Content uploaded by Jenny Joutsen
Author content
All content in this area was uploaded by Jenny Joutsen on Jan 20, 2020
Content may be subject to copyright.
Article
Heat Shock Factor 2 Protects against Proteotoxicity
by Maintaining Cell-Cell Adhesion
Graphical Abstract
Highlights
dHSF2 is required to maintain cell-cell adhesion
dHSF2 deficiency leads to downregulation of cadherin
superfamily genes
dImpaired cell-cell adhesion sensitizes cells to prolonged
proteotoxic stress
dCadherin-mediated cell-cell adhesion is a survival
determinant upon proteotoxicity
Authors
Jenny Joutsen, Alejandro Jose Da Silva,
Jens Christian Luoto, ...,
De
´lara Sabe
´ran-Djoneidi,
Eva Henriksson, Lea Sistonen
Correspondence
lea.sistonen@abo.fi
In Brief
Joutsen et al. show that heat shock factor
2 (HSF2) is essential for cell survival
during prolonged proteotoxicity. Lack of
HSF2 leads to marked misregulation of
cadherin superfamily genes and
functional impairment of cell-cell
adhesion. Cell-cell adhesion is found to
be a key determinant of proteotoxic
stress resistance.
Joutsen et al., 2020, Cell Reports 30, 583–597
January 14, 2020 ª2019 The Author(s).
https://doi.org/10.1016/j.celrep.2019.12.037
Cell Reports
Article
Heat Shock Factor 2 Protects
against Proteotoxicity
by Maintaining Cell-Cell Adhesion
Jenny Joutsen,
1,2,7
Alejandro Jose Da Silva,
1,2,7
Jens Christian Luoto,
1,2
Marek Andrzej Budzynski,
1,2
Anna Serafia Nylund,
1,2
Aurelie de Thonel,
3,4,5
Jean-Paul Concordet,
6
Vale
´rie Mezger,
3,4,5
De
´lara Sabe
´ran-Djoneidi,
3,4,5
Eva Henriksson,
1,2
and Lea Sistonen
1,2,8,
*
1
Faculty of Science and Engineering, Cell Biology, A
˚bo Akademi University, Tykisto
¨katu 6, 20520 Turku, Finland
2
Turku Bioscience Centre, University of Turku and A
˚bo Akademi University, Tykisto
¨katu 6, 20520 Turku, Finland
3
CNRS, UMR 7216 ‘‘Epigenetic and Cell Fate,’’ 75250 Paris Cedex 13, France
4
University of Paris Diderot, Sorbonne Paris Cite
´, 75250 Paris Cedex 13, France
5
De
´partement Hospitalo-Universitaire DHU PROTECT, Paris, France
6
INSERM U1154, CNRS UMR 7196, Muse
´um National d’Histoire Naturelle, Paris, France
7
These authors contributed equally
8
Lead Contact
*Correspondence: lea.sistonen@abo.fi
https://doi.org/10.1016/j.celrep.2019.12.037
SUMMARY
Maintenance of protein homeostasis, through induc-
ible expression of molecular chaperones, is essential
for cell survival under protein-damaging conditions.
The expression and DNA-binding activity of heat
shock factor 2 (HSF2), a member of the heat shock
transcription factor family, increase upon exposure
to prolonged proteotoxicity. Nevertheless, the spe-
cific roles of HSF2 and the global HSF2-dependent
gene expression profile during sustained stress
have remained unknown. Here, we found that HSF2
is critical for cell survival during prolonged proteo-
toxicity. Strikingly, our RNA sequencing (RNA-seq)
analyses revealed that impaired viability of HSF2-
deficient cells is not caused by inadequate induction
of molecular chaperones but is due to marked down-
regulation of cadherin superfamily genes. We
demonstrate that HSF2-dependent maintenance of
cadherin-mediated cell-cell adhesion is required for
protection against stress induced by proteasome in-
hibition. This study identifies HSF2 as a key regulator
of cadherin superfamily genes and defines cell-cell
adhesion as a determinant of proteotoxic stress
resistance.
INTRODUCTION
The cells in a human body are constantly exposed to environ-
mental stressors, which challenge the maintenance of protein
homeostasis, also called proteostasis. To survive insults that
disturb proteostasis, cells rely on a selection of protective mech-
anisms that can be launched upon stress exposures. The heat
shock response is a well-conserved stress protective pathway
that is induced in response to cytosolic protein damage and
mediated by heat shock transcription factors (HSFs; Joutsen
and Sistonen, 2019). Upon activation, HSFs oligomerize, accu-
mulate in the nucleus, and bind to their target heat shock ele-
ments (HSEs) at multiple genomic loci (Vihervaara et al., 2013,
2017; Mahat et al., 2016). The canonical HSF target genes
encode molecular chaperones, such as heat shock proteins
(HSPs), which assist in the maintenance of a correct protein
folding environment by refolding the misfolded proteins or di-
recting them to protein degradation machineries (Hartl et al.,
2011). In addition, HSFs are important in a variety of other phys-
iological and pathological processes and the repertoire of HSF
target genes has been shown to extend beyond the HSPs
(Hahn et al., 2004; A
˚kerfelt et al., 2010; Gonsalves et al., 2011;
Mendillo et al., 2012; Riva et al., 2012; Bjo
¨rk et al., 2016; Li
et al., 2016).
The human genome encodes six HSF family members (HSF1,
HSF2, HSF4, HSF5, HSFX, and HSFY), of which HSF1 and HSF2
are the most extensively studied (Joutsen and Sistonen, 2019).
Although these factors are homologous in their DNA-binding do-
mains, they share only a few similarities in the tissue expression
patterns, regulatory mechanisms, and signals that stimulate their
activity (Jaeger et al., 2016; Gomez-Pastor et al., 2018). HSF1 is
essential for HSP expression and cell survival under acute stress
conditions (Joutsen and Sistonen, 2019). HSF2 is an unstable
protein and its expression is highly context dependent, fluctu-
ating in different cell and tissue types (Sarge et al., 1991; Alastalo
et al., 1998), developmental stages (Mezger et al., 1994; Rallu
et al., 1997), and during the cell cycle (Elsing et al., 2014). Conse-
quently, regulation of HSF2 protein levels has been considered
to be the main determinant of its DNA-binding capacity (Mathew
et al., 1998; Budzy
nski and Sistonen, 2017). Interestingly, the
DNA-binding activity of HSF2 increases in cells exposed to lac-
tacystin- or MG132-induced proteasome inhibition (Kawazoe
et al., 1998; Mathew et al., 1998; Pirkkala et al., 2000), indicating
that HSF2 can respond to proteostasis disruption. Nevertheless,
the molecular details of the activation mechanisms of HSF2 are
currently not conclusively understood.
Cell Reports 30, 583–597, January 14, 2020 ª2019 The Author(s). 583
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
The ubiquitin-proteasome system is one of the main cellular
mechanisms regulating protein turnover, thereby affecting multi-
ple aspects of cell physiology, such as signal transduction and
apoptosis (Hershko and Ciechanover, 1998; Varshavsky, 2012).
Due to the fundamental function in cell physiology, the protea-
some complex has emerged as an important target for anti-can-
cer therapy (Deshaies, 2014). The most common drug to inhibit
proteasome function is bortezomib (BTZ; PS-341, VELCADE),
which is currently used as a standard treatment in hematological
malignancies (Chen et al., 2011). BTZ is a dipeptide boronic acid
derivative that targets the chymotrypsin-like activity of the 26S
proteasome, causing progressive accumulation of damaged pro-
teins (Kisselev et al., 2006; Chen et al., 2011; Goldberg, 2012). By
exposing human blood-derived primary cells to clinically relevant
concentrations of BTZ, Rossi and colleagues demonstrated that
prolonged proteasome inhibition results in upregulation of HSF2
at both mRNA and protein levels (Rossi et al., 2014). They also
showed that HSF2, together with HSF1, localizes to the pro-
moters of HSP70 and AIRAP (zinc finger AN1-type domain 2a)
genes (Rossi et al., 2014). In another study, sensitivity to protea-
some inhibition was linked to HSF2 deficiency in mouse embry-
onic fibroblasts (Lecomte et al., 2010), but the mechanisms by
which HSF2 promotes cell survival are currently unknown.
In this study, we show that HSF2 is critical for survival of cells
during prolonged proteasome inhibition. To our surprise, the
genome-wide expression analyses revealed that HSF2 disrup-
tion results in a profound downregulation of genes belonging
to the cadherin superfamily and subsequent functional impair-
ment of cell-cell adhesion. Furthermore, we show that failure to
form adequate cadherin-mediated cell-cell adhesion contacts
predisposes cells to proteasome inhibition-induced cell death.
These results identify HSF2 as a key regulator of cadherin genes.
Taken together, we show that by maintaining cadherin-mediated
cell-cell adhesion, HSF2 acts as an important pro-survival factor
during sustained proteotoxic stress.
RESULTS
U2OS Cells Lacking HSF2 Are Predisposed to BTZ-
Induced Proteotoxicity
To explore the role of HSF2 in prolonged proteotoxic stress, we
first examined the expression and cellular localization of HSF2
during BTZ treatment. Human osteosarcoma U2OS cells were
treated with different concentrations of BTZ (0–100 nM) for 6
or 22 h and HSF2 protein levels were examined with immuno-
blotting. The time points were selected to assess both the
Figure 1. HSF2 Is Upregulated upon Prolonged Bortezomib (BTZ) Treatment
(A) Immunoblot analysis of HSF2 expression. U2OS WT cells were treated with indicated concentrations of BTZ for 6 or 22 h. Control (C) cells were treated with
DMSO. Tubulin was used as a loading control.
(B) Confocal microscopy images of HSF2 immunofluorescence staining. U2OS WT cells were plated on coverslips and treated with 25 nM BTZ for 6, 10, or 22 h.
Control cells were treated with DMSO. Samples were fixed and stained with anti-HSF2 antibody. DAPI was used for DNA detection. The overlay of HSF2 and DAPI
maximum intensity projection signals is shown in merge. Scale bar, 100 mm.
(C) Immunoblot analysis of HSF2 in subcellular fractions. U2OS WT cells were treated with 25 nM BTZ for 6 and 22 h. Control cells were treated with DMSO. Wc,
whole cell fraction; Cy, cytoplasmic fraction; and Nu, nuclear fraction. Lamin A/C and Tubulin were used as fractionation controls.
584 Cell Reports 30, 583–597, January 14, 2020
(legend on next page)
Cell Reports 30, 583–597, January 14, 2020 585
short-term and the long-term exposure to BTZ. HSF2 was
slightly upregulated already at the 6-h time point and at 22 h
its expression was highly elevated (Figure 1A), which is in agree-
ment with a previous report (Rossi et al., 2014). Indirect immuno-
fluorescence and analysis of distinct subcellular fractions
revealed that HSF2, which is known to be both cytoplasmic
and nuclear (Sheldon and Kingston, 1993; Sistonen et al.,
1994), resides predominantly in the nucleus already under con-
trol conditions and the nuclear localization is further enhanced
during BTZ treatment (Figures 1B and 1C). These results show
that cells respond to BTZ treatment with marked increases in
HSF2 levels and accumulation in the nucleus.
Next, we asked if HSF2 is required for cell survival under sus-
tained stress conditions. We generated a U2OS HSF2 knockout
cell line (2KO hereafter), where HSF2 expression was abolished
by mutating the first exon of the HSF2 gene using the CRISPR-
Cas9 method. In these cells, the protein expression of HSF2
was completely abrogated (Figure 2A). U2OS WT and 2KO cells
were treated with indicated concentrations of BTZ for 22 h and
examined with microscopy. We observed a dramatic difference
in the viability of the wild-type (WT) and 2KO cells, since the cells
lacking HSF2 exhibited an apoptotic non-adherent phenotype in
concentrations where the WT cells remained adherent (Fig-
ure 2B). Quantitative cell viability measurements confirmed that
the survival of 2KO cells was significantly impaired upon BTZ
treatment (Figure 2C). Furthermore, 2KO cells accumulated
more cleaved PARP-1 than WT cells, demonstrating a more pro-
nounced activation of apoptosis (Ling et al., 2002)(Figure 2D).
