Selective Inhibition of the Hsp90α Isoform

Article abstract of DOI:10.1002/anie.202015422

The 90 kDa heat shock protein (Hsp90) is a molecular chaperone that processes nascent polypeptides into their biologically active conformations. Many of these proteins contribute to the progression of cancer, and consequently, inhibition of the Hsp90 protein folding machinery represents an innovative approach toward cancer chemotherapy. However, clinical trials with Hsp90 N-terminal inhibitors have encountered deleterious side effects and toxicities, which appear to result from the pan-inhibition of all four Hsp90 isoforms. Therefore, the development of isoform-selective Hsp90 inhibitors is sought to delineate the pathological role played by each isoform. Herein, we describe a structure-based approach that was used to design the first Hsp90α-selective inhibitors, which exhibit >50-fold selectivity versus other Hsp90 isoforms.

Full text of DOI:10.1002/anie.202015422

Angewandte
Communications
Chemie
How to cite: Angew. Chem. Int. Ed. 2021, 60, 10547–10551
International Edition: doi.org/10.1002/anie.202015422
German Edition: doi.org/10.1002/ange.202015422
Drug Discovery
Selective Inhibition of the Hsp90a Isoform
Sanket J. Mishra, Anuj Khandelwal, Monimoy Banerjee, Maurie Balch, Shuxia Peng,
Rachel E. Davis, Taylor Merfeld, Vitumbiko Munthali, Junpeng Deng, Robert L. Matts, and
Brian S. J. Blagg*
Abstract: The 90kDa heat shock protein (Hsp90) is a molec-
ular chaperone that processes nascent polypeptides into their
biologically active conformations. Many of these proteins
contribute to the progression of cancer, and consequently,
inhibition of the Hsp90 protein folding machinery represents
an innovative approach toward cancer chemotherapy. How-
ever, clinical trials with Hsp90 N-terminal inhibitors have
encountered deleterious side effects and toxicities, which
appear to result from the pan-inhibition of all four Hsp90
isoforms. Therefore, the development of isoform-selective
Hsp90 inhibitors is sought to delineate the pathological role
played by each isoform. Herein, we describe a structure-based
approach that was used to design the first Hsp90a-selective
inhibitors, which exhibit > 50-fold selectivity versus other
Hsp90 isoforms.
plasmic reticulum residing glucose regulated protein 94
(Grp94); and the mitochondrial isoform, tumor necrosis
factor receptor-associated protein 1 (Trap1).[1,2] Pan-inhibi-
tion of all four isoforms leads to non-specific degradation of
the entire Hsp90-dependent substrate population, and ulti-
mately leads to undesired activities.[4,9] The development of
Hsp90 isoform-selective compounds can provide a tool to
dissect the role played by each isoform towards disease
pathology and toxicities associated with pan Hsp90 inhibition.
Despite high structural similarity between the constitu-
tively expressed isoform, Hsp90b, and the inducible isoform,
Hsp90a, each of these isoforms fold select client proteins.[10]
Genetic knockdown of Hsp90a results in the degradation of
oncogenic client proteins, suggesting that the administration
of an Hsp90a-selective inhibitor could exhibit anti-cancer
activity against Hsp90a-dependent cancers.[11] In addition to
its presence in the cytosol, Hsp90a is also secreted extracell-
ularly to promote wound healing, cell adhesion, and inflam-
mation.[12] In fact, significant levels of Hsp90a are secreted
from highly invasive cancers to activate matrix metallopro-
tease 2 (MMP-2) to induce expression of MMP-3, which
drives tumor invasion.[13,14]
Extracellular Hsp90a also contributes to an inflammatory
microenvironment that supports prostate cancer progres-
sion.[14] Analysis of gene ontology traits revealed Hsp90b to
interact with a larger subset of proteins than Hsp90a.[15]
Notably, Hsp90a-dependent processes contribute to stress
adaptation or other specialized functions, while Hsp90b is
important for maintaining cell viability, clearly highlighting
the distinct evolution of each isoform.[15–17]
Initial efforts to develop isoform-selective inhibitors
focused on Grp94, due to its distinguishing structural features
that easily differentiate it from the other Hsp90 isoforms.[18–20]
However, the design of isoform-selective inhibitors against
the cytosolic isoforms, Hsp90a and Hsp90b, has been
extremely challenging as these isoforms share > 95% identity
within the N-terminal ATP-binding sites. In fact, only two
amino acids differ between these nucleotide-binding sites.
