Rational Design of Crystallization-Induced-Emission Probes To Detect Amorphous Protein Aggregation in Live Cells

Article abstract of DOI:10.1002/anie.202103674

Unlike amyloid aggregates, amorphous protein aggregates with no defined structures have been challenging to target and detect in a complex cellular milieu. In this study, we rationally designed sensors of amorphous protein aggregation from aggregation-induced-emission probes (AIEgens). Utilizing dicyanoisophorone as a model AIEgen scaffold, we first sensitized the fluorescence of AIEgens to a nonpolar and viscous environment mimicking the interior of amorphous aggregated proteins. We identified a generally applicable moiety (dimethylaminophenylene) for selective binding and fluorescence enhancement. Regulation of the electron-withdrawing groups tuned the emission wavelength while retaining selective detection. Finally, we utilized the optimized probe to systematically image aggregated proteome upon proteostasis network regulation. Overall, we present a rational approach to develop amorphous protein aggregation sensors from AIEgens with controllable sensitivity, spectral coverage, and cellular performance.

Full text of DOI:10.1002/anie.202103674

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Rational Design of Crystallization Induced Emission Probes to
Detect Amorphous Protein Aggregation in Live Cells
Authors: Di Shen, Wenhan Jin, Yulong Bai, Yanan Huang, Haochen
Lyu, Lianggang Zeng, Mengdie Wang, Yuqi Tang, Wang
Wan, Xuepeng Dong, Zhenming Gao, Hai-Long Piao, Xiaojing
Liu, and Yu Liu
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202103674
Link to VoR: https://doi.org/10.1002/anie.202103674

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
Rational Design of Crystallization Induced Emission Probes to
Detect Amorphous Protein Aggregation in Live Cells
Di Shen,[1] Wenhan Jin,[1] Yulong Bai,[1] Yanan Huang,[1] Haochen Lyu,[1] Lianggang Zeng,[1] Mengdie
Wang,[1] Yuqi Tang,[1] Wang Wan,[1] Xuepeng Dong,[3] Zhenming Gao,[3] Hai-Long Piao,[1] Xiaojing Liu,[2]
and Yu Liu*[1]
[1]
D. Shen, W. Jin, Y. Bai, Y. Huang, H. Lyu, L. Zeng, M. Wang, Y. Tang, W. Wan, H. Piao, Y. Liu
CAS Key Laboratory of Separation Science for Analytical Chemistry,
Dalian Institute of Chemical Physics, Chinese Academy of Sciences,
457 Zhongshan Road, Dalian 116023, China.
E-mail: liuyu@dicp.ac.cn
[2]
[3]
X. Liu
Institute of Molecular Sciences and Engineering,
Shan Dong University,
Jimobinhai Road, Qingdao, 266237, China
X. Dong, Z. Gao
The Second Hospital of Dalian Medical University,
467 Zhongshan Road, Dalian, 116044, China
Supporting information for this article is given via a link at the end of the document.
Abstract: Unlike amyloid, amorphous protein aggregation of no
defined structures has been challenging to target and detect in
complex cellular milieu. Herein, we rationally designed amorphous
protein aggregation sensors from aggregation induced emission
probes (AIEgens). Utilizing dicyanoisophorone as a model AIEgen
scaffold, we first sensitized the fluorescence of AIEgens to non-polar
and viscous environment mimicking the interior of amorphous
aggregated proteins. We identified a generally applicable moiety
(dimethylaminophenylene) for selective binding and fluorescence
enhancement. Regulation of the electron withdrawing groups tunes
the emission wavelength while retaining selective detection. Finally,
we utilized the optimized probe to systematically image aggregated
proteome upon proteostasis network regulations. Overall, we present
a rational approach to develop amorphous protein aggregation
sensors from AIEgens with controllable sensitivity, spectral coverage,
and cellular performance.
distinguish these conformational species are of high demand for
mechanistic studies, diagnosis, and therapeutics.
The majority of current protein aggregation sensors focus
on amyloid proteins.[6] This is mainly because amyloid proteins
are of well-defined β-sheet stacked structure revealed by state-
of-the-art microscopic techniques. Such structural guidance
allows us to rationally design chemical probes for selective
detection and intervention of amyloid aggregation.
Introduction
Figure 1. Rational design of aggregation induced emission probes (AIEgen) to
detect amorphous protein aggregation in live cell. (a) Amorphous protein
aggregates are more difficult to target by AIEgen than the amyloid ones as they
possess no defined template structure for rational design. (b) Structural
modulations revealed structural features to control fluorescence sensitivity,
fluorescence color, and cellular performance.
