Serum amyloid A and protein AA: Molecular mechanisms of a transmissible amyloidosis

1. Introduction 

The fibril protein precursor in all forms of systemic amyloidosis is expressed in one or several tissues, delivered to the blood plasma and deposited in tissues often far away from the place of synthesis [1], [2]. This is a major difference from the localized amyloid forms, e.g. those associated with type 2 diabetes or Alzheimer’s disease where the amyloid is formed close to the place of protein synthesis [3]. Exactly how the transport from the plasma to the deposition site takes place is not known. Likewise, the mechanisms behind the determination of deposition sites are almost completely unknown.

Systemic amyloidoses are life-threatening diseases in which one of about 15 different plasma proteins misfold and aggregate in many different organs throughout the body [3]. Generally, there are two known major pathways leading to the formation of the fibrils, both resulting in a concentration above the critical to favor aggregation. In one, there is a mutation in the gene for a plasma protein, giving rise to a destabilized protein which can create aggregation-prone protein species. This seems to be true for the common amyloid fibril protein transthyretin and several other proteins associated with familial forms of amyloidosis. In the other way, an otherwise normal protein is over-expressed for a sufficiently long time to nucleate and start fibrilization. This latter route seems to be mechanism behind some of the most common systemic amyloidoses including immunoglobulin light chain (AL-) and AA- (reactive or secondary) amyloidoses.

The major amyloid fibril protein in systemic AA-amyloidosis is derived from an apolipoprotein called serum amyloid A (SAA). SAA is expressed by three different genes in human [4]. Two SAA proteins are acute phase reactants, both with the ability to form amyloid in vivo. The third protein, SAA4, is constitutively expressed in a number of tissues and has been described to form amyloid when mutated [5]. The two acute phase SAA proteins, called SAA1 and SAA2 occur in several allelic forms. SAA1a seems to be particularly common in amyloid fibrils and is possibly the most amyloidogenic species. SAA1 and SAA2 are to the major part bound to high density lipoprotein [6] and are at healthy state only a minor protein component. In inflammatory conditions the production of SAA from the liver increases dramatically under influence of interleukin (IL)-1, IL-6 and tumour necrosis factor (TNF) and can reach plasma concentration 1000 times the normal [7]. Apolipoprotein A-I is then depleted and SAA becomes the major HDL-protein. Whether free SAA or apoSAA serves as substrate for fibril formation is not known.

In human, AA-amyloidosis is a complication of chronic inflammatory disorders in which the plasma SAA levels are consistently high. Infectious disorders such as tuberculosis, leprosy or malaria were previously the main causes and are still so in some countries [8], [9] but in the Western world non-infectious diseases, particularly rheumatoid arthritis, are more commonly associated with AA-amyloidosis. Also reactive amyloidosis can be associated with hereditary fevers, particularly familial mediterranean fever (FMF), caused by a mutation in the MEVF gene encoding for the protein pyrin, an IL-1beta regulator [10]. Often, many years of inflammatory disease are necessary for amyloidosis to develop. Of yet unknown reason only a minority of individuals with long-lasting inflammations develop AA-amyloidosis. One possibility is that these individuals have insufficient degradation systems for aggregated proteins. Another possibility could be that affected individuals have received one or several nucleation or seeding factors. This is what the present review deals with.

2. Prions 

In 1957, Gajdusek and Zigas described a unique disease, affecting members of the Fore people in the New Guinea Highlands [11]. The disease was progressively leading to death, usually within one year. Extensive search for postinfectious causes, dietary deficiency or toxic agents were all negative. Much in the disease, including the restricted population in which it occurred made a heredo-degenerative type of disorder most probable [12], [13]. Autopsy study revealed neurodegeneration and in about half of the cases amyloid plaque-like bodies [14]. In 1966, Gajdusek et al. published the seminal paper in which it was shown that kuru was transmissible into chimpanzees by intracerebral inoculation [15]. Later, it was shown that kuru, Creutzfeldt-Jakob diseases and scrapie, all could be transmitted into monkeys via the food chain [16]. Initially, the transmissible material, often called scrapie agent was believed to be of viral nature and given the very long incubation time, the name ‘slow virus’ was created. However, no virus was detected in spite of extensive search. Intense work by Prusiner and coworkers has led to convincing evidence that the infectious agent is not a virus, but ‘protein only’ [17]. The responsible protein was identified and given the name prion [17]. The normal prion protein, PrPC can be converted to an abnormal, transmissible PrPSc which can start a chain reaction by interaction with the normal protein.

