Squirrel Pops & Shy Spines

Back when I was a first year, I remember Craig Heller telling a story about how squirrels lose a huge proportion of their synapses during winter hibernation, which they then somehow grow back when they awaken. I've used this as cocktail party conversation since then, but only recently have I gone back and actually checked out the details about this phenomenon. It turns out it's pretty incredible.

In the winter months, squirrels hibernate, entering a state of suspended animation (torpor) for weeks at a time, interspersed with brief periods of waking. During hibernation, squirrels' core temperature can fall to zero C or below. Arctic ground squirrels are the champions at this, having been measured at core temperatures of -2.9 C in the wild, although further experiments showed that they manage to keep their brain temperature just above zero (Barnes, 1989). If you're wondering how they keep their little bodies from freezing solid, apparently they manage to turn their blood into a super-cooled solution by removing particles that ice could crystalize around (Jabr, 2012;  Reardon, 2013). I still don't totally get how this works – more research necessary.

During the torpid state, squirrels switch off most neural activity, turn down their metabolism, and slow protein synthesis and cell division (Carey et al, 2003). In addition to these amazing metabolic feats, squirrels do in fact lose a large proportion of synapses during hibernation. In one study, Heller's group has measured a 20% decrease in dendritic spine density during periods of torpor, and in another they showed a 50-60% reduction in immuno-labeled synapses when comparing torpid vs. euthermic animals (von der Ohe et al., 2006;von der Ohe et al., 2007). Interestingly the 2007 study suggests that the synapses don't vanish, but rather decouple, perhaps making it easier for these connections to rewire when the animal wakes up. In any case, these are big changes in synaptic connectivity, and this phenomenon raises some serious questions about whether the squirrel that wakes up is quite the same animal as the one who went to sleep (particularly if you are in the camp of Sebastian Seung, who argues that his individuality is encoded in the pattern of synaptic connections in his brain).

Philosophical questions aside, the real conundrum is how squirrels' memories are affected by this temporary synapse loss. Since squirrels are known for hiding nuts and acorns around the forest in preparation for their long nap, I wonder how they can manage to remember where all these tasty treats are located if during their slumber they lose so many of the synaptic connections that presumably encode this information.  If squirrels are losing up to 60% of their synapses during hibernation, how do they selectively maintain these memories? Could it be that they preferentially retain the synapses involved in nut-cache memories while allowing much of the rest of their experience to slip away? Or do the new spines that sprout upon awakening somehow "remember" how they fit into the circuit such that memories formed in autumn are reconstituted accurately in spring? If so, what structural trace stayed behind to guide the synapses back to their proper partners?

One possibility is that squirrels do not remember where they hid their acorns at all, but just trust that if every squirrel caches enough acorns for herself in autumn, there will be plenty of acorns to go around in spring for any squirrel who can sniff them out. In this scenario, each squirrel doesn't need to remember exactly which buried acorn is hers and which was her neighbor's.

Apparently for much of the 20th century, this was the assumption by anyone who cared to speculate:

'Squirrels have been criticized for hiding nuts in various places for future use and then forgetting the places. Well, Squirrels do not bother with minor details like that. They have other things on their mind, such as hiding more nuts where they can't find them' (Cuppy 1949, quoted in Jacobs & Liman, 1991).

In fact, the question of how well squirrels remember their caches is a complicated one. While they are definitely capable of detecting unfamiliar caches by smell alone, squirrels are also much more likely to retrieve food they hid themselves rather than raid adjacent caches made by their neighbors (Jacobs & Liman, 1991). However, these studies did not test whether the squirrels' memories lasted through hibernation, and results on the effects of hibernation on memory are mixed, with some studies convincingly demonstrating that squirrels' performance on operant and spatial learning tasks is significantly impaired after hibernation. On the other hand, tests of social recognition showed that the squirrels could still recognize other familiar squirrels (Roth et al., 2010,Millesi et al. 2001), suggesting that all memories are not affected equally. Similar studies in bats and marmots have shown that these hibernators do not lose their abilities on spatial learning and operant conditioning tasks (Roth et al, 2010,Clemens et al, 2009, Ruczynski and Siemers, 2011).

It seems like squirrels endowed with good memories ought to have a competitive advantage over those who must rely on smell alone. However, it may be that squirrels'  memory loss after hibernation is actually adaptive. When the squirrels wake up from hibernation, their synapses go through a massive resprouting, and for several days their performance learning new memory tasks actually improves. This suggests that the loss of synapses during hibernation could actually be an opportunity for them to learn new features of their environment that may have changed over the winter months (Weltzin et al., 2006).

This possibility, that the massive pruning during hibernation enables adaptive learning of new information on waking, is reminiscent of one of the major hypotheses about the function of sleep in animals ranging from fruit flies to humans. According to the Synaptic Homeostasis Hypothesis (SHY) proposed by Tononi & Cirelli (Tononi and Cirelli, 2012), regular sleep is a mechanism for normalizing all of the new connections an animal has made during its active period. This is an elegant but controversial theory (e.g. Frank, 2012), but essentially proposes that during the waking state, animals are building up more and more synapses, which must then be pruned back during the off-line sleep phase. Numerous molecular & imaging studies point to heightened synaptic density and strength just before sleep, and a significant pruning back during sleep (Wang et al 2011). The theory rests, of course, on the assertion that there is a real bias towards potentiation during wakefulness and towards depression and pruning during sleep, and also on the dependence of "sleep need" - the actual drive to fall asleep – on the net strength of synapses, neither of which has been conclusively shown. Still, it is tempting to speculate that there could be a general mechanism of synaptic resetting, which is perhaps performed by different mechanisms and to different degrees during sleep and hibernation.

