The hypothesis postulated that the concentration of each protein in the crowded cytosol (Fulton, 1982) has been fine-tuned over evolutionary time to approach the limits of its solubility, and yet avoid aggregating "self" with "self." Various molecular chaperones, which include the heat-shock proteins (HSPs), would quickly reverse any incipient aggregation (Fig. 1a). 

 (a) "Self" proteins (small circles, ovals and squares) approach the limits of their solubility in the crowded cytosol (area between the perimeters of the large circle representing the cell wall and the medium-sized pink circle representing the cell nucleus). HSPs (not shown) in "normal mode" act as molecular chaperones to disassemble incipient aggregates. 

 (b) "Not-self" viral proteins (small squares) more readily cross the aggregation threshold than "self" host proteins (small circles and ovals), and form larger aggregates. These aggregates constitute self/not-self discriminatory signals, resulting in a switch of HSPs (not shown) to peptide presentation mode, and the triggering of various intracellular alarms, including the upregulation of MHC class I protein expression. Aggregates are subject to proteosomal processing to form peptides which, by way of HSP-peptide intermediates, are displayed at the cell surface as MHC-peptide complexes.

 On the other hand, virus proteins would tend to overstep the aggregation threshold. This critical self/not-self discrimination event, would trip various intracellular alarms resulting in pyrexia, proteosomal degradation of aggregates to peptides, a switch of HSPs to peptide-binding mode, up-regulation of MHC protein expression, and display of MHC-viral peptide complexes at the cell surface (Fig. 1b). Thus, the cell would become a target for attack by cytotoxic T cells, the host repertoire of which would include cells able to recognize peptides from intracellular "self" proteins (Forsdyke, 1995a; Sandberg et al., 2000).

     The HSP mode switch could require either a change from the synthesis of constitutive HSPs to the synthesis of inducible HSPs (Menoret et al., 1995), or a change in properties of HSP molecules (Spiess et al., 1999). To prevent inadvertent aggregation of self- proteins following the tripping of the alarm, their synthesis would be down-regulated (e.g. the corresponding genes might have accepted mutations inhibiting the binding of RNA polymerase under alarm conditions); this would decrease the concentrations of self-proteins (a phenomenon perhaps of consequence for understanding the evolution of genetic dominance; Forsdyke, 1994a; Hurst & Randerson, 2000). Similarly, viruses would tend to accept mutations that decrease the chance of their proteins overstepping the aggregation threshold. This problem, not addressed in the initial hypothesis (Forsdyke, 1995a,b), will be considered in Section 8.

 For simplicity biological species are here considered as composed of nucleic acids and proteins, and other molecules are disregarded. Members of a biological species have many similar anatomical and physiological features and, accordingly, they have similar proteins, which are responsible for this phenotype. Some of these proteins are also found in members of other biological species, where they have similar functions. Often the name of a protein indicates its overt function. Thus the enzyme lactate dehydrogenase oxidises lactate to pyruvate, a function required in many taxa. The performance of this function places the enzyme under strict stereochemical constraint. Especially within a species, but often also between species, there is high evolutionary conservation of amino acids in certain regions of the molecule (e.g. the catalytic centre and regions which support the integrity of the catalytic centre; Fersht, 1998). Organisms with mutations in these regions are usually negatively selected ("purifying selection") in the course of evolution.

    Apart from their overt specific functions, most protein molecules have other specific functions. For example, if it is advantageous for a protein to be localized within its cell, or to form dimers, then amino acids involved in these specific intermolecular reactions will be evolutionarily conserved.

    Thus, the protein molecules of a biological species have regions that are conserved among species members, and regions that are less conserved among species members (i.e. polymorphism; Harris, 1966; Lewontin & Hubby, 1966; Forsdyke, 2001). Some less conserved regions could play a non-specific structural role, so that amino acids in these regions would act more by virtue of being neutral "placeholders" than by virtue of their specific chemical characteristics. 

     Alternatively, the less conserved regions could have no overt role, either specific or non-specific, in the regular day-to-day life of the organism (i.e. they would appear completely neutral in that their deletion would not affect overt function). Such neutral and completely neutral regions tend to be solvent accessible and surface located (Goldman et al., 1998; Bustamente et al., 2000; Alvarez-Valin et al., 2000). This indicates a potential to engage in intermolecular interactions. The greater the range of the polymorphism, the greater the range of this potential among members of a biological species.

