Inducible targeting of CNS astrocytes in Aldh1l1-CreERT2 BAC transgenic mice

Background: Studying astrocytes in higher brain functions has been hampered by the lack of genetic tools for the efficient expression of inducible Cre recombinase throughout the CNS, including the neocortex. Methods: Therefore, we generated BAC transgenic mice, in which CreERT2 is expressed under control of the Aldh1l1 regulatory region. Results: When crossbred to Cre reporter mice, adult Aldh1l1-CreERT2 mice show efficient gene targeting in astrocytes. No such Cre-mediated recombination was detectable in CNS neurons, oligodendrocytes, and microglia. As expected, Aldh1l1-CreERT2 expression was evident in several peripheral organs, including liver and kidney. Conclusions: Taken together, Aldh1l1-CreERT2 mice are a useful tool for studying astrocytes in neurovascular coupling, brain metabolism, synaptic plasticity and other aspects of neuron-glia interactions.


Introduction
Cre-mediated recombination of target genes in adult astrocytes requires the use of an inducible expression system, because many promoters of the astrocyte lineage are also active in multipotential neural stem cells in the subventricular and subgranular zones (Christie et al., 2013). Thus, transgenic mouse lines have been generated for tamoxifen-inducible Cre recombination of target genes in mature astrocytes (Chow et al., 2008;Ganat et al., 2006;Hirrlinger et al., 2006;Mori et al., 2006;Slezak et al., 2007). However, none of them achieves sufficient recombination to study the function of genes in the majority of cortical and spinal cord astrocytes.
The aldehyde dehydrogenase 1 family member L1 (Aldh1l1), also known as 10-formyltetrahydrofolate dehydrogenase (EC 1.5.1.6), converts 10-formyltetrahydrofolate to tetrahydrofolate and CO 2 together with the reduction of NADP + (Kutzbach & Stokstad, 1971). The Aldh1l1 gene is expressed in a subset of radial glia in the midline of the embryonic CNS (Anthony & Heintz, 2007) and neuronal precursors (Foo & Dougherty, 2013). By transcriptional profiling in postnatal brain, Aldh1l1 was identified to be specifically expressed in astrocytes (Cahoy et al., 2008), which increase Aldh1l1 expression about tenfold with maturation (Zhang et al., 2014). To date, Aldh1l1 is regarded a pan-astrocyte marker, as determined in BAC transgenic mice with a fluorescent reporter protein or constitutive Cre expression under control of the Aldh1l1 promoter (Heintz, 2004;Yang et al., 2011). Therefore, we selected the Aldh1l1 regulatory region and a similar BAC transgenic strategy to target transgenic expression of CreERT2 to mature astrocytes.

