Identification of Histone deacetylase (HDAC)-Associated Proteins with DNA-Programmed Affinity Labeling

Article abstract of DOI:10.1002/anie.202001205

Histone deacetylase (HDAC) is a major class of deacetylation enzymes. Many HDACs exist in large protein complexes in cells and their functions strongly depend on the complex composition. The identification of HDAC-associated proteins is highly important in understanding their molecular mechanisms. Although affinity probes have been developed to study HDACs, they were mostly targeting the direct binder HDAC, while other proteins in the complex remain underexplored. We report a DNA-based affinity labeling method capable of presenting different probe configurations without the need for preparing multiple probes. Using one binding probe, 9 probe configurations were created to profile HDAC complexes. Notably, this method identified indirect HDAC binders that may be inaccessible to traditional affinity probes, and it also revealed new biological implications for HDAC-associated proteins. This study provided a simple and broadly applicable method for characterizing protein-protein interactions.

Full text of DOI:10.1002/anie.202001205

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Identification of Histone-deacetylase (HDAC)-associated Proteins
with DNA-Programmed Affinity Labeling
Authors: Jianfu Zhang, Jianzhao Peng, Yiran Huang, Ling Meng,
Qingrong Li, Feng Xiong, and Xiaoyu Li
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.202001205
Link to VoR: https://doi.org/10.1002/anie.202001205

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
Identification of Histone-deacetylase (HDAC)-associated Proteins
with DNA-Programmed Affinity Labeling
Jianfu Zhang,[a] Jianzhao Peng,[a,b] Yiran Huang,[a] Ling Meng,[a] Qingrong Li,[a,b] Feng Xiong,[a] and
Xiaoyu Li *[a]
[a]
J. Zhang, J. Peng, Dr. Y. Huang, L. Meng, Q. Li, Dr. F. Xiong, Prof. X. Li
Department of Chemistry and the State Key Laboratory of Synthetic Chemistry, The University of Hong Kong
Laboratory for Synthetic Chemistry and Chemical Biology of Health@InnoHK
Pokfulam Road, Hong Kong SAR, China
xiaoyuli@hku.hk
[b]
J. Peng, Q. Li
Department of Chemistry
Southern University of Science and Technology China
1088 Xueyuan Road, Shenzhen, China
Supporting information for this article is given via a link at the end of the document.
Abstract: Histone deacetylase (HDAC) is
a
major class of
HDACs and the associated proteins.[5b, 6b, 7] Bantscheff, Schwarzer,
and their respective co-workers demonstrated that small
deacetylation enzymes. Many HDACs exist in large protein complexes
in cells and their functions strongly depend on the complex composition. molecules are powerful tools to profile HDAC complexes.[6b, 7g, 8]
The identification of HDAC-associated proteins is highly important in
understanding their molecular mechanisms. Although affinity probes
have been developed to study HDACs, they were mostly targeting the
direct binder HDAC, while other proteins in the complex remain
underexplored. We report a DNA-based affinity labeling method
capable of presenting different probe configurations without the need
for preparing multiple probes. Using one binding probe, 9 probe
configurations were created to profile HDAC complexes. Notably, this
method identified indirect HDAC binders that may be inaccessible to
traditional affinity probes, and it also revealed new biological
implications for HDAC-associated proteins. This study provided a
simple and broadly applicable method for characterizing protein-
protein interactions.
Recently, photo-affinity labelling (PAL) probes have been
developed to interrogate HDACs. PAL probes form covalent links
with the target, so that the captured proteins could be reliably
isolated with less interference from non-covalent interactions.[9] For
example, Cravatt and co-workers designed a series of SAHA-
BPyne probes to profile HDACs in cells.[9d, 9e] The Petukhov group
developed the “BEProFL” and “photomate” probes to study
HDACs.[9f-i] Recently, PAL probes were also used to study SIRT
proteins[10] and other HDAC complexes.[9a-c, 11] However, affinity
probes mostly capture the direct target, and the ability to capture
indirect binders is rather limited. Moreover, affinity probe has a
fixed structure; thus, the proteins it can capture are limited to the
ones accessible upon target binding. Indeed, as seen in many
studies, substantial synthesis efforts have been dedicated to
prepare multiple probes with different configurations (varying the
crosslinker, probe geometry, linker length and flexibility, etc.) to
9b-i, 12]
expand the target coverage.[7g,
For different analytical
Introduction
purposes, e.g. in-gel imaging or affinity purification, one would
need to prepare multiple probes with different tags, which is a
laborious and often highly challenging task.
Protein acetylation is an essential post-translational modification
and histone deacetylase (HDAC) is a major class of enzymes
regulating protein acetylation in cells.[1] Recent studies revealed
that HDAC-mediated acetylome is very vast and far beyond
histone modifications.[2] Aberrant HDAC activities are implicated in
numerous diseases.[3] HDACs have been intensively pursued as
drug targets and many HDAC inhibitors (HDACi’s) have been
developed to become clinical drugs.[4] Although the biological roles
of HDACs are well recognized, considerable challenges still exist
in studying their mechanisms. An important issue is that most
HDACs exist in large protein complexes in cells.[5] Being in a
complex modulates the deacetylation activity of HDACs, dictates
their recruitment to specific genomic loci, and affects the
interaction and crosstalk with other proteins.[6] Thus, efficient
approaches capable of characterizing HDAC-associated proteins
are highly desired. Previously, methods such as co-
immunoprecipitation (co-IP) and affinity pulldown were used to
study HDAC complexes. HDAC antibodies, peptides, HDACi’s,
and genetically tagged HDACs have all been used to isolate
Here we report a DNA-based affinity labelling method capable
of presenting different probe configurations without the need to
prepare multiple probes. We demonstrate the performance of this
method by profiling HDAC-associated proteins with HeLa cells.
