Response to Correspondence: Pardossi-Piquard et al., “Presenilin-Dependent Transcriptional Control of the Aβ-Degrading Enzyme Neprilysin by Intracellular Domains of βAPP and APLP.” Neuron 46, 541–554

Article (PDF Available)inNeuron 53(4):483-6 · March 2007with4 Reads
DOI: 10.1016/j.neuron.2007.01.024 · Source: PubMed
Kimberly, W.T., Zheng, J.B., Guenette, S.Y.,
and Selkoe, D.J. (2001). J. Biol. Chem. 276,
40288–40292.
Kimberly, W.T., LaVoie, M.J., Ostaszewski,
B.L., Ye, W., Wolfe, M.S., and Selkoe, D.J.
(2002). J. Biol. Chem. 277, 35113–35117.
Leem, J.Y., Vijayan, S., Han, P., Cai, D., Ma-
chura, M., Lopes, K.O., Veselits, M.L., Xu, H.,
and Thinakaran, G. (2002). J. Biol. Chem.
277, 19236–19240.
Pardossi-Piquard, R., Petit, A., Kawarai, T.,
Sunyach, C., Alves da Costa, C., Vincent, B.,
Ring, S., D’Adamio, L., Shen, J., Muller, U.,
et al. (2005). Neuron 46, 541–554.
Seiffert, D., Bradley, J.D., Rominger, C.M., Ro-
minger, D.H., Yang, F., Meredith, J.E., Jr., Wang,
Q., Roach, A.H., Thompson, L.A., Spitz, S.M.,
et al. (2000). J. Biol. Chem. 275, 34086–34091.
Welches, W.R., Brosnihan, K.B., and Ferrario,
C.M. (1993). Life Sci. 52, 1461–1480.
Zhang, Z., Nadeau, P., Song, W., Donoviel, D.,
Yuan, M., Bernstein, A., and Yankner, B.A.
(2000). Nat. Cell Biol. 2, 463–465.
Response to Correspondence: Pardossi-Piquard
et al., ‘‘Presenilin-Dependent Transcriptional
Control of the Ab-Degrading Enzyme Neprilysin by
Intracellular Domains of bAPP and APLP.’’
Neuron 46, 541–554
Raphae
¨lle Pardossi-Piquard,
1
Julie Dunys,
1
Toshitaka Kawarai,
2
Claire Sunyach,
1
Cristine Alves da Costa,
1
Bruno Vincent,
1
Jean Se
´valle,
1
Sanjay Pimplikar,
3
Peter St George-Hyslop,
2
and Fre
´de
´ric Checler
1,
*
1
IPMC, UMR6097 CNRS/UNSA, Equipe labellise
´e FRM, 660 Route des Lucioles, 06560 France
2
Centre for Research in Neurodegenerative Diseases, University of Toronto, Toronto, ON, M5S 3H2 Canada
3
Department of Pathology and Cell Biology Program, Case Western Reserve University, Cleveland, OH 44106, USA
*Correspondence: checler@ipmc.cnrs.fr
DOI 10.1016/j.neuron.2007.01.024
We recently established that neprilysin
(NEP), one of the putative amyloid
b-peptide (Ab)-degrading enzymes,
is transcriptionally upregulated by
AICDs, the C-terminal fragments of
the b-amyloid precursor protein
(bAPP/APLP) that are generated by
presenilin (PS)-dependent g- and 3-
secretase activities (Pardossi-Piquard
et al., 2005). Chen and Selkoe now re-
port lack of evidence of NEP upregula-
tion by PS-dependent g-secretase. To
explore the reasons for this difference,
we have repeated several experi-
ments, conducted new experiments,
and carefully examined the data pre-
sented by Chen and Selkoe. These
analyses reveal that (1) there is signifi-
cant consensus in key experimental
data; (2) several of Chen and Selkoe’s
negative conclusions are overreach-
ing, arising from a crucial conceptual
error on their part; and (3) there is com-
pelling, independent data from experi-
ments in nicastrin null cells and in brain
from transgenic mice overexpressing
AICD that support the notion that pre-
senilin complexes and AICD modulate
inducible NEP expression.