Despite these challenges, we recently disclosed KUNB 31 as
the first Hsp90b-selective inhibitor.[21] However, an isoform
selective inhibitor of Hsp90a has not yet been reported. Upon
analysis of the Hsp90 a and b N-terminal ATP-binding sites
bound to radicicol (Figure 1), it was determined that the
resorcinol moiety interacts uniquely with each Hsp90 isoform.
Compound 1 was chosen as lead molecule because it
represents an analog of AT13387, which is a resorcinol-
derived Hsp90 pan-inhibitor that underwent clinical evalua-
tion.[2] Computational studies with 1 bound to the N-terminal
ATP-binding site of Hsp90a (PDB: 2XAB) and Hsp90b
M
olecular chaperones are a class of proteins that regulate
proteostasis by assisting in the conformational maturation of
nascent polypeptides (referred to as client proteins) and the
renaturation of denatured proteins. The heat shock proteins
(Hsps) belong to a ubiquitously expressed molecular chaper-
one family that is highly conserved in eukaryotes. In fact, the
90 kDa Heat shock protein (Hsp90) is one of the most
abundant Hsps and is highly upregulated in stressed cells,
including cancer.[1,2] As a molecular chaperone, Hsp90 is
responsible for the conformational maturation of more than
300 client protein substrates, many of which are key
regulators of oncogenic transformation, growth, and meta-
stasis.[3–6] As a result, Hsp90 has been widely sought after as
a cancer target, which resulted in the development of 18
inhibitors that underwent clinical evaluation.[7,8] Unfortu-
nately, the clinical approval of these inhibitors was hindered
due to detrimental activities that are likely to result from the
inhibition of all four Hsp90 isoforms (pan-inhibitors). The
four Hsp90 isoforms include cytosolic isoforms Hsp90a
(inducible) and Hsp90b (constitutively expressed); the endo-
[*] Dr. S. J. Mishra, A. Khandelwal, Dr. M. Banerjee, R. E. Davis,
T. Merfeld, V. Munthali, Prof. Dr. B. S. J. Blagg
Department of Chemistry and Biochemistry
The University of Notre Dame
305 McCourtney Hall, Notre Dame, IN 46556 (USA)
E-mail: bblagg@nd.edu
Dr. M. Balch, Dr. S. Peng, Dr. J. Deng, Prof. Dr. R. L. Matts
Department of Biochemistry and Molecular Biology
246 Noble Research Center, Oklahoma State University
Stillwater, OK 74078 (USA)
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under https://doi.org/10.
1002/anie.202015422.
Angew. Chem. Int. Ed. 2021, 60, 10547 –10551
ꢀ 2021 Wiley-VCH GmbH
10547

Angewandte
Communications
Chemie
Figure 1. Structures of radicicol, AT13387, compound 1 and PU-H71.
(PDB:1UYM) revealed the presence of an extensive hydro-
gen-bond network that is coordinated by three conserved
water molecules (labeled as A–C, Figure 2). However, the
presence of Ser52 and Ile91 in Hsp90a in lieu of Ala52 and
Leu91 in Hsp90b results in formation of a smaller hydrophilic
pocket in Hsp90a and a large hydrophobic subpocket in
Hsp90b. The subtle difference between these two binding
sites were exploited to develop KUNB31, but unfortunately,
the presence of a smaller pocket in Hsp90a prevents such
investigation. Thus, Ser52/Ala52 represents the only amino
acid difference that can be utilized to develop Hsp90a-
selective inhibitors (Figure 2). Since compound 1 manifested
equipotent binding affinity towards both Hsp90a and Hsp90b
when evaluated in a fluorescence polarization (FP) assay (SI,
Figure 3S), it served as a starting point for the development of
Hsp90a-selective inhibitors.
Figure 3. Screening of phenols 2 and 3 and their proposed binding
modes in Hsp90a (PDB code: 2XAB) and IC50’s as determined with
a fluorescence polarization (FP) assay. Compounds were incubated
with Hsp90a/b and FITC-GDA in triplicate, and IC50 Æ SD was
measured.
conformation of Hsp90 (Figure 4). Replacement of the 5-
isopropyl substituent with a chlorine atom resulted in ꢀ 10-
fold decrease in affinity, while the inclusion of iodine
completely negated affinity. Introduction of bromine at the
5-position provided the highest selectivity (> 30-fold) and
a reasonable IC50 of ꢀ 2.73 mM. Intramolecular hydrogen-
bonding interactions between the fluorine and phenol hydro-
gen could explain the decreased affinity that was observed for
compound 7. Inclusion of a t-butyl group at the 5-position was
shown to manifest affinity similar to 5, and supports the
hypothesis that spheric bulk elicits selectivity for Hsp90a.
