Corresponding author: J. Raber, PhD, Oregon Health and Science University, 3181 SW Sam Jackson Pkwy, Mail Code L470, Portland, OR 97239, USA. E-mail: email@example.com
The three major human apoE isoforms (apoE2, apoE3 and apoE4) are encoded by distinct alleles (ϵ2, ϵ3 and ϵ4). Compared with ϵ3, ϵ4 is associated with increased risk to develop Alzheimer's disease (AD), cognitive impairments in Parkinson's disease (PD), and other conditions. In contrast, a recent study indicated an increased susceptibility to the recurring and re-experiencing symptom cluster of Post-Traumatic Stress Disorder (PTSD), as well as related memory impairments, in patients carrying at least one ϵ2 allele. Contextual fear conditioning and extinction are used in human and animal models to study this symptom cluster. In this study, acquisition (day 1, training), consolidation (day 2, first day of re-exposure) and extinction (days 2–5) of conditioned contextual fear in human apoE2, apoE3 and apoE4 targeted replacement and C57BL/6J wild-type (WT) mice was investigated. Male and female apoE2 showed acquisition and retrieval of conditioned fear, but failed to exhibit extinction. In contrast, WT, apoE3 and apoE4 mice showed extinction. While apoE2 mice exhibited lower freezing in response to the context on day 2 than apoE3 and apoE4 mice, this cannot explain their extinction deficit as WT mice exhibited similar freezing levels as apoE2 mice on day 2 but still exhibited extinction. Elevating freezing through extended training preserved extinction in controls, but failed to ameliorate extinction deficits in apoE2 animals. These data along with clinical data showing an association of apoE2 with susceptibility to specific symptom clusters in PTSD supports an important role for apoE isoform in the extinction of conditioned fear.
Post-traumatic stress disorder (PTSD) is one of the most common (Kessler et al. 2007) and debilitating anxiety disorders (Kessler 2000). It is frequently comorbid with other mental disorders (Koenen et al. 2008) and has been identified as a major unique identifier for suicidality (Cougle et al. 2009; Sareen et al. 2005). The prevalence of PTSD suggests the involvement of environmental risk factors (Breslau et al. 1998; Kessler et al. 1995) including socioeconomic status and education (Breslau et al. 1995), and associations between intensity and number of traumatic events (Sledjeski et al. 2008). In addition, twin studies have provided support for heritable and genetic risk factors increasing PTSD susceptibility (Chantarujikapong et al. 2001; Skre et al. 1993; Stein et al. 2002).
ApoE plays an important role in the metabolism and redistribution of lipoproteins and cholesterol. The three major human apoE isoforms are encoded by distinct alleles (ϵ2, ϵ3 and ϵ4). They differ in having cysteine (Cys) or arginine (Arg) at positions 112 and 158, and differ in metabolic properties (Mahley 1988). In the brain, apoE has been implicated in development, regeneration, neurite outgrowth and neuroprotection (Mahley 1988). Compared with ϵ2 and ϵ3, ϵ4 increases the risk of developing Alzheimer's disease (AD), cognitive impairments following neurotrauma (Nathoo 2003), those occurring with normal aging (Howieson et al. 2003) or in the context of Parkinson's disease (PD; Tang et al. 2002; Zareparsi et al. 2002). While ϵ2 is generally considered neuroprotective, compared with ϵ4, ϵ2 may increase the risk of developing sporadic PD (Huang et al. 2004). ApoE2 might also be a critical determinant of susceptibility to specific symptom clusters within the PTSD diagnosis. In a study of veterans with combat-related PTSD, compared with apoE3 and apoE4, apoE2 was associated with worse re-experiencing symptoms and memory impairments, but was not associated with a risk to develop PTSD itself (Freeman et al. 2005).
