Human Molecular Genetics, 2017, Vol. 26, No.

22 4519–4529

doi: 10.1093/hmg/ddx329
Advance Access Publication Date: 26 August 2017
Association Studies Article

ASSOCIATION STUDIES ARTICLE

Pseudoexfoliation and Alzheimer’s associated CLU risk

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

variant, rs2279590, lies within an enhancer element
and regulates CLU, EPHX2 and PTK2B gene expression
Biswajit Padhy1, Bushra Hayat1, Gargi Gouranga Nanda1,
Pranjya Paramita Mohanty2 and Debasmita Pankaj Alone1,*
1
School of Biological Sciences, National Institute of Science Education and Research (NISER) Bhubaneswar, P.O.
Jatni, Khurda, 752050, India, Homi Bhabha National Institute, Training school complex, Anushakti Nagar, Mumbai
400094, India and 2JPM Rotary Club of Cuttack Eye Hospital and Research Institute, Cuttack 753014, India
* To whom correspondence should be addressed at: School of Biological Sciences, Molecular Genetics Laboratory, National Institute of Science Education
and Research (NISER), P.O. Jatni, Khurda 752050, Odisha, India. Tel: þ91 6742494204; Fax: þ916742494004; Email: debasmita@niser.ac.in

Abstract
Genetic variants at PTK2B–CLU locus pose as high-risk factors for many age-related disorders. However, the role of these variants
in disease progression is less characterized. In this study, we aimed to investigate the functional significance of a clusterin
intronic SNP, rs2279590, that has been associated with pseudoexfoliation, Alzheimer’s disease (AD) and diabetes. We have previ-
ously shown that the alleles at rs2279590 differentially regulate clusterin (CLU) gene expression in lens capsule tissues. This
polymorphism resides in an active regulatory region marked by H3K27Ac and DNase I hypersensitive site and is an eQTL for CLU
expression. Here, we report the presence of an enhancer element in surrounding region of rs2279590. Deletion of a 115 bp
intronic region flanking the rs2279590 variant through CRISPR-Cas9 genome editing in HEK293 cells demonstrated a decreased
clusterin gene expression. Electrophoretic mobility shift and chromatin immunoprecipitation assays show that rs2279590 with
allele ‘A’ constitutes a transcription factor binding site for heat shock factor-1 (HSF1) but not with allele ‘G’. By binding to allele
‘A’, HSF1 abrogates the enhancer effect of the locus as validated by reporter assays. Interestingly, rs2279590 locus has a wide-
spread enhancer effect on two nearby genes, protein tyrosine kinase 2 beta (PTK2B) and epoxide hydrolase-2 (EPHX2); both of
which have been previously associated with AD as risk factors. To summarize, our study unveils a mechanistic role of the
common variant rs2279590 that can affect a variety of aging disorders by regulating the expression of a specific set of genes.

Introduction
The former two conditions, PEX (OMIM: 177650) and AD (OMIM:
Clusterin or Apolipoprotein-J (8p21.1) is a multifunctional pro- 104300), share similar pathological alterations like characteristic
tein with diverse roles. Stabilization of cell–cell and cell–matrix deposition of fibrillar protein aggregates and gradual deteriora-
interactions, lipid transport, apoptosis, inhibition of comple- tion of optic and brain nerves, respectively. Studies have shown
ment activation and clearance of extracellular deposits are that Clusterin accumulation differentiates the severe advanced
some of the crucial functions attributed to clusterin (CLU) (1,2). stage of PEX called pseudoexfoliation glaucoma (PEXG) from the
CLU has been associated at both genomic and proteomic levels less severe syndrome stage, pseudoexfoliation syndrome (PEXS)
with age-related disorders, such as, pseudoexfoliation (PEX), (3). Additionally, brain function in AD-affected individuals with
Alzheimer’s disease (AD) and Type 2 Diabetes mellitus (3–7). over accumulated Clusterin has been reported to deteriorate

Received: May 16, 2017. Revised: August 7, 2017. Accepted: August 22, 2017
C The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
V

4519
 4520 | Human Molecular Genetics, 2017, Vol. 26, No. 22

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

Figure 1. rs2279590 is located within an enhancer element as indicated by active regulatory marks. Region around rs2279590 viewed in UCSC genome browser using
ENCODE data. Three selected tracks: DNase hypersensitivity site, H3K27Ac (Histone H3 acetylated at lysine 27) and TFBS (Transcription factor binding site) are shown.
Location of rs2279590 is indicated by a central vertical grey line.

faster than normal individuals (8–10). It is however unclear, neighbouring AD associated candidate genes, protein tyrosine
whether CLU accumulation in such disorders is the cause or a kinase 2 beta (PTK2B) and epoxide hydrolase-2 (EPHX2); suggest-
consequence. Although studies suggest that CLU plays a crucial ing a distal enhancer effect contributed by this locus harbouring
role in preventing extracellular deposits; it also has been shown rs2279590. In conclusion, our study suggests that a common
that knockout of clusterin in PDAPP mice (AD mice model) risk variant at the CLU locus has a widespread enhancer effect
have reduced fibrillar plaque and neuritic dystrophy (11). in regulating multiple candidate genes associated with PEX/AD
Furthermore, an alternative truncated form of Clusterin called and assists in the pathogenesis of such aging disorders.
nuclear Clusterin (nCLU), is known to induce apoptosis by act-
ing as a pro-death protein (12). Although, nCLU level is in low
abundance in a healthy cell, stress induced increase in its Results
expression can lead to apoptosis of affected neurons; a patho-
Locus containing rs2279590 acts as an enhancer and
genic feature for both PEX and AD.
regulates clusterin gene expression
At the genomic level, common variants within CLU locus also
have been associated with PEX and AD (3,4,6). A variant, Using the data from ENCODE (Encyclopedia of DNA Elements)
rs2279590 within the clusterin gene is of particular importance as project in UCSC genome browser, we observed that the region
it has been picked up as a genetic risk factor for PEX, AD and dia- surrounding the polymorphic site, rs2279590 located in 7th
betes (3,4,6). Earlier, we have shown that individuals homozygous intron of clusterin gene, is within a DNase I hypersensitive site
for the risk allele ‘G’ at rs2279590 have a 2-fold higher clusterin and has modified H3K27Ac mark, suggesting the presence of
expression in lens capsule tissues than those with allele ‘A’ (3); an active regulatory region (Fig. 1). In addition, eQTL data from
suggesting a regulatory role of rs2279590 over CLU expression. GTEx project also indicate a regulatory role of rs2279590 over
Association of rs2279590 with such diverse disorders and its loca- CLU expression in both muscle-skeletal and skin-sun exposed
tion within a high-risk locus demands further investigation into tissue samples examined (Supplementary Material, Fig. S1).
its functional implications in disease progression. We, therefore, cloned a 201 bp region surrounding rs2279590
The current study focuses on finding the functional role, if into a luciferase reporter vector to check its regulatory effect
any, of rs2279590 in clusterin gene regulation. We observed that in HEK293 cells. As shown in Figure 2A, the luciferase activity
the genomic region surrounding rs2279590 is located in an was significantly higher (2-fold, P ¼ 0.03) in constructs con-
active locus marked by H3K27Ac and DNase I hypersensitivity taining the rs2279590 locus as compared with the cells with
site (ENCODE project) and is a quantitative trait loci (eQTL) for empty luciferase vector; implying a regulatory effect of this
CLU expression as evident through GTEx project. Luciferase locus.
reporter assays validated the rs2279590 region as a regulatory In addition, to check the in vivo effect of the locus on clus-
element. Deletion of a 115 bp genomic region flanking rs2279590 terin gene expression, we generated HEK293115/ cells in
polymorphic site by CRISPR/Cas9 system, leads to downregula- which 115 bp genomic region around rs2279590 was deleted
tion of clusterin gene expression in HEK293 cells; suggesting using CRISPR/Cas9 system (Fig. 2B and C). Two independent
that the variant resides within an enhancer element. Transition single-cell derived homozygous clones were subsequently
from risk allele ‘G’ to ‘A’ at rs2279590 was found to create a tran- used to check for CLU gene expression compared with that
scription factor binding site (TFBS) for heat shock factor 1 (HSF1) of control non-deleted cells. Quantitative real-time PCR (qRT-
that abrogated the enhancer activity of the region. The locus PCR) assays (Fig. 2D) confirmed a significant downregulation
surrounding this variant was also found to regulate two (P ¼ 0.01) of transcripts for secretory form of clusterin (sCLU)
 Human Molecular Genetics, 2017, Vol. 26, No. 22 | 4521

