This study demonstrates the efficacy of generating inbred Tg rats using a lentiviral vector, which is a novel approach for generating Tg animals . Despite dozens of Tg mouse models of AD-like pathology, few Tg rat lines are available for AD research [10–13]. Since the manifestation of specific genetic disorders in transgenic models is expected to be unique to each species and strain, it is essential to control phenotypic variations that stem from the genetic constitution of the background strain. Thus we chose the inbred Fischer 344 rat strain for transgenic production in order to minimize individual variation among transgenic rats. In transgenic mouse models of AD, Lehman et al.  reported that genetic background had a significant influence on the regulation of APP and Aβ deposition in Tg mice that were created on different genetic backgrounds. For our studies, we chose inbred Fischer 344 rats due to their well-defined genetics as well as their common usage in studies involving aging . As Fischer 344 rats age, their brains are increasingly susceptible to oxidative stress, which is known to correlate with many neurodegenerative diseases, including AD. However, the approach that we employed to create germline Tg APPSw/Ind lines using Fischer 344 rats can be equally applied to other strains, providing unrestricted opportunities to create disease models on various genetic backgrounds.
In this paper, we report the successful generation of APPSw/Ind Tg Fischer 344 rats expressing human APP695 containing the Swedish and Indiana mutations under the control of the Ubi-C promoter using a lentiviral vector. The APP695 form was chosen because the predominant human APP isoform expressed in neurons of the central nervous system (CNS) is the APP695 isoform . Additionally, we chose to construct the double-mutant APP695 transgene to facilitate comparisons with existing APP695 transgenic mouse and rat models, particularly the Tg2576 mice reported by Hsiao et al. .
Long-lasting neuronal expression of transgenes is an important consideration in modeling neurodegenerative diseases such as AD. Our earlier efforts to create Tg APP-and PS1-overexpressing SD rats under the control of the CMV promoter using lentiviral vectors were successful in terms of efficient gene delivery. However, the Tg rats did not have the desired level of gene product as assessed by ELISA and Western blot analysis, even at 10 months of age (unpublished data). This lack of expression may be attributable to the phenomenon of gene silencing, which has previously been observed with a number of promoters, including CMV. In the present study, the easily detectable eGFP reporter gene was used to compare eGFP expression after stereotaxic injection of lentiviral vectors containing various promoters. In these studies, CMV, PDGF, and Ubi-C promoters all drove high-level expression of eGFP in-vitro. However, when tested in vivo after stereotaxic injection or creation of Tg animals, CMV-eGFP and PDGF-eGFP expression decreased dramatically over time, whereas eGFP expression remained strong for up to 13 months when driven by the Ubi-C promoter. Characterization of transgene expression in rat brains revealed long-lasting and selective neuronal expression of genes driven by the Ubi-C promoter. Stereotaxic injections were used to qualitatively screen CMV, UbiC, and PDGF promoters for ability to drive eGFP transgene expression within the hippocampus. The choice of SD rats was based on the use of this strain in the stereotaxic mapping of the rat brain by Paxinos & Watson . The dentate gyrus injection coordinates extrapolated from the atlas have been experimentally verified in our laboratory using SD rats and therefore we used SD rats to examine eGFP expression from the CMV, UbiC, and PDGF promoters. The goal of the promoter comparisons was not to characterize the expression patterns of the CMV, UbiC, and PDGF promoter in the CNS, which has previously been well documented, but rather to show qualitative examples of eGFP expression within the hippocampus following focal injection of lentivirus. Stereotaxic injections were performed in dentate gyrus to allow qualitative assessment of gene expression within the hippocampal formation, a region particularly vulnerable to neuropathological insult in AD. We determined that the eGFP signal consistently appears to be both nuclear and cytoplasmic in vitro and in vivo, regardless of the method of transgene delivery or type of promoter. Similar results were obtained by Wei et al. , who showed that eGFP diffused bidirectionally via the nuclear pore complex across the nuclear envelope.
To our knowledge, this is the first inbred, APP-transgenic rat model of AD that has substantial quantities of Aβ in serum. Prior to the generation of APP21 and APP31 transgenic rat strains, a Fischer 344 inbred AD model expressing APP (TgAPPSw) was reported by Ruiz-Opazo et al. . However, APP expression in these TgAPPSw rats was only 56% greater than in WT rats. In addition, APP-transgenic, outbred Wistar rats expressed 2.5 times more APP in hippocampus than did control rats  In the current paper, we report 2.9 times greater APP expression in the brains of inbred Fischer 344 rats than in WT controls. Due to the higher APP expression, APP21 rats could be useful models for examining the underlying mechanisms of AD progression and for developing and testing potential therapies for AD.
