|Home | About | Journals | Submit | Contact Us | Français|
Glial fibrillary acidic protein (GFAP) is the major intermediate filament protein of astrocytes, and its expression changes dramatically during development and following injury. To facilitate study of the regulation of GFAP expression, we have generated dual transgenic mice expressing both firefly luciferase under the control of a 2.2 kb human GFAP promoter and Renilla luciferase under the control of a 0.5 kb human GAPDH promoter for normalization of the GFAP signal. The GFAP-fLuc was highly expressed in brain compared to other tissues, and was limited to astrocytes, whereas the GAPDH-RLuc was more widely expressed. Normalization of the GFAP signal to the GAPDH signal reduced the inter-individual variability compared to using the GFAP signal alone. The GFAP/GAPDH ratio correctly reflected the up-regulation of GFAP that occurs following retinal degeneration in FVB/N mice due to the rd mutation. Following kainic-acid induced seizures, changes in the GFAP/GAPDH ratio precede those in total GFAP protein. In knock-in mice expressing the R236H Alexander disease mutant, GFAP promoter activity is only transiently elevated and may not entirely account for the accumulation of GFAP protein that takes place.
GFAP is the major intermediate filament protein in astrocytes of the vertebrate central nervous system (Eng et al. 2000). Its expression is tightly regulated both during development and following injury. Increased expression of GFAP is also one of the hallmark features of gliosis, the consistent reactive response of astrocytes to injury or dysfunction in surrounding tissues or cell types (Eng and Ghirnikar 1994). What function GFAP provides to astrocytes is not well understood. However, dominant mutations in the coding region of GFAP account for nearly all cases of Alexander disease, a fatal neurogenerative disorder that usually presents as a leukodystrophy of childhood, with symptoms such as seizures, spasticity, and developmental delays (Brenner et al. 2001;Li et al. 2005). The GFAP mutations in Alexander disease appear to act in a gain-of-function fashion, and it is likely that elevated levels of GFAP, apart from expression of mutant protein per se, is central to the pathogenesis of this disorder (Hagemann et al. 2006).
Transgenic animals expressing reporter genes under the control of cell-specific promoters greatly facilitate the study of transcriptional regulation of gene expression in vivo (Cui et al. 1994). Early generation reporters included genes such as herpes thymidine kinase and human growth hormone, selected largely based on the ease with which their expression could be measured at the molecular or physiological levels. Transgenic models employing GFAP regulatory elements utilized second generation reporters that were uniquely valuable for morphological studies, such as the bacterial lacZ or jellyfish green fluorescent protein genes (Mucke et al. 1991;Brenner et al. 1994;Zhuo et al. 1997).
More recently, luciferase reporter genes have been developed that offer marked improvements over previous reporters in terms of sensitivity and dynamic range (Contag and Bachmann 2002). Zhu et al. (2004) have created transgenic mice expressing firefly luciferase under the control of the mouse GFAP promoter for the purpose of monitoring astrocyte response to injury in vivo. One can also imagine using transgenic animals as a source of cells for primary cultures. In applications of such cultures for high-throughput screening in drug discovery, a simple means for normalizing the luciferase signal would be highly desirable (Fan and Wood 2007). The availability of multiple luciferases with distinct substrate specificities allows one to introduce two transgenes at once, whereby the signal from one luciferase that is under the control of a cell-specific promoter can be normalized to the signal from a second luciferase that is under the control of a housekeeping promoter (Ullah et al. 2006). We now report the development of dual luciferase transgenic mice, using the human GFAP and GAPDH promoters to control expression of firefly and Renilla luciferase, respectively, with a goal towards creating better tools for studies of GFAP regulation both in vivo and in vitro.
To construct a transgene expressing firefly luciferase regulated by the GFAP promoter (GFAP-fLuc), we utilized the 2.2 kb human GFAP promoter derived from the GFAP-lacZ transgene previously described by Brenner et al. (1994). The GFAP promoter fragment (-2163 to +47) was subcloned into the Bgl II site in the pGL-3 basic vector (Promega) containing the firefly luciferase gene. In parallel, a transgene expressing the Renilla luciferase regulated by the human GAPDH promoter (Alexander et al. 1988) was prepared (GAPDH-RLuc), using a 0.5 kb fragment (-488 to +21) isolated from genomic DNA obtained from a human blood sample. Oligonucleotide primers were designed to generate XhoI and EcoRI restriction sites in the GAPDH PCR fragment (527 bp) that was amplified from the genomic DNA. The fragment was subcloned into XhoI and EcoRI restriction sites in the phRL-null vector (Promega) containing the Renilla luciferase gene. Both transgenes were sequenced to confirm correct orientation and sequence. Transgene fragments were isolated free of plasmid sequence and purified for microinjection using the Gene Clean Turbo kit (Q-Biogene).