     Indeed, the assumption of neutrality may not be valid. Polymorphism involving less conserved regions may have a specific function. Polymorphism within cells might serve the organism in much the same way that MHC polymorphism at the cell surface serves the organism. The fact of non-conservation implies that, like MHC polymorphism, the specificity is likely to be individual-specific, and is not shared by many other members of the species. Although this polymorphism might at one time have been neutral, it is argued here that such territory would soon have been trespassed upon to serve adaptive needs.

 The biological effect of a molecule usually depends on its functional potential and its concentration. By accepting appropriate mutations the corresponding gene can "tune" or "fine-tune" either or both of these properties, and hence modify the biological effect (change the phenotype). For present purposes, changes in functional potential will be emphasized. 

     Consider a biological species in which protein molecule A would increase the genetic fitness of species members if it interacted in the cytoplasm with protein molecule B. For example, localization of one enzyme in a metabolic pathway close to another enzyme (for which its product provides a substrate) might improve the efficiency of the pathway (Scott & Forsdyke, 1978; Kisters-Woike et al., 2000). Let there be a germline mutation resulting in a change (A to A'), such that a complex A'B can now be formed. 

     Having arranged the structure of molecule A such that it can now react with molecule B, the "hand of evolution" has achieved a significant advance. Such evolutionary "tuning" should occur quite rapidly because of the immediate selective advantage it would confer. Thus, over time species members with A' would be positively selected and would replaced species members with A.

A'  +  C  = A'C       +        Weak

A'  +  D = A'D       +        Weak

A'  +  E = A'E        +        Weak

A'  +  F =

A'  +  G  =

 .         .

 .         .

 .         .

A'  +   Z  =    

    However, the crowded cytoplasm contains many thousands of protein types (e.g. A-Z; Fig. 2). It is probable that a small subset of proteins (C, D, E) with no functional interest in A, would, by virtue of the primary change in A, by chance, now become able to interact weakly with it, to form complexes A'C, A'D and A'E. These unwanted reactions, albeit weak, might unduly immobilize the interacting molecules, to the marginal detriment of their primary overt functions. The advantage of forming A'B complexes would so far outweigh the disadvantages of forming A'C, A'D and A'E complexes, that the differential survival of individuals with A' would not be threatened. Nevertheless, it would be expect that, concomitant with and following the fixation of the gene encoding A' in the species, there would be an on-going random process of mutational counter-adaptation by genes encoding C, D and E to decrease their affinity for A', provided this could be achieved without compromising their primary functions (i.e. the mutations would affect "neutral," potentially polymorphic, residues).

     Such random "fine-tuning" (C to C', D to D', E to E') would be likely to occur in different individuals at different times, and would confer virtually insignificant adaptive advantages on those individuals. Thus individuals with constitutions A'B'CD, A'BC'D, A'BCD' would be favoured only slightly over the general population of individuals of constitution A'BCD (i.e. the adaptations would initially be "near neutral"; Ohta, 1995).

     This biological discontinuity (polymorphism) would appear relatively stable. Over many generations, however, genetic recombination would create, by chance, alliances of these fine-tuned proteins so that individuals of constitution A'B'C'D' would eventually emerge, and would be marginally advantaged. The proportion of their offspring would increase in the population, reducing the extent of the biological discontinuity.

    While this slow proximate fine-tuning was in progress, a seemingly unending chain of further fine-tunings would be occurring. In the same way that the initial change A to A' provoked counter-adaptations in C, D and E, so the changed forms C', D' and E', should each provoke corresponding counter-adaptations in individuals containing them. 

• C' might, by chance, interact with F, G and H.
    
  
• D' might, by chance, interact with I, J, and K.
    
  
• E' might, by chance, interact with L, M, and N.

 These molecules (F-N) whose functions would themselves now be marginally compromised, would in-turn need to counteradapt to F'-N'. This, in turn , would provoke further counter-adaptations, etc., etc..

    Thus, at any time-point there would be an intracellular network of proteins the potentially polymorphic regions of which the "hand of evolution" would be fine-tuning in order that the proteins not react, however weakly, with others. New positive mutations of type A to A' would from time-to-time set off new chains of adaptations and counter-adaptations. 

