Hepatitis C (HCV), HBV and HIV are major causes of premature death and morbidity globally. These infections are frequently life-long; Hepatitis viruses may result in progressive injury to the liver and cirrhosis, and death from liver failure, or hepatocellular carcinoma, while HIV causes progressive immune depletion and death from the acquired immunodeficiency syndrome (AIDS). Together, these infections cause millions of premature deaths annually, predominantly in "developing" countries. Other viruses replicating via RNA intermediaries cause similar morbidity among domestic and wild animal populations. While education, public health measures and vaccination (for HBV) have resulted in significant progress in disease control, therapy of established viral infection remains unsatisfactory.

RNA viruses and retroviruses replicate, at least in part, by RNA polymerases (RNA_pol), enzymes that lack either fidelity or proofreading function 76. During replication of hepatitis C HCV or HIV each new genome differs from the parental template by up to ten nucleotides 61 due to RNA_pol infidelity that introduces errors at ~1 × 10^-5 mutations / base RNA synthesised.

Viruses replicate by copying antigenomic intermediate templates and hence obey exponential growth kinetics, such that [RNA]_t = [RNA]_(t-1)e^k, where [RNA]_t is virus concentration at time (t) and k a growth constant. However, because of RNA_pol infidelity, wild-type (wt) virus will accumulate at [RNA_wt]_t = [RNA_wt]_(t-1)•(1-ρ)•K₁ and variant forms (mt) at [RNA_mt]_t ≈ ([RNA_wt]_(t-1)•ρ + [RNA_mt]_(t-1))•K₁, where ρ is the probability of mutation during replication and K₁ = e^k. Therefore, while wild-type virus predominates early, replication (and intracellular accumulation) of variant virus and viral proteins will accelerate (in a ratio of ([RNA_wt]_(t-1)•ρ + [RNA_mt]_(t-1))/ [RNA_wt]_(t-1)•(1-ρ) compared to wild type) and variant viral RNAs will rapidly predominate (Figure 1). Mutations progressively accumulate in RNA viruses 17 and ultimately variant RNAs and proteins, if variant RNAs are translated, will become dominant. It is also likely some variant viral proteins will resist cellular trafficking, further accelerating the intracellular accumulation of variant forms relative to wild type.

Hepatitis C, HIV, and HBV and other viruses, have broadly similar kinetics (Figure 2); initial high level viral replication that rapidly declines to relatively constant low-level viraemia 1112, typically 2–3 logs lower than at peak, for prolonged periods, a kinetic profile attributed to "immune control" 12. However, immune control is a conceptually problematic explanation for the initial decline in viral load; For example; why would potent host responses (of whatever type; humoral, cell mediated or intracellular immunity, or any combination thereof), having reduced viral load and antigenic diversity by a factor of 10^2–3 within days, falter once less than 1% of virus remains?

The stability of RNA viral quasispecies poses a major problem: During viral replication the copied genome may either identical to or a variant of parental template (Figure. 1). The probability (ρ) of a mutation occurring during replication is a function of polymerase fidelity; During one replication cycle ρ = (1-(1-M_μ)ⁿ), where (M_μ) is mutation rate and (n) genome size. Hepatitis C (a ~9200 bp RNA virus) RNA_pol introduces mutation at 10^-5 substitutions/base, ρ≈0.912. However, for multiple (θ) replications cycles, ρ = (1-(1-M_μ)ⁿ)^θ. After 20 replication cycles, occurring in <7 days in most patients 5257, the probability of any original genome remaining un-mutated is ρ_o≈7.5 × 10^-22, meaning effective loss of sequence information, an outcome that should cause quasispecies extinction 16. Persistence of stable RNA viral quasispecies is, therefore, highly paradoxical 18. This "theoretical impossibility" of RNA quasispecies stability suggests either a) the consistently reported rates of RNA_pol infidelity are incorrect (which, even if true, would only delay quasispecies extinction; if M_μ = 10^-10, ρ_o <10^-40 within 100 days etc.) or b) that innate viral mechanism(s) control RNA_pol fidelity and mediate selective replication of consensus sequence genomes. Thus, rates of viral mutation are tightly constrained by the necessity to retain sequence information. On the other hand, overly faithful template replication will restrict antigenic diversity, rendering virus susceptible to immune destruction and unresponsive to ongoing cellular changes. The necessity to retain sequence information by adequate replicative fidelity, and the later requirements (in terms of replicase ⇒ RNA_pol evolution) of viruses to access cells via evolving cell receptors and evade host defence mechanisms, has placed constraints on replicase (RNA_pol) function that dictate polymerase fidelity must be tightly, and dynamically, controlled (Figure 3a).