The phenomenon of recurring and persistent recall of traumatic memories in PTSD may be related to a failure of extinction learning (Charney et al. 1993) or a failure to modify or acquire new associations to contextual stimuli (Corcoran et al. 2005). Therefore, contextual fear conditioning has been widely used to study aspects of PTSD in humans and animal models (Siegmund & Wotjak 2006). In particular, fear conditioning is used to study the inhibition or extinction of learned fear, a process believed to underlie the recurring and re-experiencing symptoms of PTSD (Cannistraro and Rauch 2003). This study investigated the extinction of conditioned contextual fear in human apoE2, apoE3 and apoE4 targeted replacement (TR) and C57BL/6J wild-type (WT) mice. In a separate cohort of animals, open field and light–dark tests were used to assess anxiety-like behavior.
Materials and methods
Three to six-month-old Targeted Replacement (TR) apoE2 (Sullivan et al. 1998), apoE3 (Sullivan et al. 1997), and apoE4 (Knouff et al. 1999) mice, expressing human apoE under control of the mouse apoE promoter on a C57BL/6J background, and C57BL/6J WT mice (n = 219 mice) were used in this study. The breeder mice to generate 4–6 month-old TR mice for this study were generously provided by Dr. Patrick Sullivan, who maintains his colony by heterozygous and homozygous matings. The human TR mice we received from Dr Sullivan were backcrossed to C57BL/6J mice at least eight times and therefore are at least 99.6% C57BL/6J. Breeder mice to generate the WT mice were obtained from Jackson Laboratories, Bar Harbor, ME. Homozygous breeding of the mice was used to generate the experimental mice for this study. Throughout testing, all the mice were singly housed. This study was performed over a 2-year period and involved 19 apoE2, 16 apoE3, 18 apoE4 and 15 WT litters. Animals were maintained on a 1200 h light/dark schedule (lights on at 06:00). Laboratory chow (PicoLab Rodent diet 20, # 5,053; PMI Nutrition International, St. Louis, MO, USA) and water were provided ad libitum. Behavioral testing took place during the light cycle. All procedures complied with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and with IACUC approval at Oregon Health & Sciences University. Experimenters were blinded to the genotype and sex of the animals.
Freezing in mice, defined as a lack of movement except for respiration, is commonly used to assess conditioned fear (Maren 2001). The fear conditioning apparatus consisted of a Plexiglas chamber (21 cm W × 21 cm D × 24 cm H) containing a shock grid (21 × 21 cm, 30 bars 0.7 cm apart) inside a sound-attenuating chamber (55.9 cm W × 35.6 cm D × 38.1 cm H, Med Associates ENV-022V Standard Expanded PVC, St. Albans, VT, USA). Cameras and Noldus Ethovision XT 7.0 video tracking software (Noldus Information Technology, Leesburg, VA, USA) were used to automate behavioral scoring. This software has been used to measure freezing (Bast et al. 2001) and selected tracking parameters optimally correlate with traditional hand-scoring (Pham et al. 2009). Fidelity to hand-scoring is preserved under these experimental conditions (as described in Villasana et al. 2010). Lights (9.33 ± 0.88 lux) and administration of shock (scrambled) were under control of a Hamilton Kinder programmable shocker using Kinder Scientific MotorMonitor software (Kinder Scientific, Poway, CA, USA). External noise cues were further attenuated by a white noise generator running at 50 dB.
This study consisted of two contextual fear conditioning and extinction paradigms containing one day of training and four subsequent days of extinction, assessed by re-exposure to the context. In experiment 1, the mice (WT: n = 29 males and n = 14 females; apoE2: n = 11 males and n = 19 females; apoE3: n = 10 males and n = 9 females; and apoE4: n = 14 males and n = 13 females) were placed into the dark fear conditioning enclosure on the training day (day 1). Lights came on at 0:00 and went off at 5:00. 148 seconds of exposure to the context was immediately followed by a 2-seconds 0.5 mA shock. After another 148 seconds, a second 2-seconds 0.5 mA shock was delivered. Animals received 5-min extinction trials on days 2–5. Enclosures, chambers and shock grids were cleaned with 5% glacial acetic acid between mice.
Because of relatively lower freezing levels in apoE2 and WT mice, in experiment 2, 4–7 month-old male mice (WT: n = 12; apoE2: n = 6; apoE3: n = 6; and apoE4: n = 16) received four shocks (0.5 mA, 2-seconds shocks at 2:28, 4:58, 7:28 and 9:58) during a 10-min trial on the training day and 10-min extinction trials on days 2–5. Female mice were not included as there was no sex differences observed during the extinction phase of experiment 1.
