Resource
Pervasive SUMOylation of heterochromatin and
piRNA pathway proteins
Graphical abstract
Highlights
d
diGly proteomics identifies
1,500 SUMO sites in
840
proteins in the fly ovary
d
SUMO targets are highly enriched among heterochromatin
and piRNA pathway proteins
d
Loss of SUMO disrupts the nuage
d
Piwi is required for the SUMOylation of Spn-E and Mael, but
not Panx, in the germline
Authors
Maria Ninova, Hannah Holmes,
Brett Lomenick, Katalin Fejes To
́
th,
Alexei A. Aravin
Correspondence
mninova@ucr.edu (M.N.),
aaa@caltech.edu (A.A.A.)
In brief
Adapting diGly proteomics for SUMO
target discovery, Ninova et al. identify
numerous SUMOylated proteins among
heterochromatin and piRNA pathway
factors, indicating unexpectedly broad
implications of protein modification by
SUMO in cellular pathways governing
heterochromatin regulation and
transposon control.
Ninova et al., 2023, Cell Genomics
3
, 100329
July 12, 2023
ª
2023 The Author(s).
https://doi.org/10.1016/j.xgen.2023.100329
ll
Resource
Pervasive SUMOylation of heterochromatin
and piRNA pathway proteins
Maria Ninova,
1,4,
*
Hannah Holmes,
1
Brett Lomenick,
3
Katalin Fejes To
́
th,
2
and Alexei A. Aravin
2,
*
1
Department of Biochemistry, University of California Riverside, 3401 Watkins Drive, Boyce Hall, Riverside, CA 92521, USA
2
Division of Biology and Biological Engineering, California Institute of Technology, 1200 E. California Boulevard, Pasadena, CA 91125, USA
3
Proteome Exploration Laboratory of the Beckman Institute, California Institute of Technology, 1200 E. California Boulevard, Pasadena, CA
91125, USA
4
Lead contact
*Correspondence:
mninova@ucr.edu
(M.N.),
aaa@caltech.edu
(A.A.A.)
https://doi.org/10.1016/j.xgen.2023.100329
SUMMARY
Genome regulation involves complex protein interactions that are often mediated through post-translational
modifications (PTMs). SUMOylation—modification by the small ubiquitin-like modifier (SUMO)—has been
implicated in numerous essential processes in eukaryotes. In
Drosophila
, SUMO is required for viability
and fertility, with its depletion from ovaries leading to heterochromatin loss and ectopic transposon
and gene activation. Here, we developed a proteomics-based strategy to uncover the
Drosophila
ovarian
‘‘SUMOylome,’’ which revealed that SUMOylation is widespread among proteins involved in heterochromatin
regulation and different aspects of the Piwi-interacting small RNA (piRNA) pathway that represses transpo-
sons. Furthermore, we show that SUMOylation of several piRNA pathway proteins occurs in a Piwi-depen-
dent manner. Together, these data highlight broad implications of protein SUMOylation in epigenetic
regulation and indicate novel roles of this modification in the cellular defense against genomic parasites.
Finally, this work provides a resource for the study of SUMOylation in other biological contexts in the
Drosophila
model.
INTRODUCTION
Post-translational modifications (PTMs) affect a broad range of
cellular functions such as protein turnover, localization to
different subcellular compartments, or specific interactions,
and are utilized in the regulation of various molecular pathways.
PTMs involve the covalent attachment of diverse moieties from
small chemical groups to entire modifier proteins to the main
polypeptide chain. The best-known protein modifier is ubiquitin,
a
9-kDa unit that gets conjugated to target lysine side chains
via an isopeptide bond between its C-terminal carboxyl group
and the lysine epsilon amino group. Following the discovery of
ubiquitin, other small proteins that can act as modifiers have
emerged, including the small ubiquitin-like modifier (SUMO).
SUMO is a
11-kDa protein that shares structural and sequence
homology with ubiquitin and similarly gets covalently attached to
target lysines. However, SUMO conjugation is mediated by
different enzymes and serves distinct and non-redundant func-
tions (reviewed in Flotho and Melchior
1
). In brief, the
SUMOylation cascade involves activation by a dedicated E1 het-
erodimer followed by transfer to the SUMO E2 conjugating
enzyme Ubc9. Ubc9 catalyzes SUMO transfer to the final
acceptor lysine, which often (but not always) resides within a
consensus motif
c
KxE (
c
is a hydrophobic amino acid) and is
sufficient for SUMOylation
in vitro
.
2–4
Nevertheless, non-cata-
lytic SUMO E3 ligases can facilitate Ubc9 or enable substrate
specificity, and they seem to be required for SUMOylation in
some contexts and perhaps for non-consensus sites
in vivo
.
5–8
SUMO is primarily nuclear, and since its discovery has emerged
as an important regulator of different nuclear processes (re-
viewed in Xhao
9
) such as transcription factor activity, DNA repair,
rRNA biogenesis, chromosome organization, and segregation.