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

Figure 2. Enhancer effect of rs2279590 locus on clusterin gene expression. (A) Normalized luciferase activity (214.4 6 28.2) is shown for reporter construct containing
the region surrounding rs2279590 (201 bp with major allele ‘G’) compared with that of empty vector (100 6 21.9) (P ¼ 0.03). (B) A representative figure showing deletion of
115 bp region around rs2279590 by using a pair of sgRNA through CRISPR/Cas9 genome editing method in HEK293 cells which are heterozygote (AG) for rs2279590. (C)
Gel picture depicting confirmed deletion of a 115 bp region around rs2279590 generating homozygous knockout HEK293 cells. (D) qRT-PCR assay showing significant
downregulation with a P value of 0.01 for secretory form of clusterin (sCLU) in cells with deletion (0.36 6 0.1) compared with that of non deleted cells (1 6 0.02). (E)
Western blotting also showed a significant downregulation of CLU protein in cells with deletion than that of control non deleted cells. (F) An alternate transcript form
of clusterin called as nuclear clusterin (nCLU) is also found to be significantly downregulated (P ¼ 0.02) in cells with deletion (0.47 6 0.12) compared with that of non
deleted cells (1 6 0.05) is shown through qRT-PCR assay. Het and Homo KO corresponds to homozyogus and heterozygous knockout deletions, respectively. All experi-
ments were performed at least three times and values are represented as mean 6 SEM. Student’s t-test was used to calculate statistical significance between groups,
*P < 0.05.

Transition from G!A at rs2279590 creates a binding site

in HEK293115/ cells versus non-deleted cells and validated
through western blot analysis (Fig. 2E). Furthermore, we have
for HSF1
also checked the effect of this locus on another transcript var- Suspecting the presence of a transcription factor binding site
iant of clusterin coding for a proapoptotic nuclear form of (TFBS) at rs2279590, we analysed its surrounding region by online
clusterin or nCLU which only expresses during high level of TFSEARCH program. In silico results showed that heat shock fac-
cytotoxic stress (e.g. ionic radiation) including lethal heat tors (HSFs) specifically bound to the flanking region of rs2279590
shock (13). nCLU expression was also found to be downregu- with allele ‘A’ but not with allele ‘G’. To validate the same, elec-
lated (Fig. 2F) in rs2279590 locus deleted cells (HEK293115/) trophoretic mobility shift assays (EMSAs) were performed using
similar to its secretory isoform, suggesting a regulatory effect nuclear extracts from HEK293 cells and a 29 bp labelled oligo iden-
of the said locus on both sCLU and nCLU expressions. tical to surrounding genomic region at rs2279590 with allele ‘A’. It
 4522 | Human Molecular Genetics, 2017, Vol. 26, No. 22

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

Figure 3. HSF1 binds to allele ‘A’ but not to allele ‘G’ at rs2279590. EMSA was performed using heat shocked nuclear extract from HEK293 cells. (A) Unlike unlabelled
oligo with allele ‘A’, addition of unlabelled oligo with allele ‘G’ at rs2279590 couldn’t abolish the shift suggesting the specificity of binding complex to allele ‘A’. (B) Two
biotin labelled probes: a 28 bp reported heat shock element (HSE) (Lane 1-3) and one identical to 29 bp flanking region of rs2279590 with allele ‘A’ (Lane 4-6) were used.
The presence of a shift in lane 2 and 5 (arrowhead) implies binding of a protein complex from HEK293 nuclear extract to both HSE and to the genomic region surround-
ing rs2279590. However, addition of their identical but unlabelled oligos in excess (lane 3 and 6) dissolves the shift suggesting specificity of the binding complex. (C)
Specific competitive assays were done with increasing concentrations of unlabelled HSE (1 and 100 fold excess) or unlabelled rs2279590 probe with allele ‘A’ (1,
10, 50, 100, 200 and 400 fold excess). The shift abolished in the lanes 3-10 indicates that the same protein complex binds to HSE and rs2279590 region. (D)
Supershift assay with EMSA validated HSF1 antibody (Lane 4) shows elimination of the shift which signifies the binding of HSF1 to allele ‘A’ at rs2279590. All experi-
ments were replicated at least three times. Arrowhead and Starmark represent the specific and nonspecific shift, respectively.

was observed that a specific protein complex bound to the do not show a prominent shift as seen with allele ‘A’
labelled oligos (Lane 2, Fig. 3A) marked by a shift (arrowhead). (Supplementary Material, Fig. S2A).
With addition of unlabelled oligos comprising allele ‘A’ (Lane 3) A three-tier experimental validation was carried out to vali-
the shift disappears but not with unlabelled oligos with allele ‘G’ date the binding of HSFs at rs2279590. Figure 3B shows a compa-
(Lane 4); suggesting the binding complex is specific for allele ‘A’. rative shift in lanes 2 and 5 when a labelled consensus heat
Similar binding assays with labelled oligos comprising ‘G’ allele shock element (HSE) and labelled rs2279590 oligo with allele ‘A’
 Human Molecular Genetics, 2017, Vol. 26, No. 22 | 4523