Similar results were obtained with another HSF2 knockout cell
line (2KO#2 hereafter) (Figure S1A) and with Hsf2
/
MEFs
(mouse embryonic fibroblasts) (Figures S1B and S1C), confirm-
ing that the observations are not cell type specific. To verify that
the decreased survival of 2KO cells was not caused by off-target
effects of the CRISPR-Cas9 gene editing method, we trans-
fected the U2OS WT cells with scramble (Scr) or HSF2-targeting
short hairpin RNA (shRNA) plasmids and treated the cells with
BTZ for 22 h. In accordance with the results obtained with 2KO
cells, transient HSF2 downregulation significantly reduced cell
viability upon BTZ treatment and enhanced the progression of
apoptosis, which was detected by increased accumulation of
cleaved PARP-1 (Figures 2E and 2F). In contrast, re-introduction
of HSF2 to the 2KO cells resulted in significantly less cleaved
PARP-1 than in the Mock-transfected cells after BTZ treatment
(Figure 2G). Hence, we conclude that HSF2 is essential for cell
survival upon proteotoxic stress.
In addition to BTZ, treatments with MG132, a well-established
proteasome inhibitor, and amino acid analog L-canavanine,
which causes protein misfolding when incorporated into nascent
peptide chains, clearly reduced the viability of HSF2-deficient
cells (Figures S1D–S1H). Importantly, when we exposed the cells
to even more extended proteotoxic stress of 46 h, induced by
HSP90 inhibitor drugs (Geldanamycin, 17-AAG), the HSF2-defi-
cient cells exhibited reduced survival (Figures 2H and 2I).
Altogether these results demonstrate that HSF2 is critical for
cell survival upon prolonged accumulation of damaged proteins.
In contrast to HSF2, which has been found to be downregu-
lated in a subset of human cancers (Bjo
¨rk et al., 2016), the
expression, nuclear accumulation, and transcriptional activity
of HSF1 are increased in a majority of studied cancer types
(Bjo
¨rk et al., 2018;Gomez-Pastor et al., 2018). Phosphorylation
of serine 326 (pS326) in HSF1 is considered to be a marker for
its activation (Guettouche et al., 2005; Mendillo et al., 2012).
HSF1 expression and pS326 have been established as require-
ments for multiple myeloma cell survival during BTZ treatment
(Shah et al., 2016). Therefore, we examined whether the
decreased survival of 2KO cells was due to impaired HSF1
expression or phosphorylation upon proteasome inhibition.
U2OS WT and 2KO cells were treated with BTZ or MG132, and
the HSF1 protein levels and S326 phosphorylation status were
analyzed with immunoblotting. Importantly, no difference in
HSF1 expression or S326 phosphorylation between WT and
2KO cells was observed upon proteasome inhibition (Figure S1I).
These results demonstrate that although HSF1 is an essential
survival factor during acute stress (Gomez-Pastor et al., 2018),
it alone is not sufficient to protect cells against prolonged
proteotoxicity.
Figure 2. HSF2 Is Required for Cell Survival upon Prolonged Bortezomib (BTZ) Treatment
(A) Immunoblot analysis of HSF2 expression in U2OS WT and HSF2 KO (2KO) cells. Tubulin was used as a loading control.
(B) Bright-field microscopy images of WT and 2KO cells treated with indicated concentrations of BTZ for 22 h. Control cells were treated with DMSO. Scale bar,
100 mm.
(C) Calcein AM assay of WT and 2KO cells treated as in (B). Relative fluorescence was calculated against each respective control that was set to 1. The data are
presented as mean values of at least three independent experiments + SEM; *p < 0.05.
(D) Immunoblot analysis of PARP-1. Cells were treated as in (B). HSC70 was used as a loading control.
(E) Calcein AM assay of U2OS WT cells transfected with Scr or HSF2-targeting shRNA constructs (O
¨stling et al., 2007) and treated with 25 nM BTZ for 22 h.
Relative fluorescence was calculated against each respective control that was set to 1. The data are presented as mean values of three independent experi-
ments + SEM; *p < 0.05.
(F) Immunoblot analysis of HSF2 and PARP-1. Cells were transfected and treated as in (E). Tubulin was used as a loading control.
(G) Immunoblot analysis of HSF2 and PARP-1. HSF2 levels in U2OS 2KO cells were restored to those in WT cells by transiently transfecting the cells with either
Mock or HSF2 encoding plasmids. Cells were treated with 25 nM BTZ for 22 h. Control cells were treated with DMSO. HSC70 was used as a loading control.
The amount of cleaved PARP-1 relative to HSC70 was quantified with ImageJ. The data are presented as mean values of three independent experiments + SEM;
*p < 0.05.
(H) CellTiter-Glo assay of U2OS WT and 2KO cells treated with indicated concentrations of Geldanamycin (GA) for 46 h. Control cells were treated with DMSO.
Relative luminescence was calculated against each control that was set to 1. The data are presented as mean values of three independent experiments + SEM;
**p < 0.01 and ***p < 0.001.
(I) CellTiter-Glo assay of U2OS WT and 2KO cells treated with indicated concentrations of 17-AAG for 46 h. Control cells were treated with DMSO. Relative
luminescence was calculated against each control that was set to 1. The data are presented as mean values of three independent experiments + SEM; *p < 0.05.
See also Figure S1.
586 Cell Reports 30, 583–597, January 14, 2020
(legend on next page)
Cell Reports 30, 583–597, January 14, 2020 587
Induction of Heat Shock Response Is Not Sufficient to
Protect Cells against Proteotoxicity
Similarly to many other surveillance transcription factors, such
as p53 (Kubbutat et al., 1997), HIF-1a(Kallio et al., 1999), and
Nrf2 (Kobayashi et al., 2004), HSF2 is an unstable protein under
normal growth conditions (Ahlskog et al., 2010). HSF2 expres-
sion fluctuates in response to stress exposure, tumor progres-
sion, and during the cell cycle (Ahlskog et al., 2010:Elsing
et al., 2014; Bjo
¨rk et al., 2016), and high expression levels of
HSF2 correlate with its increased DNA-binding activity (Mathew
et al., 1998; Sarge et al., 1994). Due to the massive increase in
nuclear HSF2 levels upon BTZ treatment (Figures 1B and 1C),
we investigated if the impaired survival of 2KO cells was caused
by misregulation of HSF2 target genes. U2OS WT and 2KO cells
were treated with 25 nM BTZ for 6 or 10 h (Figure 3A), and the
global gene expression profiles were analyzed with RNA-seq.
It is important to note that the selected time points represent
sub-lethal proteotoxic stress conditions, at which the cell
viability is not yet compromised (Figure S2A). Before mRNA pu-
rification, the knockout phenotype was confirmed with immuno-
blotting (Figure S2B). Stress-inducible hyperphosphorylation of
HSF1 (Sarge et al., 1993) and increased HSP70 expression
were observed in both WT and 2KO cells (Figure S2B). To identify
the HSF2-dependent target genes, we first compared the induc-
ible gene expression profiles between WT and 2KO cells in
response to BTZ treatment for 6 and 10 h (Figure 3A). Differen-
tially expressed (DE) genes were determined with the Bio-
conductor R package Limma (Ritchie et al., 2015), with fold
change R3 and false discovery rate (FDR) <0.001, from quadru-
plet samples that correlated well to each other (Figure S2C).
According to the analysis, BTZ treatment resulted in a significant
upregulation and downregulation of genes in WT (>600 and
>300, respectively) and 2KO (>500 and >200, respectively) cells
(Figure 3B; Table S1). The complete dataset is available at Gene
Expression Omnibus under GEO: GSE115973.
The HSF-regulated heat shock response is one of the main
cellular survival pathways induced by proteotoxic stress (Jout-
sen and Sistonen, 2019), and it is characterized by simultaneous
upregulation of genes essential for maintaining the correct pro-
tein folding environment (Vihervaara et al., 2018). To examine
whether the impaired survival of 2KO cells is caused by a
compromised heat shock response, we analyzed the inducible
expression patterns of all human molecular chaperone genes
(Kampinga et al., 2009), in WT and 2KO cells treated with BTZ.
Intriguingly, the chaperone expression profiles of WT and 2KO
cells were nearly identical, and only HSPB2,DNAJC12, and
DNAJC18 exhibited distinct expression patterns in 2KO cells
(Figure 3C). A closer examination of the RNA-seq data for
the expression of selected chaperone genes, i.e., HSPA1A
(HSP70), HSP90AA1 (HSP90), HSPA6 (HSP70B0), and HSPB1
(HSP27), revealed equal or even higher expression levels in
2KO cells than in WT cells (Figure 3D). In response to proteotoxic
stress, HSF2 also localizes to the promoters of genes encoding
HSP90 co-chaperones and polyubiquitin (Vihervaara et al.,
2013). To study whether the regulation of these genes was
disturbed in 2KO cells, HSP90 co-chaperones PTGES3 (p23)
and AHSA1 (AHA1), as well as the polyubiquitin genes UBB
and UBC, were examined from our RNA-seq data. Since no sig-
nificant differences were observed in the expression patterns of
any of these genes (Figure 3E), we conclude that despite the
intact heat shock response, the 2KO cells were not protected
against proteotoxic stress. These findings indicate that other de-
terminants, beyond molecular chaperones, govern cell survival
during prolonged proteotoxicity.
Disruption of HSF2 Leads to Misregulation of Cell-
Adhesion-Associated Genes
To determine the differentially expressed genes between the WT
and 2KO cells, we examined the 2KO:WT comparison pair at
each experimental time point (0, 6, and 10 h) (Figure 4A). Using
the stringent cutoff criteria (fold change [FC] R3; FDR 0.001),
2KO cells were found to display significant misregulation of
819 genes already under normal growth conditions (2KO, C;
WT, C), and the proportion of upregulated and downregulated
genes remained similar throughout the BTZ treatments (2KO,
6 h; WT, 6 h: 2KO, 10 h; 2KO, 10 h) (Figure 4B; Table S1).
Gene Ontology (GO) term analysis of the misregulated genes re-
vealed a specific enrichment of terms related to cell adhesion
and cell-cell adhesion via plasma membrane adhesion mole-
cules (Figure 4C; Table S1). Similar GO terms among the com-
parison pairs implied that the genes misregulated in 2KO cells
are tightly linked to cellular adhesion properties both under con-
trol and stress conditions (Figure 4C).
To identify the adhesion molecules that are abnormally ex-
pressed in 2KO cells under both control and stress conditions,
the gene set overlaps were examined with Venn diagrams.
Among the comparison pairs, a total of 114 and 277 genes
were upregulated and downregulated, respectively (Figure 4D).
Functional cluster annotation of the 114 upregulated genes
with the DAVID analysis tool (Dennis et al., 2003) confirmed
Figure 3. Induction of the Heat Shock Response Is Not Sufficient to Protect HSF2-Deficient Cells against BTZ-Induced Proteotoxic Stress
(A) A schematic overview of the RNA-seq experiment outline. U2OS WT and 2KO cells were treated with 25 nM BTZ for 6 or 10 h. Control cells were treated with
DMSO. After treatments, mRNA was extracted and analyzed by RNA-se q. Experiments were performed in biological quadruplets. The arrows depict comparison
pairs.
(B) Differentially expressed (DE) genes in each comparison pair were determined with the Bioconductor R package Limma (Ritchie et al., 2015) (FC R3; FDR <
0.001). The upregulated and downregulated genes in a given comparison pair are indicated with red and blue bars, respectively.
(C) Normalized gene expression data for human heat shock proteins, as defined in Kampinga et al. (2009), was used to calculate the fold change of each gene in
relation WT control sample. The data are presented as a heatmap of log
2
-transformed values and were generated with GraphPad Prism7. Example s of genes that
exhibit a divergent expression pattern are framed.