     However, if the environment remained relatively constant the organism would asymptotically approach a state of perfect adaptation. Then the frequency of the positive mutations would decrease, and the initial round of proximate counter-adaptations (C', D' and E') would approach completion. Individuals of type A'B'C'D' would come to dominate the species which would loose this particular biological discontinuity. 

     In this way many natural species would be fine-tuned and appear well adapted to their environments. Some plants in this state we might designate as "weeds." They grow rapidly and form seed even in nutritionally unfavourable conditions. On the other hand, domestic plants, relative late-comers on the evolutionary scene, usually grow more slowly, even in nutritionally favourable conditions.

     Could this intrinsic tendency towards polymorphism of certain surface-located amino acid residues, which may engage in weak, random, and apparently non-functional, interactions, have been turned to the further advantage of the organism? To approach this question, we first reconsider previous work (Forsdyke, 1999) to show how, at the level of the individual, polymorphism might act to increase the effective antigenicity of mutations in oncogenes (Section 5). It will then be shown that the same principle might apply in the enhancement of the effective antigenicity of virus proteins. The designation by the host of a virus protein as "not-self," may involve both quantitative factors (i.e. its concentration exceeding a solubility threshold, which the virus might avoid by appropriate mutation), and qualitative factors (the recruitment of various polymorphic host proteins into protein aggregates, which the virus might not avoid by mutation; see Sections 6-8).

 Consider normal cells that have become potentially cancerous because of a mutation in a single oncogene. In practice, a number of such mutational switches involving different oncogenes might need to be "thrown" for cells to manifest the full cancer phenotype (Klein & Klein, 1985). Since oncogenes are a distinct subset of genes common to all members of a species, it is attractive to suppose that the primary mutation in an oncogene (A to A') would be recognized by an intracellular self/not-self discrimination system (involving differential aggregation of A'; Forsdyke, 1994b, 1995a,b) so that a peptide from A would be displayed in association with MHC proteins at the surface of the cancer cell. This MHC-peptide complex would be recognized by cytotoxic T-cells, which would then multiply and destroy the potential cancer cell or its progeny before they could form a tumour. This predicts cancer-specific antigens.

     However, the cancer antigens recognized by cytotoxic T cells tend to be neither cancer-specific, nor cancer type-specific. Rather they are specific to the individual with cancer. Antigens specific for cancers in general, or for particular cancer types, are hard to find and do not appear to serve as useful primary targets for immune attack . 

• Thus, a cancer attributed to a mutation in an oncogene encoding A in one individual (A to A') may be attacked immunologically with respect to a particular set of antigens (corresponding to genes encoding B, C and D). 
  
  
• In another individual the attack may be with respect to another set of antigens (corresponding to genes encoding E, F and G). 
  
  
• In a third individual other antigens would be involved (H, I, and J).

This suggests antigen polymorphism since, if (say) B were present as B (rather than as B') in an individual whose cancer was being attacked with respect to antigens E, F and G, then it is hard to see how B would be excluded from the attack.

     Yet, when cancer antigens were purified, many were found to correspond to heat-shock proteins, which are not polymorphic. Remarkably, the heat-shock proteins could confer individual-specific protective immunity against the cancer of origin. Srivastava et al., (1998) noted:

 The specificity of immunogenicity was found to derive from the association with the HSPs of peptide fragments of the products of non-oncogenes. In view of the widespread evolutionary conservation of HSP genes, it was suggested that HSPs might have constituted a prototypic system for the processing of peptides, which evolved into the contemporary MHC-peptide system. The modern role of HSPs might be to transfer peptides to MHC molecules, either within the cell of peptide origin for presentation by that cell to cytotoxic T cells, which would destroy the cell and proliferate (Vanbuskirk et al., 1989), or within other cells (dendritic cells) for presentation by those cells (Binder et al., 2000; Singh-Jasuja et al., 2000). By transferring the peptide signal beyond the cell of origin to a "professional" antigen-presenting cell, HSPs would serve to amplify the signal, causing specific cytotoxic T cells to proliferate further.

     To explain the apparently random recruitment of non-oncogenes into the antigenic repertoire of the cancer cell, Srivastava proposed that the cancer phenotype predisposes to mutation (Lengauer et al., 1998; Tomlinson & Bodmer, 1999) genes which are unlikely to be related functionally to the genes whose primary mutation results in cancer. These non-oncogenes, by virtue of the mutations they contained, would not affect oncogenicity per se, but would mark cancer cells as foreign for recognition by an immune system considered to have been educated to distinguish "not-self", from "self" (Srivastava, 1993, 1996). However, it was argued (Forsdyke, 1999), that this formulation is incorrect both on theoretical grounds and on experimental grounds, namely, that the individual-specific non-oncogene antigens derived from cancer cells with a mutation in an oncogene are usually not mutated (Srivastava, 1996; Srivastava et al., 1998; Zeh et al., 1999).

