The Neuronal Rho-GEF Kalirin-7 Interacts with PDZ Domain–Containing Proteins and Regulates Dendritic Morphogenesis

Neuron, Vol. 29, 229–242, January, 2001, Copyright 2001 by Cell Press

The Neuronal Rho-GEF Kalirin-7 Interacts with PDZ Domain–Containing Proteins and Regulates Dendritic Morphogenesis Peter Penzes, Richard C. Johnson, Rita Sattler,* Xiaoqun Zhang,* Richard L. Huganir,* Vikram Kambampati, Richard E. Mains,‡ and Betty A. Eipper†‡ Department of Neuroscience and * Howard Hughes Medical Institute The Johns Hopkins University School of Medicine Baltimore, Maryland 21205

Summary Spine function requires precise control of the actin cytoskeleton. Kalirin-7, a GDP/GTP exchange factor for Rac1, interacts with PDZ proteins such as PSD95, colocalizing with PSD-95 at synapses of cultured hippocampal neurons. PSD-95 and Kalirin-7 interact in vivo and in heterologous expression systems. In primary cortical neurons, transfected Kalirin-7 is targeted to spines and increases the number and size of spine-like structures. A Kalirin-7 mutant unable to interact with PDZ proteins remains in the cell soma, inducing local formation of aberrant filopodial neurites. Kalirin-7 with an inactivated GEF domain reduces the number of spines below control levels. These results provide evidence that PDZ proteins target Kalirin-7 to the PSD, where it regulates dendritic morphogenesis through Rac1 signaling to the actin cytoskeleton. Introduction Excitatory synapses are localized on dendritic spines (Harris and Kater, 1994). Rearrangements of the actin cytoskeleton are believed to underlie the generation of dendritic spines and filopodia, the likely precursors of spines (Ziv and Smith, 1996; Fiala et al., 1998). Spine shape and motility (Fischer et al., 1998; Harris, 1999; Matus, 1999; Halpain, 2000), receptor anchoring and clustering (Allison et al., 1998), synaptic transmission and long-term potentiation (LTP) (Kim and Lisman, 1999) are all affected by the actin cytoskeleton. The Rho subfamily of Ras-like small GTPases are key regulators of the actin cytoskeleton in many cell types (Kaibuchi et al., 1999). The best characterized Rho subfamily members, RhoA, Rac1, and Cdc42, are involved in regulating the morphology of dendrites and dendritic spines (Luo et al., 1996; Threadgill et al., 1997; Li et al., 2000; Nakayama et al., 2000), making their upstream regulators potential controllers of actin dynamics in dendrites. The immediate activators of Rho-like small GTP binding proteins include GDP/GTP exchange factors (GEFs) of the Dbl family (Kaibuchi et al., 1999). All Rho GEFs contain a † To whom correspondence should be addressed (e-mail: eipper@

uchc.edu). ‡ Current address: Department of Neuroscience, University of Connecticut Health Center, 263 Farmington Avenue, Farmington, Connecticut 06030.

