphate buffer. Later, brains were dissected out of the skull and postfixed for 1 hr in the same fixative. Fixed brains were embedded in low melting point agarose, and sectioned by vibratome (40 μm). For all ages, at least three brains were evaluated. Mice of both sexes were used, which were kept in standard cages with 12 hr dark/light cycle, and with access to food and water ad libitum. All animal procedures were approved by local authorities (TVV 39/2015).

| Cell culture
Primary neurons were obtained from E15 mouse brains. Briefly, the brains were removed from the embryo, the cortex dissected out and the meninges removed. The tissue was incubated with trypsin at 37 C for 15 min, and later rinsed in PBS. Mechanical dissociation was performed by pipetting several times in culture medium containing Neurobasal, Penicillin/Streptomycin, B27 and Glutamine (Gibco).

| Immunohistochemistry
Immunolabeling of brain sections was carried out using free-floating methods. Briefly, brains sections (40 μm thick) were incubated in 5% donkey serum with PBS 0.1% triton (PBST) for 1 hr at room temperature. Primary and secondary antibodies were incubated in 1% donkey serum in PBST, followed by three washes in PBST for 15 min each.
Primary antibodies were incubated for three overnights, while secondary antibodies were incubated one overnight, rotating at 4 C.
Cryosections of human brain organoids (14 μm thick) were obtained mounted onto glass slides and processed for immunolabeling similarly as for the mouse brain slices, except primary antibodies were incubated one overnight.
A detailed list of the primary antibodies used in this study can be found in Table 1. Various Alexa-dye conjugated secondary antibodies (1:500, Molecular Probes) were also used. Phalloidin coupled to Alexa dye (633) was used to detect the actin cytoskeleton, and 4 0 ,6-diamidino-2-phenylindole, dihydrochloride (DAPI) was employed to stain the nuclei.

| Specificity of the antibodies
Most of the antibodies used here were validated by the commercial providers (Table 1). As indicated in their supplier's data sheet, all antibodies showed a band of the respective molecular weight of the protein detected as revealed by Western blot. The only exception is the SARA (C137) antibody which was validated by Dr Sung (Hu et al., 2002). Also, our Western blots show a single band at~160 kDa when using the C137 antibody (or H300 antibody); which corresponds to SARA's molecular weight.
Additionally, in some control tissue sections, we omitted either the primary or the secondary antibodies. No labeling was observed in these sections.
Reverse transcription-PCR using SP6 or T7 RNA polymerases generated digoxygenin (DIG)-labeled cRNA antisense or sense probes; which cover a 663-bp region of the three SARA isoforms (between Exons 8 and 13).

| Subcellular fractioning and Western blot
Cytoplasmic and nuclear protein extractions of PC-12 cells were done using the NE-PER Extraction Reagents (Thermo Fisher, #78833), supplemented with protease inhibitor mix (Diagenode, #C12010012), following manufacturer's instructions with minor modifications. Before extracting the nuclear fraction, the pellet was washed extensively in CER I and CER II. From either fraction, 30 μg were run per lane in 4-12% Bis-Tris gels (Thermo Fisher, #NP0335BOX), transferred to PVDF membranes, and blotted with SARA antibodies (C137 or H300), actin antibody or LaminB1 antibody. Immunodetections were carried out with the ECL method (Thermo Fisher, #34577).

| Confocal imaging and analysis
Immunolabeled sections were examined with a confocal Zeiss LSM 780 microscope, and processed in ImageJ and Photoshop. Low power tile images were acquired in the confocal microscope or in a Zeiss Axio Scan with a 20× objective, and were stitched together in Zen.
Nuclear SARA intensity signal was measured in ImageJ (NIH). The area within the nucleus was selected for each cell in the DAPI channel. This area was used to determine the mean intensity gray values for the SARA signal. Background signal was negligible (<1 a.u.). At least 20 cells, from three independent cultures were measured.

| Statistical analysis
Data are presented as mean ± SEM. Statistical significance was evaluated by one-way analysis of variance (ANOVA) and a Tukey post hoc T A B L E 1 List of primary antibodies used in this study  2.10 | Nuclear localization signal analysis FASTA amino acid sequence for the different zfyve members were analyzed in NLS Mapper (Kosugi et al., 2009) to determine the sequence of putative nuclear localization signals. Peptide sequences with Scores 1 or 2 localize to the cytoplasm; that with Scores 3, 4, or 5 distribute both to the cytoplasm and the nucleus; that with Scores 6 or 7 partially localize to the nucleus; while Scores 8, 9, or 10 distribute mainly to the nucleus. 3 | RESULTS