Mechanistically, SUMOylation may lead to diverse conse-
quences, including changes in protein conformation or localiza-
tion, masking or competing with other PTMs, and, most
famously, regulating protein-protein interactions. SUMO-medi-
ated interactions typically involve an aliphatic stretch flanked
by acidic amino acids (SUMO interacting motif [SIM]) in the
partner protein and, although individually weak in their nature,
multivalent SUMO-SIM interactions were proposed to stabilize
large molecular complexes
10
,
11
and promote the formation of
phase-separated compartments such as promyelocytic leuke-
mia (PML) protein bodies.
12
However, despite being implicated
in a myriad of biological processes, our understanding of SU-
MO’s role within different molecular and cellular contexts is far
from complete.
Previous work in
Drosophila
implicated the SUMO pathway
in the regulation of heterochromatin establishment and Piwi-
interacting small RNA (piRNA)-mediated transposon (transpos-
able element, TE) silencing.
13–16
In germ cells, Piwi clade
proteins (Piwi, Aub, and Ago3 in flies) and piRNAs cooperate in
intimately linked processes that ensure transcriptional and
Cell Genomics
3
, 100329, July 12, 2023
ª
2023 The Author(s).
1
This is an open access article under the CC BY-NC-ND license (
http://creativecommons.org/licenses/by-nc-nd/4.0/
).
ll
OPEN ACCESS
post-transcriptional transposon silencing and continuous pro-
duction of mature piRNAs (reviewed in Czech and Hannon
17
).
Mature piRNA production and post-transcriptional cleavage of
transposon RNAs by Aub and Ago3 occur in a dedicated perinu-
clear structure, the nuage, similar to ‘‘germ granules’’ in other
systems. Antisense piRNAs produced in the nuage also become
loaded in Piwi, which then enters the nucleus to find transposon
nascent RNA and enforce co-transcriptional silencing at target
loci.
18–20
To date, SUMOylation is known to participate in the nu-
clear piRNA pathway in several ways. First, the SUMO E3 ligase
Su(var)2-10 was found to interact with piRNA pathway and het-
erochromatin proteins and play an essential role in the recruit-
ment of the enzymatic complex SetDB1/Wde, which deposits
the silencing epigenetic mark H3K9me3,
14
as well as with the
MEP-1/Mi-2 chromatin remodeler complex.
16
Second,
SUMOylation of Panoramix (Panx)—a co-repressor required for
H3K9me3 deposition downstream of Piwi—was found to
mediate its interaction with the general heterochromatin effector
Sov.
15
In addition to silencing of transposons, Su(var)2-10 and
SUMO were found to control H3K9me3 deposition at piRNA-in-
dependent genomic loci such as developmentally silenced tis-
sue-specific genes.
21
The pervasive effect of SUMO depletion
on global H3K9me3 levels at diverse classes of genomic targets
suggests that SUMOylation is a critical process in the regulation
of repressive chromatin integrated within different regulatory
pathways. However, further understanding of the role of
SUMOylation on chromatin requires knowledge of the full spec-
trum of SUMO targets.
We developed a proteomics approach that enables the identi-
fication of SUMOylated proteins with amino acid-level site pre-
dictions from different tissues in the classic model for piRNA
and heterochromatin studies,
Drosophila melanogaster
. Here,
we report a comprehensive dataset of SUMO targets in the fly
ovary. Notably, we identified strong enrichment of heterochro-
matin factors among SUMOylated proteins, supporting the
notion that SUMO plays a complex role in heterochromatin regu-
lation that extends to multiple targets and protein complexes.
Moreover, we find a striking enrichment of proteins specific to
the piRNA pathway among SUMO targets, including the central
effector of epigenetic silencing by piRNAs, Piwi, and several pro-
teins that localize to the nucleus and nuage. We further validated
the SUMOylation of selected piRNA pathway factors—Piwi,
Panx, Spindle-E (Spn-E), and Maelstrom (Mael)—in germ cells
and showed that the modification of Mael and Spn-E, but not
Panx, is Piwi dependent, indicative of multiple SUMO roles in
distinct steps of transposon silencing. Altogether, our findings
point to a previously unappreciated multilayered role of
SUMOylation in the piRNA pathway and heterochromatin regula-
tion that provide important clues toward our understanding of
the molecular mechanisms of genome regulation and trans-
poson control.
RESULTS
Establishing a system for the detection of SUMOylated
proteins in
Drosophila
Proteome-wide studies of PTMs have benefited from the devel-
opment of methods and reagents that enable specific enrich-
ment of modified proteins or peptides from total protein lysates.
Basic methods for the enrichment of SUMO-modified proteins
involve pull-down with anti-SUMO antibodies. However, this
approach prohibits the use of stringent washing conditions
and is therefore prone to high background. Furthermore, endog-
enous SUMO does not have trypsin cleavage sites close to its
C terminus, and trypsin digestion of SUMOylated proteins gener-
ates large, branched peptides from modified regions that
are incompatible with conventional bottom-up proteomics
(
Figure 1
A). Accordingly, specific modified residues remain
unknown and SUMOylation events can only be inferred indirectly
from the abundance of other peptides from purified SUMOylated
proteins.