catalytic domain homologous to the oncoprotein Dbl, designated the Dbl homology (DH) domain. Our laboratory previously identified Kalirin, a Dbl family member whose expression in adult rat is restricted to the CNS (Alam et al., 1996, 1997). Kalirin mRNA occurs in several alternatively spliced forms, each yielding a mosaic of protein–protein interaction and signaling domains (Figure 1A). The corresponding Kalirin proteins are localized in different subcellular regions, suggesting they subserve different functions (Johnson et al., 2000). The most abundant isoform in adult brain, Kalirin-7, is enriched in the postsynaptic density (PSD) fraction of cerebral cortex (Penzes et al., 2000). Kalirin is highly homologous to human Trio, an interactor with the transmembrane protein tyrosine phosphatase LAR (Debant et al., 1996). Drosophila Trio is essential for growth cone motility and axon guidance (Awasaki et al., 2000; Bateman et al., 2000; Liebl et al., 2000; Newsome et al., 2000), and the C. elegans Trio-like protein UNC-73 is involved in growth cone guidance and cell migration (Steven et al., 1998) (Figure 1A). Kalirin-7 and its human ortholog Duo (a huntingtin-associated protein-1 interacting protein) (Colomer et al., 1997) are multidomain proteins with a Sec14p-like putative lipid binding domain, nine spectrin-like repeats, tandem DH and pleckstrin homology (PH) domains, and a C-terminal end that includes a T/S-X-V sequence, a putative PDZ recognition motif. Kalirin-7 catalyzed nucleotide exchange on Rac1 in vitro and transfection of Kalirin-7 into NIH3T3 fibroblasts altered the actin cytoskeleton in a manner similar to Rac1 (Penzes et al., 2000). Thus, it is very likely that Kalirin-7 regulates remodeling of the actin cytoskeleton in neurons through activation of Rac1. To better understand how Kalirin-7 regulates the actin cytoskeleton at the PSD and in dendritic spines, we first sought to identify proteins that might localize Kalirin-7 to this site. To identify protein partners of Kalirin-7, we performed a yeast two-hybrid screen with the C terminus of Kalirin-7 as bait and identified 16 different known proteins, all containing PDZ domains, including the synaptic protein PSD-95. We further explored the interaction of Kalirin-7 with PSD-95, as a representative PSD interactor, and found that Kalirin-7 and PSD-95 interacted in vitro and in vivo. When transfected into cortical neurons, Kalirin-7 was specifically targeted to dendritic spine-like structures and induced generation of more and larger spine-like protrusions. In contrast, deletion of the C terminus of Kalirin-7 eliminated targeting of exogenous mutant protein to spines, while a GEF-inactive mutant reduced the number of spine-like structures. These results suggest that Kalirin-7 is a regulator of actin dynamics in dendritic spines, thus regulating their morphogenesis. Results The Putative PDZ Binding Motif of Kalirin-7 Interacts with Multiple PDZ Domain–Containing Proteins We previously reported that Kalirin-7, unlike the other Kalirin isoforms, was significantly enriched in the PSD

Neuron 230

Figure 1. The C Terminus of Kalirin-7 Interacts with Multiple PDZ Domain–Containing Proteins (A) Domain structures of Kalirin-7, -9, and -12, and related proteins human Trio, Drosophila Trio, and C. elegans UNC-73A. The indicated domains are: Sec14, Sec14p-like putative lipid binding domain; DH, Dbl-homology; PH, pleckstrin homology; SH3, Src-homology 3; FnIII, fibronectin III–like; Ig, immunoglobulin-like. The regions used to generate antibodies used in this study (characterized in Figure 2A) are indicated by bars; asterisk indicates point mutation in the DH domain of the GEFKO mutant. (B) The yeast two-hybrid screen. The C-terminal 30 residues of Kalirin-7 were used as bait for screening a rat hippocampal/cortical cDNA library (dark). The boldfaced 20 residue peptide sequence is unique to Kalirin-7; the putative PDZ domain recognition motif is underlined. (C) Proteins identified in the yeast two-hybrid screen. Domain structures of the proteins and relative position of the sequenced interactor clones are shown (thin lines below each protein). The represented domains are: PDZ, PSD95-Dlg-ZO-1 homology domain; SH3, Src-homology 3; WW domain; GuK, guanylate kinase–like; PTB, phosphotyrosine binding; PP, Pro-rich region; Ras binding and F-actin binding regions were experimentally determined and described in the cited references. Protein domains were determined by the program SMART (Schultz et al., 1998) in addition to the cited reference for each protein.

