Abstract
Sentrin/SUMO-specific protease 2 (SENP2) is a member of SENPs family
involved in maturation of SUMO precursors and deSUMOylation of specific
target, and is highly expressed in the central nervous system (CNS).
Although SENP2 has been shown to modulate embryonic development, fatty
acid metabolism, atherosclerosis and epilepsy, the function of SENP2 in
the CNS remains poorly understood. To address the role of SENP2 in the
CNS and its potential involvement in neuropathology, we generated SENP2
conditional knockout mice by crossing floxed SENP2 mice with
CaMKIIα-Cre transgenic mice. Behavioral tests revealed that SENP2
ablation induced hyper-locomotor activity, anxiolytic-like behaviors,
spatial working memory impairment and fear-associated learning defect.
In line with these observations, our RNA sequencing (RNA-seq) data
identified a variety of differential expression genes that are
particularly enriched in locomotion, learning and memory related
biologic process. Taken together, our results indicated that SENP2
plays a critical role in emotional and cognitive regulation. This SENP2
conditional knockout mice model may help reveal novel mechanisms that
underlie a variety of neuropsychiatric disorders associated with
anxiety and cognition.
Keywords: SENP2, Conditional knockout mice, Anxiety-like behavior,
Learning and memory, RNA-seq
Introduction
SUMOylation is a dynamic and reversible post-translational modification
that modulates diverse functions of target proteins, including protein
stability, protein subcellular localization, protein-protein or
protein-DNA interactions, and protein kinase activity [[33]1]. Cellular
abundance of particular SUMO-conjugated substrates is regulated by a
balance between SUMO conjugation and SUMO deconjugation.
Sentrin-specific proteases (SENPs) catalyze the removal of SUMO from
SUMO-conjugated target proteins as well as the cleavage of SUMO from
its precursor proteins, thus playing a critical role in regulating the
SUMOylation level of targets [[34]2, [35]3]. In mammals, the SENP
family consists of six members, which can be divide into three groups
(SENP1 and SENP2; SENP3 and SENP5; SENP6 and SENP7) based on homology
and function analysis [[36]3]. We recently showed that SENP1
participates in regulating nociceptive signaling in models of
inflammatory pain and attenuates I/R injury induced cell death in a
transient brain ischemia/reperfusion mouse model [[37]4, [38]5].
Moreover, SENP2 has been reported to play a role in cardiac development
[[39]6], neuronal survive [[40]7] and seizure [[41]8]. According to the
Allen Brain Atlas, SENP2 mRNAs are highly expressed in the forebrain
[[42]9], but the function of SENP2 in the central nervous system (CNS)
remains unclear. As SENP2 is required for expression of key
developmental genes, global deletion of SENP2 is embryonically lethal
[[43]6–[44]8, [45]10]. Thus, we developed a forebrain excitatory
neuron-specific SENP2 knockout mouse model to examine SENP2 functions
in the CNS. We found that these conditional knockout (cKO) animals
display hyperactivity and reduced anxiety-like behavior, impaired
learning and memory. Gene ontology (GO) analysis of the differential
expression genes revealed enrichment for numerous cellular and
molecular functional categories, including those related to “Cell
death” and “Immune response”. Taken together, our results indicate that
SENP2 plays an important role in the forebrain, and its absence leads
to molecular and behavioral changes associated with locomotion,
anxiety, learning and memory.
Results
Generation of forebrain-specific SENP2 cKO mice
To determine the functional role of SENP2 in the CNS, we crossed floxed
SENP2 (SENP2^fl/fl) mice [[46]8] with CaMKIIα-Cre transgenic mice
[[47]11] to generate SENP2 conditional knockout (cKO) mice, in which
SENP2 was selectively removed from principal neurons of postnatal
forebrain (Fig. [48]1a-b). Because the CaMKIIα-Cre transgene is
expressed between postnatal days 14–21 in excitatory neurons in the
forebrain [[49]11], this allowed us to specifically assess SENP2
function in the postnatal forebrain without disrupting its contribution
to early CNS development and/or causing embryonic lethality. In the cKO
mice, SENP2 protein levels were strongly reduced in the forebrain
(cortical and hippocampal) excitatory neurons (Fig. [50]1c). As
determined by western blot analyses, loss of SENP2 protein expression
occurred in the cortex (58.01 ± 4.90% of littermate controls) and
hippocampus (42.96 ± 6.24% of littermate controls) of 6-week-old cKO
mice, but not in the cerebellum (105.66 ± 9.99% of littermate
controls), where the Cre recombinase is not expressed (Fig. [51]1d).
Additionally, real-time quantitative PCR (qPCR) determined that SENP2
mRNA levels in the cKO mice were significantly reduced in cortex
(45.82 ± 2.94% of littermate controls) and hippocampus (35.17 ± 6.07%
of littermate controls) of 6-week-old cKO mice. However, no change was
observed in the cKO mice at 3 weeks after birth (Fig. [52]1e).
Fig. 1.
[53]Fig. 1
[54]Open in a new tab
Generation of forebrain-specific SENP2 cKO mice. a The schematic
diagram of SENP2 cKO mouse generation. b Genotyping identification of
conditional knockouts by PCR. c Confocal microscopy photomicrographs
showing double immunostaining of CaMKIIα (green) and SENP2 (red) in
excitatory neurons of 6-week-old cKO and littermate control mice. There
is significantly less SENP2 positive excitatory neurons in cortex (Ctx)
and hippocampus (Hip), but not in cerebellum (Ceb) brain slices. Scale
bar = 50 μm. d In 6-week-old cKO mice, As detected by western blots,
the SENP2 protein level is significantly reduced in the Ctx and Hip,
but not in the Ceb of 6-week-old SENP2 cKO mice (left panel).