To overcome these obstacles and obtain a high-confidence
dataset of the SUMO-modified proteins (herein referred to as
the SUMOylome) in
Drosophila
, we adapted an approach that al-
lows stringent purification, enrichment, and proteomic detection
of peptides containing a modified SUMO remnant.
22
This
approach, originally developed for the study of SUMOylation in
human cells,
22
employs ectopically expressed SUMO protein
with 6xHis-tag and a point mutation (T > R substitution, herein
referred to as SUMO-TR) that enables trypsin cleavage before
the C-terminal GG motif at the conjugation site (
Figures 1
A and
1B). In a two-step purification process, SUMO-modified proteins
are first selected based on the His tag by nickel affinity under
denaturing conditions, which eliminates the activity of SUMO
proteases and most background from noncovalently bound pro-
teins. The resulting SUMOylated protein-enriched fraction is
then trypsinized to generate a mixture of peptides including
branched peptides with short di-glycine (diGly) remnant from
cleaved SUMO-TR moiety. These diGly remnant-containing
peptides are further enriched by pull-down with a specific anti-
body, purified, and analyzed by mass spectrometry (MS). As
the diGly remnant alters the molecular mass and charge of a
given peptide, diGly-modified species can be distinguished
from other background peptides with high confidence. Of note,
ubiquitin and other ubiquitin-like protein modifiers naturally
have an arginine residue before the terminal diGly motif; there-
fore, any ubiquitinated protein that unspecifically co-purifies
with SUMOylated proteins during the His-based enrichment
step can also generate diGly peptides. However, this back-
ground can be accounted for through negative control samples
from tissues that do not express SUMO-TR (
Figure 1
B). To
enable this method for SUMOylation detection in
Drosophila
tis-
sues, we created a transgenic line that carries a modified copy of
the
smt3
locus (which encodes the single SUMO homolog in this
species) where the SUMO coding sequence has a 6xHis N-ter-
minal tag, T86 > R substitution before the C-terminal GG motif,
and a
2.5-kb upstream region containing the putative endoge-
nous promoter. The His-tagged SUMO-TR protein was detect-
able by western blotting (WB) (
Figure 1
C) and, importantly, this
transgene completely rescues the lethality of the null allele
smt3
04493
, confirming that it encodes a fully functional protein.
Identification and characterization of the
Drosophila
ovarian SUMOylome
We used flies expressing SUMO-TR to optimize the two-step
purification procedure described above (
Figure 1
B) and obtain
2
Cell Genomics
3
, 100329, July 12, 2023
Resource
ll
OPEN ACCESS
SUMOylation site-derived peptides from ovarian tissue (see
STAR Methods
for details). Following this procedure, we per-
formed three independent experiments, where each replica
involved parallel sample preparation from SUMO-TR ovaries
and wild-type controls followed by label-free tandem MS. Each
sample yielded
1,000–2,500 diGly sites (
Figure 2
A) and, alto-
gether, we detected 3,159 exact diGly sites mapping to the pro-
tein products of 1,295 genes. In comparison, MS analyses of
input samples prior to diGly enrichment yielded less than 100
diGly sites (data not shown), highlighting the importance of the
diGly enrichment step for capturing SUMOylation sites. For
each diGly site, we calculated the normalized intensity ratios be-
tween SUMO-TR and control samples. The ratios showed a
prominent bimodal distribution: approximately half of the sites
in each experiment were detected exclusively in SUMO-TR sam-
ples, indicative of genuine SUMOylation targets (
Figure 2
B). The
diGly sites detected with similar intensities in SUMO-TR and
control samples or biased to control samples likely originate
from unspecifically bound ubiquitinated proteins. Consistent
with this, motif analysis showed that diGly sites detected in
SUMO-TR samples are enriched in the canonical SUMOylation
motif, while diGly sites from the negative control do not have
any motif enrichment (
Figures 2
B and
S1
).
The sets of predicted
bona fide
SUMO targets (SUMO-TR/
control intensity ratio >10) were highly reproducible between
the three experiments, with
50% overlap on the specific site
level, and a 75% overlap on the target protein-coding gene level
(
Figure 2
C). Previously known SUMO targets, including
RanGAP1, PCNA, and Su(var)2-10/dPIAS, are present in the
high-confidence set. Specific sites that do not appear in all ex-
periments tend to have the lowest intensities (
Figure S2
A).
Considering that SUMOylation is often transient and affects a
minuscule fraction of a given protein, it is likely that sensitivity
is the major limitation to the reproducibility of detected sites
A
C
D
B
Figure 1. Proteomics approach for SUMO site detection
(A) Diagram of the peptides generated after trypsin digestion of proteins modified by wild-type and T > R mutant SUMO.
(B) Diagram of the experimental design and work flow for SUMOylation-derived diGly-modified peptide MS analysis.
(C) Diagram of the recombinant SUMO construct used to generate SUMO-TR expressing transgenic flies.
(D) Western blot of total ovary lysate confirming the expression of recombinant SUMO. Ovaries from flies overexpressing 6xHis-FLAG-tagged SUMO are us
ed as
a positive control.
Cell Genomics
3
, 100329, July 12, 2023
3
Resource
ll
OPEN ACCESS