bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
1 KIF2C regulates synaptic plasticity and cognition by mediating
2 dynamic microtubule invasion of dendritic spines
3 Rui Zheng1,2†, Yong-Lan Du1,2†, Xin-Tai Wang2, Tai-Lin Liao1,2, Zhe Zhang1,2, Na Wang1,2, Xiu-
4 Mao Li3, Ying Shen4, Lei Shi5, Jian-Hong Luo1,2*, Jun Xia6*, Ziyi Wang7*, Junyu Xu1,2*
5
6 1Department of Neurobiology and Department of Rehabilitation of the Children’s
7 Hospital, Zhejiang University School of Medicine, Hangzhou, 310058, China.
8 2NHC and CAMS Key Laboratory of Medical Neurobiology, Ministry of Education Frontier
9 Science Center for Brain Research and Brain Machine Integration, School of Brain Science and
10 Brain Medicine, Zhejiang University, Hangzhou, 310058, China.
11 3Department of Orthopedic Surgery, the Second Affiliated Hospital, Zhejiang University School
12 of Medicine, Hangzhou, 310058, China.
13 4Department of Physiology and Department of Neurology of First Affiliated Hospital, Zhejiang
14 University School of Medicine, Hangzhou, 310058, China.
15 5JNU-HKUST Joint Laboratory for Neuroscience and Innovative Drug Research, Jinan University,
16 Guangzhou, China.
17 6Division of Life Science and State Key Laboratory of Molecular Neuroscience, The Hong Kong
18 University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China.
19 7Innovative Institute of Basic Medical Sciences of Zhejiang University (Yuhang), Hangzhou,
20 310012, China.
21 †These authors contributed equally to this work.
22 *Corresponding author. Junyu Xu ([email protected]), Ziyi Wang ([email protected]), Jun Xia
23 ([email protected]) and Jian-Hong Luo ([email protected]).
24 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
25 Abstract
26 Dynamic microtubules play a critical role in cell structure and function. In nervous
27 system, microtubules specially extend into and out of synapses to regulate synaptic
28 development and plasticity. However, the detailed polymerization especially the
29 depolymerization mechanism that regulates dynamic microtubules in synapses is still
30 unclear. In this study, we find that KIF2C, a dynamic microtubule depolymerization
31 protein without known function in the nervous system, plays a vital role in the
32 structural and functional plasticity of synapses and regulates cognitive function.
33 Using RNAi knockdown and conditional knockout approaches, we showed that
34 KIF2C regulates spine morphology and synaptic membrane expression of AMPA (α-
35 amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid) receptors. Moreover, KIF2C
36 deficiency leads to impaired excitatory transmission, long-term potentiation, and
37 altered cognitive behaviors in mice. Mechanistically, KIF2C regulates microtubule
38 dynamics and microtubule invasion of spines in neurons by its microtubule
39 depolymerization capability in a neuronal activity-dependent manner. This study
40 explores a novel function of KIF2C in the nervous system and provides an important
41 regulatory mechanism on how microtubule invasion of spines regulates synaptic
42 plasticity and cognition behaviors.
43 Keywords: KIF2C, microtubule depolymerization, spine invasion, long-term
44 potentiation, cognition
2 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
45 Introduction
46 Microtubules (MTs) are eukaryotic cytoskeletons that play an important role in cell
47 function. In the central nervous system, MTs serve as a structural basis for trafficking
48 material throughout the neuron and are the primary regulators of neuronal
49 differentiation (1). The MTs dynamic instability, i.e. the growing or shrinking of MT
50 plus end, has been detected in both shaft and dendritic protrusions (2) and is involved
51 in spine formation and synaptic function (3-6). Inhibition of MT polymerization results
52 in deficits in LTP and negatively affects memory formation or retention (7, 8). The
53 abnormal MTs dynamics is associated with several neurodegenerative and psychiatric
54 diseases, including Alzheimer’s disease, Parkinson’s disease, schizophrenia, and
55 depression (9). During neuronal activities, microtubules can be regulated to extend from
56 the dendritic shaft into spines and affect synaptic structure and function (3-8), which
57 could be an important event for synaptic plasticity. However, regulatory mechanism
58 underlying microtubule dynamics in or out from synapse upon neuronal activity is still
59 unclear, nor its role in learning and memory.
60 The polymerization dynamics of MTs are essential for their biological functions
61 and can be regulated by many MT-associated proteins (MAPs) (10, 11). The kinesin-13
62 family of kinesin superfamily proteins (KIFs) has a distinct function in depolymerizing
63 MTs (12), and therefore is involved in many MT-dependent events (13, 14). KIF2C,
64 also called mitotic centromere-associated kinesin (MCAK), is the best-characterized
65 family member and is localized to the spindle poles, spindle midzone, and kinetochores
66 in dividing cells. KIF2C utilizes MT depolymerase activity to regulate spindle
67 formation and correct inappropriate MT attachments at kinetochores (15-17). KIF2C
68 uses ATP hydrolysis for MT depolymerization from both ends and directly interacts
69 with EBs to enhance its destabilizing ability (18-22). Particularly, KIF2C is associated
70 with Alzheimer’s disease and suicidality psychiatric disorders. In vitro study showed
71 that Oligomeric amyloid-β (Aβ) directly inhibits the MT-dependent ATPase activity of
72 KIF2C and leads to abnormal mitotic spindles, which may cause chromosome mis-
73 segregation and increased aneuploid neurons and thereby contributes to the
74 development of Alzheimer’s disease (23). Clinical evidence shows that KIF2C is
3 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
75 significantly decreased in suicidal ideation cohort and could be utilized as a reference
76 for predicting suicidality (24). However, the in vivo physiological function of KIF2C in
77 the central nervous system (CNS) remains to be elucidated.
78 In this study, we have found that KIF2C in the CNS is expressed in a brain-wide
79 and synaptic localized pattern, which could be regulated by neuronal activity. Using a
80 nervous system-specific KIF2C conditional knockout (cKO) mouse, we further
81 confirmed its pivotal role in spine morphology, synaptic transmission, and synaptic
82 plasticity. By overexpression of KIF2C wild-type (WT) and MT depolymerization-
83 defective mutation, we have found that KIF2C regulates synaptic plasticity by
84 mediating MT depolymerization and affecting MT synaptic invasion. Furthermore,
85 KIF2C cKO mice exhibit deficiencies in multiple cognition-associated tasks, which
86 could be rescued by overexpression of KIF2C WT but not the MT depolymerization-
87 defective mutant. Overall, our study explores a novel function of KIF2C in neurons and
88 synaptic plasticity, and reveals how MT dynamics in synapses upon neuronal activity
89 contributes to cognition.
4 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
90 Results
91 KIF2C is highly expressed in the brain and at the synapse
92 Previous studies have showed that KIF2C plays an important role during cell division
93 in mitotic cells (25-27); however, it is not clear whether KIF2C is also present in the
94 nervous system. Western blot analyses showed that KIF2C was expressed in multiple
95 regions of the brain (Figure 1A), and shows sustained high expression in the
96 hippocampus throughout the postnatal stages (Figure 1B). To identify KIF2C
97 abundance in neurons, western blot analyses were performed on whole-cell extracts of
98 cultured hippocampal neurons (Figure 1C). The results showed that KIF2C was
99 expressed in a development-related pattern in cultured neurons: detectable at day in
100 vitro 3 (DIV 3) and significantly increased at DIV 10. Moreover, the subcellular
101 fractionation analysis showed an enrichment of KIF2C in the postsynaptic density (PSD)
102 fraction (Figure 1D). Immunofluorescent staining of cultured neurons revealed KIF2C
103 expression as granular puncta of different sizes along dendrites overlapping with the
104 postsynaptic marker PSD-95 and presynaptic protein Bassoon (Figure 1E). These
105 results showed that KIF2C is abundantly expressed in CNS neurons and is enriched at
106 postsynaptic density, suggesting a possible function in synapses.
107 KIF2C knockdown affects spine morphology, synaptic transmission and long-
108 term potentiation in vitro and in vivo
109 To explore the function of KIF2C in synapses, KIF2C was knocked down in cultured
110 hippocampal neurons using viral-mediated KIF2C shRNA (shKIF2C) (Figure 2A) at
111 DIV 7–10 and the spine morphology was examined at DIV 16–18. The results showed
112 that KIF2C knockdown did not affect spine density (Figure 2B). Electrophysiological
113 recordings also showed similar frequency and amplitude of miniature excitatory
114 synaptic currents (mEPSCs) in control and knockdown neurons (Figure S1A),
115 suggesting that acute knockdown of KIF2C in cultured neurons does not affect basal
116 synaptic transmission. Thereafter, we investigated whether KIF2C regulates spine
117 morphology in an activity-dependent manner. The induction of long-term potentiation
118 (LTP) results in changes in spine structure(28-31). To test whether KIF2C is involved
5 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
119 in spine structural plasticity, we further conducted a glycine-induced chemical LTP
120 (cLTP) assay to produce synaptic enhancements that mimic LTP(32) in KIF2C
121 knockdown hippocampal neurons. Spine density did not change after cLTP stimulation
122 (Figure 2C, D). However, the spine head width significantly increased in control
123 hippocampal neurons, whereas it was not affected in the KIF2C knockdown neurons
124 after cLTP stimulation (Figure 2E), indicating that KIF2C is essential for spine
125 structural plasticity. In addition, both subcellular fractionation and
126 immunocytochemical results showed that cLTP triggered the translocation of KIF2C
127 from synaptic protrusions to the dendritic shaft (Figure 2F–I). Together, these data
128 indicate that KIF2C might be removed from postsynaptic sites by neuronal activity,
129 which may contribute to synaptic plasticity.
130 To further determine the effect of KIF2C on synaptic function, KIF2C was knocked
131 down in hippocampal CA1 neurons in vivo and the electrophysiological properties in
132 acute slices were measured. Mice were injected with either AAV-scramble shRNA
133 (control) or AAV-KIF2C shRNA (shKIF2C) in the CA1 region (Figure 3A), and
134 mEPSCs were recorded. Compared to the control, KIF2C knockdown neurons had
135 enhanced amplitude but similar frequency of mEPSCs (Figure 3B–D). This discrepancy
136 between cultured neurons and brain slices may result from activity-dependent synaptic
137 scaling in free-behaving mice after viral injection. In addition, the excitatory
138 postsynaptic current ratio of N-methyl-D-aspartic acid receptor (NMDAR)/α-amino-3-
139 hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) were decreased in
140 knockdown hippocampal slices (Figure 3E). Furthermore, we investigated whether
141 KIF2C regulates presynaptic function retrogradely by comparing paired-pulse ratios
142 (PPRs). KIF2C knockdown neurons exhibited comparable PPRs to the control,
143 suggesting that KIF2C had a minimal effect on contacting presynaptic function (Figure
144 3F). To test whether LTP is affected in KIF2C knockdown neurons, we measured
145 whole-cell LTP in CA1 pyramidal neurons by stimulating Schaffer collaterals. The
146 magnitude of LTP was significantly attenuated within 5 min in knockdown
147 hippocampal slices (Figure 3G, H). These results suggest that KIF2C is necessary for
148 synaptic plasticity in hippocampal neurons, and its insufficiency may affect the basal
6 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
149 function of NMDARs or AMPARs in the hippocampus.