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

Figure 4. Enhancer effect of rs2279590 locus was lost after binding of HSF1 to allele ‘A’ at rs2279590. (A) Chromatin immunoprecipitated samples were used in qRT-PCR
assay which shows more enrichment of rs2279590 flanking region with HSF1 antibody than with that of rabbit IgG in HEK293 cells. (B) Fold enrichment of genomic
region around rs2279590 is significantly higher (P ¼ 0.004) in IP samples with HSF1 (3.19 6 0.55) antibody than that of rabbit IgG (1.04 6 0.08). (C) Normalized luciferase
activity is shown in heat shocked, MG132 (5 mM) treated and siRNA (HSF1) treated HEK293 cells to differentiate the allelic effect on reporter activity. Cells transfected
with constructs containing allele ‘A’ show reduced reporter activity in both heat shocked (41.23 6 13.61, P ¼ 0.01) and MG132 (76.72 6 34.73, P ¼ 0.03) treated cells than
cells with allele ‘G’ (226.11 6 27.43 and 201.39 6 12.84, respectively); while it was found to be similar with P value of 0.31 between constructs with allele ‘G’ (191.6 6 7.43)
and ‘A’ (144.46 6 16.05) after knockdown of HSF1 with a pool of HSF1 specific siRNA. All experiments were performed at least three times and values are represented as
mean 6 SEM. Student’s t-test was used to calculate statistical significance between groups, *P < 0.05, **P < 0.005.

were used, respectively. A competitive EMSA on labelled oligo performed in HEK293 cells. As shown in Figure 4A and B the
containing rs2279590/A allele, when challenged with increased genomic region surrounding rs2279590 was enriched by qRT-PCR
concentrations (1- and 100-fold excess) of consensus unlabelled after immunoprecipitation with anti-HSF1 antibody but not by
HSE oligo (lanes 3 and 4, Fig. 3C) or unlabelled rs2279590/A oligo rabbit IgG (negative control). This confirms in vivo binding of HSF1
(1-, 10-, 50-, 100-, 200- and 400-fold excess) (lanes 5–10, Fig. 3C) to the sequences comprising of rs2279590.
the shift disappears; indicating that the 29 bp rs2279590/A
sequence binds to protein complexes similar to those binding to
a heat shock element. As shown in Figure 3C, 1-fold addition of Binding of HSF1 to allele ‘A’ at rs2279590 abrogates the
either unlabelled HSE or rs2279590/A (lane 3 and 5, respectively) enhancer effect of the locus
oligos was sufficient to compete for the binding complex, thereby In order to check the allele specific effect on the reporter activity, a
drastically reducing the intensity of the shift. 201 bp intronic region harbouring rs2279590 (with either ‘A’ or ‘G’
Later, when the binding complex was challenged with HSF1- allele) was cloned into pGL4.23 luciferase vector and transfected
specific antibody, disappearance of the shift was noted (lane 4, Fig. into HEK293 cells and the reporter activity was checked. As repre-
3D) indicating that the protein binding complex bound to allele ‘A’ sented in Figure 4C, changing the allele from ‘G’ to ‘A’ in HEK293
at rs2279590 comprises of HSF1 protein. However, absence of any transfected cells, under heat shock or MG-132 treatment drasti-
super-shift could be because the antibodies are bound to the epit- cally reduced the reporter activity; suggesting a loss of enhancer
ope of proteins necessary for DNA-protein complex formation. effect with rs2279590/A allele. Although a similar effect was also
Super-shift assays with other HSFs (HSF2 and HSF4) did not dis- found in non-treated cells growing at normal conditions but the
solve the shift completely, suggesting the binding complex con- difference was not found to be significant (Supplementary
sists of HSF1 but not HSF2 and HSF4 (Supplementary Material, Fig. Material, Fig. S3). To understand allele specific enhancer activity in
S2B). Additionally, to validate the binding of HSF1 to rs2279590 absence of HSF1, luciferase activity in HEK293 transfected cells
in vivo, chromatin immunoprecipitation (ChIP) assays were was checked after knocking down HSF1 expression by a pool of
 4524 | Human Molecular Genetics, 2017, Vol. 26, No. 22

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

Figure 5. Locus containing rs2279590 also regulates PTK2B and EPHX2 gene expression. (A) A stretched genomic region of 700 kb around clusterin gene in chromosome
8 is shown. Position of crucial genes relative to that of clusterin is depicted. (B) qRT-PCR assays were done for genes located in the PTK2B-CLU locus in rs2279590 region
deleted cells and control non-deleted cells. Expression of PTK2B and EPHX2 is significantly downregulated with a P value of 0.006 and 0.008 respectively in deleted cells
(0.58 6 0.08 and 0.34 6 0.04, respectively) compared with control non-deleted cells (1 6 0.05 and 1 6 0.04, respectively). qRT-PCR assay did not show any expression of
CHRNA2 in control HEK293 cells, hence omitted from analysis. Experiments were performed at least three times and values are represented as mean 6 SEM. Student’s
t-test was used to calculate statistical significance between groups, *P < 0.05.

