Spongiform Encephalopathies, Alzheimer's, and Other Diseases of Abnormal Polymerization: A Thermodynamic Model


Prional diseases certainly are candidates for the most unusual afflictions of humanity and mammals generally. Originally, they came to light as scrapie, a disease of Icelandic sheep. Marked by a long time course and bizarre behavior, these diseases were referred to as "unconventional slow viral infections". They were unconventional in the sense that they were all uniformly fatal, evoked no inflammatory or immune response, showed no eclipse period1, and resisted all attempts to isolate or visualize a virus-like agent.

Among the first human prional diseases was kuru, allegedly transmitted by cannibalism, and Creutzfeldt-Jakob disease (CJD), notorious for its resistance to hospital sterilization procedures. Although linked by a common pathology, the prional diseases include those like kuru that are entirely acquired and those like Gerstmann-Straussler syndrome that are entirely genetic.

The current paper presents a model for diseases of abnormal polymerization (DAPs), as a newly recognized class of disorders. This model explains the central mystery surrounding prional disease, namely, their contagious nature in the apparent absence of nucleic acid. The model builds upon recent explanations in the consideration of the thermodynamic and kinetic properties of the proteins involved. Building upon recent proposals that aberrant protein can stabilize a rare conformational shift that occurs spontaneously, it differs from previous models that aberrant protein actually induces normal protein to undergo a conformational change. Thus, the model focuses on structural stabilization rather than enzymatic conversion. Moreover, the model differs from previous models in suggesting that individual prion protein molecules (see below) would not be infectious because their abnormal conformation would never (practically speaking) be stabilized. Using prional disease as an example, this paper elucidates some of the thermodynamic variables that need to be considered in rigorous modeling of polymerization diseases. It illustrates how a rare molecular event can have devastating consequences for the host. It also explains how the model can be extended and compared to a number of similar as well as some apparently dissimilar conditions.


Stanley Prusiner and coworkers carried out numerous experiments demonstrating that a variety of agents that disrupt or modify nucleic acids have no apparent effect on infectivity or titer of the etiological agents of spongiform encephalopathies and that agents which affect the integrity of proteins almost uniformly diminished the infectivity of the so-called prions. Prional diseases carry the apparent distinction of being the only class of infectious disease that is transmitted independent of nucleic acid. (Recent papers by Wickner and others suggest that this distinction may be shared by so-called "non-Mendelian genetic elements" of [URE3] and [PSI] in Saccharomyces cerevisiae.) Early models included the unsatisfying suggestion that prions replicated by a novel form of nucleic independent protein synthesis.

A significant breakthrough in the study of prional diseases occurred when Prusiner et al. Developed a model replication system which greatly decreased the time course needed to do experimental studies and greatly facilitated the ability to do quantitative studies. The method consisted of directly inoculating infectious material into the brains of hamsters and using the time to clinical symptoms as a measure of infectious titer. This assay expedited attempts to purify the prional agent. Purified infectious fractions consist largely or entirely of a single protein labeled PrP for prion protein. Prp is encoded by a host gene which in humans is located on the short arm of chromosome 20. The gene has been cloned, sequenced, and monoclonal antibodies have been raised to its protein product. The protein is a 33-35 kilodalton protein which is normally found in the membrane of neurons, bound to a phosphatidyl inositol (PI) anchor which appears to facilitate trafficking of the protein between the exterior and interior of the cell. The specific function of PrP is unknown. PrP knockout animals, in which the gene has been experimentally disrupted, are relatively viable and relatively normal in their phenotype.

PrP isolated from spongiform encephalopathies consists of an isoform of PrP which is polymerized and appears to differ in its three dimensional conformation. The normal cellular form of PrP has a predominance of alpha helical secondary structure (42%) and negligible beta pleated sheet (3%). In contrast, the PrP isolated from scrapie has a decreased alpha helical content (30%) and a predominance of beta pleated sheet (43%).

