Prions: nitric oxide, superoxide, and peroxynitrite
Transport, sensory, conditioning roles for repeat region
Active site of SOD3 and regulation of NOS
Review of known copper-binding proteins (refs 1, 2)
6 June 98 webmaster opinionThe 3D structure proposed below for the prion repeat region of both avian and mammal prions are helical coils (of diameter 12.6Å and 17.2Å and mean pitches about 8.7 and 6.5 degrees, resp.) containing tightly bound copper (perhaps alternating with zinc spaced by histidine liganding through both ring nitrogens) in the central core. It takes two full repeat units to make one turn of the helix. Precise resonant positioning of transition metals in an extended shielded environment suggests -- remarkably -- that the repeat domain is the core of an enzyme. This core immediately suggests the active sites of CU-Zn superoxide dismutase. The substrate must be a small inorganic molecule involving oxygen such as superoxide, nitric oxide, or hydroxyl radical because of copper's properties unless the remainder of the prion molecule contributes more of a conventional substrate pocket.
I propose that the repeat domain of prion protein actually constitutes the active site of an unprecedented new family of superoxide dismutases (SOD4), though nitric oxide oxidoreductase (NOOR) and peroxynitrite dismutase (POND) remain viable alternatives. The normal function of prion protein is to sustain a balancing act between levels of the neurotransmitter nitric oxide and coupled reactive oxidants superoxide and peroxynitrite. Copper transport or immobilization models have a similar effect of protective conditioning of the extra-cellular matrix.
Mammals already have three superoxide dismutases (SODs) families: a cytoplasmic representative of the copper-zinc SOD1, a mitochondial representative of manganese active site SOD2a, and a secreted, interstitially located copper-zinc SOD3 (the major superoxide dismutase of human extracellular fluids). These are unrelated to each other and to prion protein in primary structure and evolution. (There is a further iron SOD2b found in E.coli and other organisms that is a recent paralogue of the manganese family.) This would make prion protein the third independent development of a Cu-Zn superoxide dismutase, the fourth independent evolution of a protein with this particular enzymatic capability, and the fifth family with this capability. (Other copper proteins, such as ceruloplasmin, transferrin and ferritin can also catalyze the dismutase reaction but at vastly lower levels.) The OMIM site carries a thorough discussion of known SODs and diseases such as ALS associated with mutant alleles.
Linear His-Cu-His-Zn-His constitutes the active site of SOD1 and quite likely the SOD3 superoxide dismutases. The earliest studies of metal binding by MP Hornshaw et al. to prion repeat domain fragments found that both copper and zinc bound with micromolar binding constants: BBRC 1995 Feb 15;207(2):621-629. Latter studies have favored cupric ion only. DR Brown et al. reported reduced overall SOD activity in null mice but did not attribute this directly to prion protein: Nature 1997 Dec 18;390(6661):684-687. However, no experiment has looked at metal content in situ in immuno-precipitated native protein, nor at full-length recombinant holo-protein reconstituted with both zinc and copper present. Without properly formed protein, in vitro measurement of metal preferences, binding constants, and enzymatic activity remain problematic. Note in avian SOD4, an interesting potential exists for an 'extended' site or multiple active sites : (Cu-His-Zn-His-)3-4; even longer strings could be formed in extra-repeat CJD. Note these extra-length proteins might be too efficient and ironically toxic: by scavenging superoxide, less is available to form peroxynitrite, meaning NO persists too long.
The metal ensemble might be called a polynuclear cluster [or if copper and zinc, "brass"], the number of reactive centers either equal to half the number of repeats or one minus the number of metal atoms (if there is no polarity to the metal pair, ie Cu-Zn-Cu is active either as Cu-Zn or Zn-Cu). A mixed metal enzyme might self-assemble correctly despite uniformity of repeats if Cu-His-Cu or Zn-His-Zn are less favorable; more interestingly, a copper chaparones might be required, such as CCS a specific copper chaperone for SOD1, Cox17 for cytochrome oxidase in yeast, Atx1 a soluble cytoplasmic copper(I) chaperone with a two- or three-coordinate metal active site in a domain related to Menkes and Wilson disease proteins, or MelC1 for apo-tyrosinase.
However, binuclear and trinuclear copper enzymes are also known, such as copper chaperone cox17, laccase, cytochrome ba3, tyrosinases, subunit II of cytochrome c oxidase, nitric oxide reductases, and certain cytochrome-c oxidases (direct Cu++/Cu+ bonds with unpaired electronic spin distributed evenly over the two Cu ions, no intervening histidine). Superoxide dismutases do not require external reducing equivalents.