Proteins need to properly adopt their three-dimensional
structures to acquire physiological functions.[1] Genetic mutations,
exotic proteome stresses, chemical modifications, and ageing
factors may lead to protein misfolding and aggregation.[2] Aberrant
protein aggregation beyond the cellular protein homeostasis
(proteostasis)[3] network capacity of refolding and clearance will
lead to numerous protein conformational diseases, including
neuromuscular degeneration, metabolic disorders, and
cardiovascular diseases.[4] Protein aggregation is a complex
multi-step liquid-to-solid phase separation process that consists
of different protein conformational states, including folded,
unfolded, misfolded soluble oligomers, and insoluble
aggregates.[5] Among these highly dynamic states, aggregated
proteins can be further categorized as amyloid and amorphous
aggregates according to their morphologies and biochemical
properties. Chemical tools that can precisely target and
Series of chemical probes have been developed to visualize
amyloid-β tangles and plagues associated with Alzheimer’s
disease.[6] Further efforts focus on distinguishing fine structural
differences among various amyloid proteins using fluorescent
probes and DNA aptamers, such as BODIPY based sensors
designed by Kim et al. that selectively target Tau amyloid tangles
over amyloid-β.[7-10] Organometallic probes pioneered by
Meade[11, 12] and Marti[13-16] not only broaden the chemical space
for rational design but also introduce multiple detection modes
other than fluorescence intensity. Kelly,[17] Nowick,[18] and
Petersson[19] et al. further resolved the oligomeric state of amyloid
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
precursors using FlAsH, triphenylmethane, and thioamide dye-
is to acquire fluorescence sensitivity towards polarity and
viscosity. To understand how to modulate AIEgen’s
environmental sensitivity, we utilized the dicyanoisophorone
(DCI) AIEgen scaffold as a model due to their compelling
structural features (Figure 2a). First, DCI scaffold harbors donor-
π-acceptor (D-π-A) structure that governs intramolecular charge
transfer (ICT) upon irradiation and the consequent polarity
sensitivity.[43] Second, the rotary bonds of DCI scaffold results in
twisted intramolecular charge transfer (TICT) at excited state that
controls its viscosity sensitivity.[44] Our initial task is to understand
how to control these two sensitivities in the DCI scaffold via
systematic structural analysis.
related fluorescence based detections.
Aggregation induced emission (AIEgen) probes developed
by Tang et al. opened a new avenue for chemists to design
fluorescent amyloid sensors with desired structural diversity,
chemical tunability, and biocompatibility (Figure 1a).[20-23] This
strategy unifies the intrinsic aggregation propensity of both
AIEgen and amyloid proteins for templated co-assembly. A
remarkable progress in the field made by Zhu et al. achieved
rational design of a blood-brain barrier permeable AIEgen to map
amyloid-β plaques in vivo.[24] Overall, the development of amyloid
sensors has advanced from case-by-case serendipity approach
to the current structure-guided rational design strategy with
precise control over sensitivity, selectivity, spectrum, etc.
However, intracellular amorphous protein aggregation with
random and indefinite structure is usually envisioned to be
“untargetable” or “undruggable” by small molecule probes, though
they carry critical roles in cellular physiology and pathology. The
lack of defined structures for amorphous protein aggregates
prevents them from being targeted via classic and novel design
strategies for chemical sensors.[25-33] Another challenge for
selective detection of amorphous aggregates is how to achieve
selectivity over other organelles in the complex cellular milieu as
they accumulate intracellularly rather than the extracellular space
for amyloid aggregates. Although AIEgens exhibit appealing
performance in detecting amyloid aggregated proteins, they failed
to detect cellular amorphous ones without the covalent linkage to
proteins exemplified by the seminal works from Hong and Zhang
et al.[34-38] Sensors of inherent selective binding towards
amorphous aggregated proteins, however, were seldom reported.
Thus, a blueprint that guides us to design amorphous aggregated
proteins sensors with controllable properties may be
complementary puzzle piece for the field.
a
In this work, we rationally designed a series of amorphous
protein aggregation sensors from the dicyanoisophorone (DCI)
AIEgen scaffold (Figure 1b).[39-42] A systematic study on the
structure-fluorescence relationship allowed us to dissect
structural moieties that regulate fluorescence sensitivity, binding
affinity, fluorescence color, and cellular performance. Through a
combination of photo-physical measurements, biophysical
analyses, and theoretical calculation, we identified three key
structural elements for an AIEgen to acquire satisfactory
properties: (1) Dimethylaminophenylene that up-regulates polarity
and viscosity sensitivity is the essential moiety to recognize
amorphous aggregates and turn on the fluorescence; (2) Electron
withdrawing moieties fine-tunes the emission wavelength; (3)
Balancing probe’s lipophilicity and solubility contributes to the
cellular signal-to-noise ratio. We further demonstrated these
features are generally applicable to other AIEgens. Finally, we
utilized the sensor from this work to image aggregated proteome
via proteostasis network regulations.
Figure 2. Regulation of the polarity- and viscosity-sensitivity of dicyanoiso-
phorone (DCI) probes by varying the symmetry and electron density of the
electron donating groups. (a) Fluorescence mechanism that underlies the
environmental sensitivity of DCI probes. (b) Series A probes contain asymmetric
aromatic moieties as the electron donating groups. (c) Series B probes contain
symmetric phenyl moieties as the electron donating groups. All probes were
listed according to their electron density from relatively electron-deficient moiety
to electron-excessive moiety.
Towards this goal, we synthesized two series of probes
based on the DCI scaffold with variations on the symmetry and
the electron density of the electron donating end (Figure 2b & 2c).
Series A harbors asymmetric aromatic moieties whereas series B
contains symmetric phenyl derivatives. The symmetry of these
substituents has been proposed to determine its bond rotation at
excited state and thus may control its sensitivity to viscosity.[45, 46]
Among each series, we further modulated the electron density in
electron donating groups resulting in relatively electron-excessive
(A5-A8, B4-B9) and electron-deficient (A1-A4, B1-B3) probes
(Figure 2b, 2c). Regulation of electron density may contribute to
the fluorescence sensitivity towards polarity as electronic
distribution can affect the intramolecular charge transfer (ICT).
Results and Discussion
Structural modulation of AIEgen’s viscosity sensitivity.
Protein misfolding and aggregation accompany with the
exposure of hydrophobic interiors (polarity change) and
misassembly into compact polypeptide chains (viscosity change).
In order to sense these micro-environmental changes, the first
challenge for AIEgens to detect amorphous protein aggregation
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
in different solvent mixtures. Statistics analysis was performed using GraphPad
Prime software and P value was calculated based on T-test.
