Report

Retrograde semaphorin signaling regulates synapse elimination in the developing mouse brain

See allHide authors and affiliations

Science  30 May 2014:
Vol. 344, Issue 6187, pp. 1020-1023
DOI: 10.1126/science.1252514

Making and breaking neuronal synapses

As the brain develops, early synapse formation is exuberant and haphazard. But as development progresses, connections are refined into functional networks. In that process, many synapses get eliminated. Uesaka et al. now show that molecules already known for axon guidance are functional later on when they regulate the synaptic pruning needed to refine the circuits connected during axon guidance.

Science, this issue p. 1020

Abstract

Neural circuits are shaped by elimination of early-formed redundant synapses during postnatal development. Retrograde signaling from postsynaptic cells regulates synapse elimination. In this work, we identified semaphorins, a family of versatile cell recognition molecules, as retrograde signals for elimination of redundant climbing fiber to Purkinje cell synapses in developing mouse cerebellum. Knockdown of Sema3A, a secreted semaphorin, in Purkinje cells or its receptor in climbing fibers accelerated synapse elimination during postnatal day 8 (P8) to P18. Conversely, knockdown of Sema7A, a membrane-anchored semaphorin, in Purkinje cells or either of its two receptors in climbing fibers impaired synapse elimination after P15. The effect of Sema7A involves signaling by metabotropic glutamate receptor 1, a canonical pathway for climbing fiber synapse elimination. These findings define how semaphorins retrogradely regulate multiple processes of synapse elimination.

Neurons form exuberant synapses with target cells early in development. Then necessary synapses are selectively strengthened, whereas unnecessary connections are weakened and eventually eliminated during postnatal development. This process, known as synapse elimination, is crucial for shaping immature neural circuits into functionally mature versions (13) and requires unidentified retrograde signaling from postsynaptic cells. In our work, we searched for retrograde signaling molecules involved in the regression of redundant synapses from climbing fibers onto Purkinje cells during postnatal development of the mouse cerebellum (2, 46).

First, we profiled genes expressed in postsynaptic Purkinje cells during the period of climbing fiber synapse elimination (figs. S1 and S2) (see supplementary materials and methods). We picked up genes of secreted or membrane-associated molecules as possible candidates that mediate retrograde signaling. We then performed loss-of-function analyses by lentivirus-mediated RNA interference knockdown in Purkinje cells in cocultures of the cerebellum and the inferior olive, the origin of climbing fibers (7, 8). We found that knockdown of Sema3A, a secreted class of semaphorin, caused a significant reduction in the number of climbing fibers innervating each Purkinje cell (Fig. 1, A and C, and fig. S3). Furthermore, the amplitude of excitatory postsynaptic currents induced by climbing fiber stimulation (CF-EPSCs) became significantly smaller after Sema3A knockdown (Fig. 1E). In contrast, knockdown of Sema7A, a membrane-bound class of semaphorin, caused a significant increase in the number of climbing fibers innervating each Purkinje cell without changing CF-EPSC amplitude (Fig. 1, B, D, and F, and fig. S3).

Fig. 1 Effects of Sema3A or Sema7A knockdown in Purkinje cells on climbing fiber synapse elimination in olivo-cerebellar cocultures.

(A and B) Sample CF-EPSCs from control and Sema3A knockdown (Sema3A-KD) (A) or Sema7A knockdown (Sema7A-KD) (B) Purkinje cells at 16 days in vitro (DIV). Scale bars, 0.5 nA (vertical) and 5 ms (horizontal). (C and D) Frequency distributions of the number of climbing fibers innervating each Purkinje cell during 15 to 18 DIV for (C) control (white columns, n = 32 cells) and Sema3A-KD (green columns, n = 38) and for (D) control (n = 38) and Sema7A-KD (orange columns, n = 40). *P < 0.05 (Mann-Whitney U test). (E and F) Average amplitudes of CF-EPSCs for (E) control (n = 32) and Sema3A-KD (n = 38) and for (F) control (n = 38) and Sema7A-KD (n = 40). *P < 0.05 (Mann-Whitney U test). Error bars indicate SEM.