fraction of rat cerebral cortex and localized to punctate structures along processes of cultured cortical neurons (Penzes et al., 2000). The TYV sequence at the C terminus of Kalirin-7 (Figure 1B) suggests binding to PDZ domains (S/T-X-V) (Songyang et al., 1997). To identify proteins that interact with the C terminus of Kalirin-7 (K7-CT), we employed the yeast two-hybrid system (Figure 1B). We screened a cDNA library prepared from the hippocampus/cortex of 3-week-postnatal rat pups. We isolated 21 individual cDNAs, 16 of which encoded proteins containing at least one PDZ domain (Table 1; Figure 1C). In each case, the Kalirin-7 interacting fragment (Figure 1C, underlined) included a PDZ domain. These interacting proteins were classified into several functional groups (Table 1). Virtually all of the proteins identified are known to bind specific transmembrane proteins (TM): PSD-95, SAP102, SAP97, Chapsyn-110, ZO-1, ZO-2, S-SCAM, spinophilin, PICK-1, X11, GIPC, syntenin, MUPP1. Many are proteins known to be synap-

tically localized (S): PSD-95, SAP102, SAP97, Chapsyn110, ZO-1, S-SCAM, spinophilin, PICK-1. Several of the proteins are known to be involved in receptor clustering (RC): PSD-95, SAP102, SAP97, Chapsyn-110, PICK-1, AF-6. A smaller subset of the proteins is associated with protein trafficking (T): X11/Mint, GIPC, syntenin. There are three actin binding proteins with which Kalirin-7 interacts (AB): neurabin, spinophilin, AF-6. Several of the proteins were detected at cell–cell junctions (CC): SAP97, ZO-1, ZO-1, AF-6, spinophilin. The remaining five clones encoded proteins with no obvious relationship to the bait and were considered false positives. These clones were myelin basic protein, arachidonate lipoxygenase, hemoglobin ␣ chain, prealbumin, and ferritin light chain subunit. Kalirin-7 Interacts with PSD-95 In Vivo and In Vitro In adult rats, the expression of Kalirin mRNA is restricted to the CNS, and Kalirin-7 is enriched in the PSD fractions

Kalirin Regulates Spine Morphology 231

Table 1. Proteins that Interact with the C-Terminal Region of Kalirin-7 Protein

Known Functions

Propertiesa

Number of Times Entrez

Referenceb

PSD-95

clustering NMDA receptors, K channels, kainate receptors, binds nNOS, SynGAP, neuroligins, CRIPT, SAPAPs, Citron, Shank binds NMDA receptors, GluR1 binds NMDA receptors, AMPA receptors (GluR1), K⫹ channel clustering, in cadherin-based junctions in epithelia clustering NMDA receptors in tight junctions in epithelia, interacts with ZO-2, ZO-3, AF-6, occludin, actin in tight junctions in epithelia, interacts with ZO-1, ZO-3, AF-6, occludin, actin binds NMDA receptors, neurologin, GKAPs, nRapGEP, delta-cetenin unknown, similar to S-SCAM binds F-actin, p70S6K binds F-actin, in dendritic spines, inhibits protein phosphatase-1 in tight and adherens junctions in epithelia, clustering Eph receptors, regulates cell–cell junction formation during embryogenesis, binds Ras, ZO-1, ZO-2, F-actin targeting C. elegans GluR and LET23 (EGF receptor), ␤-APP trafficking, Munc-18-1 interactor, in complex with CASK, Velis binds RGS-GAIP on clathrin-coated vesicles, binds semaphorin, neurophilin-1 binds syndecan, EphB, pro-TGF trafficking in Golgi, apical early endocytic compartment clustering AMPA receptors, Eph receptor, binds PKC binds 5HT2C receptor

S, RC, TM

3

P31016

Cho et al., 1992

S, RC, TM 4 S, RC, CC, TM 1

Q62936 Q62696

Muller et al., 1996 Muller et al., 1995

S, CC, TM S, CC, TM

1 2

Q63622 P39447

Kim et al., 1996 Itoh et al., 1993

S, CC, TM

1

AAD19964 Jesaitis et al., 1994

S, TM

3

AAC31124 Hirao et al., 1998

SAP102 SAP97

Chapsyn-110 ZO-1 ZO-2 S-SCAM MAGl-1 Neurabin Spinophilin AF-6/afadin

X11␣/Mint-1

GIPC Syntenin PICK-1 MUPP1

ND 1 AB 3 AB, CC, TM, S 2

AAB91996 Dobrosotskaya et al., 1997 U72994 Nakanishi et al., 1997 AAC05183 Allen et al., 1997