We investigated the viscosity sensitivity by correlating the
viscosity dependence (χ values) and relative fluorescence
intensity with different types of substitutions. The χ value
describes the gradient of fluorescence intensity enhancement in
response to viscosity increase. We found that series A probes
bearing asymmetric EDGs exhibited slightly higher averaged χ
values compared to series B probes of symmetric EDGs (Figure
3a & 3c, S5, S6). In contrast, electron density of these EDGs
made no statistically significant contribution towards χ values
(Figure 3a & 3b). These observation echoes recent reports by
Kwon and Zhang et al. that explains how asymmetric aromatic
substitutions control the viscosity dependence via regulating its
Structural modulation of AIEgen’s polarity sensitivity.
We further studied how these structural modulations
influence the polarity sensitivity by measuring the
solvatochromism (shift in emission wavelength) and relative
fluorescence intensity in solvents of different polarity. We first
quantified the shift in fluorescence emission wavelength from
polar methanol to non-polar dioxane. Different from the viscosity
scenario, probes of electron-excessive EDGs were more
sensitive to solvent polarity than electron-deficient ones in terms
of fluorescence wavelength shift (Figure 4a & 4b). Meanwhile,
these electron-excessive groups improved the maximal
fluorescence intensity in non-polar dioxane that mimics the
hydrophobic interior of aggregated proteins (Figure 4d & 4e).[47, 48]
Interestingly, the symmetry of these substituents in the EDG
regulates neither the polarity dependent wavelength shift (Figure
4c) nor the intensity enhancement (Figure 4f). Together, these
excited state rotational energy barrier (TICT process).[38,
45]
Besides viscosity dependence, probes harboring electron-
excessive EDGs showed higher average fluorescence intensity in
viscous solvent than electron-deficient ones (Figure 3d & 3e),
whereas the symmetry of these substitutions played less
determinative effect in controlling their fluorescence intensity
(Figure 3d & 3f). Thus, symmetry of the substituted EDGs
regulates viscosity dependence (χ value) and their electron
density controls the maximal fluorescence intensity.
Figure 3. The symmetry and electron density of EDGs regulates viscosity
sensitivity. (a) χ value of series A and B probes describes the viscosity
dependence. (b) Electron density of the EDG plays insignificant role in
regulating the χ value. (c) Asymmetric substituents of the EDG result in slightly
higher average χ value than the symmetric ones. (d) Relative fluorescence
intensity in pure glycerol (red bars) and methanol (blue bars). (e) Electron-
excessive substituents of the EDG exhibited higher relative fluorescence
intensity in viscous glycerol. (f) The symmetry of the EDGs has minimal impact
on the fluorescence intensity in viscous glycerol. 20 μM probes were prepared
Figure 4. The electron density of EDGs regulates polarity sensitivity. (a)
Fluorescence emission wavelength shifts from polar methanol to non-polar 1,4-
dioxane solvent. (b) Electron density of the EDGs controls the polarity
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
dependent emission wavelength shifts. (c) The symmetry of the EDGs has
minimal impact on polarity dependent emission wavelength shifts. (d) Relative
fluorescence intensity of probes in H2O and 1,4-dioxane. (e) Electron-excessive
substituents of the EDG gives higher relative fluorescence intensity in non-polar
dioxane. (f) The symmetry of the EDGs has minimal impact on the fluorescence
intensity in non-polar dioxane. 20 μM probes were prepared in different solvent
mixtures. (g) & (h) B7 is a crystal induced emission fluorophore.
suggested probes that recognize amorphous aggregated proteins
(B5-B9) must be sensitive to viscosity and polarity but not vice
versa. These results also hinted that the symmetric para-
aminophenylene moiety in the DCI scaffold may contribute to the
selective binding towards aggregated proteins.
results demonstrated that the polarity sensitivity was purely
governed by the electron density of the EDGs but not their
symmetry. This observation complies with the fact that
fluorescence polarity sensitivity of a probe is influenced by the
dipole moment of its excited state structure. Such dipole moment
is mainly determined by the electron density distribution of the
probe but not its symmetry.
AIEgen’s crystal induced emission property.
Finally, we examined the impact of the structural
modulations on the AIE property of these probes. Interestingly,
we observed that aggregation caused quenching (ACQ) effect
dominated over the AIE effect when probes were fast precipitated
out of water (Figure S7, S8). These precipitates were
characterized as non-fluorescent amorphous aggregates of
probes under fluorescence microscopy (Figure 4h, S9, S10).[49]
Interestingly, the majority of these probes (A1-A5, A7-A8, B1-B9
in Figure 2a) recovered the fluorescence in crystalline solid state
regardless of their substituents (Figure 4g & 4h, S12, S13, S15,
S16). Some of these probes (A3, A4, B1-B4) do not fluoresce in
viscous or non-polar solvents but exclusively fluoresce in
crystalline solid, suggesting that these probes require well-aligned
compaction that inhibits all possible bond rotations to turn on the
fluorescence. However, this stringent requirement cannot be
satisfied by viscous solvent (glycerol), non-polar solvent (dioxane),
amorphous solid and protein aggregates that lack of template
compact structures. These observations may explain why
AIEgens were mostly reported as amyloid sensors and
sensitization of AIEgens towards non-polar and viscous
environment in this work is necessary for them to detect
amorphous protein aggregates of no defined structure.
Together, we revealed structure features in the DCI scaffold
that may contribute to the regulation of viscosity and polarity
sensitivity, and crystal induced emission property. Electron-
excessive EDGs are critical to maintain the enhanced
fluorescence intensity in viscous, non-polar, and compact
crystalline environment that mimics the interiors of misfolded and
aggregated proteins.
