Summary: I present the long-term memory preservation hypothesis: the idea that glutaraldehyde fixation may be able to preserve in a comprehensive way the information that encodes an organism’s long-term memories.
I review evidence from clinical research and neuroscience that suggests long-term memories are encoded in the nervous system as durable changes in synaptic nanostructure, protein distribution, and gene expression, among other physical changes. By “long-term memories,” I mean observable changes in behavior that persist for longer than 24 hours, including semantic memories, long-term motor skills, and episodic memories. Helpful background can be found in (Squire 2004), which provides a more in-depth summary of long-term memory systems.
The rest of this document is organized around the following results:
- Memories outlast electrical blackouts. Long-term memories can survive a complete loss of blood flow and electrical brain activity.
- Memories outlast individual molecules. Memories are encoded by self-reinforcing brain structures whose molecular parts are interchangeably replaced as needed.
- Memories outlast individual neurons. A single memory is stored as a distributed change across an entire group of neurons. It can survive even when some of the constituent neurons are destroyed.
- Glutaraldehyde preserves detailed nanostructure and biomolecules. Glutaraldehyde can retain virtually all proteins, DNA and other biomolecules. And it can preserve every synapse in great detail.
Which suggests the the conclusion:
- Glutaraldehyde may preserve memory. We may not know how memories are encoded, or how to read them, or which specific structures matter. But we do know that long-term memory persists in enduring biochemical and structural arrangements, not in electrical patterns, brittle arrangements of individual molecules, or localized individual neurons. Because glutaraldehyde fixation is very comprehensive, there is unlikely to be a memory storage system that evades preservation. Put another way: we know from neuroscience and clinical evidence that two otherwise identical organisms that differ by single observable behavioural difference which persists for more than 24 hours, must differ from each other by at least several biochemical / structural changes at more than one synapse. But glutaraldehyde fixation is capable of preserving the physical differences between two organisms which only differed from each other by biochemical / structural changes at a single synapse, neuron, or other cell. This means that glutaraldehyde preservation must create an injective mapping between durable memory states and preserved artifacts. Since injective maps preserve information, glutaraldehyde fixation preserves comprehensive information about long-term memories.
Note: preserving is not interpreting. I make no claims that current technology can easily access or meaningfully interpret long-term memories in a preserved organism. Rather, I’m arguing that comprehensive information about an organism’s long-term observable behavior is likely to be retained by chemical fixation.
Long-term memories are physically robust; memory storage does not require sustained electrical activity
Many organisms (including humans) can retain long-term memories while surviving highly disruptive events such as temporary global ischemia (Steen et. al. 1979) or deep hypothermic circulatory arrest (DHCA) (Behringer et. al. 2003, Percy et. al. 2008). Many profound changes take place in the nervous system during these traumatic events, such as loss of the brain’s extracellular space (Thorne & Nicholson, 2006) and complete loss of electrical activity (Raichle 1983), yet long-term memories are still retained. The information of long-term memories must be stored in a durable physical/chemical form that doesn’t depend on the nervous system’s dynamic electrical activity, or else traumatic events like ischemia or DHCA would erase long-term memories.
Note: Long-term memory recall is an active process which requires electrical activity. It should not be confused with storage. Recall is disrupted during DHCA and temporary global ischemia but returns after recovery.
Key papers: (note, click the papers in boxes to expand them and review relevant quotes!)