3. Infectious amyloid 

Interestingly, about the time when Gajdusek published the transmission of kuru, a strange phenomenon was described in experimental mouse AA-amyloidosis. Many strains of mice develop, after a fairly long lag phase, this form of systemic amyloidosis when given a chronic inflammatory challenge. Ranløv and coworkers found that cells or extract from an already amyloidotic mouse injected at the time of induction of inflammation dramatically reduced this lag phase [18]. The phenomenon was verified by other studies [19]. Initially, quite complicated extraction methods to obtain AEF were used but just ordinary AA-amyloid fibril preparations [20] work efficiently [21]. From the liver of a single amyloidotic mouse, we made a fibril extract containing 1.32mg protein per ml and this material is astonishingly efficient at serial dilutions down to picograms of protein [22]. The extract from this single mouse should be enough to induce amyloidosis in millions of animals.

The finding that formation of amyloid fibrils is a nucleation dependent event is of fundamental importance and this seems to be principally the same for all forms of amyloid including that of the prion protein [23]. In a solution of an amyloid fibril protein above the critical concentration a nucleus forms after a lag phase which can be quite long. Seeding such a solution with preformed fibrils of the same nature more or less abolishes this lag phase [23], [24]. In some yet unknown way, the misfolded and aggregated protein induces a rapid misfolding and aggregation of the protein in solution. This in vitro phenomenon resembles what seems to happen when an amyloid-prone animal receives AEF. Although most data indicate that the AEF activity is exerted by the protein core in the amyloid fibril, it should be remembered that amyloid extracts also contain some additional minor components including serum amyloid P-component [25], proteoglycans [26] and lipids [27]. Although these components are believed to be important in amyloidogenesis and for the persistence of the fibrils, nothing is presently known regarding their possible role in transmission. Studies on this possibility would be of interest.

The following experiment indicates that the seeding includes a three-dimensional effect on the protein and shows that there is a possibility that nucleation may originate from inorganic material as well [28]. By heating prion-containing material above a temperature where no organic material should persist, Brown et al. obtained evidence that inorganic replicas formed [29]. The infectivity of these was quite low but the finding is of great interest.

It took quite long time until the nature of the accelerating material in AEF was understood but today, all facts indicate that the mechanism is seeding by fibrils or fibril fragments [30], similar to the prion phenomenon [22]. This conclusion was further emphasized by the finding that amyloid-like fibrils, made from synthetic short peptides, exert AEF activity [31], [32]. It was then shown that the new AA-amyloid was directly added to the injected foreign fibrils. The AEF phenomenon is not restricted to mice but has been shown for Guinea pig [33], Chinese hamster [30] and mink [34], indicating a general mechanism. Of great principal importance are studies showing that murine systemic amyloidosis of a completely different biochemical form, derived from apolipoprotein A-II, also is seedable in the same way [35], [36].

4. Routes for transmission of AA-amyloidosis 

Prion disorders are transmissible by different routes, including the oral way, most dramatically shown by the example of kuru [15], [37]. In most AEF-experiments, the ‘infectious’ material has been given intravenously or intraperitoneally. However, addition of a small amount (about 1mg/l; calculated total dose per mouse 0.2mg) of a fibril preparation to the drinking water for 5 weeks before induction of inflammation accelerated the development of AA-amyloidosis [22]. Where and how the seeding AA-amyloid fibril or particle is taken up from the gastrointestinal tract is completely unknown. It seems possible, although unproven, that the route is the same as for the prion seed. In the latter case, specialized epithelial microfold (M) cells overlying Peyer’s patches, are believed to play a central role [38] and may transport PrPSc transepithelially. From there, further transport by follicular dendritic cells may take place. Interestingly, the spleen is one organ in which PrPSc replication may occur [39] and it is also the first organ where AA-amyloidosis develops in the mouse. When radiolabelled fibrils were injected intravenously, radioactivity was particularly pronounced in the perifollicular area in the spleen [32], the site where the very first amyloid develops [33]. The route for the infectious AA-material to the spleen has not yet been clarified. Further spreading can take place by the blood since in murine AA-amyloidosis, some circulating monocytes were found to carry amyloid fibrils and had the capability to accelerate amyloidosis in recipient animals [40].