If the substrate is a nitrogen oxide and not the superoxide anion, then perhaps a role for the hydrophobic domain would be in coupling to membrane bound electron transport elements or compounds such as quinone or alpha-tocopherol (which is rapidly oxidized non-enzymatically by peroxynitrite). The question arises whether the repeat domain by itself could be toxic when uncoupled from the globular and hydrophobic domains and partially degraded. The loss of zinc from SOD1 is said to approximately double its efficiency for catalyzing peroxynitrite-mediated tyrosine nitration, ie, the partial apo-enzyme is a run-away producer of superoxide.
Thus another catalytic activities worth considering is nitric oxide oxidoreductase (NOOR) which oxides NO to nitrite: 2(NO + H2O) ---> HNO2 + H+ + 1 e-. This requires a terminal electron acceptor: the hydrophobic central domain 106-126 would mediate electron transfer to conventional membrane-bound electron transport systems. Prion distribution in the brain, vascular system, and elsewhere might then roughly mirror NO use as neurotransmitter. The function would be to speed the elimination of NO so that it does not react with superoxide to form peroxynitrite. The globular domain of prion protein could bind with regulatory molecules coupled to this neurotransmitter and modulate enzymatic activity of the repeat domain through conformational change, change that could be locked in with certain point mutations. Nitrite as end product is itself toxic (blue baby syndrome, used to color myoglobin in meat) but far less so than peroxynitrite.
One wonders if in extra-repeat CJD, NOOR could be a bit too efficient, scavenging NO so rapidly that it cannot reach its target guanyl cyclase heme in sufficient quantity to serve as neurotransmitter, with many adverse implications for neuronal health perhaps as NO (and peroxynitrite) come to be over-produced in compensation. Mutants like P102L and P105L could interfere with hypothetical feedback of NOOR by carboxy terminal domains accomplishing the same effect by locking the switch in the wrong position. The balancing act between not enough nitric oxide versus too much peroxynitrite could also be upset by quaternary dimer interactions between mutant and host prion, or gene dossage effects.
Finally, the substrate could be peroxynitrite itself and the enzymatic activity peroxynitrite dismutase (POND): NO + O2- ---> O2N0- --->NO3-, that is, nitric oxide interacting with superoxide forms peroxynitrite which is catalytically rearranged to form nitrate with no net gain or loss in reducing equivalents. This reaction requires no integration with cellular membrane electron transport. This reaction is similar to superoxide dismutase: O2- + 02- ---> HOOH + O2 and to catalase: HOOH + HOOH ---> 2H20 + O2. The regulatory issues are more conventional: loss-of-function with accumulation of a toxic metabolite. Knockout mutants would need some compensatory mechanism higher conventional SOD3?) suppressed in certain deletions of both repeat and hydrophobic domains; point mutations are recessive and not seen for an autosomal gene.
With all three potential substrates, the question is always raised, why have an enzyme when the reaction proceeds so fast on its own? (And then there is carbonic anhydrase.) Irwin Fridovitch met with considerable controversy in trying to gain acceptance for superoxide dismutase in the 1960's; today a whole industry centers around this enzyme and the aging process. The answer is, these reactive compounds can do a lot of non-specific oxidative damage in sub-microsecond time scales before they dissipate. Very fast enzymes provide an effective -- but by no means perfect -- competing reaction to self-dissipation and cell damage.
SOD4 (or NOOR or POND) has a distinctive cellular location tethered to the extra-cellular side of the plasma membrane and an extented finger-like configuration allowing the active site to be at the scene where superoxide radicals might do the most damage: membranes, receptors, and synapses. The problem is very much aggravated by the use of nitric oxide as neurotransmitter because of its very rapid interaction with superoxide anion to form peroxynitrite anion (OONO-), a potent covalent poison for tyrosine residues whose phosphorylation is critical in regulation (by tyrosine-specific protein kinases such as insulin receptor). It is feasible to measure net 3-nitrotyrosine but this has not been done in a prion context.