Bohl, Michael A., et al. “The history of therapeutic hypothermia and its use in neurosurgery.” Journal of Neurosurgery (2018)
Mizrahi, Eli M., et al. “Hypothermic-induced electrocerebral silence, prolonged circulatory arrest, and cerebral protection during cardiovascular surgery.” Electroencephalography and clinical neurophysiology (1989)
(click to expand)
Certain cardiovascular operations require cardiopulmonary bypass and prolonged circulatory arrest to provide an adequate operative field. Profound hypothermia is induced to protect brain function during these periods without cerebral perfusion…
…Electrocerebral silence (ECS), induced by profound hypothermia, is readily recognizable in the EEG…
…The EEG of most patients showed a characteristic evolution with progressive core hypothermia. In all patients, there was initial depression of amplitude and slowing of background rhythms. There was progressive depression of amplitude which, in 17 patients, was preceded by a period of episodic generalized voltage attenuation (suppression-burst activity). This was followed by the development of generalized, periodic, slow-wave transients. The transients were mono- or polyphasic, of moderate voltage, and of relatively long duration (up to 0.5 sec). The intervals between transients became more prolonged, and the transients became progressively less complex and more depressed in amplitude. This eventually led to the onset of ECS…
…Duration of circulatory arrest was 14-109 min (mean, 39 min). The time required for cooling to the onset of ECS was 20-107 min (mean, 41 min)…
Percy, Andrew, et al. “Deep hypothermic circulatory arrest in patients with high cognitive needs: full preservation of cognitive abilities.” The Annals of thoracic surgery (2009)
(click to expand)
This study presents relatively strong evidence that patients with high cognitive needs who underwent aortic surgery using DHCA (deep hypothermic circulatory arrest) experienced no perceptible cognitive change as a consequence of this procedure. This study, using a self-administered questionnaire, supplemented by a familial informant, provides direct subjective feedback by patients who underwent DHCA and their families. Our study found excellent preservation of cognitive function after surgery, according to both patient and informant responses. Although subtle deficits after DHCA might hide in individuals with less intellectually demanding professions, it is unlikely that substantive deficits could remain undetected in our high-cognitive needs group.
…High-cognitive professions were loosely defined, but were essentially limited to physicians, lawyers, doctorates, clergymen, artists, musicians, accountants, and managers…
Our results indicate that there was no significant difference in quality of work generated after surgery between patients who had DHCA and those without. This validates the initial motivation of this study, based on our clinical impression that DHCA did not seem to affect or bias patients’ postoperative performance in their work field.
Raichle, Marcus E. “The pathophysiology of brain ischemia.” Annals of Neurology: Official Journal of the American Neurological Association and the Child Neurology Society (1983)
(click to expand)
Interruption of cerebral blood flow results in loss of consciousness within 10 seconds and cessation of spontaneous and evoked electrical activity within 20 seconds. Within several minutes after the loss of electrical activity there are major disruptions of normal tissue ion homeostasis. Ion-sensitive microelectrodes placed in the extracellular fluid space of the brain record a marked increase in potassium concentration and a fall in sodium and calcium concentrations…
…Steen et al recently reported that fasted dogs subjected to complete ischemia could tolerate only 8 to 9 minutes of ischemia and go on to achieve normal electrical activity…
Wu, Xianren, et al. “Emergency preservation and resuscitation with profound hypothermia, oxygen, and glucose allows reliable neurological recovery after 3 h of cardiac arrest from rapid exsanguination in dogs.” Journal of Cerebral Blood Flow & Metabolism (2008)
Biological information storage mechanisms use biochemical “cycles” and spatially-distributed engrams to implement physically robust long-term memories
How do organisms store information for the long term in a form that can survive ischemia and deep hypothermic circulatory arrest? Science still has much to learn about how neurons and other cells process information and encode long-term memories at the systems level. But at the lowest levels of molecules and structure, decades of hard work by neuroscientists have revealed many of the physical and chemical mechanisms organisms use to store information (Bailey et. al. 2015).
We know that for long-term memory formation to occur, A) the distribution of neuronal proteins and especially synaptic proteins must change (and maintain the change), and B) these changes have to happen at multiple neurons and synapses. Ultimately these changes alter neurons’ electrical behavior, which in turn alters observable behavior (Langille & Brown 2018).
Memory is not represented by change at a single synapse, but by a series of processes involving molecular, biochemical, cellular and circuit level changes in widespread constellations of neurons throughout the brain.(pg. 3, Langille & Brown, 2018)
Cycles allow molecular changes to last
The concept of the self-reinforcing biological cycle has become clear over the last several decades. In this type of cycle, a set of molecules work together to reinforce the concentrations and states of the other molecules in the cycle. Even though individual molecules may degrade over time, the cycle itself is able to maintain its state almost indefinitely.
If you know how the cycle works, then once you know the state of one part of the cycle, you have a great deal of information about the states of the other parts of the cycle. This is especially true over the long term when the cycle is in a steady state. As an example, one important biomolecular cycle involved in learning and memory controls the number of AMPA receptors at glutamatergic (excitatory) synapses. Generally, the more AMPA receptors at an excitatory synapse, the “stronger” that synapse is and the more influence it has on the neuron to which it is attached (Bartol et. al. 2015).