Several mouse models have been generated to study the effects of APP mutations. Hsia et al.  generated Tg APPSw/Ind mice. The characterization of these Tg mice indicated that the neurotoxic effects of Aβ may not require plaque formation. APP23 mice express 7-fold more APPSw than endogenous mouse APP and develop Aβ deposits at 6 months of age . Tg APP mice were generated using tissue-specific promoters such as enolase, platelet derived growth factor, and Thy-1 [27–29]. Similarly, we show that promoter choice significantly influences the expression of the transgene, such that Ubi-C is superior to CMV and PDGF in rats. The present study describes the generation of APP Tg Fischer 344 rats under the control of the Ubi-C promoter, which drives transcription in all tissues relatively stably. The expression of APPSw/Ind transgenic mRNA was detectable in all tissues analyzed for both the APP21 and APP31 lines. The expression of transgene was greater in the APP21 line than in the APP31 line. Since these Tg rats were generated as models of AD-like pathology, the expression level of the transgene is particularly important. The APP21 line showed 3 times greater cerebral APP expression compared to WT rats. The expression of brain APP in the APP31 line was about half that of the APP21 line. In addition, serum Aβ40 levels corroborate these findings. The APP21 line had significantly greater human Aβ40 than did the APP31 line. In transgenic mouse models, the expression of APP has reached levels as high as 10-fold greater than endogenous APP . However, protein expression level is not the only determinant of AD phenotype, as lower expression of mutant forms of APP can induce early and robust deposition of Aβ in brain parenchyma and/or vasculature [29, 30].
While the anticipation of stable transgene expression was a key consideration in selecting the Ubi-C promoter for our studies, selective neuronal expression driven by this promoter was unexpected. Our observations in brain and primary cultures from eGFP-transgenic SD rats confirmed the highly preferential expression of eGFP in neurons versus glial cells (Fig. 2). Potentially even greater levels of selectivity are suggested by our characterization of APPSw/Ind expression in transgenic Fischer 344 rats. In these animals, human APP was strongly expressed in neurons, but within the hippocampus, there was a strong demarcation based on the intensity of immunostaining in the pyramidal cell layer between CA1 and CA2 (Fig. 6). The basis for this selectivity is unclear, and additional studies will be required to fully evaluate the distribution of Ubi-C promoter-driven transgene expression in brain.
We chose Northern blot analysis to assess gene expression in APP21 and APP31 transgenic and WT rats and C3-3 mice. This allowed us to confirm the size of the complete transcript in the transgenic animals. Samples collected from brain yield a higher band for APP mRNA and 18S rRNA for both rat and mouse samples. Since both APP mRNA and 18S rRNA migration was retarded in brain RNA samples compared to the rest of the organs, we suspect that brain RNA samples contain residual substances that impede the migration of brain RNA. These could be residual lipids, as the brain contains more fat compared to the other organs analyzed.
Since the sequences of human and rat APP are highly similar, the APP probe used for Northern blot analysis did not distinguish between human and rat APP. This enabled comparison of native rat APP- and human APP-transgene expression. The transgene as well as the endogenous rat APP gene expression patterns showed significant tissue-specificity. Interestingly, both the human APP transgene and endogenous rat APP mRNA were more abundant in kidney and lung than in heart and liver. Although tissue-specific expression of APP in humans was reported previously , tissue-specific expression of the APP transgene was unexpected in the transgenic rats because the Ubi-C promoter drives expression of the human APP transgene in this construct. We have generated purinergic receptor Y2- (P2RY2) transgenic rats using the same vector backbone, and expression of the P2RY2 transgene driven by the Ubi-C promoter did not show tissue-specificity (unpublished data).
Tissue specific expression of APP transgene can be due to APP mRNA stability, RNA silencing or transcription regulatory elements within APP cDNA. The 3'UTRs of human, rat and transgenic APP are substantially different which reduces the possibility of tissue specific differences in APP mRNA stability. In addition, we were unable to find a candidate sequence for RNA silencing within the rat or human genome. Thus, regulation through enhancer or silencer elements within the APP cDNA is a possible explanation for tissue-specific expression regardless of the promoter that drives the expression of APP. A recent publication by Collin and Martens  supports the presence of transcription regulatory role of APP cDNA.
In order to confirm tissue-specific expression of APP, we analyzed APP transgene expression in C3-3 transgenic mice, which is driven by the ubiquitously expressed prion protein promoter . Expression levels of endogenous prion protein are similar in various tissues. We suggest that the similarity of tissue-specific expression patterns of APP transgenes that are driven by two different promoters (Ubi-C and PrP) in two different species (rat vs. mouse) strongly supports the presence of transcription regulatory elements within the APP cDNA. The temporal and spatial expression differences in APP-transgenic mice have been attributed to the promoters (PDGF, and Thy-1) used to drive expression . However, we believe that the changes in APP expression patterns shown in our study cannot be explained by the promoter because of the stable expression driven by the Ubi-C promoter in most tissues. We hypothesize that elements within the cDNA sequence may regulate the expression of APP tissue-specifically. Identification of these elements might ultimately broaden treatment options for AD. In conclusion, these APP-transgenic rats could be a useful model in which to study the regulation of APP expression as well as pathogenic mechanisms in AD.