Controlled Cre/loxP Site-Specific Recombination in the Developing Brain in Medaka Fish, Oryzias latipes

Background Genetic mosaic techniques have been used to visualize and/or genetically modify a neuronal subpopulation within complex neural circuits in various animals. Neural populations available for mosaic analysis, however, are limited in the vertebrate brain. Methodology/Principal Findings To establish methodology to genetically manipulate neural circuits in medaka, we first created two transgenic (Tg) medaka lines, Tg (HSP:Cre) and Tg (HuC:loxP-DsRed-loxP-GFP). We confirmed medaka HuC promoter-derived expression of the reporter gene in juvenile medaka whole brain, and in neuronal precursor cells in the adult brain. We then demonstrated that stochastic recombination can be induced by micro-injection of Cre mRNA into Tg (HuC:loxP-DsRed-loxP-GFP) embryos at the 1-cell stage, which allowed us to visualize some subpopulations of GFP-positive cells in compartmentalized regions of the telencephalon in the adult medaka brain. This finding suggested that the distribution of clonally-related cells derived from single or a few progenitor cells was restricted to a compartmentalized region. Heat treatment of Tg(HSP:Cre x HuC:loxP-DsRed-loxP-GFP) embryos (0–1 day post fertilization [dpf]) in a thermalcycler (39°C) led to Cre/loxP recombination in the whole brain. The recombination efficiency was notably low when using 2–3 dpf embyos compared with 0–1 dpf embryos, indicating the possibility of stage-dependent sensitivity of heat-inducible recombination. Finally, using an infrared laser-evoked gene operator (IR-LEGO) system, heat shock induced in a micro area in the developing brains led to visualization of clonally-related cells in both juvenile and adult medaka fish. Conclusions/Significance We established a noninvasive method to control Cre/loxP site-specific recombination in the developing nervous system in medaka fish. This method will broaden the neural population available for mosaic analyses and allow for lineage tracing of the vertebrate nervous system in both juvenile and adult stages.


Introduction
Genetic mosaic analysis is a powerful tool in the fields of neuroscience and developmental biology for labeling a subset of neurons, tracing cell-lineage, and modulating neuronal function [1][2][3][4]. The recent combination of Cre/loxP and optogenetic tools allows for specific modulation of selected neurons within complex neural tissues [5,6]. Cre induction can be spatially controlled by cell type-specific promoters/enhancers and site-specific viral infection [7,8]. The lack of appropriate promoters/enhancers, however, limits the neural population available for mosaic analysis in the vertebrate brain. Site-specific viral infection requires invasive surgical procedures, which also limits free access to the entire brain. Thus, the development of noninvasive methods for controlled Cre/loxP site-specific recombination will increase the neural population available for mosaic analyses of the vertebrate nervous system. To address this issue, we focused on a heat-inducible Cre/loxP gene induction system in medaka fish. Medaka embryos have high temperature tolerance (4-35uC) compared to zebrafish (25-33uC) [9], allowing for various temperature-mediated treatments [10]. In addition, an artificial heat-shock promoter (HSP) comprising multimerized heat shock elements has very low background activity and no leak in medaka fish [11], which allows for spatiotemporal site-specific Cre induction. Local heat treatment using a metal probe and an infrared laser results in ectopic Cre induction in a small number of cells in various tissues such as the gonads and epidermal tissues in medaka fish [12][13][14]. To our knowledge, however, there are no reports of its application in neural tissue.
Here we used an infrared laser-evoked gene operator (IR-LEGO) system to induce highly regulated spatiotemporal Cre expression in neural precursor cells of medaka embryos [15]. Medaka fish have a transparent chorion that facilitates noninvasive observations of the nervous system throughout development [10], and it allows for heating a small subpopulation of differentiating neurons in the neural placode using an infrared laser (wavelength: 1480 nm). Based on the fate maps of neural placodes in the medaka embryo, analysis of clonally related cells in a specific brain region of interest can be performed by irradiating a small population of the neuronal precursor cells.
To examine whether a heat-inducible Cre/loxP gene induction system works in the medaka nervous system, we generated transgenic (Tg) medaka lines for the detection of Cre/loxP recombination with the promoter regions of medaka HuC. HuC belongs to a family of vertebrate neuronal-specific genes homologous to the Drosophila elav and serves as an early marker of differentiating neurons [16,17]. Zebrafish HuC is expressed in neuronal precursor cells during embryogenesis, and then high expression levels persist in most regions of post-hatching and larval brain [18,19]. Thus, the zebrafish HuC promoter is widely used for visualizing and/or modifying the function of neural circuits in juvenile fish [20]. The HuC promoter is also applicable for visualizing the differentiation process during adult neurogenesis, as HuC expression is restricted to newborn and differentiating neurons in the adult zebrafish brain [19,21]. In the present study, we demonstrated that stochastic Cre recombination during embryogenesis allowed for visualization of clonally-related cells in compartmentalized regions of the adult medaka brain. Further, irradiation of the developing medaka embryo brains using an infrared laser allowed for visualization of clonally-related HuCexpressing cells in both juvenile and adult medaka fish.