AB, RC, CC

1

U83230

Mandai et al., 1997

T, TM

1

O35430

Okamoto and Sudhof, 1997

T, TM

1

AAC67549 De Vries et al., 1998

T, TM

7

AAB97144 Grootjans et al., 1997

RC, S, TM TM

1 3

NP032863 Staudinger et al., 1995 CAA04681 Ullmer et al., 1998

a

S, synaptic localization; CC, at cell–cell junctions; TM, transmembrane protein binding; AB, filamentous actin binding; RC, involved in receptor clustering; T, involved in trafficking. b Due to space limitations, only the article describing the identification of the protein, linked to the protein sequence in GenBank, is listed; ND, not determined.

(Penzes et al., 2000). Therefore, we decided to explore the interaction of Kalirin-7 with PSD-95, a well-characterized protein with a similar localization. The interactor fragments of PSD-95 from the yeast two-hybrid screen contained nonoverlapping PDZ domains, indicating that at least one of the first two and the third PDZ domain interact with Kalirin-7. Similarly, other molecules such as NR2 (Kornau et al., 1995) and Syn-GAP (Kim et al., 1998) bind to multiple PDZ domains of PSD-95. We employed two antibodies to detect Kalirin: a generic antibody recognizing all isoforms, designated “Kalirin-spec,” and an antibody generated to the C terminus of Kalirin-7, which recognizes only this isoform, designated “Kalirin-7” (Figure 2A). To evaluate the distribution of total Kalirin-7 relative to total PSD-95 in each fraction from a PSD preparation, we utilized Western blotting (Figure 2B). A significant portion of the total Kalirin-7 is present in the Triton-insoluble fraction (1-Triton-P) of a crude synaptosomal fraction. This fraction also contains the largest portion of PSD-95. The fraction insoluble after a second Triton extraction contained essentially all of the Kalirin-7 and PSD-95. The fraction soluble after extraction in Sarcosyl (Triton/Sarcosyl-S), contained almost all of the Kalirin-7, while a portion of PSD-95 was not extracted, remaining in the “PSD core” (Triton/Sarcosyl-P). A significant portion of the total Kalirin-7 cofractionates with PSD-95 in the PSD, although Kalirin-7 is not as strongly associated with the PSD as PSD-95.

To test whether Kalirin-7 and PSD-95 interact with each other in vivo, we performed coimmunoprecipitation experiments using extracts from a crude synaptosomal fraction (P2, in Figure 2B). Proteins were solubilized with sodium dodecyl sulfate (SDS) in the presence of K7-CT peptide to prevent postsolubilization binding of PSD-95 to the C terminus of Kalirin-7. Incubation with Kalirin-spectrin antibody resulted in coprecipitation of PSD-95 along with Kalirin-7 (Figure 2C). Negative controls included use of preimmune serum or antibody blocked with antigen; under these conditions, PSD-95 and Kalirin-7 were not immunoprecipitated. Since multiple forms of Kalirin were precipitated with the Kalirin-spectrin antibody, we used glutathione S–transferase (GST) fusion proteins to test whether PSD-95 bound specifically to the unique C terminus of Kalirin-7. We incubated an extract of the synaptosomal fraction of rat cerebral cortex with a GST-fusion protein that included only the C-terminal 30 residues of Kalirin-7 (GST-K7-CT) (Figure 2D). PSD-95 bound to GST-K7-CT. Inclusion of K7-CT peptide during the incubation disrupted this binding. At the same time, no PSD-95 binding to GST alone was detected. To disrupt the PDZ binding motif of Kalirin-7 and to test whether the interaction occurred through the C-terminal residue of Kalirin-7, we mutated the last residue from valine to aspartate (K7-V/D). As predicted for a PDZ-mediated interaction, the GST-K7-V/D fusion protein did not bind PSD-95.

