Scholarly article on topic 'PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/β1-Integrin Signaling'

PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/β1-Integrin Signaling Academic research paper on "Biological sciences"

CC BY
0
0
Share paper
Academic journal
Developmental Cell
OECD Field of science
Keywords
{}

Abstract of research paper on Biological sciences, author of scientific article — Xingfeng Liu, Jisen Huai, Heiko Endle, Leslie Schlüter, Wei Fan, et al.

Summary Alterations in dendritic spine numbers are linked to deficits in learning and memory. While we previously revealed that postsynaptic plasticity-related gene 1 (PRG-1) controls lysophosphatidic acid (LPA) signaling at glutamatergic synapses via presynaptic LPA receptors, we now show that PRG-1 also affects spine density and synaptic plasticity in a cell-autonomous fashion via protein phosphatase 2A (PP2A)/β1-integrin activation. PRG-1 deficiency reduces spine numbers and β1-integrin activation, alters long-term potentiation (LTP), and impairs spatial memory. The intracellular PRG-1 C terminus interacts in an LPA-dependent fashion with PP2A, thus modulating its phosphatase activity at the postsynaptic density. This results in recruitment of adhesome components src, paxillin, and talin to lipid rafts and ultimately in activation of β1-integrins. Consistent with these findings, activation of PP2A with FTY720 rescues defects in spine density and LTP of PRG-1-deficient animals. These results disclose a mechanism by which bioactive lipid signaling via PRG-1 could affect synaptic plasticity and memory formation.

Academic research paper on topic "PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/β1-Integrin Signaling"

Developmental Cell

Article

PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/ß1-Integrin Signaling

Highlights

• PRG-1 affects spine density and synaptic plasticity in a cell-autonomous fashion

• Cell-autonomous PRG-1 action is mediated via its intracellular C terminus (PRG-1CT)

• PRG-1CT acts via protein phosphatase 2A (PP2A)/ITGB1 activation

• Pharmacological ITGB1 activation restores spine density and synaptic plasticity

Authors

Xingfeng Liu, Jisen Huai,

Heiko Endle.....Thomas Deller,

Robert Nitsch, Johannes Vogt

Correspondence

jisen.huai@unimedizin-mainz.de (J.H.),

robert.nitsch@unimedizin-mainz.de

(R.N.),

johannes.vogt@unimedizin-mainz.de (J.V.)

In Brief

Postsynaptic PRG-1 controls lysophosphatidic acid (LPA) signaling at glutamatergic synapses via presynaptic LPA2 receptors. Liu et al. demonstrate that PRG-1 also affects spine density and synaptic plasticity in a cell-autonomous fashion via protein phosphatase 2A (PP2A)/ITGB1 activation. In line with this, PP2A activation with FTY720 rescued spine density and LTP in PRG-1 animals.

Liu et al., 2016, Developmental Cell 38,1-16

August 8, 2016 © 2016 The Author(s). Published by Elsevier Inc.

http://dx.doi.org/10.1016/j.devcel.2016.06.019

CelPress

Developmental Cell

Article

PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/ß1-Integrin Signaling

Xingfeng Liu,18 Jisen Huai,1>8* Heiko Endle,18 Leslie Schlüter,1 Wei Fan,1 Yunbo Li,1 Sebastian Richers,1 Hajime Yurugi,2 Krishnaraj Rajalingam,2 Haichao Ji,1 Hong Cheng,1 Benjamin Rister,1 Guilherme Horta,1 Jan Baumgart,1 Hendrik Berger,1 Gregor Laube,3 Ulrich Schmitt,4 Michael J. Schmeisser,5 6 Tobias M. Boeckers,5 Stefan Tenzer,2 Andreas Vlachos,710 Thomas Deller,7 Robert Nitsch,19 * and Johannes Vogt19 *

institute of Microanatomy and Neurobiology, University Medical Center, Langenbeckstrasse 1, 55131 Mainz, Germany

2MSU-FZI, Institute for Immunology, University Medical Center, Langenbeckstrasse 1, 55131 Mainz, Germany

3Institute for Integrative Neuroanatomy, CCO, Charite - University Medicine Berlin, Chariteplatz 1, 10117 Berlin, Germany

4Department for Psychiatry and Psychotherapy, University Medical Center, Untere Zahlbacher Street 8, 55131 Mainz, Germany

5Institute for Anatomy and Cell Biology, Ulm University, Albert-Einstein-Allee 11, 89081 Ulm, Germany

6Department of Neurology, Ulm University, Oberer Eselsberg 45, 89081 Ulm, Germany

7Institute for Clinical Neuroanatomy, Neuroscience Center, Goethe University Frankfurt, Theodor Stern Kai 7, 60590 Frankfurt am Main,

Germany

8Co-first author

9Co-senior author

10Present address: Institute of Anatomy II, Faculty of Medicine, Heinrich-Heine-University, Moorenstrasse 5, 40225 Dusseldorf, Germany 'Correspondence: jisen.huai@unimedizin-mainz.de (J.H.), robert.nitsch@unimedizin-mainz.de (R.N.), johannes.vogt@unimedizin-mainz.de (J.V.) http://dx.doi.org/10.10167j.devcel.2016.06.019

SUMMARY

Alterations in dendritic spine numbers are linked to deficits in learning and memory. While we previously revealed that postsynaptic plasticity-related gene 1 (PRG-1) controls lysophosphatidic acid (LPA) signaling at glutamatergic synapses via presynaptic LPA receptors, we now show that PRG-1 also affects spine density and synaptic plasticity in a cell-autonomous fashion via protein phosphatase 2A (PP2A)/ b1-integrin activation. PRG-1 deficiency reduces spine numbers and b1-integrin activation, alters long-term potentiation (LTP), and impairs spatial memory. The intracellular PRG-1 C terminus interacts in an LPA-dependent fashion with PP2A, thus modulating its phosphatase activity at the postsyn-aptic density. This results in recruitment of adhe-some components src, paxillin, and talin to lipid rafts and ultimately in activation of b1-integrins. Consistent with these findings, activation of PP2A with FTY720 rescues defects in spine density and LTP of PRG-1-deficient animals. These results disclose a mechanism by which bioactive lipid signaling via PRG-1 could affect synaptic plasticity and memory formation.

INTRODUCTION

Spines are the principal sites of excitatory synaptic transmission, playing important roles in synaptic plasticity and memory formation (Sala and Segal, 2014). The formation and maintenance of spines is regulated by integrins, a family of hetero-

dimeric extracellular matrix (ECM) receptors that span the cytoplasmic membrane (Luo et al., 2007). Upon activation, integrins reorganize the actin cytoskeleton (Ethell and Pasquale, 2005; Shi and Ethell, 2006), which is the primary structural basis of spines (Fischer et al., 1998). While a large number of integrin family members have been detected at dendritic spines, integrin subunits p1 and p3 have been directly shown to play a role in spine remodeling of hippocampal neurons (Shi and Ethell, 2006). Integrins most likely act in concert with other molecular regulators of spine plasticity (Ethell and Pas-quale, 2005), which have been reviewed elsewhere (Bourne and Harris, 2008).

Plasticity-related gene 1 (PRG-1 or LPPR-4) is a member of an integral membrane protein family that possesses homology to the lipid phosphate phosphatases (LPPs) (Sigal et al., 2005). PRG-1 exhibits the characteristic LPP feature of three conserved domains facing the extracellular side of the plasma membrane, which enable LPPs to interact with bioactive lipid phosphates such as lysophosphatidate (LPA) or spingosine-1-phosphate (Brindley and Waggoner, 1998). Bioactive lipid phosphates initiate receptor-directed signaling cascades and regulate fundamental cellular processes (Moolenaar et al., 2004). PRG-1, however, interacts with LPA in a manner different from classical LPPs (McDermott et al., 2006) by enabling transmembrane transport of LPA to intracellular compartments (Trimbuch et al., 2009; Vogt et al., 2016). PRG-1 is expressed at postsyn-aptic sites of principal neurons and acts in a non-cell-autonomous fashion by controlling LPA in the synaptic cleft, which in turn stimulates presynaptic LPA receptors resulting in an increased release probability of glutamate vesicles at excitatory synapses (Tokumitsu et al., 2010; Trimbuch et al., 2009; Vogt et al., 2016). Previous studies have shown that various members of the PRG family play a role in regulating structural plasticity, including filopodia formation, neurite extension, and brain reorganization after lesion (Brauer et al., 2003; Broggini et al.,

Developmental Cell 38, 1-16, August 8, 2016 © 2016 The Author(s). Published by Elsevier Inc. 1 This is an open access article under the CC BY license (http://creativecommons.Org/licenses/by/4.0/).

2010; Coiroetal.,2014; Peevaetal., 2006; Savaskan etal.,2004; Sigal et al., 2007; Velmans et al., 2013). However, using ion-mobility enhanced data-independent label-free liquid chromatography (LC)-tandem mass spectrometry (MS), we have detected PRG-1 as a characteristic postsynaptic density (PSD) protein, while the other members of the PRG family were not found in a high-confidence PSD preparation (Distler et al., 2014). Interestingly, unlike other members of the type 2 phos-phatidic acid phosphatase (PAP2) superfamily, PRG-1 contains an additional C-terminal hydrophilic domain of around 400 amino acids (aa) (Brauer et al., 2003). This domain has been shown to exhibit interaction motifs important for intracellular signaling cascades (Tokumitsu et al., 2010; Vogt et al., 2016), indicating an additional cell-autonomous function of PRG-1. Since (1) PRG-1 in the mouse brain is localized to the PSD of glutamatergic neurons, (2) its expression starts at embryonic day 19 (E19), and (3) it reaches its highest abundance in the third week of life (Brauer et al., 2003), i.e., during synaptogenesis and spine maturation, we hypothesized that PRG-1 could be involved in mediating these processes in a cell-autonomous fashion via its intracellular C terminus.

Since we detected an impact of PRG-1 on spine density and activation of p1-integrin, and found altered long-term potentiation (LTP) of excitatory synapses in PRG-1-deficient animals, we aimed to dissect the molecular signaling pathway that connects PRG-1 to p1-integrin activation and determine the importance of this pathway for synaptic plasticity and memory function.

