The study was carried out in strict accordance with the guidelines of the OLAW-NIH (Office Laboratory Animal Welfare). The protocol was approved by the local Animal Care Committee of Fundación Instituto Leloir (FIL) Assurance# A5168-01.

Hemizygous transgenic McGill-R-Thy1-APP (Tg^+/−) rats, harboring the human APP751 transgene with the Swedish and Indiana mutation under the control of the murine Thy1.2 promoter (Leon et al., 2010), were provided to Fundación Instituto Leloir (FIL) by The Royal Institution for the Advancement of Learning/McGill University, Montreal, Quebec (Canada).

An in-house breeding colony was established at FIL by cross-breeding Tg^+/− males with wild-type Wistar females rats. Each litter was culled to 10 pups (up to 8 males per litter were left). At weaning (21 post-natal days), dams were removed from the litters and pups were genotyped as described below. When the results of the genotyping were available (between 22 and 23 post-natal days), rats were pseudorandomly assigned to experimental groups. To avoid the litter effect, groups were made up of pups from 3 to 4 different litters. Animals were maintained on a 12-h light/12-h dark cycle, group-housed 2–4 per cage and had ad libitum access to water and a standard rodent diet. Tissue samples were obtained by ear punches from the litters to determine the genotype of each rat. Genotyping was performed using a validated protocol. Briefly, genomic DNA was prepared from ear biopsy tissue using proteinase K digestion, followed by phenol/chloroform/isoamyl alcohol extraction (Life Technologies). DNA quantification was determined in a Nanodrop 2000 spectrofotometer. Samples with a ratio of absorbance at 260/280 and 260/230 nm above 1.80 were used for PCR to recognize whether or not the rat carried the human APP (hAPP) gene. The following sets of primers were used: APP-F 5′-AGGACTGACCACTCGACCAG-3′; APP-R 5′-CGGGGGTCTAGTTCTGGAT-3′; CyclophilinG-F 5′- TACAACAGTAGAACAAGGGAGCGAAG-3′ and CyclophilinG-R 5′-ATCCCTCCTTCTTCTCCTCCTATCTTT-3′. Cycling conditions were: 95°C (15 min) and 40 cycles of 94°C (30 s), 54°C (1 min), 72°C (1 min) followed by one step at 72°C (2 min) and kept at 4°C until processed. PCR products were resolved by 2% agarose mini-gel electrophoresis in TBE buffer and visualized with ethidium bromide (10 mg/ml) staining under UV light. Cyclophilin G amplicon (248 bp) was used as a housekeeping gene to verify DNA integrity and PCR conditions. The hAPP amplicon (387 bp) indicates the presence of the human transgene. To verify the hemizygocity of the transgenic rats, quantitative PCR (qPCR) was performed by using the same set of hAPP primers as described above and the following set of GAPDH (used as housekeeping gene) primers: GAPDH-F: 5′-GGGGAAGGACGCTGTACGGG-3′ and GAPDH-R: 5′-AAGGGGAGC AACAGCTGGGGT-3′. A total of 1 ng of genomic DNA was used per well (5 μl of a 0.2 ng/μl dilution). Samples were plated in triplicate. SYBR-Green qPCR was performed by using the KAPA SYBR^® FAST Universal 2X qPCR Master Mix (Kapa Biosystems). Reactions were run in a Stratagene Mx3005P cycler and results analyzed by the MxPro software in a Comparative Quantitation mode. In all of the runs a genomic DNA sample of a hemizygous animal was included and used as an internal calibrator. If the unknown sample is hemizygous, the relative quantity (dRn) value should be between 0.8 and 1.3; if the unknown sample is homozygous the result is close to a value of 2 (between 1.8 and 2.3).

