Rax-CreERT2 Knock-In Mice: A Tool for Selective and Conditional Gene Deletion in Progenitor Cells and Radial Glia of the Retina and Hypothalamus

To study gene function in neural progenitors and radial glia of the retina and hypothalamus, we developed a Rax-CreERT2 mouse line in which a tamoxifen-inducible Cre recombinase is inserted into the endogenous Rax locus. By crossing Rax-CreERT2 with the Cre-dependent Ai9 reporter line, we demonstrate that tamoxifen-induced Cre activity recapitulates the endogenous Rax mRNA expression pattern. During embryonic development, Cre recombinase activity in Rax-CreERT2 is confined to retinal and hypothalamic progenitor cells, as well as progenitor cells of the posterior pituitary. At postnatal time points, selective Cre recombinase activity is seen in radial glial-like cell types in these organs – specifically Müller glia and tanycytes – as well as pituicytes. We anticipate that this line will prove useful for cell lineage analysis and investigation of gene function in the developing and mature retina, hypothalamus and pituitary.


Introduction
The mammalian central nervous system (CNS) is comprised of hundreds of distinct neuronal and glial subtypes [1][2][3], and the genetic mechanisms that guide the differentiation of these cell types from neuronal progenitors are still poorly understood. In recent years, the use of mice carrying targeted conditional mutations has led to a considerable advance in our understanding of this process. The retina is a readily accessible and tractable system for studying CNS cell subtype specification. It is comprised of three cellular layers, each of which contains seven major cell types. Individual cell types can be readily identified based on morphological and molecular criteria, and each is specified from retinal progenitor cells during a discrete temporal window.
A number of mouse lines were generated in the last 15 years that use the regulatory regions of retinal progenitor-specific genes to drive expression of Cre recombinase, allowing for the selective manipulation of gene function in retinal progenitors [1][2][3][4][5][6][7]. However, each of these genetic lines has a number of inherent features that limits their usefulness for studies of gene function during retinal development. Several, such as Chx10-Cre and Pax6-aCre, are active only after neurogenesis has been initiated at approximately embryonic day (E)11, and are thus not useful for studying the genetic control of earlier stages of retinal development [4,5]. Other lines, such as Dkk3-Cre and Nes-CreER T2 , are active in many other CNS regions [6,7]. Several others, such as the Six3-Cre and Rax-Cre transgenic mice show selective retinal and ventral forebrain expression starting from E8.5 but are constitutively active, making it impossible to selectively study gene function at different developmental time points [8][9][10]. The lone exception is the Rax-IRES-tTA-tetO-Cre line [11], which expresses Cre from the 39 end of the endogenous Rax transcript in a doxycyclinerepressible manner. However, these animals show ectopic Cre expression in the developing heart, as well as variable levels of microphthalmia, making them less than ideal for certain applications. Furthermore, the need to continuously supply doxycycline to maintain repression of Cre expression imposes an additional burden in maintaining this line.
We sought to build on these previous efforts to address the still unmet need for a chemically-inducible Cre line that is selectively expressed in retinal progenitors. We generated a mouse line in which tamoxifen-inducible Cre recombinase was inserted into the endogenous Rax locus. Since Rax mRNA is also expressed in hypothalamic progenitors, we expect that these animals will also be useful for studying gene function in the developing hypothalamus, a topic which has recently begun to attract considerable attention [12][13][14][15][16]. In addition, in these animals Rax mRNA is expressed in cells of the developing neurohypophysis or posterior pituitary [17]. In adult animals, Rax mRNA is selectively expressed in hypothalamic tanycytes [18], a radial glial-like cell type that have been implicated in control of energy balance and metabolism [19][20][21], and have also been found to act as adult neural progenitors [18,22,23]. There are few useful Cre lines available for studying hypothalamic development, and no tanycyte-specific Cre lines been reported to date [7,22,24]. Mice that are heterozygous for null mutations of Rax show no obvious defects in either retinal or hypothalamic development, and only exhibit some mild changes in tanycyte gene expression and barrier function [25].