Quantification of the Western blots is shown in the right panel. e As
detected by qPCR, the SENP2 mRNA level is significantly reduced in Ctx
and Hip, but not in the Ceb of 6-week-old cKO mice. All data are
presented as mean ± SEM. Ctrl (Control): n = 3; cKO (SENP2 conditional
knockout): n = 3. Statistical analysis performed with two-way ANOVA
followed by Bonferroni’s post-hoc. *p < 0.05, **p < 0.01, ***p < 0.001
compared with littermate controls
Behavioral analyses of cKO mice
We performed a battery of behavioral tests to evaluate the behavioral
phenotype of SENP2 cKO mice. In open field test, cKO mice travelled
more distance (m) than their littermate controls during a 30 min
recording period (Ctrl (Control): 94.66 ± 4.43, n = 19; cKO:
216.26 ± 15.81, n = 12; p < 0.0001, Welch’s t-test) (Fig. [55]2a-c). As
abnormal exploratory behaviors may also be indicative of changes in
anxiety, we assessed the number of entries, duration spent and distance
traveled in center area. We observed that SENP2 cKO mice entered the
center more frequently (Ctrl: 68.11 ± 6.00, n = 19; cKO: 94.75 ± 6.48,
n = 12; p = 0.0068, Student’s t-test) (Fig. [56]2d-e), spent more time
(s) in center area (91.64 ± 9.97, n = 19; cKO: 149.04 ± 11.42, n = 12;
p = 0.0009, Student’s t-test) (Fig. [57]2f-g) and traveled more
distance in center area than littermate controls (Ctrl: 9.93 ± 0.95,
n = 19; cKO: 13.85 ± 1.00, n = 12; p < 0.0107, Student’s t-test) in a
30 min test session (Fig. [58]2h-i). Although a 5 min test session is
often sufficient to assess the critical components of general
exploratory locomotion, the most commonly used measure of overall
exploratory/locomotor activity is the total distance traveled. SENP2
cKO mice spent longer time and entered more frequency in the center
area than littermate control mice at 10 min, 15 min, 20 min(Fig.
[59]2d, f), and traveled more distance at 10 min, 20 min (Fig. [60]2h),
although no statistically significant difference between cKO and
littermate control mice was observed at 5 min, 25 min and 30 min. These
findings suggest that cKO mice showed great aspiration for exploring
the center area, and thus displayed decreased anxiety-like behavior.
Fig. 2.
[61]Fig. 2
[62]Open in a new tab
SENP2 cKO mice exhibited hyperactivity and decreased anxiety-like
behavior. a Representative exploratory tracks of either littermate
control (Ctrl) or cKO mice in the open field. b-i Analysis of open
field exploration behavior for b distance traveled (m): increased
locomotor activity of cKO mice at every 5 min block (5 min: Ctrl:
18.91 ± 1.03, n = 19; cKO: 31.69 ± 1.81, n = 12; p < 0.0001; 10 min:
Ctrl: 15.76 ± 0.73, n = 19; cKO: 35.64 ± 2.22, n = 12; p < 0.0001;
15 min: Ctrl: 15.40 ± 0.96, n = 19; cKO: 37.76 ± 2.53, n = 12;
p < 0.0001; 20 min: Ctrl: 14.60 ± 0.96, n = 19; cKO: 39.16 ± 3.13,
n = 12; p < 0.0001; 25 min: Ctrl: 14.75 ± 0.88, n = 19; cKO:
36.61 ± 3.62, n = 12; p < 0.0001; 30 min: Ctrl: 15.24 ± 0.88, n = 19;
cKO: 35.40 ± 3.92, n = 12; p < 0.0001; two-way ANOVA analysis followed
by Bonferroni’s post-hoc), and c increased locomotor activity of cKO
mice in the 30-min open field test (Ctrl: 94.66 ± 4.43, n = 19; cKO:
216.26 ± 15.81, n = 12; p < 0.0001, Welch’s t-test). d Analysis of
number of entries in center area at every 5 min: Increased number of
entries of cKO mice in center area at 10, 15 and 20 min (5 min: Ctrl:
12.16 ± 1.56, n = 19; cKO: 12.50 ± 1.73, n = 12; p > 0.9999; 10 min:
Ctrl: 10.11 ± 1.11, n = 19; cKO: 15.92 ± 1.49, n = 12; p = 0.0422;
15 min: Ctrl: 10.53 ± 1.05, n = 19; cKO: 16.33 ± 1.65, n = 12;
p = 0.0425; 20 min: Ctrl: 11.11 ± 1.35, n = 19; cKO: 18.33 ± 1.78,
n = 12; p = 0.0052; 25 min: Ctrl: 11.58 ± 1.27, n = 19; cKO:
14.92 ± 1.45, n = 12; p = 0.7143; 30 min: Ctrl: 13.33 ± 1.35, n = 19;
cKO: 16.75 ± 2.18, n = 12; p = 0.6865; two-way ANOVA analysis followed
by Bonferroni’s post-hoc), and e increased number of entries of cKO
mice in the 30-min open field test (Ctrl: 68.11 ± 6.00, n = 19; cKO:
94.75 ± 6.48, n = 12; p = 0.0068, Student’s t-test). f Analysis of
duration spent (s) in the center of the field at every 5 min: An
increased center time was observed in cKO mice at 10, 15 and 20 min
(5 min: Ctrl: 14.44 ± 2.52, n = 19; cKO: 17.35 ± 3.57, n = 12;
p > 0.9999; 10 min: Ctrl: 13.17 ± 2.30, n = 19; cKO: 24.00 ± 12.14,
n = 12; p = 0.0477; 15 min: Ctrl: 14.37 ± 2.34, n = 19; cKO:
25.12 ± 3.00, n = 12; p = 0.0497; 20 min: Ctrl: 14.47 ± 2.19, n = 19;
cKO: 33.24 ± 4.43, n = 12; p < 0.0001; 25 min: Ctrl: 17.20 ± 2.77,
n = 19; cKO: 25.00 ± 3.07, n = 12; p = 0.3334; 30 min: Ctrl:
19.00 ± 2.37, n = 19; cKO: 24.40 ± 3.26, n = 12; p > 0.9999; two-way
ANOVA analysis followed by Bonferroni’s post-hoc), and g increased time
spent (s) of cKO mice in the center of the field in the 30-min open
field test (Ctrl: 91.64 ± 9.97, n = 19; cKO: 149.04 ± 11.42, n = 12;
p = 0.0009, Student’s t-test). h Analysis of distance traveled (m) in
center area at every 5 min: Increased distance traveled of cKO mice in
the center area at 10 and 20 min (5 min: Ctrl: 1.72 ± 0.26, n = 19;
cKO: 1.94 ± 0.32, n = 12; p > 0.9999; 10 min: Ctrl: 1.44 ± 0.18,
n = 19; cKO: 2.38 ± 0.18, n = 12; p = 0.0379; 15 min: Ctrl:
1.53 ± 0.17, n = 19; cKO: 2.31 ± 0.18, n = 12; p = 0.1428; 20 min:
Ctrl: 1.61 ± 0.19, n = 19; cKO: 2.57 ± 0.33, n = 12; p = 0.0324;
25 min: Ctrl: 1.74 ± 0.19, n = 19; cKO: 2.26 ± 0.26, n = 12;
p = 0.7768; 30 min: Ctrl: 1.88 ± 0.26, n = 19; cKO: 2.39 ± 0.32,
n = 12; p > 0.9999; two-way ANOVA analysis followed by Bonferroni’s
post-hoc), and i Increased distance traveled (m) in center area of the
field in the 30-min open field test (Ctrl: 9.93 ± 0.95 m, n = 19; cKO:
13.85 ± 1.00 m, n = 12; p < 0.0107, Student’s t-test). j Representative
track of exploration in elevated plus maze of either littermate control
(Ctrl) or cKO mice. k, i Analysis of number of entries in open arms or
closed arms: more number of cKO mice entries in open arms (Ctrl:
6.84 ± 0.97, n = 19; cKO: 21.58 ± 3.91, n = 12; p = 0.0031, Welch’s
t-test), l less number of cKO mice entries in closed arms (Ctrl:
25.26 ± 1.32, n = 19; cKO: 19.00 ± 2.49, n = 12; p = 0.0213, Student’s
t-test), m, n Analysis of percentage of time spent in open arms or
closed arms: more percentage of time spent in closed arms of cKO mice
(Ctrl: 6.85 ± 1.54%, n = 19; cKO: 37.20 ± 6.90%, n = 12; p = 0.0010,
Welch’s t-test), n less percentage of time spent in closed arms of cKO
mice (Ctrl: 68.31 ± 2.94%, n = 19; cKO: 43.36 ± 7.81%, n = 12;
p = 0.0096, Welch’s t-test). o Novelty suppressed feeding. cKO mice
have less latency (s) to feeding compared with littermate control mice
(Ctrl: 294.79 ± 36.24, n = 19; cKO: 89.42 ± 38.73, n = 12; p = 0.0008,
Student’s t-test). All data presented as mean ± S.E.M.
To further explore the anxiety-like behaviors, cKO mice were evaluated
in two anxiety-related behavioral assays including elevated plus maze
and novelty suppressed feeding. In the evaluated plus maze test, cKO
mice gained more number of entries in the open arms (Ctrl: 6.84 ± 0.97,
n = 19; cKO: 21.58 ± 3.91, n = 12; p = 0.0031, Welch’s t-test) (Fig.
[63]2k) and less number of entries in the closed arms accordingly
(Ctrl: 25.26 ± 1.32, n = 19; cKO: 19.00 ± 2.49, n = 12; p = 0.0213,
Student’s t-test) (Fig. [64]2l). In addition, cKO mice spent more
percentage of time in the open arms (Ctrl: 6.85 ± 1.54%, n = 19; cKO:
37.20 ± 6.90%, n = 12; p = 0.0010, Welch’s t-test) (Fig. [65]2m) and
less percentage of time in the closed arms (Ctrl: 68.31 ± 2.94%,
n = 19; cKO: 43.36 ± 7.81%, n = 12; p = 0.0096, Welch’s t-test) (Fig.
[66]2n). Moreover, cKO mice significantly reduced the latency (s) to
feed in the novel environment compared with littermate controls (Ctrl:
294.79 ± 36.24, n = 19; cKO: 89.42 ± 38.73, n = 12; p = 0.0008,
Student’s t-test) (Fig. [67]2o). Taken together, these results
demonstrated that cKO mice exhibited anxiolytic-like behavior.
Previous studies have suggested that SUMOylation plays an important
role in learning and memory [[68]12–[69]14]. Given the critical
function of SENP2 in regulating SUMOylation status and its high-level
expression in the brain, we investigate whether the loss of SENP2 in
forebrain excitatory neurons impairs learning and memory. As first, we
assessed spatial working and reference memory of cKO mice using the
Y-maze spontaneous alternation task. Compared with littermate controls,
cKO mice displayed significantly reduced alternations (Ctrl:
72.18 ± 1.84%, n = 18; cKO: 60.35 ± 4.22%, n = 10; p = 0.0062,
Student’s t-test) (Fig. [70]3a), suggesting an impairment of spatial
working memory in cKO mice. Moreover, we examined associative learning
and memory behaviors using a contextual fear conditioning protocol
[[71]15]. As shown in Fig. [72]3b, littermate control mice responded
well in training session and exhibited freezing behavior when
reintroduced to the same context 24 h later. In contrast, cKO mice not
only showed impaired learning ability during training, but also
displayed extremely low freezing level in contextual test session,
demonstrating an impairment of cKO mice in associative learning and
memory. Furthermore, we found that the cKO mice had a substantially
decreased nesting score in the nest building assay compared with
littermate control mice (Ctrl: 4.30 ± 0.26, n = 10; cKO: 1.44 ± 0.24,
n = 9; p < 0.0001, Student’s t-test) (Fig. [73]3c). In rodents, the
nest building behavior represents a form of social behaviors, and that
impaired nest building is considered to represent a negative phenotype
of psychiatric diseases including schizophrenia [[74]16]. To determine
whether cKO mice displayed other core characteristics of the
neuropsychiatric disorders, we conducted prepulse inhibition (PPI) task
to measure the sensorimotor gating of cKO mice [[75]17]. Compared with
their littermates, cKO mice displayed normal startle reaction (Ctrl:
475.71 ± 78.32, n = 8; cKO: 522.48 ± 58.08, n = 8; p = 0.6389,
Student’s t-test) (Fig. [76]3d). When assayed in a prepulse inhibition
(PPI) task, there were no differences between cKO and littermate
control mice in the extent of PPI at 3 increasing prepulse sound
intensities (%PPI of 70 dB prepulse stimulus: Ctrl: 47.64 ± 5.23%,
n = 8; cKO: 42.90 ± 3.01%, n = 8; p > 0.9999; %PPI of 74 dB prepulse
stimulus: Ctrl: 47.86 ± 6.92%, n = 8; cKO: 41.00 ± 3.84%, n = 8;
p = 0.9895; %PPI of 78 dB prepulse stimulus: Ctrl: 47.63 ± 6.44%,
n = 8; cKO: 48.23 ± 2.21%, n = 8; p > 0.9999; two-way ANOVA analysis
followed by Bonferroni’s post-hoc) (Fig. [77]3d). Thus, cKO mice had no
obvious deficit in PPI.