    The requirement that genes must mutate to "not-self" in order to be able to trigger immune defences would not be necessary if 

• (i) the peptide products of many normal genes ("self") do not eliminate, in advance, T cells with the potential to react against them (Forsdyke, 1995a, 1999; Sandberg et al., 2000), and
   
  
• (ii) a mutated oncogene could somehow confer antigenicity on such products in an individual-specific manner. 

Thus, instead of describing potential individual-specific antigenic phenotypes as 

• A'B'C'D',
   
  

• A'E'F'G',
  
  

• A'H'I'K',

This amends an earlier formulation (Forsdyke, 1999) where differences between individuals in the original mutation (A to A', A to A'', A to A''') were invoked to produce antigenic phenotypes A'BCD, A''EFG, A'''HIK (i.e. the mutational change conferring oncogenicity could be either A', A'', or A''', and these different forms of A would recruit different sets of proteins for targeting by cytotoxic T cells).

     Following the argument of Section 4, the same mutation A' in different individuals should suffice to recruit sets of polymorphic proteins (from among B-K) into the respective antigenic phenotypes. The individual-specific nature of the antigenicity would appear readily explicable in terms of the natural polymorphic tendency of proteins with respect to amino acid residues which are not under strict functional constraint as related either to their generally-recognized named role (e.g. lactic dehydrogenase), or to their ability to engage in specific intermolecular reactions related to this function. It should be noted that the initial hypotheses of Srivastava (1993) and of Forsdyke (1999), both postulate an apparent polymorphism generated in somatic time in response to a mutated oncogene, whereas the present version postulates a real preexisting polymorphism generated over evolutionary time.

 HSPs may be involved in the presentation, not only of tumour-derived antigens, but also of antigens from virally-infected cells (Ciupitu et al., 1998). A protective role of polymorphic intracellular proteins should be very important in virus infections. A viral protein (A) can be considered in the same way as a mutant oncogene, so that host proteins B, C and D would be recruited into the antigenic repertoire of the infected cell. A mutation in a viral protein A to A' which might prevent it interacting with host proteins B, C and D, might be of little value when the virus infected its next host. In this host these proteins might exist as polymorphic forms B', C' and D', which might not interact with A. Thus, in this host the polymorphic proteins recruited into the antigenic repertoire might be E, F and G. In another host H, I and J might be recruited.

     In contrast, the extracellular recognition of a not-self viral protein ("antigen") involves a normal resident protein ("antibody"), which interacts with a discrete part ("epitope") of the intact antigen. In this case, the antibody protein (immunoglobulin) has evolved for this specific purpose and, as far as we are aware, has no other primary purpose. Its polymorphism has been developed to the extreme by the evolution of variable region genes with the potential to diversify further during somatic time (Tonegawa, 1988). 

     In the case of intracellular proteins there is no known dedicated molecular species of an antibody nature. So proteins with normal functions in the economy of the cells would be co-opted for the aggregation function. Host proteins B, C and D can be regarded as intracellular "antibodies" of very low specificity, which happen to have sufficient complementarity with A to form a complex in the crowded cytosol (an environment highly conducive to intermolecular interactions; Forsdyke, 1995a).

     Just as extracellular antigen-antibody complexes, having labelled intruders as "not-self," trigger amplifying inflammatory responses, so multimolecular intracellular aggregates (A'B, A'C or A'D) should suffice to initiate intracellular "inflammation." This would lead to MHC presentation of their peptides (Townsend et al., 1990), and the release of chemokines such as G0S19/MIP1" (Cook et al., 1995; Heximer et al., 1998). 

    To the extent that intracellular "antibodies," MHC proteins and chemokines are not polymorphic, successful viral counter-adaptations would be expected (Davis-Poynter & Farrell, 1996), and there would exist the necessary elements for an arms race between pathogen and prey, leading to positive Darwinian selection of variants of both (Forsdyke, 1995c, 1996). To the extent that the proteins are polymorphic, viral counter-adaptations would less likely be successful.