HSF1 specific siRNA (Supplementary Material, Fig. S4). As overexpression has a protective role in a variety of neurodege-
expected, no significant difference (P ¼ 0.31) was observed in the nerative disorders caused due to accumulation of misfolded
reporter activity between constructs containing either allele ‘G’ or proteins (15–17). Further, aberrant expression of HSF1 has also
‘A’ in HSF1 knocked down cells (Fig. 4C). This suggests binding of been implicated in neurodegeneration (18). Assuming an
HSF1 to allele ‘A’ abrogates the enhancer effect of the locus. involvement of proteotoxic stress in anterior eye tissues of PEX-
affected subjects, we checked the expression of HSF1.
Interestingly, HSF1 level was found to be significantly upregu-
Deletion of rs2279590 locus alters the expression of PTK2B lated in both lens capsule and conjunctiva of PEXS-affected
and EPHX2, known modulators in AD pathogenesis individuals (Fig. 6). However, we didn’t find any upregulation of
HSF1 in PEXG-affected individuals; a later stage of PEX with
GWAS studies have previously shown that the region containing
degenerated optic nerve head (ONH) cells.
CLU and PTK2B is situated within a high-risk locus for AD patho-
genesis (14). As rs2279590 is also within the same high-risk locus,
we wanted to investigate the effect of this variant in regulating Discussion
the neighbouring genes around CLU. For this, expression of nine
Clusterin, a multifunctional protein has divergent roles from
crucial genes clustered around 5’ and 3’ ends of CLU gene, within
being cytoprotective to cytotoxic. Extensive studies have reported
a stretch of around 558 kb was analysed in HEK293115/ cells
the association of CLU variants with the risk of developing vari-
lacking the regulatory locus (Fig. 5A). One of the genes, CHRNA2
ous diseases like PEX, diabetes and AD (3,4,6,7). This indicates
did not show any expression in control HEK293 cells and is there-
that a complex mechanism employed by CLU is responsible for
fore omitted from analysis. Interestingly, a significant downregu-
the progression of these diseases. Understanding the role of risk
lation in the expression of two upstream genes, PTK2B and EPHX2
variants within CLU can help in better diagnosis and treatment of
was observed in HEK293115/ cells as compared with control
such disorders. In this study, we aimed to characterize the func-
HEK293 cells (Fig. 5B). This indicates that the genomic region
tional significance of a risk variant, rs2279590, housed in 7th
around rs2279590 has an extended regulatory role with a wide-
intron of CLU gene. This particular variant has been reported to
spread enhancer effect on PTK2B and EPHX2.
be a risk factor for both AD and PEX (3,4,6). We, therefore,
intended to understand the functional role of this SNP.

Upregulation of HSF1 suggests proteotoxic stress in

anterior eye tissues of PEX-affected subjects rs2279590 is a functional intronic variant and regulates
clusterin gene expression
In response to proteotoxic stress, cell activates HSF1, which in
turn upregulates the heat shock proteins (HSPs) to prevent pro- Recent studies have reported the involvement of genetic var-
tein misfolding. Earlier studies have reported that HSF1 iants in CLU gene with the risk of developing PEX and AD.
 Human Molecular Genetics, 2017, Vol. 26, No. 22 | 4525

each with an effect size around 0.2. Altogether, the combined

effect of nearby SNPs including rs9331896 and rs11136000 may
supersede the moderate effect shown by rs2279590 in different
ethnic background. Further ongoing functional studies of these
eQTL SNPs in relation to rs2279590 will define the role of CLU in
AD and PEXG progression. This substantiates the fact that cer-
tain risk alleles in CLU gene enhance clusterin expression and
thereby contribute towards AD/PEX pathogenesis.
Studies have shown that elevated level of CLU is linked to
disease severity in both PEXG- and AD-affected patients (3,8,20).
Both in the anterior eye tissues of PEXG and in the brain of AD-
affected individuals CLU was found to be upregulated compared
with their respective controls (3,20,21). Although CLU was upre-

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

gulated, whether it’s the cause or an effect is still speculative.
Clusterin knockout in PDAPP mice has shown to be beneficial
due to reduced fibrillar plaque formation and neuritic dystrophy
Figure 6. HSF1 upregulation indicates a proteotoxic stress in anterior eye tissues (11). Nonetheless, being an extracellular chaperone, the protec-
of PEXS-affected study subjects. qRT-PCR assays were done to check the expres-
tive role of CLU cannot be undermined. A study suggests that it
sion of HSF1 in anterior eye tissues of PEX (including both PEXS and PEXG) and
is the extracellular CLU: substrate ratio that decides the fate of
control subjects. HSF1 is found to be upregulated in both lens capsule (1.39 6
0.15, P ¼ 0.02) and conjunctiva (2.68 6 0.29, P ¼ 0.0003) in PEXS individuals com- CLU to be pathogenic or protective. When substrate is in large
pared with that of control (1 6 0.03 and 1.06 6 0.12, respectively). However, no molar excess, CLU coincorporates itself within the complex in a
difference was observed between control and PEXG in lens capsule (1.01 6 0.05 failed attempt to prevent aggregation; leading to large insoluble
versus 1.11 6 0.08) and conjunctiva (1.01 6 0.06 versus 1.15 6 0.08), respectively). aggregates (22). Further, in chronic stress a nuclear form of
This implicates a proteotoxic stress in the anterior eye tissues of PEXS subjects
Clusterin (nCLU) tends to increase which initiates caspase-3
and a failed stimulation to upregulate HSF1 in PEXG individuals might be
dependent apoptosis (12). Enhancer effect associated with risk
responsible for the death of ONH cells. LC and Conj correspond to lens capsule
and conjunctiva, respectively. Sample size is denoted by ‘n’ and expression allele ‘G’ at rs2279590 locus, on both secretory and nuclear
change in fold is represented as mean 6 SEM. Student’s t-test was used to calcu- forms of Clusterin suggests a cytotoxic effect of both sCLU and
late statistical significance between groups, *P < 0.05, ***P < 0.0005. nCLU can augment neurodegeneration.

Since then, attempts have been made to identify the signifi-

cance of these risk variants in regulating CLU expression. One
HSF1 is a critical regulator in clusterin gene expression
such study reported that the AD-risk allele at CLU, rs9331888, Through bioinformatic and molecular analysis, we identified
increased the clusterin expression as compared with its coun- HSF1 to bind preferentially to protective allele ‘A’ at rs2279590
ter allele (19). Similarly, two other GWAS-associated risk var- and consequently regulate clusterin gene expression. HSF1 is a
iants; rs9331896 and rs11136000 which are located in the prominent member of a family of transcription factors called
second and third intron of CLU gene (6,14), respectively have heat shock factors known to be activated upon heat shock,
regulatory role over CLU expression as evident through eQTL stress or inflammatory triggering agents. Upon activation, it dif-
data from GTEx project. Accordingly, examined tissue samples ferentially regulates either by upregulating or downregulating a
with risk allele of both these SNPs, rs9331896 (T) and cascade of genes. It is a complex regulator and affect gene
rs11136000 (C) reportedly have elevated CLU expression than expression differently under similar conditions in different tis-
their respective reference alleles (GTEx Portal on 07/11/17). sues (23). Here, binding of HSF1 to the protective allele ‘A’ nega-
This suggests a cumulative effect of these risk variants in CLU tively regulates CLU gene expression; suggesting a suppressor
upregulation during pathogenesis of AD. However, same wasn’ effect. By negatively regulating, HSF1 might help in reducing the
t true for another AD-risk variant, rs7982; suggesting that not cytotoxicity associated with CLU overaccumulation.
all associated variants in clusterin gene have functional impli- Earlier studies have related HSF1 to various types of neuro-
cations in disease causation (10). degenerative proteinopathies. Downregulation of HSF1 acceler-
Similar to the above studies, we previously reported that ated the formation of protein aggregates in Huntington’s
individuals homozygous for risk allele ‘G’ at rs2279590 (another disease (24); whereas, activated HSF1 in R6/2 Huntington dis-
risk variant for AD/PEX) within CLU showed 2-fold increased ease mice prevents polyglutamine aggregate formation (16).
clusterin gene expression compared with ‘AA’ carriers in lens Thus, HSF1 could be regarded as a key controller of endogenous
capsules (3). In the present study, we found an enhancer effect protein aggregation (25). Similarly, in AD animal models, HSF1
of the genomic region surrounding rs2279590 with the risk allele expression was remarkably decreased in the cerebellum;
‘G’ on CLU expression. This is also supported by the eQTL data whereas, overexpression of HSF1 reduces brain b-amyloid levels
from GTEx that indicates tissue samples with genotype ‘GG’ and improves memory function (26,27). These studies suggest a
(alternative allele- G) at rs2279590 have elevated CLU expression protective role of HSF1 in such aging disorders. This was in-
than samples with that of ‘AA’ (reference allele- A) with an sync with the recent reports where the Alzheimer’s patients
effect size of 0.2 (Supplementary Material, Fig. S1). However, showed significantly lower levels of HSF1 than control individu-
reverse allelic association (allele ‘A’) in German cohort suggests als (28). In this study, we identified an upregulated HSF1 mRNA
a profound effect of other nearby SNPs compared with moder- levels in the anterior eye tissues of PEXS subjects but not in the
ate effect shown by rs2279590 (4). Analysis of eQTL data for later more severe form of disease, i.e. PEXG; suggesting the pro-
entire CLU gene (GTEx Portal on 07/11/17) also indicates that tective role of HSF1 is diminished in the later stages of the dis-
there are dozens of SNPs acting as eQTL for CLU expression ease condition and may augment severity.
 4526 | Human Molecular Genetics, 2017, Vol. 26, No. 22