RESULTS

PRG-1 Deficiency Reduced Hippocampal Spine Density and Impaired LTP

To address the role of PRG-1 in spine plasticity, we first compared spine densities in primary hippocampal neurons of wild-type (WT), PRG-1 and WT-overexpressing PRG-1 at 14 days in vitro (DIV) (Figures 1A-1C-,,2). Indeed, PRG-1 neurons showed a significant reduction in spine density of about 30% (Figure 1Ci), while overexpression of PRG-1 in WT neurons induced a significant increase in spine density, indicating a critical role of PRG-1 in spine formation and density. However, spine morphology, i.e., spine head area, was not affected (Figure 1C2). Further assessment of the neuronal morphology revealed a decreased dendritic arborization (Figures 1D and 1E) in PRG-1 neurons, which mainly affected higher-order dendritic branches while not affecting the number of dendritic end tips per se (Figures S1A and S1B). To correlate these results with a functional readout for synaptic plasticity, we probed LTP at Schaffer collateral-CA1 synapses in acute slices prepared from WT and PRG-1 animals (Figure 1F). Impairment of LTP was observed in PRG-1 animals, and hence the results of these initial experiments point toward an important role of PRG-1 in regulating spine/synapse plasticity.

PRG-1 Expression Induces Specific Cell Adhesion

To uncover molecular pathways through which PRG-1 could affect plasticity, we performed experiments in non-neuronal cells, i.e., HEK-293, which do not express PRG-1 (Figure S1C). Interestingly, stably PRG-1-expressing HEK293 cells showed

increased filopodia formation (Figure 1G) and higher vinculin expression (Figure 1H), a membrane-cytoskeletal protein involved in linking integrins to actin (Thompson et al., 2013), when compared with naive HEK293 cells. However, filopodia formation in PRG-1-expressing HEK293 cells was only observed in HEK293 cells expressing full-length PRG-1 but not in HEK293 cells expressing only the transmembrane part of PRG-1 (Figures S1E and S1F). These results indicate a role for PRG-1 in structural plasticity, which appears to be mediated by the intracellular PRG-1 C terminus. Interestingly, the function of the C terminus was only observed when it was an integral part of the PRG-1 molecule, and could not act on its own (Figure S1G). These findings are in line with increased spine numbers in PRG-1-overex-pressing WT neurons and support the notion that PRG-1 is involved in mediating structural plasticity.

To understand whether this structural plasticity is accompanied by expression of molecules involved in cell adhesion, we used a crystal violet absorbance assay and analyzed binding to various components of the ECM. PRG-1 overexpression specifically increased binding to fibronectin (FN) and laminin (LN), but not to collagens, tenascin, or vibronectin (Figure 1I). Quantitative assessment of cell adhesion revealed a robust increase in binding of PRG-1-expressing HEK293 cells to FN-coated (Figures 1J and 1K) and LN-coated surfaces (Figures 1J and 1L), respectively. Since PRG-1 is an LPA-interacting molecule, we also tested for the effect of LPA on PRG-1-induced HEK-293 cell adhesion, finding that LPA significantly reduced PRG-1 specific cell adhesion while not affecting naive HEK293 cells (Figure 1M). We conclude from these experiments that PRG-1 expression mediates a specific molecular pathway involved in cell adhesion.

PRG-1-Induced Cell Adhesion Is Mediated by b1-Integrins

Selective binding to FN and LN indicated that integrins b1 (ITGB1)-containing heterodimers are involved in PRG-1-medi-ated cell adhesion (Delwel et al., 1994). Therefore, we performed fluorescence-activated cell sorting (FACS) analysis of PRG-1-expressing HEK293 cells and indeed detected an increase in active surface ITGB1. Interestingly, total surface ITGB1 was similar in PRG-1-expressing and naive HEK293 cells (Figures 1N and 1O). Moreover, PRG-1-expressing HEK293 cells displayed a significant increased affinity to echistatin (Figure 1P), a desintegrin family member binding to activated integrins (Mon-leon et al., 2005), further indicating increased surface expression of activated ITGB1. Specific binding of PRG-1-expressing HEK cells to FN were fully blocked by echistatin (Figure 1Q) as well as by the blocking antibody P5D2, which specifically targets activated ITGB1 (Lenter et al., 1993) (Figure 1R).

To finally establish a firm link between ITGB1 and PRG-1-induced cell adhesion, we used a cell line devoid of both ITGB1 expression (GD25) (Wennerberg et al., 1996) and PRG-1 (Figure S1D). Expression of PRG-1-only did not increase adhesion to FN-coated surfaces, while such an increase was observed after PRG-1 and ITGB1 transfection of GD25 cells (Figure 1S).

PRG-1 Deficiency Leads to Reduced Active ITGB1 on Hippocampal Neurons

We further assessed the expression of endogenous activated ITGB1 on primary neurons (Figure 1Ti). PRG-1 neurons

Please cite this article In press as: Liu et al., PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/pi-Integrln Signaling, Developmental Cell (2016), http://dx.d0i.0rg/10.1016/j.devcel.2016.06.019

RïJ^JSI

• 0.4.

WT PRG-1 ^ E

i) II

?l ¿i

' ff 1

1 Y 2K_

radial distance from call body (|im) J HEK293

PRG-1/HEK-293

¿•W * * * *

• \ ? " I" . r » t-' * » * _

» •

«h fc** 1M0

HJ- „ 800

1 §600

H}- si 1400 ^ 1

Total surface ITGB1

Active surface ITGB1

Echistatin affinity

150- *

| loo- = ï

's II"

->> -V

■■'•à £

■act: ITGB1 ' \ ' . _

fei PRG-1

act. ITGB1

dendrites PSD95 punctse

V V ff'" ir

total ITGB1 —

dendritic shaft I

• 'S S*

act ITGB1 9 actin

total ITGBÎ ,

t «' î 2°'

showed a strong reduction of activated ITGB1 signal on cell bodies and dendrites when compared with WT neurons (Figures 1U1 and 1U2). Signal intensity of clusters of activated ITGB1 was also reduced in PRG-1 in comparison with WT neurons (Figure S1J). Detailed analysis of active ITGB1 expression in the dendritic shaft and in the postsynaptic compartment (marked by PSD95) confirmed the specific decrease of active ITGB1 at synaptic sites in dendrites of PRG-1 neurons (Figures 1T2, 1U3, and 1U4). Interestingly, expression of total ITGB1 was even slightly increased in neurons lacking PRG-1, indicating a specific role of PRG-1 in ITGB1 activation (Figures 1V, 1W1, and 1W2). Since we further confirmed the alteration of active ITGB1 using another independent method, i.e., hippocampal synaptic junction fractions of PRG-1-deficient mice (Figure 1X, see also

Figure S1K for further information)—together with the results obtained in HEK293 cells—we are confident in concluding that PRG-1 is linked to the activation of ITGB1.

Direct Molecular Interaction of PRG-1 with PP2A, but not with ITGB1

Since our data so far provided strong evidence for a role of PRG-1 in ITGB1 activation, we assessed direct molecular interaction between both molecules. Using a proximity ligation assay (PLA) we detected a close vicinity of PRG-1 and ITGB1 in PRG-1-expressing HEK293 cells (Figures 2A and 2B). However, immunoprecipitation (IP) studies failed to demonstrate a direct interaction between PRG-1 and ITGB1 (Figure 2C). Therefore, we used ultra-performance LC-MS to identify other putative

Figure 1. Lack of PRG-1 Alters Spine Density and LTP while PRG-1 Expression Improves Cell Adhesion to Fibronectin and Laminin via Active Integrin |!1 Upregulation

(A) Spine density In wild-type (WT) primary neurons, PRG-1 and WT neurons overexpresslng PRG-1 (WT + PRG-1). Transfectlon with GFP or PRG-1 was performed at 8 days in vitro (DIV) and imaging was performed at 14 DIV.

(B) Expression of PRG-1 fused to an FLAG tag reveals specific PRG-1 expression in dendritic spines.

(C and C2) Spine density was significantly decreased in PRG-1 neurons but increased in WT neurons after PRG-1 overexpression while spine area was not different between groups (Kruskal-Wallis with Dunn's multiple comparisons test; WT = 30 dendritic segments and 55 spines; PRG-1 = 19 dendritic segments and 41 spines; WT + PRG-1 = 29 dendritic segments and 54 spines).

(D and E) PRG-1 neurons at DIV 14 display a significantly decreased dendritic arborization (D) as shown by Scholl analysis (E, n = 19 WT and PRG-1 neurons, Mann-Whitney test; see also Figures S1A and S1B).

(F) Significantly decreased LTP in hippocampal CA1 of PRG-1 brain slices compared with WT controls (n = 13 WT and 13 PRG-1 two-way ANOVA; arrowheads indicate high frequency stimulation). Averaged excitatory postsynaptic potential (EPSP) samples during baseline (1) and 90 min post LTP induction (2) are shown above LTP graphs. Scale bars: x = 20 ms, y = 1 mV.

(G) Cellular protrusions and fllopodia extending from PRG-1-DsRed expressing HEK293 cells (lower panel) compared with MEM-DsRed expressing controls (upper panel). See also Figures S1E-S1G.

(H) Increased focal adhesion formation in PRG-1-expressing cells shown by vinculin immunostaining.

(I) Cell adhesion on different ECM substrates (collagens I, II, and IV, fibronectin [FN], laminin [LN], tenascin [TN], vibronectin [VN]) were assessed via cresyl violet cell absorbance. Data were assessed after subtraction of unspeciflc adhesion on BSA-coated substrates (Kruskal-Wallis with Dunn's multiple comparisons post hoc, n = 6 experiments for each substrate).

(J-L) (J) Cell adhesion of naive and PRG-1-expressing HEK293 cells to FN- and LN-coated surfaces. Quantification is shown in (K) and (L) (FN, n = 9 experiments and LN, n = 6 experiments each, t test).

(M) Application of 1 mM LPA reduced adhesion of PRG-1-expressing HEK293 cells on FN but did not affect naive HEK293 cells (t test, n = 6 experiments per group).

(N and O) (N) FACS assessment of total surface integrin ß1 (ITGB1; left, example of a flow cytometric assessment; right, quantification) revealed no difference between control HEK293 and PRG-1-expressing HEK293 cells while active ITGB1 was significantly higher in PRG-1-expressing HEK293 cells (O) (one-sample t test, n = 6 experiments per group).

(P) FACS assessment of cells incubated with fluorescent labeled echistatin (100 nM, 1 hr, left) shows significantly higher echistatin affinity of PRG-1-expressing HEK293 cells when normalized to the corresponding naive HEK293 cell values (one-sample t test, n = 5 experiments per group).

(Q) Quantification of cell adhesion of untreated and echistatin-pretreated (100 nM, 1 hr) PRG-1-expressing HEK cells on FN-coated plates (t test, n = 6 experiments per group).

(R) Quantitative assessment of cell adhesion of mIgG and P5D2-pretreated (active ITGBI-specific antibody, 2 mg/ml) PRG-1-expressing cells on FN-coated plates (t test, n = 6 experiments per group).