(D and E) mRNA expression levels of selected heat shock proteins (HSPA1A,HSP90AA1,HSPA6,and HSPB 1) (D), HSP90 co-chaperones (PTGES3 and AHSA1),
and stress-responsive ubiquitin genes (UBB and UBC) (E) determined with RNA-seq. The data are prese nted as mean values ±SEM relative to WT control sample
that was set to 1.
See also Figure S2 and Table S1.
588 Cell Reports 30, 583–597, January 14, 2020
(legend on next page)
Cell Reports 30, 583–597, January 14, 2020 589
the strong association to cell adhesion and an extracellular
matrix, including collagens (COL16A1 and COL18A1) and
laminins (LAMB1 and LAMA5)(Figure 4E, left panel; Figure S3).
Interestingly, the 277 downregulated genes included members
from multiple cadherin sub-families, such as protocadherins
(PCDHA1 and PCDHA7), desmosomal cadherins (DSC2), and
Fat-Dachsous cadherins (FAT2), suggesting that the cadherin-
mediated cell adhesion was extensively misregulated in 2KO
cells (Figure 4E, right panel). The most prominent changes
were detected in protocadherins, as 13 distinct protocadherin
genes were significantly downregulated in 2KO cells both under
normal growth conditions and upon exposure to BTZ-induced
stress (Figure 4E).
Cells Lacking HSF2 Display Abnormal Cadherin
Expression
Cadherins are transmembrane adhesion molecules that
mediate Ca
2+
-dependent cell-cell adhesion via the conserved
extracellular cadherin domains (Hirano and Takeichi, 2012).
The human genome encodes 110 cadherin genes, which
together form the cadherin superfamily consisting of distinct
cadherin sub-families (Hirano and Takeichi, 2012). Since the
cadherin genes appeared as an HSF2-dependent gene group
and showed significant misregulation in multiple sub-family
members, we examined the expression profiles of all cadherin
superfamily genes in 2KO cells. Normalized gene expression
data were used to generate a heatmap encompassing all cad-
herin genes encoded by the human genome. By comparing
the expression profiles of WT and 2KO cells in control and
BTZ-induced stress conditions, we observed a prominent
downregulation of the entire cadherin superfamily. At least
one member from every sub-family was found downregulated
in 2KO cells, including classical cadherins (CDH2 and CDH6),
desmosomal cadherins (DSC2 and DSG2), CDH23-PCDH15
cadherins (CDH12), Fat-Dachsous cadherins (FAT2 and FAT4),
Flamingo cadherins (CELSR1), and Calsyntenins (CLSTN2)(Fig-
ure 5A; Table S1). The most striking downregulation was de-
tected in clustered a-, b-, and g-protocadherins (Peek et al.,
2017), of which a majority were found to be abnormally ex-
pressed in 2KO cells (Figure 5A). Based on these results, we
propose that cadherins are the main adhesion molecules down-
regulated in HSF2-depleted U2OS cells.
For understanding the biological relevance of the RNA-seq an-
alyses, we determined the protein expression levels of classical
cadherins (Pan-Cadherin), N-cadherin (CDH2), and clustered
g-protocadherins (Pan-PCDHgA) by immunoblotting. As shown
in Figure 5B, classical cadherins, specifically N-cadherin, and
g-protocadherins were significantly downregulated also at the
protein level (Figure 5B), and the downregulation was maintained
throughout the BTZ treatment (Figure S4A). Since cadherins are
essential in mediating Ca
2+
-dependent cell-cell contacts, we
examined the functional impact of our observations using a cell
aggregation assay, where single cells were allowed to freely
make cell-cell adhesion contacts in suspension. U2OS WT and
2KO cells were suspended in cell aggregation buffer supple-
mented with either CaCl
2
or EDTA. WT cells supplemented
with Ca
2+
formed large cell aggregates, which were completely
abolished in Ca
2+
-chelating conditions (EDTA) (Figures 5C and
S4B). In stark contrast, 2KO cells were unable to form cell aggre-
gates even in the presence of Ca
2+
(Figures 5C and S4B), indi-
cating that HSF2 is required to maintain cadherin-mediated
cell-cell contacts.
Loss of distinct cell-cell adhesion molecules has been associ-
ated with cellular inability to form three-dimensional (3D) spher-
oids in ultra-low attachment (ULA) round bottom plates (Stadler
et al., 2018). When U2OS WT and 2KO cells were grown on ULA
plates, we found that WT cells formed compact spheroids in 48
h. In contrast, 2KO cells were not able to integrate into compact
spheres, thereby occupying a significantly larger area of the ULA
plates (Figure 5D). Similar spheroid-forming phenotypes were
observed when WT and 2KO cells were grown on bacterial plates
(Figure S4C). We further explored the spheroid-forming capacity
by culturing the cells in a 3D extracellular matrix (ECM) and the
in vivo tumor growth with chicken chorioallantoic membrane
(CAM) assay. As expected, the spheroids and tumors originating
from 2KO cells were significantly smaller than the WT counter-
parts (Figures 5E and S4D), further strengthening the findings
of functional impairment of cell-cell adhesion in the absence of
HSF2. A profound decline in the expression and function of cad-
herin superfamily proteins was also observed in 2KO#2 cells
(Figures S4E–S4H), demonstrating that the alterations are not
specific for a single-cell clone. Altogether these results show
that the lack of HSF2 leads to disrupted cadherin expression at
the mRNA and protein levels, thereby resulting in deterioration
of cadherin-mediated cell-cell adhesion.
Impaired Cell-Cell Adhesion Sensitizes Cells to
Proteotoxic Stress
Although it is well acknowledged that cadherins are essential
mediators of tissue integrity and pivotal in regulating the devel-
opment of multicellular organisms (Hirano and Takeichi, 2012;
Peek et al., 2017), their impact on proteotoxic stress resistance
Figure 4. HSF2 Regulates Expression of Genes Associated with Cadherin-Mediated Cell-Cell Adhesion
(A) A schematic overview of the U2OS WT and 2KO comparison pairs.
(B) DE genes in 2KO:WT comparison pairs (control, 6 h, and 10 h) were determined with the Bioconductor R package Limma (Ritchie et al., 2015) (FC R3; FDR <
0.001). The upregulated and downregulated genes are indicated with red and blue bars, respectively.
(C) Gene Ontology (GO) terms were analyzed with topGO and GOstats packages in Bioconductor R. Biological processes from each comparison pair were
ranked according to their p values and the five most significantly changed GO terms are shown. The number of genes associated with a given term is indicated.
(D) Venn diagrams presenting the interrelationship of significantly (FC R3; FDR < 0.001) upregulated or downregulated genes in 2KO:WT comparison pairs at
control (orange), 6-h (gray), and 10-h (green) time points. Diagrams were generated using the BioVenn web application.
(E) Gene term heatmap generated with DAVID Functional Annotation Clustering Tool based on the 114 upregulated (left panel) and the 277 downregulated (right
panel) genes in 2KO cells in all treatment conditions as shown in (D). Red and blue squares denote positive association between the gene and the keyword,GO
term, or InterPro (IPR) term. Cluster enrichment score for the upregulated gene cluster is 4.39 and for the downregulated gene cluster it is 9.71.
See also Figure S3 and Table S1.
590 Cell Reports 30, 583–597, January 14, 2020
(legend on next page)
Cell Reports 30, 583–597, January 14, 2020 591
has remained unexplored. To examine whether the observed
impairment of cell-cell adhesion in 2KO cells also contributes
to the susceptibility of the cells to BTZ-induced stress, we
restored the cellular adhesion properties by re-introducing
N-cadherin to 2KO cells. N-cadherin was selected for these ex-
periments, because it is the most abundantly expressed cad-
herin superfamily member in WT U2OS cells, according to our
RNA-seq data (GEO: GSE115973), and it was found to be down-
regulated in 2KO cells. WT and 2KO cells were transfected with
either Mock or N-cadherin plasmids, and the N-cadherin expres-
sion was examined with immunoblotting (Figure 6A). As shown in
Figure 6A, we were able to restore the N-cadherin levels in 2KO
cells, which resulted in a functional rescue of cell-cell adhesion in
2KO cells (Figure 6B). Importantly, when exposed to BTZ, the
2KO cells expressing exogenous N-cadherin displayed signifi-
cantly less cleaved PARP-1 than the Mock-transfected cells
(Figures 6C, 6D, and S5), suggesting that restoration of cell-
cell adhesion can suppress cell death caused by BTZ-induced
proteotoxic stress.
All cadherin superfamily proteins are characterized by
extracellular cadherin repeat domains, which mediate homo-
philic adhesion contacts between adjacent cells (Seong et al.,
2015). Stabilization of the extracellular domains is regulated
by Ca
2+
, which binds to the interdomain regions of the
consecutive cadherin repeats and rigidifies the ectodomain
structure. To be able to comprehensively investigate the role
of cadherins in the cellular resistance to proteotoxic stress,
we first treated WT U2OS cells and MEFs with BTZ for 20 h
to induce proteotoxic stress, after which the whole cadherin-
mediated cell-cell adhesion program was destabilized by spe-
cifically depleting the extracellular Ca
2+
with EGTA (Figure 6E).
Serum-free culture conditions were used for complete deple-
tion of extracellular Ca
2+
. We observed that Ca
2+
depletion
intensified cell death, which was evidenced by the enhanced
PARP-1 and Caspase-3 cleavage in WT U2OS cells and
MEFs, respectively, after a combined treatment with both
BTZ and EGTA (Figure 6F). Altogether, these results show
that cadherin-mediated cell-cell adhesion is a key determinant
of cell survival upon BTZ treatment and that destabilization
of cadherin contacts predisposes cells to stress-induced
proteotoxicity.
DISCUSSION
Maintenance of cellular proteostasis is fundamental for the
viability of all cells and organisms (Joutsen and Sistonen,
2019). The heat shock response is critical for promoting proteo-
stasis and it is under strict control of the HSFs, among which
HSF1 is considered as the main factor responding to acute
stress. Until now, the role of HSF2 in the cellular response to
sustained proteotoxicity has remained unknown. We hypothe-
sized that HSF2 is required to protect cells against progressive
accumulation of protein damage. To test this hypothesis, we
used proteasome inhibitors (BTZ and MG132), L-Canavanine,
and HSP90 inhibitors as our experimental tools to induce
long-term proteotoxic stress. BTZ treatment has been previ-
ously shown to upregulate HSF2 at both mRNA and protein
levels in blood-derived human primary cells and to induce
HSF2 binding at designated gene loci (Rossi et al., 2014). Our
data showed that BTZ treatment also leads to a remarkable in-
crease in HSF2 protein levels in malignant human cells. More-
over, we demonstrate that the amount of nuclear HSF2 is mark-
edly increased in BTZ-treated cells, showing that HSF2
specifically responds to proteasome inhibition. We found that
HSF2 is not only activated by BTZ-induced proteotoxicity, but
it is absolutely essential for cell survival under these conditions.
Based on our results, we conclude that HSF2 is required to
protect cells against progressive accumulation of damaged
proteins.
Elevated protein levels of HSF1 and its phosphorylation on
serine 326 were recently shown to be a prerequisite for multiple
myeloma cell survival upon BTZ treatment (Shah et al., 2016; Fok
et al., 2018). Therefore, we explored whether HSF2 depletion
sensitizes cells to BTZ through misregulated HSF1, specifically,
or the heat shock response in general. Neither difference in HSF1
levels nor serine 326 phosphorylation was detected between WT
and HSF2-depleted cells treated with proteasome inhibitors.