Compounds 10, 11 and 12 were synthesized to probe Site 1 of
Hsp90a (Figure 4, 4S), which is a flexible pocket that is lined
with hydrophobic residues, while simultaneously reducing the
distance between the 4-phenol and Ser52 to increase binding
affinity and selectivity towards Hsp90a.
The binding affinity of 10, 11 and 12 were determined and
compound 12 was found to bind Hsp90a with a IC50 of
ꢀ 4.8 mM and > 20-fold selectivity versus Hsp90b. As illus-
trated in Figure 4, the methylenedioxy ring induces the
opening of Site 1, (Figure 4A, 4S). In contrast to PU-H71,
which is a pan inhibitor that induces the opening of Site1,
Compound 1 binds to the closed form of Hsp90 (Figure 4B).
As can be seen in Figure 2, The resorcinol pharmacophore
was shown to interact with Asp93 via the 2-phenol, whereas
Ser52 hydrogen bonds with the 4-phenol via water molecules
A and B. Therefore, systematic removal of each phenol was
pursued to determine their contribution towards the binding
to each isoform. Upon preparation, the desphenol variants,
compound
2 (2-phenol) and compound 3 (4-phenol)
(Figure 3) were evaluated for isoform selectivity.[21,22] Com-
pound 3, which contains the 4-phenol, exhibited ꢀ 10-fold
greater binding affinity for Hsp90a versus Hsp90b, whereas
compound 2 bound Hsp90b with minimal preference ( ꢀ 2-
fold).[21] Excited by these results, the isopropyl moiety at the
5-position was replaced to modulate the steric and electronic
environment of the phenol, as well as to access the open
Figure 2. Resorcinol pharmacophore of compound 1 and its interac-
tions with N-terminal ATP binding site in A) Hsp90a and B) Hsp90b
C) Amino acid sequence alignment of N-terminal ATP-binding site
(details in SI).
Figure 4. A) Co-crystal structures of PU-H71 (PDB code: 2FWZ) and
B) 1 (PDB code: 2XAB) with Hsp90a.
10548 www.angewandte.org
ꢀ 2021 Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 10547 –10551

Angewandte
Communications
Chemie
Table 1: Replacement of the isopropyl group, IC50 determined using
Table 2: Installation of substituted thiophenols and evaluation of bind-
fluorescence polarization (FP) assay. Compounds were incubated with
Hsp90a/b and FITC-GDA in triplicate, and IC50 Æ SD was measured.
ing affinity IC50 determined using fluorescence polarization (FP) assay.
Compounds were incubated with Hsp90a/b and FITC-GDA in triplicates
and IC50 Æ SD was measured.
Compound
R group
IC50 Hsp90a (mM)(a) IC50 Hsp90b (mM)(a)
4
5
6
7
Cl
Br
I
4.69Æ0.27
2.73Æ0.19
>100
7.90Æ1.18
>100
>100
>100
>100
CF3
Compound IC50 Hsp90a (mM)(a) IC50Hsp90b (mM)(a) Fold Selectivity
8
9
2.62Æ0.08
>100
12a
12b
12c
12d
12e
12 f
12g
12h
0.71Æ0.03
0.52Æ0.01
0.53Æ0.07
0.84Æ0.13
0.72Æ0.08
3.03Æ0.26
0.9Æ0.02
>50
27.7Æ3.07
11.6Æ1.07
>50
>70
ꢀ53
ꢀ22
>63
ꢀ51
>16
>16
ꢀ48
0.65Æ0.04
12.92Æ1.92
10
>50
>50
37.5Æ3.8
>50
11
12
>25
>50
>50
0.46Æ0.05
22.28Æ2.8
4.87Æ0.29
>100
However, the 5-position thiophenol present in 12 is able to
access Site 1 in a conformation that overlaps with the
methylenedioxy ring of PU-H71 (Figure 4SC). Despite the
presence of an identical Site 1 in both Hsp90a and Hsp90b,
substitutions on the thiophenol were proposed to elicit
different conformations of Hsp90a, perturbing interactions
of the 4-Phenol with Ser52. Table 2 represents thiophenol
analogs that were studied as well as their respective binding
affinities (additional structure activity relationship (SAR), FP
probe binding affinities for Hsp90 isoforms and solubility data
can be found in supporting information). Compounds 12a and
12b contained methylenedioxy and ethylenedioxy rings,
respectively, which were proposed to interact with Tyr139
via an additional hydrogen bond (Figure 5a).