For time-course analysis of the behavioral phenotype, 3 cohorts of male rats consisting in Tg^+/− and wild-type (WT) littermates were used. Rats were divided in 6 experimental groups as follows: group 1 (3 months old Tg^+/− rats; n = 10); group 2 (3 months old WT rats; n = 9); group 3 (6 months old Tg^+/− rats; n = 9); group 4 (6 months old WT rats; n = 7); group 5 (12 months old Tg^+/− rats; n = 11); group 6 (6 months old WT rats; n = 11) (see Figure 2). Only male rats were used to avoid any potential effects of changes in the female estrus cycle on behavioral performance.

Three groups (n = 3/group) of naïve Tg^+/− male rats of 3, 6, and 12 months of age, and one group (n = 3) of naïve WT male rats of 12 months of age, were anesthetized with ketamine (50 mg/kg) and xylacine (10 mg/kg), perfused transaortically with 0.9% sodium chloride with 200 U/l heparine and brains were quickly removed as described previously (Leal et al., 2006). Brains were divided into left and right hemispheres. Hippocampus was isolated from left hemisphere and retained for biochemical studies. The right hemisphere was post-fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4 overnight at 4°C, and finally transferred to a solution of 30% sucrose in 0.1 M PB for 5 days, or until sectioned for immunohystochemistry into 40 μm coronal sections with a freezing sledge microtome (SM 2000R, Leica) at −20°C. None of the animals assigned to immunohistochemical and biochemical experiments were submitted to behavioral testing.

To estimate the levels of hippocampal SDS-soluble Aβ peptides, 100 mg of hippocampus were homogenized with a manual Teflon glass homogenizer in 1 ml of RIPA buffer (150 mM NaCl, 0.5% sodium deoxycolate, 1% Triton X-100, 2% SDS) containing proteases inhibitors (5 mM EDTA, 5 mM EGTA, 5 μg/ml leupeptin, 10 μg/ml aprotinin, 1 μg/ml pepstatin) and phosphatases inhibitors (50 mM sodium fluoride and 5 mM sodium orthovanadate) and centrifuged at 20,000 × g at 4°C for 45 min using a Eppendorf microcentrifuge. Supernatants were aliquoted and stored at −80°C. Individual samples were analyzed to assess SDS-soluble human Aβ40 or Aβ42 levels by using sandwich ELISA tests commercially available (Invitrogen). Results were expressed as pg of Aβ/mg of hippocampal homogenate proteins as determined with a bicinchoninic acid assay (Pierce, Rockford, IL) and were corrected by the signal of WT protein background. Determinations were performed in duplicate samples in three independent experiments. To evaluate SDS-resistant oligomeric Aβ levels, samples (250 μg/lane) were run in SDS-Tricine 12% gels and processed as previously described (Leon et al., 2010). Immunoreactivity was detected by Western blot using an enhanced chemiluminescence detection system (ECL, Thermo Scientific) and membranes analyzed and quantitated with a STORM 840 Phosphor Imager (GE Healthcare). Group values were obtained simultaneously and normalized with respect to APP immunoreactivity.

Brain sections were analyzed by using systematic random sampling. Every sixth brain section (240 μm apart) was stained using a free-floating immunohistochemistry procedure as previously described (Leon et al., 2010). iAβ label was detected with McSA1 (1:4000) mouse monoclonal antibody and sections were analyzed at 40× magnification by using an Olympus BX50 microscope. Immunoreactivity was assessed with Image J software (NIH, USA) in sections between the coordinates −2.30 and −3.8 mm from bregma for dorsal hippocampus and between −4.16 and −6.04 mm from bregma for ventral hippocampus (Paxinos and Watson, 2009) following a standard protocol. Stereological estimation of the hippocampal volume affected by iAβ is described below in Image analysis Section.