Ethics statement
The work in this paper was conducted using protocols approved by the Animal Care and Use Committee of the University of Tokyo (permit number: 12-07). All surgery was performed under cold anesthesia, and all efforts were made to minimize suffering.

Fish and breeding conditions
Medaka fish (Oryzias latipes, drR strain) and all Tg lines, Tg (HuC:loxP-DsRed-loxP-GFP) and Tg (HSP:Cre), were maintained in their respective groups in plastic aquariums (12 cm 613 cm 619 cm). All fish were hatched and bred in our laboratory. The water temperature was maintained at ,28uC and the light was provided by fluorescent lamps for 14 h per day (08:00 to 22:00).

Generation of Tg (HSP:Cre/Crs:BFP)
The dual expression vector, pPBIS19-mgfc:TagBFP-8xHSE:Cre containing fused gene of Cre recombinase with N-terminal nuclear localization sequence and the red fluorescent protein mCherry under the control of artificial heat shock inducible promoter [11]. This plasmid also had TagBFP (Wako Junyaku Kogyo Co. Ltd., Osaka, Japan) gene under the control of a mouse gamma F crystalline promoter [22] for selecting the proper transgenic lines. It allowed us to select the embryos carrying HSP:Cre transgene by blue fluorescence of their eyes without heat treatment. Insulators were inserted between the polyA sequence under the control of HSP promoter and TR3, and also inserted into the boundary region of the two promoters, HSP and Crystalline to prevent position-effects of the transgenes. The F1 embryos from wild type and vectorinjected F0 were raised to sexual maturity and screened for germline transmission by a fluorescent microscopy examining BFP expression.

Immunohistochemistry
Immunostaining was performed on 14-mm cryosections. Whole brains were embedded in OCT compound (Sakura Tissue Tek) and cut using Cryostat (Leica, CM 1850). The sections were blocked with 0.2%Triton, 1% dimethylsulfoxide, and 2% bovine serum albumin in PBS at room temperature for 1 h, then incubated in the primary antibodies diluted in the blocking buffer at 4uC overnight. Primary antibodies used in this study were mouse anti-HuC/D (1:500, Molecular Probes), rabbit anti-DsRed (1:500, Clontech). Primary antibodies were detected by subclassspecific secondary antibodies labeled with Alexa 488/548 (1:1000, Molecular Probes), respectively.

mRNA microinjection and microscopy
Cre-SV40 was cloned into a pGEM-T easy vector (Promega, Madison, WI). Template preparation and in vitro synthesis of mRNAs were performed as described previously [24]. Confocal and fluorescence microscopy analyses were performed using a Zeiss confocal microscope (LSM710; Carl Zeiss, Oberkochen, Germany) and a Leica epifluorescence microscope (MZFLIII; Leica, Tokyo, Japan). The micrographs were processed with Photoshop software (Adobe, San Jose, CA) and the projection of a confocal stack was rendered using FluoRender (http://www. fluorender.com).