(S) Cell-adhesion assay using ITGBI-deficient GD25 cells. PRG-1 expression could only improve cell adhesion in the presence of ITGB1 (one-way ANOVA with Bonferroni post hoc, n = 6 experiments per group).

(T| andT2) Immunostaining for active ITGB1 in WTand PRG-1 primary neurons (9 DIV; Ti). Active ITGB1 and PSD95 expression on WTand PRG-1 dendritic segments (9 DIV; T2). Scale bars, 20 mm (Ti and 1.5 mm (T2).

(U1-U4) (U and U2) Mean fluorescence intensity of active ITGB1 was decreased in PRG-1 neuronal cell bodies and dendrites (since WT group was also used for analysis in Figures 3B and 3C, control for multiple comparisons was performed using one-way ANOVA with Bonferroni correction for parametric data and Kruskal-Wallis test with Dunn's multiple comparisons for non-parametric data; values are calculated in % of WT levels; n = 15 WT neuronal cell bodies and n = 9 PRG-1 neuronal cell bodies; n = 28 WT dendritic segments and 23 PRG-1 dendritic segments, see also Figure S1J). (U3 and U4) Significant reduction of active ITGB1 distribution in the dendritic shaft and postsynaptic compartment of PRG-1 neurons (n = 25 WT and 25 PRG-1 dendritic segments; t test; values are calculated in % of WT levels).

(V) Immunostaining for anti-integrin ß1 (total ITGB1) shows comparable signal intensity in WT and PRG1 primary neurons (6 DIV).

(Wt and W2) Mean fluorescence intensity of cell bodies and dendrites of total integrin in PRG-1 neurons compared with WT (Mann-Whitney test, n = 40 neurons and dendrites and 43 PRG1 neurons and dendrites; values are calculated in % of WT levels).

(X) Synaptic junction fractions containing presynaptic and postsynaptic membranes revealed significant decrease of active ITGB1 in PRG-1 hippocampi compared with corresponding WT preparation (n = 6 experiments per genotype, one sample t test; see also Figure Si K).

Values represent mean ± SEM. *p < 0.05. **p <0.01, ***p <0.001, ****p < 0.0001. Scale bars represent 1.5 mm in (A) and (B), 20 mm in (D), 10 mm in (G) and (H), 1.5 mm in (T2), and 20 mm in (Tt) and (V).

Please cite this article In press as: Liu et al., PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/pi-Integrln Signaling, Developmental Cell (2016), http://dx.d0i.0rg/10.1016/j.devcel.2016.06.019

(kDa) 83

135 115

¿P ,0

gi □ hek293 □ prg-1

GST GST-PRG1CD

PRG-1/HEK293

PRG-1/HEK293

ITGB1 52.

25 > 20 *** 200 g. 150' Fi -

u 15 S CL 10 Ï 100' «t

5 n 50'

0 é> //

G -1 IR P et

> ot 30

C 5 20

(kDa) 1 5 10 1 5 10 Lysate (m!) 75- __

PPP2 R2A

IB: PPP2R2A

purified samples

(kDa) 83'

PPP2 R2A

aL u. 5

Q- Q. 10

lysate

PPP2 R2A

IB: PPP2R2A

IB: GST

HEK293

Cortex

PRG-1-/-

IP: PRG-1

*** ***

25 » 20

Ï 15 «

£ 10 5 0

(kDa) 0

(+ 1MM LPA) 10 30 60min

IP: anti-PRG1

PRG1 PPP2R2A

PPP2R2A

35— 42_

PPP2C pY307 actin

lysate

* * * * T * X

' JL t

t5 t10t30t60

t5 t10 t30 t60

+1mm lpa

+1um lpa

interaction partners of PRG-1 (Figures S2A and S2B) and confirmed a molecular interaction between PRG-1 and PPP2R2A using IP (Figure 2D). To provide evidence for a direct protein interaction, we expressed PPP2R2A in a cell-free system and performed in vitro pull-down experiments with the purified intracellular C terminus of PRG-1 bound to glutathione S-trans-ferase (GST) (GST-PRG-1CD, see also Figures S2C and 2L2). Our results strongly suggest a direct molecular interaction between both molecules (Figure 2E). In fact, PLA showed that this molecular interaction is accompanied by a close spatial vicinity of PRG-1 and the protein phosphatase complex PP2A, which is formed by PPP2R1A, PPP2R2A, and PPP2C (PLA, Figures 2F and 2G1-2G3). Using a pull-down assay to purify PPP2R2A bound to the cytoplasmic domain of PRG-1 in HEK293 cells, we detected significant phosphatase activity of this complex (Figures 2H and 2I), indicating functional relevance of the PRG-1/PP2A interaction at the membrane. Furthermore, using mouse cortical tissue, we confirmed an endogenous direct molecular interaction of PRG-1 with PPP2R2A (Figure 2J). In line with this, cortical neurons harbor PRG-1 and the protein phosphatase complex made up of PPP2R1A, PPP2R2A, and PPP2C in close spatial vicinity as shown by PLA (Figure 2K), indicating a role of this interaction in vivo.

Interaction with PP2A Requires aa 554-588 of the PRG-1 C Terminus

Besides other interaction partners, PP2A forms a complex with Ca2+/calmodulin-dependent protein kinase IV (CaMKIV) (West-phal et al., 1998), and previous work by Tokumitsu et al. (2010) identified a calmodulin-binding motif in the intracellular C termi-

nus of PRG-1 (at aa 554-588). We tested whether this motif is able to serve as a putative interaction site with PPP2R2A and analyzed HEK293 cells expressing a PRG-1 construct lacking this binding motif (PRG-1A554~588, Figure2L1). IP studies showed no interaction between PRG-1A554-588 and PPP2R2A (Figure 2L-|). Furthermore, by employing a short peptide of PRG1 (aa 554-588 coupled to TAMRA as a fluorescent dye), we demonstrated a direct interaction with PPP2R2A, thus confirming that this short aa stretch in the C terminus of PRG-1 is driving this interaction (Figures 2L2 and S2E). PLA analysis also failed to reveal close spatial vicinity between PRG-1 A554~588 and PPP2R2A (Figures 2M and 2N). Consistent with these findings, HEK293 cells expressing prg-1a554-588 did not exhibit increased adhesion to FN, which was specific for cells expressing the WT PRG-1 construct (Figure 2O).

Application of the competitive peptide mimetic of the PRG-1 aa 554-588 fused to an HIV TAT cell penetrating sequence (TAT 554-588) to WT PRG-1-expressing HEK-293 cells significantly reduced phosphatase activity after pull-down of PRG-1 (Figure 2P). Taken together, this set of experiments suggest that the molecular interaction of PRG-1 and PPP2R2A at the 554-588 binding motif is a prerequisite for PRG-1 recruitment of active PP2A to the membrane. Since PRG-1 interacts with extracellular LPA (Brauer et al., 2003; Trimbuch et al., 2009; Vogt et al., 2016), we assessed the impact of LPA on HEK293 cells expressing PRG-1 (Figures 2Q-2S). These studies revealed that PRG-1/PPP2R2A interaction was decreased following LPA application (Figures 2Q and 2R) while phosphorylation of PPP2C at aa Y307 (as a marker for PP2A activity inhibition) was increased (Figures 2Q and 2S).

Figure 2. PRG-1 Interacts with PP2A

(A and B) (A) Proximity ligation assay (PLA) of PRG-1 and ITGB1 transfected cells Indicating close vicinity of both molecules and a significant Increase In PLA signal In PRG-1-expressing HEK293 cells (B; t test, n = 22 HEK-293 and n = 23 PRG-1-expressing HEK293 cells).

(C and D) Co-immunoprecipitation (IP) using a PRG-1 antibody shows PRG-1 and PPP2R2A association (D; see also Figures S2A and S2B), while no signal was detected for integrin p1 (ITGB1, C).

(E) IP using GST-coupled PRG-1 C-terminal domain (GST-PRG-1CD) produced in E. coli and PPP2R2A produced in a cell-free protein expression system. After in vitro reaction 1, 5, or 10 ml of the cell-free lysate was used as indicated. GST-coupled protein (upper panel) and PPP2R2A (bottom panel) detected by western blotting. Right panel shows the expression level of PPP2R2A from the rabbit reticulocyte lysate (see also Figures S2C and S2D).

(Fand G) PP2A subunits A (PPP2R1A), B (PPP2R2A), and C (PPP2CA) were found in close proximity to PRG-1 in PRG-1-expressing HEK293 cells as shown by in situ PLA and fluorescence signal quantification (G1-3; PPP2R1A n = 9 HEK and 8 PRG1; PPP2R2A n = 9 HEK and 11 PRG-1; PPP2CA n = 6 HEK and 9 PRG-1; n represents number of analyzed pictures. When more than one cell was visible, the mean of the displayed cells was calculated and used for the statistical analysis; t test was applied for parametric data and Mann-Whitney test was applied for non-parametric data).

(H) PP2A activity of tandem affinity purification pull-down from PRG-1CD (PRG-1 cytoplasmic domain) expressing cells and naive HEK293 cells assessed by a serine/threonine phosphatase assay shows significant increase of PP2A activity indicating maintained PP2A activity after recruitment by PRG-1 (t test, n = 6).

(I) PP2A activity in total cell lysates was not different in PRG-1CD expressing and naive HEK293 cells (t test, n = 6). (J) IP using a PRG-1-specific antibody detected PRG-1/PPP2R2A interaction in the cortex.

(K) PP2A subunits A (PPP2R1A), B (PPP2R2A), and C (PPP2CA) were found in close proximity to PRG-1 in hippocampal neurons by PLA. 0-! and L2) (L1 Deletion of amino acids (aa) 554-588 of PRG-1CD impeded PRG-1 and PPP2R2A interaction as shown by IP using a PRG-1-specific antibody. (L2) The PRG-1 aa 554-588 peptide tagged to TAMRA as a fluorescence indicator dye binds glutathione Sepharose (see also Figure S2E) and interacts with PPP2R2A produced in a highly standardized cell-free system.

(M and N) (M) PRG-1 aa 554-588 are critical for PRG-1/PPP2R2A interaction as shown by PLA and gray value quantification (N; since PRG-1/PPP2R2A group was also used in quantitative analysis shown in Figure 2G2, control for multiple comparisons was performed by one-way ANOVA with Bonferroni post hoc correction, n = 9 for HEK-293 and n = 11 for PRG-1 and for PRG-1 i554-588-expressing HEK293 cells).

(O) Deletion of aa 554-588 significantly attenuated PRG-1-expressing cell adhesion on fibronectin-coated surfaces (one-way ANOVA with Bonferroni post hoc, n = 6 experiments per condition).