As predicted by our model, compounds 12a and 12b were
shown to exhibit improved binding affinity for Hsp90a.
Similarly, 12c, which contained the 3,4-dimethoxy group
bound Hsp90a with an IC50 value of ꢀ 530 nM and ꢀ 22-fold
selectivity versus Hsp90b. A methyl scan on the thiophenol
ring led us to conclude that substitutions at 3- and 4-positions
enhanced binding affinity. Consequently, substitutions at the
3- and 4-position of the thiophenol ring were sought, and
eventually resulted in the preparation of 12d, 12e, 12 f, and
12g. Notably, 12d and 12e were found to exhibit an IC50
ꢀ 840 nM and ꢀ 720 nM for Hsp90a, whereas, the trimethyl
substituted compound 12 f bound with reduced affinity. The 3-
methoxy-5-methyl substituted compound (12g) resulted in
ꢀ 900 nM affinity for Hsp90a. Homologation of the 3-methyl
substituent to the ethyl variant resulted in increased affinity.
However, inclusion of a propyl group at the 3-position
reduced affinity for Hsp90a (SI).
Figure 5. a) Co-crystal structure of 12b with Hsp90a (PDB code:
7LSZ) b) Co-crystal structure of 12d with Hsp90a (PDB code: 7LT0)
c) Overlay of the co-crystal structures of 12b and 1 bound to Hsp90a
(PDB: 2XAB) d) Overlay of the co-crystal structures of 12b and PU-
H71 bound to Hsp90a.
In order to study the binding mode of the Hsp90a-
selective inhibitors, compounds 12b and 12d were co-
crystalized with the N-terminal domain of Hsp90a. As
expected, the crystal structures supported our hypothesis
that the ethylenedioxy ring interacts with Tyr139 via a con-
served water molecule, which increases binding affinity. In
contrast, compound 12d does not afford secondary hydrogen
bond with Tyr139. However, the hydrophobic occupation of
Site 1 improves the binding affinity by reducing the entropic
penalty, which explains the increased affinity manifested by
12h. It should also be noted that the 4-phenol interacts with
Angew. Chem. Int. Ed. 2021, 60, 10547 –10551
ꢀ 2021 Wiley-VCH GmbH
www.angewandte.org 10549

Angewandte
Communications
Chemie
Ser52 via water molecules A and B (Figure 2 and Fig-
ure 5a,b). Therefore, the selectivity manifested by the benze-
nethiol containing compounds for Hsp90a was rationalized by
overlaying the binding pose of the non-selective compound 1,
and 12b bound to Hsp90a. As can be observed in Figure 5c,
inclusion of the benzenethiol at the 5-position and the 4-
phenol induce a conformational change that leads to the
opening of Site 1. In fact, the orientation of the thioether
fragment is different to that observed for the purine-class of
Hsp90 pan-inhibitors (Figure 5d). The co-crystal structures of
Hsp90a bound to 12b and 12d suggest the thioether to shift
toward the adenine binding pocket (Figure 5c) by bringing
the 4-phenol into close proximity to Ser52, a residue that
differentiates the Hsp90a and Hsp90b binding sites.[16,17] As
a result, Hsp90a preferentially accommodates the benzene-
thiol containing 4-phenol.
A NCI-60 cancer cell screen was performed with com-
pounds 5 and 12d to determine the cellular response of
Hsp90a-selective inhibitors. The growth of non-small cell
lung cancer cell line, NCI-H522, melanoma cell line UACC-
62, ovarian cancer cell line SK-OV-3, and renal cell carcinoma
cell line UO-31 were selectively inhibited. Hsp90 client
protein degradation was determined using NCI-H522 cells
treated with increasing concentrations of 12h. In particular,
the levels of Hsp90 client proteins Her2, Raf-1 and Akt
decreased in a dose-dependent manner.