Results
To generate Rax-CreER T2 knock-in mice, we constructed a targeting vector in which a CreER T2 cassette was inserted immediately downstream of the initiation methionine of the endogenous Rax gene, followed by a pGK-neo sequence flanked by Frt sites for selection on targeted ES cells ( Figure 1A). Southern blot analysis performed using a 59 probe external to the 7.5 kb long left arm of the targeting construct showed the presence of both 11.6 kb and 12.9 kb bands in AflII/SexAI-digested genomic DNA, while wildtype ES cells exhibited only an 11.6 kb band ( Figure 1B). Following mating to FLPeR mice [26], removal of the pGK-neo cassette was confirmed using PCR analysis (data not shown).
Rax-CreER T2 mice were then mated to R26-CAG-lox-stop-lox-TdTom (Ai9) mice, which express the red fluorescent Tdtomato protein under the control of the ubiquitous CAG promoter, following Cre-dependent excision of a transcriptional stop cassette ( Figure 1C) [27]. We first determined whether adult animals that had not been treated with 4-hydroxytamoxifen (4-OHT) during any stage of development exhibited any evidence of Cre activity. Aside from a few rare Tdtomato-positive cells in the hippocampus and hypothalamus, no evidence for Cre activity was observed in the absence of 4-OHT (data not shown).
To determine whether Cre activity could be selectively induced in Rax-expressing progenitors during early stages of CNS development, we delivered a single oral gavage treatment of 4-OHT delivered at E8.5, and analyzed Tdtomato expression at E10.5 ( Figure 2A). We observed intense mosaic expression of Tdtomato in neural progenitors of the ventral forebrain, and labeled cells were detected in both the telencephalon and diencephalon ( Figure 2C-G). Retinal progenitors are also robustly labeled ( Figure 2H). No other cell types were labeled. This pattern of Cre activity corresponds precisely to the previously reported domains of Rax mRNA expression at E8.5 [17,28], and confirms the specificity of Cre activity. Table 1 lists the dose of 4-OHT administered and the number of Rax-CreER T2 ;Ai9 mice examined for this and all other experiments in this study.
We next tested Rax-CreER T2 ;Ai9 mice at later stages of development to determine whether 4-OHT-induced Cre activity faithfully replicated endogenous Rax expression patterns [12]. A single oral gavage treatment with 4-OHT at E10.5 resulted in selective induction of Tdtomato in ventral hypothalamic progenitors by E12.5 ( Figure 3A), along with more limited induction in anteriodorsal hypothalamic progenitors ( Figure 3A, white asterisk) and newly generated hypothalamic neurons ( Figure 3A, dashed square). Hypothalamic progenitors that are located between the tuberoventral and anteriodorsal domains are unlabeled ( Figure 3A, white arrowhead), and likely correspond to Dlx1/2-expressing progenitors that give rise to structures such as the dorsomedial hypothalamic nucleus, which does not express Rax mRNA [12,13]. Robust labeling of the posterior pituitary (PP) was also observed ( Figure 3D), along with progenitors throughout the retinal outer neuroblastic layer (ONBL) and postmitotic cells of the inner neuroblastic layer (INBL) ( Figure 3E). Red fluorescence in axons in the optic nerve indicated that these were retinal ganglion cells (RGCs), one of the earliest-born retinal cell types [29]. Cre induction at E12.5, the peak period of hypothalamic neurogenesis, resulted two days later in widespread and selective labeling of ventral hypothalamic progenitor and precursor cells ( Figure 4A-D). Selective labeling of neuroretina is also observed, with red fluorescence in RGC cell bodies and axons being particularly prominent ( Figure 4E). This likewise closely matches the reported pattern of Rax expression at E12.5 [12].
In contrast, 4-OHT injection at E16.5, after embryonic hypothalamic neurogenesis is completed, resulted in Tdtomato expression that was mostly restricted to vimentin-positive radial glia-like cells lining the ventral walls of the third ventricle ( Figure 5A-C,E). A limited number of Tdtomato-positive cells were also found in the hypothalamic parenchyma ( Figure 5A-C). Tdtomato expression is also observed in vimentin-positive cells of the posterior pituitary ( Figure 5E). In the central retina, Tdtomato expression is primarily observed in the ONBL, with only an occasional positive cell in the INBL ( Figure 5D), and colocalization of Tdtomato with the RGC marker NF165 is not observed ( Figure 5H). This corresponds well with the known time course of RGC generation, which is essentially complete in the central retina by E16.5. In contrast, in the peripheral retina, more Tdtomato-positive cells and processes are seen in the INBL and some colocalization of Tdtomato is seen in the optic nerve, indicative of expression in later born RGCs of the peripheral retina ( Figure 5D). In addition, colocalization of Tdtomato with GFAP-positive retinal astrocytes was seen in the INBL ( Figure 5F). These do not derive from retinal neuroepithelium, but instead migrate into the retina along the optic nerve [30]. This observation suggests that that a subset of retinal astrocytes may be derived from Rax-expressing hypothalamic progenitor cells. Finally, a subset of Tdtomato-positive cells in the ONBL were double-labeled with Olig2, which is expressed in a subset of lineage-restricted retinal progenitor cells ( Figure 5G) [31].
We next investigated Cre activity at three postnatal time points. In the hypothalamus, Tdtomato expression is restricted to hypothalamic tanycytes following 4-OHT intraperitoneal (i.p.) injection at P7, P28 and P50 ( Figure 6A-I), which again recapitulates the previously reported pattern of Rax mRNA expression [4,12,32]. Td tomato labeled cells robustly expressed Sox2 and vimentin, both of which are prominently expressed in tanycytes ( Figure 6J). TdTomato expression is not observed in more dorsally located ependymal cells of the third ventricle at any postnatal stage. Tanycytes are more efficiently labeled at P7 and P28 than at P50 ( Figure 6A-I). At both P7 and P28, efficient induction of Tdtomato expression is observed in ventricular zones that a1, a2, b1 and b2 tanycytes ( Figure 6K). Daily i.p. injections of corn oil alone from P18-22 do not result in any Tdtomato labeling of tanycytes in Rax-CreER T2 ;Ai9mice (Figure 7A-C right panels). Considerably more tanycyte labeling was seen following a series of three 4-OHT injections from P5-P9 when compared to a single injection at P7 ( Figure 7D,E).
In the neuroretina, robust and selective Cre activation is seen in radial Müller glia following 4-OHT i.p. injection at P7 ( Figure 8A), likewise reflecting the previously reported developmental pattern of Rax mRNA expression [18,28,33]. Surprisingly, despite the fact that only very low levels of Rax mRNA expression have been reported in mature Müller glia [28,34,35], we also observe a less efficient but highly selective induction of Tdtomato in Müller glia of animals injected at both P28 and P50 ( Figure 8B-E). Finally, at all three stages, we also observe robust and selective induction of the reporter in the non-pigmented zone of the ciliary body ( Figure 8F-H).
In addition, we observed robust and selective labeling of a subset of cells in the posterior pituitary following 4-OHT i.p. injection at P28 ( Figure 9A-B). These cells are immunopositive for vimentin, but not GFAP or neurofilament, indicating that they are most likely fibrous pituicytes [36], or other glial cells of the neurohypophysis ( Figure 9C-H) [36,37]. Surprisingly, we also observed Tdtomato labeling in a subset of cells in the granule layer of the flocculus of the cerebellum when 4-OHT was delivered between P5 and P9 ( Figure 9I), though not at any point earlier or later. Though not previously reported in the literature, inspection of in situ hybridization data obtained from the Allen Brain Atlas revealed that Rax mRNA was indeed selectively expressed in the flocculus at P4 but not at later time points ( Figure 9J). Finally, we examined other organs outside the brain of the Rax-CreER T2 ;Ai9 mice, including skeletal muscle, heart, kidney, liver, intestine, and adrenal, but did not find reporter expression in any other tissue (data not shown).