Optimal viral replication is a compromise between maximising host-to-host viral transmission at each host contact versus maximising transmission at sometime during the host's life: Uncontrolled, exponential growth, as might result from the mode of viral replication, would cause rapid cell lysis, host death and a reduced likelihood of stable host-to-host transmission, a prerequisite for viral survival on an evolutionary timescale. While maximising the probability of host-to-host transmission at each contact, high-level viral replication increases the probability of host disease, thus reducing opportunity for transmission long term. Contrariwise, adverse viral outcomes may result from inadequate viral replication causing increased clearance and reduced host-to-host transmission. Viruses that cause premature host death or that are cleared by host mechanisms before transmission to, and infection of, other hosts are biological failures that have strong Darwinian pressures acting against them. Optimal long-term viral stability, therefore, dictates viral replication rates (that is, polymerase processivity) and mutation frequency (that is, polymerase fidelity) must be closely regulated (Figure 3b).

That viruses capable of chronic persistence auto-regulate replication and mutation rates by replicative homeostasis. Replicative homeostasis results when RNA polymerase end-translation products (envelope and contiguously encoded accessory proteins) interact with RNA_pol to alter processivity and fidelity.

Substantial clinical and in-vitro evidence, including the kinetic paradox indicate viruses auto-regulate. During successful antiviral treatment levels of virus fall sharply 1229525357, often becoming undetectable. However, viral replication rebounds, rapidly and precisely, to pre-treatment levels on drug withdrawal in patients 525357 and in tissue culture 1. This in-vitro data confirm replication is controlled by factors independent of either cellular or humoral immune function. Auto-regulation of HCV replication was confirmed most emphatically in patients undergoing plasmapharesis in whom 60–90% reduction in levels of virus returned to baseline, but not beyond, within 3–6 hours of plasma exchange 44. Studies suggesting autoregulation of tobacco mosaic virus replication occurred independent of interferon effects, intrinsic interference or interference by defective virus 34 confirming this phenomenon is not confined to either animal viruses or cells. These data beg the questions: How does the replicative mechanism "choose" any particular level of replication and how does it return, so accurately, to pre-treatment levels?

Most cellular enzymes are under some form of kinetic control, usually by product inhibition. While simple negative-feedback product inhibition is sufficient to control enzyme reaction velocity and the rate of product synthesis, it is inadequate to ensure the functional quality of any complex molecules – including proteins – synthesised. The functionality of RNA_pol output depends on the functionality of protein(s) translated from any RNA synthesized by RNApol. For viruses, and their polymerase, evolutionary survival – i.e. whether the polymerase, and its viral shell, avoids immune surveillance, gains access to cells, and replicates to infect other hosts – is a function of the properties that the sequence, topological variability and structural integrity of envelope proteins impart. RNA polymerase is responsive to and is influenced by accessory proteins that induce conformational changes to alter both processivity and fidelity 2031, representing partial "proof of concept" of the mechanism postulated.

Evolutionary stability requires adaptability to changing environmental circumstances. For viruses, an ability to modulate replication and mutation rates dynamically in response to cellular changes is essential. Viruses intrinsically capable of adaptation to environmental changes, including variations in host density, and evolving cell receptor polymorphisms, immune and other host responses, among other variables, will enjoy a competitive advantage over viruses lacking innate responsiveness. Contrariwise, self-replicating molecules, including viruses, that lack innate adaptability, for whom replication is contingent upon a chance confluence of appropriate cellular conditions – including permissive cell receptors, absence of cell defences and so on – are highly vulnerable to extinction by both adverse environmental changes and competition for scarce intracellular resources by molecules capable of adaptation. For viruses, this adaptability requires antigenic and structural diversity be controlled and, in turn, that means the two critical RNA_pol attributes, fidelity and processivity, be dynamically modifiable, and controllable. These linked functional requirements imply a dynamic nexus between the functional output of RNA_pol (i.e. envelope proteins) and that polymerase.