The Hsp90b-dependent substrate, CDK4, degraded at
a higher concentration of 12h in comparison to c-Src, which is
a Hsp90a-dependent client, highlighting preferential inhib-
ition of Hsp90a in the cellular environment.[21,23] Hsp90a-
dependent client survivin also decreased in a dose-dependent
manner when treated with 12h. Levels of Hsp70 were induced
albeit to a lesser extent, which is in contrast to Hsp90 pan-
inhibitors such as GDA. Hsp90 levels were induced at lower
concentrations of 12h, similar to GDA, however, it decreased
with higher concentrations of 12h. Interestingly, levels of
HSF1 also decreased upon treatment with 12h but only at
high concentrations.
The design and development of isoform-selective inhib-
itors has remained a challenging task in medicinal chemistry,
especially, when there exists significant sequence and con-
formational identity. Hsp90a and Hsp90b represent a major
challenge for medicinal chemists, as only two amino acids vary
in the ligand binding sites, and they are shielded behind two
conserved water molecules. While an Hsp90b-selective inhib-
itor was previously shown to occupy the small subpocket in
Hsp90b that results from the two amino acid difference,
Hsp90a-selective inhibitors could not be developed via
a similar approach. Therefore, the roles played by each
phenol was interrogated and led to the identification of
monophenol 3 as a new lead compound that exhibited ꢀ 10-
fold selectivity for Hsp90a. Subsequent studies led to the
development of compound 5, which exhibited greater selec-
tivity for Hsp90a, but possessed modest binding affinity (IC50
ꢀ 2.7 mM.) Improvement in the binding affinity was sought by
installation of a benzenethiol moiety to occupy Site 1. Further
exploration of the benzenethiol side chain led to 12h, which
manifested an IC50 of ꢀ 460 nM for Hsp90a and ꢀ 48-fold
selectivity over Hsp90b. NCI Screening of the Hsp90a-
Figure 6. Western blot analysis of client proteins with 12h in NCI-
H522 cell line. DMSO and 500 nM Geldanamycin (GDA) were used as
negative and positive controls, respectively.
selective inhibitors revealed cancer cell lines that are
selectively inhibited by these compounds. Moreover, it was
determined that Hsp90a-dependent substrates are readily
degraded in the presence of Hsp90a-selective inhibitors, but
Hsp90b-dependent clients were not affected until higher
concentrations were reached. In totality, the first Hsp90a-
selective inhibitor was developed via a structure-based
approach to yield a drug-like small molecule that manifests
good affinity and reasonable selectivity in both recombinant
and cellular assays. The full utility and optimization of these
inhibitors continue and the results from such studies will be
reported in due course.
Acknowledgements
The authors gratefully acknowledge financial support of this
work via the National Institutes of Health to B.S.J.B.
(CA213566), R.L.M. and J.D. (CA219907), and the Okla-
homa Agricultural Experiment Station to R.L.M.
(OKL03159) and J.D. (OKL03060). The research is also
supported in part by an Institutional Development Award
(IDeA) from the National Institute of General Medical
Sciences of the National Institutes of Health (Award
10550 www.angewandte.org
ꢀ 2021 Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 10547 –10551

Angewandte
Communications
Chemie
[13] B. K. Eustace, T. Sakurai, J. K. Stewart, D. Yimlamai, C. Unger,
P20GM103640). We thank the staff of beam-line 14 at the
Stanford Synchrotron Radiation Lightsource for their sup-
port.
C. Zehetmeier, B. Lain, C. Torella, S. W. Henning, G. Beste, B. T.
Scroggins, L. Neckers, L. L. Ilag, D. G. Jay, Nat. Cell Biol. 2004, 6,
507 – 514.
[14] J. E. Bohonowych, M. W. Hance, K. D. Nolan, M. Defee, C. H.
Parsons, J. S. Isaacs, Prostate 2014, 74, 395 – 407.
[15] A. K. Voss, T. Thomas, P. Gruss, Development 2000, 127, 1 – 11.
[16] I. Grad, C. R. Cederroth, J. Walicki, C. Grey, S. Barluenga, N.
Winssinger, B. De Massy, S. Nef, D. Picard, PLoS One 2010, 5,
e15770.
Conflict of interest
The authors declare no conflict of interest.
[17] C. Kajiwara, S. Kondo, S. Uda, L. Dai, T. Ichiyanagi, T. Chiba, S.