(P) PP2A activity of pull-down from PRG-1 CD-expressing HEK-293 cells and from TAT554-588-treated PRG-1-CD-expressing cells shows significantly lower

activity than non-treated cells, indicating decreased PP2A/PRG-1 binding upon PRG-1CD/TAT554-558 interaction (t test, n = 6 for both groups).

(Q-S) (Q) Western blot of IP after LPA stimulation (1 mM) using a PRG-1 antibody indicates diminished PPP2R2A concentrations overtime (R) and an increase of

PPP2R2A phosphorylation (Y307) upon LPA stimulation (S; data normalized to the T0 values [set at 100] and calculated for each column using a one-sample t test;

n = 6 experiments for PPP2R2A/PRG-1 association and n = 6 experiments for PP2ApY307).

Values represent mean ± SEM. **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bars, 10 mm.

Please cite this article in press as: Liu et al., PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/p1-Integrin Signaling, Developmental Cell (2016), http://dx.d0i.0rg/10.1016/j.devcel.2016.06.019

>VT ■■ t. : --i ; . • ' '* V active 01-ihtegrin WT+ TAT-PRG-1 554-588 --a "^C, * ï iL. active p1-integrin

PRG-1 + active fS1-integrin . . PRG-1'+FTY720 ■ ■■ J ;V' v... • ■ ' •

dendrites

E 0.4' f

0. 100 (0 0. №

50 100

Tims [min]

50 100

Time [min]

Figure 3. PRG-1/PP2A Interaction Induces ITGB1 Activation, Regulates Spine Density, and Enhances LTP

(A-C) (A) Active ITGB1 immunostaining of DIV9 WT hippocampal neurons treated with TAT-PRG-1-554-588 peptide (10 mM, 1 hr) display a significantly lower fluorescence intensity of active ITGB1 on neuronal cell bodies and dendrites (B and C; t test, for cell bodies WT n = 18, WT + TAT-PRG-1-554-588 n = 13; for dendrites WT n = 28, WT + TAT-PRG-1-554-588 n = 27). Scale bar, 20 mm.

(D) PP2A activator FTY720 (1 mg/ml, 1 hr preincubation) significantly improved cell adhesion while the PP2A inhibitor okadaic acid (OA) (100 nM, 2 hr preincubation) prevented PRG-1-dependent cell adhesion (t test, PRG-1, PRG-1 + OA n = 5 experiments per group; FTY720 n = 8 experiments per group).

(E-G) PRG-1 primary neurons (9 DIV) display a significantly higher fluorescence signal for active ITGB1 after stimulation with the PP2A activator FTY720 (1 mg/ml, 48 hr incubation; F and G, one-way ANOVA with Bonferroni post hoc for normal distributed data and Kruskal-Wallis with Dunn's multiple comparison test for non-parametric data; for cell bodies WT n = 15, PRG-1 n = 9, PRG-1 + FTY720 n = 13; for dendrites WT n = 28, PRG-1 n = 23, PRG-1 + FTY720 n = 23).

(H and I) FTY720 (100 nM) treatment for 6 days significantly increased spine density in dissociated PRG-1 neurons (n = 69 PRG-1 and 34 PRG-1 + FTY720 dendritic segments; t test). Scale bar, 1.5 mm.

(J and K) Vehicle-treated PRG-1 neurons and PRG-1 neurons treated for 24 hr with 1 nM OA showed no significant change in spine density.

Interaction of PRG-1 with PP2A Is Involved in Activation of ITGB1, Regulation of Spine Density, and LTP

Next, we investigated the role of the molecular interaction of PRG-1 with PPP2R2A for PRG-1-dependent activation of ITGB1 and regulation of dendritic spine density. We used the competitive binding peptide TAT 554-588 in primary hippocam-pal neurons (Figure 3A) and detected a significant reduction of activated ITGB1 on both cell bodies (Figure 3B) and dendrites (Figure 3C).

To test for the role of PPP2R2A in mediating PRG-1-depen-dent activation of ITGB1, we used established pharmacological tools to either stimulate or inhibit PP2A activity. Following treatment with FTY720, which serves as an activator of PP2A (Perrotti and Neviani, 2013), PRG-1-expressing HEK293 cells showed a significant increase in specific adhesion to FN-coated surfaces (Figure 3D). In contrast, selective inhibition of PP2A by nanomolar concentrations of okadaic acid (OA) (Takai et al., 1992) significantly decreased this adhesion (Figure 3D). Moreover, FTY720 was able to induce expression of activated ITGB1 (Figure 3E) on cell bodies and dendrites of PRG-1 hippocampal neurons to levels observed in WT neurons (Figures 3F and 3G). In line with these results, PRG-1 hippocampal neurons treated with 100 nM FTY720 over 8 days showed an increase in spine density compared with untreated control cells (Figures 3H and 3I). However, treatment of PRG-1 neurons with 1 nM OA did not alter spine density, further supporting specific PRG-1/PP2A interaction in spine formation (Figures 3J and 3K). Finally, in vivo treatment with FTY720 (5 mg/kg per day for at least 30 days) significantly rescued the deficit in LTP at Schaffer collateral-CA1 synapses in PRG-1 slices (Figure 3L), while having no effect on LTP in WT animals (Figure 3M).

PRG-1 Organizes the Focal Adhesion Complex in Lipid Rafts

PP2A is known to interact with src (Pradhan et al., 2010), a non-receptor tyrosine kinase, thereby controlling activation of integrins. To analyze whether scr is involved in PRG-1/PP2A interaction driving ITGB1-mediated adhesion, we analyzed PRG-1-induced adhesion to FN using a cell line devoid of src (SFY cells) (Klinghoffer et al., 1999), which shows no PRG-1 expression (Figure S3A). Overexpression of PRG-1 was only able to increase adhesion in SFY cells in the presence of src (Figures 4A and S3A), but not in normal, non-src-expressing SFY cells (Figure 4A), demonstrating a crucial role of src in the PRG-1-dependent ITGB1 activation. PLA studies (Figure 4B) revealed a close spatial vicinity of src and PRG-1 (Figure 4C), while the src inhibitor PP2 (Hanke et al., 1996) was able to block any PRG-1-mediated adhesion of HEK293 cells to FN (Figure 4D). Since PP2A and src can locate to lipid rafts (Xu et al., 2013), we analyzed whether PRG-1 has an impact on PPP2R2A and src in lipid rafts, and showed a significant PRG-1-dependent in-

crease of both molecules in the lipid rafts containing membrane fraction (F2) (Figures 4E-4G).

To answer the question whether PRG-1 expression results in recruitment of further members of the focal adhesion complex important for ITGB1 activation to lipid rafts, we performed a systematic PLA study for ITGB1, talin (Calderwood, 2004), paxillin (Schaller, 2001), src, and PP2A in lipid rafts (identified by caveo-lin-1 expression) of PRG-1- and PRG-1 D554-588-expressing, and naive HEK293 cells (Figure 4H). We detected a significant increase in the presence of all the aforementioned molecular components in lipid rafts (Figures 4H and 4I). Importantly, the increase of their presence in lipid rafts was dependent on the direct molecular interaction of PRG-1 with PPP2R2A, since HEK293 cells expressing PRG-1D554-588 lacking this interaction (Figure 2J) showed a significantly reduced lipid raft localization (Figures 4H and 4I).

The importance of the localization of this ITGB1 activating molecular complex in lipid rafts for PRG-1-induced adhesion of HEK293 cells to FN-coated surfaces was further demonstrated using MpCD and nystatin, known to disrupt raft formation (Wang et al., 2013), both of which were able to block this PRG-1-induced adhesion (Figure 4J). In fact, PLA studies (Figure 4K) revealed that PRG-1 is in close spatial vicinity to both talin and paxillin Figures 4L-| and 4L2). Seeding of PRG-1-expressing HEK cells on FN-coated surfaces in turn showed a time-dependent increase in paxillin phosphorylation (Figures 4M and 4N), indicating an outside-in signaling promoting paxillin action to stimulate focal adhesion complexes.

To understand the impact of PRG-1 on talin, we first assessed the total amount of talin and did not detect a difference between naive and PRG-1-expressing HEK cells (Figures 4O and 4P). FACS analysis revealed an increase in immunolabeling of talin rod and a decrease in talin head in PRG-1-expressing HEK293 cells when compared with their controls while no difference in serine-425 phosphorylation of talin could be observed (Figures 4Q and 4R). These data indicate that PRG-1 induces the talin head to form a complex with the ITGB1 cytoplasmic tail, thereby hiding the epitope for antibody binding.

PRG-1 Is Important for Hippocampal Spine Density In Vivo

Finally, we aimed to demonstrate the relevance of PRG-1 for the regulation of neuronal spine plasticity in vivo. Using electron microscopy, we detected that PRG-1 is already expressed in nascent spines at the PSD as early as P2, allowing for its involvement in spine formation (Figures S4A-S4D). Next, we assessed spine density in three different regions of the hippocampus (Figure 5A) at postnatal day 12 (P12) and P19 (Figures 5B and 5D). In all areas and at all time points investigated, our spine counts revealed significantly lower densities in PRG-1-deficient mice compared with WT littermate controls (Figures 5C and

(L) LTP In hippocampal CA1 of PRG-1 and FTY720-treated PRG-1 mice. Arrowhead Indicates stimulation (n = 6 PRG-1 and 6 FTY720 treated PRG-1 mice; two-way ANOVA). Averaged EPSP samples during baseline (1) and 90 min post LTP induction (2) are shown to the right of LTP graphs. Scale bars: x = 20 ms, y = 0.25 mV.

(M) LTP in hippocampal CA1 of WT and FTY720-treated WT mice. Arrowhead represents stimulation (n = 10 WT and 10 FTY720-treated WT mice; two-way ANOVA). Averaged EPSP samples during baseline (1) and 90 min post LTP induction (2) are shown to the right of LTP graphs. Scale bars: x = 20 ms, y = 0.5 mV. Values represent mean ± SEM. *p < 0.05, ***p < 0.001, ****p < 0.0001. Scale bars represent 20 mm in (A) and (E) and 1.5 mm in (H) and (J).

Please cite this article In press as: Liu et al., PRG-1 Regulates Synaptic Plasticity via Intracellular PP2A/p1-Integrln Signaling, Developmental Cell (2016), http://dx.d0i.0rg/10.1016/j.devcel.2016.06.019

*** T 1 100'

n ra 50' 6

HEK293 PRG-1/HEK293

(kDa) F1 F2 F3 F4 F1 F2 F3 F4

52' 130

2 P P P

PPP2R2A>j_ 10, **

.gf> N

PRG-1 PRG-1 A554-588

fwö %

□ HEK293

□ PRG-1

B PRG-1A554-588 ***

- ' '^Fi

: 50020

-- h 15

10 _*" 8'

li-H 2

4° I* 30 -X. 11

0 0 n 1

5 111 0 5 JL 111

HEK293 PRG-1

(Kda) 0 20min 0 20min 68

La hek293 1 □ prg-1

c 80 60

|S 40 20 0*

20 - = 15

Paxillin a 100.