The two transgenes were co-injected at equimolar ratios into the pronuclei of fertilized eggs on either FVB/N or FB6F1 hybrid backgrounds, using a combined concentration of 3 ng/ul. Genotyping was performed by PCR on genomic DNA isolated from tail biopsies at weaning, using the following conditions: one cycle at 95°C 3min, 32 cycles of 95°C 40s, 58°C 30s, 72°C 1min and a final extension step at 72°C for 5 min. The pr imers used for detection of the GFAP-fLuc transgene were 5’-TCTCTAAGGAAGTCGGGGAAGC-3’ (forward) and 5’-CAGCGGGAGCCACCTGATAGCCTT-3’ (reverse), and for detection of the GAPDH-RLuc transgene were 5’-CTCCCATCGGGCCAATCTCAGTCC– 3’ (forward) and 5’-GCGTTTGCGTTGCTCGGGGTCGTA-3’ (reverse). The primer locations in each transgene are shown in Figure 1A, B. The predicted sizes for the PCR fragments are 403 bp for the GFAP-fLuc transgene, and 551 for the GAPDH-RLuc transgene. Nine founder mice were identified that contained both transgenes, and were mated to FVB/N mice to establish lines. Two of the lines were used for the studies reported here, one derived from an FVB/N egg (Tg172-1) and one from an FB6F1 egg (Tg172-9). Transmission analysis from the founders to their offspring confirmed that the two genes segregated together and behaved as a single genetic locus. The Tg172-9 line of mice has been donated to the Jackson Laboratory for unrestricted distribution (stock #JR9638).
The dual luciferase reporter assay was performed according to the manufacturer’s recommendations (E1910, Promega), with minor modifications for use on extracts prepared from mouse tissues. The tissues were homogenized in the passive lysis buffer and spun at 17,500 g for 20 min at 4°C. An equal volume of the reagent including the substrate for firefly luciferase was added to 50-100 μg protein in 50 μl supernatant, and the signal was measured using a GloRunner Microplate Luminometer (Tuner Biosystems). Subsequently, the substrate for Renilla luciferase was mixed with the signal quenching solution in a 1:50 (v/v) ratio and 50 μl of this reconstituted reagent was added to each well. The luminescent signal by Renilla luciferase was measured immediately.
Frozen sections of brain (50 μm) were treated with blocking solution (10% donkey serum, 0.4% Triton-X 100, 0.02% Na-Azide in PBS) for 1 hr at RT. Sections were incubated with anti-firefly luciferase (Luc-17) mouse monoclonal antibody (1:100, Abcam) and anti-GFAP rabbit polyclonal antiserum (1:1000, DAKO) for 48 hrs at 4 °C. The sections were rinsed three times in PBS and incubated with Alexa-488 and Alexa-594 secondary antibodies for 1 hr at RT. The sections were mounted on slides using VectaShield Mounting Medium (Vector Labs). Immunofluorescent images were obtained from a Nikon C1 confocal microscope at the Waisman Center’s Cellular and Molecular Neuroscience Core facility.
Brains were divided in half down the midline and each sample homogenized in 2 ml lysis buffer containing 2% SDS, 50 mM Tris-HCl, 5 mM EDTA, pH 7.4, along with 1 mM PMSF and Complete Proteinase Inhibitor Cocktail (cat. #11836145001, Roche). The lysate was boiled for 30-40 minutes and the protein extract was prepared for GFAP quantification using a sandwich ELISA as previously described (Hagemann et al. 2006). The capture antibodies were the SMI-26 mouse anti-GFAP monoclonal cocktail (Covance Research Products), and the detection antibody was a polyclonal rabbit anti-GFAP (DAKO).