NO is a physiologically important activator of soluble guanylyl cyclase where it binds to the heme iron; in essence it is scavenged by superoxide.. Nitrotyrosine is found in Lewy bodies in Parkinson. SOD2 itself is inactivated by peroxynitrite inactivtion of a tyrosine. Nitric oxide synthase uses molecular oxygen (calcium/calmodulin-dependent synthesis converts arginine to NO and citrulline) and may itself generate superoxide byproduct. It would be curious if the two modified arginines in prion protein are citrulline. Note too that N(G)-monomethyl-L-arginine (L-NMMA) is a known nitric oxide synthase (NOS) inhibitor
In evolution, this new class of enzyme may have appeared in a recognizable form relatively late (say 550 Mya ago) accompaning needs associated with the elaboration of the nervous and vascular system in longer-lived, higher metabolic rate animals (eg, hummingbird) more susceptible to accumulated oxidative damage and possibly correlating with the appearance of nitric oxide, NO, as neurotransmitter. This scenario fits observed prion involvement in CAA and a proposed higher throughput enzyme in birds, whose long resonant repeat is an unprecedented active centre.
No homologues or hybridizing DNA for prion protein have been found in fruit fly or nematode; happily, both genomes will be fully sequenced with 12 months. However, NOS is found in goldfish, shark, and sea lamphrey, mussel, and flatworm; short-lived or cold-blooded animals may have fewer issues with reactive oxygen species . One wonders if other lineages have spun off yet other variations of the metal-binding repeat theme: marsupials at 178 Mya have a modest variation of 10-9-9-9 and so a helix of 19.7 Å diameter and pitch 5.7 degrees, though the single species sequenced may be misleading for the clade as a whole.
Ascorbic acid is for some species an alternative to SODs. L-gulonolactone oxidase, LGO, is the terminal enzyme in ascorbic acid biosynthesis of in animals. SOD activity has increased in the mammals accompanied by a decrease in the LGO activity. SOD activity is especially high in the guinea pig, bats, monkey and humans, the species lacking LGO.
Inflammatory cytokines upregulate secretion of SOD3 stimulating production of nitric oxide. The phagocytic respiratory burst of neutrophil superoxide from the NADP-dependent oxidative system could have a counterpart in microglia; the effect would be to exacerbate any defect in SODs; normally, the respiratory burst and nitric oxide synthase act synergistically to kill microbial pathogens. Homocysteine, recently implicated in Alzheimer, reacts with Cu++ to form superoxide which scavenges NO through peroxynitrite formation. Anti-oxidants are frequently suggested in CJD and Alzheimer as therapeutic approaches.
17 May 98 webmasterThe structural data are compatible with other possibilities for normal function that accomplish related reactions without being conventional enzymes; he bottom line is conditioning of the extra-cellular medium:
--As synthesized in vivo, more abundant intra-cellular zinc may be the initial bound metal; on the outer cell surface, the PPII structure may then functions literally as an ion-exchange 'column' for more avidly bound copper. If random coil does reach the cell surface, or if metal is removed as part of a functional cycle, the repeat region in effect scavenges the extra-cellular matix and synapses for copper, coiling up cooperatively as it finds metal ions. Prion protein is reported to have a 6-hour turnover.
There is also a precedent for copper chaperones, ie, proteins that function specifically to capture deliver copper in a suitable chemical form to enzymes that need them. Many neurotransmitters are synthesized or scavenged by copper-containing enzymes and copper is reported released with normal packets of neurotransmitter.
The repeat domain may be closely coupled with the following hydrophobic invariant domain, whose function remains unknown despite many studies of its role in the conformational transition in CJD. These domains may have more to do with each other than either does to the main globular region. Structures such as hairpins and amphipathic coil/sheet have been proposed for the hydrophobic invariant domain. If the prion molecule acts as a molecular switch, the invariant core is probably the heart of the switch. It is normally held in the 'off' position by the globular domain (which has a second independent role of providing feedback to the cell that it is on the job in adequate amounts). However, when the repeat region has bound metal ions, the region can 'open' the switch by the hydrophobic interactions of the aromatic residues. The core peptide has been shown in vitro to form ion-permeable channels in the cell membrane.
This suggests a 'flypaper' model: in its extended PPII conformation, the repeat region extends into the extra-cytoplasmic space or a synapse to capture copper on its 'sticky' surface. LIganding of copper to histidine/glutamine causes the repeat region to contract into a more compact double-coiled conformation whereby its aromatic residues stack in such a way as to bind the hydrophobic core domain, opening its pore and allowing copper to diffuse into the cell. When the copper has left, the repeat region extends again, losing affinity for the core domain and causing the pore to close, preventing leakage of ion gradients [which amounts to an out-of-control ATPAse, since these would have to be pumped out to maintain homeostasis], the hydrophobic domain possibly exiting the membrane, though the protein as a whole remains attached through its GPI anchor.