Results and Discussion
There is consensus on several impor-
tant points. Both groups find that NEP
expression is dramatically reduced in
PS1
/
:PS2
/
double-knockout
mouse embryonic fibroblasts (MEFs)
(Figures 1D and 1E in Pardossi-Pi-
quard et al., 2005;Figures 1A and 1B
in this paper; Figure 1C in Chen and
Selkoe, 2007). Both groups find mod-
est reductions (%25%) in NEP expres-
sion in PS1
/
:PS2
/
double-knock-
out BD8 blastocyst cells (Figures 1F
and 1G in Pardossi-Piquard et al.,
2005; Figure 1A in Chen and Selkoe,
2007). Both groups observe either
modest or no consistent change in
NEP expression in single-knockout
PS1
/
,PS2
/
,APP
/
, and APLP-
2
/
MEFs (Figures 3C and 6C in
Pardossi-Piquard et al., 2005;Figures
1A and 1B in this paper; Figures 1C
and 4C in Chen and Selkoe, 2007).
Repetition of several of our experi-
ments using different MEF cell lines
and different anti-NEP antibodies re-
veals that the results are unchanged
(Figures 1A and 1B). This argues
against an artifact arising from labora-
tory error or from differences in clones.
We have also pursued additional
experiments, both of which strongly
support the validity of our initial report.
First, we investigated NEP activity,
protein, and mRNA levels in nicastrin
knockout MEFs (NCT
/
), which do
not form functional presenilin com-
plexes and which have no g-secretase
activity (Yu et al., 2000). NEP expres-
sion, NEP activity, and NEP mRNA
levels were all dramatically reduced
in nicastrin knockout MEFs (NCT
/
)
(Pardossi-Piquard et al., 2006)(Fig-
ure 1C). However, NEP expression is
restored in these NCT
/
MEFs by
Neuron 53, February 15, 2007 ª2007 Elsevier Inc. 483
Neuron
Correspondence
transfection with NCT-
cDNA (Figure 1C). This ef-
fect was specific because
NCT cDNAs do not restore
NEP expression in PS
/
fi-
broblasts (Figure 1C). These
data are fully consistent
with the in vitro data from
both our group and from
Chen and Selkoe, which
show downregulation of
NEP in g-secretase-defec-
tive PS1
/
:PS2
/
double-
knockout MEF cells.
Second, we examined
NEP expression in the brain
of transgenic mice overex-
pressing Fe65 and AICD
(Ryan and Pimplikar, 2005).
AICD:Fe65 double-trans-
genic mice displayed en-
hanced cerebral expression
of NEP compared to Fe65
single-transgenic mice (Fig-
ure 2). These data therefore
reinforce our conclusion
that NEP can be transcriptionally con-
trolled by AICDs generated by g-sec-
retase-cleavage of bAPP and its family
members (i.e., APLPs).
Taken together, the experimental
data described above and our prior
observations (several components of
which are congruent with those of
Chen and Selkoe) support the conclu-
sion that functional presenilin/g-secre-
tase complexes mediate inducible
NEP expression via AICDs. However,
inspection of the experimental as-
sumptions and the experimental de-
sign used by Chen and Selkoe reveals
several explanations for their different
interpretation of similar experimental
data.
First, Chen and Selkoe have made
a fundamental and critic conceptual er-
ror. They failed to take into account the
fact that inducible NEP expression is
modulated by AICD products that are
derived by a highly redundant system
(Pardossi-Piquard et al., 2005). Thus,
AICDs can be generated from either
APP or the APLP homologs. g-secre-
tase activity is mediated by either PS1
or PS2. Recognition of the simulta-
neous redundancy in the presenilins
and in AICD signaling is critical to the
correct interpretation of experiments
in single-knockout cells (PS1
/
or
PS2
/
or APP
/
or APLP-2
/
only)
versus double-knockout cells (PS1
/
:
PS2
/
or APP
/
:APLP-2
/
). Thus,
Chen and Selkoe point out that PS1
/
fibroblasts did not mimic the reduc-
tion of NEP in PS1
/
:PS2
/
cells
(Figure 1C, lanes 4–6 versus 10–12 of
Chen and Selkoe, 2007) and that this
constitutes evidence that NEP expres-
sion is not modulated by PS1. The
observation itself is not in doubt. Our
data also show that NEP expression
is not altered in PS1
/
-only MEFs
(Figure 3C in Pardossi-Piquard et al.,
2005). However, the conclusion that
NEP expression is not modulated by
PS1 is overreaching and incorrect
because it makes the assumption that
the residual PS2 has no activity. While
PS1 may be ‘‘dominant’’ over PS2
for the g-secretase cleavage of APP,
it overlooks the fact that inducible
NEP expression is also regulated by
g- and 3-cleavage of APLPs. There
are no data to show that PS1 and PS2
differ in terms of g- and 3-cleavage of
APLPs.