Heat-induction in the whole body
Ten embryos were placed into a tube with 200 ml medaka hatching buffer [25] and heated at 39uC for the prescribed number of hours in a thermalcycler (TProfessional Basic; Biometra).

Heat induction in the telencephalon
Embryos were mounted at stage 24, dorsal side up, in a drop of 2% methylcellulose (M-0387; Sigma Chemical Co., St. Louis, MO) and observed using the 206 custom made objective lens (mono-coated lens of UPlanSApo; Olympus, Tokyo, Japan) for an Olympus epifluorescence microscope with IR-LEGO unit (IR-LEGO 1000; Sigma-Koki, Saitama, Japan). Heat was induced in the telencephalon by 1480-nm light generated by a high-power single mode CW Raman fiber laser (Model PYL-3-1480-M; IRE-Polus Group, Sturbridge, MA), as previously described [12,15].

Results
Generation and characterization of the HuC:loxP-DsRed-loxP-GFP Tg medaka line We generated a Tg medaka line that expresses loxP-DsRed-loxP-GFP under the control of a 3.3-kb medaka HuC promoter (HuC:loxP-DsRed-loxP-GFP). DsRed expression in the Tg embryo was first detected as early as stage 27 in the anterior brain vesicleintermediate brain vesicle (Ant-Int; Fig. 1A and Fig. S1A). After stage 34, DsRed fluorescence was observed in the whole brain (Fig. 1B, C and Fig. S1B, C). In 3-month-old adult fish, DsRedexpressing cells were prominently detected in the superficial layer of the telencephalon, cerebellum, posterior edge of the optic tectum, and the left side of habenular nucleus (Fig. 1D, Fig. S2).
To determine the details of DsRed expression, we prepared coronal sections of the Tg whole brain (HuC:loxP-DsRed-loxP-GFP; Fig. 1E, F). The surface of the telencephalon and the left side of habenular nucleus exhibited prominent DsRed expression (Fig. 1F panel I -VII). In addition, DsRed expression was detected in the olfactory bulb (panel II -IV), preoptic area (panel V), ventromedial nucleus (panel VII), optic tectum (panel VIII -XIV), marginal zone of the third ventricular zone (panel VIII -XI), hypothalamus (panel X -XII), and cerebellum (panel XIV -XV). We previously mapped ''proliferation zones'' comprising stem cells in the adult medaka brain [26] and found a large number of HuC-positive neural progenitors situated at or near the ''proliferation zones'' (Fig. S2). DsRed-positive axons were observed near the olfactory bulb, fasciculus longitudinalis medialis (Flm), nucleus glomerulosus posterioris (NGp), and corpus interpeduncularis (Ci), to which the habenular nucleus neurons project [27]. Taken together, these findings indicated that DsRed expression in the Tg (HuC:loxP-DsRed-loxP-GFP) labeled the neural progenitor cells with huC expression. Furthermore, to confirm that DsRed is expressed in HuC-positive neural progenitors, we performed co-staining with antibody against HuC/D and DsRed by immunohistochemistry. The pattern of HuC/D expression overlapped with that of DsRed, including the habenular nucleus ( Fig. 1G-I).