Besides HDACs and known binders, many potential novel HDAC
interactors were identified. Notably, this method identified a
number of indirect HDAC binders that may not be accessible to
traditional affinity probes and it also revealed new biological
implications of HDAC-associated proteins.
Results and Discussion
Previously, we reported a DNA-based affinity labelling method
named DPAL (DNA-Programmed Affinity Labelling; Figure 1a).[13]
In DPAL, a ligand is conjugated with a DNA strand as the binding
binding probe (BP) and a complementary DNA strand is
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
pocket, were also prepared. First, the IC50 of SAHA and BP-1 was
determined to be 55 and 320 nM, respectively, suggesting that
DNA conjugation lowered the activity but the probe remained
active. BP-2 showed an IC50 of 4.1 µM, possibly due to the
branched probe shape.[9e] BP-3 was inactive, precluding the
possibility of inhibition by DNA. Next, BP-1 was incubated with
recombinant HDAC1 or HDAC3/NCoR2 complex and a series of
biotinylated CPs, respectively (Figure 2c).[13b, 14b] All CPs contained
a 15-nt region complementary to BP-1 but with different spacers (n
= +15 to -24). The mixtures were briefly irradiated (365 nm, 0 °C,
2 min.) and resolved with Western blot. As shown in Figure 2d-2e,
most BP/CPs were able to label HDAC1, HDAC3, and NCoR2.
Little labelling was observed with n > +12 or n < -12, while CPs
with smaller n’s (+6 to -6) had higher efficiency. Roughly the same
amount of HDAC3 and NCoR2 were labelled, suggesting the
crosslinker could access the protein beyond the SAHA-binding site.
DNA-based probes may partially renature during electrophoresis,
14b]
therefore resulting in a dual-band pattern.[13,
The negative
controls (SAHA competition, with BP-3, no UV) did not show
significant labelling (Figure 2f). Furthermore, HeLa cell nuclear
extracts were subjected to the same labelling procedure with BP-
1/CP (n = +3), the biotinylated proteins were isolated with
streptavidin beads and analysed with Western blot. As shown in
Figure 2g, HDAC1, HDAC3, and the HDAC1/2-associated
CoREST could be captured, and the capture was significantly
reduced with SAHA competition. These results have shown that
DNA-based probes were able to specifically label HDACs and
HDAC-associated proteins.
Next, we performed more comprehensive profiling of HDAC-
associated proteins with HeLa cells. First, BP-1 was paired with a
series of CPs (n = +15 to -12), respectively, to create 9 different
probe configurations (Figure 3a). The BP/CPs were incubated with
HeLa cell nuclear extracts and irradiated as same as in Figure 2.
The biotinylated proteins were affinity-purified, during which strong
washing conditions were applied to disrupt non-covalent
interactions.[9d] The captured proteins were trypsinized and the
resulting peptides were characterized with high-resolution liquid
chromatography mass spectrometry (LC-MS/MS). The samples
from SAHA competition were processed in parallel as control. For
statistical robustness, all experiments were repeated for 3-5 times.
The raw MS data was processed using the Perseus software to
generate a volcano plot for each BP/CP (Figure S3). In each plot,
the x-axis reflected the difference in protein abundance between
the experiment and the control groups based on label free
quantification (LFQ): the LFQ ratio (experiment:control) was
calculated and binary logarithmized (log 2), and labeled as “log2
(fold change)”. The y-axis was the minus common logarithm of the
p-value of each data point and was labeled as “-log10 (p-value)”
(see figure caption and the Supporting Information for more details
on MS data processing). A two-sided Student’s T-test was
performed with a false discovery rate (FDR) = 0.05 and S0 = 1 as
the statistical threshold, shown as curved lines in each plot.[17] The
proteins that met this threshold were considered as specific ones.
The numbers of specific proteins are summarized in Figure 3b, and
the complete protein list is provided in the Supporting Information.
The MS data showed the probes with large n‘s captured fewer
proteins. Strikingly, at n = +3, the probes captured many more
proteins, suggesting that a 3-base protruding spacer might be
advantageous for the probe to access more proteins. A small
number of overlapping proteins was observed (Figure 3b),
indicating the spacer had significant impact on target coverage.