One week before starting with the behavioral battery each cohort was transported to ININCA (School of Medicine, Universidad de Buenos Aires) where a behavioral core facility is available. Throughout the study period, animals were housed in a controlled environment (21 ± 2°C, 50–60% relative humidity, 12 h light/12 h dark schedule) and all efforts were made to minimize discomfort. After acclimation, rats were handled daily (5 min/day) for three consecutive days and subjected to a basic neurological examination which consisted of a battery of neurological reflex tests: righting response after being placed on the dorsal side; eye blink (response to light touch with a small camel hair brush), ear twitch, and limb withdrawal in response to tactile stimuli (light touch with a gloved finger); orienting response to a visual stimuli (flashlight) and startle response following an auditory stimulus (a metal clicker) as previously reported (Cohen et al., 2013). Responses were scored as being present or absent. The maximum time period in which behavioral tests were carried out was between 8:00 a.m. and 5 p.m. Since there is a major circadian influence on EPM and OF performance (Verma et al., 2010), these tests were conducted between 8:00 a.m. and 12:00 p.m. Moreover, testing order of the groups was counterbalanced to avoid the confounding effect of time of the day at which animals were tested. The order of tests were as follows: Elevated Plus Maze (EPM) test, Open Field (OF) test, Novel Object Recognition Test (NORT), Y-maze spontaneous alternation (Y-Maze) test, and Morris Water Maze (MWM) test. The interval between tests was at least 24 h. After each animal was tested, the entire apparatus and objects in the NORT were cleaned with 70% ethanol to prevent a bias due to olfactory cues. All testing sessions were digitally recorded (JVC Everio GZ-HD620) and analyzed using a computerized video-tracking system (Ethovision XT, version 7, Noldus Information Technology, Wageningen, The Netherlands) or the ethological observation software JWatcher V1.0. All the experimenters were blinded to the genotype of the animals.

For stereological estimations the contours of ventral and dorsal hippocampus showing immunoreactivity for iAβ was drawn in each section with Image J software (NIH, USA) tool to determine the affected area. The Cavalieri's principle (Gundersen and Jensen, 1987) was used to calculate the affected volume by using the following formula: V = ∑AxNxE, where V = volume (mm³); ∑A = sum of area counted (mm²); N = number of sections analyzed; E = thickness of each section (mm).

All data are shown as the mean ± s.e.m. Data were analyzed by Two-Way ANOVA or by Two-Way mixed ANOVA tests followed by Tukey's HSD post-hoc tests for multiple comparisons, unless noted otherwise. The significance level was set at 5%. SPSS 15.0 for Windows software (Chicago, IL, USA) was used to perform all statistical analyses.

In order to better characterize Aβ-mediated neuropathology at early age, we quantified the levels of SDS-soluble and SDS-resistant peptide Aβ immunoreactivity in hippocampal homogenates from 3, 6, and 12 months old animals by a commercially available ELISA test and Western blotting, respectively. By ELISA, we detected the presence of SDS-soluble human Aβ40 in Tg^+/− hippocampal homogenates early at the age of 3 months (9.16 ± 1.42 pg/mg). The levels of SDS-soluble human Aβ40, however, did not differ significantly between Tg^+/− rats at 6 months (9.84 ± 1.30 pg/mg) or 12 months (8.00 ± 3.19 pg/mg) of age (F < 1), with an average of 9.00 ± 1.92 pg of Aβ40/mg of total hippocampal proteins. By contrast, we were not able to detect SDS-soluble human Aβ42 peptide in RIPA homogenates of Tg^+/− rats at any age analyzed. In addition, SDS-resistant Aβ oligomers appeared in hippocampal homogenates of 6 and 12 months old Tg^+/− rats, but not in 3 months old Tg^+/− rats (see Figure 1A). In accordance with previous reports, our data show that in Tg^+/− rats the neuropathology was restricted to iAβ (see Figure 1B, upper panel) and no significant differences were obtained in the volume fraction affected in dorsal or ventral hippocampus among different ages analyzed (F < 1 for both areas analyzed; see Figure 1B, lower panel). Our data strongly suggest a subtle phenotype in Tg^+/− rats with an “age-independent” accumulation for SDS-soluble monomeric Aβ40 and iAβ but an “age-dependent” accumulation for SDS-resistant Aβ oligomers.

Gross inspection of each rat prior to the behavioral experiments did not reveal any visible differences between WT and Tg^+/− animals. All animals appeared healthy and neurological reflexes were all scored as being present.