Baseline freezing, prior to the first shock in the training trial on day 1 was analyzed and compared with post-training freezing (acquisition of fear response) and freezing 24 h after training (consolidation and recall). Extinction of conditioned fear was measured as a decrease in freezing over time compared with the first contextual re-exposure 24 h following training.
Baseline motion and shock sensitivity
To assess differences in shock sensitivity, motion during shock, recorded using a Med Associates (Med Associates) fear conditioning system (see Anagnostaras et al. 2010 for details) was analyzed in 60 3–4 month-old male mice (WT = 20, apoE2 = 14, apoE3 = 19 and apoE4 = 21) over ten 0.35 mA shocks, separated by 60-seconds. A lower amperage was chosen in case threshold differences would not be apparent at 0.5 mA. Med Associates reports motion using a proprietary index based on pixel change. Baseline motion before shock administration was also assessed.
To assess potential contributions of anxiety-like behavior, six-month-old male apoE2 (N = 10), apoE3 (N = 8) and apoE4 (N = 9) mice were tested in both open field and light–dark mazes, as described in Villasana et al. (2006).
Data are reported as averages ± the standard error of the mean and were analyzed using SPSS 16.0 software (IBM, Armonk, NY, USA). Two-way analysis of variances (anovas) were conducted using sex and genotype as between-subject variables. Extinction curves were analyzed using repeated-measures anovas with testing day as a within subjects variable and genotype and sex as a between-subject variables. Non-normal data were log-transformed if possible; else they were analyzed using non-parametric methods. If Mauchly's test of sphericity (for repeated-measures) was not satisfied, multivariate statistics using Wilk's lambda (λ) were reported. Bonferroni adjustments were used for post-hoc multiple comparisons. Results were considered significant at an α level of 0.05. Video failure resulted in the loss of day 1 data for four apoE4 mice (two male and two female mice). These data were removed from the baseline and acquisition freezing level analysis. A software failure on day three resulted in the loss of data of 13 WT mice (five female and eight male mice) from this data point. The freezing levels of these mice on day 2 were carried over into day 3.
There was an effect of genotype on baseline freezing levels (F3,99 = 3.492, P = 0.018; Fig. 1a). ApoE3 mice showed higher freezing levels (9.52 ± 0.86) than WT (6.54 ± 0.70) (P = 0.037) or apoE2 (6.43 ± 0.60) (P = 0.038) mice. The freezing levels of apoE4 mice (7.88 ± 0.88) did not differ from those in any other group.
All groups acquire contextual fear
Acquisition of contextual conditioned fear was analyzed comparing the percentage of time spent freezing during baseline (0:00–2:28) and the time after the first shock (2:30–4:58). There was a main effect of training (F1,94 = 141.37, P < 0.0001), and a significant interaction between training and genotype (F3,94 = 3.900, P = 0.011), although all groups exhibited increased freezing (Table 1; WT: F1,29 = 33.288, P < 0.0001; apoE2: F1,29 = 41.674, P < 0.0001; apoE3: F1,18 = 14.064, P = 0.001; and apoE4: F1,22 = 127.493, P < 0.0001), suggesting intact acquisition of conditioned fear.
Table 1. Acquisition and extinction of conditioned fear in experiment 1
All genotypes exhibited acquisition of fear, as showed by a significant increase in freezing following the first training shock.
ApoE3, ApoE4 and WT mice exhibit extinction (significant decrease in freezing relative to day 2).
*P < 0.05; ***P < 0.005; ****P < 0.0001; †P = 0.059. ApoE2 mice do not exhibit extinction.