(Kda)# +

Talin Actin

Talin ROD Talin HEAD Talin pS425

lysate

T0 T20

R ROD HEAD pS425

« 200

150 §

3100 e 50 0

HEK293

N S 500

HEK293

*** ***

HEK293

5E). To provide unequivocal evidence that spine density depends on postsynaptic PRG-1, we electroporated a Cre-encod-ing construct into the brain of mice carrying a floxed prg-1 gene, thus deleting PRG-1 expression (Figure 5F). Similar to the results in the PRG-1 mice, single PRG-1-deficient neurons showed a highly significant decrease in their spine density compared with PRG-1-expressing neurons (Figures 5F and 5G).

Interestingly, in PRG-1 animals, active ITGB1 was no longer confined to dendritic spines as was the case in WT animals (Figures S4E-S4H). These findings support the hypothesis that postsynaptic PRG-1 directly modulates spine density via ITGB1. We further assessed whether the introduction of postsynaptic PRG-1 in a PRG-1 brain is sufficient to rescue the spine density. Spine analysis indeed confirmed that PRG-1 expression in single cells in a PRG-1 brain (Figure 5F) was able to reconstitute spine density in these neurons (Figures 5F and 5G). Since we have previously shown that PRG-1 modulated glutamatergic transmission in a non-cell-autonomous fashion via the presynaptic LPA2 receptor (Trimbuch et al., 2009), we now analyzed whether a PRG-1 variant (PRG-1R346T), not capable of intracellular LPA uptake (Vogt et al., 2016) and thereby unable to act in a non-cell-autonomous fashion, still has the capacity to affect spine density in PRG-1 animals. As shown in Figure 5F, in utero electroporation of PRG-1R346T induced a visible increase in spine density resulting in a rescue to WT levels (Figure 5G).

To determine whether the observed decrease in spine density is transient and subsequently normalizes with brain maturation, we also analyzed spine densities in adult animals. We found that the highly significant reduction of spine density observed in the juvenile brain was still present in adults while other features of spine morphology were not altered (Figures 5H-5I2). Quantification of VGlut1-positive axonal boutons impinging on dendritic shafts showed no differences between WT and PRG-1 neu-

rons, while density of spine-associated VGlutl-positive boutons was drastically reduced (Figures S4I-S4K). This shows that decreased axonal contacts on spines are not compensated by shaft synapses. To assess a role of PRG-1 in the dynamic regulation of spine density, we either kept WT and PRG-1 animals in isolation or exposed them to an enriched environment. Assessment of spine density showed significant increase of spine density in WT animals under enriched conditions but no changes in PRG-1 animals, showing a loss of dynamic regulation of spine density in PRG-1 animals (Figure S4L).

PRG-1 Regulates Spine Plasticity and Spatial Memory in Cell-Autonomous Fashion

Our data so far indicate a cell-autonomous action of PRG-1 on spine density mediated by molecular interaction resulting in ITGB1 activation. However, PRG-1 plays a role in the modulatory control of hippocampal excitability in a non-cell-autonomous fashion (Trimbuch et al., 2009). Since spine density can be affected by neuronal activity (McKinney et al., 1999), we have addressed a role of altered neuronal activity observed in PRG-1-deficient animals (Trimbuch et al., 2009) on spine density. The non-cell-autonomous PRG-1 role in controlling excitability results in an increase in synaptic strength of PRG-1 animals (Trimbuch et al., 2009; Unichenko et al., 2016) and requires the presence of presynaptic LPA2 receptor signaling (Trimbuch et al., 2009), and this hyperexcitability was fully reversed in LPA2 receptor (LPA2R)/PRG-1 double-deficient animals (Trimbuch et al., 2009). Hence, we tested for the possibility that alteration in spine densities can be explained by the LPA2R-mediated non-cell-autonomous effect of PRG-1 on neural excitability and analyzed spine density in PRG-1-deficient animals on an LPA2R_/~ background (Figure 5J). While spine density analyses in LPA2R_/~ mice revealed no significant differences from WT animals, LPA2R_/~/PRG-1 animals displayed

Figure 4. PRG-1 and PP2A Organize Src/Paxillin/Tallin Complex Formation, Activating ITGB1

(A) Src-deficient cell line (SFY) showed PRG-1-specific cell adhesion only after Src expression (one-way ANOVA with Bonferroni post hoc, SFY and PRG-1/SFY

n = 6, Scr/SFY n = 12, PRG-1/SRC/SFY n = 9 experiments per group; see also Figure S3A).

(B and C) (B) PLA showing in situ Src expression in close proximity to PRG-1. Quantification in (C) (t test, n = 6).

(D) Src inhibition via PP2 significantly reduced adhesion of PRG-1-expressing HEK293 cells to FN (t test, PRG-1 n = 6, PRG-1 + PP2 n = 8). (E-G) (E) Western blot showing the F2 fraction of membrane rafts preparations from naive and PRG-1-expressing HEK293 cells. Upon PRG-1 expression (see clear band in the F2 fraction), signals for PPP2R2A, Src, and ITGB1 in the membrane raft fraction (F2) were markedly enhanced. Quantification of Src and PPP2R2A in (F and G) (F2, unpaired t test; for Src n = 8 and for PPP2R2A n = 6 experiments per group).

(H1-H6 and I1-I6) (H1, I1) PRG-1 (n = 6) and PRG-1D554-588 (n = 6) show similar membrane raft localization as shown by colocalization with caveolin (Cav1) in the PLA. (H2, I2) PPP2R2A shows significantly higher membrane raft recruitment in PRG-1- (n = 7) but not in PRG-1D554-588 (n = 7)-expressing cells. (H3-5, I3-5) Src (n = 9 for PRG-1 and n = 8 for PRG-1D554-588), paxillin (n = 5 for PRG-1 and n = 6 for PRG-1D554-588), and talin (n = 8 for PRG-1 and n = 6 for PRG-1D554-588), which are located downstream of PPP2R2A, show significantly higher membrane raft recruitment in PRG-1 but not in PRG-1 D554-588-expressing cells. (H6, I6) Total ITGB1 expression in lipid rafts was significantly decreased in PRG-1 D554-588-expressing cells (n = 9) when compared with PRG-1-expressing cells (n = 10) (oneway ANOVA with Bonferroni post hoc for all experiments).

(J) Raft destruction by treatment with MpCD (5 mM, 30 min) or nystatin (25 mg/ml, 30 min) significantly reduced PRG-1-induced cell adhesion (one-way ANOVA with Bonferroni post hoc, n = 7 for PRG-1 and n = 6 for PRG-1 + MpCD or with nystatin).

(K and L) PLA showing expression of complex components for ITGB1 activation such as paxillin (n = 5 for HEK293 cells and n = 6 for PRG-1-expressing HEK293 cells) and talin (n = 5 for HEK293 cells and n = 8 for PRG-1-expressing HEK293 cells) in close proximity to PRG-1 (t test was used for parametric data, MannWhitney for non-parametric data).

(M and N) (M) Western blot showing paxillin phosphorylation (pY118) in naive and PRG-1-expressing HEK293 cells before (0 min) and 20 min after seeding on a fibronectin-coated plate. Quantification of gray values in (N) (normalized to T0 values of HEK293 cells) revealed significant increase of paxillin phosphorylation after 20 min in PRG-1-expressing HEK cells (one-sample t test for T0 of PRG-1 and unpaired t test for T20 of HEK-293 and PRG-1, n = 5 for all groups). (O and P) (O) No difference in talin expression in western blot of cell lysate of naive and PRG-1-expressing HEK293 cells. Quantification is shown in in (P) (t test, n = 6 for all groups).

(Q and R) FACS analysis using antibodies against talin head, talin rod, and S425 talin phosphorylation. Quantification of talin rod (n = 6), talin head (n = 6), and serine-425 phosphorylation (n = 7) in PRG-1-expressing HEK293 cells in (R). (one-sample t test).

Values represent mean ± SEM. *p < 0.05. **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bars represent 10 mm in (B) and 20 mm in (H) and (K).

Thy1 -EGFP

P12nWT

□ KO

P19 C]WT □ KO

□ WT Spine length I2 Spine area

□ KO

a o.5'

□ wt □ PRG-1 + ^^^ !s 100 ■ te . -s ll^, a* Liluf £ so ■

lAfld *60 ' **** ^ ^

I 300|

■s £ 200

Time [min]

0 LPAjR

£ 60' 1

0 PRGT'" LPAjR +

Y 2|/>

100' ¡80

I GO «

-40' ¡20'

learning days

O PRG-1/LPA2 dKO ■©■ LPAj-KO

reduced spine densities (when compared with WT animals) which were not different to those seen in PRG-1 -/- animals (Figures 5J and 5K). These LPA2R-/-'PRG-1-/- animals, with reduced spine density but no increase in synaptic strength, due to genetic ablation of the presynaptic LPA2R mediating the increased synaptic strength in PRG-1 -/- animals (Trimbuch et al., 2009), exhibit a reduction in LTP when compared with LPA2R-/- animals, which lack the non-cell-autonomous action of PRG-1 mediated via presynaptic LPA receptors (Trimbuch et al., 2009), but maintain the cell-autonomous function mediated by the intracellular C terminus of PRG-1 at the postsynaptic side (Figure 5L). Taken together, these findings provide additional evidence for the fact that reduced spine plasticity in PRG-1 -/- mice is not mediated by its non-cell-autonomous action regulating hyperexcitability via presynaptic LPA receptors, but rather depends on the ability of PRG-1 to induce a cell-autonomous intracellular signaling pathway upon LPA interaction mediated via the PRG-1 intracellular C terminus.

Finally, we carried out behavioral experiments using a Morris water maze (MWM) to test whether the observed deficits in spine density and LTP in the hippocampus of PRG-1 -/- mice translate to alterations in spatial memory. PRG-1 -/- mice showed a significantly lower performance in the learning phase and no significant spatial memory formation in the probe trial (Figures 5N and 5M). We also used LPA2R-/-'PRG-1 mice to rule out a non-cell-autonomous role of PRG-1 mediated by presynaptic LPA2-Rs. Indeed, deficits in spatial memory observed in

PRG-1 -/- mice were not rescued by additional deletion of the presynaptic LPA2 receptor (Figures 5M-5O), confirming a direct, cell-autonomous function of PRG-1 in these features.