Kainic acid (A.G. Scientific, San Diego) was administered intraperitoneally (15 mg/kg) to male mice from the Tg172-9 line at 10-14 weeks (n = 20). Control mice (n = 4) were from the same transgenic line but injected with vehicle only (PBS). Seizures were evaluated at 5 minute intervals over a 2 hr period based on a clinical scoring system modified from that of Racine (1972): 0, normal; 1, immobile; 2, rigidity/tail extension; 3, repetitive motion/forelimb tremor; 4, rearing/falling; 5, multiple episodes of category 4 within 5 minute period; 6, tonic seizure; 7, death.
To generate Dual-Glo transgenic mice for the purpose of monitoring expression of GFAP, two transgenes were constructed separately, with the firefly luciferase placed under the control of the human GFAP promoter and the Renilla luciferase placed under the control of the human GAPDH promoter (Figure 1A, B). The transgenes were co-injected into the pronuclei of fertilized eggs, and integration was detected by observation of the predicted fragments from PCR of tail DNA (Figure 1C). Nine founders were identified that carried both transgenes, and lines were established from 7 of these founders. Firefly luciferase activity was measured in brains taken from mice of the F1 generation. Five of the lines exhibited relatively high levels of expression for both transgenes (Figure 2). Subsequent studies reported here utilized mice from the Tg172-1and Tg172-9 lines.
To investigate tissue specificity of expression of both transgenes, firefly and Renilla luciferase activities were measured in the same samples from mice of the Tg172-9 line. For the GFAP-fLuc transgene, activity was highest in brain, very low but detectable in heart, and undetectable in kidney, muscle, lung, and liver (Figure 3A). In contrast, the GAPDH-RLuc transgene exhibited comparable high levels of activity in brain and heart, and low but detectable levels in the other tissues (Figure 3B). Non-transgenic littermates yield values that are typically < 35 in raw units (or < 1% of the transgenic signal)(data not shown). These results indicate that the GFAP promoter is relatively brain-specific, and that the GAPDH promoter is variable but active in multiple tissues.
To determine whether the brain-specificity of GFAP-fLuc reflects expression in astrocytes, we attempted immunofluorescent localization of luciferase protein. The level of expression in un-injured mice was below the limit of detection by this method (data not shown). However, promoter activity is increased following exposure to the convulsant kainic acid (see below). Firefly luciferase protein becomes readily detectable under these conditions, and displays extensive co-localization with GFAP (Figure 4).
Initial measurements of firefly luciferase activities in brain revealed significant inter-individual variability (see Figure 3A). Co-integration and co-expression of the firefly and Renilla luciferase transgenes allowed us to test whether normalization of the GFAP signal to the corresponding GAPDH signal in the same sample would reduce this variability. We performed sequential measurements of the two luciferases in brains of mice from both the Tg172-1 and Tg172-9 lines. The individual values for each animal are shown for the GFAP-fLuc (Figure 5A) and GAPDH-RLuc (Figure 5B) activities. Normalization of the GFAP signal to the GAPDH signal dramatically reduced the variability (Figure 5C). To further assess the improvement in precision offered by this normalization, we calculated means, standard deviations, and coefficients of variation before and after normalization (Table 1). The coefficients of variation were reduced approximately 3-fold in both transgenic lines by normalization.
To determine whether the GFAP-fLuc transgene is responsive to a genetically-induced stimulus for gliosis, we took advantage of the fact that the FVB/N strain which was used for generating the lines is homozygous for the rd mutation. This mutation induces post-natal loss of photoreceptors in the retina and up-regulation of GFAP in Muller cells (Sarthy and Fu 1989). Simple F1 crosses to strains such as C57BL/6J will introduce one wild type allele at the rd locus and preserve normal retinal architecture, thus preventing gliosis. We therefore compared luciferase activities in mice carrying the transgenes in either the parent FVB/N background vs. FB6F1 hybrids. Previously the same strain comparison was used to document responsiveness of a GFAP-GFP transgene to genetic injury (Zhuo et al. 1997).
Retinal GFAP promoter activity in FVB mice was elevated compared to their F1 counterparts, before and after normalization (Figure 6A and C). Similar results were found in males (data not shown). To determine whether this difference is specific to the retina, we also measured GFAP promoter activity in samples of brain from mice in the two genetic backgrounds. The signal in brains of FVB mice was indeed increased compared to the F1 hybrids (1.4-fold), but not to the same degree as found in the retina. Interestingly, GFAP protein levels in whole brain are also elevated 1.7-fold in FVB mice compared to F1 hybrids (data not shown). Considering the GAPDH promoter alone, we found a 6-fold difference in the levels of Renilla luciferase activity in brain between FVB/N and FB6F1 mice (Figure 6B), but the basis for this difference is not known. However, the GAPDH promoter activity was not increased in response to photoreceptor degeneration (Figure 6B).