We next evaluated the roles of Sema3A and Sema7A in climbing fiber synapse elimination in vivo. We verified that Sema3A and Sema7A were strongly expressed in Purkinje cells during synapse elimination (fig. S4 and table S1). We injected lentiviruses expressing a microRNA (miRNA) against Sema3A or Sema7A into the cerebellar vermis of neonatal mice. We then examined climbing fiber innervation in acute cerebellar slices prepared from mice at various ages. We found that Sema3A knockdown caused a significant reduction in the number of climbing fibers (Fig. 2, A and C). This effect was seen from postnatal day 8 (P8) to P18 but not during P6 and P7 or P21 to P30 (fig. S5), indicating that Sema3A knockdown accelerated synapse elimination without affecting axon guidance or initial synapse formation. Furthermore, the amplitude of CF-EPSCs was significantly smaller in Sema3A knockdown Purkinje cells (table S2). There was no change in basic electrophysiological parameters of CF-EPSCs, including paired-pulse ratio (table S2), which suggests that the change is mostly of postsynaptic origin. The effect of Sema3A knockdown was rescued by co-injection of lentiviruses for the expression of a miRNA-resistant Sema3A (Sema3A-rescue) (Fig. 2, A and C). Moreover, the number of climbing fiber terminals around the somata of Sema3A knockdown Purkinje cells was smaller than that of control Purkinje cells (Fig. 2, E and G, and fig. S6). Our morphological analysis, however, was not sensitive enough to capture the intermediate state of climbing fiber terminals undergoing elimination. Taken together, these results indicate that Sema3A maintains or strengthens somatic climbing fiber synapses and opposes their elimination from P8 to P18, which covers both the early and the late phases of climbing fiber elimination (2, 913). In the present experimental condition, only a subset of Purkinje cells express Sema3A knockdown constructs, and they are surrounded by normal Purkinje cells that secrete Sema3A. Therefore, the diffusion of Sema3A is considered to be limited and does not affect climbing fibers that form synapses on neighboring Purkinje cells.

Fig. 2 Postsynaptic Sema3A and Sema7A have opposite effects on climbing fiber synapse elimination in vivo.

(A to D) Sample CF-EPSCs [scale bars, 0.5 nA (vertical) and 5 ms (horizontal)] and frequency distributions of the number of climbing fibers innervating each Purkinje cell (C and D) for control (white columns, n = 70), Sema3A-KD (dark green columns, n = 79), and Sema3A-rescue (Sema3A-RES, light green columns, n = 35) during P12 to P15 (A and C), and for control (n = 93), Sema7A-KD (orange columns, n = 84), and Sema7A-rescue (Sema7A-RES, light orange columns, n = 43) during P21 to P30 (B and D). *P < 0.05, **P < 0.005, ***P < 0.001 (Mann-Whitney U test). (E and F) Confocal microscopic images showing immunoreactivities for calbindin, a Purkinje cell marker (magenta), and vesicular glutamate transporter type 2 (VGluT2), a marker of climbing fiber terminals (green), in the same slice from a P14 mouse cerebellum for Sema3A-KD (E) and from a P21 mouse cerebellum for Sema7A-KD (F). Scale bars, 10 μm. (G and H) Frequency distributions of the number of perisomatic climbing fiber terminals on Purkinje cells for (G) control (white columns, n = 40) and Sema3A-KD (green columns, n = 38) during P14 and P15 and for (H) control (n = 30) and Sema7A-KD (orange columns, n = 44) during P21 and P22. *P < 0.05, ***P < 0.001 (Mann-Whitney U test).

