PIB and other radioligands for deposited amyloid are being used with increasing frequency to help identify persons with amyloid pathology, yet the specific binding sites for these ligands remain uncertain. This paper examines binding in tissue from humans dying with AD, humans dying without AD, and nonhuman primates with insoluble amyloid. The results may indicate that PIB binds preferentially to the amyloid characteristic of human AD.

View all comments by Richard C. Mohs

This study adds further weight to previously posed hypotheses that not all aggregates of Aβ are structurally identical—at least not in terms of PIB binding.

This phenomenon was first recognized early in the development of PIB-related tracers, when it was noticed that the Bmax for PIB binding to synthetic Aβ1-40 was very low with a stoichiometry on the order of 1 mole of PIB to 500 molecules of Aβ (then unpublished data). While this phenomenon was initially brushed off as some artifact unique to in vitro fibrils, it raised its head again when we (and others) tried to employ several different amyloid-depositing transgenic (Tg) mouse models in our preclinical amyloid imaging development program only to find that PIB didn’t bind well in these models despite levels of insoluble Aβ that greatly exceeded those found in human AD brain (1,2).

Although Maeda et al. later teased out detectable PIB binding in the APP23 mouse in an elegant study (3), it remained clear that amyloid plaques in Tg mouse brain are very different from those in human AD brain, at least by the nature of...  Read more

This study adds further weight to previously posed hypotheses that not all aggregates of Aβ are structurally identical—at least not in terms of PIB binding.

This phenomenon was first recognized early in the development of PIB-related tracers, when it was noticed that the Bmax for PIB binding to synthetic Aβ1-40 was very low with a stoichiometry on the order of 1 mole of PIB to 500 molecules of Aβ (then unpublished data). While this phenomenon was initially brushed off as some artifact unique to in vitro fibrils, it raised its head again when we (and others) tried to employ several different amyloid-depositing transgenic (Tg) mouse models in our preclinical amyloid imaging development program only to find that PIB didn’t bind well in these models despite levels of insoluble Aβ that greatly exceeded those found in human AD brain (1,2).

Although Maeda et al. later teased out detectable PIB binding in the APP23 mouse in an elegant study (3), it remained clear that amyloid plaques in Tg mouse brain are very different from those in human AD brain, at least by the nature of having hundreds of times lower density of PIB binding sites. Maeda and colleagues suggested that PIB binding may be determined by the form of Aβ that is missing amino acids Asp-1 and Ala-2 and has the glutamate in position-3 cyclized into a pyroglutamate, i.e., Aβ-N3(pE). While this may be true, it is not proven, and synthetic Aβ-N3(pE) fibrils do not reproduce the stoichiometry found in AD brain.

In the Rosen et al. study, this issue is pushed several steps closer to humans by the use of aged primate models of Aβ deposition. Rosen et al. have shown that, although non-human primates produce and accumulate Aβ that is identical in amino acid sequence to human Aβ, no non-human species has been shown to exhibit the full behavioral or pathological characteristics of AD (4). So why not? Is it the amount of Aβ, the nature of the primate brain, or the characteristics of the Aβ?

This paper clearly shows that at least some aged primates deposit AD-like levels of insoluble Aβ40 and Aβ42. Whether the oligomeric composition of this Aβ is similar to that in humans is a question likely to be answered soon by several labs skilled at these measurements. However, Rosen et al. show that the stoichiometry of PIB binding in non-human primate brain is orders of magnitude lower than that in AD brain; a finding very similar to that in transgenic mice. The authors go on to suggest that “the high-affinity binding of PIB may be selective for a pathologic, human-specific conformation of multimeric Aβ, and thus could be a useful experimental tool for clarifying the unique predisposition of humans to Alzheimer’s disease.”

This suggests an application of PIB binding that was not foreseen and raises another logical question: will there be pathologic or non-pathologic forms of insoluble Aβ detected in human brain? That is, will there be people without clinical AD who prove to have histologically detectable Aβ postmortem? There are already early hints that this will prove true and will pose a question that will need to be sorted out. Is it just a matter of postmortem histology being more sensitive than in-vivo PET scans? That would certainly be expected, but Rosen et al. suggest that there may be a more interesting explanation. That is, for reasons we do not understand and did not anticipate, PIB may bind much more effectively to a particularly pathological form of Aβ tertiary structure than to a myriad of other possible Aβ aggregates. Much work lies ahead to sort out these questions.