249.45 ± 28.23
30.09 ± 2.03
29.38 ± 1.84
33.07 ± 2.26
32.47 ± 1.89
277.86 ± 49.00
60.00 ± 3.87
50.30 ± 3.44*
54.59 ± 3.96
46.73 ± 3.30*
395.10 ± 67.14
58.94 ± 3.06
56.32 ± 2.97†
47.64 ± 2.16****
46.27 ± 1.86****
344.08 ± 59.91
44.54 ± 2.6
40.47 ± 2.54
38.77 ± 2.32
37.26 ± 2.19***
Compared with apoE3 and apoE4, apoE2 and WT mice exhibit reduced contextual fear 24 h after training
Consolidation and recall of conditioned contextual fear were measured as the percent of time freezing during the 300 seconds trial on day 2 (first day of re-exposure; Fig. 1b). There was an effect of genotype (F3,111 = 23.171, P < 0.0001), as well as a significant interaction of sex and genotype (F3,111 = 3.985, P = 0.010). Therefore, the sexes were then analyzed separately. There was an effect of genotype in males (F3,60 = 13.008, P < 0.0001), with apoE2 mice freezing less than WT, apoE3 and apoE4 mice (P < 0.0001). There was also an effect of genotype in females (F3,51 = 11.158, P < 0.0001), with female apoE4 and apoE3 mice exhibiting greater freezing than WT (WT vs. apoE3: P = 0.027; WT vs. apoE4: P < 0.001) and apoE2 mice (apoE2 vs. apoE3: P < 0.003; apoE2 vs. apoE4: P = 0.0001). Overall (Fig 1b), apoE2 mice froze less than WT, apoE3 and apoE4 mice (P < 0.0001), and WT mice froze less than apoE3 (P = 0.011) and apoE4 mice (P = 0.006).
ApoE2 mice exhibit impaired extinction compared with controls
Freezing during extinction trials (Fig 1c–f, Table 1) showed an effect of day (λ = 0.751, F3,109 = 12.034, P < 0.0001), and a significant interaction of day and genotype (λ = 0.749, F9,333 = 3.581, P < 0.0001). There was no significant interaction with sex and day or main effect of sex in extinction. Next, each genotype was analyzed separately. There was an effect of day in WT (λ = 0.689, F3,39 = 5.879, P = 0.002; with lower freezing on day 5 compared with day 2 (P = 0.004)), apoE3 (λ = 0.541, F3,15 = 4.240, P = 0.023; with lower freezing on day 5 and 3 compared with day 2 (P = 0.015 and P = 0.022, respectively), and apoE4 mice (λ = 0.454, F3,23 = 9.217, P < 0.0001; with lower freezing on day 5 vs. day 2 (P < 0.0001) and day 4 vs. day 2 (P < 0.0001), with a trend toward a significant decrease on day 3 compared with day 2 (P = 0.059)). There was no effect of day, and thus no decrease in freezing, within apoE2 mice.
ApoE2 mice exhibit impaired extinction compared with controls
Increasing the number of shocks during training elevates the fear response to the context 24 h after training
Doubling the length of the trial and the number of shocks during training on day 1 elevated freezing levels on day 2 above those seen in the first experiment in all genotypes (Mann–Whitney U-test, WT: Md1 = 25.72, Md2 = 36.17, U = 160, z = −1.997, P = 0.046; apoE2: Md1 = 16.13, Md2 = 30.33, U = 19, z = −3.014, P = 0.003; apoE3: Md1 = 10.58, Md2 = 20.67, U = 11, z = −2.927, P = 0.003; and apoE4; Md1 = 17.07, Md2 = 30.31, U = 160, z = −3.342, P = 0.001; Fig. 1b). There was an effect of genotype on freezing levels on day 2 (F3,36 = 15.221, P < 0.0001), similar to that seen in experiment 1; apoE3 and apoE4 mice exhibited greater freezing than WT or apoE2 mice (P < 0.0001).
