Summary

It’s helpful to have rich mental models of the structures of the brain; otherwise, it is difficult to integrate empirical data when reasoning about what happens to the brain during the dying and preservation processes. Based on my research, I currently feel that the most important set of structures in the brain are water-soluble gel-like networks of biomolecules that are in a constant state of remodeling. We can study the properties of these gel-like structures: how strong they are, how they might break down or be strengthened, how other biomolecules are anchored to them, and how they can be measured. It seems to be better when structures in the brain that contain the information content for valued cognitive processes, like engrams, are integrated into or anchored in stronger gel-like networks, because stronger gel-like networks seem to be relatively better maintained during the dying and preservation processes.

The structures of the brain during everyday life

Like other organs, the brain is obviously not a bag of liquid. And neither are the individual cells inside of the brain. And yet the brain and its constituent cells are clearly not solids either. So on a most basic level, the brain’s native structure must be something in between a liquid and a solid. A common way to model the material properties of the brain is via gelatin (Ploch et al., 2016), and an every day example of gelatin is jello.

One type of structure that is between a liquid and a solid is a gel. Gels can also be thought of as a solid with a liquid dispersed throughout it. Gels are a classic type of soft matter that are omnipresent in our lives and in biological systems (Douglas, 2018). As (Douglas, 2018) describes it:

“Gels are ubiquitous—many foods (e.g., gelatin, cheese, ketchup), consumer products (e.g., toothpaste, cosmetics, shaving cream, etc.) and industrial products (e.g., adhesives, asphalt) can be defined as belonging to a rheologically-defined class of materials termed”gels”. Moreover, the gel state is also characteristic of biological materials since the cytoplasm of eukaryotic cells is typically a gel, the cell as a whole has gel-like rheological properties, collagen extracts forms gels similar to those found in the cell extracellular matrix, and, in some cases, gel-like rheology extends to animal tissues composed of ensembles of cells “glued” together by the extracellular matrix, and it is thus not surprising that diverse forms of biological material such as foods, biofilms, soils, etc. have a gel-like nature. Gels are then the quintessential form of soft matter.”

Biological systems are often described as “gel-like” because there might be slight deviations from gel behavior or because the gel behavior may not have been conclusively shown. I often use this terminology of “gel-like” so as to not claim to be more precise than is warranted by the data.

Most organs, including the brain, can be thought of as biological gels or more specifically hydrogels. Hydrogels are gel structures that can maintain their shape when dissolved in water (Liu et al., 2020). The sol phase of a hydrogel is a flowing fluid (Jeong et al., 2012). The gel phase maintains a particular 3D network structure and does not flow or change its shape on practical timescales. So, a hydrogel has solid-like properties despite being contained in a liquid water solvent.

Transition of particles from the sol to gel state; gelierung = German for gelation; Image source

Biological hydrogels tend to be soft, elastic to a degree but also viscous to a degree, and contain a high concentration (60-90%) of water (Liu et al., 2020). As (Liu et al., 2020) notes, the body can be thought of as a living machine made up of mostly skeletons and different types of hydrogels.

Basic terminology of gel-like materials

What is a gel? Of all of the fields I have learned about in writing these essays, perhaps none has less precise definitions than the materials science of gels (Almdal et al., 1993). Defining what a gel is appears to be a Sisyphean task for humanity. I will try to avoid this quagmire by using somewhat non-rigorous, not always consistent definitions and begging forgiveness. As one of the historical pioneers in the field, Dorothy Lloyd, noted:

“The colloidal condition, the”gel”, is one which it is easier to recognize than to define, and even recognition is confused by the fact that the limits between gel and sol, on the one hand, and gel and what may be termed curd, on the other, are not precise, but consist of a gradual change. For this reason some workers classify as “gels” systems which others exclude. Only one rule seems to hold for all gels, and that is that they must be built up from two components, one of which is a liquid at the temperature under consideration, and the other of which, the gelling substance proper, often spoken of as the gelator, is a solid. The gel itself has the mechanical properties of a solid, i.e., it can maintain its form under the stress of its own weight, and under any mechanical stress it shows the phenomenon of strain.” (from Lloyd, 1926, as cited in (Almdal et al., 1993)).