Model Proteins

In building the polymerization model for the spongiform encephalopathies, it is useful to consider the alpha and beta forms of keratin. Keratin can exist in at least two kinetically stable forms. The alpha forms, made up of large regions of alpha helices, is considerably more thermodynamically stable. Therefore, in nature, keratin exists almost exclusively in the alpha form. Under certain conditions, it is possible to produce a predominance of the beta form, in which regions that were alpha helical are now stabilized as beta pleated sheets. Although transiently stable, keratin will eventually revert back to the more thermodynamically stable alpha form.

It is also instructive to examine the pathogenesis of another disease of abnormal polymerization, namely, sickle cell anemia (SCA). In SCA, abnormal polymerization of hemoglobin can occur if two conditions exist simultaneously. The first condition is the predominance of genetically altered beta chains (HbS) in which the acid amino acid glutamate is replaced with the hydrophobic amino acid valine at amino acid position 6. The second condition is a low partial pressure of oxygen resulting in a conformational change associated with the deoxygenated state. In the sticky patch model (presented in model), each of these conditions results in formation on Hb of a complementary hydrophobic region. Although neither condition alone is sufficient, together they allow the hemoglobin to polymerize into form long insoluble fibrils. These fibrils consist of large aggregates of tetrameric hemoglobin. The fibrils can be massive enough to deform the red blood cell from a biconcave disk to rigid sickles or other shapes, which cannot pass through capillary beds leading to the formation of emboli and loss of red cells resulting in clinical symptoms.

Polymerization in Spongiform Encephalopathies

The model begins by assuming that PrP can exist in two forms: a more compact, more highly soluble alpha form characterized by the predominance of alpha helices, and a more fibrous, more insoluble beta form characterized by the predominance of beta sheet secondary structures. Under normal conditions, the two forms exist in equilibrium, with the equilibrium favoring the alpha form by many orders of magnitude so that the beta form rarely appears at all, and the probability of two beta form proteins encountering each other is virtually nil. If beta forms were to encounter each other, however, there is the possibility that additional bonds would form, thereby stabilizing the molecules in the beta form. Even under these conditions, there is a certain finite probability of depolymerization. However, if the stability is above a certain threshold, the small beta protofibil complex is more likely to encounter yet another beta molecule than to undergo depolymerization. Initial nucleation would require a small number of proximate PrP molecules simultaneously undergoing alpha to beta transition. Additional PrP may be "recruited". Recruitment will lead to a very slow polymerization which will eventually result in cytopathology and cell death. Alternatively, additional PrP molecules may be "induced" to undergo conformational change, which is to say the presence of a nucleation site increases the probability of an alpha to beta transition much as an enzyme might speed the rate of a reaction. If the induction rate were high enough, it would allow for the possibility of a single rare alpha to beta transition to serve as a nucleation site. It would still not allow individual monomers to be infectious as they revert back to the more stable alpha form. This explains the consistent finding that protein denaturing agents such as SDS, disrupt the infectivity of PrP-scrapie.

As fibrils accumulate, one might expect that an increased number of nucleation sites would increase the probability of polymerization leading to an acceleration of disease, with exponential kinetics. Since exponential growth can only occur if the number of nucleation sites increase, it is useful to look at the mechanisms by which that may occur. In one model, each beta form might have multiple binding sites which would produce an elaborate branching pattern of polymerization and an exponential growth rate considerably steeper than second order. In the alternative model, growth would be largely linear (with the possibility of forming weaker lateral interactions). However, linear aggregates would display a finite probability of breaking in half (or otherwise fragmenting), thereby doubling the number of nucleation sites. Breakage could occur if a beta form within a fibril make a rare transition back to the alpha form. These two modes of generating nucleation sites should lead to differences in the electron micrographic appearance and in the kinetics of accumulation.

Table 3 lists some of the variables that may be important in the modeling of prional diseases. Making a number of assumptions, it may be possible to make a crude estimate of the equilibrium constant for the alpha to beta form transition.