To determine whether the GFAP-fLuc transgene is responsive to chemical injury of the brain, and the temporal relation between changes in promoter activity and protein levels, we utilized kainic acid-induced seizures as a means to acutely activate astrocytes in the hippocampus. Adult mice from the Tg172-9 line were injected IP with kainic acid to induce seizures, and then luciferase activities and GFAP protein levels were measured during the subsequent two day period. Mice reliably began to show clinical signs of seizure activity at 10-15 min. after injection, reached maximum clinical scores at 75 min. post-injection, and then gradually recovered (Figure 7A). As expected, kainic acid injection resulted in increases in both GFAP promoter activity (normalized to the GAPDH signal) and GFAP protein levels (Figure 7B, one-way ANOVA, p<0.001). To evaluate the time course of these changes, we used t-tests as post-hoc assessments of promoter activity or protein levels at each time point in reference to the values at the 0 hr. time point. The GFAP promoter activity was significantly increased at 6 hrs after injection (9.86 ± 1.14 vs 6.39 ± 0.86, unpaired t-test, p<0.01) and then declined to baseline levels by 24 hrs. In contrast, GFAP protein accumulated more slowly, reaching a significant increase at 24 hours (762.3 ± 72.4 ng, unpaired t-test, p<0.001) and remaining elevated at 48 hours (535.8 ± 39.1 ng, unpaired t-test, p<0.01), compared to initial levels (289.1 ± 13.5 ng)(Figure 7B). Again, the GAPDH promoter activity showed no significant change at any of the time points measured (Figure 7C). These results indicate that the GFAP-fLuc transgene is responsive to chemically-induced gliosis, and correctly anticipates subsequent changes in protein levels.
Heterozygous mutations in the coding region of GFAP cause Alexander disease, yet a key factor in pathogenesis is thought to be elevated levels of the GFAP protein itself (Quinlan et al. 2007). To determine whether activation of transcription at the level of the GFAP promoter contributes to this increase, we crossed the Tg172-9 mice with a mouse strain that carries a common Alexander disease-associated mutation as a knock-in at the endogenous Gfap locus (R236H, equivalent to the human R239H mutation). Previous studies of the R236H mice have shown 6-fold elevations of GFAP mRNA, and 3.7-fold elevations of GFAP protein, at 8 weeks of age (Hagemann et al. 2006). However, as shown in Figure 8, the GFAP promoter activity (normalized again to the GAPDH signal) was only modestly increased in 2-3 week-old mice, and declined to normal levels by 4 weeks. These results suggest that promoter activation only partially accounts for the excessive accumulation of GFAP that occurs in these mice, and that other post-transcriptional mechanisms also play a role.
We describe generation of transgenic mice expressing both firefly and Renilla luciferase reporter genes under the control of the astrocyte-specific GFAP and housekeeping GAPDH promoters, respectively. The dual reporter assay in these mice reduces inter-animal variability and correctly reflects the direction and time course of gliosis-related changes in activity of the GFAP promoter. Previously, Zhu et al. (2004) described the generation of a transgenic line expressing just the firefly luciferase under the control of the mouse GFAP promoter. This line has subsequently been used for studying changes in GFAP promoter activity in live animals following injection of kainic acid (Zhu et al. 2004), Streptococcus pneumoniae infection (Kadurugamuwa et al. 2005), ischemia (Cordeau et al. 2008), and EAE (Luo et al. 2008). However, direct comparisons between the mouse of Zhu et al. (2004) and the ones reported here are difficult due to differences in methods of injury and quantitation of the reporter.