Neuron 232

Figure 2. Kalirin Interacts with PSD-95 In Vitro and In Vivo (A) Comparison of Kalirin antibodies. Extracts of cerebral cortex and hippocampus (25 ␮g protein) were analyzed. Kalirin was detected by Western blotting with the Kal-spec and Kalirin-7 antibodies (Figure 1A). (B) Relative abundance of Kalirin and PSD95 in rat cerebral cortex postsynaptic density fractions. Aliquots representing 0.04% of S1, S2, and P2, or 0.4% of each PSD fraction (S and P) were separated by SDS–PAGE and analyzed by Western blotting using Kalirinspectrin (upper) and PSD-95 (lower) antibodies. Fractions indicated are: S1, postnuclear supernatant; S2, supernatant from the crude synaptosomal fraction; P2, crude synaptosomal pellet. The P2 fraction, further separated on a discontinuous sucrose gradient, was extracted in 0.5% Triton X-100 (1-Triton) and centrifuged to give supernatant (S) and pellet (P). Pellet P was split in half and reextracted in 0.5% Triton X-100 (2-Triton) or 3% Sarcosyl (Triton-Sarcosyl) and each extract was centrifuged yielding a supernatant (S) and pellet (P). (C) PSD-95 is coimmunoprecipitated with Kalirin-7. A rat cerebral cortex synaptosomal fraction (P2) was solubilized in 2% SDS in the presence of K7-CT peptide. After neutralization with Triton X-100, samples were incubated with Kalirin-spectrin antibody. Negative controls were preimmune serum and preblocked antibody. Samples representing 10% of the input and 40% of the bound fractions were analyzed by Western blotting with a PSD-95 monoclonal antibody (upper) and Kalirin-7-specific antibody (lower). (D) PSD-95 interacts with the C terminus of Kalirin-7. GST fusion proteins tested included GST-K7-CT (in absence or presence of K7-CT peptide), GST-K7-V/D, or GST. The Sarcosyl extract of a rat cerebral cortex synaptosomal fraction (500 ␮g) was neutralized with Triton X-100 and incubated with the indicated GST fusion protein (5 ␮g). Aliquots of the input (90%) and bound (60%) fractions were analyzed by Western blotting with a PSD-95 monoclonal antibody.

Thus, Kalirin-7 interacts with PSD-95 in rat brain synaptosomal membranes and this interaction is mediated by the C-terminal residues of Kalirin-7. Kalirin-7 Colocalizes with PSD-95 and Filamentous Actin at Synaptic Sites in Cultured Hippocampal Neurons Having shown that Kalirin-7 interacts with PSD-95, we explored the localization of Kalirin-7 in neurons that also express PSD-95 (Figure 3). We visualized Kalirin-7 in 24day-old cultured primary hippocampal neurons, which have mature synapses with fully differentiated PSDs (Goslin and Banker, 1991). Kalirin-7 was strikingly abundant in punctate structures arrayed along dendrites (Figure 3A). Kalirin-7 colocalized with PSD-95 in the majority of these structures. This staining pattern suggests the presence of Kalirin-7 in synapses on dendritic spines, and colocalization with PSD-95 is consistent with enrichment in PSDs. Kalirin-7 was visualized in a larger number of punctate structures than PSD-95, suggesting that many of these have Kalirin-7 but not PSD-95 (Figure 3A, small arrows). The higher magnification image re-

veals overlapping (arrowhead) but not identical distributions for Kalirin-7 and PSD-95, consistent with their different behavior upon subcellular fractionation. In addition to its synaptic localization, Kalirin-7 localizes to punctate structures in the dendritic shafts, to processes without PSD-95 (most likely axons) (Figure 3A, arrows), and to cell bodies (S). This is consistent with the subcellular localization of Kalirin-7 established by biochemical methods (Penzes et al., 2000) and with its multiple interactors. Kalirin regulates remodeling of the actin cytoskeleton when transfected into several cell types (Mains et al., 1999; Penzes et al., 2000), and Kalirin-7 may perform a similar function in dendritic spines. To compare the localization of F-actin and Kalirin-7, we stained cultured hippocampal neurons with FITC-conjugated phalloidin and the Kalirin-7 antibody (Figure 3B). F-actin exhibited a patchy distribution along neuronal processes and was significantly enriched in the dendritic spines compared to the nearby shaft. Kalirin-7 localization overlapped that of F-actin in most spines (Figure 3B, arrowheads). To confirm the synaptic localization of Kalirin-7, we simulta-