Dissecting the structural elements that recognize
amorphous protein aggregates.
In this work, we aimed to identify the structural moieties that
selectively recognize amorphous aggregates. We selected E. coli
dihydrofolate reductase (DHFR) as a model protein as it
aggregates into amorphous morphology as shown by the
transmission electron microscopy (TEM, Figure 5a inset). We
examined the fluorescence response of the studied probes upon
heat induced DHFR aggregation. Overall, B5 to B9 probes that
contain electron-excessive symmetric phenyl substitutions
exhibited higher fluorescence intensity and larger fold-of-change
in aggregated proteins than other probes (Figure 5a, S17).
Importantly, probes harboring asymmetric electron-excessive
moieties in A series (A5-A8) did not respond to amorphous protein
aggregation, though they are highly fluorescent in both viscous
and non-polar solvents (Figure 3d, 4d). These observations
Figure 5. Para-aminophenylene group is essential for recognizing protein
amorphous aggregates. (a) Structure-fluorescence relationship towards
sensing heat induced aggregated DHFR protein at 60 oC. (b) Fluorescence
spectra of B7 (25 μM) in folded and aggregated DHFR (50 μM, 60 oC, 5 min).
(c) Detection limits and linear range of B7 in detecting DHFR aggregation. (d)
All designed probes are not capable of detecting amyloid aggregates of α-
synuclein protein (140 μM, 1350 rpm for 72 h, 25 oC). (e) B7 is generally
sensitive to amorphous protein aggregation. (f) DHFR inhibitor trimethoprim
(TMP, 100 μM) stabilizes mut-DHFR (50 μM) quantified by the fluorescence of
B7. (g) & (h) B7 is generally selective to amorphous protein aggregation over
amyloid one. (i) & (j) B7 selectively partitions into insoluble aggregates. Error
bars: standard error (n=3).
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
Considering both the relative fluorescence intensity (3rd
highest) and the fold-of-change (28-fold, 3rd highest), B7 was the
optimal probe for further investigation (Table S2). Photo-physical
analysis of B7 probe in aggregated DHFR revealed a 28-fold
increase in fluorescence intensity (Figure 5b, exmax. 537 nm,
emmax. 665 nm, Φ = 0.17). B7 probe offered a wide linear range
from 0.45 μM to 25 μM to quantitively report on DHFR aggregation
(Figure 5c). Selectivity of B7 probe over amyloid protein
aggregates was also demonstrated using α-synuclein and
transthyretin as model proteins that forms amyloid fibrils upon
aggregation (Figure 5d, S18, S19). Thioflavin T (ThT) was utilized
to monitor α-synuclein amyloid aggregation over 72 h at ambient
temperature. Other than ThT, no probe in A and B series showed
obvious fluorescence increase as α-synuclein formed amyloid
protein aggregates. These results confirmed that B5-B9 exhibited
exclusive selectivity to amorphous protein aggregates. Finally, we
examined whether this selectivity was universal to other
amorphous aggregated proteins including mut-DHFR,[50] sortase,
superoxide
dismutase
(SOD1-V31A)[37]
,
and
human
immunoglobulin (Figure 5e) in a thermal shift assay. At elevated
temperatures, we observed increased fluorescence of B7
throughout all model proteins, from which we quantified the
melting temperatures (Tm) and revealed their thermodynamic
stabilities. In addition, we also demonstrated the application of B7
in quantifying the ligand engagement by measuring trimethoprim
(TMP) stabilization of mut-DHFR (Figure 5f). B7 holds the
selectivity of amorphous aggregates over amyloid aggregates
throughout different model proteins (Figure 5g, 5h, S18). Most
importantly, we showed B7 selectively partitioned into aggregated
proteins in presence of folded ones (Figure 5i, 5j), showing its
intrinsic binding affinity to aggregated proteins. Collectively, B7
showed exclusive selectivity to amorphous protein aggregates
over amyloid protein aggregates (Figure S19, S20, S23). These
lines of evidence also supported that the para-aminophenylene
group in the EDGs plays an important role in recognizing
amorphous protein aggregates.
We next investigated the origin of the fluorescence as
protein aggregation is a multi-step biochemical process involving
unfolded (U), misfolded oligomeric (M), and insoluble species (I)
(Figure 6a). First, urea mediated DHFR unfolding resulted in no
fluorescence (Figure 6b), indicating that the fluorescence is not
caused by unfolded proteins. This is different from the commercial
SYPRO orange dye (Figure S21) and other covalent AIEgens that
detect unfolded proteins.[34, 36] Second, misfolded oligomers were
captured by chemical crosslinking experiment (Figure 6c), and the
fluorescence intensity from misfolded oligomers was only half of
the one from insoluble aggregates (Figure 6c, S22), suggesting
insoluble aggregates is responsible for complete turn-on of the
fluorescence. Last, in a quantitative thermal shift assay, the
thermal shift curves from the fluorescence of B5-B9 probes
aligned well with the turbidity curve from OD330 readouts that
measured the insoluble protein aggregation, resulting in similar
melting temperature Tm (Figure 6d). In addition, biochemical
fractionation experiments supported that the fluorescence of B7
primarily retained in the insoluble DHFR aggregation (Figure 6e,
6f). Together, these results indicated that the fluorescence of the
probes initially turns on upon the formation of misfolded oligomers
but saturates at the insoluble aggregation state.