Strikingly, the classical heat shock response, as characterized
by the global upregulation of molecular chaperones, HSP90
co-chaperones, and polyubiquitin genes, was not compromised
in cells lacking HSF2. These results indicate that HSF2 promotes
cell survival independently of HSF1. Thus, we provide evidence
that the ability to survive proteotoxic stress does not solely
Figure 5. HSF2 Controls Cellular Adhesion Properties through Cadherin Superfamily Proteins
(A) Normalized gene expression data from the RNA-seq analysi s for cadherin superfamily genes, as defined in Hirano and Takeichi (2012), was used to calculate
the fold change of each gene in relation to respective expression in the WT control sample. The data are presented as a heatmap of log
2
-transformed fold changes
and were generated with GraphPad Prism7. N-cadherin and protocadherin gamma subfamily A (PCDHgA) were chosen for further analyses.
(B) Immunoblot analysis of classical cadherins, N-cadherin, and the members of PCDHgA in U2OS WT and 2KO cells. Lack of HSF2 expressio nin 2KO cells was
confirmed and HSC70 was used as a loading control. The amount of cadherins relative to respective HSC70 level was quantified with ImageJ. The data are
presented as mean values of three independent experiments + SEM; ***p < 0.001 and ****p < 0.0001.
(C) Cell aggregation assay of U2OS WT and 2KO cells suspended in cell aggregation buffer supplemented with 3 mM CaCl
2
(Ca
2+
) or 3 mM EDTA. Cells were
rotated for 2.5 h at 37C and visualized with bright-field microscopy. Scale bar, 1 mm.
(D) Bright-field microscopy images of U2OS WT and 2KO cells cultured in ULA plates. Cells were imaged after 24 and 48 h. Scale bar, 200 mm. The size of the
spheroid area was quantified with ImageJ. The data are presented as mean values of three independent experiments + SEM. *p < 0.05 and **p < 0.01.
(E) Confocal microscopy images of U2OS WT and 2KO cells. Cells were cultured in 3D in Matrigel for 5, 8, and 13 days. At the indicated days, spheroids were
fixed, and F-actin was stained with Alexa 488-labeled phalloidin (green). DAPI was used to stain the nuclei (blue). Zstacks of the spheroids were imaged with a
spinning disc confocal microscope. The maximum intensity projection images represent the average spheroid size for each cell line at indicated time points from
three biological repeats. Scale bar, 10 mm. The volume of the spheroids was quantified with Image J with the 3D Object Counter v2.0 plugin (Bolte and Cordelie
`res,
2006). The data are presented as mean values of three independent experiments + SEM. ***p < 0.001.
See also Figure S4.
592 Cell Reports 30, 583–597, January 14, 2020
(legend on next page)
Cell Reports 30, 583–597, January 14, 2020 593
depend on the induction of molecular chaperones but engages a
larger repertoire of cellular pathways and properties.
To our surprise, despite the stringent cutoff criteria (FC R3;
FDR 0.001), we found a considerable number of genes display-
ing altered expression profiles in cells lacking HSF2. Among
the most prominently misregulated genes were those belonging
to the cadherin superfamily. Here, we demonstrate that lack of
HSF2 leads to a profound downregulation of cadherins both at
mRNA and protein levels, identifying HSF2 as a key regulator
of cadherin genes. Cadherins are a large group of transmem-
brane adhesion molecules, which mediate Ca
2+
-dependent
cell-cell adhesion and thereby function as essential mediators
of tissue integrity (Hirano and Takeichi, 2012). We found that
HSF2-deficient cells display functional impairment of cadherin-
mediated cell-cell adhesion already under normal growth
conditions. Together with earlier results of HSF2 displaying
DNA-binding capacity already in the absence of stress (Sarge
et al., 1991; A
˚kerfelt et al., 2008; Vihervaara et al., 2013), these
results suggest that HSF2 has a physiological role in regulating
cadherin functions. Excitingly, impaired migration and misposi-
tioning of neurons have been shown to underlie the corticogen-
esis defects in Hsf2
/
mice (Kallio et al., 2002; Chang et al.,
2006), and cadherin superfamily proteins are fundamental for
correct neuronal migration (Hayashi and Takeichi, 2015). Thus,
it is tempting to speculate that the HSF2-dependent disruption
of cadherin-mediated cell-cell contacts contributes to the
abnormal corticogenesis of Hsf2
/
mice.
The downregulation of cadherin gene expression raises
important questions about the mechanisms by which HSF2 reg-
ulates these genes. Genome-wide mapping of HSF2 binding
sites has been previously determined with chromatin immuno-
precipitation sequencing (ChIP-seq) in human K562 erythroleu-
kemia cells (Vihervaara et al., 2013) and in mouse testis (Korfanty
et al., 2014). Remarkably, both studies identified HSF2 occu-
pancy on multiple cadherin superfamily genes. Since non-
adherent K562 cells are deficient of endogenously expressed
classical cadherins and distinct protocadherins (Ozawa and
Kemler, 1998), it is not surprising that HSF2 was found to occupy
only the CLSTN gene under control growth conditions (Viher-
vaara et al., 2013). However, upon acute heat stress, HSF2 bind-
ing was observed at classical cadherins (CDH4), desmogleins
(DSG2), Fat-Dachous cadherins (DCHS2), Flamingo cadherins
(CELSR2), and CDH23-PCDH15 cadherins (CDH23)(Vihervaara
et al., 2013), demonstrating that multiple genes belonging to the
cadherin superfamily can be targeted by HSF2 in human cells. In
mouse testis, HSF2 was also shown to occupy several cadherin
genes, including CDH15,CDH5,CDH18,CDH13,FAT1,PCDH9,
PCDH17, and PCDHA1 (Korfanty et al., 2014). Importantly, we
now demonstrate the functional relevance of HSF2-mediated
cadherin regulation and propose HSF2 as a central regulator of
cadherin genes.
Failure in the maintenance of proteostasis is a hallmark of aging
and neurodegenerative diseases (Douglas and Dillin, 2010).
Intriguingly, in a mouse model of Huntington’s disease, lack of
HSF2 was shown to predispose mouse brain to poly-Q aggre-
gates and reduce lifespan (Shinkawa et al., 2011), suggesting
that HSF2 is required to protect neurons from progressive accu-
mulation of damaged proteins. Cell survival upon proteotoxic
stress has been conventionally considered to depend on induc-
ible transcriptional programs, such as the heat shock response
or the unfolded protein response (Walter and Ron, 2011; Go-
mez-Pastor et al., 2018). However, in this study, we show that
HSF2-dependent maintenance of cell-cell adhesion is an essen-
tial determinant of proteotoxic stress resistance. Our results indi-
cate that misregulation of distinct cellular properties already
under normal growth conditions can sensitize cells to proteotox-
icity. HSF1 and HSF2 represent the two arms of the cellular resis-
tance toward proteotoxic stress; HSF1 as an acute responder to
protein damage and HSF2 as a factor maintaining the long-term
stress resistance. Notably, in a meta-analysis of transcriptional
changes associated with Alzheimer’s disease and aging, HSF2
was identified as a gene commonly downregulated during aging
(Ciryam et al., 2016). Therefore, it is possible that the age-associ-
ated downregulation of HSF2 and subsequent disruption of cad-
herin-mediated cell-cell adhesion participates in sensitizing cells,
such as neurons, to aggregate mismanagement.
STAR+METHODS
Detailed methods are provided in the online version of this paper
and include the following:
dKEY RESOURCES TABLE
dLEAD CONTACT AND MATERIALS AVAILABILITY
Figure 6. Impaired Cell-Cell Contacts Sensitize Cells to BTZ-Induced Proteotoxic Stress
(A–D) N-cadherin levels in U2OS 2KO cells were restored to those in WT cells by transiently transfecting the cells with either Mock or N-cadherin plasmids co-
expressing GFP.
(A) Immunoblot analysis of HSF2 and N-cadherin. HSC70 was used as a loading control.
(B) Cell aggregation assay was performed as in Figure 5C. Cell aggregates were imaged with bright-field (BF) and fluorescence filters (GFP). Scale bar, 500 mm.
(C) For immunoblot analysis of HSF2 and PARP-1, cells were treated with 25 nM BTZ for 22 h. Control cells were treated with DMSO. HSC70 was used as a
loading control. The amount of cleaved PARP-1 relative to the respective HSC70 level was quantified with ImageJ. The data are presented as mean values of
three independent experiments + SEM; *p < 0.05.
(D) Flow cytometry analysis of fluorescently labeled cleaved PARP-1 antibody. Cells were treated with 25 nM BTZ for 22 h. Control cells were treated with DMSO.
The data are presented as mean values of three independent experiments + SEM; *p < 0.05. The statistical analysis was performed with a Student’s t test.
(E) A schematic overview of the calcium-depletion experiments.
(F) U2OS WT cells were treated with or without 25 nM BTZ for 20 h in serum-free growth medium after which the extracellular calcium was depleted with 4 mM
EGTA and BTZ treatment continued for 2 h. MEFs were treated with 5 or 10 nM BTZ for 20 h in serum-free growth medium after which the extracellular calcium
was depleted with 2 mM EGTA and BTZ continued for 2 h. Control cells were treated with DMSO. PARP-1 and Caspase-3 cleavage was assessed with
immunoblotting. Cells were imaged with a bright-field microscope. Scale bar, 200 mm.
See also Figure S5.
594 Cell Reports 30, 583–597, January 14, 2020
dEXPERIMENTAL MODEL AND SUBJECT DETAILS
BGeneration of HSF2 knock-out U2OS cells with
CRISPR-Cas9
BCell culture
BChicken chorioallantoic membrane (CAM) assay
dMETHOD DETAILS
BTreatments
BTransfections
BImmunoblotting
BImmunofluorescence
BSubcellular fractionation
BCell viability measurements
BCell aggregation assays
BRNA-sequencing
BFlow cytometry
BQuantitative RT-PCR (qRT-PCR)
B3D cell culture and immunofluorescence
BVisualization of the data
dQUANTIFICATION AND STATISTICAL ANALYSIS
BBioinformatic analysis of the RNA-seq data
BOther data analyses
dDATA AND CODE AVAILABILITY
SUPPLEMENTAL INFORMATION
Supplemental Information can be found online at https://doi.org/10.1016/j.
celrep.2019.12.037.
ACKNOWLEDGMENTS
We thank Joshua Weiner (University of Iowa, Iowa, US) for helpful discussions
and advice and for providing us with the anti-PanPCDHgA antibody. Mikael
Puustinen from the Sistonen laboratory is acknowledged for his assistance
with the CAM-assay. All members of the Sistonen laboratory and Ve
´ronique
Dubreuil from the Mezger laboratory are thanked for their valuable comments
and critical review of the manuscript. Imaging was performed at the Cell Imag-
ing Core, Turku Bioscience Centre, University of Turku and A
˚bo Akademi Uni-
versity. The instruments used in this project belong to the infrastructure of Bio-
center Finland. We thank Markku Saari and Jouko Sandholm from the Cell
Imaging Core of Turku Bioscience Centre for technical assistance and advice.
The Bioinformatics unit of Turku Bioscience Centre is acknowledged for their
assistance with the RNA-seq data analysis. The Bioinformatics unit is sup-
ported by University of Turku, A
˚bo Akademi University, and Biocenter Finland.
This study has been funded by the Academy of Finland (L.S.); Sigrid Juselius
Foundation (L.S.); Turku Doctoral Network in Molecular Biosciences
(A.J.D.S. and J.C.L.); Finnish Cultural Foundation (J.J.); Cancer Foundation
Finland (J.J. and L.S.); A
˚bo Akademi University Research Foundation (J.J.
and M.A.B.); Magnus Ehrnrooth Foundation (A.J.D.S. and L.S.); Tor, Joe and
Pentti Borg Memory Foundation (J.J.); Ida Montin’s Foundation (J.J.); Otto
A. Malm Foundation (J.J. and A.J.D.S.); the Medical Research Foundation
Liv och Ha
¨lsa (J.J. and L.S.); K. Albin Johansson’s Foundation (J.J.,
A.J.D.S., and E.H.); Agence Nationale Recherche (Program SAMENTA ANR-
13-SAMA-0008-01; A.d.T., V.M., and D.S.-D.); Short Researcher Mobility
France Embassy/MESRI-Finnish Society of Science and Letters (V.M.); and
CNRS/Project International de Coope
´ration Scientifique PICS 2013-2015
(A.d.T., V.M., and D.S.-D.).