Stochastic Cre/loxP recombination by injection of a low concentration of Cre mRNA
We examined whether stochastic recombination can be induced by injection of Cre mRNA into the Tg embryos at the 1-cell stage, which would allow us to label the cell lineage. First, we confirmed that Cre recombination was induced in most cells of various tissues, including muscle and brain, by Cre mRNA injection (100 ng/ml) into Tg (beta actin:loxP-DsRed-loxP-GFP) under the control of a ubiquitous beta-actin promoter [7] (Fig. S3A-E). Next, a small amount of Cre mRNA (0.2, 2 and 20 ng/ml) was injected into the Tg (HuC:loxP-DsRed-loxP-GFP) at the 1-cell stage (30 min after fertilization). Recombination by Cre mRNA injection led to changes in the expression of DsRed to that of GFP ( Fig. 2A). The mosaic expression of DsRed and GFP (white arrowhead) was detected at the adult stage (Fig. 2B, C). The induction of GFP expression increased dose-dependently. The distribution of GFPexpressing cells differed among individuals (Fig. S3F), indicating that stochastic recombination can be induced in some stem cells by the injection of Cre mRNA. We were unable to precisely determine the number of stem cells in which recombination occurred, however, because it takes at least 12 h to express mature GFP [28], during which multiple cell divisions may have occurred.
To visualize the three-dimensional distribution of the DsRedand GFP-expressing cells, we obtained optical images of vibratome sections (120 mm) and generated three-dimensional images of the telencephalon of the Cre mRNA (20 ng/ml)-injected fish (Fig. 2C  white dashed line, Fig. 2D, and Movie. S1). Interestingly, subpopulations of GFP-expressing cells were detected in spatially distinct segments. These findings suggest that subpopulations (defined as clonally-related cells) derive from a single or a few adjacent cells.

Heat induction of Cre/loxP recombination in the nervous system
To induce the recombination by heat treatment, we established a Cre-inducible Tg (HSP:Cre/Cry:BFP) with an artificial HSP promoter containing eight consecutive artificial heat-shock response elements and beta-globin minimum promoter derived from Xenopus laevis. Expression of Cre in the Tg (HSP:Cre) is reported to be inducible by a brief heat shock treatment. The construct expressed blue fluorescent protein (BFP) under the control of the mouse Crystallin (Cry) promoter as a screening marker, and an insulator sequence was inserted between HSP:Cre and Cry:BFP (Fig. 3A) [29] to avoid cross talk between the artificial HSP and Cry promoters which are located in the same construct of the vector. We generated Tg (HuC:loxP-DsRed-loxP-GFP/ HSP:Cre/Cry:BFP) by genetic crossing.
We examined the efficiency of the heat-inducible recombination during embryonic development. We exposed the embryos to heat shock (39uC, 3 h) at the early blastula stage (stage 9-10, 0,1 days post fertilization [dpf]). GFP fluorescence was detected at the time of hatching in the whole brain of Tg (HuC:loxP-DsRed-loxP-GFP/ HSP:Cre/Cry:BFP) embryos (Fig. 3B, upper panel). In contrast, no GFP fluorescence was detected in either heat-exposed Tg (HuC:loxP-DsRed-loxP-GFP) (Fig. 3B, middle panel) or non-treated Tg (HuC:loxP-DsRed-loxP-GFP/HSP:Cre/Cry:BFP) embryos (Fig. 3B, lower panel). These findings indicated that Cre protein expressed by HSP in response to heat induction led to the recombination. In addition, the efficiency of the heat-induced recombination apparently decreased with development (Fig. 3C).

Spacially controlled induction of Cre/loxP recombination using an IR-LEGO system
To control Cre/loxP recombination spatially, we used the infrared laser-evoked gene operator (IR-LEGO) system (Fig. 4A) to induce local Cre expression by laser irradiation [12,15]. We heated the surface of the right neural placode of the telencephalon in stage 24 Tg (HuC:loxP-DsRed-loxP-GFP/HSP:Cre/Cry:BFP) embryos (2 dpf, Fig. 1A) by irradiation with an incident laser power of 21.6 mW for 1 sec. GFP fluorescence was detected in a subpopulation of neurons in the right telencephalon in the heatexposed Tg (HuC:loxP-DsRed-loxP-GFP/HSP:Cre/Cry:BFP) juvenile and adult fish only in the right side ( Fig. 4C and 4D). The juveniles were observed after hatching. In negative controls, GFP fluorescence was not detected in either heat-exposed Tg (HuC:loxP-DsRed-loxP-GFP) or non-treated Tg (HuC:loxP-DsRed-loxP-GFP/HSP:Cre/Cry:BFP) fish (Fig. 4C) and also in the left side of the treated Tg. These findings indicated that GFP-expressing neurons in both juvenile and adult brains were clonally-related cells derived from the micro area of neural placodes exposed to the infrared laser.