The same conceptual error afflicts
the interpretation of Chen and Sel-
koe’s results for APP
/
and APLP
/
single-knockout mice. In these sin-
gle-knockout tissues, the
failure to observe a change
in NEP levels is more parsi-
moniously explained by re-
dundant and persistent PS-
signaling via AICDs from
the other APP/APLP family
members. Indeed, although
we reported an 25% de-
crease in NEP expression
in APP
/
or APLP-2
/
sin-
gle-knockout MEFs, as
would be expected, major
reductions in NEP expres-
sion were only seen in
our original report in
APP
/
:APLP-2
/
double-
knockout MEFs (Figures 6C
and 6D in Pardossi-
Piquard et al., 2005).
A second source of differ-
ences in interpretation de-
rives from the cell lines cho-
sen by Chen and Selkoe for
their work. They have based
much of their work on blas-
tocyst-derived cells. We deliberately
chose not to use BD8 cells as our
main cell system for two reasons. First,
the data from both groups clearly
show that (1) blastocyst-derived cells
(including both wild-type PBD8 cells
and PS1
/
:PS2
/
double-knockout
BD8 cells) have high constitutive levels
of NEP expression and (2) there is only
a minimal but consistent reduction
(<25%) in inducible expression
in PS1
/
:PS2
/
double-knockout
BD8 cells (Figures 1F and 1G of
Pardossi-Piquard et al., 2005; Figure 1A
of Chen and Selkoe, 2007). The small
proportion of NEP that is PS inducible
and the high level of constitutive NEP
expression clearly make it difficult to
reliably detect small (<25%) changes
in inducible NEP expression. The sec-
ond reason that we chose not to use
PBD8 and BD8 is that they have poor
and variable transfection efficiencies.
Inspection of Figure 1B from Chen
and Selkoe reveals clear evidence of
such variability in BD8 cells. This trans-
fection variability has significant ef-
fects on PS1 expression and g-secre-
tase complex formation. For instance,
in lanes 8–9 in Figure 1B, which are la-
beled as BD8 cells transfected with
both PS1 and PS2 cDNA, there is in
Figure 1. NEP Expression in HEK293 Cells (293), Wild-Type
(PS
+/+
), and PS-Deficient (PS
/
) and NCT-Deficient Murine
Embryonic Fibroblasts
(A) Comparison of NEP in indicated cell lines by Western blot by means
of two monoclonal antibodies, 56C6 (Chen and Selkoe) and 18B5 (Par-
dossi-Piquard et al., 2005). (B) The same differences in neprilysin ex-
pression were robustly observed in homogenates from wild-type or
PS-deficient fibroblasts, regardless of the detergent used. (C) Wild-
type MEFs with intact presenilin complexes (i.e., NCT
+/+
or PS
+/+
)
have abundant NEP expression and activity. PS
/
and NCT
/
knockout MEFs both lacking functional presenilin complexes and
lacking g-secretase activity have only low levels of constitutive NEP
expression and activity. However, transfection with V5-tagged NCT
cDNA allows selective recovery of NEP expression and activity in
NCT
/
but not in PS
/
MEFs. Error bars show means ± SEM of three
to five experiments.
484 Neuron 53, February 15, 2007 ª2007 Elsevier Inc.
Neuron
Correspondence
fact (1) essentially no PS1
expression, (2) only a very
small percentage of nicas-
trin in maturely glycosylated
forms (a marker of incorpo-
ration into mature functional
PS complexes), and (3) only
minimal decrease in
APP-CTFs, which repre-
sents at best a 30% reduc-
tion from baseline and is
not better than PS2 comple-
mentation alone (N.B. PS2
does not normally fully com-
plements APP processing in
PS1 deficiency). The impor-
tant issue here, however, is
not the PS1 and PS2 levels
but whether biologically ef-
fective g-secretase activity
has been restored in rele-
vant subcellular compart-
ments by their transfection.