Figure 6. Origin of probes’ fluorescence. (a)&(b) Unfolded DHFR is not capable
of turning on the fluorescence of B7. Insoluble aggregation was induced by
heating DHFR (50 μM) at 60 oC for 5 min, and DHFR unfolding was triggered
by urea (6 M) at 4 oC for 24 h. (c) Misfolded oligomers partially turned on the
fluorescence of B7. (d) Thermal shift assay using either OD330 turbidity
(insoluble) or the fluorescence of the tested probes offers similar Tm values upon
DHFR aggregation. (e)&(f) Fractionation experiment revealed the fluorescence
retained in the insoluble DHFR. T: total; S: soluble; I: insoluble. Error bars:
standard error (n=3).
General applicability of key structural elements in
developing protein amorphous aggregation sensors.
The above results suggested that dimethylaminophenylene
moiety is essential to recognize amorphous protein aggregates.
We next demonstrated this key structural element is generally
applicable to the design of other sensors, in particular from
AIEgen scaffold. Towards this goal, we incorporated this
structural moiety onto fluorescent protein chromophore
analogues,[37, 51, 52] chromone, and pyridinium derivatives through
stilbene connection to generate AIEgen probes C1 to C5 (Figure
7a, Figure S14). Meanwhile, we also connected this structural
element with barbituric acid, thiazolidinedione, and fluorescent
protein chromophore via styrene to afford C6, C7 and C8. Both
connection modes are common in developing AIEgen sensors for
amyloid protein aggregation.[6] First, higher fluorescence intensity
in glycerol was observed in probes with stilbene connection than
probes with styrene connection (Figure 7b & 7c). Although both
connections possess rotary bonds, our results supported that
stilbene connection sensitizes the probe towards the TICT
process upon irradiation, resulting in greater viscosity sensitivity.
Second, probes with stilbene connection (C1-C5) exhibited
stronger response to amorphous aggregates than probes with
styrene connection (C6-C8, Figure 7d & 7e). Collectively, these
results
confirmed
that
stilbene-connected
dimethylaminophenylene is essential to recognize amorphous
aggregates, which may be useful to guide future design of probes
targeting protein amorphous aggregates (Figure S23, S24).
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
emission wavelength from 554 nm to 687 nm in aggregated WT-
DHFR with blue shifts (Figure 4d), indicating a less polar
environment in aggregated proteins than glycerol. Theoretical
density functional theory (DFT) calculation of the HOMO-LUMO
energy gaps revealed that D1 has the largest energy gap (3.28
eV) and D6 has the smallest energy gap (2.70 eV), supporting our
observation in fluorescence emission wavelength (Figure 8e,
S25). The longest emission wavelength of D6 may arise from the
extension of conjugation via the barbituric acid moiety. Finally, Tm
values measured from the fluorescence signals of D1 to D6
(ranging from 325.96 to 327.88 K) are comparable with the one
from OD330 turbidity readouts (326.64 K, Figure 8f). These lines
of evidence demonstrated that variation of electron withdrawing
groups could fine-tune the emission wavelength in detecting
protein aggregation but not their sensitivity towards amorphous
aggregated proteins.
Figure 7. Stilbene-connected dimethylaminophenylene is essential to recog-
nize amorphous aggregates. (a) The chemical structures of C1-C5 with stilbene
connection, C6-C8 with styrene connection, and B7. (b) Fluorescence of B7,
C1-C8 in glycerol. (c) Probes of stilbene connection are more fluorescent in
viscous glycerol than styrene connection. (d) Fluorescence of B7, C1-C8 in
folded and aggregated DHFR proteins. (e) Probes of stilbene connection are
more fluorescent in aggregated proteins than styrene connection. 20 μM probes
were prepared in glycerol or 25 μM probes were mixed with 50 μM DHFR
proteins for aggregation. Statistics analysis was performed using GraphPad
Prime software. Error bars: standard error (n=3).
Fine-tuning the electron withdrawing capacity to control
fluorescence emission wavelength.
We are also interested in understanding how to fine tune the
color of these probes while retaining its response towards
amorphous aggregated proteins. We next attempted to tune the
fluorescence emission wavelength of probe B7 (D5) in detecting
protein aggregation by structural modulation of the HOMO-LUMO
electronic energy gaps. As stated above, the para-
dimethylaminophenylene group plays an essential role in
fluorescence response towards aggregated proteins. We fixed
this essential structural moiety and varied the electron
withdrawing capacity to control the HOMO-LUMO energy gaps,
Figure 8. Tunable fluorescence emission wavelength of D1-D6 by structural
modulation of the electron-withdrawing groups. (a) The chemical structures of
D1-D6. (b) Structure-fluorescence relationship in heat induced aggregated WT-
DHFR proteins at 60 oC. (c) Normalized emission spectra of D1-D6 in glycerol.
(d) Normalized emission spectra of D1-D6 in aggregated WT-DHFR (60 oC). (e)
Theoretical DFT calculation of HOMO/LUMO energy gap of D1 and D6. (f)
Thermal shift assay measured the melting temperatures of DHFR using OD330
turbidity assay or D1-D6 fluorescence. Error bars: standard error (n=3).
rendering
D
series probes harboring different electron
withdrawing groups (EWGs, Figure 8a).
First, we showed all the probes retained satisfactory
fluorescence responses towards aggregated proteins with
different fluorescence intensity (Figure 8b). This result indicated
that the structures of EWGs on the probe posed less
determinative impact on the probes’ binding affinity with the
aggregated protein compared to the EDGs. Second, the emission
wavelength of D1 to D6 in viscous glycerol spanned from 606 nm
to 723 nm (Figure 8c). Similarly, we observed fluorescence
Balancing the solubility and lipophilicity to optimize the
cellular performance.