Ishido, T. Koji, H. Udono, Biol. Open 2012, 1, 977 – 982.
[18] A. S. Duerfeldt, L. B. Peterson, J. C. Maynard, C. L. Ng, D.
Eletto, O. Ostrovsky, H. E. Shinogle, D. S. Moore, Y. Argon,
C. V. Nicchitta, B. S. J. Blagg, J. Am. Chem. Soc. 2012, 134, 9796 –
9804.
[19] P. D. Patel, P. Yan, P. M. Seidler, H. J. Patel, W. Sun, C. Yang,
N. S. Que, T. Taldone, P. Finotti, R. A. Stephani, D. T. Gewirth,
G. Chiosis, Nat. Chem. Biol. 2013, 9, 677 – 684.
[20] S. J. Mishra, S. Ghosh, A. R. Stothert, C. A. Dickey, B. S. Blagg,
ACS Chem. Biol. 2017, 12, 244 – 253.
[21] A. Khandelwal, C. N. Kent, M. Balch, S. Peng, S. J. Mishra, J.
Deng, V. W. Day, W. Liu, C. Subramanian, M. Cohen, J. M.
Holzbeierlein, R. Matts, B. S. J. Blagg, Nat. Commun. 2018, 9,
425.
Keywords: drug discovery · Hsp90 alpha · isoform selectivity ·
structure-based drug design
[1] M. Taipale, D. F. Jarosz, S. Lindquist, Nat. Rev. Mol. Cell Biol.
2010, 11, 515 – 528.
[2] H.-K. Park, N. G. Yoon, J.-E. Lee, S. Hu, S. Yoon, S. Y. Kim, J.-H.
Hong, D. Nam, Y. C. Chae, J. B. Park, B. H. Kang, Exp. Mol.
Med. 2020, 52, 79 – 91.
[3] M. V. Powers, P. Workman, FEBS Lett. 2007, 581, 3758 – 3769.
[4] L. Neckers, P. Workman, Clin. Cancer Res. 2012, 18, 64 – 76.
[5] V. C. da Silva, C. H. Ramos, J. Proteomics 2012, 75, 2790 – 2802.
[6] J. M. Holzbeierlein, A. Windsperger, G. Vielhauer, Curr. Oncol.
Rep. 2010, 12, 95 – 101.
[7] K. Jhaveri, T. Taldone, S. Modi, G. Chiosis, Biochim. Biophys.
Acta Mol. Cell Res. 2012, 1823.
[22] C. W. Murray, M. G. Carr, O. Callaghan, G. Chessari, M.
Congreve, S. Cowan, J. E. Coyle, R. Downham, E. Figueroa,
M. Frederickson, B. Graham, R. McMenamin, M. A. OꢀBrien, S.
Patel, T. R. Phillips, G. Williams, A. J. Woodhead, A. J. A.
Woolford, J. Med. Chem. 2010, 53, 5942 – 5955.
[8] A. Yuno, M. J. Lee, S. Lee, Y. Tomita, D. Rekhtman, B. Moore,
J. B. Trepel, Methods Mol. Biol. 2018, 1709, 423 – 441.
[9] D. S. Hong, U. Banerji, B. Tavana, G. C. George, J. Aaron, R.
Kurzrock, Cancer Treat. Rev. 2013, 39, 375 – 387.
[10] A. Hoter, M. E. El-Sabban, H. Y. Naim, Int. J. Mol. Sci. 2018, 19,
2560.
[23] W. Liu, G. A. Vielhauer, J. M. Holzbeierlein, H. Zhao, S. Ghosh,
D. Brown, E. Lee, B. S. Blagg, Mol. Pharmacol. 2015, 88, 121.
[11] X. Wang, X. Song, W. Zhuo, Y. Fu, H. Shi, Y. Liang, M. Tong, G.
Chang, Y. Luo, Proc. Natl. Acad. Sci. USA 2009, 106, 21288 –
21293.
[12] W. Li, D. Sahu, F. Tsen, Biochim. Biophys. Acta Mol. Cell Res.
2012, 1823, 730 – 741.
Manuscript received: November 20, 2020
Accepted manuscript online: February 23, 2021
Version of record online: March 26, 2021
Angew. Chem. Int. Ed. 2021, 60, 10547 –10551
ꢀ 2021 Wiley-VCH GmbH
www.angewandte.org 10551