150 KIF2C knockout affects spine morphology, synaptic transmission, and long-term
151 potentiation in vivo
152 To further investigate the functions of KIF2C in vivo, we generated cKO mice using the
153 CRISPR/Cas9 system (Figure 4A). To obtain cKO mice, KIF2Cflox/flox mice were mated
154 with Nestin-Cre transgenic mice, which resulted in KIF2C deletion only in the nestin
155 cell lineage (Figure S2A). The knockout efficiency was confirmed using a hippocampal
156 tissue RT-PCR assay (Figure S2B). There were no significant differences in body
157 weight or brain size between Nestin-Cre;KIF2CF/F (cKO) and KIF2CF/F (WT) mice
158 (Figure S2C). In addition, using Nissl staining, we found that the thickness of
159 hippocampal CA1 and CA3 regions did not change significantly in cKO mice (Figure
160 4B). Further examination of neuronal morphology of CA1 pyramidal cells by Golgi
161 staining and Sholl analysis showed no significant differences in dendritic complexity
162 or spine density between KIF2C cKO mice and WT mice (Figure 4C–E). However, the
163 density of mushroom-type spines decreased in basal and apical dendrites of
164 hippocampal neurons in cKO mice, whereas the density of filopodia-type spines
165 increased (Figure 4F). Furthermore, electron microscopy analysis revealed that the
166 length and thickness of the PSD fraction were remarkably reduced in the hippocampus
167 of cKO mice (Figure 4G, H). Mushroom spines are commonly considered mature
168 spines(33, 34); therefore, these results indicate impaired spine maturation in the
169 hippocampal CA1 region of cKO mice.
170 To determine whether such impaired spine maturation may affect synaptic
171 transmission and synaptic plasticity in cKO mice, mEPSCs were first examined in CA1
172 hippocampal pyramidal neurons. Consistent with the shRNA knockdown experiments
173 in CA1, the amplitude of mEPSCs increased in cKO mice compared to that in WT
174 controls (Figure 5A, B); whereas the frequency of mEPSCs was similar between the
175 two groups (Figure 5C). The increased mEPSC amplitude was not due to altered
176 presynaptic activity because PPRs were not affected in cKO neurons (Figure 5D),
177 indicating a postsynaptic influence in mEPSC property changes. Therefore, we
7 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
178 examined the NMDA/AMPA ratio of the CA1 neurons and found a decreased
179 NMDA/AMPA ratio in cKO mice (Figure 5E). In addition, we performed cell-surface
180 biotinylation experiments to assess possible changes in the surface expression of these
181 two receptors. The levels of surface AMPA receptors significantly increased in the
182 hippocampus of cKO mice, with NMDA receptors showing an increasing trend (Figure
183 5F), which may explain the increased mEPSC amplitude. We also compared long-term
184 plasticity in the hippocampal CA1 region between WT and cKO mice and found that
185 the LTP magnitude was significantly attenuated within 1 h in cKO hippocampal slices
186 (Figure 5G, H), while long-term depression was not altered (Figure 5I, J), suggesting
187 that KIF2C plays a critical role in LTP induction except for LTD.
188 KIF2C regulates synaptic transmission and plasticity by mediating dynamic
189 microtubule invasion of dendritic spines
190 Extensive studies on KIF2C in dividing cells have shown that KIF2C regulates
191 microtubule dynamics by mediating microtubule depolymerization (15-17). However,
192 dynamic microtubules are reported to be involved in regulating dendritic spine
193 morphology and synaptic plasticity (35). Therefore, KIF2C may regulate synaptic
194 plasticity through dynamic microtubules. To further explore this, we transfected EB3-
195 tdtomato in DIV 7–10 WT or cKO hippocampal neurons to mark the MT plus end and
196 observed the MT dynamics in mature neurons at DIV 18–20 using live-cell imaging.
197 The results showed that the speed of EB3 was faster in cKO dendrites (WT: 8.9 ± 0.8
198 μm / min, cKO: 12.4 ± 0.5 μm / min) (Figure 6A, B), indicating an increased rate of
199 MT polymerization in cKO mice. Using the plusTipTracker software, which is designed
200 to track MT growth parameters(36), we analyzed the polymerization properties of the
201 EB3-labeled MT plus ends in dendrites. Depletion of KIF2C significantly increased MT
202 plus-end growth speed (WT: 18.1 ± 0.7 μm / min, cKO: 32.3 ± 1.4 μm / min); however,
203 it decreased the growth lifetime (WT: 10.2 ± 0.4s, cKO: 6.4 ± 0.1s) (Figure 6B). MT
204 plus end growth length was not significantly altered (WT: 2.7 ± 0.1 μm, cKO: 2.9 ± 0.1
205 μm) (Figure 6B). These results suggest that KIF2C regulates MT depolymerization in
206 hippocampal neurons and that MT dynamics become disordered with KIF2C deficiency.
8 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
207 MTs remain dynamic in mature neurons and are capable of invading dendritic
208 protrusions. MT invasion of dendritic protrusions depends on neuronal activity(37). To
209 observe MT invasion, we used 50 ng/mL brain-derived neurotrophic factor (BDNF) to
210 induce spine plasticity, and labeled newly polymerized MTs with an antibody against
211 tyrosinated-α-tubulin. In the basal state, invading MTs were present more in the
212 dendritic spines of cKO neurons (Figure 6C), suggesting increased MT invasion events
213 in cKO synapses. After BDNF treatment, MT invasion robustly increased in WT
214 neurons; however, no change was observed in cKO (Figure 6C). Based on our previous
215 result showing KIF2C translocation from synapses to dendrites upon neuronal activity
216 (Figure 2F–I), these results indicate that KIF2C is critical for activity-dependent
217 microtubule invasion of spines. Synapses deficient in KIF2C have a moderate increase
218 in basal microtubule invasion; however, they lose their ability to respond to neuronal
219 activities due to the lack of further robust MT dynamics.
220 In mitotic cells, KIF2C binds to the end of MTs and couples ATP hydrolysis to initiate
221 MT depolymerization(38). To further confirm whether the MT depolymerization
222 property of KIF2C is critical for regulating synaptic transmission and plasticity, we
223 constructed an MT depolymerization loss-of-function KIF2C mutant (KIF2C G491A).
224 Its human homologous KIF2C G495A mutant has been previously reported to lose ATP
225 hydrolyzing and hence MT depolymerization abilities (20, 39). The KIF2C(G491A)
226 mutant was first tested in HEK 293T cells for its MT depolymerization function.
227 Immunocytochemical analysis showed that KIF2C(G491A) transfected cells had intact
228 cell morphology, while the KIF2C(WT) successfully depolymerized MTs and
229 maintained cell rounding (Figure S3A). Furthermore, we expressed KIF2C(WT) or
230 mutant in the hippocampal CA1 region of cKO mice by AAV-mediated gene delivery
231 (Figure S3B, C) and examined synaptic transmission and plasticity using an
232 electrophysiological approach. The results showed that re-expression of KIF2C(WT)
233 significantly reduced the elevated amplitude of mEPSCs (Figure 6D–F) and restored
234 LTP in cKO mice (Figure 6G, H). However, without MT depolymerization activity, the
235 KIF2C (G491A) mutant failed to show significant differences in mEPSC properties
236 (Figure 6I–K) or LTP induction in the cKO CA1 neurons (Figure 6L, M). These findings
9 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
237 indicate that KIF2C regulates synaptic transmission and plasticity by mediating
238 dynamic microtubule invasion of dendritic spines.
239 KIF2C plays an important role in CA1-dependent cognitive behaviors
240 KIF2C deficiency leads to abnormalities in synaptic transmission and synaptic
241 plasticity; we speculated whether KIF2C plays a role in higher-order brain function.
242 Therefore, we examined KIF2C cKO mice for several behavioral tests. The KIF2C cKO
243 mice showed normal motion and anxiety levels in the open field test and elevated zero
244 maze compared to those of the WT mice (Figure S4A, B), and had normal grooming
245 frequencies in the commonly used repetitive behavioral test (Figure S4C). The Morris
246 water maze task also elicited similar learning curves and probe performance in both
247 KIF2C cKO and WT mice (Figure S4D), suggesting normal spatial learning and
248 memory ability in mice with KIF2C deficiency. However, when we subjected the mice
249 to the Y-maze test, KIF2C cKO mice showed significant impairment in spontaneous
250 alternation behavior compared to that of the WT mice (Figure 7A), indicating that
251 KIF2C cKO mice were deficient in short working memory, which could be due to the
252 special role of CA1 in temporal processing (40-42).
253 To investigate whether this learning deficit would translate to other cognitive
254 disorders, we examined the mice for cued conditioned fear memory and social memory.
255 The conditioned fear learning test showed that KIF2C cKO mice displayed a
256 comparatively weaker freezing response to the context and tone conditioned stimulus
257 (CS) during the acquisition stage (Figure 7B). We monitored the conditioned freezing
258 response to the background context and CS until 14 or 15 days after acquisition,
259 respectively, and found that cKO mice still showed less freezing compared to WT mice,
260 particularly in context-dependent freezing (Figure 7C). These results suggest that the
261 maintenance of fear memory in KIF2C cKO mice was impaired. In a prepulse inhibition
262 test, WT and cKO mice showed the same extent of inhibition as the prepulse (Figure
263 S4E), showing that the cKO mice had normal sensorimotor gating. In the three-
264 chambered social test, both WT and KIF2C cKO mice exhibited no position preference
265 in habituation stage (Figure S4F). KIF2C cKO mice exhibited a normal preference for
10 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
266 the chamber in which a stranger mouse (S1) was present, compared with the empty
267 chamber (Figure 7D). However, with the introduction of a second stranger (S2), KIF2C
268 cKO mice exhibited no preference for the novel stranger (S2), compared with the
269 familiar mouse (S1) (Figure 7E). These results showed that KIF2C cKO mice showed
270 normal sociability but significantly reduced social novelty, indicating that KIF2C is
271 essential for social memory maintenance.