References:
1. Klunk WE, Lopresti BJ, Ikonomovic MD, et al. Binding of the positron emission tomography tracer Pittsburgh compound-B reflects the amount of amyloid-beta in Alzheimer's disease brain but not in transgenic mouse brain. J Neurosci 2005;25:10598-10606. Abstract

2. Toyama H, Ye D, Ichise M, et al. PET imaging of brain with the beta-amyloid probe, [11C]6-OH-BTA-1, in a transgenic mouse model of Alzheimer's disease. Eur J Nucl Med Mol Imaging 2005;32:593-600. Abstract

3. Maeda J, Ji B, Irie T, et al. Longitudinal, quantitative assessment of amyloid, neuroinflammation, and anti-amyloid treatment in a living mouse model of Alzheimer's disease enabled by positron emission tomography. J Neurosci 2007;27:10957-10968. Abstract

4. Rosen RF, Farberg AS, Gearing M, et al. Tauopathy with paired helical filaments in an aged chimpanzee. J Comp Neurol 2008;509:259-270. Abstract

View all comments by William Klunk

Since PIB is a ThT analog, perhaps something can be learned from observations of how ThT fluorescence yields vary among various aggregated forms of Aβ. My lab has been routinely reporting ThT responses of aggregates on an aggregate weight-normalized basis. Even though our data derive from measurements with ThT in the low micromolar range, we see fluorescence differences among aggregates that seem relevant to this discussion. For mature amyloid fibrils grown under identical conditions, we find that single point mutations in Aβ(1-40) lead to fibrils that vary over 1-2 logs in ThT fluorescence (1). In some cases, of course, this might conceivably be due to local effects of the mutation on the stereoelectronic structure of the (still poorly understood) ThT binding site. However, the mutations that yield elevated or suppressed ThT binding vary so much in their chemical natures, and in their locations in the primary sequence of Aβ, that it seems more likely that they are due to amyloid fibril conformational differences.

More recently, we have confirmed the conformational...  Read more

Since PIB is a ThT analog, perhaps something can be learned from observations of how ThT fluorescence yields vary among various aggregated forms of Aβ. My lab has been routinely reporting ThT responses of aggregates on an aggregate weight-normalized basis. Even though our data derive from measurements with ThT in the low micromolar range, we see fluorescence differences among aggregates that seem relevant to this discussion. For mature amyloid fibrils grown under identical conditions, we find that single point mutations in Aβ(1-40) lead to fibrils that vary over 1-2 logs in ThT fluorescence (1). In some cases, of course, this might conceivably be due to local effects of the mutation on the stereoelectronic structure of the (still poorly understood) ThT binding site. However, the mutations that yield elevated or suppressed ThT binding vary so much in their chemical natures, and in their locations in the primary sequence of Aβ, that it seems more likely that they are due to amyloid fibril conformational differences.

More recently, we have confirmed the conformational dependence of ThT binding by generating a number of different polymorphic forms of wild-type Aβ(1-40), all of them mature amyloid fibrils, that differ in their ThT fluorescence yields over a 20-fold range (unpublished). In our hands, protofibrils of Aβ(1-40) are also very weak in their weight-normalized ThT binding (2), although different labs have reported quite different results on this.

Taken together, the hypothesis, as put forward by Rosen et al., that Aβ might aggregate differently in different settings, and that these various forms might exhibit differences in both PIB binding and toxicity, seems quite reasonable. Directly probing how pathogenicity of well-characterized aggregated states might correlate with their PIB binding is a daunting task for the future.

Since polymorphic fibrils, if truly homogeneous, tend to grow (elongate) with retention of conformation, the source of some of these different forms will likely prove to be subtleties in how the local environment influences the early fibril nucleation events that “typecast” the aggregates. This might include the presence or absence of factors that can influence the aggregation pathway (2,3). Mixing experiments of the kind conducted by Rosen et al. is useful for ruling out major influences by species-specific factors that might bind to aggregates and block PIB binding sites. Elucidating species differences in brain structures or molecules that influence nucleation—the initiation of spontaneous aggregation into specific self-propagating polymorphic structures—will require other types of studies.

References:
1. Shivaprasad, S. and R. Wetzel, Scanning cysteine mutagenesis analysis of Abeta-(1-40) amyloid fibrils. J Biol Chem, 2006. 281(2): p. 993-1000. Abstract

2. Williams, A.D., M. Sega, M. Chen, I. Kheterpal, M. Geva, V. Berthelier, D.T. Kaleta, K.D. Cook, and R. Wetzel, Structural properties of Abeta protofibrils stabilized by a small molecule. Proc Natl Acad Sci U S A, 2005. 102(20): p. 7115-20. Abstract

3. Kodali, R. and R. Wetzel, Polymorphism in the intermediates and products of amyloid assembly. Curr Opin Struct Biol, 2007. 17: 48-57. Abstract

View all comments by Ronald Wetzel

Comparative biochemical, radiochemical, and neuropathological assessments of Aβ depositions across species, as elegantly conducted by Rosen and colleagues, may provide clues to identification of neurotoxic Aβ subforms characteristic of human AD brains. This issue was initially highlighted by the observations in murine APP and APP/PS1 transgenic models that the composition of N- and C-terminal Aβ heterogeneities (1) and in-vivo accessibility to plaques by radiolabeled β-sheet ligands, [3H]PIB and [11C]PIB, substantially differed from those in AD patients (2-4), and that these animals did not prominently exhibit fibrillar tau pathologies and progressive neuronal loss.