Fig. 3.
[78]Fig. 3
[79]Open in a new tab
SENP2 ablation impaired working memory, contextual fear learning and
nest building activity. a-b Cognitive test of cKO mice. a Y maze test.
cKO mice showed reduced percentage of accurate spontaneous alternations
over total number of alternations among the three arms (Ctrl:
72.18 ± 1.84%, n = 18; cKO: 60.35 ± 4.22%, n = 10; p = 0.0062,
Student’s t-test). b Contextual fear condition test. cKO mice showed
low levels of freezing behaviors in training session (pre-shock: Ctrl:
3.25 ± 0.94%, n = 19; cKO: 2.66 ± 1.47%, n = 11, p > 0.9999; 1st shock:
Ctrl: 7.83 ± 2.59%, n = 19; cKO: 0.89 ± 0.52%, n = 11, p > 0.9999; 2nd
shock: Ctrl: 23.20 ± 4.06%, n = 19; cKO: 0.86 ± 0.48%, n = 11,
p = 0.0009; 3rd shock: Ctrl: 39.05 ± 5.48%, n = 19; cKO: 1.11 ± 0.63%,
n = 11, p < 0.0001; 4th shock: Ctrl: 48.83 ± 5.80%, n = 19;
cKO:1.87 ± 1.11%, n = 11, p < 0.0001; 5th shock: Ctrl: 51.21 ± 4.96%,
n = 19; cKO: 2.61 ± 1.01%, n = 11, p < 0.0001; two-way ANOVA analysis
followed by Bonferroni’s post-hoc). In contextual fear retrieval
session, cKO mice display very lowly freezing behaviors (2 min: Ctrl:
42.92 ± 4.97%, n = 19; cKO: 21.79 ± 4.21%, n = 11, p = 0.0481; 4 min:
Ctrl: 50.61 ± 5.81%, n = 19; cKO: 14.55 ± 5.14%, n = 11, p < 0.0001;
6 min: Ctrl: 51.57 ± 5.35%, n = 19; cKO: 13.10 ± 4.79%, n = 11,
p < 0.0001; 8 min: Ctrl: 52.49 ± 5.41%, n = 19; cKO: 14.05 ± 7.04%,
n = 11, p < 0.0001; 10 min: Ctrl: 47.86 ± 5.02%, n = 19; cKO:
11.57 ± 4.60%, n = 11, p < 0.0001; two-way ANOVA analysis followed by
Bonferroni’s post-hoc). c Nest building test. cKO mice display lower
nest scores in the nest building test (Ctrl: 4.30 ± 0.26, n = 10; cKO:
1.44 ± 0.24, n = 9; p < 0.0001, Student’s t-test). d Acoustic startle
response and prepulse inhibition in cKO mice. cKO mice display normal
startle reflex in 120 dB acoustic stimulus compared with littermate
controls (Ctrl: 475.71 ± 78.32, n = 8; cKO: 522.48 ± 58.08, n = 8;
p = 0.6389, Student’s t-test) and showed similar PPI of the startle
response than littermate control mice at the prepulse level of 70 dB,
74 dB, 78 dB (70 dB: Ctrl: 47.64 ± 5.23%, n = 8; cKO: 42.90 ± 3.01%,
n = 8; p > 0.9999; 74 dB: Ctrl: Ctrl: 47.86 ± 6.92%, n = 8; cKO:
41.00 ± 3.84%, n = 8; p = 0.9895; 78 dB: Ctrl: 47.63 ± 6.44%, n = 8;
cKO: 48.23 ± 2.21%, n = 8; p > 0.9999; two-way ANOVA analysis followed
by Bonferroni’s post-hoc). All data presented as mean ± S.E.M.
Identification of SENP2-regulated transcripts in the cerebral cortex
To elucidate the molecular mechanism underlying the behavioral deficits
exhibited in the cKO mice, we performed high-throughput RNA sequencing
(RNA-seq) to identify genes with altered expressions when SENP2 is
selectively removed in the forebrain excitatory neurons. RNAs were
prepared from cerebral cortex tissues isolated from the brains of
6-week-old cKO and their littermate control mice. We sequenced RNA
libraries from 3 biological replicates per genotype and evaluated the
data by pearson correlation coefficient. Differential expression genes
(DEGs) analysis revealed consistent changes between the two genotypes
across all 3 replicates (Fig. [80]4a). We observed 863 up-regulated
genes and 170 down-regulated genes in cerebral cortex of cKO mice using
an adjusted p value < 0.05 and relative gene expression level > 2 fold
change (Fig. [81]4b). A detailed comparative analysis of the gene
expression profiles appears in Additional file [82]1: Table 1.
Fig. 4.
[83]Fig. 4
[84]Open in a new tab
RNA-seq analysis of cKO. a The pearson correlation coefficient between
cKO and littermate control mice. Results showed that correlation
coefficient of 3 pairs cortical samples from cKO and littermate control
mice was more than 0.95. b The differential expression genes between
cKO and littermate control mice. The red dots represent up-regulated
863 genes, while the green dots represent down-regulated 170 genes
using the most stringent criteria (intersection of adjusted p < 0.05
and the relative gene expression level > 2 fold change, the gene was
considered as the differential expression gene)
GO and KEGG enrichment analysis for DEGs
Gene ontology (GO) analysis of the DEGs revealed SENP2 ablation altered
the expression of genes involved in several biological processes,
including locomotion, learning and memory (Fig. [85]5a). Additionally,
GO analysis of DEGs revealed enrichment for numerous cellular and
molecular functional categories, including those related to “Cell
death” (Fig. [86]5b) and “Immune response” (Fig. [87]5c). A detailed GO
enrichment analysis of the related GO terms is included in the
Additional file [88]1: Table 2. Together, these results suggest that
DEGs between cKO and littermate control mice are enriched for
locomotion, learning and memory, cell death risk related genes. To
further explore the molecular signaling pathway related to behavioral
phenotypes, we introduce KEGG (Kyoto Encyclopedia of Genes and Genomes)
enrichment analysis for DEGs (Fig. [89]5d). A detailed KEGG enrichment
analysis of the KEGG signaling pathways appears in the Additional file
[90]1: Table 3. KEGG enrichment analysis showed that “Immune system”
related pathways were significantly enriched such as “Inflammatory
bowel disease (IBD)”, “Th17 cell differentiation”, “Hematopoietic cell
lineage”, “Cytosolic DNA-sensing pathway” and so on. These results were
in accordance with the results of the GO enrichment analysis of “Immune
response related biologic process” (Fig. [91]5c). On the other hand, we
observed that “Cell growth and death” related pathways were also
enriched such as “Cellular senescence” and “P53 signaling pathway”.