Kalirin Regulates Spine Morphology 233

Figure 3. Kalirin-7 Colocalizes with PSD-95 and Filamentous Actin at Synapses and in Dendritic Spines of Cultured Hippocampal Neurons Twenty-four-day-old cultured hippocampal neurons were immunostained simultaneously with the affinity-purified Kalirin-7 antibody and PSD95 monoclonal antibody (A), FITC-phalloidin (B), or synaptophysin monoclonal antibody (C). Examples of sites of colocalization of Kalirin-7 and PSD-95, filamentous actin, or synaptophysin are indicated by large arrowheads on inset figures. Small arrows in (A) indicate fibers staining for Kalirin-7 alone. Scale bar, 10 ␮m.

neously stained neurons for Kalirin-7 and synaptophysin, a protein enriched in presynaptic terminals (Figure 3C). Kalirin-7 colocalized with synaptophysin at a large fraction of synapses (Figure 3C, arrowheads). The shapes of the structures visualized by Kalirin-7 and synaptophysin were generally different. While synaptophysin stained elongated ellipsoidal structures, Kalirin-7 stained more punctate structures. Kalirin-7 Interacts with PSD-95 in Transfected NIH3T3 Fibroblasts To examine the interaction of Kalirin-7 with PSD-95 in a cellular system, we transiently transfected fibroblasts with Kalirin-7 and PSD-95, individually or together. Transient expression of Kalirin-7 (Figure 4A) resulted in a phenotype characteristic for activation of Rac1, as induced by the constitutively active mutant Rac1Q61L (Figure 4B) (Penzes et al., 2000). Most of the Kalirin-7

appeared to be cytoplasmic, with little staining at the membrane in lamellipodia. Staining of filamentous actin revealed a reduction in the number of stress fibers in Kalirin-7-expressing cells compared to control (nontransfected) cells. When expressed alone in NIH3T3 cells, PSD-95 was mostly cytosolic and was without effect on cell morphology (Figure 4C). In cells expressing both Kalirin-7 and PSD-95, a significant fraction of each protein was membrane associated (Figure 4D). We detected colocalization of the two proteins at the edges of lamellipodia in the majority of cells (Figures 4D–4F, arrows). The sites of colocalization of Kalirin-7 and PSD-95 were also enriched in F-actin, (Figures 4E and 4F, open arrows). These results suggest that PSD-95 interacts with Kalirin-7 at specific sites; these sites are marked by local concentrations of the two proteins along with F-actin. A C-terminally truncated Kalirin-7 (Kal7-⌬CT) was used to evaluate the impor-

Neuron 234

Figure 4. Kalirin-7 Interacts with PSD-95 in Transfected Fibroblasts (A–H) NIH3T3 fibroblasts transiently expressing Myc-Kalirin-7 (A), PSD-95 (B), Myc-Kalirin-7, and PSD-95 (D–F), Myc-Kal7-⌬CT (G) or MycKal7-⌬CT and PSD-95 (H) were serum starved over night and stained with Kalirin-spec antibody, PSD-95 monoclonal antibody, and FITCphalloidin, alone, or in combinations of two (as indicated on pictures). Arrows show membrane-associated clusters of Kalirin-7 and PSD-95, and open arrows indicate F-actin in these clusters. Scale bar, 50 ␮m. Quantitative analysis. (I) Plot shows the relative fraction of the different phenotypes. Sites scored as concave edges (pound sign), extensive lamellipodia (L), protrusions (p), and lamellipodia on protrusions (lp) are indicated. (J) Plot of average cell shape complexities represented by P2A values (circle has minimum value of 4␲.). The differences between Kalirin-7 and nontransfected (nt) (p ⬍ 0.001) and between Kalirin-7 and Kalirin-7⫹PSD-95 (p ⬍ 0.0027) were statistically significant. Differences between Kalirin-7 plus PSD-95 and control cells were not significant (p ⫽ 0.094). Bars represent averages of ⵑ25–40 cells in two different experiments; error bars represent standard deviations between values for the same type cells. Cell shapes were measured with the IP Labs 3.0 software (Scanalytics).