5. Non-experimental seeding of systemic AA-amyloidosis 

Acceleration of the development of AA-amyloidosis was originally described from laboratory work with mice. The question can then be raised whether this is a type of laboratory artifact. The fact that this kind of acceleration is efficient not only in rodents but also in a carnivorous animal contradicts this. Furthermore, in a recent paper the possible horizontal spreading of AA-amyloidosis in captive cheetah was described [41]. Interestingly, as early as 1976 and at a time when seeding of amyloid was not known, Benditt wrote about a strange phenomenon in Peking ducks. He pointed out that in his colony of ducks, none had developed amyloidosis after a year although two-thirds of animals of this species in other colonies were reported to have amyloidosis in their second year [42]. He also performed transmission experiments by injection of amyloid-containing liver homogenate into ducks and found that they in due time developed more severe AA-amyloidosis than animals receiving liver without amyloid. Unfortunately, no more experiments seem to have been reported. Also in several other avian species AA-amyloidosis is common. A few years ago, there was a mass-death of birds, particularly herring gulls, in Southern Sweden. A number of herring gulls were analyzed and AA-amyloidosis was identified in about 50% of them (Désirée Jansson et al., manuscript in preparation). The animals suffered from different infections and infestation and the direct cause of death was determined to be botulism [43]. However, the prevalence of AA-amyloidosis was remarkable and a horizontal transmission can be suspected but certainly not proven.

6. Transmission of systemic AA-amyloidosis between species 

Prion diseases can be transmitted between species, a well-known fact since the work by Gajdusek and coworkers [15] but particularly since the occurrence of human variant spongioform encephalopathy (variant Creutzfeldt-Jakob disease) [44], originating from cattle and transmissible to mice [45]. Given the resemblance of prion protein with other amyloid fibril proteins’ behavior in vitro [46], it is natural to question whether AA-amyloidosis is transmissible between species. The answer is yes. As early as 1969, Shirahama et al. reported transfer of experimental amyloidosis by human splenic amyloid homogenate to mice [47]. In more extensive studies, AA-amyloid fibrils from several different mammalian and one avian species were found to accelerate experimental murine amyloidosis [48], [49]. In our studies, we found a considerable difference in “infectivity” between donor species and while AA-fibrils extracted from donkey were efficient as amyloid enhancers, human AA-fibrils were not (Westermark et al., unpublished results). Similar to the results obtained by Cui et al. [49], we found that bovine AA-fibrils have the ability to induce AA-amyloidosis in mice. However, fibrils from none of the other tested species were as efficient as those of mice in agreement of a species barrier, similar to that described for prions [50].

7. Possible inoculation in human 

AA-amyloidosis occurs as a complication to chronic inflammatory disorders. The reported incidence varies substantially between different parts of the world and only very few really reliable studies have been reported. One major problem is that the disease often is severely under-diagnosed. It is, however, clear that only a proportion of individuals in the high risk zone develop AA-amyloidosis and the reason for this is unknown. One possibility is that AA-amyloidosis develops as a result of a stochastic event, similar to what has been proposed for sporadic Creutzfeldt-Jakob disease. However, there is a definite possibility that at least in some cases, the disease is induced by a seeding agent. Direct seeding with amyloid fibrils from another individual is possible, but probably rare. However, Sponarova et al. [40] found in amyloidotic mice scattered monocytes, containing definite AA-amyloid particles. Such cells were able to transmit AA-amyloidosis to recipient mice. Furthermore, blood cells from humans with AA-amyloidosis but not healthy individuals accelerated AA-amyloidosis in mice (G.T. Westermark et al., to be published). Transmission of variant, but not classical, Creutzfeldt-Jakob disease via blood transfusion has occurred [51], [52], [53]. A transfer of AA-fibrils by transfusion is theoretically quite possible. Organ transplantation is another risk factor as shown by us [54]. A third, possible route of transmission of AA-amyloidosis could be by heterologous seeding. Systemic AA-amyloidosis is common in many mammalian and avian species and may enter our food chain. Thus, Tojo et al. found a high incidence of AA-amyloidosis in slaughtered Japanese cows [55]. This form of amyloidosis is particularly common in several species of birds and is a problem in duck and goose industry. AA-amyloid was found in commercially available duck liver and in paté de foie gras [56] and fibrils from such material accelerated amyloidosis in mice over-expressing IL-6, both when administered intravenously and given orally [56]. Similar to the situation with bovine spongioform encephalopathy where a large human population is believed to have ingested infected meat from such animals, it is very likely that many, perhaps a majority of individuals in many countries have eaten food containing AA-amyloid fibrils which may, if the conditions are the right, induce amyloidosis in the recipient.