Let’s review some properties of gels so that we can develop mental models of their biophysics. Intuitively, gels have a consistency similar to jello and have a “squishy” response to stress (Douglas, 2018).

One important property of gels is that they are elastic. Elasticity is the propensity of a material to reversibly resist a distorting force by changing its shape and then returning to its original shape when the force is removed – sort of like a trampoline. This is as opposed to plasticity, which refers to the propensity of a material to undergo an irreversible change in shape in response to a force.

One way to quantify the elasticity of a gel is by its Young’s modulus. This quantifies the relationship between the stress (i.e., the force per the area) and the strain (i.e., the amount of deformation) in a material in response to a uniaxial force – meaning a push or pull in one direction relative to the object.

The Young’s modulus is the slope of the linear portion of the stress-strain curve for a material; Source: Nicoguaro

Gels are generally called viscoelastic when they have both viscous and elastic properties. One way to quantify the viscoelasticity of a gel is by its response to an oscillatory force. In response to an oscillatory force, a more elastic gel-like object will exhibit strain that is in phase with the force, so that they occur simultaneously. In contrast, for a more viscous gel-like object, there will be more of a delay between the strain observed and the stress applied. Brain tissue, like that of most soft organs, is viscoelastic because it exhibits both elastic and viscous responses to an oscillatory type of stress.

The storage modulus and loss modulus of a gel-like object quantify the relationship between the stress and the strain in response to an oscillatory force.

The storage modulus:

The loss modulus:

One convention is that during gel formation, a gel is said to have formed when the ratio of the storage modulus to the loss modulus is greater than one. As a gel breaks down over time, it will exhibit more viscous behavior and less elastic behavior.

As the degree of crosslinking in a gel increases, the gel will exhibit more elastic behavior and less viscous behavior. Generally speaking, crosslinking of the constituent molecules is usually required for gel formation; however, there are some exceptions in which crosslinks are not required for gel formation, in which the topological arrangement alone of long, stiff, long-lasting biomolecules such as filamentous proteins can produce gel-like rheological behavior (R. Raghavan et al., 2012).

Highly viscous materials without significant elastic components generally act similar to glass-like materials. On longer time scales, cells in the brain tend to act like glass-like materials, insofar as they demonstrate slow, locally non-elastic rearrangements in response to a force (Deng et al., 2006). This is because on longer time scales, the components that make cells elastic, such as the actin cytoskeleton, are able to actively dissociate and remodel to change their shape in response to a given force (Hohmann et al., 2019).

Yet another important parameter of gels is the yield point. The yield point indicates the location on a stress-strain curve where a material stops demonstrating elastic behavior – temporarily deforming in response to a stress – and begins to demonstrate plastic behavior, which is permanently deforming in response to a stress.

When a gel reaches its yield point, it “melts,” “fluidizes,” or “liquefies” into its component particles or polymers (Douglas, 2018). This phenomenon is known as shear thinning, wherein the viscosity of the material increases as the stress increases. An everyday example of fluidization of a viscous material is ketchup and how it stubbornly refuses to flow out of the bottle until enough stress is applied and it begins to flow out of the bottle rapidly all at once.

The cytoskeleton as a gel

The intracellular space has gel-like properties that appear to be largely due to the cytoskeleton. The human cytoskeleton is primarily made of proteins, which can be categorized into actin, intermediate filaments, microtubules, and supporting proteins such as crosslinkers. The precise protein composition of the cytoskeleton varies significantly based on the cell type. For example, neurons have specialized intermediate filament proteins called neurofilaments, in astrocytes the primary intermediate filament is a protein called GFAP (Pekny et al., 2004), and oligodendrocytes do not have intermediate filaments (Bauer et al., 2009).