tance of the PDZ domain to the observed interaction. Cells expressing Kal7-⌬CT were circular, with extensive lamellipodia, similar in morphology to Kalirin-7-expressing cells. However, cotransfection with PSD-95 did not dramatically alter their phenotype. Both proteins had a diffuse localization, with limited colocalization in ruffles, suggesting a reduced interaction of PSD-95 with Kal7⌬CT. The interaction between Kalirin-7 and PSD-95 is partially mediated by the C terminus of Kalirin-7. The cellular phenotype induced by Kalirin-7 is clearly altered by coexpression of PSD-95 (Figures 4I and 4J). Cells expressing Kalirin-7 alone were flat (Figure 4A), with simple convex edges and extensive lamellipodia (L), while cells expressing PSD-95 alone (Figure 4C) or both proteins exhibited few lamellipodia (Figure 4I). Cells expressing both proteins exhibited complex jagged edges, with at least two large indentations per cell (Figure 4, pound sign). A large fraction of the cells expressing PSD-95 alone exhibited longer protrusions, while only a small percentage of Kalirin-7-expressing cells showed such structures. Approximately two-thirds of

the cells expressing both proteins showed cellular protrusions (p, protrusions), and many formed lamellipodia at the distal end of these protrusions (lp, lamellipodia on protrusions). We assessed the shape of each type of cell quantitatively by plotting the cell shape complexity (P2A values, ratios between perimeter2 and area of cells) (Figure 4J). Cells expressing Kalirin-7 alone had smaller P2A values than cells expressing both proteins, due to the presence of jagged edges, indentations, and protrusions on the cells simultaneously expressing Kalirin-7 and PSD-95. Differences between Kal7-⌬CT or Kal7⌬CT/PSD-95 and Kalirin-7 or nontransfected cells were not significant; the phenotypes of cells expressing Kal7⌬CT or Kal7-⌬CT/PSD-95 were intermediate between those of Kalirin-7 and Kalirin-7/PSD-95, suggesting that the interaction between Kalirin-7 and PSD-95 was only partially eliminated. PSD-95 Reduces Activation of Rac1 by Kalirin-7 The different cellular phenotype resulting from coexpression of PSD-95 with Kalirin-7 suggested that PSD-

Kalirin Regulates Spine Morphology 235

Figure 5. PSD-95 Modulates the GEF Activity of Kalirin-7 (A) Myc.Kalirin-7, PSD-95, Myc.Kalirin-7 and PSD-95, Myc.Kal7-⌬CT, or Myc.Kal7-⌬CT and PSD-95 were transiently expressed in HEK293 PEAK-Rapid cells. Cells were serum starved overnight, and activation of endogenous Rac1 was measured 24 hr after transfection using the PBD assay. Bound Rac1 protein was analyzed by Western blotting with Rac1 monoclonal antibody. Controls were mock-transfected cells (control) and cell homogenates treated with GTP␥S. Expression of transfected proteins was evaluated by Western blotting with the Myc antibody for the various Kalirin proteins or PSD-95 antibody for PSD-95. (B) Plot of averages of Rac1 activation from two or three experiments; error bars are standard deviations. Values for Rac1 activation were normalized to the expression levels of the Myc-tagged Kalirin proteins. The differences between Kalirin-7 versus control (asterisk, p ⬍ 0.025), Kalirin-7 versus Kalirin-7 plus PSD-95 (double asterisk, p ⬍ 0.01), and Kal7-⌬CT versus Kal7-⌬CT plus PSD-95 (triple asterisk, p ⬍ 0.01576) were statistically significant (Student’s t test).