In principle, expressing Renilla luciferase under the control of a housekeeping promoter in the Dual-Glo assay should prove especially valuable in the context of assays where significant changes in global gene expression of astrocytes may take place. In the case of screening for compounds that reduce expression of GFAP, focusing on changes in the firefly luciferase signal alone might be mis-interpreted when non-specific toxicity causes down-regulation of many genes and perhaps even cell death. Unfortunately, there is considerable disagreement over which RNA/protein serves as the best internal control to use in normalizing expression data (Suzuki et al. 2000). In addition, while housekeeping promoters are widely used in cell culture to achieve uniform and high-level expression, they have proved problematic in transgenic animals with highly variable patterns of expression (Kisseberth et al. 1999). One luciferase transgenic line has been made utilizing a CMV-chicken β-actin promoter, but was noted to express the reporter only in a selected group of tissues that did not include brain (Cao et al. 2005). We chose the GAPDH promoter for control of the Renilla luciferase based on its frequent use in normalizing expression data, and recent studies indicate its utility for studies of chronic neurodegenerative disease in the mouse (Calvo et al. 2008). We utilized a relatively short fragment (~0.5 kb) of the human GAPDH promoter to control expression of Renilla luciferase, and found considerable variation between tissues, with brain expression being the highest. However, this variability is unlikely to be due to the length of the promoter, as non-uniform expression was also found in a mouse expressing firefly luciferase under the control of an 11.8 kb fragment from the human gene (Xenogen, unpublished observations). With the exception of skeletal muscle, our comparison of multiple tissues displayed a similar ranking to that found in a survey of GAPDH mRNA levels in human (Barber et al. 2005).
We employed three distinct injury models by which to induce gliosis in astrocytes and/or retinal Muller cells and cause up-regulation of GFAP expression. While the dual luciferase reporter mice described here detect increased levels of GFAP promoter activity in all three models, it is important to consider more closely how faithfully they report the precise time courses and levels of induction. For instance, we found using kainic acid that GFAP promoter activity was significantly increased at 6 hours after injection, but had declined to normal levels by 24 hours. Mucke et al. (1991), using lacZ as a reporter, also found rapid induction of the GFAP promoter following a stab wound injury. However, two other studies of kainic acid injury in mice found significant elevations of GFAP mRNA, and in one case GFAP driven luciferase activity, with slower and more sustained time courses (Kaasinen et al. 2000;Zhu et al. 2004), although the doses were also higher and genetic background may play a role. Secondly, the level of induction achieved by our dual luciferase reporter mice may be lower than that obtained with the mouse of Zhu et al. (2004). Using 30 mg/kg of kainic acid for injury, these investigators found approximately 10-fold elevations of both GFAP and luciferase mRNA levels at 48 hrs after injection, and ~40-fold elevation of luciferase signal as detected by imaging. In contrast, we observed a 2-fold change in luciferase activity, and only at 6 hours post-injection (though our dose of kainic acid was lower and we used tissue extracts for measuring luciferase activity). Similarly, in the rd model of gliosis we observed a 4.9-fold increase in GFAP promoter activity, whereas others have reported a 10-fold elevation of GFAP mRNA (although the age of the mice in this study was not specified) (Sarthy and Fu 1989).
In the case of a mouse expressing the R236H mutant GFAP that is a model of Alexander disease, we found that GFAP promoter activity is only modestly elevated, and only at early developmental time points, in contrast to the higher and more sustained elevations of GFAP mRNA and protein found in earlier studies of the same mice (Hagemann et al. 2006). One explanation for these discrepant results is that post-transcriptional mechanisms account for much of the elevated expression that takes place in the context of this disease. GFAP is a long-lived protein in vivo, with a half-life of approximately 8-9 weeks in spinal cord (DeArmond et al. 1986), and therefore may persist much longer than the luciferase reporter. Other studies in cell lines indicate that the accumulation of GFAP that occurs in Alexander disease in part reflects inhibition of proteasomal function (Tang et al. 2006). However, an alternative possibility to explain our results with the luciferase reporter is that the 2.2 kb human GFAP promoter, while capable of detecting up-regulation of GFAP expression in gliosis (Brenner et al. 1994;Zhuo et al. 1997), does not contain all the regulatory elements that contribute to up-regulation of the endogenous gene.
Finally, one appeal of luciferase transgenic mice is the opportunity to visualize reporter gene expression in living animals, as originally demonstrated using the firefly luciferase gene (Contag et al. 1998). With suitable substrates it may prove possible to perform dual imaging of firefly and Renilla luciferase activities in the same mouse (Bhaumik and Gambhir 2002).
We thank Denice Springman and Channi Kaur for technical assistance, and Michael Brenner for comments on the manuscript. This work was supported by NIH grants HD046599, NS060120, and NS042803 (to A.M.), and HD03352 (to the Waisman Center).
Publisher's Disclaimer: This is an Accepted Article that has been peer-reviewed and approved for publication in the Journal of Neurochemistry, but has yet to undergo copy-editing and proof correction. Please cite this article as an “Accepted Article”