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
Moving forward into complex intracellular environment, we
found that the original B7 (E1) probe is of higher cellular
fluorescence background when detecting amorphous proteome
aggregation induced by inhibiting proteosome degradation
function using MG132 (Figure 9e). We attributed this high
fluorescence background to its poor solubility that may cause self-
assembly driven by the hydrophobic effect and the inherent nature
of AIEgens. Thus, we generated E2 and E3 probes of improved
solubility via installation of hydroxy groups (Figure 9a) to minimize
the inherent self-aggregation nature of these AIEgen probes. E2
and E3 were more resistant towards precipitation from water
measured by residual absorbance over 40 h incubation (Figure
9b). Though all three probes retained fluorescence enhancement
upon DHFR aggregation, we observed trade-off between
fluorescence intensity and fold-of-change as the solubility of the
probes increases (Figure 9c). Importantly, poor solubility of E1 led
to significant cytotoxicity at high concentrations and E2 of median
solubility possesses the best cellular compatibility (Figure 9d).
Both E2 and E3 achieved less cellular fluorescent background in
detecting amorphous proteome aggregation (Figure 9e). Given
the balanced fluorescence intensity, fold-of-change, cellular
compatibility, and background fluorescence, E2 was chosen for
further live-cell applications. Our results also highlight balancing
hydrophilicity and hydrophobicity is essential for optimal cellular
performance.
of this probe in imaging the cellular aggregated proteome upon
proteostasis regulations by small-molecule regulators. In both
HEK293T cells and HeLa cells, we in-situ stained aggregated
proteome with E2 together with proteostasis regulators. We
observed distinct morphology and subcellular locations of
aggregated proteomes formed under different conditions (Figure
10b), among which the role of spliceosome in compromising
proteostasis was first observed here and may deserve further
investigation. Fractionation experiment confirmed cellular E2
fluorescence originates from the insoluble proteome (Figure S26).
Interestingly, in unstressed cancerous cell line (HeLa), we always
observed high basal fluorescence than non-cancerous cell line
(HEK293T), indicating that cancerous cells consume proteostasis
network capacity and manipulation of proteostasis network can
be a promising anti-cancer treatment strategy (Figure S27).
Quantification of fluorescent puncta numbers and size per cell
hinted the extent of proteostasis collapse (Figure 10c, S28). In
addition, the number
Figure 9. Balancing probe lipophilicity and solubility to optimize cellular
performance in detecting proteome aggregation. (a) Hydroxyl group improves
the solubility of the probe. (b) Resistance to precipitation measured by residual
UV-Vis absorbance at 497 nm. 50 μM in H2O. (c) Relative fluorescence intensity
and fold-of-change upon DHFR aggregation. (d) MTT assay quantifies the
cytotoxicity of probes (0, 1, 10, 20, 40 and 60 μM). Error bars: standard error
(n=3). (e) Fluorescent background of probes in live cells (over-exposed) without
media washes. (f) Relative mean gray value of fluorescent cellular background
(n = 4, mean±SD; ***P <0.001, ****P <0.0001).
Imaging aggregated proteome upon proteostasis regulation.
Protein homeostasis (proteostasis) network consists of
protein chaperonins, chaperones and co-chaperones, folding
enzymes, proteases, and proteosome that assist protein folding,
refolding, and degradation for proper surveillance of cellular
fitness (Figure 10a).[3] Imbalanced proteostasis often leads to
aberrant proteome aggregation. Using the amorphous proteome
sensor E2 developed herein, we demonstrated facile applications
Figure 10. Cellular proteome aggregation profiles upon small molecule
regulation of proteostasis network. (a) Proteostasis network and its small
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
molecule regulators for different pathways. (b) Confocal images of cellular
aggregated proteome in HEK293T and HeLa cells upon proteostasis collapse
induced by the regulators. (c) Quantification of puncta number in stressed
HEK293T cells. Error bars: standard error (n=3). (d) Dose-dependent
fluorescent puncta number upon increasing cycloheximide concentration.
Puncta number was analyzed (n = 4, mean±SD; ****P <0.0001). Experiment
details see SI.
[6]
A. Aliyan, N. P. Cook, A. A. Marti, Chem. Rev. 2019, 119, 11819-
11856.
[7]
[8]
[9]
W. Tan, M. J. Donovan, J. Jiang, Chem. Rev. 2013, 113, 2842-2862.
L. Ge, Y. Tian, Anal. Chem. 2019, 91, 3294-3301.
P. Verwilst, H. R. Kim, J. Seo, N. W. Sohn, S. Y. Cha, Y. Kim, S.
Maeng, J. W. Shin, J. H. Kwak, C. Kang, J. S. Kim, J. Am. Chem.
Soc. 2017, 139, 13393-13403.
[10] P. Verwilst, H. S. Kim, S. Kim, C. Kang, J. S. Kim, Chem. Soc. Rev.
2018, 47, 2249-2265.
of fluorescent puncta imaged by E2 in stressed cells is dose-
dependent on the amount of protein synthesis inhibitor cyclo-
heximide (Figure 10d, S29), showing its potential as a quantitative
indicator of proteostasis network capacity. These phenotypic
observations urge us to develop suitable chemical tools that can
in-situ capture, distinguish, and regulate the contents inside
aggregated fraction of cellular proteome under different stresses
in the future.
[11] M. C. Heffern, P. T. Velasco, L. M. Matosziuk, J. L. Coomes, C.
Karras, M. A. Ratner, W. L. Klein, A. L. Eckermann, T. J. Meade,
Chembiochem 2014, 15, 1584-1589.