These pathways may involve in modulating the programmed cell death of
cKO mice [[92]18–[93]20]. Previous studies revealed that “MAPK
signaling pathway” involved in regulating anxiety and depression-like
behavior in mice [[94]21, [95]22]. Our observation also showed that
SENP2 ablation decrease anxiety-like behavior (Fig. [96]2a-o). Based
upon the enriched KEGG pathways, previous studies and our findings, the
“MAPK signaling pathway” could be one of a possible mechanism in
modulating anxiety-like behavior in cKO mice.
Fig. 5.
[97]Fig. 5
[98]Open in a new tab
GO functional and KEGG pathway enrichment analysis of DEGs. a-c
Separate GO enrichment analysis was carried out with “Goatools” using
Fisher’s exact test and Benjamini-Hochberg was used to multiple testing
corrections. The up-regulated and down-regulated genes were classified
into different functional categories according to the GO term
enrichment analysis for behavior related biological process, such as
“Locomotion”, “Learning and memory” related biologic processes were
enriched (a), cell death related biologic process, such as “Programmed
cell death” related biologic processes were enrichment (b), Immune
response related biologic process, such as “Immune response” related
biologic processes were enrichment (c). d KEGG signing pathway
enrichment analysis of differential expression genes. KEGG enrichment
analysis was performed with “KOBAS 2.0” using Fisher’s exact test and
Benjamini-Hochberg was used to multiple testing correction. The “Immune
system”, “Cell growth and death” related pathway significantly
enriched. Adjusted p value < 0.05 represents GO term or KEGG signing
pathway significantly enriched. Rich factor means the ratio of enriched
genes in background genes. The larger value of rich factor, the more
enrichment of GO term or KEGG signing pathway. Count indicates the
number of genes and “padj” means adjusted p value
Multiple gene expression changes associated with anxiety-related behavior
As the SENP2 cKO mice exhibit decreased anxiety-like behavior (Fig.
[99]2), we attempted to screen out the genes associated with
anxiety-like behavior from DEGs. However, to the best of our knowledge,
there are no available databases that record anxiety disorder related
genes. Thus, we decided to retrieve the 1033 DEGs using the PubMed
database. By searching for the literature, we identified that 17 genes
associated with anxiolytic-like effect on behavior are up-regulated
(Fig. [100]6a), and 9 genes associated with anxiety-like effect on
behavior are down-regulated (Fig. [101]6b). In addition, the 26 genes
of anxiety-like or anxiolytic-like effect on behavior have been further
confirmed by transgenic mice. Taken together, these genes may have an
important role in modulating the level of anxiety. The 17 genes
associated with anxiolytic-like effect on behavior, including that APOE
[[102]23], TSPO [[103]24], UCP2 [[104]25], MT1 [[105]26], MT2
[[106]27], AIM2 [[107]28], CNR2 [[108]29], LCN2 [[109]30], DLK1
[[110]31], CNTF [[111]32], MYD88 [[112]33], HDC [[113]34], ALOX5
[[114]35], TLR4 [[115]36], ROCK1 [[116]37], MOV10 [[117]38], DDC
[[118]39]. The 9 genes associated with anxiety-like effect on behavior,
including that LAMP5 [[119]40], KALRN [[120]41], CRHR1 [[121]42], CCK
[[122]43], SNAP25 [[123]44], OLFM2 [[124]45], FAAH [[125]46], IQSEC2
[[126]47], EMX1 [[127]48]. These data suggest that either the
anxiolytic-like effects gene up-regulated or anxiety-like effects gene
down-regulated could be a reduced level of anxiety. However, to
elucidate the detailed mechanism of decreased anxiety-like behavior,
further studies are needed to screen and determine from the 26
candidates and confirm its role in regulating the level of anxiety in
cKO mice.
Fig. 6.
[128]Fig. 6
[129]Open in a new tab
Multiple gene expression changes associated with anxiety-related
behavior. 26 genes were associated with anxiety or anxiolytic-like
effects on behavior based on 1033 DEGs. a 17 up-regulated genes
associated with anxiolytic-like effects on behavior. b 9 down-regulated
genes associated with anxiety-like effects on behavior. All data
presented as mean ± S.E.M.
Discussion
Animal models are extremely useful in establishing causality between
genetic mutations, synaptic changes, circuit dysfunctions and abnormal
behaviors, thereby aiding us in understanding the pathogenesis of
neurological and neuropsychiatric diseases [[130]49, [131]50]. SENP2 is
a member of the sentrin/SUMO-specific proteases (SENPs) family and
implicated in embryonic development [[132]6, [133]10], fatty acid
metabolism [[134]51], atherosclerosis [[135]52], and epilepsy [[136]8].
However, the function of SENP2 in CNS and its potential contribution to
the neuropathology remains unclear. In this study, we generated
conditional knockout of SENP2 in excitatory neurons in the postnatal
forebrain and determined that these cKO mice display comprehensive
behavioral phenotypes including hyperactivity, reduced anxiety-like
behavior, impaired learning and memory (Figs. [137]2 and [138]3).
Consistently, RNA-seq results showed that the loss of SENP2 is
associated with moderate changes in gene transcripts related to
“locomotion”, “learning and memory” (Fig. [139]5a) and multiple gene
expression changes associated with anxiety-related behavior (Fig.