The overall metaphor is the sticky tongue of a frog extending out to capture an insect, contracting the tongue, swallowing, then sticking the tongue out for another meal (of copper). Quantitative copper transport may not the purpose -- rather, protection of sensitive structures such as synapses from metal-catalyzed oxidative damage. A similar idea differently motivated was put forward by Brown et al. In knockout mice, the cell might compensate somewhat by increasing classical copper transporters, by increased non-lethal turnover of damaged components, or increasing SOD3, etc. Partial deletions of repeat and pore domains would not function but the residual fragment might signal the cell not to compensate. In effect, the repeat domain condition the extra-cellular medium where reduced glutathione does not reach.
Note that in the copper capture scenario, as the repeat domain cycles between two conformations [PPII extended, PPII contracted] while it captures and delivers copper, the hydrophobic core domain also cycles between two conformational states [closed-pore surface, open pore integral membrane]. This may be an accident waiting to happen. When the conformations are decoupled by partial proteolysis or by mutation, the domain may get stuck in the [lethal] open pore conformation which leads to cell death by energy depletion. Prolines 102 and 105 could mediate communication between the two flanking domains, switching between cis and trans conformation during the copper cycle to signal the pore domain.
Could the repeat region, in conjunction with the following hydrophobic domain, itself penetrate the membrane or possibly make a round trip? While the length of PPII coil is adequate as approximately two bilayers and the two lysine caps might favorably interact with phospholipid, the hydrophobicity is so-so, the polar backbone is not satisfied with hydrogen bonding, and the region is released to the extra-cellular milieu when phospholipase clips the GPI anchor.
A variation of this model simply has the repeat region collect but not transport copper. After loading, a conformational change signals to the cell to clip the GPI anchor and release the protein to the extra-cellular medium. Next, the sialic acids are cleaved. In many soluble glycoproteins, such as the copper protein ceruloplasmin, loss of sialic acid on the glycosylation sites is a signal to remove the protein from the circulation. This amounts to transport of unwanted copper (or cadmium etc.) from the brain to the liver, which is better equipped to detoxify transition metals.
Note that the PPII coil itself is so extended [unlike alpha helix] that it is actually 3x longer on average than random coil itself [statistical mechanics given in T. Creighton, 'Proteins: Structures and Molecular Properties' W.H. Freeman, 1993, pg 178], so length considerations are just backwards (though the random coil does have very long, if rare, extensions). This suggests that the random coil structure has no physiological signficance as a conformer.
Could PPII form a passive bridge or anchor or spacer, connecting the cell surface to the extra-cellular matrix, another neuron, or pinch off a membrane pocket or anchor a raft of specialized lipid? This seems quite feasible, the highly basic ends perhaps have affinity for negatively charged phospholipid or some receptor. This scenario fits the overall conservation of length in mammal and bird repeat (125 angstroms). However, no role is then left for copper (except in a one-time contractile mechanism to draw structure together); conventional collagen-like helix could accomplish these tasks, so what explains 310 million years of evolutionary conservation of the histidines? There are no known examples of structural copper, a trace element. Only stoichiometric amounts of copper are immobilized (though the protein turns over in 6 hours). PPII does not have the structural domains of elastin, not needed in the brain. Note PPII coil is recognized by binding proteins in other contexts so an additiional role could be an exposed target for a SH3 homology domain.
Could the PPII structure be a sensory finger? The resonance structure makes the repeat almost an antenna (or lightning rod). In the protein HPRG, a high histidine-glycine protein with tandem repeats of consensus G(H/P)(H/P)PH, the function (of local pH detection) is thought to be modulated by copper and zinc. Here the special pK of histidine is exploited. The repeat binds heparin through the histidines, a strikingly pH-sensitive reaction having a titration curve with midpoint pH 6.8 and little binding to heparin at physiological pH in the absence of metals. HPRG is interpreted as a local pH sensor, interacting with negatively charged GAGs on cell surfaces only when it acquires a net positive charge by protonation and/or metal binding, providing a mechanism to detect local acidosis (e.g. ischemia or hypoxia). [J Biol Chem 1998 Mar 6;273(10):5493-5499]
The graphic below represents the main ideas in these alternatives: repeat region copper capture with a switchable pore:
Some predictions of the model:
-- lineage chimeras (chicken repeat substituted for mammalian repeat or vice versa) might function quite satisfactorily because the hydrophobic domain is identical and the repeat region may interact only with this; lengths, hydrophobic profiles, and copper-binding capability are the same.