DISCUSSION

In this study we report a physiological role of PRG-1 for neuronal spine density, LTP, and spatial memory. We demonstrate that PRG-1 -/- mice have reduced spine density compared with their WT littermates. Spine density could be recovered by interfering with a signaling pathway by which a specific intracellular PRG-1 domain, via direct interaction with PPP2R2A, resulted in activation of ITGB1. This PRG-1 action is independent of its known extracellular role in controlling LPA receptor-mediated synaptic hyperexcitability (Trimbuch et al., 2009; Unichenko et al., 2016), and essential for proper spatial memory. These data provide evidence for the fact that PRG-1 controls hippo-campal synaptic plasticity that is important for related memory function in a cell-autonomous fashion.

It is well established that actin rearrangement drives the formation and loss of dendritic filopodia, which are developmental precursors of spines (Ziv and Smith, 1996). This led us to hypothesize that molecular interaction occurring at the long cytoplasmic C terminus of PRG-1 influences actin regulatory proteins in an as yet unknown, cell-autonomous manner. Here we show that PRG-1 controls spine density via interaction of its cytoplasmic calmodulin-binding domain (CBD; Tokumitsu et al.,

Figure 5. PRG-1"'" Animals Display Decreased Spine Density and Altered Spatial Memory

(A) Hippocampus of a Thy-1 L21 EGFP mouse showing areas of spine assessment (highlighted by red squares): stratum radiatum (sr, apical dendrites) and stratum oriens (so, basal dendrites) of the CA1 region, and the outer two-thirds of stratum moleculare (sm) of the dentate gyrus (DG).

(B-E) (B and D) Spine density at P12 and P19 indicate lower spine density in PRG-1 -'- mice at P12 (C, n = 16 WTand 37 PRG-1 -'- for sm; n = 35 WTand 35 PRG-1-'- for sr; n = 35 WTand 33 PRG-1-'- for so) and at P19(E, n = 23 WTand 33 PRG-1-'- forsm; n = 18 WTand 26 PRG-1-'- forsr; n = 21 WTand 21 PRG-1-'- for so; n represents analyzed dendritic segments; t test).

(F) Images of dendritic segments of an in utero GFP-transfected WT neuron and a constitutive PRG-1-'- neuron, of a single-cell PRG-1-'- neuron induced by Cre-electroporation in a pRG-l"°x"l°x brain (single-cell cre knockin, scCre-KI) and of a PRG-1 reconstituted neuron in a PRG-1-'- brain (scPRG-1-KI) with corresponding immunostaining for PRG-1, confirming effective PRG-1 re-expression, and of a PRG-1R346T reconstituted neuron in an PRG-1-'- brain (from left to right).

(G) Single-cell PRG-1 deletion (scCre-KI) via in utero electroporation of a Cre-encoding construct in a conditional PRG-1""1 mouse showed a spine density reduction similar to that in PRG-1-'- animals. Re-expression of PRG-1 (scPRG-1 KI) rescued spine density in PRG-1-'- neurons to WT levels. Afunction-deficient PRG-1 variant (PRG-1R346T), unable to take up LPA, also rescued density, pointing to PRG-1's cell-autonomous effect (n = 18 for WT, n = 20 for PRG-1-'-, n = 26 for single-cell PRG-1-'-, n = 13 for single-cell PRG-1-reconstitution (KI), and n = 18 for PRG-1R346T reconstituted PRG-1-'- neurons; n represents analyzed dendritic segments; one-way ANOVA with Bonferroni post hoc).

(H) Adult PRG-1-'- animals displayed significantly reduced spine densities in all analyzed regions (n = 18 WTand 35 PRG-1-'- forsm; n = 17 WTand 23 PRG-1-'-for sr; n = 19 WT and 32 PRG-1-'- for so; n represents analyzed dendritic segments; t test; see also Figures S4I-S4K).

(It and I2) Spine length and spine head area were not altered in adult CA1 hippocampal neurons. n = mean of spine length in 36 dendritic segments for WT and of 55 dendritic segments for PRG-1-'- neurons. Spine area was measured in n = 207 spines of WT neurons and in 202 spines of PRG-1-'- neurons. Data were calculated using a t test or a Mann-Whitney test.

(J and K) (J) Dendrite of a LPA2 receptor-deficient (LPA2R-'-) and a PRG-1 -'-/LPA2R-'- CA1 pyramidal neuron. Quantitative analysis (K) shows significantly lower spine density in PRG-1-'- and PRG-1-'-/LPA2R-'- but not in LPA2R-'- neurons (n = 51 forWT, n = 43 for PRG-1-'-, n = 58 for PRG-1-'-/LPA2R-'-, and n = 66 for LPA2R-'-; n represents analyzed dendritic segments; one-way ANOVA with Bonferroni post hoc).

(L) LTP in hippocampal CA1 of LPA2-R-'- and LPA2-R-'-/PRG-1-'- mice. Arrowhead represents stimulation (n = 7 LPA2-R-'- slices and n = 6 LPA2-R-'-/ PRG-1-'- slices; two-way ANOVA). Averaged EPSP samples during baseline (1) and 45 min post LTP induction (2) are shown to the right of LTP graphs. Scale bars: x = 20 ms, y = 0.5 mV.

(M) Spatial memory in the Morris water maze was significantly decreased in PRG-1-'- mice when compared with WT litters (n = 19 WTand n = 12 PRG-1-'- mice; two-way ANOVA).

(N) PRG-1-expressing, LPA2R-'- animals revealed a typical learning behavior but a significantly decreased learning by additional PRG-1 deletion (PRG-1-'-/ LPA2R-'-) (n = 11 LPA2R-'- and n = 6 PRG-1 -'-/LPA2R-'- mice; two-way ANOVA).

(O) Probe trials show that PRG-1-'- and PRG-1 -'-/LPA2R-'- spent significantly less time in the quadrant that previously contained the platform than WT and LPA2R-'- mice, indicating altered spatial memory in PRG-1-'- and PRG-1-'-/LPA2R-'- mice. Dotted line represents chance level (25%) (n = 18 WT, n = 10 PRG-1-'-, n = 11 LPA2R-'- and n = 6 PRG-1 -'-/LPA2R-'- mice; Kruskal-Wallis with Dunn's multiple comparisons test was used for non-parametric data [target] and one-way ANOVA with Bonferroni post hoc was used for normally distributed data).

Values represent mean ± SEM. *p < 0.05, **p < 0.01, ****p < 0.0001; n.s., not significant. Scale bars represent 150 mm in (A) and 1 mm in (B), (D), (F), and (J).

2010) with the PP2A holoenzyme to activate ITGB1. PP2A holo-enzymes, a family of serine-threonine phosphatases involved in controlling a diverse array of cellular processes including cell adhesion, consist of three subunits. The catalytic subunit C and a structural subunit A form the PP2A core enzyme dimer, while the third subunit B is believed to recruit the core enzyme dimer to form the holoenzyme and direct its localization to a particular subcellular compartment and, hence, modulate substrate selectivity and catalytic activity. To date, four unrelated families of B subunits have been identified: B/B55/ PR55/PPP2R2, B'/B56/PR61/PPP2R5, B"/PR72/PPP2R3, and B"'/PR93/SG2NA/PR110/Striatin (Lambrecht et al., 2013). We found that PRG-1 itself is localized in plasma membrane rafts and acts as a scaffolding molecule to organize integrin adhesome assembly via its CBD associated with PPP2R2A (PPP2R2 isoform 1) and accordingly the holoenzyme. The molecular interaction between PRG-1 CBD and PP2A seems to be critical for organization of the adhesome, since disruption of this interaction by CBD deletion from PRG-1 also abolished the association of PRG-1 with other important adhesome components such as src, talin, and paxillin. Importantly, ITGB1 was also shown to be recruited to membrane rafts in a PRG-1-depen-dent fashion and, in line with present data on the adhesome components described here, all molecules were identified in a comprehensive proteome assessment of the PSD (Distler et al., 2014). These data indicate that the PRG-1/PP2A complex forms the core of the PRG-1-organized adhesome while the other components are attracted to proximity associated with the core and distributed around it to activate ITGB1.

It is known that integrin can be physically activated by talin binding to its cytoplasmic domain (Calderwood, 2004). Therefore, there is reason to assume that PRG-1 can activate ITGB1 via recruitment of talin through PP2A to the adhesome, and that ITGB1 itself is recruited to the proximity of membrane rafts where the PRG-1-organized adhesome further stimulates the activation of ITGB1 (Figure S5). Besides talin, src and PP2A were also reported to directly activate ITGB1 (Kim et al., 2004; Sakai et al., 2001). Furthermore, PP2A is able to stabilize ITGB1-mediated cell adhesion not only by dephosphorylation of ITGB1 but also by recruitment of IQGAP1 to ITGB1-Rac, thereby leading to ITGB1 anchoring to F-actin assemblies (Taka-hashi and Suzuki, 2006).

During neuronal development, spines are formed on den-drites reaching the highest density at the peak of synaptogen-esis around the third week of rodent life. A generally accepted view is that neuronal activity modulates spine density, and recent findings show a direct correlation between spines and memory (Yang et al., 2009). This is in line with our findings in PRG-1 mice, which show decreased spine density and impaired spatial memory. Although the role of ITGB1 in spine development has been debated (Huang et al., 2006), recent data show that ITGB1 is critical in spine maturation (Bourgin et al., 2007; Ning et al., 2013). In line with their role for spine remodeling, integrins are critical for synaptic plasticity such as LTP, which is a correlate for memory formation (Chan et al., 2003). Further studies using knockout (KO) mice have identified integrins important for LTP and memory formation showing that ITGB1s (Chan et al., 2006; Huang et al., 2006) in combination with, e.g., a3-integrins (Chan et al., 2007; Kerrisk et al., 2013),

are critically important for specific phases of LTP (Babayan et al., 2012; Chan et al., 2006; Huang et al., 2006). The downstream ITGB1 mechanism involves cofilin inactivation and actin polymerization (Wang et al., 2008). Our data provide evidence for the fact that the dynamics of spine formation as occuring under physiological conditions such as in an enriched environment is dependent on proper PRG-1/active ITGB1 signaling. The role for ITGB1 for dendritic spine plasticity, proper LTP induction, and hippocampal-dependent memory as described by previous work is in line with our data showing that activation of the PRG-1/PP2A/ITGB1 signaling is necessary for proper spine density and spatial memory.