In contrast to Sema3A, Sema7A knockdown caused a significant increase in the number of climbing fibers (Fig. 2, B and D). This effect was not seen before P14, became obvious during P15 to P18, and persisted into adulthood (fig. S7 and table S3). There was no change in basic electrophysiological parameters of CF-EPSCs (table S3). The number of climbing fiber terminals around the somata of Sema7A knockdown Purkinje cells was larger during P21 to P30 than that of control Purkinje cells (Fig. 2, F and H, and fig. S6), yet the intermediate state of climbing fiber terminals undergoing elimination was not found. These results indicate that Sema7A knockdown specifically impairs elimination of somatic climbing fiber synapses after P15. Our results thus suggest that Sema7A facilitates the late phase of climbing fiber elimination (2, 5, 11, 13).

We next tested whether the Sema7A signaling for climbing fiber synapse elimination is downstream of metabotropic glutamate receptor 1 (mGluR1), P/Q-type voltage-gated calcium channel (P/Q-VGCC), or glutamate receptor δ2 (GluD2), which act in Purkinje cells and are crucial for different aspects of climbing fiber synapse elimination (5, 14, 15). We found that the effect of double knockdown of Sema7A and mGluR1 was the same as that of single knockdown of mGluR1 (Fig. 3, A and D, and fig. S8), whereas double knockdown of Sema7A and either P/Q-VGCC or GluD2 had additive effects (Fig. 3, B, C, E, and F, and fig. S8). Moreover, Sema7A expression was reduced in mGluR1 knockout mice (Fig. 3, G to I). Overexpression of Sema7A into mGluR1 knockdown Purkinje cells rescued the effect of mGluR1 knockdown (Fig. 3, J and K). These results suggest that Sema7A mediates synapse elimination downstream of mGluR1 signaling.

Fig. 3 Sema7A mediates synapse elimination downstream of mGluR1.

(A to F) Sample CF-EPSCs [scale bars, 1 nA (vertical) and 5 ms (horizontal)] and frequency distributions of the number of climbing fibers innervating each Purkinje cell (D to F) during P21 to P30 for control (white columns, n = 49), mGluR1 knockdown (mGluR1-KD, darker blue columns, n = 43), and mGluR1/Sema7A double knockdown (mGluR1/Sema7A-DKD, lighter blue columns, n = 45) (A and D); for control (n = 43), P/Q-VGCC knockdown (P/Q-KD, green columns, n = 42), and P/Q/Sema7A double knockdown (P/Q/Sema7A-DKD, yellow columns, n = 39) (B and E); and for control (n = 43), GluD2 knockdown (GluD2-KD, red columns, n = 42), and GluD2/Sema7A double knockdown (GluD2/Sema7A-DKD, magenta columns, n = 39) (C and F). *P < 0.05; ns indicates no significant difference (Mann-Whitney U test). (G to I) Total and surface expression of Sema7A in mGluR1 knockout (KO) and wild-type (WT) mice were analyzed by Western blotting (n = 4 mice each). *P < 0.05 (Student’s t test). Error bars indicate SEM. (J and K) Sample CF-EPSCs [scale bars, 1 nA (vertical) and 5 ms (horizontal)] and frequency distributions of the number of climbing fibers innervating each Purkinje cell (K) for control (n = 23) and combined Sema7A overexpression with mGluR1 knockdown (mGluR1-KD/Sema7A-OE, blue columns, n = 28).

We examined the effects of Sema3A or Sema7A knockdown on other types of synapses onto Purkinje cells. Knockdown of Sema3A but not Sema7A enhanced excitatory transmission from parallel fibers to Purkinje cells (fig. S9, A and B). Neither Sema3A nor Sema7A knockdown affected inhibitory synaptic transmission to Purkinje cells (fig. S9, C to H).