Results and discussion
Efficiency of recombination in astrocytes of Aldh1l1-CreERT2 transgenic mice We generated Aldh1l1-CreERT2 transgenic mice by inserting a CreERT2 cassette (Sauer, 1994) under control of the Aldh1l1 promoter in a murine BAC (BAC RP23-7M9). Targeting the first coding exon of Aldh1l1 by homologous recombination, we substituted the open reading frame of exon 2 with the CreERT2 cDNA ( Figure 1a). Three lines of BAC transgenic mice were obtained by pronuclear injection, and crossbred with the Cre reporter mice ROSA26-Tdto or ROSA26-Eyfp (Madisen et al., 2010;Srinivas et al., 2001). Based on the degree of expression, one of the three lines of Aldh1l1-CreERT2 mice was selected for detailed characterization of double-transgenic offspring.
First, we determined the leakiness of reporter expression in adult Aldh1l1-CreERT2 mice. After corn oil injections in Aldh1l1-CreERT2*ROSA26-Tdto mice, we found very few labeled cells (less than 5 per section), demonstrating that the inducible Cre system operates tightly. In parallel experiments, adult Aldh1l1-CreERT2 mice were analyzed 7 days after tamoxifen induction. Sagittal brain sections revealed numerous tdTomato Cre reporter expressing cells, which in the forebrain exhibited the typical morphology of protoplasmic astrocytes ( Figure 1). Co-labeling revealed that almost all S100beta (S100 calcium-binding protein B) positive cells in hippocampus and cerebral cortex expressed tdTomato ( Figure 1, Table 1).
For comparison, when using a less sensitive EYFP Cre reporter line (Srinivas et al., 2001) in corresponding experiments, only two thirds of all S100beta positive cells in the cortex were also EYFP positive (Table 1). Thus, although both Cre reporter lines were generated as a knock-in into the endogenous ROSA26 locus, the recombination efficacy achieved is clearly different, in agreement with previous reports (Madisen et al., 2010;Srinivas et al., 2001). This finding illustrates the need to determine recombination efficiency individually for each combination of Cre allele and floxed target gene.
The expression pattern of some astroglial marker proteins, such as GFAP (glial fibrillary acidic protein), differs between protoplasmic astrocytes in the cortex and fibrous astrocytes in white matter. We therefore assessed the efficacy of Cre recombination separately for the corpus callosum, fimbria, hippocampus and spinal cord. Again, in all these regions a large majority of astrocytes, as defined by S100beta or GFAP, expressed Cre reporter, e.g. 85±1% in the corpus callosum and 94±2% in the fimbria (n=3 animals) ( Figure 2, Table 1). Co-labeling with GFAP was not used for cell counts because of the protein's low abundance in cell bodies which makes unequivocal quantification difficult.
In the cerebellum, a large fraction (89 ± 1%) of S100beta positive Bergman glia cells expressed the Cre reporter EYFP ( Figure 2c, Table 1). While 3.3 ± 0.3% of parvalbumin positive interneurons of the molecular layer expressed the tdTomato Cre reporter, none was double positive in corresponding experiments using the EYFP Cre reporter, confirming the sensitivity of the tdTomato reporter with a tendency for off-target recombination. Cre reporter expression was also observed in some neurons in the dentate gyrus and olfactory bulb, likely reflecting some recombination in adult neural stem cells in the subgranular and subventricular zone, followed by the migration of labeled progeny through the rostral migratory stream (Figure 1c).
Next, we compared Aldh1l1-CreERT2 mediated recombination with the expression pattern of EGFP in Aldh1l1-Egfp transgenic mice, generated with a similar BAC based strategy (Heintz, 2004). As expected, reporter and EGFP expression was nearly identical in the cortex, confirming the high efficiency of CreERT2 mediated induction of the tdTomato reporter ( Figure 3a).  Table 1. Efficiency and specificity of Aldh1l1-CreERT2 mediated recombination in brain. Efficiency and specificity of inducible Cre mediated recombination in adult Aldh1l1-CreERT2 mice crossbred with Cre reporter ROSA26-Tdto or ROSA26-Eyfp. For each value shown (average percentage), cells were counted on eight confocal images and two sections for each of n=4 animals. Efficiency is expressed as percent Cre reporter positive cells of all S100beta labeled cells. Specificity is expressed as percentage of all Cre reporter positive cells that lack immuno-labeling for S100beta.

Cellular specificity of Cre expression
Next, we tested the cell-type specificity of the Aldh1l1-CreERT2 transgene. Co-localization of tdTomato with markers for neurons (NSE, neuron specific enolase) or microglia (Iba1, ionized calcium binding adaptor molecule 1) was virtually absent ( Figure 4, Table 2). However, we observed a small fraction of Cre reporter positive cells co-localizing with Olig2 (oligodendrocyte lineage transcription factor 2), a transcription factor found in all oligodendrocyte lineage cells, including oligodendrocyte precursor cells (Figure 4b). Similarly, in triple transgenic mice that additionally express EYFP under control of the endogenous NG2 (neural/glial antigen 2) promoter (Karram et al., 2008), we identified 3.4% of double labeled cells, presumably oligodendrocyte precursor cells based on their localization and morphology. However, co-localization with a marker of mature oligodendrocytes (CAII, carbonic anhydrase 2) was negligible 12d after tamoxifen injections, and did not increase in mice that were analyzed 27 weeks after recombination (tamoxifen induction at 16 weeks of age). This suggests that the small percentage of Aldh1l1-CreERT2 expressing NG2 glia does not give rise to oligodendrocytes. An independently generated line of Aldh1l1-CreERT2 mice (Srinivasan et al., 2016) shows some Olig1, Olig2, CNP and CAII but no NG2 expression, as determined by ribotag-dependent transcriptome profiling (Sanz et al., 2009). Whether this dissimilarity is caused by the different detection methods employed remains to be determined.

Cre recombination in peripheral organs
Aldh1l1 is an enzyme of folate metabolism that is expressed in various peripheral organs (Krupenko & Oleinik, 2002). In agreement, we detected Cre reporter expression in liver, kidney, lung, and small intestine by direct immunofluorescence and Western blotting ( Figure 5). Cre reporter was not detected in heart muscle.