One interesting phenomenon, noted in the mouse, is that there may be a long delay between the intake of amyloid enhancing material and the development of disease. When amyloid fibrils were given intravenously into animals without an inflammatory challenge, the mice stayed healthy. However, if inflammation was induced with silver nitrate as much as 4months later, amyloidosis developed almost immediately [22]. That means the individuals may be primed for the amyloid disease in the case that a chronic inflammatory disease is acquired later. A similar primed state has been described in murine AA-amyloidosis after resolution of the deposits [57]. It will be very difficult to prove an association of ingestion of amyloid and development of AA-amyloidosis if this kind of delayed effect is true also in human. It is of interest to note that the incubation period for kuru has been proven to be as long as over 50years [58].

8. Additional amyloid induction possibilities 

Not only amyloid fibrils may induce AA-amyloidosis in susceptible animals. There are many examples of amyloid-like protein assemblies in nature. Spider silk contain beta-sheet fibrils [59] and fibrils of the prion-like Saccharomyses protein Sup35 [60] and bacterial curli [61] have typical amyloid fibril properties. Silk [62], [63], curli, and Sup35 fibrils [63], all have AEF properties in the mouse model. One line in nano-technology for tissue engineering includes construction of a fibrillar scaffold, made from assembled short synthetic peptides [64], [65]. Such fibrils forming hydrogels have a great potential to provide a scaffold for new tissue to grow on. They can have great advantages over some other synthetic materials, for example they are biodegradable. There are different ways to construct the materials and one is through the assembly of peptides to beta-sheet fibrils. These are ultrastructurally amyloid-like, bind Congo red and show green birefringence after this staining. Interestingly, such intentionally amyloid-like fibrils may accelerate systemic AA-amyloidosis in the mouse (Westermark et al., submitted manuscript). This should be a concern given the purpose to use the assembled peptides in vivo. Methods to test possible amyloid-accelerating properties of any material constructed for in vivo use obviously have to be developed.

Taken together, these different facts show that we in the natural and cultural environment are surrounded by and probably ingest and inhale many different protein assemblies which may start a systemic AA-amyloidosis in a susceptible individual, i.e. any with a long-lasting inflammation leading to persistently high plasma concentration of SAA. Humans are probably not as sensitive as the experimentally used mouse strains and more rarely are the SAA levels in humans persistently as high as those reached in the laboratory animals. In addition, there is nothing known about barriers between other animals and human. We may also have protection systems against aggregation of proteins into amyloid fibrils, including immunity [66] (Nyström et al., submitted manuscript). Finally, one may speculate whether seeding with foreign materials may play a role in the pathogenesis of other amyloid-related disorders.

9. Variations in systemic AA-amyloidosis – a strain phenomenon? 

Most commonly, human AA-amyloid is derived from one single gene product, SAA1a [67], [68]. In spite of this, the disease is not uniform. Interestingly, there are at least two distinctive phenotypes, particularly well recognized in the kidney [69], [70], [71]. In the most common, glomeruli are severely affected (Fig. 1a and b), resulting in early proteinuria. In the other type, there is massive vascular amyloidosis but very little glomerular deposits (Fig. 1c and d) and patients with this form may present with renal insufficiency without glomerular leakage. The two histological types are strongly associated with protein cleavage pattern [71], [72] and particularly long and short AA-proteins are characteristic of the vascular type. These two morphological variants have also different staining properties. It should be pointed out that in these patients amyloid deposits at other locations, e.g. spleen and liver are identical with that in the kidneys. The reason for this variation is completely unknown but may be related to the strain phenomenon in prion diseases. It has been known for long time that prions may vary in their incubation time and in their electrophoretic behavior after proteinase K digestion. These differences in properties can be propagated repeatedly in mice with identical prion genes. In vitro experiments with recombinant truncated prion proteins have convincingly shown that the propagation is conformational driven by conformation templating, i.e. the protein adapts the three dimensional structure of the seed [73], [74]. The generated fibrils acquired similar secondary structure and morphology as the seed. Similar results have been obtained with the Saccharomyces prion protein Sup35, a protein that has no sequence identity with the mammalian prion protein but which can aggregate into amyloid-like fibrils [75], [76]. These similar findings with two completely unrelated proteins making amyloid-like fibrils can point to a universal mechanism which may be attributable to human amyloid forms of other protein nature, including systemic AA-amyloidosis. If this holds true, we should expect differences in fibril morphology in experimental amyloidosis obtained with different seed. However, this possibility seems not to have been studied as yet but is a subject of our laboratories.

View Large Image Download to PowerPoint

Fig. 1. A possible strain phenomenon in systemic AA-amyloidosis. In a and b is shown a section of a kidney with widely spread deposits in glomeruli (arrows). There is also amyloid in arterioli. The section in c and d is from a kidney with almost no amyloid in the glomeruli (arrows) but pronounced vascular amyloidosis. Congo red with normal light in a and c and polarized light in b and d.