When a new long-term memory such as a fear memory is formed, the strengths of many synapses must change to alter the long-term behavior of the organism (Abdou et. al. 2018). To change its strength, a glutamatergic synapse employs a huge variety of molecules all working together to change the “set point” of the number of AMPA receptors present. The components of the “AMPA cycle” include (among others): NMDA receptors, CaMKII, MAPK, PKA, BDNF, CREB, mTOR, PDK1, NSF, PIN1, PI3K, PKMζ, NSF, PICK1, and AMPA receptors themselves, as well as the physical size of the synapse itself. All these components are part of a self-reinforcing cycle that sets, stabilizes, and maintains the strength of each synapse (Sacktor 2011).
The end result of this cycle is that synapses can alter their strength and maintain this alteration even though the individual molecular components of the synapse are always in flux. Of course, the nervous system has many biochemical cycles that work in parallel to maintain information (and homeostasis). The “AMPA cycle” is one such cycle with particular relevance to long-term memory.
Engrams encode long-term memories using multiple synapses and neurons
A second important concept from neuroscience is the concept of an engram or attractor state, formed when an assembly of neurons alter the strengths of their synaptic connections during memory encoding (Choi et. al. 2018). Although there is still much to learn about how engrams work at a high level, recent work has been very impressive: Engram cells can be tagged and later manipulated (Poo et. al. 2016), and this technique has been used to erase long-term memories (Han et. al. 2009) or even create false long-term memories (Liu et. al. 2014). The key point is that creating a new engram requires changes in multiple neurons and synapses (Xu et. al. 2009).
The prevailing view is that the formation of an engram involves strengthening of synaptic connections between populations of neurons that are active during encoding, leading to the formation of a neuronal ensemble. This increase in synaptic strength between neurons increases the likelihood that the same spatiotemporal pattern of neural activity that occurred during encoding will be recreated at a later time (retrieval). Engrams need not be confined to a single brain region, but rather may be composed of widely distributed networks of neuronal ensembles.(p. 1 Josselyn et. al. 2015)
Most notably, engrams can survive the destruction of some of their constituent neurons / synapses (Hayashi-Takagi et. al. 2015).
How can the information of long-term memories survive traumatic events like ischemia and DHCA? Biochemical cycles and the physical redundancy of engrams play a big role. Engrams can survive the death of a few neurons (Han et. al. 2009) because they are themselves composed of many neurons and synapses working together. Individual synapses and neurons can survive and recover from some molecular damage, since their biomolecular storage cycles encode information redundantly in the states of mutually reinforcing macromolecules such as proteins, mRNAs, nanostructural components and even modifications to DNA (Chen et. al. 2014).
Josselyn, Sheena A., Stefan Köhler, and Paul W. Frankland. “Finding the engram.” Nature Reviews Neuroscience (2015)
Memories are thought to be encoded as enduring physical changes in the brain, or engrams. Most neuroscientists agree that the formation of an engram involves strengthening of synaptic connections between populations of neurons (neuronal ensembles).
…not only can contemporary rodent studies claim to have found the engram, but also have identified means to control it.
Langille, Jesse J., and Richard E. Brown. “The Synaptic Theory of Memory: A Historical Survey and Reconciliation of Recent Opposition.” Frontiers in systems neuroscience (2018)
Synaptic change is thus the first step in a series of events which link molecular activity at the synapse and the subsequent intracellular biochemical cascades and cellular changes to the cognitive aspects of memory. The neurobiological basis of memory exists as a series of synapse-specific molecular and biochemical changes, including de novo protein synthesis, phosphorylation, up-regulation of synaptic receptors and synaptic growth within and between cell assemblies which results in long-term changes to synaptic efficacy…
…Any memories stored in the neurons of the cell assembly or phase sequence are expressed through synaptic modifications…
…even if one was to concede that memory may reside in individual cells or in genomic and epigenomic modifications, these cells form a cell assembly and the expression of that memory requires the activation of the synapses linking the cells. In order to facilitate the formation of a cell assembly, synaptic activation occurs and as a memory is consolidated into a cell assembly, synaptic modifications occur. Without these synaptic modifications, memories could not be expressed…
Bailey, Craig H., Eric R. Kandel, and Kristen M. Harris. “Structural components of synaptic plasticity and memory consolidation.” Cold Spring Harbor perspectives in biology (2015)
Perhaps the most striking finding in the cell biology of memory is that the consolidation and long-term storage of memory involves transcription in the nucleus and structural changes at the synapse. These structural components of learning-related synaptic plasticity can be grouped into two general categories: (1) remodeling and enlargement of preexisting synapses, and (2) alterations in the number of synapses, including both the addition and elimination of synaptic connections.