     It should be noted that polymorphic MHC proteins usually display differential peptide binding (Townsend et al., 1990), with the result that one individual may display a peptide which another (with a different MHC haplotype) cannot. It is not envisioned here that this cause of the differential presentation of peptides from self-antigens plays as important a role as the intracellular "antibodies." Consistent with this, species with a limited number of MHC genes can remain healthy (Kaufman, 2000).

 Solubility, a fundamental protein property, is at the centre of the differential aggregation hypothesis of self/not-self recognition (Forsdyke, 1995a, b). As their limit of solubility is approached, molecules of resident cellular protein types would tend to aggregate. The aggregations would be protein-specific, in that proteins aggregate like-with-like, leaving other protein types unaggregated (Lauffer, 1975; Leikin & Parsegian, 1994). The growing aggregates would tend to become insoluble forming precipitates or cellular "inclusion bodies." This phenomenon has long been exploited by biochemists to purify different protein types from mixtures by differential precipitation (e.g. with polyethylene glycol). The specificity appears similar to that required for the formation of pure crystals from molecular mixtures, a process which seems to require structural regularities, and perhaps molecule-specific vibrations, or resonances (Israelachvili & Wennerstrm, 1996).

 
    Recently, aggregation has been associated with cell dysfunction in various neurological diseases, which include the prion diseases where a conformational isomer with a tendency to aggregate, when transferred to a new organism, triggers a similar conformational change with ensuing aggregation of the corresponding host protein (i. e. a prion-precursor molecule without the tendency to aggregate). This can lead to insolubilization (precipitation). Experimentally it is found that, in a cytosol containing more than one type of prion-precursor molecule, addition of a few prion molecules of one type causes conversion and aggregation only of the corresponding normal type. The other types remain unaggregated (Santoso et al., 2000). Thus one type of prion can act as a "seed", catalysing the formation of a specific prion aggregate. The crowded cytosol can be viewed as poised on the verge of concentration "criticality" (Fulton, 1982; Forsdyke, 1995a, b). The unremitting activity of molecular chaperones, which include members of the heat-shock protein family, can usually ensure that protein conformations are kept in soluble-mode, so that the criticality threshold is not crossed (Fig. 1a; Forsdyke, 1999; Newnam et al., 1999; Warrick et al. 1999).

 
    However, the prion phenomenon provides an instructive model for a process by which a normal resident intracellular protein species ("self") can serve as a recognition device for an extrinsic protein ("not self") with which it shares some property. The recognition is expressed as an aggregation which appears ultimately overwhelming in the case of prion diseases, and is correlated with adverse consequences. It is suggested here that, in principle, the process might be turned to the advantage of the organism, with the shared property in this case being weak antigen-antibody-like complementarity between a resident intracellular protein and a foreign ("not-self") protein originating from a virus or a mutated "self" protein (oncogene).

 When the sequences of two alleles randomly chosen from the human population are compared, there are approximately two variations per kb. Thus, in a large population, many 1 kb sequences would be expected to differ from the canonical sequence (Rowen et al., 1997; Sunyeav et al., 2000). Polymorphism is widespread, but is the extent of polymorphism sufficient for the role considered here? For simplicity we have considered only one viral gene (A), and the recruitment of three host proteins (e.g. B, C and D). However, viruses generally synthesize many different proteins at different stages of their life cycle, each with the potential to recruit a distinct subset of host proteins. Furthermore, the recruitment of only one host protein should suffice to trigger an antiviral response in that host, while, because of polymorphism, the same protein might not be recruited in a subsequent host (which might present peptides from another polymorphic protein). 

     Thus, although a virus might accept mutations which change its proteins quantitatively (i.e. so that they do not exceed a concentration threshold beyond which self-aggregation would occur), its acceptance of mutations which change its proteins qualitatively (i.e. change its protein sequences to decrease their ability to recruit host proteins into aggregates) might confer no adaptive advantage if such host proteins were polymorphic. When confronted with such polymorphism, to evade detection by the host more profound decreases in viral protein quantities would be required.

 Unlike viruses, an oncogene activated by mutation is unlikely to mutate further, so confusing the defences of the host cell. Indeed, it would be beneficial for one or more self-proteins, to the extent that their primary functions were not impaired, to accept mutations which promote their interactions with (and hence potential coaggregation with) such mutated oncogene proteins. If primary oncogene mutations have in the past diminished reproductive success, this inherited adaptation of non-oncogenic self-proteins would tend to make cancer a disease of post-reproductive life. 