272 Based on the behavioral tests, we found that KIF2C plays an important role in CA1-
273 mediated cognitive and memory functions. The deficiencies observed in the Y-maze
274 test, cued conditioned fear memory, and three-chambered social test of KIF2C cKO
275 mice were highly likely due to the lack of LTP in the CA1 region (Figure 5G, H). As
276 re-expression of KIF2C in the CA1 region restored LTP (Figure 6G, H), we speculated
277 whether such CA1-specific rescue could also ameliorate the behavioral abnormalities
278 in KIF2C cKO mice. In our pretests, over-expression of KIF2C in mice for a long time
279 would lead to neuronal death (data not shown). Therefore, we re-expressed KIF2C in
280 the cKO mice CA1 region using a doxycycline-induced viral expression system (Figure
281 S4G) and examined the rescuing effects in the Y-maze test and three-chambered social
282 test. The results showed that cKO mice with KIF2C(WT) overexpression displayed a
283 normal alternation in the Y-maze test compared to cKO mice with control virus (Figure
284 7F). Furthermore, cKO mice with KIF2C(WT) overexpression also showed normal
285 interaction with the stranger mouse in the social novelty test (Figure 7G, H). In addition,
286 we tested whether the expression of the functional-null mutant KIF2C(G491A) in the
287 CA1 region of cKO mice would have any rescue effect. The results showed that cKO
288 mice injected with KIF2C(G491A) virus still had impaired working and social memory
289 (Figure 7I–K). Therefore, these results suggest that KIF2C is critical for cognitive
290 function, and this functional role is mediated by its MT depolymerization ability in
291 synapses during LTP.
11 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
292 Discussion
293 KIF2C is an ATP-dependent microtubule depolymerase that regulates cell division in
294 many aspects of mitosis, such as spindle assembly, MT dynamics, correct kinetochore-
295 MT attachment, and chromosome positioning and segregation(38). However, its
296 function in the CNS remains unclear. This study elucidated the functions of KIF2C in
297 regulating spine morphology, synaptic transmission, and plasticity in hippocampal
298 neurons and its critical role in CA1-mediated cognitive behaviors. We demonstrated
299 that the mechanism mediated by KIF2C is highly correlated with its known function in
300 MT depolymerization and MT invasion of dendritic spines. This study expands the
301 understanding of the function of KIF2C in non-mitotic cells and also unveils its critical
302 role in the CNS.
303 The stabilization of dendritic spines is an important structural basis for synaptic
304 plasticity. We found that both acute knockdown or genetic deficiency of KIF2C
305 eliminated LTP except LTD expression. On a molecular basis, KIF2C in knockdown or
306 cKO hippocampal CA1 neurons exhibited a higher AMPAR-mediated mEPSC
307 amplitude and an increase in surface AMPARs expression, with a trend of increase in
308 NMDARs. It is possible that increased expression of ionic glutamate receptors,
309 particularly AMPARs, on synapses leads to LTP occlusion. It would be interesting to
310 further explore the mechanism of KIF2C-involved receptor trafficking or synaptic
311 expression. As important architectural elements in neurons, MTs are also responsible
312 for intracellular transport. It has been reported that NMDA receptors can be transported
313 by KIF17 (43) and AMPA receptors can be transported by KIF5 along MTs in dendrites
314 (44). It is possible that KIF2C regulates membrane receptor trafficking by mediating
315 MT dynamics through the regulation of MT stability. We observed microtubule
316 dynamics using live cell imaging of KIF2C cKO hippocampal neurons. We observed
317 that depletion of KIF2C induced faster, shorter-lived microtubule growth in dendrites.
318 KIF2C deficiency also increased the frequency of MT invasion under basal conditions,
319 which was probably due to the loss of MT polymerization suppression. Therefore, the
320 increased AMPA receptor expression might be due to increased MT invasion, which
321 brings AMPA receptors into synapses (Figure 8). Another hypothesis is that such
12 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
322 changes in MT polymerization kinetics might affect the stability of AMPA receptor-
323 containing vesicles on MT structures (Figure 8). It might cause the release of vesicles
324 from MTs in the dendrite shaft, which may be distributed into spines by membrane
325 infusion and later diffusion, similar to a previous report on the MT-dependent
326 trafficking and synaptic entry of synaptotagmin-IV (45).
327 Activity-dependent MT invasion of dendritic spines becomes disabled after BDNF
328 stimulation in cKO hippocampal neurons. Our findings reveal that during cLTP
329 induction KIF2C was removed from the synapses (Figure 2F). Although the underlying
330 mechanism is unclear, translocation of KIF2C from synapse to dendrite or depletion of
331 synaptic KIF2C during synaptic activation is found to be required to ensure robust MT
332 invasion into spines and where and when MT depolymerization events should be
333 inhibited. Even though the genetic depletion of KIF2C increases the basal MT invasion
334 into spines, it largely prevents further increase in MT invasion during LTP, resulting in
335 synaptic plasticity failure and, finally, cognitive inflexibility (Figure 8). This hypothesis
336 was further supported by our rescue experiments using the KIF2C (G491A) mutant, in
337 which ATP hydrolyzing and MT depolymerization capabilities are abolished(20). Re-
338 expression of KIF2C (WT), except for KIF2C (G491A) mutant, rescued the mEPSC
339 hyper-amplitude in KIF2C cKO mice. KIF2C (G491A) also failed to restore the
340 impaired LTP or cognitive function in KIF2C cKO mice during working memory or
341 social memory tests as in the WT mice. Therefore, the MT depolymerization function
342 of KIF2C plays a critical role in regulating synaptic plasticity and cognitive behaviors.
343 Interestingly, Oligomeric amyloid-β directly inhibits the MT-dependent ATPase
344 activity of KIF2C in vitro, and thereby leads to the generation of defective mitotic
345 structures which may contributes to the development of Alzheimer’s disease (23). Aβ
346 stabilizes MT by reducing catastrophe-frequency (46), which is similar to KIF2C
347 conditional knockout phenotype. Furthermore, KIF2C was recently discovered as a
348 blood biomarker for suicidal ideation in psychiatric patients, such as patients with
349 schizophrenia (24).Therefore, KIF2C could be a therapeutic target for Alzheimer’s
350 disease or suicidality psychiatric disorders. Nevertheless, whether KIF2C plays other
351 important role in higher-order brain functions remain to be explored.
13 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
352 Materials and Methods
353 Generation of KIF2C flox/flox mice and animal maintenance
354 KIF2Cflox/flox mouse was generated via CRISPR/Cas9 system in the Model Animal
355 Research Center of Nanjing University (Nanjing, China). In brief, Cas9 mRNA, sgRNA
356 and donor were co-injected into zygotes. sgRNA-5’ end: 5’- TCC ACT GTG GAA TGG
357 TCT AA -3’; sgRNA-3’ end: 5’- CTG GTT GTG AGA CAC CTG TA-3’. Mouse kif2c
358 gene is composed of 20 exons. The sgRNA directs Cas9 endonuclease cleavage in
359 intron 1-2 and intron 8-9 and create a DBS (double-strand break). DBS breaks were
360 repaired by homologous recombination, and resulted in LoxP sites inserted in intron 1-
361 2 and intron 8-9 respectively. The mouse pups were genotyped by PCR and Southern
362 blot followed by sequence analysis. Conditional knockout mice (KIF2Cflox/flox; Nestin-
363 Cre) were obtained by crossing KIF2Cflox/flox mice with Nestin-Cre mice (Wang et al.,
364 2019). The resulting offspring were genotyped by PCR of genomic DNA. The primers
365 information: KIF2C floxP fragment, F: 5’-GGT CCA GCT CTT TAC TGA TGT GTT
366 C -3’; R: 5’-ACA AAG CAA GTC CAG GTC CAA G-3’; Nestin-Cre, F: 5’-TGC AAC
367 GAG TGA TGA GGT TC-3’; R: 5’- GCT TGC ATG ATC TCC GGT AT-3’. Mice were
368 kept under temperature-controlled conditions on a 12:12 h light/dark cycle with
369 sufficient food and water ad libitum. Experiments were done in male littermates at the
370 age of 1~3 month. All animal experiments were performed in accordance with the
371 ethical guidelines of the Zhejiang University Animal Experimentation Committee, and
372 were in complete compliance with the National Institutes of Health Guide for the Care
373 and Use of Laboratory Animals.
374
375 Antibodies and reagents
376 Antibody to KIF2C (12139-1-AP, RRID: AB_2877829) was purchased from
377 Proteintech (Rosemont, IL). The anti-PSD95 (ab2723, RRID: AB_303248), anti-MAP2
378 (ab5392, RRID: AB_2138153) and anti-EB3 (ab157217, RRID: AB_2890656) were
379 purchased from Abcam (Cambridge, UK). The anti-α-tubulin, tyrosinated (MAB1864-
380 I, RRID: AB_2890657), anti-synaptophysin (MAB368, RRID: AB_94947), anti-
381 GAPDH (MAB374, RRID: AB_2107445), anti-GluA1(04-855, RRID: AB_1977216)
14 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
382 and anti-GluA2 (MAB397, RRID: AB_2113875) were purchased from Millipore
383 (Massachusetts, USA). The anti-GluN2A (4205, RRID: AB_2112295), anti-GluN2B
384 (4207, RRID: AB_1264223), anti-GluN1 (5704, RRID: AB_1904067) were from Cell
385 Signaling Technology (Danvers, MA). The anti-flotillin was from BD (610820, RRID:
386 AB_398139). The anti-β-tubulin (sc-5274, RRID: AB_2288090) was from Santa Cruz
387 (Dallas, TX). The anti-bassoon was from Enzo Life Sciences (ADI-VAM-PS003-F,
388 RRID: AB_11181058). The phalloidin was from Yeasen (40734ES75, Shanghai, China).
389 Horseradish peroxidase-conjugated secondary antibodies for immunoblotting were
390 from Jackson and for immunostaining were from Invitrogen (Carlsbad, CA).
391 Dulbecco’s modified Eagle’s medium (DMEM), Neurobasal medium, B27 were from
392 Gibco. GlutMax, 4',6-diamidino-2-phenylindole (DAPI), and Alexa Fluor-conjugated
393 secondary antibodies were from Invitrogen (Carlsbad, CA). Nissl was from Beyotime
394 (Shanghai, China). Protease inhibitor cocktail was from Merck Chemicals. Other
395 chemicals were from Sigma (St. Louis, MO) unless stated otherwise.