Elevating freezing does not ameliorate extinction deficits in apoE2 mice
Freezing during extinction trials exhibited an effect of day (F3,108 = 25.225, P < 0.0001) and an interaction between day and genotype (F9,108 = 2.589, P = 0.010; Table 2). When each genotype was analyzed separately, there was an effect of day in WT (F3,33 = 7.511, P = 0.001; day 5 vs. day 2, (P = 0.040); day 4 vs. day 3 (P = 0.046)), apoE3 (F3,15 = 12.992, P < 0.0001; day 5 vs. day 2, P = 0.038; day 5 vs. day 3: P = 0.005), and apoE4 mice (F3,45 = 26.503, P < 0.0001; day 4 (P < 0.000) and day 5 (P = 0.003) vs. day 2; and a trend toward a difference between day 3 and day 2 (P = 0.059)). Changes in freezing in apoE2 mice did not reach significance.
Table 2. Extinction of conditioned fear in experiment 2
[Correction added on 14 September 2012, after first online publication: ApoE2 values in Day 3 has been changed from 53.61 ± 438 to 53.61 ± 4.38].
ApoE3, ApoE4 and WT mice exhibit extinction (significant decrease in freezing relative to day 2).
*P < 0.05; **P < 0.01. ApoE2 mice do not exhibit extinction.
52.06 ± 5.01
53.61 ± 4.38
48.71 ± 5.58
45.39 ± 3.25
82.00 ± 3.62
72.68 ± 6.14
63.75 ± 7.61
60.17 ± 6.96*
74.59 ± 2.01
66.35 ± 2.41
60.53 ± 1.66**
63.69 ± 1.83**
55.82 ± 3.80
56.61 ± 3.32
50.04 ± 4.18
45.83 ± 4.43*
Freezing differences are not due to shock sensitivity
ApoE2 mice show greater baseline motion (492.59 ± 21.85) than all other groups (effect of genotype: F3,70 = 20.726, P < 0.0001); WT: 350.89 ± 12.00, apoE3: 346.17 ± 12.50, apoE4: 309.27 ± 18.72; P < 0.0001 for apoE2 vs. any other genotype; Fig. 2a). With regards to shock sensitivity, there was only a trend of an effect of genotype on average motion during shock (F3,70 = 2.381, P = 0.077). This was driven by a trend for WT mice (1172.70 ± 58.02) to display less motion (P = 0.082) than E3 (1383.037 ± 59.52) mice. No differences were found between apoE4 (1225.66 ± 76.42) and E2 (1219.98 ± 69.34) mice and any other groups (Fig 2a).
ApoE2 mice do not exhibit elevated anxiety-like behavior compared with apoE3 or apoE4 mice
There was an effect of genotype on time spent in the open (anxiety-inducing) portion of the open field arena (F2,24 = 4.440, P = 0.023). ApoE4 mice spent less time in the center of the arena compared with apoE2 counterparts (P = 0.041), with a trend toward greater anxiety-like behavior relative to apoE3 mice (P = 0.061; Fig. 2b). An effect of genotype on distance moved was also noted (F2,24 = 8.266, P = 0.002, with apoE2 mice exhibiting higher activity than apoE4 mice (P = 0.001; Fig. 2c). There were no effects of genotype on time spent in the light (anxiety-inducing) portion of the light–dark maze (Fig. 2d), nor on distance moved (Fig. 2c).
In this study, male and female ApoE2 mice showed acquisition of conditioned fear but failed to exhibit extinction. ApoE2 mice also exhibited lower freezing in response to re-exposure to the context on day 2 than apoE3 or apoE4 mice. This cannot explain their extinction deficit, as WT mice exhibited similar freezing levels as apoE2 mice on day 2 but still exhibited extinction. Increased anxiety has been associated with greater activation of the amygdala during extinction, and a deficit thereof. As apoE2 mice did not show higher anxiety levels than apoE3 or apoE4 mice in the open field or light–dark tests, enhanced anxiety levels cannot explain the extinction deficit in apoE2 mice. In contrast to apoE2 mice, WT, apoE3 and apoE4 mice, showed extinction. This deficit seems specific to fear conditioning and extinction learning, as apoE2 mice show comparable spatial learning and memory in the water maze as compared with apoE3 mice (Villasana et al. 2006; Siegel et al. 2010). These data, along with clinical data showing an association of apoE2 with susceptibility to specific symptom clusters in PTSD (Freeman et al. 2005), support an important role of ApoE isoform in the extinction of conditioned fear.
ApoE2 mice show greater baseline motion than all other groups. It is unlikely that greater activity would significantly affect freezing levels if memory and fear response were intact, as freezing is a specific behavior rather than simply reduced activity levels. This is also indicated by the distinct pattern of genotype differences in baseline motion as compared with that of baseline freezing. Nevertheless, elevating freezing levels on day 2 failed to produce extinction in the same time-frame as was observed in the other genotypes. Increased activity would not affect a decrease in freezing over time, and one might even expect to see accelerated reduction in freezing over time if greater activity was present. Furthermore, reduced fear on day 1 cannot explain the deficit in extinction, as elevating freezing levels in the second paradigm did not ameliorate this apparent deficit.
ApoE4 mice showed higher levels of contextual freezing than other genotypes. Consistent with this result, apoE4 mice require fewer trials to learn the passive avoidance task compared with apoE2 and E3 mice (Villasana et al.2006, 2008). ApoE2 mice did not show elevated anxiety levels as compared with apoE3 or apoE4 mice in the open field or light–dark tests. Given the lack of differences in fear conditioning between apoE4 mice and other genotypes (E2 notwithstanding), differences in emotional learning and memory and anxiety may be unrelated to the extinction deficits of apoE2 mice in this study.
Contrary to our results, Kornecook et al. (2010) found reduced freezing levels in apoE4 compared with apoE3 mice in animals also over-expressing human APP. Aβ accumulation in these mice is apoE isoform-dependent (apoE4 > apoE3; Holtzman et al., 2000) and differential interactions between hAPP and apoE, rather than apoE isoform might affect contextual fear conditioning in this model.
In the apoE TR model in which apoE is expressed under control of the mouse apoE promoter, apoE4 mice do not show reduced cognitive performance under baseline conditions but enhanced susceptibility to cognitive impairments following environmental challenges such as cranial irradiation (Villasana et al. 2006). In contrast, in mice expressing apoE in either neurons or astrocytes, apoE4 mice do show cognitive impairments compared with to apoE3 mice (Hartman et al. 2001; Raber et al. 1998; van Meer et al. 2003). Part of these convergent data might be due to the fact that apoE4 TR mice outperform GFAP-apoE4 mice in spatial learning and memory in the water maze (Siegel et al. 2010).
Contextual fear conditioning is frequently used to assess hippocampus-dependent memory (Anagnostaras et al. 2001), particularly in the context of acquisition and consolidation. Extinction processes similar to the initial learning have been proposed, involving acquisition, retrieval and consolidation, to mediate the formation of new, or modulation of original, fear memories (Ji & Maren 2007). Hippocampal long-term-potentiation (LTP) has been implicated in the acquisition of conditioned fear (Maren et al. 1994). Attenuation of LTP in the Prefrontal Cortex-Hippocampus pathway has been shown to impair extinction in conditioned fear (Farinelli et al. 2006; Herry & Garcia 2002). Hippocampal LTP is reduced in apoE2 mice compared with ApoE4 mice (Trommer et al. 2004), and this might contribute to the extinction deficit of E2 mice. In addition to the hippocampus (Corocan et al. 2005), lesion studies have implicated the amygdala in the formation of new associations between stimuli (Knight et al. 2004) as well as in the expression of the fear response (Cheng et al. 2006) and a role of the prefrontal cortex in retrieval of the extinction memory (Quirk & Mueller 2008), and there is support for interdependence of these three brain regions in fear conditioning and extinction as well (Ji & Maren 2007).
Consistent with clinical data showing an association of apoE2 with susceptibility to specific symptom clusters in PTSD, these data support an important role for apoE isoform in the extinction of conditioned fear. Deficits in extinction in apoE2 mice provide an example of aberrant processing of fear memory that may represent extant differences prior to developing the condition. Future studies are warranted to determine the brain region(s) involved in the isoform-dependent effects of apoE on contextual fear conditioning and extinction.
This work was supported by NNJ05HE63G (NASA) and the OHSU Development Account of Dr. Jacob Raber. The assistance of research intern Karen Li is appreciated. There are no conflicts of interest to declare.