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

Figure 7. Proposed model explaining the effect of SNP-dependent HSF1 binding at rs2279590, target gene expression and AD/PEX risk. Genomic region containing
rs2279590/G resides within an active enhancer element and increases gene expression of CLU, PTK2B and EPHX2. However, binding of HSF1 to protective allele ‘A’ at
rs2279590 diminishes the enhancer effect of the locus; which leads to decrease in target gene expression with a lowered risk of developing PEX/AD.

rs2279590 enhancer element also regulates two distal HSF1 to the protective allele ‘A’ abolishes the enhancer effect of
genes: PTK2B and EPHX2 the locus, which consequent to decreased expression of the target
genes; thereby implying a lowered risk of developing PEX or AD.
GWAS studies have previously shown an association of PTK2B-
CLU loci with AD (14). Protein tyrosine kinase 2 beta or PTK2B,
belongs to a non-receptor protein kinase family involved in cal- Summary and Conclusion
cium induced regulation of ion channels and MAPK pathway
activation. It plays a role in inducing phosphorylation of GSK3 With this study, we confirm that a PEX/AD associated risk var-
(Glycogen synthase kinase 3) which then promotes Tau fibrillar iant, rs2279590, resides within an enhancer element and regu-
pathology. In hippocampus of AD-affected individuals, height- lates the expression of three candidate genes CLU, PTK2B and
ened phosphotyrosine immunoreactivity was found in the EPHX2, which were previously known to be modulators in the
neuritic plaques, tangle-bearing neurons and microglia, which progression of AD. Increase in CLU expression by the risk allele
are characteristic features of increased PTK2B activity (29). at rs2279590 provides a mechanistic insight into the cytotoxic
Accumulation of PTK2B was also reported in the early event of role of CLU in PEXG individuals. Future studies are needed to
AD pathogenesis with progressive Tau pathology. However, it is explore the role of PTK2B and EPHX2 genes in PEX pathogenesis.
unclear whether accumulation of PTK2B in aiding the Tau path-
ology, is a consequence or a cause (30). In this study, we found
decreased levels of PTK2B mRNA in cells with the deleted Materials and Methods
enhancer element containing rs2279590; suggesting a distal Study subjects recruitment
enhancer effect of the locus over PTK2B expression. Although,
regulatory effect of rs2279590 on PTK2B was not established in This study was approved by the ethics review boards of the
eQTL data from GTEx project, it may depend on various con- National Institute of Science Education and Research and JPM
founding factors including the type of tissue analysed. Further, Rotary Club of Cuttack Eye Hospital and Research Institute,
earlier reports have shown an increased PTK2B expression with India and adhered to the tenets of the Declaration of Helsinki.
AD associated risk alleles at rs28834970 and rs2718058 (a cis- All participants underwent a detailed ocular examination,
pQTL within PTK2B locus and a trans-pQTL within NME8 locus) including slit lamp, ocular biometry, Goldman applanation
(31). Cumulatively, these results suggest that AD risk variants tonometry, þ90 D biomicroscopic fundus evaluation and 4 mir-
are capable of regulating PTK2B expression to promote AD ror gonioscopy. Inclusion and exclusion criteria for the grouping
pathogenesis. of control and PEX-affected individuals were followed as
Deletion of rs2279590 locus also leads to downregulation of reported previously (3). Anterior eye tissues (lens capsules and
EPHX2, another candidate gene for AD pathogenesis. EPHX2 or conjunctiva) from PEX-affected study subjects and age matched
epoxide hydrolase-2 metabolises epoxyeicosatrienoic acids which controls were collected in RNAlater stabilisation solution
are proven to be neuroprotective. Upregulation of hydrolase activ- (Invitrogen, USA) during cataract surgery and stored at 80 C
ity of soluble EPHX2 leads to an increased OGD-induced (oxygen- until further use.
glucose deprived) neuronal cell death (32). Recently, EPHX2 too has
been associated with AD as a risk factor through GWAS (33).
According to all the cumulative data known till date, a model Cell culture
is being proposed as shown in Figure 7 that rs2279590 resides The human cell line, HEK293 was cultured in HiGlutaXL
within a regulatory genomic region and with risk allele ‘G’ shows Dulbecco’s Modified Eagle Medium, High Glucose (AL007G) with
a widespread enhancer effect on three PEX/AD risk associated 10% fetal bovine serum (RM9952) and 1% penicillin (100 U/ml)
candidate genes; CLU, PTK2B and EPHX2. However, binding of and streptomycin (0.1 mg/ml) (A001), maintained at 37 C and 5%
 Human Molecular Genetics, 2017, Vol. 26, No. 22 | 4527