[12] A. Iscen, C. R. Brue, K. F. Roberts, J. Kim, G. C. Schatz, T. J. Meade,
J. Am. Chem. Soc. 2019, 141, 16685-16695.
[13] N. P. Cook, V. Torres, D. Jain, A. A. Marti, J. Am. Chem. Soc. 2011,
133, 11121-11123.
[14] N. P. Cook, M. Ozbil, C. Katsampes, R. Prabhakar, A. A. Marti, J.
Am. Chem. Soc. 2013, 135, 10810-10816.
[15] A. Aliyan, B. Kirby, C. Pennington, A. A. Marti, J. Am. Chem. Soc.
2016, 138, 8686-8689.
[16] B. Jiang, A. Aliyan, N. P. Cook, A. Augustine, G. Bhak, R.
Maldonado, A. D. Smith McWilliams, E. M. Flores, N. Mendez, M.
Shahnawaz, F. J. Godoy, J. Montenegro, I. Moreno-Gonzalez, A. A.
Marti, J. Am. Chem. Soc. 2019, 141, 15605-15610.
Conclusion
Unlike amyloid sensors, protein amorphous aggregation
sensors were seldom reported due to the lack of targeting
strategies. In this work, we have demonstrated how to derive
fluorescent sensors to selectively target protein amorphous
aggregates from crystal induced emission dicyanoisophorone
(DCI) scaffold. Via rational design and structure-fluorescence
study, we revealed structural features that regulate viscosity and
polarity sensitivity, binding affinity to aggregated proteins,
fluorescent color, and cellular performance. Our work not only
provides a facile tool to evaluate cellular proteostasis but also
complements the current lack of chemical strategies to target
intracellular protein aggregation. These key structural elements
revealed herein may serve as guiding groups to facilitate further
design of imaging sensors, proteomics probes, and therapeutic
reagents for amorphous aggregated proteins.
[17] J. Lee, E. K. Culyba, E. T. Powers, J. W. Kelly, Nat. Chem. Biol.
2011, 7, 602.
[18] P. J. Salveson, S. Haerianardakani, A. Thuy-Boun, S. Yoo, A. G.
Kreutzer, B. Demeler, J. S. Nowick, J. Am. Chem. Soc. 2018, 140,
11745-11754.
[19] J. M. Goldberg, S. Batjargal, B. S. Chen, E. J. Petersson, J. Am.
Chem. Soc. 2013, 135, 18651-18658.
[20] Y. Hong, J. W. Lam, B. Z. Tang, Chem. Commun. 2009, 4332-4353.
[21] Y. Hong, J. W. Y. Lam, B. Z. Tang, Chem. Soc. Rev. 2011.
[22] Y. Hong, L. Meng, S. Chen, C. W. Leung, L. T. Da, M. Faisal, D. A.
Silva, J. Liu, J. W. Lam, X. Huang, B. Z. Tang, J. Am. Chem. Soc.
2012, 134, 1680-1689.
[23] H. B. Cheng, Y. Li, B. Z. Tang, J. Yoon, Chem. Soc. Rev. 2020, 49,
21-31.
[24] W. Fu, C. Yan, Z. Guo, J. Zhang, H. Zhang, H. Tian, W. H. Zhu, J.
Am. Chem. Soc. 2019, 141, 3171-3177.
[25] E. Y. Zhou, H. J. Knox, C. J. Reinhardt, G. Partipilo, M. J. Nilges, J.
Chan, J. Am. Chem. Soc. 2018, 140, 11686-11697.
[26] L. P. Smaga, N. W. Pino, G. E. Ibarra, V. Krishnamurthy, J. Chan,
J. Am. Chem. Soc. 2020, 142, 680-684.
Probe synthesis, supporting figures, experimental methods in the
Supporting Information.
[27] A. K. Yadav, C. J. Reinhardt, A. S. Arango, H. C. Huff, L. Dong, M.
G. Malkowski, A. Das, E. Tajkhorshid, J. Chan, Angew. Chem. Int.
Ed. 2020, 59, 3307-3314.
Acknowledgements
[28] C. S. Wijesooriya, J. A. Peterson, P. Shrestha, E. J. Gehrmann, A.
H. Winter, E. A. Smith, Angew. Chem. Int. Ed. 2018, 57, 12685-
12689.
This work was supported, in part, by funds National Natural Sci-
ence Foundation of China (21907091), the LiaoNing
Revitalization Talents Program from the Liaoning province of
China (XLYC1907048), China Postdoctoral Science Foundation
[29] J. A. Peterson, C. Wijesooriya, E. J. Gehrmann, K. M. Mahoney, P.
P. Goswami, T. R. Albright, A. Syed, A. S. Dutton, E. A. Smith, A. H.
Winter, J. Am. Chem. Soc. 2018, 140, 7343-7346.
Grant
(2019M661138),
Dalian
Innovation
Fund
[30] J. Tang, M. A. Robichaux, K. L. Wu, J. Pei, N. T. Nguyen, Y. Zhou,
T. G. Wensel, H. Xiao, J. Am. Chem. Soc. 2019, 141, 14699-14706.
[31] J. V. Jun, E. J. Petersson, D. M. Chenoweth, J. Am. Chem. Soc.
2018, 140, 9486-9493.
(2020JJ26GX027), Pfizer ASPIRE award for transthyretin
amyloidosis basic research, The Taishan Scholars Project of
Shandong Province (no.tsqn201909017).
[32] J. V. Jun, D. M. Chenoweth, E. J. Petersson, Org. Biomol. Chem.
2020, 18, 5747-5763.