On short time scales, actin and microtubule polymers are crosslinked to form a stable mesh by a large number of crosslinking helper proteins. So over short time scales, the cytoskeleton is said to exhibit elastic behavior, meaning that when a stress is applied, the network will bend temporarily but return to its original configuration once the stress is removed (Mogilner et al., 2018).

Over long time scales, the cytoskeletal crosslinks tend to dissociate, meaning that the polymer units can move around and the mesh can remodel prior to crosslinking once again. The turnover of cytoskeletal crosslinking helper proteins is generally in the range of seconds to minutes. So over longer time periods of seconds, the cytoskeleton is said to exhibit viscous behavior (Mogilner et al., 2018).

Importantly, the major type of cytoskeletal remodeling that causes it to display viscous behavior during life, as opposed to elastic behavior, is an active process. For example, one of the main ways that actin filaments break down is via the binding of a particular protein, called ADF-cofilin. This binding increases the local mechanical stress in an actin filament and thereby increases its fragmentation rate (Hohmann et al., 2019).

The binding of ADF-cofilin to actin requires energy in the form of ATP. ATP can be thought of as the molecular unit of currency, because it can be “spent” in order to stimulate chemical reactions in a cell. After clinical death, cellular energy supplies in the form of ATP are rapidly depleted. As a result, once ATP has been depleted after clinical death, this form of regulated breakdown of actin filaments will not occur. The cytoskeleton can still decompose as a result of autolytic processes such as hydrolysis; however, these will likely take a longer amount of time.

Of the three major cytoskeletal proteins, the actin cytoskeleton is often thought to be the primary determinant of the mechanical strength of cells (Ananthakrishnan et al., 2006). One study found that disrupting actin filaments decreases the mechanical strength by threefold, whereas disrupting microtubules or intermediate filaments does not materially change this value (as cited in (Ananthakrishnan et al., 2006)). The cell membrane has a bending rigidity of around 1000 times less than the cortical actin network (as cited in (Ananthakrishnan et al., 2006)).

Schematic of the actin cortex cytoskeleton in axons; (Leite et al., 2016)

Cytoskeletal networks created by neurofilaments, which are the intermediate filaments found in neurons, are also found to be quite stable over time (Yuan et al., 2009). Neurofilaments have unique sidearms that allow them to form parallel arrays that can support the highly polar shapes of neurons. As opposed to actin and microtubules, neurofilaments form a highly stationary and a metabolically stable network – they don’t turn over as much (Mages et al., 2018).

The cytoskeleton tethers the plasma membrane in place through a specialized portion of it called the cell cortex. In oligodendrocytes, which have a specialized wrapping extensive of its plasma membrane called myelin, myelin-specific proteins such as MBP interact closely with cytoskeletal proteins to maintain their association (Bauer et al., 2009).

The proteins that make up the cytoskeleton are properly referred to as biomolecules. Other biomolecules are frequently attached to the cytoskeleton directly through covalent bonds (Mogilner et al., 2018). This is a major form of what we can consider anchoring. It stands to reason that biomolecules which are directly attached to the cytoskeleton will retain their positions longer during the dying process. As far as I can tell, this is generally what is found: for example, proteins embedded in the cell membrane, which is directly attached to the cytoskeleton, are reported to be relatively stable even when profiled some period of time after death in the brain. As (Waldvogel et al., 2006) describe it:

Post-mortem human tissue, due to the nature of its preservation, is unpredictable in its immunohistochemical labeling compared to animal tissue. So far we have not been able to demonstrate that any particular factor such as post-mortem delay, age or sex of the subject, pH or various fixative regimes governs staining ability and antigen preservation in human tissue. It appears more likely to be influenced by the “agonal state” of the subject, that is, the tissue oxygenation and general state of health of the subject leading up to death. Receptor proteins and other proteins which are embedded in the cell membranes are generally stable, and structural proteins such as components of the cytoskeleton are also quite stable, whereas neurotransmitter candidates and certain enzymes are much more labile and less likely to be preserved.