Systems capable of homeostatic regulation (auto-regulation) have the following characteristics: i) an efferent arm that effects changes in response to perturbations of an equilibrium; ii) an afferent arm that measures the systems response to those changes; iii) mechanism(s) by which i) and ii) communicate. The mechanism of viral autoregulation – Replicative Homesostasis – described here requires: i) that viral envelope (Env) proteins interact with viral RNA polymerases (RNA_Pol); ii) that these Env :RNA_Pol interactions alter both polymerase processivity and fidelity; iii) that wild-type (consensus sequence) Env_wt :RNA_Pol complexes cause more rapid, less faithful RNA replication than variant (variant) Env_mt :RNA_Pol complexes. There is solid evidence for each requirements of replicative homeostasis.

A large body of literature, for many viruses, establishes an important relationship between envelope and polymerase proteins and documents that Env proteins influence both RNA_Pol processivity and fidelity.

First, for HIV, overwhelming evidence suggests HIV polymerases properties, and those of related retroviruses – for example, simian immunodeficiency virus (SIV) and the feline immunodeficiency virus (FIV) – are influenced by Env proteins (for example, 91535. Broadly, these indicate heterologous Env proteins – when administered as live attenuated vaccines 71, adjuvant enhanced protein vaccine 83, or as recombinant Env proteins in cell culture 64 – dramatically alter viral load, and both replication and mutation rates of wild-type virus. Specific examples include data demonstrating HIV Env regions obtained from different patient isolates, when cloned into common HIV-1 backbones, conferred a spectrum of replication kinetics and cytotropisms characteristic of the original Env clone, and independent of either the clones' ability to raise antibody 51, or the replicative characteristics of the 'native' polymerase backbone 51. Similarly, chimeric HIV-1 viruses expressing heterologous Env, again with a common polymerase backbone, have replication kinetics and cell tropism phenotypes identical to the parental Env clone 39, suggesting the Env is a critical determinant of polymerase function. Similar results obtained with SIV clones 36 strongly support conclusions drawn from feline immunodeficiency virus 37 data. Fine mapping of HIV envelope proteins identified 6 mutations within the V1-V3 loop that increased viral replication in a manner independent of nef 77, confirming other work examining HIV Env recombinants 14, and extending earlier work that demonstrated a single amino acid substitution (at position 32 of the V3 Env domain) was sufficient to change a low replication phenotype into high-replicating phenotype 13. Finally, for HIV, co-transfection with Env variants at 10 fold excess dramatically inhibited replication of wild-type virus 75, providing direct evidence for both the interaction and differential affinity for wild-type and variant Env for polymerases. Critically, many of these observations are from in-vitro systems, indicating the effects are independent of either cellular or humoral immune influence. Many studies report the effect of Env/polymerase interactions in terms of altered viral tropisms, and did not examine changes to polymerase fidelity explicitly. However, virus replication can alter in only two ways; either there is more or less virus, or the viral genomic sequence may be changed by altered polymerase fidelity. Variant viruses expressing altered envelope proteins will have altered cell receptor affinities and hence, variable cell tropisms.

Second, for HCV, many separate observations document HCV replication and polymerase functionality is dependent on envelope proteins: i) HCV viral genotypes are defined by sequences of either envelope or polymerase regions 437374 and these are necessarily acquired together – a genetic nexus implying a functional relationship. ii) Observations that a) co-infection with multiple HCV genotypes occurs less frequently than predicted by chance and b) certain HCV genotypes become progressively dominant in populations both suggest – at a population level – replicative suppression of some HCV genotypes by others 68. These observations are supported by observations of both homotypic 33 and heterotypic HCV super-infection 32 documenting genotype-dependent replicative suppression of one HCV genotype by another in individual patients. iii) Functional infectious chimeric viruses with polymerase and Env proteins derived from different genotypes have not been reported. iv) Full-length HCV chimeras, engineered with deletions of p7 envelope proteins, are replication deficient and non-infections, indicating intact genotype-specific HCV envelope sequences are essential for proper HCV replication. Specific replacement of p7 of the 1a clone with p7 from an infectious genotype 2a clone was replication defective, suggesting a genotype-specific interaction between the p7 envelope protein and other genomic regions 66. v) In two independent chimpanzees studies HCV inoculation resulted in persistent infection

only

highly

Third, HBV envelope and polymerase protein genes have overlapping open reading frames and significant alterations in envelope and polymerase gene and protein sequences cannot, therefore, occur independently, a genetic nexus again implying an important functional relationship. Mutations in envelope sequences occurring spontaneously 82 following therapy of HBV with lamuvidine and immunoglobulin prophylaxis 672 or after vaccine escape 8 are frequently associated with high level viral replication, although replication-deficient mutations are described 47. These data are generally interpreted to mean polymerase gene mutation(s) cause altered polymerase protein sequence and, hence, abnormal polymerase function. While this is probably partially true if the functionally relevant HBV RNA polymerase is an envelope/polymerase heterodimer (analogous to the p66/p51heterodimer of HIV RT 30), then an equally valid interpretation is that mutations in