CO2. All cell culture chemicals were procured from HiMedia, biotin 5’-end labelling [Integrated DNA Technologies (IDT), Iowa,
Mumbai, India. USA]. Sequences of the oligomers are listed in Supplementary
Material, Table S1. Annealing of complementary oligos was done

by incubating them at 95 C for 5 min, followed by step-cooling to
Luciferase reporter assays room temperature.
For reporter assays, pGL4.23 luciferase reporter vector with min- Nuclear extract from heat shocked (1 h at 42  C) HEK293 was
imal promoter and pGL4.74 renilla vector were used (Promega, prepared by using NE-PER kit (Thermo Fisher Scientific) and pro-
Madison, Wisconsin). Genomic DNA was extracted from tein concentrations were estimated by Bradford’s assay. EMSA
peripheral blood leucocytes of the study subjects by phenol- was performed using LightShift Chemiluminescent EMSA Kit
chloroform extraction method. An intronic region of 201 bp (Thermo Fisher Scientific) following the manufacturer’s instruc-
surrounding rs2279590 variant (harbouring either ‘AA’ or ‘GG’ tions. DNA-protein binding assays were carried out using 3 mg of
genotype at the polymorphic site) was PCR-amplified using a total nuclear extract and 100 fmol of biotinylated annealed oli-
specific primer pair (Supplementary Material, Table S1) from the gonucleotides for each 20 ml total reaction volume. Competitive

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

extracted genomic DNA. The amplified products were then EMSA was done using 400-fold excess (40 pmol) of unlabelled
cloned into pGL4.23 vector by double digestion at KpnI-XhoI site double-stranded oligonucleotides. For supershift assays, 2 mg of
(KpnI-HF and XhoI-HF, New England Biolabs, Ipswich, MA, USA). EMSA validated HSF1, HSF2 and HSF4 antibody (sc-9144X, sc-
For transfection, HEK293 cells were seeded in a 12-well plate. 13056 and sc-366983; respectively) were procured from Santa
At 80% confluency, the cells were transiently co-transfected Cruz Biotechnology, Dallas, Texas and were incubated with the
(Lipofectamine 2000, Invitrogen) with luciferase constructs final reaction mixture for an additional 30 min on ice. Protein-
(1 mg) and renilla vector (pGL4.74, 10 ng). Transfection efficiency DNA complexes were separated on 6% native polyacrylamide
was normalized by renilla reporter activity. After 24 h of gels in 0.5X TBE, transferred to nylon membranes (Thermo
post-transfection, cell lysates were prepared following Dual- Fisher Scientific) and observed by chemiluminescent detection
LuciferaseV R Reporter Assay System (Promega). Reporter activ-
methods post-UV crosslinking.
ities were measured with VarioskanV R Flash Multimode reader

(Thermo Fisher Scientific, Waltham, Massachusetts) following

Chromatin immunoprecipitation
manufacturer’s instructions. Values for luciferase activity, post
normalisation with Renilla reporter activity, were used for fur- Pierce Agarose ChIP kit (Thermo Fisher Scientific) was used for
ther analysis. Each of the experiments was repeated independ- in vivo ChIP assays and the protocol suggested by the manufac-
ently with at least three replicates. turer was followed. Briefly, 2x106 HEK293 cells were seeded and
used for each experiment. Micrococcal nuclease (20 U) was used
for efficient digestion of formaldehyde fixed chromatin.
CRISPR/Cas9 construct preparation and genome editing Digested chromatin with an average length of 500 bp was subse-
Genome editing of HEK293 was done by CRISPR/Cas9 system as quently used for immunoprecipitation (IP). For each IP, 25 mg of


described previously (34). A pair of sgRNA was designed (http:// digested chromatin was incubated overnight at 4 C with 5 mg of
crispr.mit.edu/; date last accessed September 7, 2016) with a lit- ChIP validated HSF1 antibody (Santa Cruz Biotechnology, sc-
tle off-target specificity to delete a 115 bp region around the 9144X). As negative control, normal Rabbit IgG (1 ml) provided in
SNP, rs2279590 (Supplementary Material, Table S1). Annealed the ChIP kit was used. Later, 20 ml of ChIP grade protein A/G plus


oligonucleotides were phosphorylated and ligated into BbsI agarose beads was added to each IP and incubated for 2 h at 4 C
digested PX459 (Plasmid #62988; Addgene, Cambridge, with continuous rocking. IP complexes were then washed twice
Massachusetts). with each wash buffer supplied in the kit. Elution of the IP com-


At 50% confluency, the HEK293 cells were transfected with plex was done by incubating in IP elution buffer at 65 C for 1 hr.
2.5 mg of each CRISPR construct (with sgRNA1 and sgRNA2) Reverse crosslinking and protein digestion were done by NaCl
simultaneously using lipofectamine 2000 (Invitrogen, Carlsbad, and Proteinase K, respectively. Subsequently, DNA was purified
California). Transfected cells were selected after 24-h post- by using DNA Clean-Up column supplied in the kit. Fold enrich-
transfection in complete media supplemented with 2.5 mg/ml ment of the target region in IP compared with that of input was
puromycin. Single cell clones were then isolated and cultured assayed by a specific primer set (Supplementary Material, Table
by dilution cloning and subsequently genomic DNA was iso- S1) through quantitative real time PCR (qRT-PCR).
lated using DNeasy Blood & Tissue Kit (Qiagen, Hilden,
Germany) following manufacturer’s instructions. Screening was
Knockdown assays
done to find out homozygous deletion of genomic region con-
taining rs2279590 by using a specific set of primers siRNA pool for targeting HSF1 was procured from Santa Cruz
(Supplementary Material, Table S1) outside the targeted sites Biotechnology (sc-35611) and transfection was done as per man-
and subsequently confirmed by sequencing. Positive clones ufacturer’s protocol. For luciferase assays, reporter vector and
were used for subsequent experiments. siRNA pool were transfected together, cell lysates were pre-
pared after 36 h and subsequently checked for reporter activity.