     Since such adaptation of self-proteins could confer an advantage, the corresponding alleles would be positively selected and monomorphism should be developed more than expected under a strictly neutral mutation hypothesis. Modelling studies tend to support this view, but it is noted that "what the selection might be remains unclear" (Eyre-Walker, 1999; Akashi, 1999). Currently, most models do not distinguish between the many potential, sometimes competing, non-neutral pressures on genomes, which include GC-pressure, purine-loading pressure, fold-pressure, and the "polymorphic pressure" considered here (Forsdyke & Mortimer, 2000).

 A virus appears in a cell both as nucleic acid and as protein. While beyond the scope of the present discussion, it should be noted that the formation of double-stranded RNAs between a virus RNA and host RNAs (for refs. see Forsdyke & Mortimer, 2000; Lao & Forsdyke, 2000; Cristillo et al., 2001), provides another potential self/not-self discriminatory signal which might supplement or synergize with the postulated differential protein aggregation signal (Forsdyke, 2000; (Click Here) and see End Note (below)).

AKASHI, H. (1999). Within- and between-species DNA sequence variation and the footprint of natural selection. Gene 238, 39-51.

ALVAREZ-VALIN, F., TORT, J. F. & BERNARDI, G. (2000). Nonrandom spatial distribution of synonymous substitutions in the GP63 gene from Leishmania. Genetics 155, 1683-1692.

BINDER, R. J., HAN, D. K. & SRIVASTAVA, P. K. (2000). CD91: a receptor for heat shock protein gp96. Nature Immun. 1, 151-155.

BUSTAMENTE, C. D., TOWNSEND, J. P. & HARTL, D. L. (2000). Solvent accessibility and purifying selection within proteins of Escherichia coli and Salmonella enterica. Mol. Biol. Evol. 17, 301-308.

CIUPITU, A. M., PETERSSON, M., ODONNELL, C. L., WILLIAMS, D. K., JINDAL, S., KIESSLING, R. & WELSH, R. M. (1998). Immunization with a lymphocytic choriomeningitis virus peptide mixed with heat shock protein 70 results in a protective antiviral immunity and specific cytotoxic lymphocytes. J. Exp. Med. 187, 685-691.

COOK. D. N., BECK, M. A., COFFMAN, T. M., KIRBY, S. L., SHERIDAN, J. F., PRAGNELL, I. B. & SMITHIES, O. (1995). Requirement of MIP1" for an inflammatory response to viral infection. Science 269, 1583-1585.

CRISTILLO, A. D., MORTIMER, J. R., BARRETTE, I. H., LILLICRAP, T. P. & FORSDYKE, D. R. (2001). Double-stranded RNA as a not-self alarm signal: to evade, most viruses purine-load their RNAs, but some (HTLV-1, EBV) pyrimidine-load. J. theor. Biol. 208, 475-491.

DAVIS-POYNTER, N. J. & FARRELL, H. E. (1996). Masters of deception: a review of herpesvirus immune evasion strategies. Immun. Cell Biol. 74, 513-522.

EYRE-WALKER, A. (1999). Evidence of selection on silent site base composition in mammals: potential implications for the evolution of isochores and junk DNA. Genetics 152, 675-683.

FERSHT, A. (1998). Structure and Mechanism in Protein Science. New York: W. H. Freeman.

FORSDYKE, D. R. (1991). Early evolution of MHC polymorphism. J. theor. Biol. 150, 451-456.

FORSDYKE, D. R. (1994a). The heat-shock response and the molecular basis of genetic dominance. J. theor. Biol. 167, 1-5.

FORSDYKE, D. R. (1994b). Relationship of X chromosome dosage compensation to intracellular self/not-self discrimination: a resolution of Mullers paradox? J. theor. Biol. 167, 7-12.

FORSDYKE, D. R. (1995a). Entropy-driven protein self-aggregation as the basis for self/not-self discrimination in the crowded cytosol. J. Biol. Sys. 3, 273-287.

FORSDYKE, D. R. (1995b). Fine tuning of intracellular protein concentrations, a collective protein function involved in aneuploid lethality, sex-determination and speciation? J. theor. Biol. 172, 335-345.