Still, there is one final, tantalizing connection here, which is that the effects of hibernation on synaptic density and dendritic morphology are highly reminiscent of effects seen by a number of labs in brain slices immersed in ice-cold solutions – spines retract and cells shrink, followed by a profusion of resprouting when the slice is warmed again (Roelandse & Matus, 2004, Kirov et al., 2004). This suggests that cold itself could drive the synaptic pruning that occurs during hibernation. Likewise, during sleep the body's core temperature drops to its lowest point in the circadian cycle. Maybe this is reaching a little, but it's tempting to wonder whether there is some deeper relationship between metabolic regulation of temperature and energy and homeostatic regulation of synapse density (Roth et al., 2010).

I've obviously just touched the tip of the ice-berg here, and I hope to come back to this topic in future posts. Let me know what you think in the comments!


Barnes, B.M. (1989). Freeze avoidance in a mammal: body temperatures below 0 degree C in an Arctic hibernator. Science 244 (4912):1593-1595. http://dx.doi.org/10.1126/science.2740905

Carey, Andrews, and Martin (2003). Mammalian Hibernation: Cellular and Molecular Responses to Depressed Metabolism and Low Temperature. Physiol Rev  vol. 83 no. 4 1153-1181. http://dx.doi.org/10.1152/physrev.00008.2003.

Clemens LE, Heldmaier G, Exner C.(2009) Keep cool: memory is retained during hibernation in Alpine marmots. Physiol Behav. 2009 Aug 4;98(1-2):78-84. http://dx.doi.org/10.1016/j.physbeh.2009.04.013.

Cuppy, W. (1949). How to Attract the Wombat.  Chicago: University of Chicago Press

Frank, M.G. (2012). Erasing Synapses in Sleep: Is It Time to Be SHY? Neural Plasticity 2012 (2012), Article ID 264378. http://dx.doi.org/10.1155/2012/264378

Jabr, F. (2012) What the Supercool Arctic Ground Squirrel Teaches Us about the Brain's Resilience. Scientific American, June 26, 2012. http://www.scientificamerican.com/article.cfm?id=arctic-ground-squirrel-brain

Jacobs, L.F. & Liman, E.R. (1991). Grey squirrels remember the locations  of buried nuts.  Anim. Behav. 41, 103 - 110. http://dx.doi.org/10.1016/S0003-3472(05)80506-8

Kirov, S.A., Petrak, L.J., Fiala, J.C., and Harris, K.M (2004). Dendritic spines disappear with chilling but proliferate excessively upon rewarming of mature hippocampus. Neuroscience 127: 69–80. http://dx.doi.org/10.1016/j.neuroscience.2004.04.053

Millesi, E., Prossinger, H., Dittami, J.Pl, Fleider, M. (2001). Hibernation Effects on Memory in European Ground Squirrels (Spermophilus citellus). Biol Rhythms  16 (3):264-271. http://dx.doi.org/10.1177/074873040101600309

Reardon, S. (2013). Zoologger: Supercool squirrels go into the deep freeze. New Scientist, January 25, 2013. http://www.newscientist.com/article/dn23107-zoologger-supercool-squirrels-go-into-the-deep-freeze.html

Roelandse, M. and  Matus, A. (2004). Hypothermia-Associated Loss of Dendritic Spines. J. Neurosci. 24(36): 7843-7847. http://dx.doi.org/10.1523/JNEUROSCI.2872-04.2004.

Roth, T.C., Rattenborg, N.C., and Pravosudov, V.V. (2010) The ecological relevance of sleep: the trade-off between sleep, memory and energy conservation. Philos Trans R Soc Lond B Biol Sci. 2010 March 27; 365(1542): 945–959. http://dx.doi.org/10.1098/rstb.2009.0209

Ruczynski and Siemers (2011). Hibernation does not affect memory retention in bats. Biol Lett. 23; 7(1): 153–155. http://dx.doi.org/10.1098/rsbl.2010.0585

Tononi G. and Cirelli C. (2012). Time to Be SHY? Some Comments on Sleep and Synaptic Homeostasis. Neural Plasticity 2012 (2012), Article ID 415250. http://dx.doi.org/10.1155/2012/415250

von der Ohe CG, Darian-Smith C, Garner CC, Heller HC. (2006) Ubiquitous and Temperature-Dependent Neural Plasticity in Hibernators. J. Neurosci.,  26(41): 10590-10598. http://dx.doi.org/10.1523/JNEUROSCI.2874-06.2006

von der Ohe CG, Garner CC, Darian-Smith C, Heller HC. (2007). Synaptic protein dynamics in hibernation. J Neurosci. 3, 27(1):84-92. http://dx.doi.org/10.1523/NEUROSCI.4385-06.2007

Wang, G., Grone, B., Colas, D., Appelbaum, L., and Mourrain, P. (2011) Synaptic plasticity in sleep: learning, homeostasis, and disease. Trends Neurosci. 34(9): 452–463. http://dx.doi.org/10.1016/j.tins.2011.07.005

Weltzin, M.M., Zhao, H.W., Drew, K.L. ,Bucci, D.J. (2006). Arousal from hibernation alters contextual learning and memory.Behavioural Brain Research 167, 1 (15):128–133. http://dx.doi.org/10.1016/j.bbr.2005.08.021,