First, we will assume that a single protofibril will lead to clinical disease over the course of 10-20 years. The estimate will be based on the probability of developing a protofibril within the first 50 years (1.6 x 109 seconds) of life. The incidence of sporadic CJD is 1 per million (10-6). This gives an estimate of the spontaneous occurrence rate. Using crude estimates of 1013 susceptible neurons, we derive protofibril formation rate estimates of @10-28 neurons/sec. Though the rate is astronomically low, it would result in hundreds of cases worldwide.

The cellular event also represents a combined probability of independent rate events. It is like that the rate of nucleation is much lower than the rate of monomer accumulation. Otherwise, we would not expect to see significant polymeric fibril accumulation. Also, if nucleation and subsequent monomer accumulation both resulted from individual alpha to beta transitions, then the subsequent rate of accumulation would be too slow to result in clinical disease. The problem is resolved if, as hypothesized, nucleation results from a number of simultaneous alpha to beta transitions, while subsequent accumulation results from individual events. If, for instance, we estimate that six independent alpha to beta transitions are needed for nucleation, and we use a crude estimate of 108 PrP molecules per cell, then the probability that an individual PrP will be involved in nucleation is @10-36 PrP/sec, and the probability of an individual alpha to beta transition is @10-6 PrP/sec. The rate could result in a reasonable if not high rate of fibril accumulation. Indeed, it may be possible to do thermodynamic modeling studies to estimate the size of a PrP-beta form oligomer that would be stable enough to favor polymerization over depolymerization.

As noted above, the exponential growth kinetics depend on the means by which additional nucleation sites are generated. If nucleation sites are generated by breakage, two interesting conclusions follow. First, the rate of accumulation must exceed the rate of breakage, otherwise fibrils would depolymerize more rapidly than they polymerize. Second, given a higher rate of monomer accumulation, growth is largely independent of the monomer accumulation rate, but rather it is a function of the breakage, so that doubling the breakage rate will double the growth rate. If each beta form has multiple binding sites from which it can grow, growth will be a function of monomer accumulation and large independent of the breakage rate.

The model described neither requires nor excluded the possibility of alpha to beta conformational change induced by preexisting fibrils or protofibrils. It is possible that induced conformation would be mostly likely to occur during the synthesis of new PrP when bond stabilization may allow fibrils to act as chaperonins for the beta conformation.

Three Scenarios for the Pathogenesis of Spongiform Encephalopathies

I would like to consider three scenarios for the development of spongiform encephalopathies: 1) spontaneous development, 2) genetic predisposition, and 3) infectious.

As noted above, rare spontaneous nucleations followed by polymerization explain the sporadic forms of prional disease including a subset of the Creutzfeldt-Jakob cases.

In the inherited forms of the disease (such as Gerstmann-Straussler, fatal familial insomnia, and certain forms of CJD), the individual encoded a form of PrP in which the equilibrium is shifted towards the beta form, though still greatly favoring the alpha form. This shift increases the probability that beta forms will encounter each other and form a stable complex nucleating further polymerization. Dosage can be related to time course of disease because increased inoculum has more nucleation sites, placing it further along the exponential expansion curve. The polymerization model explains the dominant phenotype of these diseases.

As an infectious agent, (such as kuru, scrapie, bovine spongiform encephalopathy - BSE, and certain forms of CJD), an individual is exposed to a small aggregation of stabilized beta PrP. This serves as the nidus for the recruitment of additional molecular of alpha PrP into the complex. The absence of an eclipse phase would be expected since even cell lysis would not decrease the number of beta form fibrils that could serve as nucleation sites. Transmission would require that PrP protofibrils are capable of gaining access to the vasculature and thence back across the blood brain barrier. Transmission of scrapie may be facilitated by expression in muscle and peripheral nerve cells. Because of the need for vascular access, one might predict that the larger aggregates may not be contagious without fragmenting. Another strong prediction of this model is that enteric uptake in cannibalism (as suggested in the case of kuru) would not be an effective route of transmission since the probability of absorbing undigested, polymerized fibrils and having them subsequently gain vascular and neurological access is negligible. However, burial rituals involving manipulation or consumption of neural tissue would greatly increase the probability of parenteral exposure via penetrating lesions of the skin or mouth. Ancillary proteins may affect the rate of polymerization. Thus germline or somatic mutations may alternatively facilitate or inhibit the development of spongiform encephalopathies. Ancillary proteins may be responsible for differences in lesion distribution between the inherited forms of the disease.