AMPA (a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptors have been shown to play a role in spine growth and maintenance (McKinney et al., 1999), and activation of various NMDA (N-methyl-D-aspartic acid) receptors caused rapid spine retraction (Halpain et al., 1998). However, our own studies on PRG-1 do not provide evidence of changes in the classical molecular machinery at the synapse in PRG-1-KO mice. Immunofluorescent staining and western blot data on PSD-preparations showed no differences in the expression of presynaptic markers, such as synaptophysin, VGlut1, VGat, and GAD67, or postsynaptic markers, such as AMPA receptor subunits (GluR1, GluR2/3, GluR4), PSD-95, and the NMDA receptor subunits NR1 and NR2A/B in PRG-1 and their WT litters. Furthermore, the AMPA/NMDA ratio and the holding current changes on AMPA wash-in did not differ between the WT and PRG-1 KO mice (Trimbuch et al., 2009).

PRG-1 seems to act via LPA2 receptors and has a modulatory role in excitatory transmission in a non-cell-autonomous fashion. Lack of postsynaptic PRG-1 has been shown to lead to an increased release of glutamate from the presynaptic axon terminal mediated via presynaptic LPA receptors (Trimbuch et al., 2009). We were able to rule out, however, that a loss of spines and alteration in synaptic plasticity as observed in the PRG-1-deficient mouse is due to increased synaptic activity. The fact that spine density, LTP, as well as spatial memory were not recovered in LPA2/PRG-1 double-deficient mice, which exhibit normal synaptic activity (Trimbuch et al., 2009), provides evidence for a cell-autonomous action of PRG-1. In turn, genetically modified PRG-1 devoid of its LPA-uptake function but able to interact with PP2A (Trimbuch et al., 2009; Vogt et al., 2016), was still capable of mediating proper spine formation. This appears to be mediated via direct interaction of its intracellular C terminus with PPP2R2A and subsequent activation of ITGB1 regulating spine density.

Together with the functional and behavioral data, our molecular and morphological studies provide evidence for the fact that PRG-1 drives a cell-autonomous signaling pathway involved in the regulation of spine density, and subsequently memory formation.

EXPERIMENTAL PROCEDURES

DNA Constructs and Cell Lines, and Primary Neuronal Cultures

Constructs and establishment of stable expressing cell lines were obtained using standard procedures. GD25 cells were provided by Reinhard Faessler and MEF-SFY by Mirko Schmidt. Primary neuronal cultures were prepared as described by Vogt et al. (2012). For details see Supplemental Experimental Procedures.

Synaptic Junction Preparation

Synaptic junction fraction was isolated as described by Distler et al. (2014). For details see Supplemental Experimental Procedures.

Immunoprecipitation and Immunoblotting

Immunoprecipitation was performed according to standard procedures. Primary antibodies: PRG-1 (1:3,000, custom-made antibody against aa 624-637, NP_808332) (Trimbuch et al., 2009); ITGB1, PPP2R1A (81G5), PPP2R2A (100C1), PPP2C (52F8), Src (32G6), talin (C45F1), and paxillin pY118 (1:1,000), all from Cell Signaling; caveolin-1 (Cav1) (1:2,000) from Sigma; PPP2C pY307 (E155, 1:1,000) from Abcam; and p-actin (1:5,000) from MP Biomedicals. For details see Supplemental Experimental Procedures.

Flow Cytometry

Flow cytometry analyses were carried out on a FACS Canto II cytometer (BD Biosciences). Antibodies are described in detail in Supplemental Experimental Procedures.

Proximity Ligation Assay

PLA was performed using Duolink reagents (Sigma) according to standard protocols. Imaging was performed on the Leica platform of TCS SP8 confocal laser scanning microscope equipped with LAS AF software. For the evaluation of the specific number of dots per cell, pictures of representative cells with positive PLA were used.

Lipid Raft Preparation

Lipid rafts were extracted according to the instruction manual of Caveolae/ Rafts Isolation Kit (Sigma).

Cell-Adhesion Assay

The cell-adhesion assay was adopted from the instruction manual of ECM Cell Adhesion Array Kit (Chemicon). For details see Supplemental Experimental Procedures.

Serine/Threonine Phosphatase Assay

Serine/threonine phosphatase studies were performed using a commercially available assay system (Promega).

In Utero Electroporation and Expression Plasmids for IUE

The in utero electroporation experiments were carried out as described by Trimbuch et al. (2009) in accordance with a protocol approved by the local animal welfare committee.

Immunofluorescence and Confocal Microscopy

Immunofluorescent staining of HEK-293 cells expressing PRG-1, PRG-1-FLAG, or primary neurons transfected with different constructs was performed as described elsewhere (Vogt et al., 2016). Antibodies are described in detail in Supplemental Experimental Procedures. Nuclear staining (DAPI; Invitrogen) was added for better visualization of HEK-293 cells and neurons.

Long-Term Potentiation

LTP was performed as described in Supplemental Experimental Procedures. Morris Water Maze

All experimental procedures were carried out in accordance with the European Communities Council Directive regarding care and use of animals for experimental procedures and were approved by Landesuntersuchungsamt Koblenz, Germany.

Spatial learning and memory were tested by the MWM hidden platform task using the same maze and protocol as described in Supplemental Experimental Procedures.

Statistics

Data analysis was performed using GraphPad Prism 6 (GraphPad Software). After assessing for normal distribution (using the Kolmogorov-Smirnov test), data were analyzed using an unpaired two-tailed Student's t test for normally distributed data or a Mann-Whitney test for non-parametric data. Data normalized to control values (rendering control values as 100) were calculated using a

one-sample t test. For comparison of multiple groups, normally distributed data were assessed using a one-way ANOVA with Bonferroni correction for multiple comparisons, or a Kruskal-Wallis test with Dunn's test for non-parametric data. For comparison of groups containing matched values (LTP + MWM) a two-way ANOVA was used. Differences were considered statistically significant at p < 0.05.

SUPPLEMENTAL INFORMATION

Supplemental Information includes Supplemental Experimental Procedures and five figures and can be found with this article online at http://dx.doi.org/ 10.1016/j.devcel.2016.06.019.

AUTHOR CONTRIBUTIONS

Conceptualization, R.N., J.H., and J.V.; Methodology, J.H., J.V., R.N., T.D., A.V., and K.R.; Investigation, X.L., L.S., Y.L., H.E., H.J., H.C., W.F., B.R., S.R., H.Y., G.L., S.T., J.B., H.B., M.J.S., G.H., G.L., T.M.B., and U.S.; Writing -Original Draft, R.N., J.H., and J.V.; Writing - Review and Editing, J.V. and R.N.; Supervision, R.N., J.H., J.V., and A.V.; Funding Acquisition, R.N., J.H., J.V., A.V., and T.D.

ACKNOWLEDGMENTS

We thank Darragh O'Neill for proofreading the manuscript, and Melanie Pfeifer and Nicolai Schmarowski for technical assistance. We further thank Reinhard Fassler, Mirko Schmidt, and Jerold Chun for providing cell lines and animals, respectively. Data from this study are part of the PhD thesis of X.L., the master's thesis of H.E., and the doctoral thesis of B.R. R.N., together with J.V., and T.D., together with A.V., dedicate this study to their mentor Michael Frotscher, who shaped their thinking on the intimate relation between structure and function in the nervous system. This work was supported by the Deutsche Forschungsgemeinschaft (SFB 1080) to R.N., J.V., J.H., A.V., and T.D. and the European Research Council (ERC-AG "LiPsyD") to R.N. K.R. is a Heisenberg Professor of the DFG (RA1739/4-1) and GFK, and a BIF-PLUS3 fellow.

Received: January 27, 2016 Revised: May 4, 2016 Accepted: June 14, 2016 Published: July 21, 2016

REFERENCES

Babayan, A.H., Kramar, E.A., Barrett, R.M., Jafari, M., Haettig, J., Chen, L.Y., Rex, C.S., Lauterborn, J.C., Wood, M.A., Gall, C.M., et al. (2012). Integrin dynamics produce a delayed stage of long-term potentiation and memory consolidation. J. Neurosci. 32, 12854-12861.

Bourgin, C., Murai, K.K., Richter, M., and Pasquale, E.B. (2007). The EphA4 receptor regulates dendritic spine remodeling by affecting betal-integrin signaling pathways. J. Cell Biol. 178, 1295-1307.

Bourne, J.N., and Harris, K.M. (2008). Balancing structure and function at hippocampal dendritic spines. Annu. Rev. Neurosci. 31, 47-67. Brauer, A.U., Savaskan, N.E., Kuhn, H., Prehn, S., Ninnemann, O., and Nitsch, R. (2003). Anew phospholipid phosphatase, PRG-1, is involved in axon growth and regenerative sprouting. Nat. Neurosci. 6, 572-578. Brindley, D.N., and Waggoner, D.W. (1998). Mammalian lipid phosphate phos-phohydrolases. J. Biol. Chem. 273, 24281-24284.

Broggini, T., Nitsch, R., and Savaskan, N.E. (2010). Plasticity-related gene 5 (PRG5) induces filopodia and neurite growth and impedes lysophosphatidic acid- and nogo-A-mediated axonal retraction. Mol. Biol. Cell 21, 521-537. Calderwood, D.A. (2004). Talin controls integrin activation. Biochem. Soc. Trans. 32, 434-437.

Chan, C.S., Weeber, E.J., Kurup, S., Sweatt, J.D., and Davis, R.L. (2003). Integrin requirement for hippocampal synaptic plasticity and spatial memory. J. Neurosci. 23, 7107-7116.

Chan, C.S., Weeber, E.J., Zong, L., Fuchs, E., Sweatt, J.D., and Davis, R.L. (2006). Beta 1-integrins are required for hippocampal AMPA receptor-dependent synaptic transmission, synaptic plasticity, and working memory. J. Neurosci. 26, 223-232.

Chan, C.S., Levenson, J.M., Mukhopadhyay, P.S., Zong, L., Bradley, A., Sweatt, J.D., and Davis, R.L. (2007). Alpha3-integrins are required for hippocampal long-term potentiation and working memory. Learn. Mem. 14, 606-615.

Coiro, P., Stoenica, L., Strauss, U., and Brauer, A.U. (2014). Plasticity-related gene 5 promotes spine formation in murine hippocampal neurons. J. Biol. Chem. 289, 24956-24970.

Delwel, G.O., de Melker, A.A., Hogervorst, F., Jaspars, L.H., Fles, D.L., Kuikman, I., Lindblom, A., Paulsson, M., Timpl, R., and Sonnenberg, A. (1994). Distinct and overlapping ligand specificities of the alpha 3A beta 1 and alpha 6A beta 1 integrins: recognition of laminin isoforms. Mol. Biol. Cell 5, 203-215.

