Frameshift Mutants of β Amyloid Precursor Protein and Ubiquitin-B in Alzheimer's and Down's Syndrome Patients
Chris Weihl led this live discussion on 28 May 1999. Readers are invited to submit additional comments by using our Comments form at the bottom of the page.
Article under discussion
van Leeuwen FW, de Kleijn DPV, van den Hurk HH, Neubauer A, Sonnemans MAF,
Sluijs JA, Koycu S, Ramdjielal RDJ, Salehi A, Martens GJM, Grosveld FG,
Peter J, Burbach H, Hol EM. Frameshift mutants of β amyloid precursor
protein and ubiquitin-B in Alzheimer's and Down's syndrome patients. Science
1998 Jan 9;279 (5348): 242-247. Abstract
View Transcript of Live Discussion — Posted 6 September 2006
By Chris Weihl
Familial Alzheimer's Disease (FAD) has been associated with a variety of
point mutations in three genes, APP (amyloid precursor protein), PS1 (presenilin
1) and PS2 (presenilin 2). These point mutations are single base pair changes
at the genomic DNA level and are inherited from generation to generation.
The mutated gene thus encodes a mutant protein that results in FAD. Conventional
genetic analysis of AD patients has primarily focused upon the identification
of these types of genomic mutations.
Recently mutations have been discovered in several proteins exclusively
associated with neurodegenerative diseases, that are not found at the genomic
level, suggesting that neurons are capable of altering genetic information
post transcriptionally. It has been known that mammalian neurons are unique
in their ability to "edit" their mRNA transcripts in order to
increase phenotypic variability (O'Connell, 1997). In particular, glutamate
and serotonin receptors are edited at specific bases (adenine to guanine)
creating amino acid changes not found at the DNA level (ibid). While this
form of RNA editing has not been associated with disease, two other novel
forms of post transcriptional RNA editing have recently been associated
with sporadic amyotrophic lateral sclerosis (ALS) and AD.
Rothstein and colleagues identified multiple splice variants in the mRNA
of EAAT2 (excitatory amino acid transporter type 2 abundant in glia) resulting
from aberrant splicing events exclusively in sporadic ALS patients (Lin
et al., 1998). The splice variants were not the result of changes at the
genomic DNA level but were created post-transcriptionally. In the paper
under discussion van Leeuwen and colleagues identified a novel form of RNA
editing resulting in a frameshift mutation of transcribed mRNAs that was
not found at the genomic DNA level and was highly specific for AD and Down's
syndrome patients (van Leeuwen et al., 1998).
In order to understand the potential mechanisms involved in these pathogenic
RNA editing events, it is necessary to review the steps involved in cellular
protein synthesis. In brief, genomic DNA is transcribed into pre-mRNA which
contains exons (protein coding sequences) and introns (non coding sequences)
in the nucleus of a cell. Introns are then spliced from the pre-mRNA via
a complex mechanism mediated through a spliceosome into mRNA. mRNA is then
shuttled to the cytoplasm and translated via ribosomes into functional proteins.
Hence alterations that result in mutant protein can be introduced at any
of the aforementioned steps. In addition, all of the events involved in
protein synthesis are mediated by multiple protein complexes, and mutations
in any of these proteins may result in a mutant protein. In fact two other
neurodegenerative diseases, spinal muscular atrophy (SMA) and fragile X
syndrome (FMRP), are the result of mutations in the proteins SMN (associated
with splicing machinery) and FRAXA (associated with RNA trafficking from
the nucleus to the cytoplasm) respectively, suggesting that post-transcriptional
mRNA changes may be a general mechanism in the pathogenesis of disease (Feng
et al., 1997; Liu et al., 1997; Siomi et al., 1993).
Discussion of paper
Previous studies have alluded to a novel mechanism of mutant protein reversion
in vasopressin deficient rats (VP-/-) (Evans et al., 1994). Genetic analysis
revealed that VP-/- rats contain a GA deletion in a GAGAG motif in the vasopressin
allele resulting in an out of frame nonfunctional vasopressin protein. However,
a small number of magnocellular neurons (the neurons that secrete vasopressin)
express immunoreactivity for both the mutant and normal vasopressin protein
in VP-/- rats. Further characterization revealed that magnocellular neurons
in VP-/- rats are capable of reverting this phenotype to VP-/+ resulting
in an in-frame vasopressin mRNA transcript and functional protein. Furthermore
wild-type rats also alter their vasopressin transcripts in the opposite
manner creating mutant vasopressin suggesting that this phenomenon was not
induced by the diseased state of VP-/- rats.