Figure 1. a) Scheme of DNA-Programmed Affinity Labelling (DPAL). b) Multiple
CPs hybridize at different sites on BP for capturing the direct and indirect binders
in protein complex. c) CPs may have a “protruding” (n > 0) or “recessive” (n < 0)
configuration. L: a known ligand; star: photo-crosslinker.
conjugated with an affinity tag and a photo-crosslinker as the
capture probe (CP). Upon target binding, BP hybridizes with CP
and UV-irradiation triggers the capture of the target protein. DPAL
has been used to identify the target of small molecules, peptides,
nucleic acids, and aptamers.[13-14] With a similar concept, the
Gothelf group developed an elegant method for site-specific
antibody-DNA conjugation,[15] and the Tan group used aptamers to
guide DNA-templated protein conjugation.[16] However, DPAL has
not been exploited to study protein-protein interactions (PPIs). We
reason that, instead of one BP/CP pair, hybridizing multiple CPs at
different positions on the BP may capture not only the direct target,
but also the proteins at different locations in the complex (Figure
1b). Changing the BP/CP hybridization position could flexibly vary
the probe configuration without additional probe synthesis as only
a single BP is required. Although multiple CPs are necessary, they
are independent from the ligand and can be pre-prepared and used
right off the shelf. As shown in Figure 1c, CP may be either
“protruding” (n > 0) or “recessive” (n < 0); and CPs with large “n”
values may capture the proteins distal to the ligand-binding site.
We reason this approach may be suitable for characterizing
HDAC-associated proteins, considering the multitude of diversity
of HDAC complexes in cells.
First, we conjugated SAHA (suberoylanilide hydroxamic acid) to
DNA as the BP (BP-1; Figure 2a). SAHA is a pan-HDAC inhibitor
and can engage multiple HDACs;[8a] and the 4’-position of its
phenyl ring could be modified without abolishing HDAC binding. A
non-DNA affinity probe (BP-2) and a control probe (BP-3) without
the critical hydroxamate, which chelates the Zn2+ in the catalytic
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 2. a) Structures of SAHA, BP-1, BP-2, and BP-3. b) HDAC activity assay results. c) BP-1 and multiple CPs were used to label HDAC and HDAC complexes.
Labeled proteins were analyzed with Western blot. Conditions: DNA, 1.0 µM; protein, 0.3 µM; 1x PBS (phosphate-buffered saline), pH 7.4, 0.1 M NaCl; UV: 365 nm,
0 °C, 2 min. L: SAHA. d)-e) Western blots of the labeled proteins; n values are as marked. f) Labeling specificity test. g) Labeling was conducted with BP-1/CP (n =
+3) in HeLa nuclear extracts (2 µg/µL); biotinylated proteins were isolated with streptavidin beads and analyzed with Western blot. Lane 1: BP-1/CP; lane 2: with 20-
fold excess SAHA. IB: immunoblotting. See the SI for details.
HDAC1, 2, 3, and 6 were detected (Table S1), which was
consistent with the previous reports also using SAHA-based
probes.[6b, 9a, 9d] HDAC3 exhibited very weak MS signal but the
labelling could be verified with Western blot,[9a, 9d, 11] and we also
observed similar phenomenon (Figure S4); the weak signal from
HDAC6 may be due to its low abundance in nuclear extracts. Also
similar to previous reports, Class-IIa HDACs were not identified,
possibly because of its low affinity to hydroxamates.,[6b, 9a, 9c, 9d, 11]
In addition, several known components of major corepressor
HDAC complexes were identified (Table S1).
In total, 9 BP/CPs identified 267 unique proteins, which
included HDACs, HDAC-associated proteins, and SAHA-binding
proteins, such as the ones sharing the similar ligand space as
HDAC or completely unrelated ones. To narrow down the list, the
proteins were subjected to several filters (Figure 3d). First, we
compared the list with the interactome of human HDAC1-11.[5b]
Surprisingly, only 11 overlapping proteins was found, probably
because of the different cell types and the probe structures. Next,
the remaining 256 proteins was compared with the HDAC binders
identified with small-molecule-based probes.[6b, 9a, 9d, 9e] Again, a
small overlap of 16 proteins was observed, suggesting the DNA-
based probes captured different sets of proteins from the small
molecule probes. Furthermore, we conducted comprehensive
literature survey and found that 128 proteins were either reported
to interact with HDAC or annotated as deacetylase substrates
based on acetylome analysis.[18] Finally, 112 proteins remained,
which were considered as potential novel HDAC interactors,
although it may still contain some SAHA binders.[9a] We tested the
non-DNA probe BP-2;[9d] very few proteins were identified and only
one passed the threshold (Figure S5), possibly due to the
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 3. a) Protein profiling with 9 BP/CP configurations in HeLa cell lysates.
Conditions were the same as Figure 2. b) Number of the specific proteins
identified. c) Venn diagram showing the overlapping proteins for n = +9, +6, +3,
and -3; others were not compared due to low protein number. d) Passing the
proteins through three filters. e) Keywords analysis with DAVID functional
annotation. Volcano plots are shown in Figure S3. Full protein list and
experimental details are provided in the SI.
Figure 4. a) Structures of SAHA and TSA. b-d) Volcano plots (n = +12, +9, +6)
from the SAHA and TSA experiments were compared. The curves denote the
statistical threshold (two-sided Student’s T-test; FDR < 5%, S0 = 1). x-axis: log2
(fold change) = log2 (LFQ mean of the experiment group/LFQ mean of the control
group); LFQ: label-free quantitation; y-axis: -log10 (p-value) of each data point.
Red: proteins that passed the threshold; purple: overlapping proteins. e) Venn
diagrams summarizing the identified proteins. See the Supporting Information for
MS data processing and plotting details.
low binding affinity of BP-2. Finally, we performed keyword
analysis using DAVID functional annotation (Figure 3e).[19] The top
category is acetylation, covering 76.8% of the proteins with the
most significant enrichment (p-value = 1.40 × 10-105), which
corroborated the profiling experiments where deacetylation
complexes were targeted.