Discussion
This study has identified Rax-CreER T2 knock-in mice as potentially useful for studies of both retinal and hypothalamic development, as well as for studying the biology of Müller glia and tanycytes. Heterozygous Rax-CreER T2 knock-in mice are phenotypically normal and show no ectopic Cre activation. In addition, the pattern of 4-OHT induced Cre activity precisely recapitulates the expression pattern of the endogenous Rax gene. Indeed, we find that this line is selectively active in mature Müller glia and in cells of the ciliary margin, which express Rax mRNA at very low levels [38][39][40]. We anticipate that this mouse line will be particularly useful for analyzing temporally dynamic functions of genes that are important for retinal development; examples include homeodomain factors such as Pax6, Chx10 and Lhx2. These factors are both broadly expressed in RPCs and also thought to act in discrete temporal windows to control specification of individual retinal cell types [41][42][43][44][45]. It will also likely prove useful for studies of the genetic mechanisms guiding hypothalamic cell specification during embryonic development. Because Cre is selectively active in Müller glia and tanycytes, both of which may act as neural progenitors in adults [18,[46][47][48], this mouse line makes it possible to prospectively label these cells for lineage analysis and determine the fates of the cells they generate. Furthermore, this mouse line can be used to selectively disrupt the function of genes that are candidates for regulating proliferative   Figure   E8.5RE10.5 1.5 mg Gavage into mother 3 Figure 2 E10.5RE12.5 1.5 mg Gavage into mother 2 Figure 3 E12.5RE14.5 1.5 mg Gavage into mother 3 Figure 4 E16.5RE18.5 1.5 mg Gavage into mother 4 Figure