Two-Way ANOVA tests revealed significant main effects of age for “total distance moved” and “number of closed arm entries” [F_{(2, 51)} = 8.36, p = 0.001; F_{(2, 51)} = 9.29, p < 0.001, respectively]. By contrast, neither the main effects of genotype nor the age × genotype interactions were significant (F < 1 for all cases). These results suggest that locomotor activity changed with age, regardless of the genotype. Post-hoc analysis confirmed that both 6- and 12-month-old WT and Tg^+/− rats moved less distance and had a lower number of entries into closed arms than 3-month-old groups (p < 0.05 for all comparisons; see Figures 3A,B). Regarding “percentage of open arm entries” and “percentage of time spent in open arms,” neither the main effects of age nor the main effects of genotype nor the age × genotype interactions were significant (F < 1 for all cases; see Figures 3C,D), indicating that levels of anxiety were similar among all experimental groups.

When “total distance moved” and “number of rears” were analyzed by Two-Way ANOVA tests, the main effects of age were significant [F_{(2, 51)} = 13.98, p < 0.001; F_{(2, 51)} = 11.10, p < 0.001, respectively] but neither the main effects of genotype nor the age × genotype interactions were significant (F ≤ 1 for all cases). Post-hoc multiple comparisons confirmed that locomotor and rearing activity declined with age, since 6- and 12-month-old WT and Tg^+/− rats moved less distance and presented a lower rearing frequency than 3-month-old groups (p < 0.05 for all comparisons; see Figures 4A,B). Relative to anxiety levels, the Two-Way ANOVA tests for “number of entries into the center” and “time spent in central area” indicated that neither the main effects of age nor the age × genotype interactions were significant (F ≤ 1 for all cases). However, the main effects of genotype showed significant differences [F_{(2, 51)} = 8.46, p = 0.005; F_{(2, 51)} = 13.87, p < 0.001, respectively]. Post-hoc analyses revealed that 6- and 12-month-old Tg^+/− rats showed a lower number of entries and spent less time in the center of the OF than WT groups (p < 0.05 for all comparisons; see Figures 4C,D). In contrast, 3-month-old Tg^+/− rats did not differ from their WT littermates in none of the variables analyzed (see Figures 4C,D). Since the interactions age × genotype did not reach statistical significance, the lack of differences between 3 month-old WT and Tg^+/− groups should be taken with caution (Tybout et al., 2001).

The discrimination index (d1) and discrimination ratio (d2) were not statistically different between genotypes or ages [Two-Way ANOVA tests. d1. Genotype: F_{(1, 51)} = 2.25, p = n.s.; Age: F_{(2, 51)} < 1; Genotype × age: F_{(2, 51)} < 1. d2. Genotype: F_{(1, 51)} < 1; Age: F_{(2, 51)} = 1.25, p = n.s.; Genotype × age: F_{(2, 51)} < 1]. One sample t-tests indicated that d1 and d2 scores were significantly different from those expected by chance for all experimental groups (p < 0.05 for all cases; see Figures 5A,B). These results demonstrate that WT and Tg^+/− rats were able to recognize the novel object from the familiar one at all ages tested.

A significant main effect of age for “total number of arm entries” [F_{(2, 51)} = 17.52, p < 0.001] was revealed by a Two-Way ANOVA test. Neither the main effect of genotype nor the age × genotype interaction were significant (F < 1 for both cases). Regardless of the genotype, post-hoc analysis showed that 12-month-old WT and Tg^+/− rats displayed a lower number of arm entries than 3- and 6-month-old groups (p < 0.01 for all comparisons; see Figure 6A). When the “percentage of alternation” was also analyzed by a Two-Way ANOVA test, the main effects of age and genotype and the interaction were all significant [F_{(2, 51)} = 4.48, p = 0.016; F_{(2, 51)} = 7.81, p = 0.007; F_{(2, 51)} = 3.13, p = 0.05, respectively]. Post-hoc tests indicated that 6- and 12-month-old Tg^+/− rats showed a lower percentage of alternation than their WT littermates (p = 0.020 and p = 0.006, respectively; see Figure 6B), while 3-month-old groups displayed similar levels of alternation (see Figure 6B). These results indicate that Tg^+/− rats displayed a spontaneous alternation impairment at 6 and 12 months of age.