Electrophoretic mobility shift assay

Quantitative real-time PCR
Online program, TFSEARCH (http://www.cbrc.jp/research/db/
TFSEARCH.html; date last accessed December 27, 2014) was used Total RNA was isolated from individual lens capsules or HEK293
for candidate search with a default threshold score set to 85. Two cells by using an RNA extraction kit (RNeasy Mini Kit, QIAGEN).
pairs of complementary 29-mer oligonucleotides, centred around cDNA was synthesized with 1 mg of total RNA, using a Reverse
rs2279590 with allele ‘A’ or ‘G’ and a previously reported heat Transcription Kit (Verso cDNA Synthesis Kit - AB1453A; Thermo
shock element (HSE) were ordered (35); both with and without Fisher Scientific). Gene specific primers (Supplementary
 4528 | Human Molecular Genetics, 2017, Vol. 26, No. 22

Material, Table S1) overlapping exon-exon junction were References

designed by using PrimerQuest Tool (IDT). qRT-PCR was per-
1. Rosenberg, M.E. and Silkensen, J. (1995) Clusterin: physio-
formed using 7500 Real time PCR Systems (Applied Biosystems,
logic and pathophysiologic considerations. Int. J. Biochem.
Foster city, California). Total 5 ng of cDNA and 0.8 mM each of
Cell Biol., 27, 633–645.
forward and reverse primers were used per 20 ml reaction vol-
2. Jones, S.E. and Jomary, C. (2002) Clusterin. Int. J. Biochem. Cell.
ume in triplicate for each sample. Amplification specificity of
Biol., 34, 427–431.
the PCR product was checked via melt curve analysis and
3. Padhy, B., Nanda, G.G., Chowdhury, M., Padhi, D., Rao, A. and
sequencing. DDCt method was used to calculate expression
Alone, D.P. (2014) Role of an extracellular chaperone, clus-
change in fold for each target gene. For normalization, GAPDH
terin in the pathogenesis of pseudoexfoliation syndrome
expression was taken as an endogenous control.
and pseudoexfoliation glaucoma. Exp. Eye Res., 127, 69–76.
4. Krumbiegel, M., Pasutto, F., Mardin, C.Y., Weisschuh, N.,
Western blotting Paoli, D., Gramer, E., Zenkel, M., Weber, B.H., Kruse, F.E.,

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

Schlotzer-Schrehardt, U. et al. (2009) Exploring functional
Cytosolic extract from HEK293 cells was prepared using NE-PER candidate genes for genetic association in german patients
kit (Thermo Fisher Scientific). Denatured cytosolic extract (10 mg) with pseudoexfoliation syndrome and pseudoexfoliation
was then loaded on a 12% SDS-PAGE and subsequently trans- glaucoma. Invest. Ophthalmol. Vis. Sci., 50, 2796–2801.
ferred onto a PVDF membrane (Immobilion-P PVDF from Merck 5. Chen, L.H., Kao, P.Y., Fan, Y.H., Ho, D.T., Chan, C.S., Yik, P.Y.,
Millipore, Billerica, Massachusetts). Subsequent steps were fol- Ha, J.C., Chu, L.W. and Song, Y.Q. (2012) Polymorphisms of
lowed as previously described (3). A polyclonal antibody for CR1, CLU and PICALM confer susceptibility of Alzheimer’s
Clusterin (sc-6419; Santa Cruz Biotechnology) was used as pri- disease in a southern Chinese population. Neurobiol. Aging,
mary antibody and HRP-conjugated rabbit anti-goat IgG 33, 210 e211–217.
(480011730; Imgenex, India) was used as secondary antibody. 6. Lambert, J.C., Heath, S., Even, G., Campion, D., Sleegers, K.,
GAPDH antibody (6665 A; Imgenex, India) was used for endoge- Hiltunen, M., Combarros, O., Zelenika, D., Bullido, M.J.,
nous control experiments. Detection was done using chemilu- Tavernier, B. et al. (2009) Genome-wide association study
minescence kit (Super Signal Femto Maximum Sensitivity identifies variants at CLU and CR1 associated with
Substrate, Thermo Fisher Scientific) in a Chemi-Doc (Bio-Rad, Alzheimer’s disease. Nat. Genet., 41, 1094–1099.
Hercules, California). 7. Daimon, M., Oizumi, T., Karasawa, S., Kaino, W., Takase, K.,
Tada, K., Jimbu, Y., Wada, K., Kameda, W., Susa, S. et al.
(2011) Association of the clusterin gene polymorphisms with
DNA sequencing type 2 diabetes mellitus. Metabolism, 60, 815–822.
Bidirectional Sanger’s sequencing of all the constructs was done 8. Giri, M., Zhang, M. and Lu, Y. (2016) Genes associated with
using BigDye Terminator v.3.1 cycle sequencing kit (Applied Alzheimer’s disease: an overview and current status. Clin.
Biosystem), with their respective primers (Supplementary Interv. Aging, 11, 665–681.
Material, Table S1) on 3130xl Genetic Analyser (Applied 9. Schrijvers, E.M., Koudstaal, P.J., Hofman, A. and Breteler,
Biosystem) platform. For sequencing analysis, BioEdit v7.1 M.M. (2011) Plasma clusterin and the risk of Alzheimer dis-
(http://www.mbio.ncsu.edu/bioedit/bioedit.html; date last ease. JAMA, 305, 1322–1326.
accessed March 4, 2016) and sequence analysis software v5.3 10. Karch, C.M., Jeng, A.T., Nowotny, P., Cady, J., Cruchaga, C.
(Applied Biosystem) were used. and Goate, A.M. (2012) Expression of novel Alzheimer’s dis-
ease risk genes in control and Alzheimer’s disease brains.
PloS One, 7, e50976.
Statistical analysis 11. DeMattos, R.B., O’Dell, M.A., Parsadanian, M., Taylor, J.W.,
Harmony, J.A., Bales, K.R., Paul, S.M., Aronow, B.J. and
All experiments were repeated at least three times. Descriptive
Holtzman, D.M. (2002) Clusterin promotes amyloid plaque
data were presented as mean 6 SEM. Group wise results for stat-
formation and is critical for neuritic toxicity in a mouse
istical significance was compared by Student’s t-test and
model of Alzheimer’s disease. Proc. Natl. Acad. Sci. U.S.A, 99,
P < 0.05 was considered as statistically significant.
10843–10848.
12. Caccamo, A.E., Scaltriti, M., Caporali, A., D’Arca, D.,
Supplementary Material Scorcioni, F., Astancolle, S., Mangiola, M. and Bettuzzi, S.
(2004) Cell detachment and apoptosis induction of immor-
Supplementary Material is available at HMG online. talized human prostate epithelial cells are associated with
early accumulation of a 45 kDa nuclear isoform of clusterin.
Biochem. J., 382, 157–168.
Acknowledgement 13. Trougakos, I.P., Djeu, J.Y., Gonos, E.S. and Boothman, D.A.
The authors thank the study participants for their contribution (2009) Advances and challenges in basic and translational
and consent for this study. research on clusterin. Cancer Res., 69, 403–406.
14. Lambert, J.C., Ibrahim-Verbaas, C.A., Harold, D., Naj, A.C.,
Conflict of Interest statement. None declared. Sims, R., Bellenguez, C., DeStafano, A.L., Bis, J.C., Beecham,
G.W., Grenier-Boley, B. et al. (2013) Meta-analysis of 74, 046
individuals identifies 11 new susceptibility loci for
Funding Alzheimer’s disease. Nat. Genet., 45, 1452–1458.
National Institute of Science Education and Research, 15. Verma, P., Pfister, J.A., Mallick, S. and D’Mello, S.R. (2014)
Department of Atomic Energy (India); Council of Scientific and HSF1 protects neurons through a novel trimerization- and
Industrial Research (India) Grant no. 27(0317)/16/EMR-II. HSP-independent mechanism. J. Neurosci., 34, 1599–1612.
 Human Molecular Genetics, 2017, Vol. 26, No. 22 | 4529