[140]6). In addition, GO enrichment analysis identified changes in
genes related to “Cell death” and “Immune response” (Fig. [141]5b-c).
Taken together, our results demonstrate that SENP2 plays important
functional roles in the forebrain.
SUMOylation is a dynamic and reversible posttranslational protein
modification that regulates the functions of target proteins [[142]53].
In the CNS, neuron-specific SUMO1–3 knockdown mice show less
exploration of center area in open field test [[143]12]. Additionally,
previous studies showed that hippocampus-dependent learning and memory
is impaired by overexpression of a dominant negative Ubc9 peptide in
the hippocampal CA1 area [[144]13]. These results suggested that the
balance of SUMOylation/deSUMOylation is critical for forebrain mediated
function, including hippocampus-dependent learning and memory, and
anxiety-like behaviors. SENPs catalyze the removal of SUMO from
SUMO-conjugated proteins, thus playing a critical role in regulating
the SUMOylation level of targets [[145]2, [146]3]. Previous study
revealed that neuron-specific SENP2 knockout mice display hyperactivity
and sudden death [[147]8]. Consistently, we report here that the
forebrain excitatory neuron-specific SENP2 knockout mice displayed
hyper-locomotor activity in the open field test. Moreover, using
several different behavioral assays, we identified a reduced anxiety
phenotype in the SENP2 cKO mice (Fig. [148]2j-o).
Previous study has reported that death of matured neurons in the
forebrain increases the level of anxiety [[149]54]. To further explore
the molecular mechanism related to behavioral phenotypes of cKO mice,
we conducted RNA-seq on the cortex of cKO animals and observed that
numerous DEGs related to programmed cell death biologic processes were
enriched (Fig. [150]5b), suggesting that ablation of SENP2 may lead to
neuronal death. However, we don’t yet have any direct evidence showing
that conditional knockout of SENP2 led to programmed cell death in the
forebrain excitatory neurons. Further study is certainly needed to
investigate the potential linkage between neuronal death and
anxiolytic-related behaviors in the SENP2 cKO mice. On the other hand,
we identified 26 genes from 1033 DEGs that are implicated in
anxiety-related behaviors. These include 17 up-regulated genes known to
have anxiolytic-like effects and 9 down-regulated genes have
anxiety-like effects (Fig. [151]6a-b). This finding thus should shed
some light on the potential molecular mechanism for the anxiolytic
behaviors exhibited by the SENP2 cKO mice.
The SENP2 cKO mice also exhibited deficits in associative learning and
spatial working memory functions, which were assessed in both
contextual fear conditioning test and Y-maze spontaneous alternation
task respectively (Fig. [152]3a-b). Consistent with the observed
cognitive impairments, the RNA-seq studies we conducted on the cortex
of cKO animals detected 20 DEGs that are specifically related to
learning and memory-related tasks (Fig. [153]5a). These findings
suggest that SENP2-dependent protein modifications are important for
learning and memory, and dysfunction of SENP2 contributes to impaired
cognitive functions in mice. Moreover, GO analysis of the DEGs in the
SENP2 cKO mice revealed enrichment of genes for several cellular and
molecular functional categories, including those related to “Cell
death” and “Immune response” (Fig. [154]5b-c). Given that SENP2 has
been implicated in neural disorders such as epilepsy and
neurodegeneration [[155]7, [156]8], future studies should be directed
to examine the role of these SENP2-regulated molecular pathways in the
observed behaviors.
As the subcellular localization of SENP2 is predominantly nuclear, we
performed the differential analysis of gene and transcript expression
using high throughput RNA-seq. These analyses suggest that SENP2
knockout results in changes of gene transcription that are associated
with various behaviors including locomotion, learning and memory,
anxiety. One interpretation of these results is that SENP2 dependent
post translational modification of proteins, such as transcription
factors and co-factors, may lead to alterations in gene expression.
However, it is unclear whether these changes in gene transcription are
directly caused by the loss-of-function of SENP2. On the other hand, it
is very likely that SENP2 can directly regulate neuronal functions by
altering the status of SUMOylated proteins. We hope that, in future
studies, different proteomics approaches will help further gain insight
into the role of SENP2 plays in the CNS.
In summary, we generated a line of cKO mice to investigate the
functions of SENP2 in the CNS. Our results demonstrate that dysfunction
of SENP2 in forebrain excitatory neurons leads to behavioral changes
associated with emotional and cognitive functions.
Materials and methods
Experimental animals
All animal procedures were carried out in accordance with the
guidelines for care and use of laboratory animals of Shanghai Jiao Tong
University School of Medicine and approved by the Institutional Animal
Care and Use Committee (IACUC). All mice used in this study were
C57BL/6 background. SENP2^fl/fl mice [[157]8] were crossed with
CaMKIIα-Cre mice [[158]11] to generate SENP2 conditional knockout mice
(cKO). PCR primers used for genotyping are the following: SENP2 loxP
Forward: 5′-CTTCTGCTTCTCTTAGTGCT-3′ and SENP2 loxP Reverse:
5′-CTTCTGCTTCTCTTAGTGCT-3′, with the expected product sizes of 183 and
149 bps for SENP2^fl/fl and WT mice respectively. The presence of
CaMKIIα-Cre was identified by PCR using the following primers: Cre
Forward: 5′-CGCTGGGCCTTGGGACTTCAC-3′ and Cre Reverse:
5′-CAGCATTGCTGTCACTTGGTC-3′, with the expected PCR product size of
403 bps.
Immunofluorescence staining
Immunofluorescence staining was performed as described previously
[[159]55], with minor modifications. Mice were perfused intracardially
with 4% paraformaldehyde. After an overnight postfixation in the same
fixative at 4 °C, brain tissues were embedded in 2% agar and cut into
50 μm sections with a vibratome (Leica, VT1200S). Brain sections were
blocked with 0.3% Triton X-100 and goat serum in PBS for 1 h at room
temperature and then incubated with SENP2 (1:100, Abcam, ab58418) and
CaMKIIα (1:400, ThermoFisher Scientific, MA1–048) primary antibody
overnight at 4 °C. Next, brain sections were incubated with Alexa
Fluor546-conjugated goat anti-rabbit (Invitrogen, 1:500, A-11035) or
Alexa Fluor488-conjugated goat anti-mouse secondary antibodies
(Invitrogen, 1:500, A-11029) for 1 h at room temperature and mounted on
glass slides using a small brush. The fluorescence images were acquired
using the confocal microscope (Leica, TCS SP8).