Ethics statement
All mice used in these studies were maintained and euthanized according to protocols approved by the Institutional Animal Care and Use Committee at the Johns Hopkins School of Medicine.

Mice
C57BL/6J wild type mice were purchased from Charles River and were bred with chimeric mice to determine germline transmission of the Rax-CreER T2 targeted locus. The neomycin cassette was removed using FLPeR mice [26] donated by Dr. Jeremy Nathans (Johns Hopkins). The FLPeR was bred out of the mouse line after the neomycin cassette elimination. R26-CAG-loxstop-lox-TdTom (Ai9) mice were a generous gift from Dr. Xinzhong Dong (Johns Hopkins).

Construction of Rax-CreER T2 targeting and control vectors
We used recombineering to retrieve a 14 kb fragment from a C57BL/6 background Bacterial Artificial Chromosome (BAC) containing the Rax locus. Briefly, SW106 cells containing the BAC were electroporated with a pGK-DTa vector containing two previously cloned 500 bp retrieving arms (a 39 arm and a 59 arm) homologous to the ends of the chosen 14 kb BAC region. In parallel, we generated a control vector with a longer short arm, so as to allow testing of PCR primers used in screening ES cells with correctly integrated targeting cassettes. The control vector was designed with a 500 bp 39 arm that was homologous to the BAC region 500 bp downstream from the 39 arm used in the construction of the CreER T2 targeting vector. We activated the recombineering machinery of the SW106 cells (containing the BAC and the pGK-DTa vector) as described [49], which induced the retrieval of the 14 kb BAC region by homologous recombination between the BAC and the 39 and 59 500 bp arms of the pGK-DTa vector. The resulting vector (pGKDTa-LA-Rax-SA) consisting of the Rax locus (4.5 kb), an upstream 7.5 kb long arm (LA), a short 2 kb arm (SA) and the negative selection cassette pGK-DTa, was then used for a second round of recombineering with the vectors containing the CreER T2 -neo targeting constructs. The Rax-CreER T2 -neo constructs were generated using traditional cloning as follows: first, we cloned two 500 bp arms homologous to genomic sequence upstream (59 arm) and downstream (39 arm) from the eighth nucleotide downstream of the first base in the initiator methionine codon of Rax (UCSC genome browser July 2007 (NCBI37/mm9) Assembly) into pBluescript. We next introduced a neomycin cassette between these two arms. We then cloned the CreER T2 (obtained from Addgene -pCAG-CreER T2 , Plasmid 14797) upstream of the neomycin cassette, to generate the CreER T2 -neo construct. The resulting CreER T2 -neo cassette was inserted into the Rax locus by homologous recombination between the 500 bp arms of the CreER T2 -neo construct and the pGK-DTa-LA-Rax-SA vector.
The control construct was generated in the same way as the targeting vectors, except that the control construct contained only the neomycin cassette and its short arm was 1 kb longer. A longer short arm allowed the standardization of a PCR protocol to test whether the 39 portion of the targeting construct had been correctly inserted into the endogenous Rax locus. In this PCR protocol, the forward primer targeted the neomycin cassette and the reverse primer targeted the second intron of Rax; a 3 kb fragment was produced only when the constructs were inserted in the correct location. PCR primer pairs suitable for identifying the correctly inserted short arm were then identified using the control construct as a template. Following this, the Rax-CreER T2 -neo targeting vector was fully sequenced, confirmed as correctly designed. This vector was then submitted to the Johns Hopkins ES Cell Core Facility for electroporation into MC1 mouse ES cells derived from 129S6/SvEvTac, using previously described protocols [50].