AUTHOR CONTRIBUTIONS
J.J., A.J.D.S., and L.S. designed the research; J.J., A.J.D.S., J.C.L., M.A.B.,
A.S.N., and E.H. performed the experiments; A.d.T., J.-P.C., V.M., and
D.S.-D. generated and analyzed the U2OS cell lines; J.J., A.J.D.S., J.C.L.,
E.H., and L.S. analyzed the data; and J.J., A.J.D.S., and L.S. wrote the manu-
script with all authors providing feedback.
DECLARATION OF INTERESTS
The authors declare no competing interests.
Received: June 19, 2019
Revised: October 15, 2019
Accepted: December 12, 2019
Published: January 14, 2020
REFERENCES
Ahlskog, J.K., Bjo
¨rk, J.K., Elsing, A.N., Aspelin, C., Kallio, M., Roos-Mattjus, P.,
and Sistonen, L. (2010). Anaphase-promoting complex/cyclosome partici-
pates in the acute response to protein-damaging stress. Mol. Cell. Biol. 30,
5608–5620.
A
˚kerfelt, M., Henriksson, E., Laiho, A., Vihervaara, A., Rautoma, K., Kotaja, N.,
and Sistonen, L. (2008). Promoter ChIP-chip analysis in mouse testis reveals Y
chromosome occupancy by HSF2. Proc. Natl. Acad. Sci. USA 105, 11224–
11229.
A
˚kerfelt, M., Morimoto, R.I., and Sistonen, L. (2010). Heat shock factors: inte-
grators of cell stress, development and lifespan. Nat. Rev. Mol. Cell Biol. 11,
545–555.
Alastalo, T.P., Lo
¨nnstro
¨m, M., Leppa
¨, S., Kaarniranta, K., Pelto-Huikko, M.,
Sistonen, L., and Parvinen, M. (1998). Stage-specific expression and cellular
localization of the heat shock factor 2 isoforms in the rat seminiferous epithe-
lium. Exp. Cell Res. 240, 16–27.
Alexa, A., and Rahnenfuhrer, J. (2019). topGO: Enrichment Analysis for Gene
Ontology. R package version 2.37.0. https://bioconductor.org (Bioconductor).
Andrews, S. (2010). FastQC: a quality control tool for high throughput
sequence data. Babraham Bioinformatics. www.bioinformatics.babraham.
ac.uk/projects/fastqc.
Bjo
¨rk, J.K., Ahonen, I., Mirtti, T., Erickson, A., Rannikko, A., B€
utzow, A., Nor-
dling, S., Lundin, J., Lundin, M., Sistonen, L., Nees, M., and A
˚kerfelt, M.
(2018). Increased HSF1 expression predicts shorter disease-specific survival
of prostate cancer patients following radical prostatectomy. Oncotarget 9,
31200–31213.
Bjo
¨rk, J.K., A
˚kerfelt, M., Joutsen, J., Puustinen, M.C., Cheng, F., Sistonen, L.,
and Nees, M. (2016). Heat-shock factor 2 is a suppressor of prostate cancer
invasion. Oncogene 35, 1770–1784.
Bolte, S., and Cordelie
`res, F.P. (2006). A guided tour into subcellular colocal-
ization analysis in light microscopy. J. Microsc. 224, 213–232.
Budzy
nski, M., and Sistonen, L. (2017). Versatile Functions of Heat Shock Fac-
tors: It Is Not All about Stress. Curr. Immunol. Rev. 13, 4–18.
Chang, Y., O
¨stling, P., A
˚kerfelt, M., Trouillet, D., Rallu, M., Gitton, Y., El Fatimy,
R., Fardeau, V., Le Crom, S., Morange, M., et al. (2006). Role of heat-shock
factor 2 in cerebral cortex formation and as a regulator of p35 expression.
Genes Dev. 20, 836–847.
Chen, D., Frezza, M., Schmitt, S., Kanwar, J., and Dou, Q.P. (2011). Bortezo-
mib as the first proteasome inhibitor anticancer drug: current status and future
perspectives. Curr. Cancer Drug Targets 11, 239–253.
Ciryam, P., Kundra, R., Freer, R., Morimoto, R.I., and Dobson, C.M. (2016). A
transcriptional signature of Alzheimer’s disease is associated with a meta-
stable subproteome at risk for aggregation. Proc. Natl. Acad. Sci. USA 113,
4753–4758.
Dennis, G., Jr., Sherman, B.T., Hosack, D.A., Yang, J., Gao, W., Lane, H.C.,
and Lempicki, R.A. (2003). DAVID: Database for Annotation, Visualization,
and Integrated Discovery. Genome Biol. 4,3.
Deshaies, R.J. (2014). Proteotoxic crisis, the ubiquitin-proteasome system,
and cancer therapy. BMC Biol. 12,94.
Douglas, P.M., and Dillin, A. (2010). Protein homeostasis and aging in neuro-
degeneration. J. Cell Biol. 190, 719–729.
Cell Reports 30, 583–597, January 14, 2020 595
Elsing, A.N., Aspelin, C., Bjo
¨rk, J.K., Bergman, H.A., Himanen, S.V., Kallio,
M.J., Roos-Mattjus, P., and Sistonen, L. (2014). Expression of HSF2 decreases
in mitosis to enable stress-inducible transcription and cell survival. J. Cell Biol.
206, 735–749.
Falcon, S., and Gentleman, R. (2007). Using GOstats to test gene lists for GO
term association. Bioinformatics 23, 257–258.
Fok, J.H.L., Hedayat, S., Zhang, L., Aronson, L.I., Mirabella, F., Pawlyn, C.,
Bright, M.D., Wardell, C.P., Keats, J.J., De Billy, E., et al. (2018). HSF1 Is essen-
tial for myeloma cell survival and a promising therapeutic target. Clin. Cancer
Res. 24, 2395–2407.
Gentleman, R.C., Carey, V.J., Bates, D.M., Bolstad, B., Dettling, M., Dudoit, S.,
Ellis, B., Gautier, L., Ge, Y., Gentry, J., et al. (2004). Bioconductor: open soft-
ware development for computational biology and bioinformatics. Genome
Biol. 5, R80.
Goldberg, A.L. (2012). Development of proteasome inhibitors as research tools
and cancer drugs. J. Cell Biol. 199, 583–588.
Gomez-Pastor, R., Burchfiel, E.T., and Thiele, D.J. (2018). Regulation of heat
shock transcription factors and their roles in physiology and disease. Nat.
Rev. Mol. Cell Biol. 19, 4–19.
Gonsalves, S.E., Moses, A.M., Razak, Z., Robert, F., and Westwood, J.T.
(2011). Whole-genome analysis reveals that active heat shock factor binding
sites are mostly associated with non-heat shock genes in Drosophila mela-
nogaster. PLoS ONE 6, e15934.
Guettouche, T., Boellmann, F., Lane, W.S., and Voellmy, R. (2005). Analysis of
phosphorylation of human heat shock factor 1 in cells experiencing a stress.
BMC Biochem. 6,4.
Hahn, J.-S., Hu, Z., Thiele, D.J., and Iyer, V.R. (2004). Genome-wide analysi s of
the biology of stress responses through heat shock transcription factor. Mol.
Cell. Biol. 24, 5249–5256.
Ha
¨rma
¨, V., Virtanen, J., Ma
¨kela
¨, R., Happonen, A., Mpindi, J.P., Knuuttila, M.,
Kohonen, P., Lo
¨tjo
¨nen, J., Kallioniemi, O., and Nees, M. (2010). A comprehen-
sive panel of three-dimensional models for studies of prostate cancer growth,
invasion and drug responses. PLoS ONE 5, e10431.
Hartl, F.U., Bracher, A., and Hayer-Hartl, M. (2011). Molecular chaperones in
protein folding and proteostasis. Nature 475, 324–332.
Hayashi, S., and Takeichi, M. (2015). Emerging roles of protocadherins: from
self-avoidance to enhancement of motility. J. Cell Sci. 128, 1455–1464.
Hershko, A., and Ciechanover, A. (1998). The ubiquitin system. Annu. Rev. Bio-
chem. 67, 425–479.
Hirano, S., and Takeichi, M. (2012). Cadherins in brain morphogenesis and wir-
ing. Physiol. Rev. 92, 597–634.
Huang, W., Sherman, B.T., and Lempicki, R.A. (2009). Systematic and integra-
tive analysis of large gene lists using DAVID bioinformatics resources. Nat.
Protoc. 4, 44–57.
Hulsen, T., de Vlieg, J., and Alkema, W. (2008). BioVenn - a web app lication for
the comparison and visualization of biological lists using area-proportional
Venn diagrams. BMC Genomics 9, 488.
Jaeger, A.M., Pemble, C.W., 4th, Sistonen, L., and Thiele, D.J. (2016). Struc-
tures of HSF2 reveal mechanisms for differential regulation of human heat-
shock factors. Nat. Struct. Mol. Biol. 23, 147–154.
Joutsen, J., and Sistonen, L. (2019). Tailoring of Proteostasis Networks with
Heat Shock Factors. Cold Spring Harb. Perspect. Biol. 11, a034066.
Kallio, P.J., Wilson, W.J., O’Brien, S., Makino, Y., and Poellinger, L. (1999).
Regulation of the hypoxia-inducible transcription factor 1aby the ubiquitin-
proteasome pathway. J. Biol. Chem. 274, 6519–6525.
Kallio, M., Chang, Y., Manuel, M., Alastalo, T.P., Rallu, M., Gitton, Y., Pirkkala,
L., Loones, M.T., Paslaru, L., Larney, S., et al. (2002). Brain abnormalities,
defective meiotic chromosome synapsis and female subfertility in HSF2 null
mice. EMBO J. 21, 2591–2601.
Kampinga, H.H., Hageman, J., Vos, M.J., Kubota, H., Tanguay, R.M., Bruford,
E.A., Cheetham, M.E., Chen, B., and Hightower, L.E. (2009). Guidel ines for the
nomenclature of the human heat shock proteins. Cell Stress Chaperones 14,
105–111.
Kawazoe, Y., Nakai, A., Tanabe, M., and Nagata, K. (1998). Proteasome inhi-
bition leads to the activation of all members of the heat-shock-factor family.
Eur. J. Biochem. 255, 356–362.
Kim, D., Pertea, G., Trapnell, C., Pimentel, H., Kelley, R., and Salzberg, S.L.
(2013). TopHat2: accurate alignment of transcriptomes in the presence of in-
sertions, deletions and gene fusions. Genome Biol. 14, R36.
Kisselev, A.F., Callard, A., and Goldberg, A.L. (2006). Importance of the
different proteolytic sites of the proteasome and the efficacy of inhibitors varies
with the protein substrate. J. Biol. Chem. 281, 8582–8590.
Kobayashi, A., Kang, M.I., Okawa, H., Ohtsuji, M., Zenke, Y., Chiba, T., Igara-
shi, K., and Yamamoto, M. (2004). Oxidative stress sensor Keap1 functions as
an adaptor for Cul3-based E3 ligase to regulate proteasomal degradation of
Nrf2. Mol. Cell. Biol. 24, 7130–7139.
Korfanty, J., Stokowy, T., Widlak, P., Gogler-Piglowska, A., Handschuh, L.,
Podkowi
nski, J., Vydra, N., Naumowicz, A., Toma-Jonik, A., and Widlak, W.
(2014). Crosstalk between HSF1 and HSF2 during the heat shock response
in mouse testes. Int. J. Biochem. Cell Biol. 57, 76–83.