396
397 qRT-PCR
398 For single-cell analysis, the tip of a conventional patch-clamp pipette was placed tightly
399 on the soma of a selected individual CA1 neuron. Gentle suction was applied to the
400 pipette. After complete incorporation of the soma, the negative pressure was released
401 and the pipette was quickly removed from the bath. The harvested contents were
402 subjected to qRT-PCR using OneStep Kit (Qiagen, Hilden, Germany).
403 For tissue analysis, the hippocampus tissue RNA was extract by using RNeasy®
404 Mini Kit (Qiagen, Hilden, Germany). In brief, hippocampus tissue was homogenized
405 in buffer RLT. The lysate was centrifuged at 12,000× g (4°C for 3 min). The supernatant
406 was transferred and 1 volume of 70% ethanol was added. Then RNA was absorbed on
407 RNeasy® Mini spin column and purified by washing with buffer RW1 and RPE. The
408 harvested contents were subjected to Reverse Transcription Kit (Vazyme). The qPCR
409 reaction was performed on HiScript® II 1st Strand cDNA Synthesis Kit (R212-02)
410 (Vazyme, Nanjing, China) using ChamQ™ Universal SYBR® qPCR Master Mix
411 (Vazyme, Q711). Primer information: GAPDH: F 5’- AGG TCG GTG TGA ACG GAT
15 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
412 TTG-3’; R 5’- TGT AGA CCA TGT AGT TGA GGT CA-3’. KIF2C: F 5’-TGC CGT
413 TGT TGA TGG TCA GTG-3’; R 5’-GGA GAC ACT TGC TGG GAA CAG-3’.
414 shKIF2C: F 5’-TGG ATC GAA GGA GGT ACC AC-3’; R 5’- CAC TGA CCA TCA
415 ACA ACG GCA-3’. The following parameters were used: 95°C for 30s, followed by
416 39 cycles of 95°C for 10s, 60°C for 30s in accordance with the manufacturer’s protocol.
417
418 Hippocampal neuron culture and transfection
419 Primary cultures of hippocampal neurons were prepared from embryonic day 18 mice.
420 The isolated hippocampi were dissociated with 2.5% trypsin at 37°C for 20 min. The
421 neurons were quantified and centrifuged at 1,800 rpm for 6 min and then plated on the
422 glass coverslips pre-coated with poly-L-lysine (Sigma). The culture medium was
423 Neurobasal (Gibco) supplemented with 2% B27 (Gibco), GlutMax (Invitrogen) and 1%
424 Penicillin-Streptomycin in a 5% CO2 atmosphere at 37°C. Neurons were transfected at
425 DIV 7-10 using a modified calcium phosphate protocol(47). More than 10 days after
426 transfection, neurons were directly subjected to live imaging under a Nikon A1R
427 confocal scanning microscope.
428
429 DNA constructs
430 The shRNA targeting sequence is designed against position 820-838 of mouse KIF2C
431 (Gene ID: 73804) open reading frame (5’-CCGGATGATCAAAGAATTT-3’), using
432 Bgl Ⅱ/SalⅠ restriction enzymes. KIF2C(WT) uses the Nhe Ⅰ/BamH Ⅰ restriction enzymes.
433 KIF2C point mutations were replaced a G to A at 491 position, using the Nhe Ⅰ/BamH
434 Ⅰ restriction enzymes. Plasmid synthesis and virus construction were completed by the
435 company (OBIO, Shanghai, China). KIF2C shRNA was sub-cloned into AAV backbone
436 pAKD-CMV-bGlobin-mcherry-H1-shRNA. KIF2C (WT) was sub-cloned into AAV
437 backbone pAAV-CMV-MCS-3FLAG-P2A-mNeonGreen-CW3SL and rTTA-TRE3G-
438 mcherry (BrainVTA, Wuhan, China). KIF2C (G491A) was sub-cloned into AAV
439 backbone pAAV-CMV-MCS-3FLAG-P2A-mNeonGreen-CW3SL (OBIO, Shanghai,
440 China).
441
16 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
442 Adeno-associated virus microinjection
443 For virus microinjection, WT or KIF2C cKO mice were randomly allocated to
444 experimental groups and processed. A small craniotomy was performed after
445 anesthetizing the mouse. The flow rate (40 nl/min) was controlled by a micro-injector
446 (World Precision Instruments). Virus was microinjected in both left and right
447 hippocampal CA1 region (from bregma, 3 weeks: ML: 1.45 mm; AP: 1.8 mm; DV: 1.5
448 mm; 2 months: ML: 1.65 mm; AP: 1.82 mm; DV: 1.62 mm). For electrophysiology, 4–
449 6-week-old mice were subjected for experiments at least 3 days after injection. For
450 KIF2C overexpression mice subjected for behavioral tests, mice were injected and bred
451 until 2–3-month-old, then took water containing (40 μg/ml) doxycycline for 3 days to
452 induce KIF2C expression before tests.
453
454 Glycine-induced LTP
455 Hippocampal neurons were infected by shRNA virus at DIV 7-10, and subjected for
456 chemical LTP induction at DIV 16-18. Briefly, neurons were firstly incubated in
457 extracellular solution (ECS; in mM: 140 NaCl, 2 CaCl2, 5 KCl, 5 HEPES, 20 glucose)
458 supplemented with 500 nM TTX, 1 μM strychnine, 20 μM bicuculline (Buffer A; pH
459 7.4) at 37°C for 5 min, and treated with 200 μM glycine in Buffer A for 10 min. Then
460 neurons were incubated in Buffer A for another 5-20 min. The cells were then harvested
461 for experiments(32, 48, 49).
462
463 Western blots
464 After determining protein concentration with BCA protein assay (Bio-Rad), equal
465 quantities of proteins were loaded and fractionated on SDS-PAGE and transferred to
466 NT nitrocellulose membrane (Pall Corporation), immunoblotted with antibodies, and
467 visualized by enhanced chemiluminescence (Pierce Biotechnology). Primary antibody
468 dilutions used were GluA1 (1:1,000), GluA2 (1:2,000), GluN1 (1:2,000), GluN2A
469 (1:2,000), GluN2B (1:2,000), flotillin (1:1,000), KIF2C (1:500), EB3 (1:10,000),
470 PSD95 (1:1,000), β-tubulin (1:2,000), GAPDH (1:10,000), and secondary antibodies
471 (1:10,000). Film signals were digitally scanned and quantitated using Image J software.
17 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
472
473 Immunocytochemistry
474 DIV 18-20 hippocampal neurons were fixed in 4% paraformaldehyde (PFA) containing
475 4% sucrose for 20 min, and then treated with 1% Triton X-100 in PBS for 10 min at
476 room temperature (RT). After washing with 1x PBS, neurons were transferred into
477 blocking solution (10% normal donkey serum, NDS in PBS) for 1 h at RT. Neurons
478 were then incubated with primary antibodies at 4°C overnight and incubated with
479 secondary antibodies for 3 h at RT. The samples were subjected to confocal laser
480 scanning microscope Nikon A1R. Primary antibody dilutions used for
481 immunocytochemistry were PSD95 (1:1,000), KIF2C (1:500), bassoon (1:1,000),
482 tyrosinated α-tubulin (1:500), and MAP2 (1:10,000), and secondary antibodies
483 (1:1,000). All antibodies were diluted in 1x PBS containing 3% NDS. Measures of spine
484 head width and spine density in cultured hippocampal neurons were performed using
485 Imaris software and colocalization analysis was performed by Metamorph software.
486
487 Immunohistochemistry
488 Mice were anesthetized and transcranially perfused with 4% PFA. Coronal slices with
489 30 μm thickness were prepared and placed in blocking solution for 1 h at RT. After
490 washing with 1x PBS, slices were incubated with primary antibodies overnight at 4°C
491 and incubated with secondary antibodies for 3 h at RT. Images were acquired using an
492 automated fluorescence microscope Olympus BX53.
493
494 Nissl staining
495 Nissl staining was performed using Nissl staining Kit (Beyotime, C0117). Coronal
496 slices (30 μM thickness) were immersed in Nissl staining solution for 10 min at 37°C,
497 rinsed with ddH2O, dehydrated in ethanol, and cleared in xylene. Images of cortex and
498 hippocampus were captured using an Olympus VS120 microscope. Hippocampus CA1
499 and CA3 thickness was calculated by Image J software.
500
501 Golgi staining
18 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
502 Golgi-Cox staining of hippocampus CA1 tissue was performed using the FD Rapid
503 GolgiStain™ Kit (FD NeuroTechnologies, PK401) following the manufacturer’s
504 protocol. In brief, mice were deeply anesthetized with intraperitoneal injection of 4%
505 chloral hydrate (10 μl/g), the brain was rapidly removed and rinsed briefly in ddH2O.
506 The tissues were immersed in impregnation solution made by mixing equal volumes of
507 solution A and B, and then stored at RT for 2 weeks in the dark. Tissues were transferred
508 to solution C and store at 4°C in dark for 7 days. Coronal slices (150 µm thickness)
509 were prepared using a vibrating tissue slicer (Leica VT1000S) in solution C. Slices were
510 mounted on gelatin-coated slides (FD NeuroTechnologies, PO101) with solution C and
511 dry naturally at RT. Slices were rinsed in ddH2O twice and placed in a mixture
512 consisting of 1 part solution D, 1 part solution E and 2 part ddH2O for 10 min. Slices
513 were rinsed in ddH2O twice, dehydrated in 50%, 75%, 95% and 100% ethanol for 4
514 min, cleared with xylene for 3 times, sealed with permount (Sigma, 06522). Images of
515 hippocampus CA1 pyramidal neurons were captured by an Olympus BX61 microscope
516 with 100x objective lens. The sholl analysis, spine density and the morphological
517 classification of the spines were calculated by Image J software(50). The morphological
518 classification of spines as previously described(34, 51, 52): mushroom, big protrusions
519 with a well-defined neck and a very voluminous head; stubby, short protrusions without
520 a distinguishable neck and head; long-thin, protrusions with a long neck and a clearly
521 visible small bulbous head; filopodia, thin, hair-like protrusions without bulbous head.
522
523 Preparation of subcellular fractionation
524 Hippocampal tissues were homogenized in buffer C (320 mM sucrose, 20 mM Tris-
525 HCl [pH=8.0], 2 mM EDTA and 200 μg/ml PMSF) supplemented with protease
526 inhibitors (Merck Chemicals) and centrifuged at 800x g for 10 min at 4°C to result in
527 P1 and supernatant S1. The supernatant S1 was centrifugated at 12,000x g for 15 min
528 at 4°C to yield P2 and S2. Then P2 fraction was resuspended in buffer D (20 mM Tris-
529 HCl [pH=8.0], 1 mM EDTA, 100 mM NaCl, and 1.3% Triton X-100) and centrifuged
530 at 100,000x g for 1 h to produce a supernatant S3 and a pellet P3. The pellet represented
531 PSD fraction, and the protein concentration was measured using BCA protein assay kit
19 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
532 (Thermo, 23225).