Bourne, Jennifer, and Kristen M. Harris. “Do thin spines learn to be mushroom spines that remember?” Current opinion in neurobiology (2007)
Hayashi-Takagi, Akiko, et al. “Labelling and optical erasure of synaptic memory traces in the motor cortex.” Nature (2015)
…In vivo imaging of AS-PaRac1 revealed that a motor learning task induced substantial synaptic remodelling in a small subset of neurons. The acquired motor learning was disrupted by the optical shrinkage of the potentiated spines, whereas it was not affected by the identical manipulation of spines evoked by a distinct motor task in the same cortical region. Taken together, our results demonstrate that a newly acquired motor skill depends on the formation of a task-specific dense synaptic ensemble.
Han, Jin-Hee, et al. “Selective erasure of a fear memory.” Science (2009)
Litt, Abninder, et al. “Is the brain a quantum computer?” Cognitive Science (2006)
There is substantial evidence that the timescales for individual quantum events in the brain are not in accord with the temporal requirements for influencing neural firing in a consequential (i.e., non-noise) manner…
…The power of a quantum computer lies in its ability to maintain superposed qubit states long enough to facilitate superparallel computation. The maintenance of an extraordinarily high degree of isolation from even minute environmental interactions is a vital prerequisite for preventing decoherence, which is the decay of coherent quantum-state superpositions caused by such interaction…
…Standing in stark contrast to these physical requirements are the conditions that exist in and around animal brains. Brains are warm, wet, biological constructs, honed by evolution to exhibit the sort of robustness and durability needed for survival in the world. Although to some extent they are protected from the environment by a thick skull, cushioning fluid layer, and so on, this isolation is nowhere near sufficient to maintain large-scale quantum coherence at the neuronal level for computationally significant periods of time…
…there appear to be no special quantum mechanical properties needed to explain psychological and neurological phenomena. The onus is on those who would appeal to quantum theory to show the existence of aspects of the brain that are not explained by neurocomputational theories, and that can be explained by quantum computation or associated mechanisms.
Glutaraldehyde is amazingly effective at preserving biological tissue in great detail
For over 50 years, glutaraldehyde and other similar aldehydes have been used to preserve biological specimens for detailed analysis. Glutaraldehyde reacts rapidly with tissue to form a densely crosslinked, stable gel-like form which can withstand major changes in pH, temperature, osmotic stress, and other ordinarily destructive insults (Hopwood 1973). Virtually all proteins and mRNAs can be labeled and analyzed after aldehyde fixation using techniques like in situ hybridization (for mRNAs) (Rosenberg et. al. 2018) and immunohistochemistry (for proteins) (Murray et. al. 2015). Both of these techniques have been used with great success to study the vast array of biomolecules which make up cells (Griffiths 1993).
Aldehyde fixation must retain virtually all proteins and mRNAs, since both immunohistochemistry and in situ hybridization techniques start with aldehyde fixation, and these techniques would not be able to analyze these biomolecules if they were lost during the first step of processing.
Formaldehyde fixation has been shown to retain all proteins (within the limits of Ostasiewicz et. al. ability to measure) as well as post-translational modifications of proteins such as phosphorylation and glycosylation. Formaldehyde protects proteins even from the very harsh conditions of paraffin embedding, which involves complete dehydration of the tissue, exposure of the dehydrated tissue to the powerful solvent xylene, and infiltration of hot (60° C) molten paraffin wax into the tissue (Ostasiewicz et. al. 2010). Heckman & Barrnett 1973 have shown using radiolabeling that glutaraldehyde fixation retains all lipids (within their capability to measure). Glutaraldehyde fixation has long been the primary means of preparing tissue for electron microscopy because of its ability to preserve nanostructure in near-perfect detail (Hayat 1981).