It is apparent from both the
minimal levels of maturely
glycosylated NCT and the
small reduction in APP-CTF
levels in lanes 4–9 of Figure 1B of
Chen and Selkoe that g-secretase ac-
tivity has been only poorly restored in
these transfected BD8 cells. This fail-
ure to robustly restore g-secretase ac-
tivity severely limits the negative con-
clusions that can be safely drawn
from BD8 cells. Despite these two con-
founders, Chen and Selkoe derive
much of their data and conclusions
from BD8 blastocyst-derived cells.
A third source of concern relates to
the magnitude of the change in PS-
dependent signaling that is necessary
to produce a discernable change in
inducible NEP expression. Because
the threshold for induction of NEP is
unknown (and is likely cell-type spe-
cific), reasonable grounds for rejection
of the hypothesis that PS-dependent
signaling modulates inducible NEP
expression would be the demonstra-
tion of no change in NEP expression
following unambiguous changes in
PS-dependent signaling. Many of the
PS-complementation and g-secretase
inhibitor studies in Chen and Selkoe do
not achieve robust changes in PS-
dependent signaling. For instance,
high expression of PS1 in PS1
/
:
PS2
/
BD8 cells in Figure 1B of
Chen and Selkoe or in PS1
/
:PS2
/
MEFs in their Figure 1D caused (1)
only minimal induction of maturely
glycosylated nicastrin and (2) only min-
imal reduction in APP-CTFs (<30% re-
duction) compared to untransfected
PS1
/
:PS2
/
cells. It is unclear that
significant amounts of g/3-secretase
activity were restored to biologically
relevant compartments in these
knockout cells. Similar concerns arise
about many of their g-secretase inhib-
itor studies (Figure 2A and Figure S2 of
Chen and Selkoe, 2007). In many of
these studies, the post-dose accumu-
lation of APP-CTF (a marker of the effi-
ciency of g-secretase activity) is at
best minimal. The same argument
also pertains to the AICD transfection
studies (Figures 3A and 3B of Chen
and Selkoe), where AICD is at trace
levels in most cells, even when triple
transfected with Fe65+Tip60+AICD.
There are, however, two figures
from Chen and Selkoe that might, at
first pass, be taken to support their
interpretation. Figure 2C of Chen and
Selkoe appears to show no decrease
in NEP levels in MEF cells treated
with the g-secretase inhibitor DAPT,
despite a robust increase in APP-
CTFs. However, closer in-
spection reveals that this
experiment is seriously
flawed and is uninterpret-
able without additional con-
trols. Specifically, the un-
treated neprilysin knockout
MEFs show APP-CTFs
levels that are identical to
those in MEFs treated with
g-secretase inhibitors (com-
pare Chen and Selkoe,
2007: Figure 2C, lane 13
versus lanes 7–12). This
makes little sense and
throws into question the
design of this experiment,
raising the possibility that
DMSO treatment of the
‘‘control’’ MEFs (Chen and
Selkoe, 2007: Figure 2C,
lanes 1–6) has had some
complex toxic effect. Even
if one disregards this, since
the amount of DAPT used
by Chen and Selkoe is low
(500 nM versus 2 mM in our
study), it is far from clear that this
experiment will have significantly
affected 3-cleavage-dependent pro-
duction of AICDs. Data on AICD levels
were not shown.
Similar technical concerns pertain to
Figure 3C of Chen and Selkoe, which
appears to show robust expression of
AICD (C60) in HEK293T cells triple
transfected with AICD/C60, Fe65, and
Tip60 (Chen and Selkoe, 2007):
Figure 3C, lanes 11 and 12). However,
the identity of this band is unclear.
Close inspection of the AICD Western
blot in Figure 3C reveals an AICD/C60
immunoreactive band pattern that is
very different from the AICD/C60 pat-
tern observed with the same antibody
in their other figures (Figures 2A, 3A,
and 3B of Chen and Selkoe). The latter
figures all show a dark invariant band
and a light AICD band beneath it.