There are several reports on murine models of AD, aimed at understanding the impact of amyloid formation on learning and memory but few are focused on emotional behavior. In this study, we analyzed the effect of genetic modification of APP on EPM and OF tests in Tg^+/− rats in view of the tendency of AD patients to be either apathetic or agitated (Chung and Cummings, 2000; Senanarong et al., 2004). We were unable to detect differences in anxiety-related behaviors between Tg^+/− and WT rats in the EPM test in the present study. Unchanged anxiety levels assessed by EPM were described in transgenic mice lacking amyloid plaques (Moran et al., 1995; Lalonde et al., 2002; Lee et al., 2004; Boon et al., 2010) and in two models with Aβ plaques (Arendash et al., 2001; Le Cudennec et al., 2008; Blanchard et al., 2009). It is important to note that behavior in EPM was not assessed in any of the transgenic rat models of AD described so far (Do Carmo and Cuello, 2013) and that our results in Tg^+/− rats, lacking extracellular Aβ deposition even at late stages, are in agreement with the concept that changes of EPM behavior appear only in transgenic models of AD with an extensive neuropathological phenotype (Savonenko et al., 2003; Lee et al., 2006; Lalonde et al., 2012). It is important to note that the levels of exploration of open arms seem to be low as compared to other reports (see Figures 3C,D). This effect could be attributable to the levels of illumination used (80–90 lux in the open arms). To rule out the possibility of a floor effect that could be masking an anxiety phenotype in Tg^+/− rats in the EPM, future studies are needed to investigate the behavior of these animals in this test under different experimental conditions.

Only one transgenic rat model of AD (the Tg6590) has been previously characterized for its performance in the OF. This transgenic rat model showed increased Aβ precursor protein (APP), cerebrovascular deposits, few diffuse plaques and increased phosphorylated tau at 15 months of age. At 9 months old these animals showed no differences either in the number of entries into center or in the time spent in the central area of the OF, as compared to control group (Kloskowska et al., 2010). By contrast, here we show that Tg^+/− rats are more anxious in the OF test at 6 months of age. This behavior was also observed at 12 months of age, suggesting the robustness of this phenotype. It is worth noting that this anxious phenotype was observed in the OF test but not in the EPM test. Previously, it was reported that the OF and the EPM operationalize different aspects of emotional behavioral (Trullas and Skolnick, 1993; Ramos et al., 1998). This lack of inter-test correlations has been attributed to the multidimensional nature of emotional behavior (Ramos, 2008). Therefore, it could be possible that the higher anxiety levels detected in Tg^+/− rats in the present study, reflect an specific dimension of emotional behavior linked to OF activity. Finally, it is important to note that both tests (EPM and OF) were sensitive to the age-related decline in spontaneous locomotor activity and exploration (see Figures 3A,B, 4A,B).

In summary, even if the emergency (Holmes, 2001) or the light/dark box (Pinton et al., 2011) tests were not assessed in this study, our report is the first analysis of the anxiety disturbances in a transgenic rat model of early AD that does not develop Aβ plaques. It is of interest that the anxiety pattern described here for the Tg^+/− rats appear to closely mimic neuropsychiatric symptoms described in some pre-clinical AD patients.