533
534 Surface biotinylation assay
535 Acute mouse coronal slices containing hippocampus region (250 μm) were prepared
536 from anesthetic mice using a vibrating tissue slicer (Leica VT1000S) in ice-cold
537 artificial cerebrospinal fluid (ACSF) containing: 125 mM NaCl, 2.5 mM KCl, 1.25 mM
538 NaH2PO4, 1 mM MgCl2, 2 mM CaCl2, 26 mM NaHCO3 and 25 mM D-glucose, bubbled
539 with 95% O2 / 5% CO2. After recovery for 30 min at 37°C, slices were incubated with
540 0.5 mg/ml EZ-link-sulfo-NHS-SS-Biotin (Thermo, 21331) in ACSF for 30 min. The
541 biotin solution was then removed, and the remaining unreacted biotin was quenched by
542 the addition of 50 mM Tris solution (pH=7.4) for 2 min. The slices were then washed
543 with ice-cold PBS and lysed with RIPA buffer supplemented with protease inhibitors.
544 Biotinylated proteins were separated from other proteins using streptavidin agarose
545 beads (Thermo, 20349) overnight at 4°C. The agarose beads were then washed three
546 times with RIPA buffer and 500× g, 2 min centrifuge. The biotinylated proteins were
547 extracted with SDS-PAGE loading buffer and boiled for 3 min before western blot.
548
549 Transmission Electron Microscopy (TEM)
550 Mice (P30) were transcranial perfused with saline and ice-cold fixative (4% PFA
551 containing 0.1% glutaraldehyde), brains were removed and stored at 4°C for 2.5 h in
552 fixative. Coronal slices containing hippocampus region (250 μm) were cut on a
553 vibratome and kept in ice-cold fixative solution. Small blocks of tissue (1 mm2) from
554 the CA1 region were excised under a microscope. The CA1 tissues were then rinsed for
555 6 times with 0.1 M PB and post-fixed in 1% OsO4 for 30 min, rinsed for 3 times with
556 ddH2O and stained with 2% uranyl acetate for 30 min at room temperature. After
557 dehydrating through a graded series of ethanol (50%, 70%, 90% and 100%), the
558 samples were embedded in an epoxy resin. Ultrathin sections (90 nm) were cut using
559 an ultra-microtome (Leica), stained with lead citrate solution, and mounted on grids.
560 EM images were captured at 11,000× and 68,000× magnification using a Tecnai
561 transmission electron microscope (FEI, Hillsboro, OR). CA1 synapses were identified
20 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
562 by the presence of a presynaptic bouton with at least two synaptic vesicles within a 50
563 nm distance from the cellular membrane facing the spine, a visible synaptic cleft, and
564 PSD(53). Image J software was used to analyze the PSD length and PSD thickness.
565
566 Electrophysiology
567 We used 1% barbital sodium to anesthetize mice and removed the brain of mice rapidly
568 and place it in ice-cold, high-sucrose cutting solution containing (in mM): 194 Sucrose,
569 30 NaCl, 4.5 KCl, 0.2 CaCl2, 2 MgCl2, 1.2 NaH2PO4, 26 NaHCO3, 10 Glucose. We cut
570 slices coronally in the high-sucrose cutting solution with vibratome, and transferred the
571 slices to an incubation chamber with artificial cerebrospinal fluid (ACSF) containing
572 (in mM): 119 NaCl, 2.5 KCl, 2.5 CaCl2, 1.3 MgSO4, 1.0 NaH2PO4, 26.2 NaHCO3, 11
573 Glucose. Slices were incubated at 34 °C for 30 min and then transferred to 27°C for at
574 least 1 hour before recording. During recording, the slices were placed in a recording
575 chamber constantly perfused with warmed ACSF（32°C）and gassed continuously
576 with 95% O2 and 5% CO2. Pipettes for Whole-cell recordings (3-5 MΩ) were filled
577 with a solution containing (in mM): 115 CsMeSO4, 20 CsCl, 10 HEPES, 2.5 MgCl2, 4
578 Na-ATP, 0.4 Na2-GTP, 10 Na2phosphocreatine, 0.6 EGTA, 5 QX-314 (pH 7.2-7.4,
579 osmolarity 290–310). Data were collected with a MultiClamp 700B amplifier and
580 analyzed by pClamp10 software (Molecular Devices, Sunnyvale, USA). The initial
581 access resistance was <25 MΩ and was monitored throughout each experiment. Data
582 were discarded if the access resistance changed 20% during whole recording. Data were
583 filtered at 2 kHz and digitized at 10 kHz.
584 For field excitatory postsynaptic potential recording, DG zone was cut down before
585 recording to avoid the generation of epileptic wave. A concentric bipolar stimulating
586 electrode was placed in the stratum radiatum to evoke EPSP and the recording electrode
587 filled with ACSF was placed at 200-400 µm away from the stimulus electrode at the
588 same region. Increasing stimulus intensity step by step to find the maximal response
589 and then the baseline response was adjusted at 1/3 to 1/2 of the maximal response. After
590 the steady baseline response was recorded for 30 min, a strain of 100 Hz-100 pulses
21 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
591 high frequency stimulus was delivered to induce long-term potential (LTP) and a train
592 of 2 Hz-900 pulses low frequency stimulus was delivered to induce long-term
593 depression (LTD). The induced post response was recorded for 1 hour. The slope within
594 2 ms at initial EPSP was calculated as the value for field EPSP. The magnitude of LTP
595 and LTD was calculated based on the EPSP values 50-60 min after the end of the
596 induction protocol.
597 For whole-cell LTP recording, 100 µmol/L picrotoxin was added in recycling ACSF
598 to block GABAR-mediated current. The stimulus electrode was placed in the stratum
599 radiatum to evoke EPSC at 0.1 HZ and AMPAR-mediated current was recorded at -70
600 mV. After the steady baseline response was recorded for 5 min, the cell was held at 0
601 mV and received 2 strains of 100 Hz-100 pulses high frequency stimulus separated by
602 20 s. And then the cell was held at -70 mV to record for 40 min. Here the HFS must be
603 given within 10 min of achieving the whole-cell recording to avoid “wash-out” of LTP.
604 The magnitude of LTP was calculated based on the EPSC values 35-40 min after the
605 end of the induction protocol.
606 For NMDAR-mediated current and AMPAR-mediated current recording, 100
607 µmol/L picrotoxin was added in recycling ACSF to block GABAR-mediated current.
608 AMPAR-mediated current was recorded at -70 mV. A mixed current for AMPAR-
609 mediated current and NMDAR-mediated current was recorded at +40 mV. The current
610 at 50 ms after stimulus was counted for NMDAR-mediated current. The
611 AMPAR/NMDAR ratio was calculated as the peak of the averaged AMPAR-mediated
612 EPSC (30-40 consecutive events) at -70 mV divided by the averaged NMDAR-
613 mediated EPSC (30-40 consecutive events).
614 For mini excitatory synaptic transmission recording, 100 µmol/L picrotoxin and 1
615 µmol/L TTX was added into recycling ACSF to block GABA currents and sodium
616 currents respectively. Recording was conducted at -70 mV. Mini analysis software was
617 used to count the mini EPSCs.
618 Paired-pulse facilitation (PPF) was examined by paired stimulations at different
619 intervals.
620
22 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
621 Behavioral experiments
622 All behavioral experiments were performed with the age of 2- to 3-month-old male
623 littermates. All animals were given at least 1 hours of habituation time before testing.
624 Open field test
625 The open field test was used to evaluate the autonomous behavior, exploratory behavior
626 and tension of animals in a new environment. Through dynamic monitoring of
627 exploratory behavior and spontaneous activity, it is often used to test basic exercise
628 volume and anxiety level. The mice were put in an area of 40 cm× 40 cm× 40 cm box
629 and exploring freely for 30 min using a video tracking system. Data were analyzed
630 using ANY-MAZE software.
631 Y maze
632 The Y-maze was used to evaluate discriminative learning, short-term working memory
633 and reference memory test. It is based on the innate curiosity of the rodents to explore
634 an arm that has not been previously visited. The Y-shaped maze is composed of three
635 equal-length arms (40 cm×9 cm×16 cm). The angle between each two arms is 120
636 degrees. The inner arms and the bottom of the maze are painted dark grey. The mice
637 were placed in the center of the maze and allowed to freely explore the three arms for
638 8 min. Take the animals’ four feet into the arm as the standard. One alternation would
639 be recorded if all arms were entered successively at a time. The numbers of arm entries
640 were recorded to evaluate the locomotion of animals in the Y-maze. Spontaneous
641 alternation was calculated as total alternations/the number of maximum alternations
642 (total arm entries-2).
643 Elevated zero maze
644 Elevated zero maze was used to investigate the animal's anxiety state which was based
645 on the animal's exploring characteristics of the new environment and the fear of hanging
646 open arms to form conflicting behaviors. The annular platform (60 cm in diameter, 5
647 cm track width) was evaluated 50 cm above the floor and divided equally into four
648 quadrants. Two opposite quadrants were “open” and others were surrounded by 16 cm
649 high blue, opaque walls. The mice were randomly placed in closed quadrants and
650 allowed to freely to explore for a period of 5 min using an overhead video tracking
23 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
651 system. Dependent measures were: time in open quadrants, open entrances, number of
652 head dips, number of stretching and number of four paws in open section.
653 Morris water maze
654 Morris water maze is the most classic behavioral experiment that evaluates animal
655 learning, memory, cognition, and space exploration ability. The water maze consists of
656 a circular pool (radius 120 cm) filed with water containing nontoxic titanium pigment
657 to make the platform invisible. Mice were trained to find the hidden platform (5 cm)
658 and placed gently in different quadrants each time. The maximum searching period
659 allotted was 60s. If a mouse did not find the platform, it will be placed on the platform
660 by the experimenter and remained for 10s. Training was performed for five days and
661 tested on the sixth day. For the training sessions, the time taken to find the platform,
662 velocity and distance traveled were recorded. For the test session, the platform was
663 removed. The time spent in the four quadrants of the pool were recorded.
664 Self-grooming
665 The mice were placed individually in a clean cage with fresh bedding. After 10 min
666 habituation, mice were monitored with a camera and recorded for 10 min to score the
667 spontaneous grooming behavior. Time of grooming was recorded.