Even if this concern is dismissed,
Chen and Selkoe have not proven
that the highly labile AICD transfected
into these cells was delivered to the nu-
cleus. Without this positive control, it is
premature to conclude that transfec-
tion of cells with AICD had no effect
on NEP transcription. By contrast, we
clearly showed that transcriptional
Figure 2. NEP Expression Is Increased in Fe65/AICD
Transgenic Mice Brains
Expression of NEP (A), Fe65 (A), and AICD (B) were analyzed as de-
scribed in the Supplemental Data in single (Fe65) or doubly transgenic
(Fe65/AICD) mice brains.
Neuron 53, February 15, 2007 ª2007 Elsevier Inc. 485
Neuron
Correspondence
upregulation of NEP was accompanied
by nuclear AICD immunoreactivity in
AICD-transfected cells (Figure 5D in
Pardossi-Piquard et al., 2005).
In their final paragraph, Chen and
Selkoe argue on theoretical grounds
that PS-dependent induction of NEP
via AICD is improbable because nepri-
lysin degrades other biologically active
peptides, because AICD is also a prod-
uct of non-Ab-generating proteolysis
of APP, and because there is unlikely
to have been evolutionary pressure
for the development of a specific
mechanism just to upregulate Abdeg-
radation by NEP. None of these theo-
logical arguments compellingly dis-
miss the biological evidence that
AICDs, generated by PS-dependent
g-secretase activity, do in fact modu-
late NEP transcription in some cells.
Without knowing the functions of
APP/APLP and of their ligands, and
without knowing the identity of the
downstream targets of AICD and Ab,
it is difficult to know what underlying
biological purpose was originally in-
tended in evolution for this pathway.
However, the rather narrow Ab/APP-
centered interpretation proposed by
Chen and Selkoe seems unnecessary.
It is more probable that AICD-medi-
ated upregulation of NEP is designed
to allow proteolytic regulation of a vari-
ety of biologically active peptides, in-
cluding Ab.
On balance, therefore, given the
persistence of experimental support
both in vitro (recovery of reduced
NEP expression in NCT
/
cells by
transfection with NCT) and in vivo
(higher NEP levels in brain tissue from
mice overexpressing AICD), our inter-
pretation remains that many, but per-
haps not all, cells show PS-dependent
inducible NEP expression.
Supplemental Data
The Supplemental Data for this article can be
found online at http://www.neuron.org/cgi/
content/full/53/4/483/DC1/.
ACKNOWLEDGMENTS
R.P.-P. is supported by the the Fondation pour
la Recherche Me
´dicale. J.D. is supported by the
APOPIS integrated project, and C.S. is funded
by the association France Alzheimer. This
work was supported by the Centre National de
la Recherche Scientifique, by an EU contract
LSHM-CT-2003-503330 (APOPIS), and by The
Fondation pour la Recherche Me
´dicale.
REFERENCES
Chen, A.C., and Selkoe, D.J. (2007). Neuron
53, this issue, 479–483.
Pardossi-Piquard, R., Dunys, J., Yu, G., St
George-Hyslop, P., Alves da Costa, C., and
Checler, F. (2006). J. Neurochem. 97, 1052–
1056.
Pardossi-Piquard, R., Petit, A., Kawarai, T.,
Sunyach, C., Alves da Costa, C., Vincent, B.,
Ring, S., D’Adamio, L., Shen, J., Muller, U.,
et al. (2005). Neuron 46, 541–554.
Ryan, K.A., and Pimplikar, S.W. (2005). J. Cell
Biol. 171, 327–335.
Yu, G., Nishimura, M., Arawaka, S., Levitan, D.,
Zhang, L., Tandon, A., Song, Y.-Q., Rogaeva,
E., Chen, F., Kawarai, T., et al. (2000). Nature
407, 48–54.
