Adaptive changes of the ground squirrel retina during hibernation

冬眠的哺乳动物在极度低温下存活而不受伤，这是细胞保存的一项非凡壮举，对潜在的医学应用具有重要意义。我们着手研究冬眠的十三行地松鼠（TLGS）对神经损伤的反应是否与清醒的鼠不同。我们选择视神经压迫（ONC）损伤模型，因为它是典型的轴索损伤模型，没有血管损伤的并发症，而且是一种小手术，可以在冬眠的TLGSs中进行，不干扰冬眠状态。活性tlgs损伤后14天，大部分RGCs死亡。然而，对于处于静止状态的tlgs，大多数RGCs在损伤后存活。然后，我们将重点放在局部ONC损伤模型上，通过计算损伤后不同时间点鼻部（损伤侧）/颞部（未损伤侧）的RGC比率来量化存活率。对于活性tlgs， RGC数量从第3天开始下降，呈指数增长，直到第21天达到平台，约有20%的RGC仍然存在，类似于小鼠和大鼠ONC模型中报告的细胞死亡动力学。与此形成鲜明对比的是，在冬眠的tlgs中，多达90%的RGCs在ONC损伤后第21天存活。此外，这些存活的rgc不仅仅是待清除的死细胞体。当我们使用多电极阵列（MEA）记录活跃的tlgs在视网膜鼻部的RGC自发放电活动时，正如预期的那样，我们观察到与颞部相比，RGC活动急剧减少。然而，对于迟钝的tlgs，视网膜鼻半部分的RGC活动保持在与颞半部分相当的水平，表明它们仍然活跃。

Hibernating mammals survive profound hypothermia without injury, a remarkable feat of cellular preservation that bears significance for potential medical applications. We set out to examine whether hibernating thirteen-lined ground squirrel (TLGS) respond to neural injury differently from awake ones. We chose optic nerve crush (ONC) injury model as it is a classic model of axonal injury without the complication of blood vessel damage, and it is a minor surgery that can be performed in hibernating TLGSs without disturbing the hibernation status. When crushed totally, majority of RGCs died 14 days after the injury in active TLGSs. For TLGSs in torpid, however, most of RGCs survived the injury. We then focused on the partial ONC injury model so that we can quantify the survival rate by calculating the nasal (injured side)/ temporal (uninjured side) RGC ratio at the different time points after the injury. For active TLGSs, the decline in RGC population started at day 3 and went on exponentially until reaching a plateau at day 21 with about 20% of RGCs remained, similar to the cell death kinetics reported in the mouse and rat ONC models. In stark contrast, as much as 90% of RGCs survived the ONC injury by day 21 in hibernating TLGSs. Moreover, these surviving RGCs were not merely dead cell bodies yet to be cleared away. When we used multi-electrode array (MEA) to record the spontaneous RGC firing activities in the nasal half of the retina from active TLGSs, as expected, we observed a sharp reduction of RGC activities compared to that of the temporal half. For torpid TLGSs, however, the RGC activities in the nasal half of retina remained at a level comparable to that of the temporal half, suggesting that they remain active. To understand why RGCs in active and torpid TLGSs respond so differently to the same axonal injury, we collected injured optic nerve samples from both active and torpid conditions and subjected them to RNAseq. We first identified differentially expressed genes (DEGs) in response to ONC and then categorized them based on cell types (McKenzie et al., 2018). Interestingly, in samples from active TLGSs (3 days after ONC), over half of the up-regulated DEGs (80 out of 136) are microglia-related. In contrast, microglia-related genes only account for less than 10% of the upregulated DEGs in samples from torpid animals. Among the up-regulated microglia-related DEGs, typical microglial activation genes such as CD68 (Hendrickx et al., 2017), CD74 (Wang et al., 2014) , C1QB (Stephan et al., 2012) and TREM2 (Wang et al., 2015) were identified, all of which are in fact downregulated in torpid animals (Figure 2B and Data S1). This polarized transcriptome pattern of microglia-related genes in active and torpid animals prompted us to directly examine the microglial reaction near the crush site. In samples from active TLGSs, we indeed observed massive aggregation of Iba1+ positive microglial cells at the injury site. Such microglial accumulation stared as early as day 1 after ONC and last as long as we sampled (day 21). Moreover, these Iba1+ cells are mostly positive for CD68 labeling, and other macrophage-like markers, such as F4/80 (Carson et al., 1998) and MFGE8 (Liu et al., 2013), indicating that they are activated microglial cells. In contrast, such microglial aggregation was completely absent in samples from torpid TLGSs. Instead, a cell-sparse region at the crush site is apparent as revealed by DAPI labeling of nuclei. These results confirmed the transcriptomic analysis that microglial response at the crush site is a significant difference between active and torpid TLGSs. However, RGC soma situate in the retina, some distance away from the crush site. Therefore, we further examine the dynamics of microglial in the retina in response to the ONC at different time points up to 21 days. In retina samples from the active TLGSs with ONC, along with the progressive RGC loss, there is a continuing accumulation of microglial cells, as well as gradually increase of CD68 expression in microglial cells. In contrast, in samples from torpid animals, there is little RGC loss, and neither microglia aggregation nor CD68 overexpression occurs up to 21 days post ONC. Another prominent feature observed from the active TLGSs is a substantial increase of Iba1 positive microglial processes in the nerve fiber layer after ONC. Many of these processes run parallel with dendrites of astrocytes and make numerous contacts, suggesting possible interactions between microglia and astrocytes that may lead to neuronal death. These results revealed a remarkably different innate immune response to axonal injury in hibernating animals. We will further investigate the mechanism(s) of such neural survival after injury in the hibernating condition. This will help to develop therapeutic strategies for treating neural injury and degeneration.