16. Fujimoto, M., Takaki, E., Hayashi, T., Kitaura, Y., Tanaka, Y., 26. Jiang, Y.Q., Wang, X.L., Cao, X.H., Ye, Z.Y., Li, L. and Cai, W.Q.
Inouye, S. and Nakai, A. (2005) Active HSF1 significantly sup- (2013) Increased heat shock transcription factor 1 in the
presses polyglutamine aggregate formation in cellular and cerebellum reverses the deficiency of Purkinje cells in
mouse models. J. Biol. Chem., 280, 34908–34916. Alzheimer’s disease. Brain Res., 1519, 105–111.
17. Neef, D.W., Jaeger, A.M. and Thiele, D.J. (2011) Heat shock 27. Gouras, G.K. (2013) mTOR: at the crossroads of aging, chaper-
transcription factor 1 as a therapeutic target in neurodege- ones, and Alzheimer’s disease. J. Neurochem., 124, 747–748.
nerative diseases. Nat. Rev. Drug Discov., 10, 930–944. 28. Goetzl, E.J., Boxer, A., Schwartz, J.B., Abner, E.L., Petersen,
18. Kim, E., Wang, B., Sastry, N., Masliah, E., Nelson, P.T., Cai, H. R.C., Miller, B.L., Carlson, O.D., Mustapic, M. and
and Liao, F.F. (2016) NEDD4-mediated HSF1 degradation Kapogiannis, D. (2015) Low neural exosomal levels of cellular
underlies alpha-synucleinopathy. Hum. Mol. Genet., 25, survival factors in Alzheimer’s disease. Ann. Clin. Transl.
211–222. Neurol., 2, 769–773.
19. Szymanski, M., Wang, R., Bassett, S.S. and Avramopoulos, D. 29. Kohler, C., Dinekov, M. and Gotz, J. (2013) Active glycogen
(2011) Alzheimer’s risk variants in the clusterin gene are synthase kinase-3 and tau pathology-related tyrosine phos-

Downloaded from https://academic.oup.com/hmg/article-abstract/26/22/4519/4095348 by guest on 19 June 2019

associated with alternative splicing. Transl. Psychiatry, 1, e18. phorylation in pR5 human tau transgenic mice. Neurobiol.
20. Zhou, Y., Hayashi, I., Wong, J., Tugusheva, K., Renger, J.J., Aging, 34, 1369–1379.
Zerbinatti, C. and Götz, J. (2014) Intracellular clusterin inter- 30. Dourlen, P., Fernandez-Gomez, F.J., Dupont, C., Grenier-
acts with brain isoforms of the bridging integrator 1 and Boley, B., Bellenguez, C., Obriot, H., Caillierez, R., Sottejeau,
with the microtubule-associated protein Tau in Alzheimer’s Y., Chapuis, J., Bretteville, A. et al. (2017) Functional screen-
disease. PloS One, 9, e103187. ing of Alzheimer risk loci identifies PTK2B as an in vivo mod-
21. Zenkel, M., Kruse, F.E., Junemann, A.G., Naumann, G.O. and ulator and early marker of Tau pathology. Mol. Psychiatry, 22,
Schlotzer-Schrehardt, U. (2006) Clusterin deficiency in eyes 874–883.
with pseudoexfoliation syndrome may be implicated in the 31. Chan, G., White, C.C., Winn, P.A., Cimpean, M., Replogle,
aggregation and deposition of pseudoexfoliative material. J.M., Glick, L.R., Cuerdon, N.E., Ryan, K.J., Johnson, K.A.,
Invest. Ophthalmol. Vis. Sci., 47, 1982–1990. Schneider, J.A. et al. (2015) CD33 modulates TREM2: conver-
22. Yerbury, J.J., Poon, S., Meehan, S., Thompson, B., Kumita, J.R., gence of Alzheimer loci. Nat. Neurosci., 18, 1556–1558.
Dobson, C.M. and Wilson, M.R. (2007) The extracellular chap- 32. Koerner, I.P., Jacks, R., DeBarber, A.E., Koop, D., Mao, P.,
erone clusterin influences amyloid formation and toxicity Grant, D.F. and Alkayed, N.J. (2007) Polymorphisms in the
by interacting with prefibrillar structures. faseb J., 21, human soluble epoxide hydrolase gene EPHX2 linked to
2312–2322. neuronal survival after ischemic injury. J. Neurosci., 27,
23. Wang, X., Cattaneo, F., Ryno, L., Hulleman, J., Reixach, N. and 4642–4649.
Buxbaum, J.N. (2014) The systemic amyloid precursor trans- 33. Chapuis, J., Flaig, A., Grenier-Boley, B., Eysert, F., Pottiez, V.,
thyretin (TTR) behaves as a neuronal stress protein regu- Deloison, G., Vandeputte, A., Ayral, A.M., Mendes, T., Desai,
lated by HSF1 in SH-SY5Y human neuroblastoma cells and S. et al. (2017) Genome-wide, high-content siRNA screening
APP23 Alzheimer’s disease model mice. J. Neurosci., 34, identifies the Alzheimer’s genetic risk factor FERMT2 as a
7253–7265. major modulator of APP metabolism. Acta Neuropathol., 133,
24. Anckar, J. and Sistonen, L. (2011) Regulation of HSF1 function 955–966.
in the heat stress response: implications in aging and dis- 34. Ran, F.A., Hsu, P.D., Wright, J., Agarwala, V., Scott, D.A. and
ease. Annu. Rev. Biochem., 80, 1089–1115. Zhang, F. (2013) Genome engineering using the CRISPR-Cas9
25. Lechler, M.C., Crawford, E.D., Groh, N., Widmaier, K., Jung, R., system. Nat. Protoc., 8, 2281–2308.
Kirstein, J., Trinidad, J.C., Burlingame, A.L. and David, D.C. 35. Ortner, V., Ludwig, A., Riegel, E., Dunzinger, S. and Czerny, T.
(2017) Reduced Insulin/IGF-1 Signaling Restores the (2015) An artificial HSE promoter for efficient and selective
Dynamic Properties of Key Stress Granule Proteins during detection of heat shock pathway activity. Cell Stress
Aging. Cell Rep., 18, 454–467. Chaperones, 20, 277–288.