-- mild mutations of conserved residues in the repeat region to similar amino acids should nonetheless be dysfunctional (though not necessarily related to CJD). For example, PHNPGY to PKNPGY, PHDPGY, PHNAGY, PHNPPY, PHNPGV should totally fail to function.
-- the second modified arginine will also occur in birds, where it also occupies a strategic position at the very beginning of the repeat region. The nature of this modification remains unknown, limiting speculation on its role.
-- 3-nitro-tyrosine levels will be elevated in knockout mice. One wonders what effect nitrosylation would have on the tyrosine of the first beta strand and whether this modification is found generally in abnormal conformer.
7 June 98 webmasterThe extra-cellular SOD3 sounds so very similar to prion SOD4 -- are the proteins related or even redundant? The SOD3 enzyme is synthesized with an 18-amino acid signal peptide preceding the 222 amino acids in the mature tetrameric enzyme. The protein has a single carbohydrate attachment site at position 107 and a disulphide between 125 and 207. Copper-binding histidines are at 114, 116, 131, and 181; zinc histidines are at 139, 142, with the aspartic acid at 145. There is no obvious homology or repeat structure and the 3D structure has been undeterminable, yet one internal stretch is quite suggestive of avian prion repeat:
MLALLCSCLL LAAGASDAWT GEDSAEPNSD SAEWIRDMYA KVTEIWQEVM QRRDDDGTLH AACQVQPSAT LDAAQPRVTG VVLFRQLAPR AKLDAFFALE GFPTEPNSSS RAIHVHQFGD LSQGCESTGP HYNPLAVPHP QHPGDFGNFA VRDGSLWRYR AGLAASLAGP HSIVGRAVVV HAGEDDLGRG GNQASVENGN AGRRLACCVV GVCGPGLWER QAREHSERKK RRRESECKAA
In the vascular system, SOD3 appears to be located on the endothelial cell surface. The characteristic distinguishing SOD3 from SOD1 and SOD2 is the heparin-binding capacity. SOD3 binds on the surface of endothelial cells through the heparan sulfate proteoglycan and is thought to detoxify superoxide from the oxidative system of neutrophils.
Getting into the chemistry of it all, we learn [Chem Res Toxicol 1998 Apr;11(4):243-246]
"The rate constant of homolysis of peroxynitrite, ONOO-, into O2*- and NO* was determined to be 0.017 s-1 at 20 degrees C.This value yields the Gibbs free energy of formation of ONOO-, 16.6 kcal/mol. Thus peroxynitrous acid homolyzes to yield nitrogen dioxide (NO2*) and hydroxyl (OH*) free radicals with free energy 7.7 kcal/mol. The rate constant of the reaction between NO* and ONOO- was found to be 5 x 10(-)2 M-1 s-1 at most. "
There seems to be only one paper relating prion disease and nitric oxide:
J Biol Chem 1996 Jul 12;271(28):16856-16861 Effect of scrapie infection on the activity of neuronal nitric-oxide synthase in brain and neuroblastoma cells. Ovadia H, Rosenmann H, Shezen E, Halimi M, Ofran I, Gabizon R"Nitric-oxide synthase (NOS) is responsible for the synthesis of nitric oxide which serves as a neural messenger in the central nervous system. NOS activity was markedly inhibited in brains of mice and hamsters and neuroblastoma cells infected with scrapie ...."
Science 1996 Nov 1;274(5288):774-777 PIN: an associated protein inhibitor of neuronal nitric oxide synthase. Jaffrey SR, Snyder SHThe neurotransmitter functions of nitric oxide are dependent on dynamic regulation of its biosynthetic enzyme, neuronal nitric oxide synthase (nNOS). By means of a yeast two-hybrid screen, a 10-kilodalton protein was identified that physically interacts with and inhibits the activity of nNOS. This inhibitor, designated PIN, appears to be one of the most conserved proteins in nature, showing 92 percent amino acid identity with the nematode and rat homologs. Binding of PIN destabilizes the nNOS dimer, a conformation necessary for activity. These results suggest that PIN may regulate numerous biological processes through its effects on nitric oxide synthase activity.
MCDRKAVIKN ADMSEEMQQD SVECATQALE KYNIEKDIAA HIKKEFDKKY NPTWHCIVGR NFGSYVTHET KHFIYFYLGQ VAILLFKSG