| Endogenous SARA localizes to nuclei of early born neurons
To first assess endogenous SARA distribution, we immunolabeled coronal brain slices of mouse embryo as early as embryonic Day (E) 13. We found that it distributed into a punctate pattern throughout the cortex (Figure 1a,b). Interestingly, SARA highly localized to nuclei in the developing cortical plate (CP), following the lateral-to-medial axis of neurogenesis (Takahashi, Goto, Miyama, Nowakowski, & Caviness Jr, 1999). The nuclear distribution of SARA overlapped with that of the deep-layer marker Ctip2, but the cells in the proliferative area (Sox2+)  To test whether SARA might also localize to nuclei of later born neurons, we immunolabeled neonatal (P0-P2) coronal brain sections.
We found that within the cortex SARA distributed to all layers of the CP within EEs (Figure 2a,b). Interestingly, only neurons of layers V/VI were positive for SARA also in their nucleus, together with the classical early born neuron marker Ctip2 (Figure 2a-c). Of note, SARA and Ctip2 immunolabeling in the deep layers were complementary. That is, while Ctip2 was expressed at higher levels in Layer V than VI, SARA expression was higher in Layer VI compared to Layer V (Figure 2c).
Outside the cerebral cortex, we found that SARA also localized to neuronal nuclei in the thalamus (Figure 2d). To determine SARA nuclear localization along the rostrocaudal axe, we analyzed sagittal P0 brain sections coimmunolabeld with the upper layers (II-IV) marker Satb2. The examination of posterior and anterior regions confirmed the specific distribution of SARA in Layers V and VI along the brain cortex (Figure 2e-g). Consistent with our previous results, independently of the SARA protein subcellular localization either within EEs and/or in the nucleus, SARA mRNA expression was highest at the CP within the neonatal brain cortex (Figure 2h-j).
Our findings were confirmed by two different SARA antibodies ( Figure 3a). Both rabbit antibodies recognize separate regions of SARA, and their results were comparable. A strong nuclear SARA localization in deep-layers neurons was detected using either one of the antibodies (Figure 3b).

| Human deep-layer neurons are also positive for SARA
Next, we evaluated whether the human SARA homologue might also target a specific cell population similarly as in the mouse cortex. To this end, we analyzed cryosections of human brain organoids. At developmental Day 54, SARA localized into vesicular structures which were enriched at the ventricular border; comparable with our observations in E15 mouse brain sections. Also in the human cells, SARA colabeled the early generated neurons along with Ctip2; while the proliferative Sox2+ area was negative for SARA nuclear immunolabeling (Figure 6a,b).
These results point that SARA functions as a novel cortical deep-layer marker in both, mouse and human tissue.

| SARA identifies early born cortical neurons grown in culture
Subsequently, we tested SARA distribution in neuronal cells grown in culture. We examined primary mouse cortical neurons, rat pheochromocytoma (PC-12) cells, and Neuro-2a cells. Primary neurons were obtained from E15 mouse brains. At this timepoint, most of the Tuj1 positive primary neurons were also immunoreactive for Ctip2 (not shown). In these cells, SARA localized to vesicle-like structures; namely, EEs. Consistent with our brain slices immunolabeling, the early born neurons also exhibited an SARA nuclear distribution (Figure 7a). Notably, the nucleolus and heterochromatin were predominantly devoid of SARA staining.

Similar to the cultured neurons, in undifferentiated PC-12 cells SARA
showed a punctate distribution throughout the cytoplasm; and it was also highly enriched in the nucleus of the cells (Figure 7b). Contrarily, in Neuro-2a cells SARA distributed mainly into endosomes, while it was not evidently accumulated in the cell nucleus (Figure 7c). These results point that comparably to the different cells of the cortex, the nuclear distribution of SARA might be differentially regulated between cell types.