Sema3A and Sema7A act on Plexin A2 (PlxnA2) and/or A4 (PlxnA4) and Plexin C1 (PlxnC1)/Integlin B1 (ItgB1), respectively (16). We found that mRNAs of receptors for Sema3A or Sema7A were expressed in inferior olivary neurons (fig. S10). We knocked down respective molecules in subsets of climbing fibers (figs. S11 and S12). Purkinje cells surrounded by climbing fibers with PlxnA4 knockdown were innervated by a significantly smaller number of climbing fibers when compared with control Purkinje cells sampled from where PlxnA4 knockdown climbing fibers were absent in the same slices (Fig. 4, A and B). Furthermore, the amplitude of CF-EPSCs was significantly smaller in PlxnA4 knockdown regions (table S2). Scrambled PlxnA4 miRNA did not alter climbing fiber innervation (PlxnA4-SCR) (Fig. 4, A and B), and the effect of PlxnA4 knockdown was rescued by coexpression of a miRNA-resistant PlxnA4 (PlxnA4-rescue) (fig. S13). Double knockdown of Sema3A in Purkinje cells and PlxnA4 in climbing fibers had the same effect as single knockdown of Sema3A (Fig. 4, G and H). The number of climbing fiber terminals around Purkinje cell somata for PlxnA4 knockdown was smaller than that for PlxnA4-SCR during P12 to P15 (fig. S14). Moreover, PlxnA4 knockdown enhanced parallel fiber to Purkinje cell transmission (fig S13). In contrast, PlxnA2 knockdown had no effect on climbing fiber innervation (fig. S15). These results indicate that Sema3A from postsynaptic Purkinje cells strengthens climbing fiber synapses and/or opposes synapse elimination through PlxnA4 in climbing fibers.

Fig. 4 Sema3A and Sema7A regulate synapse elimination through their receptors on climbing fibers.

(A to F) Sample CF-EPSCs [scale bars, 1 nA (verticlal) and 5 ms (horizontal)] and frequency distributions of the number of climbing fibers innervating each Purkinje cell (B, D, and F) for control (white columns, n = 48), PlxnA4 knockdown (PlxnA4-KD) in climbing fibers (darker green columns, n = 35), and PlxnA4-SCR in climbing fibers (lighter green columns, n = 38) during P12 to P15 (A and B); for control (n = 44), PlxnC1 knockdown (PlxnC1-KD) in climbing fibers (orange columns, n = 45), and PlxnC1-SCR in climbing fibers (light orange columns, n = 40) during P21 to P30 (C and D); and for control (n = 40), ItgB1 knockdown (ItgB1-KD) in climbing fibers (darker blue columns, n = 39), and ItgB1-SCR in climbing fibers (lighter blue columns, n = 42) during P21 to P30 (E and F). (G to L) Sample CF-EPSCs [scale bars, 1 nA (vertical) and 5 ms (horizontal)] and frequency distributions of the number of climbing fibers innervating each Purkinje cell (H, J, and L) for control (n = 30), Sema3A-KD (dark green columns, n = 21), and Sema3A/PlxnA4 double knockdown (Sema3A/PlxnA4-DKD, olive green columns, n = 23) during P9 to P12 (G and H); for control (n = 36), Sema7A-KD (orange columns, n = 21), and PlxnC1/Sema7A double knockdown (PlxnC1/Sema7A-DKD, purple columns, n = 30) during P21-P30 (I and J); and for control (n = 31), Sema7A-KD (orange columns, n = 21), and ItgB1/Sema7A double knockdown (ItgB1/Sema7A-DKD, blue columns, n = 29) during P21 to P30 (K and L). *P < 0.05, **P < 0.01, ***P < 0.001; ns indicates no significant difference (Mann-Whitney U test).