Conclusion
Aldh1l1 is a general marker for astrocytes within the CNS, and our new line of tamoxifen-inducible Aldh1l1-CreERT2 transgenic mice can be used to genetically target astrocytes in the mature CNS with high efficiency and specificity. When the corresponding genomic recombination in peripheral tissues is well tolerated, this line is suitable to study gene functions in astroglial cells of adult mice. Aldh1l1-CreERT2 mice will be made freely available upon request to the corresponding author.

Transgenic mice
All animal studies were performed at the Max Planck Institute of Experimental Medicine in compliance with the animal policies of the Max Planck Institute of Experimental Medicine and were approved by the German Federal State of Lower Saxony. All animals were housed in individually ventilated cages in groups of 3-5 mice per cage, kept in a room with controlled temperature (~23°C) under . We used a total of 26 mice of both sexes at the age of 7-10 weeks unless otherwise stated (20 -30 g body weight). All mice were analyzed as heterozygotes for the respective transgenic allele.
Generation of Aldh1l1-CreERT2 mice By PCR we introduced 50 bp of the Aldh1l1 intron 1/ exon 2 sequence 5' of the CreERT2 open reading frame. The bovine growth hormone poly A sequence (bGH pA), the frt (flippase recognition site) flanked kanamycin resistance cassette, and 50 bp of Aldh1l1 genomic sequence was inserted into an Nhe1 site 3' to the ERT2 sequence. The combined construct was introduced into exon 2 of the Aldh1l1 gene on the BAC RP23-7M9 (BACPAC Resources of the Children's Hospital Oakland Research Institute in Oakland), in frame with the start ATG, by homologous recombination in bacteria (EL250) as described (Lee et al., 2001). Excision of the resistance cassette was done by arabinose induced flippase expression. The BAC insert was excised by Not I digestion and purified by size exclusion chromatography using a sepharose column. Pronucleus injection gave rise to 5 transgenic founder mice. Genotyping was done by PCR of purified tail genomic DNA under standard conditions with the primers (5'-3', final concentration 0.25 µM) CAACTCAGTCACCCTGTGCTC and TTCTTGCGAACCTCAT-CACTCG amplifying the 3' part of intron1 of the Ald1l1 gene to the 5' part of the Cre open reading frame. Three out of five founder mice that were crossed with reporter mice showed expression in brain. Only one line (Aldh1l1-CreERT2 line 02) showed robust expression in forebrain astrocytic cells and minimal expression in other cell types of the brain.

Tamoxifen administration
Tamoxifen (Sigma, T5648) was dissolved in corn oil (Sigma, C8267) at a concentration of 7.5 mg/ml and injected intraperitoneally at 75 µg/g body weight on 5 consecutive days. We used a total of 26 mice of both sexes at the age of 7-10 weeks unless otherwise stated (20 -30 g body weight). Mice were analyzed 12 (immunohistochemistry) and 20 days (electrophysiology) after tamoxifen induction.

Antibodies
The following primary antibodies were used in this study: S100beta (
Author contributions JW performed most of the experiments and analyzed data. TD, SAB and SKS were involved in tissue preparation, immunoblotting and histology. Electrophysiological recordings were done together with SH. KAN initiated this project and edited the manuscript. GS designed experiments, performed analyses, and wrote the manuscript. All authors were involved in the revision of the draft manuscript and have agreed to the final content.

Competing interests
No competing interests were disclosed.

Grant information
This work was funded by the Deutsche Forschungsgemeinschaft (SFB/TR43 to KAN, SPP1757 SA 2014/2-1 to GS), by an ERC advanced grant to KAN, and by the CNMPB to SH. 1.