envelope

Fourth, studies of the coliphage Qβ demonstrate phage coat proteins bind to genomic RNA 86to strongly inhibit (association K_ic ≈ 10^7–8 M^-1, inhibition K_i ≈ 10⁹ M^-1s^-1) 79 RNA replication by direct suppression of polymerase activity by envelope proteins 18. This interaction is dependent on the binding site conformation, but not RNA sequence86, suggesting interaction avidity will vary as an inverse function of protein sequence divergence from wild type, an intuitive expectation confirmed experimentally 79. An impressive body of literature documents similar relationships between envelope and polymerase function in swine fever, tobacco mosaic 34, brome mosaic 2 and other RNA viruses. Importantly, studies of the tobacco mosaic virus confirmed this effect to be host-independent and virus-specific inhibition of viral RNA synthesis and to be quite distinct from any interferon effects, intrinsic interference or interference by defective virus 34. Thus, there exists solid evidence for each necessary component of replicative homeostasis for HCV, HBV and HIV, and other viruses.

Replicative homeosatsis results from differential interactions of wild-type (Wt) and variant (Mt) envelope proteins on RNA_pol in a series of feedback epicycles linking RNA_pol function, RNA replication and protein synthesis (Figure 4, 5). Intracellular accumulation of variant viral proteins causes progressive, direct, inhibition of RNA_pol and also block Env_Wt:RNA_pol interactions that increase replication and mutation. Progressive blockade of RNA_pol by variant envelope results in a less processive, more faithful, polymerase, increasing the relative output of wild-type envelope RNAs, and, subsequently, translation of wild-type envelope proteins and, hence, an inexorable progression to stable equilibria. Quasispecies stability, and other consequences (including immune escape and low-level basal replication), are inevitable outcomes that result from equilibria reached because of these interactions (Figure 5). We suggest these interactions, and the resulting equilibria, are important therapeutic targets, and the effective ligands – envelope proteins or topologically homologous molecules – implicit within this hypothesis.

Viral polymerases are clearly the effector mechanism – the efferent arm – that determines rate of viral RNA replication and mutation. The afferent arm needs to measure both the rate of viral replication and degree of viral mutation. Intracellular envelope concentrations are a direct function of effective viral replication, while competition between wild-type and variant envelope proteins for interaction with RNA_pol allows determination of viral mutation rates. Envelope proteins, as opposed to other viral products, are the obvious products to examine for functional variability, and must form part of the afferent arm necessary to "sense" perturbations in the viral equilibrium. While other viral products could be "sensed" to gauge effective viral replication, only functional measurement of envelope protein concentration and topological variability simultaneous measures both the rate of viral replication and envelope functions – properties determined by envelope structure and antigenic diversity – essential for viral survival; immune escape and cell access. Furthermore, envelope and polymerase proteins are typically coded at transcriptionally opposite ends of the viral genome; replication contingent upon a dynamic nexus between envelope and polymerase proteins is, therefore, a functional check of the integrity of the entire viral genome. Importantly, this facet of replicative homeostasis is a direct mechanism of Darwinian selection operating at a molecular level, that ensurs preferential selection and replication of "fit" viral genomes, and maintenance of genotypes (species).

Viruses, notably HIV, produce many accessory proteins (such as HIV Nef, gag, rev and HBeAg) that affect viral replication and mutation rate. However, these proteins are encoded within envelope open reading frames (ORFs) or are contiguous with them and are likely to alter

functionally

This hypothesis is simply tested. Manoeuvres that increase intracellular concentrations of variant envelope proteins or decrease wild-type envelope proteins should inhibit viral replication and reduce mutation rates. Conversely, manoeuvres increasing intracellular [Env_Wt] or reducing intracellular [Env_mt] should accelerate viral replication and mutation. In fact, observations relevant to every aspect of this hypothesis have been reported in a variety of systems and circumstances. All outcomes are completely consistent with those predicted by replicative homeostasis. Replicative homeostasis predicts, for example, HCV E2 proteins derived from genotype 1 HCV sequences would reduce HCV replication when administered to patients with heterologous HCV infection (genotypes 2,3 or 4, for example) and studies examining heterologous envelope proteins as direct RNA_pol inhibitors are underway.