Western blot
Western bolt was performed as previously described with minor
modifications [[160]4]. Brain tissues were collected on ice and
homogenized with lysis buffer I (50 mM Tris-HCl, pH 6.8, 2% SDS, 40 mM
DTT, and 5% glycerol). After denaturation for 15 min at 95 °C, the
samples were diluted 10-fold with lysis buffer II (50 mM Tris-HCl,
pH 7.4, 150 mM NaCl and 1% NP-40) and ultra-sonicated for 5 s, followed
by centrifugation for 10 min at 13,000×g (4 °C). The supernatant was
transferred to a new tube and boiled with loading buffer for 15 min.
The proteins were separated by SDS-PAGE, transferred to polyvinylidene
fluoride (PVDF) membranes, blocked with 5% non-fat milk, and
immunoblotted with SENP2 (1:1000, Santa Cruz, sc-376,731) and β-actin
(1:10000, Santa Cruz, sc-130,065) antibodies.
Real-time quantitative polymerase chain reaction (qPCR)
qPCR was performed as previously described with minor modifications
[[161]5]. Brain tissues (cortex, hippocampus and cerebellum) were
isolated on ice from cKO and littermate control mice. The total RNAs
were extracted from these brain tissues using TRIzol Reagent (Tiangen,
Beijing, China). After removing genomic DNAs, the total RNAs were
reverse transcribed to complementary DNA (cDNA) using the PrimeScript
RT reagent Kit (Takara, Dalian, China) according to the manufacturer’s
protocol. The cDNAs were then used as templates for the qPCR reactions,
which were performed in a 10 μl volume with Power SYBR Green PCR Master
Mix (CWBIO, Beijing, China) and 0.2 μM primers using the LightCycler
480 real-time PCR system (Roche, CA, USA). The qPCR primers used for
cKO and littermate control mice were as follows: SENP2 forward:
5′-TTCTCGGCACCATTCTTCGCTTGT-3′, SENP2 reverse:
5′-TGCTGCAGGATCCAGAACTCATCA-3′. GAPDH forward: 5′-CATGGCCTTCCGTGTTCC-3′
and GAPDH reverse: 5′-GCCTGCTTCACCACCTTCTT-3′.
RNA sequencing (RNA-seq)
RNA extraction: Three pairs of cortical tissue were isolated on ice
from 6-week-old cKO and littermate control mice and quickly frozen in
liquid nitrogen. The samples were delivered to the company (Majorbio
Bio-pharm Technology, Shanghai, Chain) to prepare RNA samples and
conduct high-throughput RNA-seq. The RNA quality was determined by 2100
Bioanalyser (Agilent Technologies, Tokyo, Japan) and quantified using
the ND-2000 (NanoDrop Technologies, DE, USA). The RNA integrity was
also confirmed with agarose gel electrophoresis.
Library construction and sequencing: 5 μg high-quality RNA samples were
used to construct sequencing library. Briefly, messenger RNA was
isolated from total RNA samples and then fragmented by fragmentation
buffer firstly. Secondly, double-stranded cDNA was synthesized using a
SuperScript double-stranded cDNA synthesis kit (Invitrogen, CA, USA)
with random primers (Illumina, CA, USA). Thirdly, the synthesized cDNA
was subjected to end-repair, phosphorylation and ‘A’ base addition
according to Illumina’s library construction protocol. Then, 200–300 bp
cDNA target fragments were isolated using 2% agarose followed by PCR
amplified, the isolated cDNA fragments were selected to construct
sequencing library. Libraries quantified by TBS380, and sequenced with
the Illumina HiSeq 4000 (Illumina, CA, USA).
Read mapping: The raw paired end reads were trimmed by “SeqPrep”
software and quality controlled by “Sickle software. Then clean reads
were separately aligned to reference genome with orientation mode using
“TopHat” software. The criteria for mapping were that sequencing reads
should be uniquely matched to the genome with less than 3 mismatches,
without insertions or deletions. Then, the gene regions were expanded
following depths of sites and the operon was obtained. In addition, the
whole genome was segmented into multiple 15 kb fragments that share the
same 5 kb fragments. If more than 2 consecutive fragments were without
overlapped region and at least 2 reads mapped per fragment in the same
orientation, we consider it was new transcribed region of gene
[[162]56].
Data analysis: To identify differentially expressed genes (DEGs)
between cortical tissues of cKO and littermate control mice, each
transcript expression level was calculated according to the fragments
per kilobase of transcript per million mapped reads (FPKM) method. The
gene abundances were quantified using the RSEM software, while the
EdgeR software was used for differential expression analysis and
quantified transcript read counts. In addition, the differential
expression analysis was carried out on an online platform
([163]www.majorbio.com) followed by multiple check calibration (BH). A
differentially expressed gene is identified as relative read counts > 2
fold change and adjusted p value < 0.05. Similarly, Go enrichment
analysis was conducted using the “Goatools” software followed by
Fisher’s exact test. KEGG enrichment analysis was using the “KOBAS 2.0”
software followed by Fisher’s exact test. If the adjusted p value
< 0.05, we consider the GO terms or KEGG signing pathways were
significantly enriched.
Experimental design for behavioral tests
Behavioral tests were performed on male mice of 8 to 12-weeks of age.
Mice were housed in a room with 12 h light/dark circadian rhythm,
suitable temperature (22–28 °C), adequate water and food. The
behavioral tests were performed in the light-on phase of the cycle
(10:00 a.m.-18:00 p.m.) except for otherwise noted. The experimenters
were blinded to the genotype of each mouse during all tests and data
analyses. The tests were performed in the following sequence: open
field, elevated plus maze, novelty suppressed feeding, nest building, Y
maze, and contextual fear condition. The startle response/prepulse
inhibition test was performed using another set of male mice without
any stressors. Tests were repeated using at least two different cohorts
of mice. These tests were performed at intervals of 2–4 days.