4.
astrocyte-specific, and yet present in the absolute majority of astrocytes homogeneously throughout the CNS. Prior models have reportedly suffered from either lower astrocyte specificity (such as certain Glast-CreERT lines), or lower levels of astrocytic expression, with significant differences depending on the brain area (such as in GFAP-CreERT2 lines). The authors commit to making their new mouse line freely available on request.
On balance, the new Aldh1l1-CreERT2 mouse line from Nave/Saher, together with a similar line independently generated and recently published by the Khakh lab have a large potential to advance astroglial research. Both labs have done a great service to the field by making their mice freely available. Consequently, these are likely to be highly useful tools for generating and studying astrocyte-specific knockouts and even knock-ins.
Detailed comments: In this paper the authors used 5 Tamoxifen injections to induce recombination in Aldh1l1-CreERT2 animals crossbred with either Rosa-26-tdTomato or -YFP mice, and quantified the % recombination in S100beta-positive cells (putative astrocytes) across different brain regions after 7 days by immunohistochemistry. A subset of cells was additionally examined after 20 days by electrophysiology. Other controls were also performed.
Co-staining with neuronal markers shows very limited overlap with neurons; recombination in a small (3%) subset of parvalbumin-positive cells of the cerebellum is judged to be the artifact of the tdTomato reporter line as it was absent in the YFP line. Neuronal expression in areas of adult neurogenesis is also reported, similar to what was already observed in prior astrocyte lines. Peripheral organs were also examined, and the authors report expected recombination in some of them.
The manuscript is technically sound and of high scientific quality. The paper provides a thorough characterization of the new line which has the potential to be highly useful for the field.
Below we list some minor suggestions for improving the paper, as well as some general points applicable for the entire field.
Issues to be addressed for Aldh1l1-CreERT2 mice: The authors could add more information regarding image quantification (Tables 1 & 2). Do "eight confocal pictures" mean single-plane images of different areas of the slice, or stacks? Of what thickness? What was the zoom, axial/lateral resolution, field of view?
The authors provide a helpful electrophysiological confirmation of reporter-expressing cells as passive astrocytes. As a very minor methodological point, they could indicate whether they used liquid junction potential correction.
Definitive marker for "astrocytes." Can the authors discuss why they think that S100Beta is the best marker of astrocytes? This is relevant in particular for correctly interpreting the existing S100Beta-negative population as astrocytic or non-astrocytic: indeed, the authors report that about 4-19% of S100Beta-negative cells also express reporters (Table 1). What are those cells?
Related to the previous point, the authors have performed GFAP co-staining but do not currently report the % co-labelling due to difficulty in quantification. An image from spinal cord is shown in Fig. 2d. However, some summary statement, even based on a limited number of manually analyzed cells, would be helpful. Do they see some of the S100beta-negative (reporter-positive) cells also positive for GFAP? "Titration" curve for tamoxifen induction One of the strengths of this mouse line is the very high . recombination efficiency across diverse astrocyte populations: ~90% of likely astrocytes after just five Tamoxifen injections. The authors emphasize the need to determine the % recombination for each individual line. Additionally, it would be useful in the future to know how differences in tamoxifen treatment regime correspond to different levels of recombination for some common lines (e.g. Rosa26-tdTomato). For instance, is a single injection sufficient to cause recombination in the bulk of astrocytes? Is it also possible to achieve sparser expression of the reporter (for single-cell imaging studies) with a reduced Tamoxifen administration (e.g. single-day)? Can the % recombination be raised over 90%, and if so, after how many injections? Obviously, the ultimate % recombination will depend also on a chosen reporter line (as apparent from Table 1), but more preliminary information for common lines would already be helpful.

David H. Rowitch
Department of Paediatrics, Wellcome Trust-Medical Research Council Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK This manuscript by Winchenbach described generation and characterization of a new transgenic tool et al to investigate astrocyte biology. The field is currently limited by cre recombinase driver transgenic reagents that show poor coverage and/or temporal control during central nervous system development. The current paper uses ALDH1L1 cre ERT2 BAC transgenic mice to address limitations in the field, developing a useful new tool. The data presented are technically sound and included the use of crosses with two conditional reporter alleles, single cell patch clamp electrophysiological analysis to confirm astrocyte features and inclusionary and exclusionary immunohistochemistry. The conclusions are supported by data and also highlight the utility of this new transgenic allele, compared to a commonly-used (GLAST-CRE) transgenic driver mouse, such that I think will be of significant interest in this new mouse line from the glial biology community.

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. Competing Interests:

Version 1
Reader Comment 04 Feb 2017 , University of Hamburg, Germany Christian Lohr Very nice piece of work! I'm sure the mouse line will be of great help for the community.
In case the authors prepare a revised version according to the reviewer's suggestions, I would like to add my two pence and suggest that the authors specify which colour is used for each channel in the dual colour images. The reader can only assume that, e.g., tdTomato is depicted in red and anti-S100B in green. Giving the information in the figure legends will increase the understandability of the figures. I disclose any competing interests. Competing Interests: