Fear is one of the most potent emotional experiences of our lifetime and is an adaptive component of response to potentially threatening stimuli, serving a function that is critical to the survival of higher vertebrates (Davis, 1992; LeDoux, 2000). Too much or inappropriate fear, however, accounts for many common psychiatric problems (Kent et al., 2003; Millan, 2003; Uys et al., 2003). A fearful experience can establish an emotional memory that results in permanent behavioral changes and emotional memories have been observed in many animal groups (Blanchard et al., 1993). The brain mechanisms underlying fear are similar in different species and the fear system will respond similarly in a person or a rodent, using a limited set of defense response strategies (LeDoux, 1996). The memory of learned fear can be assessed quantitatively using a Pavlovian fear-conditioning paradigm (Davis, 1992; LeDoux, 2000). During fear conditioning, an initially neutral conditioned stimulus (CS, e.g. an auditory tone) acquires biological significance by becoming associated with an aversive unconditioned stimulus (US, e.g. a footshock). After learning this association, an animal responds to the previously neutral CS with a set of defensive behavioral responses, such as freezing. Cumulative evidence suggests that the amygdala plays a central role in the acquisition, storage and expression of fear memory.

Here, an inducible striatal neuron ablation system in transgenic mice was developed. The G-protein γ7 subunit mRNA is expressed predominantly in medium spiny neurons of the caudate-putamen (CP) and nucleus accumbens (NAc) and neurons of the olfactory tubercle (Watson et al., 1994). To develop a striatal neuron-specific gene manipulation system, Gγ7-Cre and Gγ7-mCrePR mouse lines were produced by inserting the gene encoding Cre recombinase or Cre recombinase-progesterone receptor fusing protein (CrePR) into the translational initiation site of the G-protein γ7 subunit gene (Gng7) through homologous recombination in embryonic stem cells derived from the C57BL/6 strain (Mishina and Sakimura, 2007). Then, the Gγ7-Cre and Gγ7-mCrePR mice were crossed with the CAG-CAT-Z11 reporter mouse (Tsujita et al., 1999). Brain slices prepared from Gγ7-Cre × CAG-CAT-Z11 mice were stained for β-galactosidase activity to monitor the Cre recombinase activity. Strong β-galactosidase staining was found predominantly in the CP, NAc and olfactory tubercle. The Gγ7-mCrePR mouse was crossed with a knock-in mouse (Eno2-STOP-DTA) in which the Cre-inducible diphtheria toxin A gene (DTA) was introduced into the neuron-specific enolase gene (Eno2) locus (Kobayakawa et al., 2007). In Gng7(+/mCrePR) mice, one allele retains the intact Gng7 gene, and the other is inactivated by insertion of the CrePR gene. RU-486 was injected into the peritoneum of Gγ7-mCrePR × Eno2-STOP-DTA mice at postnatal day 42 (P42) to induce the recombinase activity of CrePR (Tsujita et al., 1999; Takeuchi et al., 2005; Mishina and Sakimura, 2007). Mock-injected mice served as controls. Thirteen days after RU-486 treatment, the ablation of striatal neurons was then quantitatively examined by immunohistochemical staining for NeuN, a marker protein for neurons. The density of NeuN-positive neurons in the CP drastically decreased by 13 day after RU-486 injection (F(6,54) = 99.5, P < 0.001, one-way ANOVA) and remained at a very low level thereafter. The number of NeuN-positive cells in the NAc core and shell also decreased with a similar time course. However, NeuN immunostaining signals in other brain regions including the amygdala were comparable between mock- and RU-486-treated mice. Along with the NeuN-immunohistochemistry, present results suggest that induction of CrePR-mediated DTA expression by RU-486 injection successfully ablated almost completely the medium spiny neurons that comprise approximately 90% of the NeuN-positive striatal neurons within 13 days. In subsequent analyses, I used Gγ7-mCrePR × Eno2-STOP-DTA mice from 13 to 22 days after RU-486 administration as striatal neuron-ablated mutant mice and corresponding mock-injected littermates served as controls.

Despite that approximately 90% of striatal neurons were ablated, the motor performance of the mutant mice appeared to be comparable to that of control mice in stationary thin rod (F(1,15) = 1.38, P = 0.26, repeated measures ANOVA) and rotating rod tests (F(1,14)= 3.57, P = 0.08, one-way ANOVA). Mutant mice were subjected to auditory fear conditioning to examine the possible involvement of striatal neurons in the formation of the emotional memory. Fourteen days after RU-486 treatment, mutant mice were trained for auditory fear conditioning. The ablation of striatal neurons hardly affected the auditory fear learning under the standard condition (control, 31.6 ± 5.1%; mutant, 28.0 ± 5.1%; F1,15 = 0.27, P = 0.61) in agreement with previous studies. When conditioned with a low-intensity unconditioned stimulus, however, the formation of long-term fear memory but not short-tem memory was impaired in striatal neuron-ablated mice (P < 0.05, mutant vs. control; P < 0.01, mutant vs. RU-486 control). Consistently, the ablation of striatal neurons 24 h after conditioning with the low-intensity unconditioned stimulus, when the long-term fear memory was formed, diminished the retention of the long-term memory (mock-injected mice, 37.6 ± 3.9%; RU-486-injected mice, 11.5 ± 2.6%; F(1,13) = 41.9, P < 0.001). These results reveal a novel form of the auditory fear memory depending on striatal neurons at the low-intensity unconditioned stimulus.

In addition, I report that post-conditioning infusions of NMDA receptor inhibitors into the striatum of wild-type animals disrupt consolidation of auditory fear memory (F(1,18) = 5.3, P = 0.034, repeated measures ANOVA) when mice were trained with a low-intensity unconditioned stimulus. Furthermore, intra-striatum infusions of protein synthesis blocker immediately (F(1,13) = 9.3, P = 0.009, repeated measures ANOVA) or 1 hour ( F(1,15) = 5.15, P = 0.038) after the conditioning prevent auditory fear memory whereas infusions 3 hours after conditioning do not affect the freezing level (F(1,12) = 0.22, P = 0.65), indicating that there is a critical time window of protein synthesis for memory consolidation. These findings demonstrate that NMDA receptors and de novo protein synthesis in the striatum are crucial for the consolidation of auditory fear memory with low-intensity shock.

本研究は、恐怖条件付け学習に線条体が関与するか否かを明らかにすることを目的に、分子遺伝学的手法を用いて線条体特異的かつ誘導的に神経細胞を除去可能なマウスを作製し、恐怖条件付け学習により記憶の解析を試みたものであり、下記の結果を得ている。

1. まず線条体特異的に神経細胞を誘導的に除去可能なマウスを作製した。線条体に選択的に発現するGタンパク質γ7サブユニットをコードする遺伝子のプロモーター下にCre組換え酵素とプロゲステロン受容体のリガンド結合領域を融合させたCrePR遺伝子を挿入し、Gγ7-mCrePRマウスを作製した。Gγ7-mCrePRマウスをCre依存的にジフテリア毒素Aサブユニット を発現するEno2-STOP-DTAマウスに掛け合わせ、6週齢マウスに抗プロゲステロン製剤RU-486を投与することによりCrePRの活性化を誘導した。

2. RU-486投与後10日目に細胞死のマーカーであるTUNEL陽性細胞が線条体特異的に誘導できた。さらに、神経細胞のマーカーである NeuNの免疫染色は、投与後13日目には線条体においてほぼ完全に消失し、NeuN免疫染色シグナルの消失は線条体特異的であった。線条体神経細胞の約90%を占める中型有棘神経細胞(medium spiny neuron)には、黒質網様部への直接経路および淡蒼球を解した間接経路が存在する。線条体神経細胞の除去を誘導したマウスにおいて、直接経路で発現するsubstance Pと間接経路で発現するenkephalinは黒質網様部と淡蒼球のそれぞれにおいて消失していたことから、線条体の中型有棘神経細胞はほぼ完全に消失していることが支持された。

3. RU-486投与後13日目以降のマウス(線条体特異的神経細胞除去マウス)は見た目の運動失調やジストニアの症状は示さず、thin rodにもバランス良く乗ることができた。一方で加速式のrotarodではより厳しい条件になるとパフォーマンスの向上は見られなかった。

4. 線条体特異的神経細胞除去マウスで恐怖条件付けを行った。刺激強度0.5 mAの電気ショックを音と対提示したところ、線条体神経細胞除去マウスは対照群同様に強いすくみ(freezing)反応を示した。しかしながら、刺激強度0.3 mAの条件下では、対照群のマウスは24時間後の音提示において強いfreezing反応を示したものの、線条体神経細胞除去マウスではfreezing反応は著しく低下していた。電気ショックに対する感受性をflinch とjumpを示す電流強度の閾値を指標に調べたところ、有意な変化は認められなかった。これらの結果から、線条体は刺激強度が弱い条件下における恐怖条件付け学習に重要な役割を担っていることが明らかとなった。

5. 線条体が記憶学習のどのプロセスに関与するかを調べるために、刺激強度0.3 mAの条件下で短期記憶のテストを行ったところ、対照群と線条体神経細胞除去マウスで有意な差は認められなかった。さらに同じ刺激強度の条件で、学習の24時間後にRU-486を投与し、その2週間後にテストを行ったところ、線条体神経細胞除去マウスでfreezing反応は有意に低下した。このことから、線条体は刺激強度が弱い条件下における恐怖条件付け学習の長期記憶の保持に関与していることが示唆された。

6. 次に、記憶の固定時に線条体においてNMDA受容体の活動が必要かどうかを調べた。刺激強度0.3 mAの条件下で、学習直後に線条体にNMDA受容体の阻害剤であるAPVを投与した群は人工脳脊髄液(ACSF)投与の対照群に比べ、24時間後のfreezing反応は有意に低下した。

7. 記憶の固定時に線条体において新規タンパク質合成が必要かどうかを調べた。刺激強度0.3 mAの条件下で、学習直後に線条体にタンパク質合成阻害剤であるアニソマイシン(ANI)を投与した群はACSF投与群に比べ、24時間後のfreezing反応は有意に低下した。また学習1時間後にANIを投与した群は翌日のfreezing反応は有意に障害されたのに対し、学習3時間後にANIを投与した群は高いfreezing反応を示し、ACSF群と比べ有意差は見られなかった。これらの結果より、記憶固定時に必要とされるタンパク質合成には時間枠があることが示唆された。

以上、本論文は線条体が刺激強度の弱い条件下における恐怖条件付け学習の長期記憶に関与すること、またその記憶の固定時に線条体においてNMDA受容体の活性化及び新規タンパク質合成が必要であることを明らかにした。本研究は刺激強度が弱い条件下における恐怖記憶に関わる新たな神経経路を解明し、恐怖情動記憶のメカニズム解明に重要な貢献をなすと考えられ、学位の授与に値するものと考えられる。