Distler, U., Schmeisser, M.J., Pelosi, A., Reim, D., Kuharev, J., Weiczner, R., Baumgart, J., Boeckers, T.M., Nitsch, R., Vogt, J., et al. (2014). In-depth protein profiling of the postsynaptic density from mouse hippocampus using data-independent acquisition proteomics. Proteomics 14, 2607-2613. Ethell, I.M., and Pasquale, E.B. (2005). Molecular mechanisms of dendritic spine development and remodeling. Prog. Neurobiol. 75, 161-205. Fischer, M., Kaech, S., Knutti, D., and Matus, A. (1998). Rapid actin-based plasticity in dendritic spines. Neuron 20, 847-854.

Halpain, S., Hipolito, A., and Saffer, L. (1998). Regulation of F-actin stability in dendritic spines by glutamate receptors and calcineurin. J. Neurosci. 18, 9835-9844.

Hanke, J.H., Gardner, J.P., Dow, R.L., Changelian, P.S., Brissette, W.H., Weringer, E.J., Pollok, B.A., and Connelly, P.A. (1996). Discovery of a novel, potent, and Src family-selective tyrosine kinase inhibitor. Study of Lck- and FynT-dependent T cell activation. J. Biol. Chem. 271, 695-701. Huang, Z., Shimazu, K., Woo, N.H., Zang, K., Muller, U., Lu, B., and Reichardt, L.F. (2006). Distinct roles of the beta 1-class integrins at the developing and the mature hippocampal excitatory synapse. J. Neurosci. 26, 11208-11219. Kerrisk, M.E., Greer, C.A., and Koleske, A.J. (2013). Integrin alpha3 is required for late postnatal stability of dendrite arbors, dendritic spines and synapses, and mouse behavior. J. Neurosci. 33, 6742-6752.

Kim, S.M., Kwon, M.S., Park, C.S., Choi, K.R., Chun, J.S., Ahn, J., and Song, W.K. (2004). Modulation of Thr phosphorylation of integrin beta1 during muscle differentiation. J. Biol. Chem. 279, 7082-7090.

Klinghoffer, R.A., Sachsenmaier, C., Cooper, J.A., and Soriano, P. (1999). Src family kinases are required for integrin but not PDGFR signal transduction. EMBO J. 18, 2459-2471.

Lambrecht, C., Haesen, D., Sents, W., Ivanova, E., and Janssens, V. (2013). Structure, regulation, and pharmacological modulation of PP2A phospha-tases. Methods Mol. Biol. 1053, 283-305.

Lenter, M., Uhlig, H., Hamann, A., Jeno, P., Imhof, B., and Vestweber, D. (1993). A monoclonal antibody against an activation epitope on mouse integrin chain beta 1 blocks adhesion of lymphocytes to the endothelial integrin alpha6 beta 1. Proc. Natl. Acad. Sci. USA 90, 9051-9055.

Luo, B.H., Carman, C.V., and Springer, T.A. (2007). Structural basis of integrin regulation and signaling. Annu. Rev. Immunol. 25, 619-647. McDermott, M.I., Sigal, Y.J., Crump, J.S., and Morris, A.J. (2006). Enzymatic analysis of lipid phosphate phosphatases. Methods 39, 169-179. McKinney, R.A., Capogna, M., Durr, R., Gahwiler, B.H., and Thompson, S.M. (1999). Miniature synaptic events maintain dendritic spines via AMPA receptor activation. Nat. Neurosci. 2, 44-49.

Monleon, D., Esteve, V., Kovacs, H., Calvete, J.J., and Celda, B. (2005). Conformation and concerted dynamics of the integrin-binding site and the C-terminal region of echistatin revealed by homonuclear NMR. Biochem. J. 387, 57-66.

Moolenaar, W.H., van Meeteren, L.A., and Giepmans, B.N. (2004). The ins and outs of lysophosphatidic acid signaling. Bioessays 26, 870-881.

Ning, L., Tian, L., Smirnov, S., Vihinen, H., Llano, O., Vick, K., Davis, R.L., Rivera, C., and Gahmberg, C.G. (2013). Interactions between ICAM-5 and beta1 integrins regulate neuronal synapse formation. J. Cell Sci. 126, 77-89. Peeva, G.P., Angelova, S.K., Guntinas-Lichius, O., Streppel, M., Irintchev, A., Schutz, U., Popratiloff, A., Savaskan, N.E., Brauer, A.U., Alvanou, A., et al. (2006). Improved outcome of facial nerve repair in rats is associated with enhanced regenerative response of motoneurons and augmented neocortical plasticity. Eur. J. Neurosci. 24, 2152-2162.

Perrotti, D., and Neviani, P. (2013). Protein phosphatase 2A: a target for anticancer therapy. Lancet Oncol. 14, e229-e238.

Pradhan, S., Alrehani, N., Patel, V., Khatlani, T., and Vijayan, K.V. (2010). Cross-talk between serine/threonine protein phosphatase 2Aand protein tyro-sine phosphatase 1B regulates Src activation and adhesion of integrin alphaIIbbeta3 to fibrinogen. J. Biol. Chem. 285, 29059-29068. Sakai, T., Jove, R., Fassler, R., and Mosher, D.F. (2001). Role of the cytoplasmic tyrosines of beta 1A integrins in transformation by v-src. Proc. Natl. Acad. Sci. USA 98, 3808-3813.

Sala, C., and Segal, M. (2014). Dendritic spines: the locus of structural and functional plasticity. Physiol. Rev. 94, 141-188.

Savaskan, N.E., Brauer, A.U., and Nitsch, R. (2004). Molecular cloning and expression regulation of PRG-3, a new member of the plasticity-related gene family. Eur. J. Neurosci. 19, 212-220.

Schaller, M.D. (2001). Paxillin: a focal adhesion-associated adaptor protein. Oncogene 20, 6459-6472.

Shi, Y., and Ethell, I.M. (2006). Integrins control dendritic spine plasticity in hippocampal neurons through NMDA receptor and Ca2+/calmodulin-dependent protein kinase II-mediated actin reorganization. J. Neurosci. 26, 1813-1822. Sigal, Y.J., McDermott, M.I., and Morris, A.J. (2005). Integral membrane lipid phosphatases/phosphotransferases: common structure and diverse functions. Biochem. J. 387, 281-293.

Sigal, Y.J., Quintero, O.A., Cheney, R.E., and Morris, A.J. (2007). Cdc42 and ARP2/3-independent regulation of filopodia by an integral membrane lipid-phosphatase-related protein. J. Cell Sci. 120, 340-352. Takahashi, K., and Suzuki, K. (2006). Regulation of protein phosphatase 2A-mediated recruitment of IQGAP1 to beta1 integrin by EGF through activation of Ca2+/calmodulin-dependent protein kinase II. J. Cell. Physiol. 208, 213-219.

Takai, A., Murata, M., Torigoe, K., Isobe, M., Mieskes, G., and Yasumoto, T. (1992). Inhibitory effect of okadaic acid derivatives on protein phosphatases. A study on structure-affinity relationship. Biochem. J. 284, 539-544. Thompson, P.M., Tolbert, C.E., and Campbell, S.L. (2013). Vinculin and meta-vinculin: oligomerization and interactions with F-actin. FEBS Lett. 587, 12201229.

Tokumitsu, H., Hatano, N., Tsuchiya, M., Yurimoto, S., Fujimoto, T., Ohara, N., Kobayashi, R., and Sakagami, H. (2010). Identification and characterization of PRG-1 as a neuronal calmodulin-binding protein. Biochem. J. 431, 81-91. Trimbuch, T., Beed, P., Vogt, J., Schuchmann, S., Maier, N., Kintscher, M., Breustedt, J., Schuelke, M., Streu, N., Kieselmann, O., et al. (2009). Synaptic PRG-1 modulates excitatory transmission via lipid phosphate-mediated signaling. Cell 138, 1222-1235.

Unichenko, P., Kirischuk, S., Yang, J.W., Baumgart, J., Roskoden, T., Schneider, P., Sommer, A., Horta, G., Radyushkin, K., Nitsch, R., et al. (2016). Plasticity-related gene 1 affects mouse barrel cortex function via strengthening of Glutamatergic Thalamocortical transmission. Cereb. Cortex 26, 3260-3272.

Velmans, T., Battefeld, A., Geist, B., Farres, A.S., Strauss, U., and Brauer, A.U. (2013). Plasticity-related gene 3 promotes neurite shaft protrusion. BMC Neurosci. 14, 36.

Vogt, J., Glumm, R., Schluter, L., Schmitz, D., Rost, B.R., Streu, N., Rister, B., Bharathi, B.S., Gagiannis, D., Hildebrandt, H., et al. (2012). Homeostatic regulation of NCAM polysialylation is critical for correct synaptic targeting. Cell. Mol. Life Sci. 69, 1179-1191.

Vogt, J., Yang, J.W., Mobascher, A., Cheng, J., Li, Y., Liu, X., Baumgart, J., Thalman, C., Kirischuk, S., Unichenko, P., et al. (2016). Molecular cause and

functional impact of altered synaptic lipid signaling due to a prg-1 gene SNP. EMBO Mol. Med. 8, 25-38.

Wang, X.B., Bozdagi, O., Nikitczuk, J.S., Zhai, Z.W., Zhou, Q., and Huntley, G.W. (2008). Extracellular proteolysis by matrix metalloproteinase-9 drives dendritic spine enlargement and long-term potentiation coordinately. Proc. Natl. Acad. Sci. USA 105, 19520-19525.

Wang, R., Bi, J., Ampah, K.K., Zhang, C., Li, Z., Jiao, Y., Wang, X., Ba, X., and Zeng, X. (2013). Lipid raft regulates the initial spreading of melanoma A375 cells by modulating betal integrin clustering. Int. J. Biochem. Cell Biol. 45, 1679-1689.

Wennerberg, K., Lohikangas, L., Gullberg, D., Pfaff, M., Johansson, S., and Fassler, R. (1996). Beta 1 integrin-dependent and -independent polymerization of fibronectin. J. Cell Biol. 132, 227-238.

Westphal, R.S., Anderson, K.A., Means, A.R., and Wadzinski, B.E. (1998). Asignaling complex of Ca2+-calmodulin-dependent protein kinase IV and protein phosphatase 2A. Science 280, 1258-1261.

Xu, J., Xu, Z., Zhou, J.Y., Zhuang, Z., Wang, E., Boerner, J., and Wu, G.S. (2013). Regulation of the Src-PP2A interaction in tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL)-induced apoptosis. J. Biol. Chem. 288, 33263-33271.

Yang, G., Pan, F., and Gan, W.B. (2009). Stably maintained dendritic spines are associated with lifelong memories. Nature 462, 920-924.

Ziv, N.E., and Smith, S.J. (1996). Evidence for a role of dendritic filopodia in synaptogenesis and spine formation. Neuron 17, 91-102.