In contrast to PlxnA4, knockdown of PlxnC1 in climbing fibers caused a significant increase in the number of climbing fibers innervating each Purkinje cell (Fig. 4, C and D, and table S3). Knockdown of ItgB1 also impaired climbing fiber synapse elimination (Fig. 4, E and F, and table S3). Scrambled PlxnC1 miRNA or ItgB1 miRNA did not alter climbing fiber innervation (PlxnC1-SCR, ItgB1-SCR) (Fig. 4, C to F). The effects of PlxnC1 or ItgB1 knockdown were rescued by coexpression of miRNA-resistant PlxnC1 or ItgB1 (fig. S16). Moreover, the number of climbing fiber terminals around Purkinje cell somata for PlxnC1 or ItgB1 knockdown was larger than that for PlxnC1-SCR or ItgB1-SCR, respectively, during P21 to P30 (fig. S17). Double knockdown of Sema7A in Purkinje cells and either PlxnC1 or ItgB1 in climbing fibers had the same effects as single knockdown of Sema7A (Fig. 4, I to L). The effect of double knockdown of PlxnC1 and ItgB1 was the same as that of single knockdown of either molecule (fig. S18), suggesting that PlxnC1 and ItgB1 share the same signaling pathway. We further examined whether cofilin and focal adhesion kinase (FAK) function as signals downstream of PlxnC1 and ItgB1, respectively (17, 18). Expression of constitutive-active cofilin (cofilin-CA) and knockdown of FAK caused a significant increase in the number of climbing fibers (figs. S19 and S20). Moreover, simultaneous knockdown of PlxnC1 with overexpression of cofilin-CA and double knockdown of ItgB1 and FAK had the same effects as cofilin-CA overexpression alone and single knockdown of FAK, respectively (fig. S21). We thus conclude that Sema7A facilitates elimination of climbing fiber synapses from Purkinje cell somata through acting on PlxnC1 and ItgB1 in climbing fibers and regulating cofilin and FAK signaling.

Whereas the importance of semaphorins as axon guidance molecules has been well established (16), their roles in activity-dependent refinement of neural circuits have been unclear. Here, we have disclosed that Sema3A and Sema7A function as retrograde signaling molecules that regulate developmental synapse elimination in the cerebellum. Our results suggest that Sema3A and Sema7A have opposite effects and are involved in different stages of synapse elimination (fig. S22). Because semaphorins and their receptors are expressed widely in the brain, it is highly likely that semaphorins play important roles in developmental synapse elimination in various brain areas.

Supplementary Materials

www.sciencemag.org/content/344/6187/1020/suppl/DC1

Materials and Methods

Figs. S1 to S22

Tables S1 to S3

References (2026)

References and Notes

  1. Acknowledgments: We thank A. Nienhuis for the gifts of the lentiviral backbone vector and the packaging plasmid. A Purkinje cell–tropic viral vector has been patented (19). We also thank K. Kitamura, K. Hashimoto, Y. Sugaya, and M. Mahoney for helpful discussions and K. Matsuyama, M. Sekiguchi, M. Watanabe, S. Tanaka, and A. Koseki for technical assistance. This work was supported by Grants-in-Aid for Scientific Research (21220006 and 25000015 to M.K., 19100005 and 24220007 to M.W., and 23650160 to N.U.), the Funding Program for Next Generation World-Leading Researchers (LS021) to H.H., the Strategic Research Program for Brain Sciences (Development of Biomarker Candidates for Social Behavior), Comprehensive Brain Science Network, and the Global Center of Excellence Program (Integrative Life Science Based on the Study of Biosignaling Mechanisms) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. H.H. and T. Torashima are inventors on a Japanese and a U.S. patent application for a Purkinje cell–tropic viral vector (the modified L7 promoter sequences that enable robust transgene expression specifically in cerebellar Purkinje cells, PCT/JP2007/055017 filed on 7 March 2007 and US20100146649), which is owned by Japan Science and Technology Agency (19). Materials described here are available from H.H. and A. Nienhuis, subject to a material transfer agreement (MTA) with Gunma University for the modified L7 promoter and an MTA with St. Jude Children’s Research Hospital and the George Washington University for the pCL20c MSCV-GFP. Additional data can be found in the supplementary materials.

Stay Connected to Science

Navigate This Article