With this approach, the probes may possibly reach distal
locations in the complex to capture indirect HDAC binders. Thus,
we focused on the proteins identified with longer spaces (n = +12,
+9, and +6). Since the proteins identified in Figure 3 may contain
not only HDAC binders but also the ones that bind to SAHA,[9a] we
performed an additional series of profiling experiments using
trichostatin A (TSA; Figure 4a) as the competitor. TSA is also a
pan-HDAC inhibitor;[20] thus, the proteins identified with both SAHA
and TSA competitions are expected to be HDAC binders. The
experiments using TSA were conducted the same as in Figure 3,
and the volcano plots were shown side-by-side with the ones from
the SAHA experiments (Figure 4b-d). The proteins that passed the
threshold are highlighted in red and the overlapping ones are in
purple.
passing through the filters, two potential novel binders were
identified: MEK1 and COF1. MEK1 is a mitogen-activated protein
kinase (MAPK) kinase. Interestingly, although many studies
described the synergistic effects of MEK1 and HDAC inhibitors in
anti-cancer therapy,[21] MEK1-HDAC binding was not reported. We
performed co-IP in cell lysates using anti-HDAC1, 2, 3, 6
antibodies, respectively, which showed that MEK1 bound to
HDAC1, but not the other HDACs; the reversed co-IP using anti-
MEK1 antibody further confirmed the interaction (Figure 5a and
S6a-d). Next, co-IP with recombinant HDAC1 and MEK1 confirmed
that MEK1 directly binds to HDAC1. COF1 (cofilin-1) is an actin
regulator in cells.[22] Co-IP screening with HDAC1/2/3/6 antibodies
and reciprocal co-IP experiments showed that COF1 bound to
HDAC1 in cell lysate but not between recombinant proteins (Figure
5b and S6e-i), suggesting that COF1 may bind HDAC1 indirectly
through other proteins. At n = +6, four proteins were identified
(Figure 4d), in which ANXA2 (annexin A2) was a potential novel
binder.
At n = +12, cyclophilin A (CYPA), a known HDAC1/2 binder,
was the only commonly identified protein (Figure 4b). At n = +9,
SAHA and TSA identified many more proteins (Figure 4c). After
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 6. a) Reciprocal co-IP to validate HDAC1-PARP1 binding. b) Schematic
representation of Dox-triggered PARP1 release. BP-1/CP was used to capture
PARP1 in cells untreated or treated with Dox (0.5 µM, 16 hours). c) MS reporter
ion intensities of PARP1 in b) were compared; exp.: without SAHA; control: with
20-fold excess SAHA. y-axis: MS intensity ratios = untreated cells/Dox-treated
cells. Error bars represented three replicates. d) Top: BP-1/CP were used to
affinity-purify PARP1 from HeLa cells; bottom: PARP1 expression level. Full gel
images are shown in Figure S7.
Collectively, these results strongly suggested that COF1,
ANXA2, and CALR bound to HDAC1 indirectly and demonstrated
the method’s capability in identifying indirect binders in HDAC
complexes.
Figure 5. Co-IP experiments to validate the association of HDACs with selected
proteins: a) HDAC/MEK1. b) HDAC/COF-1. c) HDAC1/ANXA2 with and without
PCNA. d) HDAC1/CALR. M: pure protein or lysate, used as marker. IgG:
Immunoglobulin G. Full gel images are shown in Figure S6.
Furthermore, we employed this method to explore biological
implications of HDAC-associated proteins. PARP1 (Poly[ADP-
ribose] polymerase 1) is a nuclear protein regulating gene
transcription. Although PARP1 is a known HDAC binder,[24] it had
not been isolated by affinity probes. Here, PARP1 was captured
by CP at n = 0; although below the statistical threshold, PARP1
showed low p-value (Figure S3) and high score/peptide coverage
in MS data (Table S2). Reciprocal co-IP also verified the
endogenous HDAC1-PARP1 interaction (Figure 6a). Under
cellular stress, HDAC1-PARP1 binding would be disrupted,
resulting in PARP1 release from HDAC1 complex (Figure 6b).[25]
We investigated whether DPAL probes could detect such a
process. HeLa cells were treated with the cytotoxic doxorubicin
(Dox) to trigger PARP1 release,[25] then the nuclear extracts were
labelled with BP-1/CP. The labelled proteins were affinity-purified
and analysed with MS. By comparing the MS reporter ion
intensities, significantly reduced PARP1 capture was observed in
Dox-treated cells, while the control experiments with SAHA
competition showed little difference (Figure 6c). Western blot also
showed less PARP1 capture in Dox-treated cells, and Dox did not
alter PARP1 expression in HeLa cells (Figure 6d). Although
Interestingly, PCNA (proliferating cell nuclear antigen), a known
direct HDAC1 binder, was also identified (albeit slightly below the
threshold); ANXA2 has been found to bind PCNA,[23] but its
interaction with HDAC1 was not reported. Thus, we hypothesized
that ANXA2 might bind HDAC1 indirectly through PCNA. As
shown in Figure 5c and S6j-l, reciprocal co-IP using anti-HDAC1
and anti-ANXA2 antibodies confirmed the HDAC1-ANXA2
binding, while removing PCNA completely abolished the
interaction. Moreover, co-IP with anti-HDAC2, 3, 6 antibodies did
not show any interaction (Figure S6m). These results indicated
that ANXA2 was an indirect HDAC1 binder and PCNA was
required for binding. Lastly, we tested an additional protein CALR
(calreticulin), which was only slightly below the threshold (Figure
4d). CALR is a chaperone protein participating in transcriptional
regulation but not known to bind HDAC. Similarly, screening with
anti-HDAC1/2/3/6 antibodies and reciprocal co-IP experiments
showed that CALR was an indirect HDAC1 binder (Figure 5d and
S6n-q).