These documentations led to a tentative conclusion that an elevated level of total Aβ peptides, particularly in an Aβ40-dominant manner, and/or particularly as a consequence of the artificial transgenic overexpression, does not induce intensification of PIB binding sites and devastating neurodegeneration. This view has been now revised by the monkey study showing that an age-related accumulation of endogenous Aβ...  Read more

Comparative biochemical, radiochemical, and neuropathological assessments of Aβ depositions across species, as elegantly conducted by Rosen and colleagues, may provide clues to identification of neurotoxic Aβ subforms characteristic of human AD brains. This issue was initially highlighted by the observations in murine APP and APP/PS1 transgenic models that the composition of N- and C-terminal Aβ heterogeneities (1) and in-vivo accessibility to plaques by radiolabeled β-sheet ligands, [3H]PIB and [11C]PIB, substantially differed from those in AD patients (2-4), and that these animals did not prominently exhibit fibrillar tau pathologies and progressive neuronal loss.

These documentations led to a tentative conclusion that an elevated level of total Aβ peptides, particularly in an Aβ40-dominant manner, and/or particularly as a consequence of the artificial transgenic overexpression, does not induce intensification of PIB binding sites and devastating neurodegeneration. This view has been now revised by the monkey study showing that an age-related accumulation of endogenous Aβ in an Aβ42-dominant form was not accompanied by enhanced PIB binding and abundant tau lesions. One can also presume that the lack of full behavioral AD characteristics in aged monkeys is attributable to the paucity of either an Aβ subtype constituting high-affinity PIB binding sites or pathological tau aggregates, apart from the issue of the presence or absence of certain oligomeric Aβ components accountable for the neurotoxicity.

While these animal data indicate that the C-terminal heterogeneity of Aβ may not have an intimate association with the affinity for PIB, there has been evidence supporting involvement of an N-terminally truncated and pyroglutamylated Aβ, AβN3(pE), in the formation of PIB-accessible amyloid plaques (4). The majority of Aβ peptides regularly undergo proteolytic cleavages by endopeptidases exemplified by neprilysin, but a small subset could be N-terminally processed possibly by aminopeptidases followed by pyroglutamylation of the third amino acid residue by chemical reaction or enzymatic activity of glutaminyl cyclase (QC), resulting in gradual accumulation of peptidase-resistant AβN3(pE) peptides.

The slow nature of the AβN3(pE) deposition may not allow emergence of plaques enriched with this element within the lifespan of an animal model, unless either downregulation of Aβ-degrading endopeptidases or upregulation of QC occurs. Given that high-affinity PIB binding components primarily consist of AβN3(pE) fibrils, efficacies of a novel therapeutic approach to AD by a QC inhibitor (5) would be specifically monitored by [11C]PIB-PET measurements.

The potential significance of the N-terminal portion in both PIB-PET and Aβ immunotherapy also suggest that Aβ is polymerized with its N-terminus facing outward. The loss of one positive and two negative electric charges by the conversion from full-length Aβ to AβN3(pE) may therefore increase the lipophilicity of the outer surface of Aβ assemblies, permitting entrance of lipophilic agents such as PIB into the β-sheet-rich interior (6). The increased N-terminal lipophilicity could also facilitate interaction of Aβ aggregates with endogenous lipophilic molecules, further modifying the accessibility by exogenous β-sheet ligands.

Rosen et al. previously demonstrated by a mass spectrometric assay that N-terminally truncated Aβ peptides predominated in human brains as compared with aged non-human primate brains (7), and this might additionally support the role of the N-terminal Aβ heterogeneity in AD pathogenesis. The relative abundance of AβN3(pE) peptides in aged rodents, monkeys, and humans is yet to be biochemically and immunohistochemically examined.