668 Three-chambered social test
669 Three-chamber social interaction test was carried out as described previously(54).
670 Briefly, two empty steel cylinders were respectively placed in the left and right
671 rectangular. The test mice were individually placed in the center chamber for a 10 min
672 habituation. Then the mice encountered a stranger C57BL /6Nmouse (S1) in the cage
673 for 10 min. The social novelty test was then arranged and another stranger C57BL/6N
674 (S2) was placed in the cage and let the test mouse explored freely for 10 min. The close
675 interaction time of the test mouse spend in each chamber was recorded.
676 Cued fear conditioning
677 Mice from each genotype were placed to the conditioning chamber for a 160s
678 habituation and then received 3 CS-US trails inter-spaced by 40s inter-trail intervals.
679 They were given 60s stimulus-free interval before returned to their home cage. The
680 conditional stimulus (CS) comprised a 20s discrete tone (80 dB, 2.8 kHz), which was
24 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
681 followed by an unconditional stimulus at the last 2s (US, 0.55 mA foot shock). To
682 examine the condition response to the background contextual cues, the mice were
683 placed back in the conditioning chamber for 6 min in the absence of the CS and US
684 stimuli after training 24h, 7d and 14d. To examine the condition response to the CS, the
685 mice were placed in a distinct and modified context and given a 6 min CS tone after
686 training 48h, 8d and 15d. For all procedures, the animal freezing behaviors were
687 monitored using a manufacturer’s software.
688 Prepulse inhibition (PPI) test
689 PPI was performed as described previously(55). Briefly, in PPI test, prepulses (PP) were
690 set at 70,74,78 and 82 dB, and pulse (P) was set at 120 dB. Percentage PPI was
691 calculated as [(P amplitude – PP amplitude)/ P amplitude] x 100%.
692 Statistics
693 Data were analyzed using GraphPad Prism 8.0 (GraphPad Software, San Diego, CA),
694 Excel 2017 (Microsoft, Seattle, WA), and sigmaplot 12.0 (Systat Software, Erkrath,
695 Germany). Data analysts were blind to experimental conditions until the data were
696 integrated. Standard deviations for controls were calculated from the average of all
697 control data. Statistical difference was determined using two-sided unpaired Student’s
698 t test or One-way ANOVA. The accepted level of significance was p < 0.05. n represents
699 the number of preparations, experimental repeats or cells as indicated in figure legends.
700 Data are presented as mean ± SEM. 701
25 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
702 Acknowledgments:
703 We are grateful to research assistant Shuang-shuang Liu from the Core Facilities of
704 Zhejiang University School of Medicine, Drs. San-hua Fang and research assistant
705 Daohui Zhang from the Core Facilities of Zhejiang University Institute of Neuroscience
706 and research assistant Bei-bei Wang from the center of Cryo-Electron Microscopy of
707 Zhejiang University.
708 Funding:
709 The Natural Science Foundation of China (31970902, 3192010300, 32000692 and
710 31871418), the Natural Science Foundation of Zhejiang Province (LD19H090002 and
711 LR19H090001), and the Key-Area Research and Development Program of Guangdong
712 Province (2019B030335001).
713 Author contributions:
714 R.Z. and J.X. conceived and designed the study. R.Z. and Y.-L.D. performed
715 experiments. R.Z., Y.-L.D., X.-T.W., T.-L.L., Z.Z., N.W., X.-M.L., Y.S., J.-H.L., J.X.,
716 Z.W. and J.X. analyzed and discussed the data. R.Z. and J.X. wrote the initial
717 manuscript, Y.S., J.-H.L., J.X., Z.W., L.S. and J.X. revised the manuscript. All authors
718 commented on the manuscript.
719 Competing interests:
720 The authors declare no competing interests.
721
26 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
722 References 723 1. E. W. Dent, P. W. Baas, Microtubules in neurons as information carriers. J Neurochem 129, 724 235-239 (2014). 725 2. T. Mitchison, M. Kirschner, Dynamic instability of microtubule growth. Nature 312, 237- 726 242 (1984). 727 3. X. Hu, C. Viesselmann, S. Nam, E. Merriam, E. W. Dent, Activity-dependent dynamic 728 microtubule invasion of dendritic spines. J Neurosci 28, 13094-13105 (2008). 729 4. J. Gu, B. L. Firestein, J. Q. Zheng, Microtubules in dendritic spine development. J Neurosci 730 28, 12120-12124 (2008). 731 5. E. B. Merriam et al., Dynamic microtubules promote synaptic NMDA receptor-dependent 732 spine enlargement. PLoS One 6, e27688 (2011). 733 6. T. Tada, M. Sheng, Molecular mechanisms of dendritic spine morphogenesis. Curr Opin 734 Neurobiol 16, 95-101 (2006). 735 7. G. P. Shumyatsky et al., stathmin, a Gene Enriched in the Amygdala, Controls Both Learned 736 and Innate Fear. Cell 123, 697-709 (2005). 737 8. J. Jaworski et al., Dynamic microtubules regulate dendritic spine morphology and synaptic 738 plasticity. Neuron 61, 85-100 (2009). 739 9. K. R. Brunden, J. Q. Trojanowski, A. B. Smith, 3rd, V. M. Lee, C. Ballatore, Microtubule- 740 stabilizing agents as potential therapeutics for neurodegenerative disease. Bioorg Med 741 Chem 22, 5040-5049 (2014). 742 10. A. Akhmanova, M. O. Steinmetz, Control of microtubule organization and dynamics: two 743 ends in the limelight. Nat Rev Mol Cell Biol 16, 711-726 (2015). 744 11. H. Bowne-Anderson, A. Hibbel, J. Howard, Regulation of Microtubule Growth and 745 Catastrophe: Unifying Theory and Experiment. Trends Cell Biol 25, 769-779 (2015). 746 12. J. Howard, A. A. Hyman, Microtubule polymerases and depolymerases. Curr Opin Cell Biol 747 19, 31-35 (2007). 748 13. S. L. Kline-Smith, A. Khodjakov, P. Hergert, C. E. Walczak, Depletion of centromeric MCAK 749 leads to chromosome congression and segregation defects due to improper kinetochore 750 attachments. Mol Biol Cell 15, 1146-1159 (2004). 751 14. N. Homma et al., Kinesin Superfamily Protein 2A (KIF2A) Functions in Suppression of 752 Collateral Branch Extension. Cell 114, 229-239 (2003). 753 15. G. J. Gorbsky, Mitosis: MCAK under the aura of Aurora B. Curr Biol 14, R346-348 (2004). 754 16. J. Helenius, G. Brouhard, Y. Kalaidzidis, S. Diez, J. Howard, The depolymerizing kinesin 755 MCAK uses lattice diffusion to rapidly target microtubule ends. Nature 441, 115-119 756 (2006). 757 17. A. L. Knowlton, W. Lan, P. T. Stukenberg, Aurora B is enriched at merotelic attachment 758 sites, where it regulates MCAK. Curr Biol 16, 1705-1710 (2006). 759 18. L. Wordeman, Microtubule-depolymerizing kinesins. Curr Opin Cell Biol 17, 82-88 (2005). 760 19. C. E. Walczak, The Kin I kinesins are microtubule end-stimulated ATPases. Mol Cell 11, 761 286-288 (2003). 762 20. W. Wang et al., New Insights into the Coupling between Microtubule Depolymerization 763 and ATP Hydrolysis by Kinesin-13 Protein Kif2C. J Biol Chem 290, 18721-18731 (2015). 764 21. A. T. Moore et al., MCAK associates with the tips of polymerizing microtubules. J Cell Biol 765 169, 391-397 (2005).
27 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
766 22. T. Lee, K. J. Langford, J. M. Askham, A. Bruning-Richardson, E. E. Morrison, MCAK 767 associates with EB1. Oncogene 27, 2494-2500 (2008). 768 23. S. I. Borysov, A. Granic, J. Padmanabhan, C. E. Walczak, H. Potter, Alzheimer Abeta disrupts 769 the mitotic spindle and directly inhibits mitotic microtubule motors. Cell Cycle 10, 1397- 770 1410 (2011). 771 24. A. B. Niculescu et al., Understanding and predicting suicidality using a combined genomic 772 and clinical risk assessment approach. Mol Psychiatry 20, 1266-1285 (2015). 773 25. P. D. Andrews et al., Aurora B regulates MCAK at the mitotic centromere. Dev Cell 6, 253- 774 268 (2004). 775 26. W. Lan et al., Aurora B phosphorylates centromeric MCAK and regulates its localization 776 and microtubule depolymerization activity. Curr Biol 14, 273-286 (2004). 777 27. M. Sanhaji et al., Polo-like kinase 1 regulates the stability of the mitotic centromere- 778 associated kinesin in mitosis. Oncotarget 5, 3130-3144 (2014). 779 28. E. Fifkova, A possible mechanism of morphometric changes in dendritic spines induced 780 by stimulation. Cell Mol Neurobiol 5, 47-63 (1985). 781 29. T. G. Oertner, A. Matus, Calcium regulation of actin dynamics in dendritic spines. Cell 782 Calcium 37, 477-482 (2005). 783 30. A. Vlachos et al., Synaptopodin regulates plasticity of dendritic spines in hippocampal 784 neurons. J Neurosci 29, 1017-1033 (2009). 785 31. D. A. Fortin et al., Long-term potentiation-dependent spine enlargement requires 786 synaptic Ca2+-permeable AMPA receptors recruited by CaM-kinase I. J Neurosci 30, 787 11565-11575 (2010). 788 32. W. Lu et al., Activation of synaptic NMDA receptors induces membrane insertion of new 789 AMPA receptors and LTP in cultured hippocampal neurons. Neuron 29, 243-254 (2001). 790 33. J. N. Bourne, K. M. Harris, Balancing structure and function at hippocampal dendritic 791 spines. Annu Rev Neurosci 31, 47-67 (2008). 792 34. N. L. Rochefort, A. Konnerth, Dendritic spines: from structure to in vivo function. EMBO 793 Rep 13, 699-708 (2012). 794 35. J. Gu, J. Q. Zheng, Microtubules in Dendritic Spine Development and Plasticity. Open 795 Neurosci J 3, 128-133 (2009). 796 36. K. A. Myers, K. T. Applegate, G. Danuser, R. S. Fischer, C. M. Waterman, Distinct ECM 797 mechanosensing pathways regulate microtubule dynamics to control endothelial cell 798 branching morphogenesis. J Cell Biol 192, 321-334 (2011). 799 37. X. Hu et al., BDNF-induced increase of PSD-95 in dendritic spines requires dynamic 800 microtubule invasions. J Neurosci 31, 15597-15603 (2011). 801 38. M. Sanhaji, C. T. Friel, L. Wordeman, F. Louwen, J. Yuan, Mitotic centromere-associated 802 kinesin (MCAK): a potential cancer drug target. Oncotarget 2, 935-947 (2011). 803 39. W. Wang et al., Kif2C minimal functional domain has unusual nucleotide binding 804 properties that are adapted to microtubule depolymerization. J Biol Chem 287, 15143- 805 15153 (2012). 806 40. J. Hoge, R. P. Kesner, Role of CA3 and CA1 subregions of the dorsal hippocampus on 807 temporal processing of objects. Neurobiol Learn Mem 88, 225-231 (2007). 808 41. J. Ji, S. Maren, Differential roles for hippocampal areas CA1 and CA3 in the contextual 809 encoding and retrieval of extinguished fear. Learn Mem 15, 244-251 (2008).