The absence of an immune response stems from the fact that no foreign protein is actually encountered, and because the initial pathogenic process occurs so insidiously that tolerance may be developed even against abnormal forms of PrP. In a similar manner, no antibody response is generated against the polymerized form of HbS hemoglobin. Two other factors may mitigate against a T cell response directed against PrP beta form. The insoluble fibrils may resist antigen presentation. Moreover, since antigen presentation occurs as peptides rather than complete proteins, it is highly unlikely that the peptide would assume the beta configuration which is significantly less thermodynamically stable.

In this model, PrP is not infectious in the typical sense of the word. The protein is not actually being replicated. It is being expressed from the cellular gene and then recruited into the growing beta complex, rather than being degraded or turned over in the typical fashion. This suggests that treatments which suppress the expression of the PrP gene such as code blockers (antisense RNA or triplex DNA agents) would slow the progression of the disease. Reversal of the disease would require agents that favor depolymerization of the complexes. Although even these could not reverse neurological damage. And because it is not likely that depolymerizing agents will completely eliminate protofibrils (priostatic rather than priocidal), such treatment would need to be take on an ongoing basis.

The polymerization model also explains the remarkable finding that PrP knock out animals do not develop spongiform encephalopathy. They do not have any substrate for polymerization. Thus spongiform encephalopathies do not result from loss of function per se, but rather from the acquisition of an additional abnormal (and potentially toxic) structure. Additional useful animal models may be devised by creating transgenic animals with various normal and/or abnormal PrP genes from humans or other mammals. Genes with promoters that up-regulate gene expression would be expected to accelerate the speed of fibril formation. As always, the addition of beta form fibrils should serve as nucleation sites to accelerate the disease process.

This model suggests that it would be possible to develop agents that disrupt further polymerization by binding at the polymerization site. An example might be anti-beta form antibodies raised by inoculation of other mammalian species. Although such agents may prevent further pathology and may even reverse the polymerization process and break up fibrils, they would be expected to have little impact on neurological damage that had already taken place. To that extent, recovery would be little by the degree of neural plasticity. Such agents would also be limited by their ability to gain access to the fibrils. Four factors may limit such access: blood brain barrier, endothelial barriers to the interstitial spaces, membrane barriers to intracellular processes, and the inability to penetrate beyond the such of the insoluble fibrils.

Application of the Polymerization Model to Other Diseases

Diseases of abnormal polymerization may extend beyond prional disease and sickle cell anemia to various other nonneurological amyloidoses and other conditions whose etiology has remained obscure. One virtue of this model is that it can provide the long sought after link between prions and Alzheimer's disease. Rather than being directly causally linked, they resemble each other in terms of their mode of pathogenesis. In this case, clinical abnormality and cell death results not from an abnormal polymerization of PrP, but rather from an abnormal polymerization of amyloid precursor protein (APP).

For Alzheimer's disease histopathological correlates include so called "senile plaques" and neurofibrillary tangles. As in the case of spongiform encephalopathies, there are hereditary and acquired forms of the disease, with strong familial associations.

This model can be used to account for all of the current AD pathological associations, at least in a hand-waving fashion. Aluminum may act either to help stabilize the polymers or conversely the polymers may bind aluminum as a more or less innocent bystander with little impact on the clinical course of the disease. These alternatives are directly testable. Estrogen is predicted to limit the expression of APP, thus decreasing the substrate for polymerization. Conversely, the extra chromosomal dosage of APP (located on chromosome 21) in Down's syndrome, may produce an increase in substrate available for polymerization.

The absence of useful animal models for AD may derive from the exceedingly long kinetics of plaque formation or possibly from genetic differences in human APP that permit the formation of plaque at all. As suggested for prional disease, transgenic animals may provide interesting animal models for AD.