Kubbutat, M.H., Jones, S.N., and Vousden, K.H. (1997). Regulation of p53 sta-
bility by Mdm2. Nature 387, 299–303.
Lecomte, S., Desmots, F., Le Masson, F., Le Goff, P., Michel, D., Christians,
E.S., and Le Dre
´an, Y. (2010). Roles of heat shock factor 1 and 2 in response
to proteasome inhibition: consequence on p53 stability. Oncogene 29, 4216–
4224.
Li, J., Chauve, L., Phelps, G., Brielmann, R.M., and Morimoto, R.I. (2016). E2F
coregulates an essential HSF developmental program that is distinct from the
heat-shock response. Genes Dev. 30, 2062–2075.
Liao, Y., Smyth, G.K., and Shi, W. (2013). The Subread aligner: fast, accurate
and scalable read mapping by seed-and-vote. Nucleic Acids Res. 41, e108.
Ling, Y.H., Liebes, L., Ng, B., Buckley, M., Elliott, P.J., Adams, J., Jiang, J.D.,
Muggia, F.M., and Perez-Soler, R. (2002). PS-341, a novel proteasome inhib-
itor, induces Bcl-2 phosphorylation and cleavage in association with G2-M
phase arrest and apoptosis. Mol. Cancer Ther. 1, 841–849.
Mahat, D.B., Salamanca, H.H., Duarte, F.M., Danko, C.G., and Lis, J.T. (2016).
Mammalian heat shock response and mechanisms underlying its genome-
wide transcriptional regulation. Mol. Cell 62, 63–78.
Mali, P., Yang, L., Esvelt, K.M., Aach, J., Guell, M., DiCarlo, J.E., Norville, J.E.,
and Church, G.M. (2013). RNA-guided human genome engineering via Cas9.
Science 339, 823–826.
Mathew, A., Mathur, S.K., and Morimoto, R.I. (1998). Heat shock response and
protein degradation: regulation of HSF2 by the ubiquitin-proteasome pathway .
Mol. Cell. Biol. 18, 5091–5098.
Mendillo, M.L., Santagata, S., Koeva, M., Bell, G.W., Hu, R., Tamimi, R.M.,
Fraenkel, E., Ince, T.A., Whitesell, L., and Lindquist, S. (2012). HSF1 drives a
transcriptional program distinct from heat shock to support highly malignant
human cancers. Cell 150, 549–562.
Mezger, V., Rallu, M., Morimoto, R.I., Morange, M., and Renard, J.P. (1994).
Heat shock factor 2-like activity in mouse blastocysts. Dev. Biol. 166, 819–822 .
O
¨stling, P., Bjo
¨rk, J.K., Roos-Mattjus, P., Mezger, V., and Sistonen, L. (2007).
Heat shock factor 2 (HSF2) contributes to inducible expression of hsp genes
through interplay with HSF1. J. Biol. Chem. 282, 7077–7086.
Ozawa, M., and Kemler, R. (1998). Altered cell adhesion activity by pervana-
date due to the dissociation of alpha-catenin from the E-cadherin.catenin
complex. J. Biol. Chem. 273, 6166–6170.
Peek, S.L., Mah, K.M., and Weiner, J.A. (2017). Regulation of neural circuit for-
mation by protocadherins. Cell. Mol. Life Sci. 74, 4133–4157.
Pirkkala, L., Alastalo, T.-P., Zuo, X., Benjamin, I.J., and Sistonen, L. (2000).
Disruption of heat shock factor 1 reveals an essential role in the ubiquitin pro-
teolytic pathway. Mol. Cell. Biol. 20, 2670–2675.
596 Cell Reports 30, 583–597, January 14, 2020
Rallu, M., Loones, M., Lallemand, Y., Morimoto, R., Morange, M., and Mezger,
V. (1997). Function and regulation of heat shock factor 2 during mouse
embryogenesis. Proc. Natl. Acad. Sci. USA 94, 2392–2397.
Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., and Smyth, G.K.
(2015). limma powers differential expression analyses for RNA-sequencing
and microarray studies. Nucleic Acids Res. 43, e47.
Riva, L., Koeva, M., Yildirim, F., Pirhaji, L., Dinesh, D., Mazor, T., Duennwald,
M.L., and Fraenkel, E. (2012). Poly-glutamine expanded huntingtin dramati-
cally alters the genome wide binding of HSF1. J. Huntingtons Dis. 1, 33–45.
Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010). edgeR: a Bio-
conductor package for differential expression analysis of digital gene expres-
sion data. Bioinformatics 26, 139–140.
Rossi, A., Riccio, A., Coccia, M., Trotta, E., La Frazia, S., and Santoro, M.G.
(2014). The proteasome inhibitor bortezomib is a potent inducer of zinc finger
AN1-type domain 2a gene expression: role of heat shock factor 1 (HSF1)-heat
shock factor 2 (HSF2) heterocomplexes. J. Biol. Chem. 289, 12705–12715.
Rueden, C.T., Schindelin, J., Hiner, M.C., DeZonia, B.E., Walter, A.E., Arena,
E.T., and Eliceiri, K.W. (2017). ImageJ2: ImageJ for the next generation of sci-
entific image data. BMC Bioinformatics 18, 529.
Sarge, K.D., Zimarino, V., Holm, K., Wu, C., and Morimoto, R.I. (1991). Cloning
and characterization of two mouse heat shock factors with distinct inducible
and constitutive DNA-binding ability. Genes Dev. 5, 1902–1911.
Sarge, K.D., Murphy, S.P., and Morimoto, R.I. (1993). Activation of heat shock
gene transcription by heat shock factor 1 involves oligomerization, acquisition
of DNA-binding activity, and nuclear localization and can occur in the absence
of stress. Mol. Cell. Biol. 13, 1392–1407.
Sarge, K.D., Park-Sarge, O.K., Kirby, J.D., Mayo, K.E., and Morimoto, R.I.
(1994). Expression of heat shock factor 2 in mouse testis: potential role as a
regulator of heat-shock protein gene expression during spermatogenesis.
Biol. Reprod. 50, 1334–1343.
Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch,
T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Fiji: an
open-source platform for biological-image analysis. Nat. Methods 9, 676–682.
Seong, E., Yuan, L., and Arikkath, J. (2015). Cadherins and catenins in dendrite
and synapse morphogenesis. Cell Adhes. Migr. 9, 202–213.
Shah, S.P., Nooka, A.K., Jaye, D.L., Bahlis, N.J., Lonial, S., and Boise, L.H.
(2016). Bortezomib-induced heat shock response protects multiple myeloma
cells and is activated by heat shock factor 1 serine 326 phosphorylation.
Oncotarget 7, 59727–59741.
Sheldon, L.A., and Kingston, R.E. (1993). Hydrophobic coiled-coil domains
regulate the subcellular localization of human heat shock factor 2. Genes
Dev. 7, 1549–1558.
Shinkawa, T., Tan, K., Fujimoto, M., Hayashida, N., Yamamoto, K., Takaki, E.,
Takii, R., Prakasam, R., Inouye, S., Mezger, V., and Nakai, A. (2011). Heat
shock factor 2 is required for maintaining proteostasis against febrile-range
thermal stress and polyglutamine aggregation. Mol. Biol. Cell 22, 3571–3583.
Sistonen, L., Sarge, K.D., and Morimoto, R.I. (1994). Human heat sho ck factors
1 and 2 are differentially activated and can synergistically induce hsp70 gene
transcription. Mol. Cell. Biol. 14, 2087–2099.
Stadler, M., Scherzer, M., Walter, S., Holzner, S., Pudelko, K., Riedl, A., Unger,
C., Kramer, N., Weil, B., Neesen, J., et al. (2018). Exclusion from spheroid for-
mation identifies loss of essential cell-cell adhesion molecules in colon cancer
cells. Sci. Rep. 8, 1151.
Varshavsky, A. (2012). The ubiquitin system, an immense realm. Annu. Rev.
Biochem. 81, 167–176.
Vihervaara, A., Sergelius, C., Vasara, J., Blom, M.A.H., Elsing, A.N., Roos-
Mattjus, P., and Sistonen, L. (2013). Transcriptional response to stress in the
dynamic chromatin environment of cycling and mitotic cells. Proc. Natl.
Acad. Sci. USA 110, E3388–E3397.
Vihervaara, A., Duarte, F.M., and Lis, J.T. (2018). Molecular mechanisms
driving transcriptional stress responses. Nat. Rev. Genet. 19, 385–397.
Vihervaara, A., Mahat, D.B., Guertin, M.J., Chu, T., Danko, C.G., Lis, J.T., and
Sistonen, L. (2017). Transcriptional response to stress is pre-wired by pro-
moter and enhancer architecture. Nat. Commun. 8, 255.
Walter, P., and Ron, D. (2011). The unfolded protein response: from stress
pathway to homeostatic regulation. Science 334, 1081–1086.
Zhang, B., Groffen, J., and Heisterkamp, N. (2007). Increased resistance to
a farnesyltransferase inhibitor by N-cadherin expression in Bcr/Abl-P190
lymphoblastic leukemia cells. Leukemia 21, 1189–1197.
Cell Reports 30, 583–597, January 14, 2020 597
STAR+METHODS
KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
Rabbit polyclonal anti-GAPDH Abcam Cat#Ab9485; RRID:AB_307275
Rat monoclonal anti-HSC70 Enzo Life Sciences Cat#ADI-SPA-815; RRID:AB_10617277
Rabbit polyclonal anti-HSF1 Enzo Life Sciences Cat#ADI-SPA-901; RRID:AB_10616511
Rabbit polyclonal anti-HSF1 p326 Abcam Cat#Ab76076; RRID:AB_1310328
Rabbit polyclonal anti-HSF2 Sigma-Aldrich Cat#HPA031455; RRID:AB_10670702
Mouse anti-HSP70 Enzo Life Sciences Cat#ADI-SPA-810; RRID:AB_10616513
Rabbit monoclonal anti-N-cadherin Millipore Cat#04-1126; RRID:AB_1977064
Rabbit polyclonal anti-N-cadherin Abcam Cat#Ab76057; RRID:AB_1310478
Mouse monoclonal anti-PARP-1 Santa Cruz Biotechnology Cat#Sc-8007; RRID:AB_628105
Rabbit polyclonal anti-cleaved Caspase 3 Abcam Cat#Ab2302; RRID:AB_302962
Mouse monoclonal anti-Pan-PCDHgA NeuroMab Cat#75-178; RRID:AB_2159447
Rabbit polyclonal anti-Pan-Cadherin Abcam Cat#Ab6529; RRID:AB_305545
Mouse monoclonal anti-Lamin A/C Cell Signaling Technology Cat#4777S; RRID:AB_10545756
Mouse monoclonal anti- b-tubulin Sigma-Aldrich Cat#T8328; RRID:AB_1844090
Mouse monoclonal anti-cleaved PARP antibody conjugated
to BV421
Cat#564129; RRID:AB_2738611
Goat anti-rabbit Alexa Fluor 488 Invitrogen Cat# R37116; RRID:AB_2556544
Chemicals, Peptides, and Recombinant Proteins
Bortezomib Santa Cruz Biotechnology Cat#sc-217785
MG132 Peptide Institute Inc. Cat#317-V
17-AAG InvivoGen Cat#anti-agl-5
Geldanamycin InvivoGen Cat#anti-gl-5
L-Canavanine sulfate salt Sigma-Aldrich Cat#C9758
Alexa Fluor 488 Phalloidin Thermo Fisher Scientiffic Cat#A12379; Cat #A22287
Critical Commercial Assays
CellTiter-Glo reagent Promega Cat#G7570
Calcein AM R&D Systems Cat#4892-010-K
AllPrep DNA/RNA/miRNA Universal Kit QIAGEN Cat#80224
Matrigel Corning Cat#356231
RNeasy mini kit QIAGEN Cat#74106
iScript cDNA Synthesis Kit Bio-Rad Cat#1708891
GenJet SignaGen Laboratories Cat#SL100489-OS
Deposited Data
RNA-seq raw data This paper GEO: GSE115973
Experimental Models: Cell Lines
U2OS wild-type This paper N/A
U2OS HSF2 knock out This paper N/A
U2OS HSF2 knock out clone 2 This paper N/A
Mouse embryonic fibroblasts wild-type O
¨stling et al., 2007 N/A
Mouse embryonic fibroblasts Hsf2
/
O
¨stling et al., 2007 N/A
Experimental Models: Organisms/Strains
Chicken: fertilized white Leghorn chicken eggs Munax Oy N/A
(Continued on next page)
e1 Cell Reports 30, 583–597.e1–e6, January 14, 2020
LEAD CONTACT AND MATERIALS AVAILABILITY
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Lea Sis-
tonen (lea.sistonen@abo.fi). The resources are shared for research and educational purposes without restriction.
Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Oligonucleotides
Primers for generating HSF2 knock-out U2OS cells
with CRISPR-Cas9 see Table S2
This paper N/A
Primer RNA18S5 forward: GCAATTATTCCCCATGAACG Sigma-Aldrich N/A
Primer RNA18S5 reverse: GGGACTTAATCAACGCAAGC Sigma-Aldrich N/A
Probe RNA18S5: FAM-TTCCCAGTAAGTGCG GGTC-BHQ Sigma-Aldrich N/A
Primer DSC2 forward: ATCCATTAGAGGACACACTCTGA Sigma-Aldrich N/A
Primer DSC2 reverse: GCCACCGATCCTCTTCCTTC Sigma-Aldrich N/A
Primer PCDHA6 forward: TGACTGTTGAATGATGGCGGA Sigma-Aldrich N/A
Primer PCDHA6 reverse: TCGGGTACGGAGTAGTGGAG Sigma-Aldrich N/A
Primer PCDH10A forward: AGGCATCAGCCAGTTTCTCAA Sigma-Aldrich N/A
Primer PCDH10A reverse: GAGAGCAGCAGACACTGGAC Sigma-Aldrich N/A
Recombinant DNA
pMLM3636, Human-gRNA-Expression Vector Keith Joung laboratory, Addgene RRID:Addgene_43860
pcDNA3.3-TOPO hCas9 Mali et al., 2013; Addgene RRID:Addgene_41815
pEGFP-N1 Clontech N/A
shRNA against HSF2 in pSUPERIOR O
¨stling et al., 2007 N/A
shRNA scrambled in pSUPERIOR O
¨stling et al., 2007 N/A
N-Cadherin in pCCL-c-MNDU3c-PGK-EGFP Zhang et al., 2007; Addgene RRID:Addgene_38153
Software and Algorithms
FastQC version 0.20.1 Andrews, 2010; FastQC. http://www.bioinformatics.babraham.
ac.uk/projects/fastqc/
TopHat2 version 2.1.0 Kim et al., 2013 https://ccb.jhu.edu/software/tophat/
index.shtml
Subreads version 1.5.0 Liao et al., 2013
R: A language an Environment for Statistical Computing R Core Tean https://www.r-project.org/
Bioconductor Gentleman et al., 2004 http://www.bioconductor.org/
Bioconductor R package edgeR Robinson et al., 2010 https://bioconductor.org/packages/
release/bioc/html/edgeR.html
Bioconductor R package Limma Ritchie et al., 2015 https://bioconductor.org/packages/
release/bioc/html/limma.html
Bioconductor R package topGO Alexa and Rahnenfuhrer, 2019 https://bioconductor.org/packages/
release/bioc/html/topGO.html
Bioconductor R package GOstats Falcon and Gentleman, 2007 https://bioconductor.org/packages/
release/bioc/html/GOstats.html
ImageJ v1.51n Rueden et al., 2017 https://imagej.net/Citing
Fiji Schindelin et al., 2012 https://imagej.net/Citing
3D Object Counter Bolte and Cordelie
`res, 2006 https://imagej.net/Citing
FlowJo Version 10 https://www.flowjo.com/
DAVID Bioinformatic Tool Huang et al., 2009 https://david.ncifcrf.gov/home.jsp
GraphPad Prism Software Version 7 and 8 https://www.graphpad.com/
BioVenn Hulsen et al., 2008 http://www.biovenn.nl
Cell Reports 30, 583–597.e1–e6, January 14, 2020 e2
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Generation of HSF2 knock-out U2OS cells with CRISPR-Cas9
Guide RNAs (gRNA) targeting the exon 1 of HSF2 were designed using CRISPOR software (http://crispor.tefor.net/) and cloned into
pMLM3636 gRNA expression plasmid (a gift from Keith Joung, Addgene plasmid #43860). Human osteosarcoma U2OS cells were
transfected with Cas9 and gRNA expression plasmids using Amaxa electroporation as recommended by the manufacturer (Lonza).
The hCas9 was a gift from George Church (Addgene plasmid #41815; http://addgene.org/41815; RRID:Addgene_41815). One week
after transfections, cells were seeded at single cell density. Clones were genotyped by DNA sequencing of PCR products spanning
the targeted region of the HSF2 gene. The selected U2OS clones presented 3 different outframe mutations on HSF2 exon 1, each
corresponding to a different allele (Table S2). Guide RNA sequence targeting the 1
st
AUG of the HSF2 exon 1: 50-UGCGCCGC
GUUAACAAUGAA-30. Following primers were used for PCR for validation: forward (hHSF2_Cr_ATG_F): 50-AGTCGGCTCCTGG
GATTG-30and reverse (hHSF2_Cr_ATG_R): 50-AGTGAGGAGGCGGTTATTCAG-30. For the experiments, we utilized HSF2 knock-
out cell clone 1 (hereafter 2KO) and HSF2 knock-out cell clone 2 (hereafter 2KO#2).
Cell culture
U2OS cells and mouse embryonic fibroblasts (WT and Hsf2
/
MEFs, O
¨stling et al., 2007) were cultured in DMEM (Dulbecco’s Modi-
fied Eagle’s media, D6171, Sigma-Aldrich), supplemented with 10% fetal calf serum, 2 mM L-glutamine and 100 mg/ml penicillin-
streptomycin, and grown in 5% CO
2
at 37C. Culture media for MEFs were also supplemented with 1 X MEM non-essential amino
acid solution (M7145, Sigma-Aldrich).
Chicken chorioallantoic membrane (CAM) assay
The CAM-assay was performed as in Bjo
¨rk et al. (2016). Briefly, fertilized white Leghorn chicken eggs were incubated at 37C under
60% humidity (embryo development day 0, EDD0). Separation of the developing CAM was induced on EDD4. On EDD8, 1 310
6
U2OS WT and 2KO cells were mixed with Matrigel in 1:1 ratio and implanted on the CAM. On EDD11, the tumors were photographed
in ovo. Tumor area was measured in blind using ImageJ.
METHOD DETAILS
Treatments
Proteasome inhibition was induced with Bortezomib (BTZ, sc-217785, Santa Cruz Biotechnology) or MG132 (Z-Leu-Leu-Leu-H,
317-V, Peptide Institute Inc.). For HSP90 inhibition, 17-AAG (anti-agl-5, InvivoGen) and Geldanamycin (anti-gl-5, InvivoGen) were
used. All inhibitors were diluted in DMSO (dimethyl sulfoxide, D8418, Sigma-Aldrich) and applied to cells in final concentrations indi-
cated in the figures. Control cells were treated with DMSO only. To induce protein misfolding with amino acid analogs, cells were
starved for 17 h in L-arginine free culture medium (A14431-01, GIBCO) supplemented with 10% fetal calf serum, 2 mM L-glutamine
and 100 mg/ml penicillin-streptomycin. Following that, L-Canavanine sulfate salt (C9758, Sigma-Aldrich) was applied to the cells in
final concentrations indicated in the figure. Cells were treated for 3 or 6 h. After the treatments, cells were visualized with Leica phase
contrast microscope, an EVOS FL Cell Imaging System (Thermo Fisher Scientific), or an Axio Vert A1-FL LED microscope (Carl Zeiss)
and harvested for further analyses.
Transfections
For transfections, 6 310
6
U2OS WT or 2KO cells were suspended in 400 mL of Opti-MEM (11058-021, GIBCO) and subjected to
electroporation (230 V, 975 mF) in BTX electroporation cuvettes (45-0126, BTX). To downregulate HSF2 in WT cells, HSF2 targeting
shRNA and Scr vectors as previously described (O
¨stling et al., 2007), were used. For restoring the protein levels of HSF2 and N-Cad-
herin in 2KO cells, HSF2 in pcDNA3.1/myc-His(-)A vector and N-cadherin in pCCL-c-MNDU3c-PGK-EGFP (Zhang et al., 2007) (a gift
from Nora Heisterkamp; Addgene plasmid #38153; http://addgene.org/38153; RRID:Addgene_38153) were used. Empty pcDNA3.1/
myc-His(-)A vector was used as Mock. One day after transfection, cells were trypsinized, counted, re-plated, and let to recover for
24 h before BTZ treatments.
For cell aggregation assays, cells were transfected with GenJet (#SL100489-OS, SignaGen Laboratories) according to man-
ufacture
rs instructions. Briefly, cells were plated 18 to 24 h prior to transfections to ensure 80% confluency, and fresh culture media
with supplements was added to the cells before transfections. The N-Cadherin encoding vector (described above) was used for
transfections, and pEGFP-N1 (Clontech) was used as a Mock. The plasmids and the GenJet reagent were diluted in serum free me-
dia, and applied to the cells in a ratio of 1:2 (DNA:GenJet reagent). Cells were incubated with the DNA:GenJet mixture for 4 h, washed
with PBS, and supplemented with complete culture media. Cells were let to recover for 24 h before the cell aggregation experiments.
Immunoblotting
Cells were collected in culture media, washed with PBS (L0615, BioWest) and lysed in lysis buffer [50 mM HEPES, pH 7.4, 150 mM
NaCl, 1 mM EDTA, 2 mM MgCl
2
, 1% Triton X-100, 10% glycerol, 1 x complete Protease Inhibitor Cocktail (04693159001, Roche
Diagnostics), 50 mM NaF, 0.2 mM Na
3
VO
4
]. Protein concentration of the lysates was determined with Bradford assay. Equal amounts
e3 Cell Reports 30, 583–597.e1–e6, January 14, 2020
of cell lysates were resolved on 4%–20% or 7.5% Mini-PROTEANTGX precast gels (Bio-Rad) and the proteins were transferred to
a nitrocellulose membrane. For HSF2 detection, membranes were boiled for 15 min in MQ-H
2
O and blocked in 3% milk-PBS-
Tween20 solution for 1 h at RT. Primary antibodies were diluted in 0.5% BSA-PBS-0.02% NaN
3
and the membranes were incubated
in respective primary antibodies overnight at 4C. The following antibodies were used: anti-GAPDH (ab9485, Abcam), anti-HSC70
(ADI-SPA-815, Enzo Life Sciences), anti-HSF1 (ADI-SPA-901, Enzo Life Sciences), anti-HSF1 pS326 (ab76076, Abcam), anti-
HSF2 (HPA031455, Sigma-Aldrich), anti-HSP70 (ADI-SPA-810, Enzo Life Sciences), N-cadherin (04-1126, Millipore or ab76057,
Abcam), anti-PARP-1 (F-2, sc-8007, Santa Cruz Biotechnology), anti-Caspase-3 (ab2302, Abcam), anti-Pan-PCDHgA (75-178,
NeuroMab), anti-Pan-Cadherin (ab6529, Abcam), anti-Lamin A/C (4777S, Cell Signaling Technology), and anti-b-Tubulin (T8328,
Sigma-Aldrich). Secondary antibodies were HRP-conjugated and purchased from Promega, GE Healthcare or Abcam. All immuno-
blotting experiments were performed at least three times.
Immunofluorescence
2310
5
U2OS WT cells were plated on coverslips or MatTek plates (P35GC-.5-14-C, MatTek Corporation) 24 h before treatments.