28 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
810 42. A. Farovik, L. M. Dupont, H. Eichenbaum, Distinct roles for dorsal CA3 and CA1 in memory 811 for sequential nonspatial events. Learn Mem 17, 12-17 (2010). 812 43. M. Setou, T. Nakagawa, D. H. Seog, N. Hirokawa, Kinesin superfamily motor protein KIF17 813 and mLin-10 in NMDA receptor-containing vesicle transport. Science 288, 1796-1802 814 (2000). 815 44. M. Setou et al., Glutamate-receptor-interacting protein GRIP1 directly steers kinesin to 816 dendrites. Nature 417, 83-87 (2002). 817 45. D. P. McVicker et al., Transport of a kinesin-cargo pair along microtubules into dendritic 818 spines undergoing synaptic plasticity. Nat Commun 7, 12741 (2016). 819 46. X. Qu et al., Stabilization of dynamic microtubules by mDia1 drives Tau-dependent 820 Abeta1-42 synaptotoxicity. J Cell Biol 216, 3161-3178 (2017). 821 47. M. Jiang, G. Chen, High Ca2+-phosphate transfection efficiency in low-density neuronal 822 cultures. Nat Protoc 1, 695-700 (2006). 823 48. E. B. Merriam et al., Synaptic regulation of microtubule dynamics in dendritic spines by 824 calcium, F-actin, and drebrin. J Neurosci 33, 16471-16482 (2013). 825 49. X. M. Zhang et al., Activity-induced synaptic delivery of the GluN2A-containing NMDA 826 receptor is dependent on endoplasmic reticulum chaperone Bip and involved in fear 827 memory. Cell Res 25, 818-836 (2015). 828 50. T. A. Ferreira et al., Neuronal morphometry directly from bitmap images. Nat Methods 11, 829 982-984 (2014). 830 51. K. Okamura et al., Cadherin activity is required for activity-induced spine remodeling. J 831 Cell Biol 167, 961-972 (2004). 832 52. J. I. Malone et al., Hyperglycemia not hypoglycemia alters neuronal dendrites and impairs 833 spatial memory. Pediatr Diabetes 9, 531-539 (2008). 834 53. G. M. Spano et al., Sleep Deprivation by Exposure to Novel Objects Increases Synapse 835 Density and Axon-Spine Interface in the Hippocampal CA1 Region of Adolescent Mice. J 836 Neurosci 39, 6613-6625 (2019). 837 54. J. L. Silverman, M. Yang, C. Lord, J. N. Crawley, Behavioural phenotyping assays for mouse 838 models of autism. Nat Rev Neurosci 11, 490-502 (2010). 839 55. M. Maekawa et al., Analysis of strain-dependent prepulse inhibition points to a role for 840 Shmt1 (SHMT1) in mice and in schizophrenia. J Neurochem 115, 1374-1385 (2010). 841
29 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
842 Figures
843
844 Figure 1. Expression of KIF2C in the central nervous system.
845 (A) Expression of KIF2C in different mouse brain regions by immunoblotting. (B)
846 Expression of KIF2C and EB3 in mouse hippocampus at different day in vitro. (C)
847 Developmental expression patterns of KIF2C and EB3 in cultured hippocampal
848 neurons. (D) Subcellular distribution of KIF2C in mouse hippocampus. H, homogenate;
849 S1, low-speed supernatant; P1, nuclei; S2, microsomal fraction; P2, synaptosomal
850 fraction; S3, presynaptosomal fraction; P3, postsynaptic density and synapto,
30 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
851 synaptophysin. (E) Representative images of hippocampal neurons double-labeled with
852 antibodies against KIF2C, PSD95 (postsynaptic marker) and Bassoon (presynaptic
853 marker). Scale bar, 5 μm.
31 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
854 855 Figure 2. KIF2C displays translocation after cLTP and is required for synaptic
856 structure plasticity.
857 (A) Electrophoresis of KIF2C and GAPDH amplicons from individual control and
858 shKIF2C hippocampal neurons. Histograms show percentage changes of KIF2C
859 mRNA levels (% of control, ***p < 0.001, n = 3 experimental repeats). (B) Cultured
860 hippocampal neurons treated with control shRNA and shKIF2C at DIV 7-10, and
32 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
861 captured at DIV 18-20. Scale bar, 1 μm. Right, Spine number per 1 μm, n = 20 neurons
862 from 3 independent culture. (C-E) Cultured hippocampal neurons treated with control
863 shRNA and shKIF2C at DIV10, and incubated with ECS or glycine (200 μM) for 10
864 min at DIV16-18. Scale bar, 1 μm. Spine density (D) and spine head width (E) were
865 calculated. Data are mean ± SEM. Spine density, n = 17 neurons from 3 independent
866 culture. Spine head width, n = 24 neurons from 3 independent culture. n.s., p >0.05, *p
867 < 0.05, ****p < 0.0001; Student’s t-test for spine density. One way-ANOVA for spine
868 head width. (F) Subcellular fraction of KIF2C in cultured hippocampus neurons after
869 10 min of glycine treatment. N-cadherin (N-cad) was served as negative control. H,
870 Homogenate; S2, microsomal fraction; P2, synaptosomal fraction and P3, postsynaptic
871 density from ECS and cLTP treated neurons. (G) Histograms show percentage changes
872 of KIF2C and β-tubulin in neurons with or without cLTP treatment. *p <0.05; Student’s
873 t-test, n = 3 experimental repeats. (H) Representative images of co-localization of
874 KIF2C (red), PSD95 (green) and MAP2 (blue). Scale bar, 5 μm. (I) Histograms show
875 percentage changes of co-localization of KIF2C and PSD95. n =20 neurons from 3
876 cultures. ***p < 0.001; Student’s t-test.
33 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
877 878 Figure 3. Knockdown KIF2C impairs synaptic transmission and plasticity
879 (A) Diagram of mcherry-shKIF2C injection, and coronal section image of the CA1
880 region. Scale bar, 100 μm. Histograms show percentage changes of KIF2C mRNA
881 levels (% of control, *p < 0.05, n = 10 cells). (B-D) mEPSCs recorded from control (n
882 = 10) and shKIF2C (n = 10) CA1 pyramidal neurons. n.s., p > 0.05, **p < 0.01,
883 Student’s t-test; For the cumulative probability curves, *p < 0.05, ****p < 0.0001,
884 Kolmogorov-Smirnov test. (E) AMPAR and NMDAR EPSCs in CA1 pyramidal
885 neurons. Left, sample sweeps illustrating NMDAR-EPSCs recorded at +40 mV (top
34 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
886 trace) and AMPAR-EPSCs recorded at −70 mV (bottom trace). Right, histograms show
887 ratio of NMDAR-EPSCs amplitude / AMPAR-EPSCs amplitude. *p = 0.0276. (F)
888 Paired-pulse stimuli evoked EPSPs with different interval. (G)Time course of
889 percentage changes of EPSCs amplitudes in control and shKIF2C CA1 pyramidal
890 neurons. Each data point represents the average of 16 successive EPSCs evoked. (H)
891 Histogram shows fEPSC slope of control and shKIF2C neurons (35-40 min). ***p <
892 0.001; Student’s t-test.
35 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
893
894 Figure 4. Abnormal spine formation in KIF2C conditional knockout mice.
895 (A) Targeting strategy to generate KIF2Cflox/flox mice. sgRNA direct Cas9 endonuclease
896 cleavage in intron 1-2 and 8-9 and create a double -strand break. (B) Nissl staining of
897 adult mice brain coronal sections and magnified images of hippocampal CA1 and CA3
898 regions. Scale bar, 50 μm. (C-F) Golgi staining of WT and cKO hippocampus CA1
899 pyramidal neuron. n = 4 mice per genotypes; Scale bar, 2 μm. (D) Sholl analysis of
900 dendritic arborization. The number of intersections related to distance from soma was
901 quantified. p = 0.2824, two-way ANOVA RM and post hoc comparisons. (E)
36 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
902 Histograms show spine number per 1 μm of WT and cKO CA1 pyramidal neuron; p =
903 0.3809. (F) The density of four types of spines in WT and cKO CA1 pyramidal neuron.
904 Mushroom spine density was 0.32 ± 0.009 μm-1 (WT) and 0.23 ± 0.01 μm-1 (cKO)
905 (****p < 0.0001). Filopodia spine density was 0.16 ± 0.009 μm-1 (WT) and 0.22 ± 0.01
906 μm-1 (cKO) (***p < 0.001). Stubby spine density was 0.12 ± 0.009 μm-1 (WT) and 0.13
907 ± 0.01 μm-1 (cKO) (p = 0.3104). Thin spine density was 0.41 ± 0.009 μm-1 (WT) and
908 0.43 ± 0.01 μm-1 (cKO) (p = 0.5511). (G) Transmission EM of synapse in hippocampal
909 CA1 region from WT and KIF2C cKO mice (n = 3 mice per genotype) at 6-8 weeks.
910 Scale bar, 500 nm. (H) Quantification of PSD thickness and length. PSD thickness:
911 40.02 ± 0.93 nm (WT) and 36.37 ± 0.58 nm (cKO). PSD length: 301.4 ± 6.8 nm (WT)
912 and 247.7 ± 5.4 nm (cKO). n = 3 mice per genotype. ***p < 0.001; ****p < 0.0001;
913 Student’s t-test.
37 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
914 915 Figure 5. KIF2C deficiency impairs synaptic transmission and plasticity
916 (A-C) mEPSCs recorded from WT (n = 21) and cKO (n = 21) CA1 pyramidal neurons.
38 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
917 ***p < 0.001, Student’s t-test; For the cumulative probability curves, n.s. p > 0.05,
918 ****p < 0.0001, Kolmogorov-Smirnov test. (D) Paired-pulse stimuli evoked EPSPs
919 with different interval. (E) AMPAR and NMDAR EPSCs in CA1 pyramidal neurons.