Immunostaining data in a study by Maeda et al. indicated that AβN3(pE) is concentrated in diffuse as well as cored/neuritic plaques in AD brains, in clear contrast to the localization of AβN3(pE) restricted to dense cores in APP transgenics (4). This is in agreement with the labeling of diffuse plaques with PIB in humans, and may explain an increased [11C]PIB signal in a subpopulation of non-demented elderly subjects. Then one might raise concerns as to whether these AβN3(pE)-rich diffuse plaques are neurotoxic, and whether [11C]PIB-positive non-demented subjects show relatively lower cognitive performances notwithstanding that they do not even meet the criteria for MCI.

Finally, molecular mechanisms implicated in the conversion of Aβ to a more toxic mode could be pursued in rodent and primate models by genetically or pharmacologically manipulating these animals, and could be visualized by PIB, if the toxic components also participate in the formation of PIB binding sites.

References:
1. Kawarabayashi T, Younkin LH, Saido TC, et al. Age-dependent changes in brain, CSF, and plasma amyloid (beta) protein in the Tg2576 transgenic mouse model of Alzheimer's disease. J Neurosci 2001;21:372-381. Abstract

2. Klunk WE, Lopresti BJ, Ikonomovic MD, et al. Binding of the positron emission tomography tracer Pittsburgh compound-B reflects the amount of amyloid-beta in Alzheimer's disease brain but not in transgenic mouse brain. J Neurosci 2005;25:10598-10606. Abstract

3. Toyama H, Ye D, Ichise M, et al. PET imaging of brain with the beta-amyloid probe, [11C]6-OH-BTA-1, in a transgenic mouse model of Alzheimer's disease. Eur J Nucl Med Mol Imaging 2005;32:593-600. Abstract

4. Maeda J, Ji B, Irie T, et al. Longitudinal, quantitative assessment of amyloid, neuroinflammation, and anti-amyloid treatment in a living mouse model of Alzheimer's disease enabled by positron emission tomography. J Neurosci 2007;27:10957-10968. Abstract

5. Schilling S, Zeitschel U, Hoffmann T, et al. Glutaminyl cyclase inhibition attenuates pyroglutamate Abeta and Alzheimer's disease-like pathology. Nat Med 2008;14:1106-1111. Abstract

6. Higuchi M. Visualization of brain amyloid and microglial activation in mouse models of Alzheimer's disease. Curr Alzheimer Res 2009;6:137-143. Abstract

7. Rosen RF, LeVine III H, Pohl J, et al. Mass spectrometric detection of modified cerebral Abeta peptides in Alzheimer’s disease, aged humans, and aged nonhuman primates. Society for Neuroscience 2006, Washington, DC (Program No. 170.22).

View all comments by Makoto Higuchi

AstraZeneca presented the preclinical data on [18F]AZD4694 at the 9th International AD/PD meeting in Prague. In our preclinical studies, AZD4694 shows high affinity to amyloid plaque with very low non-specific interactions with white matter regions devoid of amyloid plaque. This low non-specific background provides a higher contrast and should support the potential to detect very low levels of amyloid. AZD4694 is currently in a Phase 1 study (in collaboration with Karolinska Institutet, Stockholm) with the objective to test clinical utility of this ligand. Our first data look very promising. We are supportive of ADNI, which has already made significant contributions to the field, and we have planned to make our ligand available for the sites in ADNI 2. The elegant study by Rosen et al., 2009, showing that PIB may be selective for pathological human-specific conformation of aggregated Aβ, indicates that we should be cautious when comparing results from different methods for evaluating Aβ plaque load in vitro (i.e., in vitro PET tracer...  Read more

AstraZeneca presented the preclinical data on [18F]AZD4694 at the 9th International AD/PD meeting in Prague. In our preclinical studies, AZD4694 shows high affinity to amyloid plaque with very low non-specific interactions with white matter regions devoid of amyloid plaque. This low non-specific background provides a higher contrast and should support the potential to detect very low levels of amyloid. AZD4694 is currently in a Phase 1 study (in collaboration with Karolinska Institutet, Stockholm) with the objective to test clinical utility of this ligand. Our first data look very promising. We are supportive of ADNI, which has already made significant contributions to the field, and we have planned to make our ligand available for the sites in ADNI 2. The elegant study by Rosen et al., 2009, showing that PIB may be selective for pathological human-specific conformation of aggregated Aβ, indicates that we should be cautious when comparing results from different methods for evaluating Aβ plaque load in vitro (i.e., in vitro PET tracer binding, ELISA for insoluble Aβ, and IHC). The data from Rosen et al. suggest that one should include PIB or another amyloid selective PET tracer when evaluating plaque load in vitro, to support translation to the in-vivo situation. Clearly more work is needed to unravel the molecular mechanism of amyloid ligand binding.

View all comments by Samuel Svensson