Keywords: aggregation induced emission, protein aggregation,
[33] G. Lukinavicius, K. Umezawa, N. Olivier, A. Honigmann, G. Yang,
T. Plass, V. Mueller, L. Reymond, I. R. Correa, Jr., Z. G. Luo, C.
Schultz, E. A. Lemke, P. Heppenstall, C. Eggeling, S. Manley, K.
Johnsson, Nat. Chem. 2013, 5, 132-139.
fluorescence, sensors, protein homeostasis
[1]
[2]
[3]
F. Chiti, C. M. Dobson, Annu. Rev. Biochem. 2006, 75, 333-366.
M. Stefani, Biochim. Biophys. Acta. 2004, 1739, 5-25.
W. E. Balch, R. I. Morimoto, A. Dillin, J. W. J. s. Kelly, Science 2008,
319, 916-919.
[34] M. Z. Chen, N. S. Moily, J. L. Bridgford, R. J. Wood, M. Radwan, T.
A. Smith, Z. Song, B. Z. Tang, L. Tilley, X. Xu, G. E. Reid, M. A.
Pouladi, Y. Hong, D. M. Hatters, Nat. Commun. 2017, 8, 474.
[35] S. Zhang, M. Liu, L. Y. F. Tan, Q. Hong, Z. L. Pow, T. C. Owyong,
S. Ding, W. W. H. Wong, Y. Hong, Chem. Asian. J. 2019, 14, 904-
909.
[4]
[5]
F. U. Hartl, Annu. Rev. Biochem. 2017, 86, 21-26.
F. Chiti, C. M. Dobson, Annu. Rev. Biochem. 2017, 86, 27-68.
8
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
[36] T. C. Owyong, P. Subedi, J. Deng, E. Hinde, J. J. Paxman, J. M.
White, W. Chen, B. Heras, W. W. H. Wong, Y. Hong, Angew. Chem.
Int. Ed. 2020, 59, 10129-10135.
[37] Y. Liu, C. H. Wolstenholme, G. C. Carter, H. Liu, H. Hu, L. S.
Grainger, K. Miao, M. Fares, C. A. Hoelzel, H. P. Yennawar, G. Ning,
M. Du, L. Bai, X. Li, X. Zhang, J. Am. Chem. Soc. 2018, 140, 7381-
7384.
[38] C. H. Wolstenholme, H. Hu, S. Ye, B. E. Funk, D. Jain, C. H. Hsiung,
G. Ning, Y. Liu, X. Li, X. Zhang, J. Am. Chem. Soc. 2020, 142,
17515-17523.
[39] Z. Guo, C. Yan, W. H. Zhu, Angew. Chem. Int. Ed. 2020, 59, 9812-
9825.
[40] H. Li, Q. Yao, W. Sun, K. Shao, Y. Lu, J. Chung, D. Kim, J. Fan, S.
Long, J. Du, Y. Li, J. Wang, J. Yoon, X. Peng, J. Am. Chem. Soc.
2020, 142, 6381-6389.
[41] Y. Zhao, K. Wei, F. Kong, X. Gao, K. Xu, B. Tang, Anal. Chem. 2019,
91, 1368-1374.
[42] W. Zhang, F. Huo, C. Yin, J. Mater. Chem. B. 2018, 6, 6919-6929.
[43] R. Misra, S. P. Bhattacharyya, Intramolecular Charge Transfer,
2018.
[44] S. Sasaki, G. P. C. Drummen, G.-i. Konishi, J. Mater. Chem. C. 2016,
4, 2731-2743.
[45] S. C. Lee, J. Heo, J. W. Ryu, C. L. Lee, S. Kim, J. S. Tae, B. O.
Rhee, S. W. Kim, O. P. Kwon, Chem. Commun. 2016, 52, 13695-
13698.
[46] S. C. Lee, C. L. Lee, J. Heo, C. U. Jeong, G. H. Lee, S. Kim, W.
Yoon, H. Yun, S. O. Park, S. K. Kwak, S. H. Park, O. P. Kwon, Chem.
Eur. J. 2018, 24, 2888-2897.
[47] Y. Liu, M. Fares, N. P. Dunham, Z. Gao, K. Miao, X. Jiang, S. S.
Bollinger, A. K. Boal, X. Zhang, Angew. Chem. Int. Ed. 2017, 56,
8672-8676.
[48] Y. Liu, K. Miao, Y. Li, M. Fares, S. Chen, X. Zhang, Biochemistry
2018, 57, 4663-4674.
[49] CCDC 2056339 (B7) contain the supplementary crystallographic
data for this paper. These data can be obtained free of charge from
The Cambridge Crystallographic Data Centre.
[50] M. Iwamoto, T. Bjorklund, C. Lundberg, D. Kirik, T. J. Wandless,
Chem. Biol. 2010, 17, 981-988.
[51] W. Wan, W. Jin, Y. Huang, Q. Xia, Y. Bai, H. Lyu, D. Liu, X. Dong,
W. Li, Y. Liu, Anal. Chem. 2021, 93, 1717-1724.
[52] W. Wan, Y. Huang, Q. Xia, Y. Bai, Y. Chen, W. Jin, M. Wang, D.
Shen, H. Lyu, Y. Tang, X. Dong, Z. Gao, Q. Zhao, L. Zhang, Y. Liu,
Angew. Chem. Int. Ed. 2021, 60, 11335-11343.
9
This article is protected by copyright. All rights reserved.

10.1002/anie.202103674
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
Aggregation illuminates aggregation: aggregation induced emission dyes can be rationally designed to detect protein amorphous
aggregation with controllable sensitivity, color, and cellular performance.
10
This article is protected by copyright. All rights reserved.