FORSDYKE, D. R. (1995c). Conservation of stem-loop potential in introns of snake venom phospholipase A2 genes: an application of FORS-D analysis. Mol. Biol. Evol. 12, 1157-1165.

FORSDYKE, D. R. (1996). Stem-loop potential in MHC genes: a new way of evaluating positive Darwinian selection. Immunogenetics 43, 182-189.

FORSDYKE, D. R. (1999). Heat shock proteins as mediators of aggregation-induced danger signals: implications of the slow evolutionary fine-tuning of sequences for the antigenicity of cancer cells. Cell Stress Chaperones 4, 205-210.

FORSDYKE, D. R. (2000). Double-stranded RNA and/or heat-shock as initiators of chaperone mode switches in diseases associated with protein aggregation. Cell Stress Chaperones 5, 375-376.

FORSDYKE, D. R. (2001). Functional constraint and molecular evolution. Encyclopedia of Life Sciences. Macmillan Reference Ltd., London.

FORSDYKE, D. R. & MORTIMER, J. R. (2000). Chargaffs legacy. Gene 261, 127-137.

FULTON, A. (1982) How crowded is the cytoplasm? Cell 30, 345-347.

GOLDMAN, N., THORNE, J. L. & JONES, D. T. (1998). Accessing the impact of secondary structure and solvent accessibility on protein evolution. Genetics 149, 445- 458.

HAFNER, M. S., & NADLER, S. A. (1988). Phylogenetic trees support the coevolution of parasites and their hosts. Nature 332, 258-259.

HARRIS, H. (1966). Enzyme polymorphisms in man. Proc. R. Soc. Lond. B. 164, 298-310.

HEXIMER, S. P., ERNST, B. D., RUSSELL, L. & FORSDYKE, D. R. (1998). The normal copy of the G0S19-3-associated, CpG island-containing, upstream sequence is downstream of G0S19-2/MIP1a in association with a TRE17 oncogene. DNA Cell
Biol. 17, 61-68.

HURST, L. D. & RANDERSON, J. P. (2000). Dosage, deletions and dominance: simple models of the evolution of gene expression. J. theor. Biol. 205, 641-647.

ISRAELACHVILI, J. & WENNERSTR_ M, H. (1996). Role of hydration and water structure in biological and colloidal interactions. Nature 379, 219-224.

KAUFMAN, J. (2000). The simple chicken major histocompatibility complex: life and death in the face of pathogens and vaccines. Phil. Trans. R. Soc. Lond. B. 355, 1077-1084.

KIMURA, M. (1968). Evolutionary rate at the molecular level. Nature 217, 624-626.

KING, J. L. & JUKES, T. H. (1969). Non-Darwinian evolution. Science 164, 788-798.

KISTERS-WOIKE, B., VANGIERDEGOM, C. & MULLER-HILL, B. (2000). On the conservation of protein sequences in evolution. Trends. Biochem. Sci. 25, 419-421.

KLEIN, G. & KLEIN, E. (1985). Evolution of tumours and the impact of molecular oncology. Nature 315, 190-195.

LAO, P. J. & FORSDYKE, D. R. (2000). Thermophilic bacteria strictly obey Szybalskis transcription direction rule and politely purine-load RNAs with both adenine and guanine. Genome Res. 10, 228-236.

LAUFFER. M. A. (1975). Entropy-Driven Processes in Biology. New York: Springer-Verlag.

LEIKIN, S. & PARSEGIAN, V. A. (1994). Temperature-induced complementarity as a mechanism for biomolecular assembly. Proteins 19, 73-76.

LENGAUER, C., KINZLER, K. W. & VOGELSTEIN, B. (1998). Genetic instabilities in human cancers. Nature 396, 643-649.

LEWONTIN, R. C. & HUBBY, J. L. (1966). A molecular approach to the study of genic heterozygosity in natural populations. II. Amount of variation and degree of heterozygosity in natural populations of Drosophila pseudoobscura. Genetics 54, 595-
609.

MENORET, A., PATRY, Y., BURG, C. & LE PENDU, J. (1995). Co-segregation of tumor immunogenicity with expression of inducible but not constitutive hsp70 in rat colon carcinomas. J. Immun. 155, 740-747.