One key difference between CJD and AD is the infectious nature of CJD PrP. APP may be prevented from being contagious by lacking a mechanism for intracerebral transmission. This would occur if APP particular in the insoluble form were unable to cross the blood brain barrier. We might also make the prediction that direct intracerebral inoculation of AD plaques would induce additional plaque formation. This hypothesis could be tested using the transgenic animal model described above.


I would like to acknowledge the helpful comments of Jeremy Moss and Tom Davis.


Spongiform encephalopathy, polymerization, prions, Creutzfeldt-Jakob disease, Alzheimer's disease, recruitment


1 The "eclipse period" refers to the observation that a cell which is disrupted shortly after being infected with a virus, will produce no infectious virus. This is time between old virus has been uncoated and new virus has not yet been replicated.


Cohen, Pan, Huang, Baldwin, Fletterick, and Prusiner "Structural Clues to Prion Replication" Science vol. 264, pp 530-531, April 22, 1994.

Durham, William Coevolution: Genes, Culture, and Human Diversity Stanford University Press, Stanford, California, 1991. (appendix A).

Gajdusek, D. Carleton "Subacute Spongiform Virus Encephalopathies Caused by Unconventional Viruses" in Subacute Pathogens of Plants and Animals: Viroids and Prions Academic Press, 1985.

Mestel, Rosie "Putting Prions to the Test" Science vol 273, pp 184-189, July 12, 1996.

Prusiner, Stanley "The Prion Diseases" Scientific American vol 272 No. 1, pp 48-57, January, 1995.

Wickner, Reed B [URE3] as an Altered URE2 Protein: Evidence for a Prion Analog in Sarccharomyces cerevisiae Science vol 264 pp. 566-569, April 22, 1994.

Table 1

List of Spongiform Encephalopathies


Creutzfeldt-Jakob disease (CJD)
Gerstmann-Straussler syndrome (Gerstmann-Straussler-Scheinker Syndrome)
Fatal familial insomnia


Scrapie (sheep)
Bovine spongiform encephalopathy (BSE)
Transmissible mink encephalopathy
Chronic wasting disease of mule deer and elk
Feline spongiform encephalopathy
spongiform encephalopathies in other animals

Table 2

Properties of Prions

(from Gadjusek, 1985)

I. Biological properties

A. No inflammatory response
B. Chronic progressive pathology (slow infection)
C. No remissions or recoveries: always fatal
D. "Degenerative" histopathology: amyloid plaques, gliosis
E. No inclusion bodies
F. Doubling time of 5.2 days (in hamster brain), similarly slow in mouse brain
G. No eclipse phase
H. No interferon production or interference with interferon production by other viruses
I. No interferon sensitivity

II. Physicochemical properties

A. Ultrafiltration through filters with an average pore size of 20-100 nm
B. Extremely resistant to inactivation by UV or ionizing radiation
1. Suggests very small nucleic acid genome size
C. Resistant to nucleases
D. Protease resistance - conflicting results
1. Sensitivity suggests important protein component
2. Resistant to proteinase K
3. Sensitive to several proteases in the presence of SDS

Table 3

Variables in the Modeling of Prional Disease

Probability of alpha to beta transition
Probability of nucleation
Probability of beta to alpha transition in the nonpolymerized state
Rate of monomer accumulation
Probability of induced conformational change (alpha to beta)
Probability of fibril branching (number of monomer binding sites per beta molecule)
Stability of oligomers (rate of depolymerization)
Probability of fibril breakage (kinetics may vary as a function of size)
Probability of beta to alpha transition in the polymerized state (closely related to: stability of monomers, and the probability of fibril breakage)
Probability that nucleation will lead to clinical disease
Number of susceptible cells
Number of PrP molecules per cell

Other possible factors include:
transition from membrane to nonmembrane state

Figure 1

Conversion of PrP alpha to beta

Figure 2

Polymerization of PrP

Figure 3

Exponential Expansion Curve

Figure 4

Rate Constants Involved in Nucleation

More coming...

| Return to top of page | Return to Bob's Home Page |

Last modified: March 14, 1997