The paper under discussion investigates sporadic AD patients' mRNA for
frameshift mutations not found at the genomic level. Van Leeuwen and colleagues
surmised that two proteins associated with AD, APP and ubiquitin-B, that
contained multiple GAGAG motifs, might exhibit frameshift mutants in AD
patients. In order to examine the presence of putative +1 proteins (frameshifted
proteins), they generated antibodies to mutant proteins APP+1 and ubiquitin-B+1
and immunostained tissue sections from AD, Down's and normal brains. Increased
immunoreactivity of the frontal cortex, temporal cortex and hippocampus
were strongly correlated with AD and Down's syndrome brain tissue. In addition,
this correlation was not present in other neurodegenerative diseases or
in unaffected brain regions (i.e. striatum) suggesting specificity to AD
and Down's syndrome. Co-analysis of +1 protein immunoreactivity with neuropathology
suggested that APP+1 and ubiquitin+1 co-existed in the same neurons and
were strongly immunoreactive in dystrophic neurites, neuropil threads and
neurofibrillary tangles. Western analysis of the immunodetected proteins
demonstrated a 38 kDa APP+1 protein and 11 kDa ubiquitin-B+1 protein from
AD brain homogenates corresponding to their expected molecular masses.
In an effort to identify the frequency and the specific mRNA mutations
involved, van Leeuwen performed RT-PCR and sequencing of APP and ubiquitin-B
transcripts in normal and AD patients. As expected, GA deletions, as well
as some GT and CT deletions were found more often in AD and Down's syndrome
brain homogenates than in normal control brains. However, the frequency
of these mutant transcripts was low (1/2000 sequenced transcripts).
It is possible that the mutations found in the mRNA transcripts are due
to mutations at the genomic level in single neurons or mRNA editing during
or after transcription. In order to determine if the GA deletions were present
in neuronal DNA, van Leeuwen performed two rigorous PCR control experiments.
One strategy involved PCR amplification and subsequent sequencing of APP
and ubiquitin-B genes from genomic DNA and the second strategy used mutation
specific PCR primers to detect the GA deletion in APP and ubiquitin-B. Both
strategies failed to detect genomic alterations, supporting the hypothesis
that frameshift deletions in APP and ubiquitin-B occur during or after transcription.
Mechanisms of RNA editing are slowly emerging as potential mediators of
neurodegenerative disease. In particular, the current paper addresses a
novel finding of frameshift deletions present in two proteins associated
with AD. Further investigation into the mechanisms of RNA modification as
well as similar phenomena in other degenerative diseases will provide clues
to the pathogenesis of sporadic AD.
1) Was the crossreactivity of the antibodies tested? Perhaps AD
brains (and dystrophic neurites, etc.) contain non-specific immunoreactive
proteins that were detected by the APP+1 and ubiquitin-B+1 antibodies.
Reply from van Leeuwen: Yes, we extensively tested the crossreactivity
of the antibodies. APP+1 antibody revealed a band of 38 kDa on a Western
blot, as expected by translation of the mRNA sequence into the protein
sequence. Although this might be a degradation product of the APP+1 protein,
as due to the acidic domain in APP one could also expect a MW of 60-70
kDA. However APP+1 produced in stably transfected cells has also an apparent
MW of 38 kDA, as shown with the APP+1 antibody on a Western blot. Ubi B+1
stained an 11 kD protein on the Western blot and in addition, HPLC fractionation
in combination with a radioimmunoassay, showed that recombinant Ubi B+1
co-eluted with Ubi-B+1 immunoreactivity from a temporal cortex homogenate
of an AD patient. We tried immunoprecipitation with the APP+1 antibody
but failed. Therefore we consider trying to immunoprecipitate with 22C11
antibody, which recognizes the the N-terminal part of APP+1. Moreover one
must realize that a band at a certain position does not automatically mean
that the antiserum is specific(see our letter to the Editor in the J.Histochem.Cytochem.25,
2) Was there a correlation between +1 immunoreactivity and age,
considering that aged nondemented patients contain some AD-like pathology?
Reply from van Leeuwen: The next important question is: what
is first? neuropathology or +1 proteins? At the next meeting of the Society
for Neuroscience in Los Angeles, we will present new data on this issue.
A correlation between +1 immunoreactivity and age exists. As soon as UbiB+1
immunoreactivity is apparent, also some neuropathology can be seen. You
can find these details on our website.
3) Have you investigated the presence of other mutant transcripts
not associated with AD that might suggest a more generalized phenomenon
in these patients?
Reply from van Leeuwen: Yes, we investigated the presence of
tubulin immunoreactivity in AD pathology. No staining was obtained. However,
it might still be that we are dealing with a general phenomenon, and that
also in these proteins dinucleotide deletions occur. In that case our choice
of the epitope (to make the antibody) might not have been optimal.
Questions for the authors
1) In your previously published paper "Frameshift mutations
at two hotspots in vasopressin transcripts in post-mitotic neurons"
which addresses the novel mechanism of GA deletions in a GAGAG motif as
a reversion phenomenon in vasopressin-null Brattleboro rats, you argue against
an RNA editing mechanism and suggest that the mutations have occurred in
the genome of solitary neurons. However in strong contrast, the current
report on frameshift mutations associated with AD suggests that these mutations
are generated via a novel RNA editing mechanism. What evidence supports
either a genomic mutation in individual neurons or RNA edited transcripts?