Biomolecules can also be indirectly constrained to particular positions as a result of diffusion restrictions imposed by the cytoskeleton. The diffusion of protein molecules is restricted by the pore size of the actin mesh (Mogilner et al., 2018). The pore size is the amount of space in a 3D network of actin filaments that is devoid of actin.

One study estimates that for actin monomers, the decrease in diffusion speed due to the actin mesh compared to pure water is around an order of magnitude (as cited in (Mogilner et al., 2018)). While this slows down diffusion somewhat, it makes sense that biomolecules only constrained in place by membranes and diffusion constraints due to the cytoskeletal mesh, such as small molecule neurotransmitters, would be less likely to be retained during the dying processes, during which membranes are likely to be damaged.

Gel-like structures all the way down

Consistent with the gel-like structure of cells in general, subcellular areas also contain gel-like structures. For example, the synaptic bouton and post-synaptic density both form gel-like structures (Zeng et al., 2016). Protein clustering of these structures at membranes causes phase separation in a way that seems to be important for their function (Feng et al., 2019).

When a “ribosome” is seen on a microscopy image, that has either been frozen or fixed prior to visualization, what is really being captured on a molecular level is the aggregation of the molecules that comprise the ribosome into a gel-like network in a stereotyped shape and position. When we see the expected shape of this type of aggregation under the microscope, we might say something like “the ribosomes are present.”

Whether a structure can be seen under the microscope after fixation, therefore, is based on the degree of aggregation of its biomolecular components. (Wang et al., 1987) have a nice table of whether different structures are expected to be seen under the microscope, based on their degree of aggregation.

The expected preservation quality of cellular sub-structures depends on their degree of aggregation; (Wang et al., 1987)

My understanding of the reason for this is that the loose structures are weaker gel-like structures or even not gel-like at all. As will be discussed in a later essay, crosslinking-based fixation, which is a key part of processing for routine light microscopy, strengthens and stabilizes gel-like networks through covalent crosslinking of constituent biomolecules. However, if the gel-like network is not strong enough to begin with, then the fixation process will not be able to stabilize it enough to retain it.

One related consequence of fixation is it can potentially introduce gel-like structures that are not necessarily present in native tissue. This is one reason that artifacts can occur in microscopic images of fixed cells. Historically, this may have happened with the “mesosome”, a structure associated with bacterial cell membranes, that only appears in tissue prepared via chemical fixation, but not cryopreservation. However, there appears to still be some uncertainty in the field about whether the mesosome is actually or always an artifact.

Cell membranes in fixed or frozen B. subtilis; (Nanninga, 1971)

Liquid-like and gel-like structures in the cell

Not everything in the cell is best modeled as being a part of a gel-like structure. To explore this, let’s hone in on the nucleolus, a structure in the nucleus that plays a role in making ribosomes (Lafontaine et al., 2021) (Riback et al., 2022).

The nucleolus has been described as a prototypical biomolecular condensate. Biomolecular condensates are organizational structures in cells that lack membranes and have a mixture of liquid-like and solid-like properties. The nucleolus is surrounded by the less dense nucleoplasm. The structure of the nucleolus is still an active area of investigation, and one can imagine which of its potential properties would make it more liquid-like vs solid-like; liquid-like properties would include mixing of its components and exchange of its components with the surroundings (Riback et al., 2022).

Theoretical liquid-like vs solid-like properties of the nucleolus; (Riback et al., 2022)

(Riback et al., 2022) is a study that sheds light on how the nucleolus is organized at the biomolecular level. They use diffusion monitoring techniques in live cells to monitor how biomolecules move about the nucleolus.

One of the major components of the nucleolus is ribosomal RNA, or rRNA for short. Nascent, random coil structured rRNA is much larger in diameter than more mature folded rRNA, as quantified by a higher radius of gyration, Rg. As rRNA matures, it leaves the nucleolus. (Riback et al., 2022) find that random coil rRNA has more aggregation and therefore forms a gel-like structure that endows the nucleolus with higher degrees of gel-like elasticity.