Screening of electroporated ES cells
Embryonic stem cell colonies electroporated with either the targeting construct Rax-CreER T2 or the control construct. These were then screened for correct locus insertion by PCR amplification of a 3 kb DNA fragment using a forward primer located in the neomycin cassette 59-CTTCTATCGCCTTCTTGACG-39 (P1) and a reverse primer located in chromosome 18 within the second intron of the Rax gene forward 59-TTTGTCATTTCCCCTCG-TAG-39 (P2). This genotyping was performed using the following PCR touchdown protocol: 95uC 5 min, 94uC 30 min, 70uC 1 min; minus 1uC each cycle (15 times), 72uC 3 min, 94uC 30 sec, 55uC 30 sec for 30 cycles, 72uC 3 min, 72uC 7 min, 4uC hold.

Southern blotting
ES cell colonies that tested positive by PCR for correct 39 insertion into the Rax locus and the ES cell colonies electroporated with the control vector were grown further. DNA was extracted using the Blood & Cell Culture DNA Mini Kit from Qiagen (Cat. 13323), following the manufacturer's protocol. Purified DNA was digested with AfIII and SexAI (Rax-CreER T2 -neo construct) and used for Southern blotting. We designed a 511 bp probe amplified by PCR from a wild type C57BL/6J DNA using the following primers: forward 59-ATGCATTTAGATGCCTGATTGC-CAAT-39 and reverse 59-ACGCGTCAAAACCACAGTAAAC-CAAG-39. This probe hybridized 5.4 kb upstream of the long arm of the targeted locus. The probe labeling and Southern blot were performed using the DIG-High Prime DNA Labeling and Detection Starter Kit I from Roche (Cat. 11745832910) following the manufacturer's protocol. Karyotyping and ES cells pronuclear injection into ICR host embryos were performed at the Johns Hopkins Transgenic core as previously described [50].
Genotyping DNA for mouse genotyping was obtained from tail tips collected carefully with fresh razor blades to avoid cross contamination. Tails were incubated at 55uC overnight in lysis solution containing proteinase K (0.1 mg/ul). DNA extraction was performed as previously described [25].
Cre induction by 4-hydroxytamoxifen in Rax-CreER T2 ;Ai9 mice 4-OHT was prepared by sonicating 50 mg of 4-OHT (Sigma) in 1.5 mL of 100% ethanol at 33uC. 90 ml of 4-OHT/ethanol was vortexed with 300 ml of corn oil, and ethanol removed by vacuum centrifugation. Transgenic mice, double heterozygous for Rax-CreER T2 and Ai9, were given varying doses of 4-OHT depending on the mouse age (see Table 1). Cre activation in embryonic mice was induced through oral gavage of the mother. Postnatal mice were given 4-OHT by i.p. injections.

Tissue collections and histology
Mice were harvested at multiple embryonic and postnatal time points. For embryonic mice, gestational age was determined by counting the morning of the vaginal plug detection as embryonic day 0.5 (E0.5). Crown to rump lengths were also checked to verify gestational age. Embryos were dissected and fixed in 4% paraformaldehyde (PFA) in 16PBS (w/v) overnight at 4uC. Embryos were then washed three times with PBS and immersed in 30% sucrose in 16PBS (weight/volume) at 4uC for several days. Postnatal mice were anesthetized using a cocktail of ketamine, acepromazine, and xylazine (9:9:3 volume ratio), and transcardially perfused with 4% PFA. The brain, eyes, and pituitary gland were removed and post-fixed overnight at 4uC with 4% PFA on a rotating platform, followed by 24 hours immersion in in 30% sucrose/16 PBS. Tissues were then embedded in OCT (Tissue Tek, Sakura Finetek), following which either 25 mm (for embryonic tissue) or 40 mm (for postnatal tissue) sections were cut on a cryostat.