Consider again our previous example of the biological information storage cycle involving AMPA receptors: Sacktor 2011 specifically calls out NMDA receptors, CaMKII, MAPK, PKA, BDNF, CREB, mTOR, PDK1, NSF, PIN1, PI3K, PKMζ, PKMζ mRNA, PICK1, and AMPA receptors as important parts of the cycle. How well does aldehyde fixation work to preserve these biomolecules?
Quite well indeed. All of the biomolecules referenced by Sacktor 2011 are retained by aldehyde fixation:
Note 1: Glutaraldehyde does present some challenges with visualizing certain kinds of biomolecular detail. The strong and dense crosslink network it forms can actually prevent probes such as antibodies from diffusing to or even attaching to preserved proteins. This is why formaldehyde (which forms more “loose” crosslinks) is often used instead of glutaraldehyde in immunohistochemistry. However, for our purposes we are more concerned with whether biomolecules are retained in-place rather than whether they can be easily visualized with any particular immunostaining protocol. In general, because glutaraldehyde forms denser crosslinks than formaldehyde, any biomolecule retained by formaldehyde is also extremely likely to be retained by glutaraldehyde, even though some biomolecules might be harder to visualize with glutaraldehyde than with formaldehyde.
Note 2: Aldehyde fixation in general does cause some morphological changes, the most prominent of which is the loss of the brain’s extracellular space (Korogod et. al. 2015). However, these structural changes closely mirror the structural changes that happen during transient global ischemia (Tao-Cheng et. al. 2007). As discussed earlier, the morphological changes which accompany ischemia do not erase long-term memories.
Griffiths, Gareth. Fine structure immunocytochemistry. Springer Science & Business Media, 2012.
Murray, Evan, et al. “Simple, scalable proteomic imaging for high-dimensional profiling of intact systems.” Cell (2015)
Ostasiewicz, Paweł, et al. “Proteome, phosphoproteome, and N-glycoproteome are quantitatively preserved in formalin-fixed paraffin-embedded tissue and analyzable by high-resolution mass spectrometry.” Journal of proteome research (2010).
Both qualitatively and quantitatively we were not able to detect differences in protein levels between fresh or fixed and paraffin embedded proteomes.
…Using the FFPE-FASP protocol we demonstrated for the first time that the phosphoproteome is preserved in fixed and paraffin embedded tissue.
Taken together, clinical, neuroscientific, and biochemical evidence suggests that glutaraldehyde fixation comprehensively preserves the information that encodes an organism’s long-term memories
Consider the changes that must take place in the nervous system of an organism to create a long-term memory.
Our current understanding of neuroscience implies that any empirically observable long-term change in an organism’s behavior would have to involve durable changes in nanostructure and protein distribution at multiple synapses. Redundant biochemical cycles ensure that these changes are robust enough to survive the natural turnover of biomolecules, as well as catastrophic events like temporary global ischemia and deep hypothermic circulatory arrest.
Now, consider preserving an organism with glutaraldehyde. Glutaraldehyde fixation is so comprehensive that it allows differentiation between even slight differences in mRNA, protein distribution and nanostructural changes at a single synapse, or changes in gene expression in a single neuron. These minor differences are far below the level of physical changes which would be behaviourally observable in a living organism.
In summary, the signal the nervous system employs to create a long-term memory (robust self-perpetuating biochemical changes at multiple synapses) is greater than the noise introduced by glutaraldehyde fixation (which can preserve even functionally irrelevant changes at a single synapse).
For an organism to have behaviourally observable information that glutaraldehyde fixation doesn’t preserve, it would need to have an independent memory-storage mechanism with two properties:
1) Robust enough to survive ischemia, deep hypothermic circulatory arrest, concussions, anesthetic, MRI scans, and all other situations where long-term memory survives; and
2) Every component of the biochemical cycles that implement the memory system would have to be lost without a trace during glutaraldehyde fixation.
Practically speaking, this means that the memory-storage mechanism could not be implemented via changes to proteins, gene expression, or alterations in nanostructure, since these aspects are preserved by glutaraldehyde. No modern neuroscientific theory of long-term memory seriously suggests that such a memory system exists.
Thus we conclude that glutaraldehyde fixation may be able to preserve in a comprehensive way the information that encodes a typical organism’s long-term memories.
Want to get in touch? Contact us at firstname.lastname@example.org.