NEWNAM, G. P., WEGRZYN, R. D., LINDQUIST, S. L. & CHERNOFF, Y. O. (1999). Antagonistic interactions between yeast chaperones Hsp104 and Hsp70 in prion curing. Mol. Cell. Biol. 19, 1325-1333.

OBST, R., NETUSCHIL, N., KLOPFER, K., STEVANOVI, S. & RAMMENSEE, H-G. (2000). The role of peptides in T cell alloreactivity is determined by self-major histocompatability complex molecules. J. Exp. Med. 191, 805-812.

OHTA, T. (1995). Synonymous and nonsynonymous substitutions in mammalian genes and the nearly neutral theory. J. Mol. Evol. 40, 56-63.

ROWEN, L., MAHAIRAS, G. & HOOD, L. (1997). Sequencing the human genome. Science 278, 605-607.

SANDBERG, J. K., FRANKSSON, L., SUNDBACK, J., MICHAELSSON, J., PETERSSON, M., ACHOUR, A., WALLIN, R. P. A., SHERMAN, N.E., BERGMAN, T., JORNVALL, H., HUNT, D. F., KIESSLING, R. & KARRE, K. (2000). T cell tolerance based on avidity thresholds rather than complete deletion allows maintenance of maximal repertoire diversity. J. Immun. 165, 25-33.

SANTOSO, A., CHIEN, P., OSHEROVICH, L.Z. & WEISSMAN, J. S. (2000). Molecular basis of a yeast prion species barrier. Cell 100, 277-288.

SCOTT, F. W. & FORSDYKE, D. R. (1978). The rate of deoxyribonucleic acid synthesis by cultured Chinese Hamster ovary cells. Biochem. J. 170, 545-549.

SINGH-JASUJA, H., SCHERER, H.U., HILF, N., ARNOLD-SCHILD, D., RAMMENSEE, H.-G., TOES, R.E.M. & SCHILD, H. (2000). The heat shock protein gp96 induces maturation of dendritic cells and down-regulation of the gp96 receptor.  Eur.J. Immunol.  30, 2211-2215.

SPIESS, C., BELL, A. & EHRMANN, M. (1999). A temperature-dependent switch from chaperone to protease in a widely conserved heat-shock protein. Cell 97, 339-347.

SRIVASTAVA, P. K. (1993). Peptide-binding heat shock proteins in the endoplasmic reticulum: role in immune response to cancer and in antigen presentation. Adv. Cancer Res. 62, 153-177.

SRIVASTAVA, P. K. (1996). Do human cancers express shared protective antigens? Or the necessity of remembrance of things past. Sem. Immunol. 8, 295-302.

SRIVASTAVA, P. K., MENORET, A., BASU, S., BINDER, R. T. & MCQUADE, K. L. (1998). Heat shock proteins come of age: primitive functions acquire new roles in an adaptive world. Immunity 8, 657-665.

SUNYEAV, S. R., LATHE, W. C., RAMENSKY, V. E. & BORK, P. (2000). SNP frequencies in human genes: an excess of rare alleles and differing modes of selection. Trends. Genet. 16, 335-337.

TOMLINSON, I. & BODMER, W. (1999). Selection, the mutation rate and cancer: ensuring that the tail does not wag the dog. Nature Med. 5, 11-12.

TONEGAWA, S. (1988). Somatic generation of immune diversity. Biosci.Rep. 8, 3-26.

TOWNSEND, A., ELLIOTT, T., CERUNDULO, V., FOSTER, L., BARBER, B. & TSE, A. (1990). Assembly of MHC class I molecules analyzed in vitro. Cell 62, 285-295.

WARRICK, J. M., CHAN, H.Y.E., GRAY-BOARD, G. L., CHAI, Y., PAULSON, H.L. & BONINI, N. (1999). Suppression of polyglutamine-mediated neurodegeneration in Drosophila by the molecular chaperone HSP70. Nature Genetics 23, 425-428.

VANBUSKIRK, A., CRUMP, B. L., MARGOLIASH, E. & PIERCE, S. K. (1989). A peptide-binding protein having a role of antigen presentation is a member of the HSP70 heat shock family. J. Exp. Med. 170, 1799-1809.

ZEH. H. J., PERRY-LALLEY, D., DUDLEY, M. E., ROSENBERG, S. & YANG, J. C. (1999). High avidity CTLs for two self-antigens demonstrate superior in vitro and in vivo antitumor efficiency. J. Immunol. 162, 989-994.