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
PARP1 may also interact with other HDACs, the dissociation from
HDAC1 is a major reason for the decreased PARP1 capture.
Finally, we investigated whether the method could reveal novel
cellular PPIs. PKM2 (pyruvate kinase isozymes M2) was
identified at n = -3 (Figure 7a and S8). Previously, PKM2 was
found to be associated with HDAC3 in glioblastoma and colon
cancer cells;[26] however, the HDAC3-PKM2 interaction was
highly dependent on cell stimulation. Upon EGF (epidermal
growth factor) treatment, PKM2 formed a complex with HDAC3 to
regulate histone H3 acetylation; without EGF, the HDAC3-PKM2
association was completely abolished (Figure 7b).[26a, 26b] Since
the cells used here were not EGF-stimulated, we hypothesized
that PKM2 might bind to other HDACs or bind to HDAC3 without
the need for EGF-stimulation in HeLa cells. To investigate this,
we first verified the status of EGFR signalling pathway. Serum-
starved HeLa cells, either non-stimulated or EGF-stimulated,[27]
were lysed, and then three protein phosphorylation markers in the
EGFR pathway: EGFR/Y1068, Akt/S473, and ERK/T202/Y204,[28]
were analysed with the respective assays (Abcam; EGFR:
ab126438; Akt: ab253299; ERK: ab176660), which measured
and compared the quantities of total protein and the
phosphorylated protein. As shown in Figure S10, non-stimulated
HeLa cells exhibited very low phosphorylation level for all three
phosphorylation markers, while the total protein expression
remained largely unaffected upon stimulation. These results
indicated the EGF/EGFR pathway was suppressed in non-
stimulated HeLa cells. Next, we performed co-IP of HDACs and
PKM2 using anti-HDAC-1, 2, 3, 6 antibodies with the non-
stimulated cells. Results showed that PKM2 could bind to
HDAC1/3, but not HDAC2/6 (Figure 7c-f). To the best of our
knowledge, it is the first time PKM2 is identified as an HDAC1-
associated protein. Moreover, previous studies showed that
HDAC3-PKM2 association deactivated HDAC3 and increased
H3K9 acetylation in EGF-stimulated glioblastoma cells, while the
effect was reversed in colon cancer cells (Figure 7b),[26]
suggesting the function of HDAC3-PKM2 complex is highly
context-dependent. We knocked-down PKM2 expression with
small interfering RNA in non-stimulated HeLa cells and observed
little change in the level of HDAC1, HDAC3, histone H3, and
acetylated H3K9 (Figure 7g). Interestingly, the acetylation at
H3K18, another major HDAC1/3 deacetylation site,[29] exhibited
markedly decrease, suggesting PKM2 association might
deactivated HDAC1/3, and the corresponding complexes might
regulate histone H3 at different positions (e.g. H3K18) in HeLa
cells. Also, studies showed EGF-stimulation triggered PKM2
translocation into nucleus.[26] In HeLa cells, little change of PKM2
in both cytoplasm and nucleus was observed upon EGF
stimulation; the nucleus of non-stimulated cells appeared to
already contain substantial amount of PKM2 (Figure S11).
Although further studies are certainly needed, these results
suggested that PKM2 in HeLa cells may function through a
different mechanism. Collectively, using PARP1 and PKM2 as
representatives, we have demonstrated that this method could
reveal new biological implications for HDAC-associated proteins.
Figure 7. a) Volcano plot of the identified proteins at n = -3. The curves denote
the statistical threshold (two-sided Student’s T-test; FDR < 5%, S0 = 1; x-axis:
log2 (fold change) = log2 (LFQ mean of the experiment group/LFQ mean of the
control group); LFQ: label-free quantitation; y-axis: -log10 (p-value) of each data
point. b) EGF stimulation is required for PKM2-HDAC3 binding, leading to either
HDAC3 deactivation or activation HDAC3 in different cells.[26] c-f) Co-IP to
examine PKM2-HDAC interactions in non-stimulated HeLa cells. M: cell lysate.
g) Western blots to analyze the effects of PKM2 knockdown in HeLa cells;
control: with scrambled siRNA; lysate: HeLa cell lysate. h) Results indicate that
PKM2-HDAC1/3 interaction exists in HeLa cells without stimulation, which may
deactivate HDAC1/3 and lead to increased H3K18 acetylation. Full gel images
are provided in Figure S9. See the Supporting Information for details on MS
data processing and plotting.
capture indirect binders in protein complexes remained
underexplored. Here, using only one binding probe, 9 different
probe configurations were created to profile HDAC complexes,
and a large number of HDAC interactors were identified; notably,
three of the four proteins tested (Figure 5) appeared to be indirect
binders. This method is also responsive to cellular context change
and could reveal new biological implications for HDAC-associated
proteins. In addition, the probe system is modular and the guiding
ligand could be easily changed to interrogate other protein
complexes beyond HDACs.