Discussion
The present study demonstrated that injection of a low concentration of Cre mRNA induces stochastic Cre/loxP recombination. There are two possible reasons for the stochastic Cre/ loxP recombination. The first is that the injected mRNA may be sparsely distributed. The second is that a low concentration in the cells may cause stochastic recombination, which is consistent with previous reports showing that low transcriptional Cre activity leads to stochastic recombination [30]. In the present study, we were unable to demonstrate the recombination at a single cell level, perhaps because more time (.12 h) is required for cells to express detectable levels of GFP than the cell cycle time of neural progenitors [28]. Thus, we cannot exclude the possibility that GFP-expressing cells comprise more than one clonal unit derived from a single stem cell.
Next, we established a noninvasive method for controlled Cre/ loxP site-specific recombination in the nervous system during medaka embryogenesis, which allowed us to visualize and/or modify the formation of a lineage-dependent structure. Recent studies with an advanced Gal4-UAS system in zebrafish promoted the genetic dissection of neural circuits [31], revealing how a subpopulation of neurons produces behaviors in juvenile fish [32]. The relationship between functional neural circuits and clonallyrelated neurons has, however, remained largely unknown. In the adult fruit fly brain, many areas of the brain neuropil are formed by the combination of distinct clonal units [33][34][35]. Recently, in the mouse visual cortex, clonally-related neurons derived from the same radial glia were found to have similar electrophysiologic properties (orientation selectivity) in response to a visual stimulus, in contrast to the response of the nearest-neighbor non-clonally related neurons [36,37]. These findings suggest that clonallyrelated neurons, rather than identifiable macroscopic brain structures, constitute the functional modules of neural circuits. Several behavioral systems to assess optomotor response, schooling, shoaling, and mating partner preference were recently established in medaka fish [38]. Combinations of these behavioral systems and the present gene manipulation method will contribute to our understanding of how a subset of clonally-related neurons (a lineage-dependent structure) is involved in behavior.
Furthermore, in the adult brain, medaka HuC promoter activity is restricted to newborn and differentiating neurons. Thus, heatinducible Cre/loxP site-specific recombination is also applicable for investigating adult neurogenesis in medaka fish. In contrast to mammals with limited neurogenesis in the adult brains, teleost fish such as medaka fish and zebrafish constitutively generate newborn neurons in numerous proliferating zones across the whole brain throughout life [21,26,[39][40][41]. As the distribution of proliferation zones in the adult brain is mostly conserved among teleost fish, adult neurogenesis is believed to be important for the maintenance and development of the fundamental structure of the fish brain throughout life [21,41]. Here we showed that stochastic and/or site-specific Cre/loxP recombination led to labeling of clonallyrelated neural progenitors that seem to form compartmentalized blocks in the telencephalon (Fig. 5). Although it is technically impossible to hit the same neuronal progenitor cells reproducibly, lineage analysis can be achieved by heat-induction in the same micro areas based on embryo morphology or the detection of fluorescent proteins. The use of a low laser intensity to theoretically target a single cell would allow us to induce recombination in a single or a few adjacent cells [15,42]. Determination of the distribution of GFP-expressing cells by repeated analysis would enable us to identify clonal units derived from the target micro area. The presence of identical subpopulations of clonally-related cells in multiple samples could reveal clonal units. Considering that the GFP-expressing cells comprise a single, or possibly a few clonal units, all the distribution patterns could be described by a combination of identified clonal units. Thus, this work provides a promising method for lineage analysis of neural progenitors in the adult brain and allows for further investigation of the involvement of such lineage-dependent structures in the formation of functional neural circuits mediated by adult neurogenesis.   Movie S1 Serial optical sections corresponding to Fig 2D. (AVI)