486 Neuron 53, February 15, 2007 ª2007 Elsevier Inc.
Neuron
Correspondence
    • "Specifically, work by De Strooper and colleagues (Hebert et al., 2006) has shown that expression of several previously defined AICD targets genes was at best indirectly and weakly affected by APP processing. Further, the role of AICD in regulating neprilysin expression (Pardossi-Piquard et al., 2005;Pardossi-Piquard et al., 2007) was not reproduced by another report (Chen and Selkoe, 2007). Finally, a recent study has shown that secreted APP ectodomain APPsα is sufficient to rescue several anatomical, behavioral, and electrophysiological abnormalities seen in APP-KO mice (Ring et al., 2007). "
    [Show abstract] [Hide abstract] ABSTRACT: Mutations in the amyloid precursor protein (APP) cause early-onset Alzheimer's disease (AD), but the only genetic risk factor for late-onset AD is the varepsilon4 allele of apolipoprotein E (apoE), a major cholesterol carrier. Using Cre-lox conditional knockout mice, we demonstrate that lipoprotein receptor LRP1 expression regulates apoE and cholesterol levels within the CNS. We also found that deletion of APP and its homolog APLP2, or components of the gamma-secretase complex, significantly enhanced the expression and function of LRP1, which was reversed by forced expression of the APP intracellular domain (AICD). We further show that AICD, together with Fe65 and Tip60, interacts with the LRP1 promoter and suppresses its transcription. Together, our findings support that the gamma-secretase cleavage of APP plays a central role in regulating apoE and cholesterol metabolism in the CNS via LRP1 and establish a biological linkage between APP and apoE, the two major genetic determinants of AD.
    Full-text · Article · Nov 2007
    • "AICD, together with its binding partners Fe65 and Tip60, is considered to be involved in transcriptional regulation (Cao and Sudhof, 2001). Putative target genes of AICD signaling have been suggested (Baek et al., 2002; Kim et al., 2003; von Rotz et al., 2004; Pardossi-Piquard et al., 2005; Ryan and Pimplikar, 2005; Muller et al., 2007), although results for some of these genes are controversial (Hass and Yankner, 2005; Hebert et al., 2006; Chen and Selkoe, 2007; Pardossi-Piquard et al., 2007). One potential AICD target gene is the A␤-degrading enzyme neprilysin (Pardossi-Piquard et al., 2005, 2006), a metalloprotease that is one of the main A␤-degrading enzymes in the brain (Carson and Turner, 2002). "
    [Show abstract] [Hide abstract] ABSTRACT: Amyloid-beta (Abeta) deposition is a major pathological hallmark of Alzheimer's disease. Gleevec, a known tyrosine kinase inhibitor, has been shown to lower Abeta secretion, and it is considered a potential basis for novel therapies for Alzheimer's disease. Here, we show that Gleevec decreases Abeta levels without the inhibition of Notch cleavage by a mechanism distinct from gamma-secretase inhibition. Gleevec does not influence gamma-secretase activity in vitro; however, treatment of cell lines leads to a dose-dependent increase in the amyloid precursor protein intracellular domain (AICD), whereas secreted Abeta is decreased. This effect is observed even in presence of a potent gamma-secretase inhibitor, suggesting that Gleevec does not activate AICD generation but instead may slow down AICD turnover. Concomitant with the increase in AICD, Gleevec leads to elevated mRNA and protein levels of the Abeta-degrading enzyme neprilysin, a potential target gene of AICD-regulated transcription. Thus, the Gleevec mediated-increase in neprilysin expression may involve enhanced AICD signaling. The finding that Gleevec elevates neprilysin levels suggests that its Abeta-lowering effect may be caused by increased Abeta-degradation.
    Full-text · Article · Oct 2007
  • [Show abstract] [Hide abstract] ABSTRACT: Epidemiological studies suggest that a high intake of polyunsaturated fatty acids, such as docosahexaenoic acid (DHA), is associated with a reduced risk of Alzheimer's disease. Here, we examined the effects of DHA on amyloid precursor protein (APP) processing in cellular models of Alzheimer's disease by analysing levels of different APP fragments, including amyloid-beta (Abeta). DHA administration stimulated non-amyloidogenic APP processing and reduced levels of Abeta, providing a mechanism for the reported beneficial effects of DHA in vivo. However, an increased level of APP intracellular domain was also observed, highlighting the need to increase our knowledge about the relevance of this fragment in Alzheimer's disease pathogenesis. In conclusion, our results suggest that the proposed protective role of DHA in Alzheimer's disease pathogenesis might be mediated by altered APP processing and Abeta production.
    Full-text · Article · Sep 2007
Show more