This method has some limitations that may be further pursued.
First, since DNA is membrane impermeable, the probes cannot
be used in live cells, precluding the characterization of the PPIs
that are disrupted during cell lysis. Encouragingly, the recent
report on intracellular screening of DNA-encoded chemical library
(DEL) showed cell-penetrating peptides could enable intracellular
operation of DNA-conjugated small molecules.[30] Second, this
Conclusion
In summary, we have developed an affinity labelling approach
for characterizing PPIs in protein complexes. Affinity probes are
powerful tools to explore biology, but they are mostly designed to
engage the direct target protein. Using affinity probes to
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
Dreger, H. Koester, Proteomics 2011, 11, 4096-4104; b) C. Xu,
E. Soragni, C. J. Chou, D. Herman, H. L. Plasterer, J. R. Rusche,
J. M. Gottesfeld, Chem. Biol. 2009, 16, 980-989; c) V. E. Albrow,
R. L. Grimley, J. Clulow, C. R. Rose, J. Sun, J. S. Warmus, E.
W. Tate, L. H. Jones, R. I. Storer, Mol. Biosyst. 2016, 12, 1781-
1789; d) C. M. Salisbury, B. F. Cravatt, P Natl Acad Sci USA
2007, 104, 1171-1176; e) C. M. Salisbury, B. F. Cravatt, J. Am.
Chem. Soc. 2008, 130, 2184-2194; f) B. He, S. Velaparthi, G.
Pieffet, C. Pennington, A. Mahesh, D. L. Holzle, M. Brunsteiner,
R. van Breemen, S. Y. Blond, P. A. Petukhov, J. Med. Chem.
2009, 52, 7003-7013; g) H. Abdelkarim, M. Brunsteiner, R.
Neelarapu, H. Bai, A. Madriaga, R. B. van Breemen, S. Y. Blond,
V. Gaponenko, P. A. Petukhov, ACS Chem. Biol. 2013, 8, 2538-
2549; h) T. W. Hanigan, S. M. Aboukhatwa, T. Y. Taha, J. Frasor,
P. A. Petukhov, Cell Chem Biol 2017, 24, 1356-1367 e1358; i) S.
M. Aboukhatwa, T. W. Hanigan, T. Y. Taha, J. Neerasa, R.
Ranjan, E. E. El-Bastawissy, M. A. Elkersh, T. F. El-Moselhy, J.
Frasor, N. Mahmud, A. McLachlan, P. A. Petukhov,
ChemMedChem 2019; j) S. Pan, S. Y. Jang, D. Wang, S. S. Liew,
Z. Li, J. S. Lee, S. Q. Yao, Angew. Chem. Int. Ed. 2017, 56,
11816-11821; k) D. P. Murale, S. C. Hong, M. M. Haque, J. S.
Lee, Proteome Sci. 2016, 15, 14.
study used pan-HDAC inhibitors to engage multiple HDACs and
identify more proteins, but additional steps were necessary to
deconvolute the specific interactions. Using isoform-specific
ligands would provide more specific information on individual
targets. We will perform more in-depth studies and explore the
biological applications of this method in future studies.
Acknowledgements
This work was supported by grants from the Research Grants
Council of the Hong Kong SAR (AoE/P-705/16, 17321916,
17302817, 17301118, 17111319), Laboratory for Synthetic
Chemistry and Chemical Biology of Health@InnoHK of ITC,
HKSAR, grants from the National Natural Science Foundation of
China (21572014, 21877093, 91953119). We thank Drs. Qian
Zhao, Xiucong Bao, Xiaomeng Li at HKU, Nan Chen, Xin Zeng,
Prof. Chu Wang at PKU for assistance with MS analysis. We
thank Dr. Eva Fung at The University of Hong Kong and Dr. Wen
Zhou at the Analytical Instrumentation Center at Peking University,
for MS analysis support.
[10] a) Y. Xie, J. Ge, H. Lei, B. Peng, H. Zhang, D. Wang, S. Pan, G.
Chen, L. Chen, Y. Wang, Q. Hao, S. Q. Yao, H. Sun, J. Am.
Chem. Soc. 2016, 138, 15596-15604; b) Y. Xie, L. Chen, R.
Wang, J. Wang, J. Li, W. Xu, Y. Li, S. Q. Yao, L. Zhang, Q. Hao,
H. Sun, J. Am. Chem. Soc. 2019, 141, 18428-18436; c) E.
Graham, S. Rymarchyk, M. Wood, Y. Cen, ACS Chem. Biol.
2018, 13, 782-792.
Keywords: DNA-protein conjugation • DNA-templated synthesis
• histone deacetylase • affinity probes • photo-affinity labelling
[11] B. Shan, C. Xu, Y. Zhang, T. Xu, J. M. Gottesfeld, J. R. Yates,
3rd, J Proteome Res 2014, 13, 4558-4566.
[12] R. Neelarapu, D. L. Holzle, S. Velaparthi, H. Bai, M. Brunsteiner,
S. Y. Blond, P. A. Petukhov, J. Med. Chem. 2011, 54, 4350-4364.
[13] a) G. Li, Y. Liu, L. Chen, S. Wu, X. Li, Angew. Chem. Int. Ed.