| Functional aspects of nuclear SARA distribution
We next evaluated whether SARA nuclear localization might be dependent on PI3P levels. To assess this, we incubated Neuro-2a  Table 2. GO terms such as circadian rhythm and phosphatase complex is representative of the known functions of SARA. SARA has been shown to interact with Rhodopsin in the light-sensing organelle of the eye (Chuang et al., 2007), and to be part of a complex together with Smad2/3 and the phosphatase PP1c (Bennett & Alphey, 2002).
Interestingly, the most enriched GO term identified was nucleus; in agreement with the evidence presented here. In this sense, transcription regulation was another term associated with SARA binding We also performed immunofluorescence with a RUFY1 antibody (sc-398740). In our hands, it did not show a clear and consistent nuclear distribution within the cortex (not shown). Importantly, analysis of nuclear localizing signals among the zfyve members studied here identified the amino acid sequences with high score of putative nuclear localization (Table 3).
Despite extensive research in different model organisms and cell types, SARA itself was never reported to distribute to the nucleus.
It has been shown before that a pool of nuclear phosphoinositides, including PI3P, exist independently of the cytosolic pool (Gonzales & Anderson, 2006). Also, electron microscopy revealed that PI3P distributes within the nucleus (Gillooly et al., 2000). Importantly, the probe used for these experiments derived from the FYVE domains of two zfyve members: Hrs and EEA1 (zfyve2). However, the current view supports the idea that the nucleus does not attract zfyve proteins perhaps due to its neutral or mildly alkaline pH compared to the lower pH found in endosomes, phagosomes, and Golgi apparatus (Kutateladze, 2006;Lee et al., 2005). Nevertheless, the findings presented here add evidence that challenges that notion.
A seminal work identified the first zfyve member to localize in the cell nucleus. RUFY2 (zfyve13), although not in the brain, exhibits a nuclear localization in the mouse primordial cartilage and otic capsule at very specific ages, from E12 to E14 (Dunkelberg & Gutierrez-Hartmann, 2001). Also, unlike most other zfyve members that localize to endosomes, WDFY3 mainly localizes to the nuclear envelope in HeLa cells at steady state, but shifts to the autophagosome upon starvation (Simonsen et al., 2004).
Our work reveals that SARA also has the ability to localize to the nucleus in a very specific time-and cell-type dependent manner.  (Dragich et al., 2016).
While the classical deep-layer marker Ctip2 is also found in neuronal nuclei of the striatum, the hippocampus, and the olfactory bulb (Arlotta et al., 2005), SARA was not evident in the nuclei of these extracortical cells. Nevertheless, neuronal nuclei in the thalamus were also immunoreactive for SARA. Interestingly, neurons from Layer VI and some of the Layer V conforms the corticofugal neurons that project to the thalamus (Tau & Peterson, 2010). This observation raises the possibility that SARA might be relevant for the development of corticothalamic circuitry.
In line with the evidence presented here, analysis of the GO terms associated with SARA binding partners identified "nucleus" and "transcription regulation" as enriched annotated terms. These findings suggest a novel role for SARA in the nucleus, additionally to the longknown function in membrane transport. A possible contribution of SARA to the transcriptional regulation, and the identity of putative genes regulated by SARA remains to be investigated.
As a proof-of-concept, we analyzed four other zfyve members and evaluated whether they also localized to the cell nucleus. Three of which; FYCO1, WDFY3, and Hrs, show a nuclear distribution within different cells of the developing cortex. Except for WDFY3, whose nuclear localization was known in cultured HeLa cells (Simonsen et al., 2004); this work shows for the first time that FYCO1 and Hrs can distribute to subcellular structures other than their previously known localization into late endosomes and lysosomes (Pons et al., 2008;Raiborg et al., 2015).
Altogether, these results strongly point that the nuclear localization might be a feature extended among the zfyve family.
Notably, some zfyve members have been related to neurodevelopmental diseases such as autism; including WDFY3, FYCO1, and FGD1 (zfyve3) (Orosco et al., 2014;Orrico et al., 2004;Voineagu et al., 2011). For example, a mutation found to generate autism-like features; including forebrain overgrowth and enlarged ventricles, introduces a stop site right before the WD40 and FYVE domains at the end of the WDFY3 gene. Despite the important function of WDFY3 in selective macroautophagy, mutant embryos did not show a disruption in the autophagy pathway (Orosco et al., 2014). Whether the developmental defects observed after loss of WDFY3 in mutant mice are due to its nuclear localization in the cortical wall as shown here, constitutes a hypothesis to be further tested. Also, it will be interesting to evaluate if other zfyve members are involved in brain developmental disorders as a result of their role in the cell nucleus.
Structurally, it was demonstrated that the isolated FYVE domain of several zfyve members, mainly localize to nucleus and the cytoplasm, but upon dimerization stimulation the FYVE domains shift to the endosomes (Hayakawa et al., 2004;Hayakawa, Hayes, Leonard, Lambright, & Corvera, 2007). This implies that oligomerization favors endosomal localization. It remains to be elucidated whether the opposite holds true; that is, do FYVE domains shift to the nucleus upon oligomerization inhibition? T A B L E 3 Nuclear localization signal scores. The zfyve members studied here, their putative sequences responsible for nuclear localization and corresponding score, as identified by NLS mapper