920 Left, sample sweeps illustrating NMDAR-EPSCs recorded at +40 mV (top trace) and
921 AMPAR-EPSCs recorded at −70 mV (bottom trace). Right, Histograms show ratio of
922 NMDAR-EPSCs amplitude / AMPAR-EPSCs amplitude. p = 0.0322. (F) Cell-surface
923 biotinylation shows a significant increase in GluA1 and GluA2 surface expression in
924 cKO hippocampus compared with WT. Histograms show percent change of GluA1,
925 GluA2, GluN1, GluN2A, GluN2B in total and biotinlation fraction. Flotillin was the
926 internal control. *p < 0.05. (G) Time course of percentage changes of fEPSCs
927 amplitudes in control and cKO CA1 slices. WT, n = 9 slices from 7 mice; cKO, n = 7
928 slices from 6 mice. (H) Histogram shows fEPSC slope of WT and cKO neurons (55-60
929 min). *p < 0.05, Student’s t-test. (I) Time course of percentage changes of fEPSCs
930 amplitudes before (baseline) and after LTD stimulation in control and cKO CA1 slices.
931 WT, n = 7 slices from 5 mice; cKO, n = 8 slices from 6 mice. (J) Histogram shows
932 fEPSC slope of WT and cKO neurons (55-60 min). p = 0.7643, Student’s t-test.
39 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
933 934 Figure 6. Abnormal MT dynamics cause deficit in synaptic transmission and
935 plasticity
936 (A) Time-lapse recordings of EB3-tdtomato-infected neurons showing EB3-tdtomato
937 comets in dendrite of WT and cKO hippocampus neuron. Subsequent images show low-
938 pass-filtered time series in row. Scale bar is 1 μm, time in seconds. (B) Analysis of the
939 EB3 comet dynamics in dendrites of WT and cKO hippocampus neuron. n = 108 from
940 7 independent experiments. Histograms show EB3 speed velocity (**p < 0.01), MT
941 growth speed (****p < 0.0001), MT growth lifetime (****p < 0.0001) and MT growth
942 length (p = 0.094). (C) Tyrosinated α-tubulin antibody-labeled MTs (white arrows) are
943 detected in a small percentage of phalloidin (F-actin) -labeled spines under basal
944 conditions (DIV 20). Quantification of the number of spines containing MTs before and
945 after BDNF treatment. Scale bar, 5 μm. 2.4 ± 0.32% (WT), 5.9 ± 0.5% (WT + BDNF),
946 3.6 ± 0.44% (cKO), 2.9 ± 0.29% (cKO + BDNF). *p < 0.05, ****p < 0.0001, one-way
947 ANOVA. (D-F) mEPSCs recorded from control (n = 21) and KIF2C(WT) (n = 21) of
40 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
948 cKO CA1 pyramidal neurons. n.s. p > 0.05, *p < 0.05, Student’s t-test; For the
949 cumulative probability curves, ***p < 0.001, Kolmogorov-Smirnov test. (G) Time
950 course of percentage changes of fEPSCs amplitudes in cKO + control and
951 cKO+KIF2C(WT) CA1 pyramidal neurons. cKO + control, n = 9 slices from 6 mice;
952 cKO+KIF2C(WT), n = 8 slices from 6 mice. (H) Histogram shows fEPSC slope of cKO
953 + control and cKO + KIF2C(WT) neurons (55-60 min). **p < 0.01. (I-K) mEPSCs
954 recorded from cKO + control (n = 18) and cKO + KIF2C(G491A) (n = 15) CA1
955 pyramidal neurons. n.s. p > 0.05, Student’s t-test; For the cumulative probability curves,
956 n.s. p > 0.05, ***p < 0.001, Kolmogorov-Smirnov test. (L) Time course of percentage
957 changes of fEPSCs amplitudes in cKO + control and cKO + KIF2C(G491A) CA1
958 pyramidal neurons. cKO, n = 7 slices from 6 mice; KIF2C(G491A), n = 8 slices from
959 6 mice. (M) Histogram shows fEPSC slope of cKO + control and cKO + KIF2C(G491A)
960 neurons (55-60 min). n.s., p = 0.1791, Student’s t-test.
41 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
961 962 Figure 7. KIF2C cKO mice exhibit abnormal cognitive behaviors
963 (A) Y maze test from WT and cKO. n = 13 per genotype. n.s., p > 0.05, *p < 0.05,
964 Student’s t-test. (B-C) Cued fear conditioning test. B, WT and cKO mice showed no
965 significant freezing response to the tone CS across three CS-US pairings. C, Expression
966 of conditioned freezing response to the background context or CS at indicated time. WT:
967 n = 12; cKO: n = 10. Training, p > 0.05, two-way ANOVA. n.s., p > 0.05; *p < 0.05;
968 **p < 0.01; ***p < 0.001, Student’s t-test. (D-E) Three-chamber test. D, WT and cKO
969 mice spent more time with stranger 1 in sociability test. **p < 0.01; ***p < 0.001. E,
970 cKO mice did not display a preference for stranger 2 in social novelty test. n =11 per
971 genotype. n.s., p > 0.05, **p < 0.01, Student’s t-test. (F) cKO mice with KIF2C(WT)
972 overexpression displayed a normal alternation in the Y-maze test. cKO+control, n = 9;
973 cKO+KIF2C(WT), n = 7. n.s., p > 0.05, *p < 0.05, Student’s t-test. (G-H) KIF2C(WT)
974 re-expression in CA1 can restore social novelty in cKO mice. cKO+control: n = 9;
975 cKO+KIF2C(WT): n = 7. *p < 0.05, ***p < 0.001. (I) cKO mice with KIF2C(G491A)
976 overexpression displayed similar deficiency in the Y-maze test. n.s., p > 0.05, Student’s
977 t-test. n = 13 per group. (J-K) cKO mice with KIF2C(G491A) overexpression displayed
978 similar deficiency in social novelty test. n = 13 per group. n.s., p > 0.05, ***p < 0.001,
979 ****p < 0.0001. Student’s t-test.
42 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
980 981 Figure 8. Model of KIF2C govern synaptic receptor expression by regulating
982 dynamic MT synaptic invasion
983 KIF2C acts as a microtubule destabilizer in hippocampal neurons and highly expresses
984 in the synapse. In basal state, KIF2C depolymerize MT to prevent excessive MT
985 invasion into spine. During elevated synaptic activity, KIF2C translocate from synapse
986 to dendritic shafts for more stable MT invasion into synapse in WT hippocampal
987 neurons. In cKO neurons, loss of KIF2C leads to increased AMPA receptors membrane
988 expression and the probability of MT invasion, impairs synaptic structure plasticity and
989 destroy MTs activity-dependent invasion. 990
43 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
991
992 Supplementary Information Figure S1: miniature excitatory synaptic currents
993 (mEPSCs) recordings of cultured WT and KIF2C knockdown (shKIF2C) neurons.
994 (A) mEPSCs recorded from WT (n = 14) and shKIF2C (n = 16) of cultured
995 hippocampus neuron at DIV16-18. Averages of amplitude were 23.0 ± 2.76 pA (control)
996 and 23.67 ± 2.45 pA (shKIF2C), p =0.7123, Student’s t-test; p > 0.05, Kolmogorov-
997 Smirnov test. (B) Averages of frequency were 7.34 ± 1.02 Hz (control) and 8.13 ± 1.03
998 Hz (shKIF2C). p = 0.5828 Student’s t-test. *p < 0.05, Kolmogorov-Smirnov test.
44 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
999 1000 Supplementary Information Figure S2: Nestin-Cre;KIF2Cflox/flox mice.
1001 (A) Nestin-Cre;KIF2Cflox/flox mouse was crossed with Ai9 reporter mouse lines and the
1002 expression of Cre-recombinase was characterized by observing tdTomato reporter in
1003 Nestin-Cre;Ai9 mice (1-month). We found that tdTomato fluorescence was present in
1004 the cortex, hippocampus, cerebellum and many other brain regions. Scale bar: 500 μm.
1005 (B) Electrophoresis of KIF2C (387 bp), and GAPDH (220 bp) amplicons from
1006 individual WT (n = 10) and cKO (n = 10) hippocampus tissues. Right histograms show
1007 percentage changes of KIF2C mRNA levels (WT: 100.0 ± 8.8%; cKO: 5.7 ± 0.9%, ***p
1008 < 0.001, Student’s t-test). (C) cKO mice (2-month) displayed normal body weight and
1009 brain size. Body weight: WT, 22.77 ± 0.6 g; cKO, 22.8 ± 0.5 g, n = 10 per genotype.
1010 n.s., p > 0.05, Student’s t-test. Brain weight: WT, 0.45 ± 0.01 g; cKO, 0.44 ± 0.01 g, n
1011 = 8 per genotype. n.s., p > 0.05, Student’s t-test.
45 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
1012 1013 Supplementary Information Figure S3: KIF2C(WT) and KIF2C(G491A)
1014 expression.
1015 (A) GFP vector, KIF2C(WT) and KIF2C(G491A) mutant were transfected into HEK
1016 293T cells and α-tubulin was stained. White arrow points to transfected cells, yellow
1017 arrows point to intact cells. Scale bar: 10 μm. (B) Coronal section of the CA1 region
1018 image after mcherry-rAAV-KIF2C virus injection. (C) Expression of Flag-KIF2C (WT)
1019 in CA1 region total protein, GAPDH was the internal control.
46 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.05.455197; this version posted August 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
1020 1021 Supplementary Information Figure S4: WT and KIF2C cKO mice behavioral tests.
1022 (A) Open field test from WT and cKO mice. Quantification of velocity, distance of
1023 activities and exploration time. n = 13 per genotype. (B) Elevated-zero maze test. n =
1024 11 per genotype. (C) Self-grooming test. WT: n = 15; cKO: n = 18. (D) WT and cKO
1025 mice displayed similar latency to platform over training days in Morris water maze test.
1026 T, target; AL, adjacent left; AR, adjacent right; OP, opposite. n.s., p > 0.05; two-way
1027 ANOVA. (E) Prepulse inhibition (PPI) test. Pulse intensity was set to 120 dB. WT: n =
1028 10; cKO: n = 9. Data are represented as means ± SEMs. n.s., p >0.05, Student’s t-test.
1029 (F) Both WT and KIF2C cKO mice show no position preference in habituation stage.
1030 n.s., p > 0.05. (G) Time line of virus injection, doxycycline (Dox) administration and
1031 behavior test. Coronal image of rAAV-TRE3G- mcherry-KIF2C re-expression in CA1.
1032 Scale bar: 800 μm.
47