2013, 52, 9544-9549; b) G. Li, Y. Liu, X. Yu, X. Li, Bioconjug.
Chem. 2014, 25, 1172-1180.
[1] R. Marmorstein, M. M. Zhou, Cold Spring Harb. Perspect Biol.
2014, 6, a018762.
[2] T. Narita, B. T. Weinert, C. Choudhary, Nat. Rev. Mol. Cell Biol.
2019, 20, 156-174.
[3] K. J. Falkenberg, R. W. Johnstone, Nat. Rev. Drug Discov. 2014,
13, 673-691.
[4] H. T. Qin, H. Q. Li, F. Liu, Expert Opin. Ther. Pat. 2017, 27, 621-
[14] a) D. Y. Wang, Y. Cao, L. Y. Zheng, L. D. Chen, X. F. Chen, Z.
Y. Hong, Z. Y. Zhu, X. Li, Y. F. Chai, Chem. Eur. J. 2017, 23,
10906-10914; b) Y. Liu, W. Zheng, W. Zhang, N. Chen, Y. Liu, L.
Chen, X. Zhou, X. Chen, H. Zheng, X. Li, Chem. Sci. 2015, 6,
745-751; c) Y. Huang, W. Zheng, X. Li, Anal. Biochem. 2018,
545, 84-90; d) X. Bai, C. Lu, J. Jin, S. Tian, Z. Guo, P. Chen, G.
Zhai, S. Zheng, X. He, E. Fan, Y. Zhang, K. Zhang, Angew.
Chem. Int. Ed. 2016, 55, 7993-7997; e) W. Bi, X. Bai, F. Gao, C.
Lu, Y. Wang, G. Zhai, S. Tian, E. Fan, Y. Zhang, K. Zhang, Anal.
Chem. 2017, 89, 4071-4076; f) X. Bai, W. Bi, H. Dong, P. Chen,
S. Tian, G. Zhai, K. Zhang, Anal. Chem. 2018, 90, 3692-3696.
[15] a) C. B. Rosen, A. L. Kodal, J. S. Nielsen, D. H. Schaffert, C.
Scavenius, A. H. Okholm, N. V. Voigt, J. J. Enghild, J. Kjems, T.
Torring, K. V. Gothelf, Nat. Chem. 2014, 6, 804-809; b) T. B.
Nielsen, R. P. Thomsen, M. R. Mortensen, J. Kjems, P. F.
Nielsen, T. E. Nielsen, A. L. B. Kodal, E. Clo, K. V. Gothelf,
Angew. Chem. Int. Ed. Engl. 2019, 58, 9068-9072.
[16] a) C. Cui, H. Zhang, R. Wang, S. Cansiz, X. Pan, S. Wan, W.
Hou, L. Li, M. Chen, Y. Liu, X. Chen, Q. Liu, W. Tan, Angew.
Chem. Int. Ed. 2017, 56, 11954-11957; b) R. Wang, D. Lu, H.
Bai, C. Jin, G. Yan, M. Ye, L. Qiu, R. Chang, C. Cui, H. Liang, W.
Tan, Chem. Sci. 2016, 7, 2157-2161; c) L. Li, X. Chen, C. Cui, X.
Pan, X. Li, H. S. Yazd, Q. Wu, L. Qiu, J. Li, W. Tan, J. Am. Chem.
Soc. 2019, 141, 17174-17179.
[17] T. P. Krogager, R. J. Ernst, T. S. Elliott, L. Calo, V. Beranek, E.
Ciabatti, M. G. Spillantini, M. Tripodi, M. H. Hastings, J. W. Chin,
Nat. Biotechnol. 2018, 36, 156-159.
[18] C. Choudhary, C. Kumar, F. Gnad, M. L. Nielsen, M. Rehman, T.
C. Walther, J. V. Olsen, M. Mann, Science 2009, 325, 834-840.
[19] W. Huang da, B. T. Sherman, R. A. Lempicki, Nat. Protoc. 2009,
4, 44-57.
[20] N. Khan, M. Jeffers, S. Kumar, C. Hackett, F. Boldog, N.
Khramtsov, X. Qian, E. Mills, S. C. Berghs, N. Carey, P. W. Finn,
L. S. Collins, A. Tumber, J. W. Ritchie, P. B. Jensen, H. S.
Lichenstein, M. Sehested, Biochem. J. 2008, 409, 581-589.
[21] A. Suraweera, K. J. O'Byrne, D. J. Richard, Front. Oncol. 2018,
8.
636.
[5] a) C. J. Millard, P. J. Watson, L. Fairall, J. W. R. Schwabe,
Trends Pharmacol. Sci. 2017, 38, 363-377; b) P. Joshi, T. M.
Greco, A. J. Guise, Y. Luo, F. Yu, A. I. Nesvizhskii, I. M. Cristea,
Mol. Syst. Biol. 2013, 9, 672.
[6] a) E. Verdin, F. Dequiedt, W. Fischle, R. Frye, B. Marshall, B.
North, Methods Enzymol. 2004, 377, 180-196; b) M. Bantscheff,
C. Hopf, M. M. Savitski, A. Dittmann, P. Grandi, A. M. Michon, J.
Schlegl, Y. Abraham, I. Becher, G. Bergamini, M. Boesche, M.