Ideally, genetically modified animal models of AD not only have to replicate some of the hallmark neuropathology, such as plaque-like amyloid accumulations and tau deposition, but also should reproduce some of the cognitive impairments relevant to the disease. In this context, near all of the transgenic mice models with excessive Aβ either via transgenic overexpression of mutant hAPP and/or its processing enzymes, typically present static cognitive deficits that are evident early in life and progressive deficits that are age-related and associated with increased concentrations of deposited or unbound assemblies of Aβ (Kobayashi and Chen, 2005). By contrast to transgenic mice in which a wide battery of cognitive tests were performed, a narrower battery of tests were used in transgenic rat models of AD. These included MWM, Fear Conditioning, and Novel Object Recognition and Location (NORL) test (Leon et al., 2010; Iulita et al., 2014). Early description of the cognitive performance of hemizygous transgenic McGill-R-Thy1-APP rat in the MWM showed that at 3 months of age (n = 4) the learning curve was undistinguishable from that of WT (n = 8) rats while at 13 months of age, Tg^+/− rats (n = 4) exhibited a mild spatial learning deficit reflected by a significantly increased escape latency, compared to WT (n = 12), during the third day of the learning phase. Moreover, spatial reference memory, assessed during the probe trial, was not impaired at both ages, although a trend could be seen in the 3 month-old group (Leon et al., 2010). To further characterize the cognitive abilities of Tg^+/− rats, in this study we analyzed the performance of these animals in three different tasks including MWM, Novel Object Recognition (NOR) and Y-maze tests at 3, 6 and 12 months old. Using a larger number of Tg^+/− animals, we were able to detect, in the MWM, a spatial reference memory deficit early at 3 months of age, which persisted at 6 and 12 months. This deficit cannot be explained by a slower learning curve, as in contrast to the original report (Leon et al., 2010), we did not find in Tg^+/− rats any impairment in the learning phase of the MWM test. These discrepancies can be explained both by the number of animals used and by the gender of the animals utilized (males and females in the first report while males only in our study).

Memory tasks sensitive to hippocampal functions such as contextual fear conditioning (FC), NORL and Y-maze alternation have been described in different transgenic mice models of AD (Corcoran et al., 2002; Dineley et al., 2002; Ohno et al., 2004; Devi and Ohno, 2012; Mitani et al., 2013; Nagahara et al., 2013). However, none of these tests were assessed in transgenic rats until a recent report in which hemizygous transgenic McGill-R-Thy1-APP (Tg^+/−) rats showed reduced auditory Fear Conditioning (FC) and reduced contextual and cued FC recall (Iulita et al., 2014). In the same study, Tg^+/− rats displayed an impairment in the Novel Object Recognition and Location (NORL) task (Iulita et al., 2014). The impairment described in the NORL test was not reproduced by Tg^+/− in the NOR task performed in our study. This could be due to the differential stringency of the tasks as five objects were used in the previous report (Iulita et al., 2014). Although both tests are hippocampal-dependent, the NORL is a more complex task that includes the spatial component that is absent in the classic NORT test, and may explain the discrepancy between results. While in the former test one unfamiliar object had to be discriminated among a total of five objects during the novel object recognition phase of the task, we employed a more classical approach with two dissimilar objects (one unfamiliar and one familiar) that had to be discriminated during the retention trial. Discrepancies in results reveal that when the complexity of the task is increased, recognition memory may be also impaired in this transgenic rat model.

Spontaneous alternation in the Y-maze test estimates the willingness to explore novel stimuli, or to avoid familiar stimuli, and is an appropriate paradigm to asses working memory and/or behavioral disinhibition, signs also manifest in AD. Also, the total number of arm entries in the Y-maze is a parameter used to assess sensoriomotor capacities. In this study, we did not find significant differences in the total number of arm entries between Tg^+/− and WT rats. However, a significant decline was detected in the percentage of alternation in Tg^+/− rats at 6 and 12 months old. We could not ascribe this reduced rate in spontaneous alternation to behavioral disinhibition since performance of Tg^+/− rats in locomotor and exploratory parameters in the EPM and OF did not differ from WT. Therefore, it is likely that reduced spontaneous alternation reflects spatial working memory impairment. Poor performance in Y-maze appears in young mice of several single transgenic AD models at 2–4 months, when there are no plaque-like deposits, but these mice have had high levels of soluble Aβ in their nervous systems from early postnatal life (Lalonde et al., 2012). These observations are in agreements with our results, since Tg^+/− rats lack parenchymal Aβ plaques but thus accumulate soluble oligomers intraneuronally.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.