Cells were fixed with 4% paraformaldehyde (PFA) for 15 min, permeabilized in 0.1% Triton X-100 in PBS and washed three times with
PBST (PBS-0.5% Tween20). Cells were blocked with 10% FBS in PBS for 1 h at RT and incubated overnight at 4C with a primary
anti-HSF2 antibody (HPA031455, Sigma-Aldrich), which was diluted 1:20 in 10% FBS-PBS. Secondary goat anti-rabbit Alexa Fluor
488 (R37116, Invitrogen) was diluted 1:500 in 10% FBS-PBS and the cells were incubated for 1 h in RT. Cells were washed three times
with PBST, incubated with 300 nM DAPI diluted in PBS or mounted in Mowiol-DABCO or VECTASHIELD mounting medium, and
imaged with a 3i CSU-W1 spinning disc confocal microscope (Intelligent Imaging Innovations).
Subcellular fractionation
2310
6
U2OS WT cells were plated and cultured overnight. The following day, cells were treated with 25 nM BTZ for 6 or 22 h. Control
cells were treated with DMSO for 22 h. Cells were collected in culture media and washed with PBS. 20% of the suspended cells were
collected for preparation of the whole cell lysate and lysed. The remaining 80% were collected for subcellular fractionation. Cyto-
plasmic and nuclear fractions were prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagents (78833, Thermo Fisher
Scientific) according to manufacturer’s instructions. Briefly, suspended cells were washed with cold PBS. The cell pellet was sus-
pended in 200 mL of cytoplasmic extraction reagent I. After incubation on ice, 11 mL of cytoplasmic extraction reagent II was added.
The suspension was incubated on ice and centrifuged (16 000 g, 5 min). The supernatant was collected and the pellet was resus-
pended in 100 mL of nuclear extraction reagent, incubated on ice and centrifuged (16 000 g, 10 min). The supernatant (nuclear extract)
was collected and the protein concentrations were determined by BCA assay (23225, Thermo Fischer Scientific).
Cell viability measurements
5310
3
U2OS WT and HSF2 KO cells were cultured in clear bottom 96-well plate (6005181, Perking Elmer) in complete culture media.
Cells were treated with indicated concentrations of Bortezomib or MG132 for 22 h. For calcium-depletion, cells were treated with or
without 25 nM BTZ (U2OS cells) or 5 and 10 nM BTZ (MEFs) for 20 h in serum free media. The extracellular calcium was depleted with
4 mM EGTA (U2OS cells) and 2 mM EGTA (MEFs) in calcium-free media and the Bortezomib treatment was continued for 2 h. Control
cells were treated with DMSO. After treatments cells were washed with PBS and incubated for 30 min at 37C with Calcein AM (4892-
010-K, R&D Systems) diluted 1:1000 in PBS. Fluorescence intensity was measured with Hidex Sense microplate reader (HIDEX Corp)
with excitation and emission wavelengths 485 nm and 535 nm, respectively. Alternatively, CellTiter-Glo reagent (G7570, Promega)
was added to the wells in 1:1 ratio and the luminescence was measured with Hidex Sense microplate reader. Respective blank values
were subtracted from the sample values and the viability of untreated control samples was set to value 1. All measurements were
repeated at least three times.
Cell aggregation assays
After trypsinization, 5 310
5
U2OS WT and 2KO cells were suspended in 2 mL of aggregation assay buffer (137 mM NaCl, 5.4 mM KCl,
0.63 mM, Na
2
HPO
4
, 5.5 mM glucose, and 10 mM HEPES, pH 7.4) supplemented with either 3 mM CaCl
2
or 3 mM EDTA. Cells were
rotated for 2.5 h in 150 rpm at 37C, after which the aggregates were imaged with the EVOS FL Cell Imaging System (Thermo Fisher
Scientific) or with an Axio Vert A1-FL LED microscope (Carl Zeiss). Cell aggregation assays were performed in biological triplicates.
The area of the three biggest aggregates in each sample was measured with ImageJ (U. S. National Institutes of Health, Bethesda,
Maryland, USA) for quantification purposes. All cell aggregation experiments were repeated at least three times.
RNA-sequencing
2310
6
U2OS WT and HSF2 KO cells were plated and cultured overnight. Following day, cells were treated with 25 nM BTZ for 6 or 10
h. Control cells were treated with DMSO. Cells were collected, and total RNA was purified with AllPrep DNA/RNA/miRNA Universal
Kit (80224, QIAGEN) according to manufacture
rs instructions. Genomic DNA from mRNA columns was digested with DNase I. The
RNA library was prepared according to Illumina TruSeqStranded mRNA Sample Preparation Guide (part #15031047). Briefly,
poly-A containing mRNA molecules were purified with poly-T oligo magnetic beads and fragmented with divalent cations under
elevated temperatures. For first-strand cDNA synthesis, RNA fragments were copied using reverse transcriptase and random
Cell Reports 30, 583–597.e1–e6, January 14, 2020 e4
primers. Unique Illumina TrueSeq indexing adapters were ligated to each sample. The quality and concentration of cDNA samples
were analyzed with Advanced Analytical Fragment Analyzer and Bioanalyzer 2100 (Agilent, Santa Clara, CA, USA) and QubitFluo-
rometric Quantitation (Life Technologies). Samples were sequenced with Illumina HiSeq 3000 (Illumina). All the experimental steps
after the RNA extraction were conducted in the Finnish Microarray and Sequencing Center, Turku, Finland. RNA-sequencing was
performed from four independent sample series.
Flow cytometry
0.5 310
6
U2OS WT and 2KO cells were fixed at 4C with BD Cytofix/Cytoperm (554722, BD Bioscience) and washed with cold BD
Perm/Wash (554723, BD Bioscience) solution. Cells were incubated over night at 4C with anti-cleaved PARP antibody conjugated to
BV421 (564129, BD Horizon), which was diluted 1:250 in BD Perm/Wash solution. Fluorescence was analyzed with a BD LSRFor-
tessa flow cytometer (BD Bioscience) using a standard Pacific Blue filter set (450/50 nm). The flow cytometry profiles were analyzed
using FlowJo 10 software.
Quantitative RT-PCR (qRT-PCR)
RNA was isolated using a RNeasy mini kit (74106, QIAGEN) according to the manufacturer’s instructions and quantified using a
NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific). Following that, 1 mg of total RNA was reverse transcribed with
an iScript cDNA Synthesis Kit (#1708891, Bio-Rad). SensiFAST Probe Lo-ROX and SensiFAST SYBRLo-ROX kits (Bioline) were
used for qRT-PCRs that were performed with QuantStudio 3 Real-Time PCR system (Applied Biosystems, Thermo Fisher Scientific).
All primers and probes were purchased from Sigma Aldrich. The following forward (f) and reverse (r) primers, and probes (pr) were
used: fRNA18S5, 50-GCAATTATTCCCCATGAACG-30; rRNA18S, 50- GGGACTTAATCAACGCAAGC-30; prRNA18S5, 50-FAM-
TTCCCAGTAAGTGCG GGTC-BHQ-30; fDSC2; 50-ATCCATTAGAGGACACACTCTGA-30; rDSC2, 50- GCCACCGATCCTCTT
CCTTC-30; fPCDHA6, 50-TGACTGTTGAATGATGGCGGA-30; rPCDHA6, 50-TCGGGTACGGAGTAGTGGAG-30; fPCDHA10, 50- AGG
CATCAGCCAGTTTCTCAA-30; rPCDHA10, 50-GAGAGCAGCAGACACTGGAC-30. The mRNA expression levels were normalized
against the respective 18S RNA (RNA18S5) expression in a given sample. All reactions were run in triplicate from samples derived
from four biological replicates.
3D cell culture and immunofluorescence
5310
3
U2OS WT and 2KO cells were cultured on Clear Round Bottom Ultra-Low Attachment (ULA) Microplates (#7007, Corning).
1310
6
cells were used for bacterial plates. After 24 and 48 h, cells were imaged with Axio Vert A1-FL LED microscope (Carl Zeiss).
For 3D in Matrigel, cells were embedded in growth factor reduced Matrigel (#356231, Corning) and cultured in Angiogenesis m-slides
(#81501, Ibidi) as described previously (Ha
¨rma
¨et al., 2010). Briefly, wells were filled with 10 ml of Matrigel:culture medium (1:1 ratio),
which was polymerized at 37C for 60 min. WT, 2KO, or 2KO#2 cells were seeded on top of the gel at a density of 700 cells per well, let
to attach at 37C for 2 h, and covered with 20 ml of Matrigel:culture medium (1:4 ratio). The upper layer of Matrigel:culture medium was
polymerized at 37C overnight, and appropriate humidity was ensured by adding droplets of MQ-H
2
O between the wells. Culture
medium was changed every second day, and cell growth was monitored by imaging the cultures with a Zeiss Axio Vert A1-FL
LED microscope (Carl Zeiss).
For immunofluorescence, spheroids were washed with 40 ml of PBS and fixed with 25 ml of 4% PFA for 20 min at RT, followed by
three washes with 40 ml of PBS. Spheroids were stained with 25 ml of 0.7% Triton X-100, 1:500 Alexa Fluor 488 Phalloidin (#A12379,
#A22287, Thermo Fisher Scientific), 300 nM DAPI in PBS at RT for 1 h. The stained spheroids were stored in PBS at 4C until imaging.
The spheroids were imaged as z stacks with a 3i CSU-W1 spinning disc confocal microscope (Intelligent Imaging Innovations) using
the same settings between the repeats. Spheroid volume was calculated based on the phalloidin staining using ImageJ v1.51n (Rue-
den et al., 2017) software with the 3D Object Counter v2.0 (Bolte and Cordelie
`res, 2006) plugin. The threshold for background and
object voxels were manually adjusted for each image in order to capture the whole volume of each spheroid.
Visualization of the data
Heatmaps were generated with GraphPad Prism 7 Software (GraphPad Prism Software, La Jolla California USA, https://www.
graphpad.com). Venn diagrams were generated with BioVenn web application (http://www.biovenn.nl/). DAVID Bioinformatic Re-
sources 6.7 (https://david-d.ncifcrf.gov/home.jsp) was used for functional annotation clustering.
QUANTIFICATION AND STATISTICAL ANALYSIS
Bioinformatic analysis of the RNA-seq data
The quality of the raw sequencing reads was confirmed with FastQC version 0.20.1 and aligned against the hg38 human genome
assembly using TopHat2 version 2.1.0. Subreads version 1.5.0 was used to calculate gene level expression counts according to
RefSeq-based gene annotations. The downstream analysis was carried out with R and Bioconductor. The data were normalized
with TMM normalization method on the edgeR package. In all sample groups, the Spearma
ns correlation value was above 0.97, indi-
cating high reproducibility. Statistical testing between the sample groups was carried out using Bioconductor R package Limma
(Ritchie et al., 2015) and the differentially expressed genes were filtered using fold change R3 and false discovery rate (FDR) of
e5 Cell Reports 30, 583–597.e1–e6, January 14, 2020
0.001 as cutoff. Enrichment analysis for the differentially expressed (DE) filtered genes was performed with topGO and GOstats pack-
ages. GO terms in each comparison pair were ranked according to their significance (lowest p value) and the most significantly
changed terms were selected for the figures. Additional information regarding the term IDs can be found from http://
geneontology.org.
Other data analyses
Statistical analyses were performed with GraphPad Prism 7 and 8 Software (GraphPad Prism Software, La Jolla California USA,
https://www.graphpad.com). The statistical significance was analyzed with two-way ANOVA and Holm-Sidak’s post hoc test unless
indicated differently. For details, see Figures 1,2,3,4,5, and 6.
DATA AND CODE AVAILABILITY
The original data are available at Gene Expression Omnibus (GEO) database under accession number GSE115973.
Cell Reports 30, 583–597.e1–e6, January 14, 2020 e6