Delling, B. Dumpelfeld, D. Eberhard, C. Huthmacher, T.
Mathieson, D. Poeckel, V. Reader, K. Strunk, G. Sweetman, U.
Kruse, G. Neubauer, N. G. Ramsden, G. Drewes, Nat.
Biotechnol. 2011, 29, 255-265.
[7] a) G. Padige, A. T. Negmeldin, M. K. Pflum, J. Biomol. Screen.
2015, 20, 1277-1285; b) H. Y. Kuo, T. A. DeLuca, W. M. Miller,
M. Mrksich, Anal. Chem. 2013, 85, 10635-10642; c) J. K. Tong,
C. A. Hassig, G. R. Schnitzler, R. E. Kingston, S. L. Schreiber,
Nature 1998, 395, 917-921; d) A. You, J. K. Tong, C. M.
Grozinger, S. L. Schreiber, Proc. Nat. Acad. Sci. USA 2001, 98,
1454-1458; e) M. K. Pflum, J. K. Tong, W. S. Lane, S. L.
Schreiber, J. Biol. Chem. 2001, 276, 47733-47741; f) P. A. Cole,
Nat. Chem. Biol. 2008, 4, 590-597; g) A. Dose, J. Sindlinger, J.
Bierlmeier, A. Bakirbas, K. Schulze-Osthoff, S. Einsele-Scholz,
M. Hartl, F. Essmann, I. Finkemeier, D. Schwarzer, Angew.
Chem. Int. Ed. Engl. 2016, 55, 1192-1195; h) J. Taunton, C. A.
Hassig, S. L. Schreiber, Science 1996, 272, 408-411; i) M.
Yoshida, M. Kijima, M. Akita, T. Beppu, J. Biol. Chem. 1990, 265,
17174-17179; j) M. Kijima, M. Yoshida, K. Sugita, S. Horinouchi,
T. Beppu, J. Biol. Chem. 1993, 268, 22429-22435; k) J. Taunton,
J. L. Collins, S. L. Schreiber, J. Am. Chem. Soc. 1996, 118,
10412-10422.
[8] a) I. Becher, A. Dittmann, M. M. Savitski, C. Hopf, G. Drewes, M.
Bantscheff, ACS Chem Biol 2014, 9, 1736-1746; b) J. Seidel, T.
Meisinger, J. Sindlinger, P. Pieloch, I. Finkemeier, D. Schwarzer,
ChemBioChem 2019.
[9] a) J. J. Fischer, S. Michaelis, A. K. Schrey, A. Diehl, O. Y.
Graebner, J. Ungewiss, S. Horzowski, M. Glinski, F. Kroll, M.
[22] J. J. Bravo-Cordero, M. A. O. Magalhaes, R. J. Eddy, L. Hodgson,
J. Condeelis, Nat. Rev. Mol. Cell Biol. 2013, 14, 405-415.
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
[23] S. N. Naryzhny, H. Lee, FEBS Lett. 2010, 584, 4292-4298.
[24] P. O. Hassa, S. S. Haenni, C. Buerki, N. I. Meier, W. S. Lane, H.
Owen, M. Gersbach, R. Imhof, M. O. Hottiger, J. Biol. Chem.
2005, 280, 40450-40464.
[25] M. Fujimoto, R. Takii, E. Takaki, A. Katiyar, R. Nakato, K.
Shirahige, A. Nakai, Nat. Commun. 2017, 8, 1638.
[26] a) W. Yang, Y. Xia, H. Ji, Y. Zheng, J. Liang, W. Huang, X. Gao,
K. Aldape, Z. Lu, Nature 2011, 480, 118-122; b) W. Yang, Y. Xia,
D. Hawke, X. Li, J. Liang, D. Xing, K. Aldape, T. Hunter, W. K.
Alfred Yung, Z. Lu, Cell 2012, 150, 685-696; c) A. Hamabe, M.
Konno, N. Tanuma, H. Shima, K. Tsunekuni, K. Kawamoto, N.
Nishida, J. Koseki, K. Mimori, N. Gotoh, H. Yamamoto, Y. Doki,
M. Mori, H. Ishii, Proc. Nat. Acad. Sci. USA 2014, 111, 15526-
15531.
[27] S. I. Liang, B. van Lengerich, K. Eichel, M. Cha, D. M. Patterson,
T. Y. Yoon, M. von Zastrow, N. Jura, Z. J. Gartner, Cell Rep.
2018, 22, 2593-2600.
[28] A. Citri, Y. Yarden, Nat. Rev. Mol. Cell Biol. 2006, 7, 505-516.
[29] X. Zhang, W. Wharton, Z. Yuan, S. C. Tsai, N. Olashaw, E. Seto,
Mol. Cell. Biol. 2004, 24, 5106-5118.
[30] B. Cai, D. Kim, S. Akhand, Y. Sun, R. Cassell, A. Alpsoy, E. C.
Dykhuizen, R. M. van Rijn, M. K. Wendt, C. Krusemark, J. Am.
Chem. Soc. 2019.
8
This article is protected by copyright. All rights reserved.

10.1002/anie.202001205
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
DNA-based affinity labeling probes enabled the capture and characterization of HDAC-associated proteins, including the direct and
indirect binders in HDAC complexes. This method provides a simple and broadly applicable chemical tool to study protein-protein
interactions.
9
This article is protected by copyright. All rights reserved.