Open field test
The open field test was performed as previously described with minor
modifications [[164]57]. Mice were habituated in the testing room for
60 min and then introduced to the open field apparatus
(40 × 40 × 30 cm) (MED Associates). The test mice were allowed to
freely explore for 30 min in apparatus without interference. The
distance traveled, number of entries in center area (20 × 20 cm), time
spent in center area and distance traveled in center area were
automated recorded by monitor system and software (EthoVision XT 12).
Elevated plus maze
Elevated plus maze test was performed as previously described
[[165]58]. Mice were habituated in the testing room for 60 min and
placed in the intersection of open and closed arms with the mouse head
toward the open arm. The test mice were allowed to freely explore for
5 min in the apparatus (MED Associates) without interference. Monitor
system and software (EthoVision XT 12) automated recorded the time that
mice spent in open and closed arms as well as the number of entries in
open and closed arms respectively.
Novelty suppressed feeding
Novelty suppressed feeding test was performed as previously described
[[166]59] with minor modifications. Briefly, mice were fasted for 24 h
in the home cage before testing and then placed in a new feeding box in
which food was fixed on one piece of round filter paper (10 cm in
diameter) at the center of the plastic chamber (30 × 40 cm). The mice
were allowed to freely explore and eat the food for 5 min. A monitor
system (EthoVision XT 12) recorded the process of food eating including
the latency to feeding.
Y maze
The Y maze has three identical opaque arms (40 cm-long, 10 cm-wide, and
15 cm-high). We conducted the Y maze test as previously described
[[167]60]. Briefly, mice were habituated the testing room for 60 min
and placed in the distal end of one arm. The mice were allowed to
freely explore for 5 min without any interference. The monitor system
(EthoVision XT 12) recorded the locus of mice movement, while the
sequence of entries was manually recorded and analyzed by the
experimenter. The alternation ratio was calculated as follows:
[MATH: Alternation(%)=Alternatednumbers/(Totalnumberofentriesin−2)x100%. :MATH]
Contextual fear condition
Contextual fear condition was conducted as previously described with
minor modifications [[168]61]. Briefly, mice were habituated in the
testing room for 60 min and then placed in a test chamber with a black
and white plaid sticker on all sides for 10 min to habituate the
apparatus (Ugo Basile) that constantly presents 100 lx bright light for
2 consecutive days. In the training session, mice were placed in the
test chamber with a metal grid floor and received footshock (0.5 mA,
2 s) for 5 trials with 120 s interval. After the footshock, mice
remained in the conditioning chamber for another 30 s and then were
placed back to their home cage for 24 h. In the retrieval session, mice
were placed into the same contextual chamber for 10 min test without
footshock. The freezing values were recorded every 2-min block by the
monitor system and software (ANY-maze).
Nest building
Nest building test was performed as previously described with minor
modifications [[169]62]. Briefly, mice were individually housed for
12 h (20:00 p.m.-8:00 a.m.) while providing a piece of cleansing paper
into the cage. The next day, the nesting score was assessed based on
the integrity of the paper by an experimenter blinded to mouse
genotypes [[170]63]. Nests were given a score of 0–5 according to the
following criteria: 1 = more than 90% of paper was intact; 2 = 50–90%
of paper remained intact; 3 = more than 50% of paper was torn, but no
identifiable nest site; 4 = more than 90% of paper was torn and a flat
nest was built; 5 = more than 90% of paper was torn and the paper was
transformed into a tridimensional nest.
Prepulse inhibition (PPI)
Prepulse inhibition test was carried out as previously described
[[171]64] with minor modifications. Mice were habituated to the testing
room for 60 min, and placed into the test chamber for 5 min to
acclimate the apparatus (MED Associates) that present the constant
background white noise of 65 dB for 2 consecutive days. In stage I of
the PPI session, we replaced the mice into the chamber for a 5 min
acclimation. In stage II of PPI session, we presented 10 trials of high
acoustic stimulus (120 dB) with an interval of 20 ms to make the mice
accommodate the high acoustic stimulus. In stage III of PPI session, we
random presented seven types of acoustic stimulus including: 1) the
high acoustic stimulus (120 dB) only or 2) the low acoustic stimulus
only (70 dB, 74 dB and 78 dB) or 3) the high acoustic stimulus paired
with a low acoustic stimulus (70 dB paired with 120 dB or 74 dB paired
with 120 dB or 78 dB paired with 120 dB) with a random interval of
30–100 ms. In our protocol, six blocks containing seven acoustic
stimulus types were presented in a pseudorandom order and each acoustic
stimulus type was presented once within a block. The acoustic reflex
was measured by SR-Lab system (San Diego Instruments) and counted using
the largest peaks of amplitude that the signal recorded in a 300 ms
window after presented the acoustic stimulus. We respectively counted
the startle amplitude of high acoustic stimulus (120 dB only) and the
paired acoustic stimulus (70 dB paired with 120 dB or 74 dB paired with
120 dB or 78 dB paired with 120 dB). The startle amplitude of 120 dB
stimulus represented the acoustic reflex of mice, and was used as the
baseline value of PPI. The PPI was calculated as follows:
[MATH: PPI(%)=(Startleamplitudeof120dBacousticstimulus−Startleamplitudeofpairedacousticstimulus)/(Startleamplitudeof120dBacousticstimulus)x100%. :MATH]
Statistical analysis
All data are presented as mean ± S.E.M and analyzed by GraphPad Prism8
software. Two-group comparison was processed with two-tailed, unpaired
Student’s t test when the variance is equal (Fig. [172]2e, g, i, l, o,
Fig. [173]3a, c) or Welch’s t-test when the variance is unequal (Fig.
[174]2c, k, m, n). Multiple group comparison was performed with two-way
ANOVA analysis followed by Bonferroni’s post-hoc to determine
significance (Fig. [175]1d-e, Fig. [176]2b, d, f, h, Fig. [177]3b, d).
*p < 0.05, **p < 0.01, ***p < 0.001 compared with littermate controls.
Data availability statement
The data that support the findings of this study are available from the
corresponding author upon reasonable request.
Supplementary information
[178]13041_2020_591_MOESM1_ESM.xlsx^ (319.1KB, xlsx)
Additional file 1: Table S1. List of DEGs (Differentially Expressed
Genes) from RNA-Seq analysis results. Table S2. List of GO term from
RNA-Seq analysis results. Table S3. List of KEGG signing pathway from
RNA-Seq analysis results.
Acknowledgments