Reply from van Leeuwen: Initially we thought that the mutations
are introduced at the genomic level (see Mutation Research 338 (1995) 173-182,
p. 178). RNA-editing was supposed to take place in all cells and not only
in a few cells. Up till now we have no evidence which is in favor of the
hypothesis of a genomic somatic mutation. We extensively studied the genomic
DNA of Brattleboro rats and of AD and DS patients, but we couldn't find
any mutation in the genomic sequence. In addition, we neither have evidence
that the mutation is taking place at the transcript level. However, circumstantial
evidence led us to conclude that we were dealing with a mutational event
that occurs at the transcript level. First of all, APP+1 and Ubi+1 are
colocalized and it is highly unlikely that 2 somatic mutations in 2 different
transcripts occur in the same neuron. Secondly, in the Brattleboro rat
we found out that substitution with vasopressin resulted in a decrease
in the number of cells expressing vasopressin +1 immunoreactivity. This
is most probably caused by a down-regulation in the transcription of the
2) A recent paper by Rothstein has shown that aberrant splice
mutants of EAAT2 generated post-transcriptionally result in a change in
the normal proteins localization and function when overexpressed in mammalian
cells. Considering that neurons may express both wild type and +1 frameshift
mutants, is there evidence suggesting that APP+1 or ubiquitin-B+1 may elicit
a dominant negative effect on the wild-type proteins? Or perhaps a decrease
in the functional APP and ubiquitin-B protein results in a recessive defect
in the neurons, as you suggest in the paper?
Reply from van Leeuwen: We have no evidence for a dominant negative
effect of APP+1 and Ubi-B+1. For Ubiquitin-B+1 one can imagine that the
+1 protein interferes with the function of the wild-type protein. This
is currently under investigation. In case of APP+1 it is more complex,
as the function of the protein is not known. We started transgenesis studies
and we hope that these studies will provide us with data that can give
an answer on the question of the dominance of the mutation.
3) +1 protein immunoreactivity in non-AD patients as well as one
non-neuropathological Down's syndrome patient demonstrated little presence
of mutant protein. However patients with pathology showed abundant immunoreactivity.
Could the presence of +1 proteins be a downstream effect or consequence
of AD pathology? Do you speculate that a mechanism of RNA frameshift mutations
is causally related to sporadic AD?
Reply from van Leeuwen: As +1 immunoreactivity coincides with
neuropathology (see our website),
it could as well be a cause or consequence of the neuropathology. Studies
with transgenic mice might give the answer to this question. The mechanism
of the frameshift mutations might be causally related to neuropathology,
but it might also be possible that a failure in the mechanisms that normally
degrade aberrant RNA and protein is causally related to AD. In both cases,
ALS and AD a similar failure in the RNA degradation mechanism might lead
to the presence of mutant RNA and might be be the cause of neurodegeneration.
4) Your previous study in Brattleboro rats suggested that wild-type
rats as well as VP-/- rats exhibited the same phenomenon (albeit under opposite
circumstances) of frameshift mutations. Why in the present study do normal
and non-demented controls not show immunoreactivity to APP+1 or ubiquitin+1
antibodies since the phenomenon is not related to disease state as suggested
by the Brattleboro rats.
Reply from van Leeuwen: Note that aged non-demented controls
(>72 years) do show ubiquitinB+1 immunoreactivity (see our
website). The phenomenon is related to overexpression of genes as is
seen in homozygous Brattleboro rats and Down syndrome patients (see also
5) You speculate that the increase in frameshifted proteins in
AD and Down's syndrome brains may be a result of increased transcription
of these proteins. If this is true why does your data not suggest an increase
in APP+1 protein in Down's syndrome (containing trisomy of chromosome 21
and APP) versus AD patient brains?
Reply from van Leeuwen: If you look to table I of the Science
paper you will notice that the percentage of DS patients showing immunoreactivity
for APP+1 in the frontal and temporal cortex and the hippocampus, is much
higher compared to AD patients. In fact the difference is larger if we
exclude the non-demented Down syndrome patient.
6) RNA editing via single base pair changes is mediated by a family
of enzymes known as dsRNA-specific editases. What proposed mechanism is
involved in this novel form of frameshifted RNA editing found in AD patients?
Are flanking RNA sequences important considering that exon 9 GA deletions
were more prevalent than exon 10 GA deletions from APP mRNA?
Reply from van Leeuwen: The mechanism by which the mutation occurs
is not clear to us. We searched for consensus sequences in the RNA flanking
GAGAG-sequence, but we could not find a clear motif that occurred in APP,
UbiquitinB and vasopressin. Instead of RNA-editing the dinucleotide deletion
can be explained by the occurrence of RNA-polymerase slippage or stuttering.
The rate of this process (GA-deletion) can be promoted by increased transcription.
In the case of APP we detected more GA-deletions in exon 9 than in exon
10, this is probably due to an extended GAGAG motif: GAGAGAGA.
Submit a Comment on this Live Discussion