95 might affect the ability of Kalirin-7 to activate Rac1. To test this hypothesis, we transiently transfected pEAKRapid HEK293 cells with Kalirin-7, PSD-95, or Kalirin-7 and PSD-95. We measured the activation of Rac1 using the p21 binding domain (PBD) of PAK1 (Benard et al., 1999). We monitored Kalirin-7 and PSD-95 expression using Myc and PSD-95 antisera, respectively (Figure 5A). Expression of Kalirin-7 results in activation of Rac1 (Figure 5B). In contrast, Rac1 was not activated to the same extent in cells coexpressing Kalirin-7 and PSD-95 (Figure 5B, columns 3 and 4). Rac1 activation in control cells was indistinguishable from Rac1 activation in cells expressing PSD95 alone or Kalirin-7 and PSD-95. Cotransfection of cells with Kal7-⌬CT and PSD-95 yielded less Rac1 activation than observed in cells expressing Kal7-⌬CT alone, suggesting that PSD-95 interacts with other regions of Kalirin-7, in addition to the C terminus. Kalirin-7 Overexpressed in Cortical Neurons Is Targeted to Dendritic Spine-Like Structures and Affects Their Morphology To evaluate the function of Kalirin-7 in neurons, we transfected 1-week-old primary cortical neurons with MycKalirin-7 together with GFP. The neurons expressing Kalirin-7 and GFP (Figures 6A and 6B) had dendritic morphologies different from neurons expressing GFP alone (Figures 6C and 6D). The dendrites of the Kalirin-7expressing neurons had a large number of protrusions with larger sizes and more complex shapes than control neurons. At this stage, neurons have mostly dendritic filopodia, and few “true” spines, as revealed by transfection with GFP alone (Figures 6C and 6D) (Papa et al., 1995; Ziv and Smith, 1996; Fiala et al., 1998). For ease of discussion, we refer to the structures induced both proximally and distally on dendrites by overexpression of Kalirin-7 and its mutants as “spine-like structures”. A significant proportion of the exogenous Kalirin-7 was localized to multiple punctate structures distributed along the dendrites (Figure 6B, arrows). The dendritic shafts contained small puncta (pound sign), while large clusters that stained for Myc were often offset from the shafts (asterisk). Better visualization of neuronal morphology with an antibody to GFP confirmed targeting of Kalirin-7 to dendritic protrusions (arrowheads). There

was much less Myc-Kalirin-7 in the dendritic shaft than at the tips of these protrusions. Overexpression of Kalirin-7 resulted in the generation of spine-like structures of divergent morphology (Figure 6E): massive lamellipodia (a and h), long and highly branched (b), short and highly branched (c and d), small and very densely placed (e), resembling small growth cones (f), and mushroom shaped (g). Most were dramatically different from the dendritic filopodia on control neurons (Figure 6D). The C Terminus and GEF Activity Are Essential for Mediating the Localization and Morphological Effects of Kalirin-7 To evaluate the role of the C-terminal PDZ binding motif of Kalirin-7 in its localization and function in neurons, we expressed Myc-Kalirin-7-⌬CT together with GFP in cortical neurons (Figures 7A and 7B). Myc-Kalirin-7-⌬CT was concentrated in the cell body and was less abundant than Myc-Kalirin-7 in dendrites. The somas of a large fraction of the Kalirin-7-⌬CT-expressing neurons exhibited abundant aberrant neurites (an), a feature less often detected in nontransfected neurons. These aberrant neurites were much longer (10–20 ␮m) than the filopodia on dendrites of control or Kalirin-7-expressing neurons. Myc-Kalirin-7-⌬CT behaved like a soluble protein, diffusely distributed in the dendritic shaft, but not entering the filopodia (Figure 7B, ampersands). The morphology of the dendritic filopodia on Myc-Kalirin-7-⌬CTexpressing neurons (Figure 7B, arrows) is largely indistinguishable from that of the control neurons (Figure 6D). These results indicate that Kalirin-7 is targeted to dendritic spines by interactions mediated by its C terminus. Left in the cell soma, Kalirin-7-⌬CT causes aberrant process formation. To evaluate the role of the GEF activity of Kalirin-7 in its effects on dendritic morphology, we replaced Lys1393, an amino acid in the DH domain of Kalirin-7, with Ala to generate Kalirin-7-GEFKO. Mutation of the homologous residue in the first DH domain of Trio (Lys1372) significantly reduced its GEF activity (Liu et al., 1998). We expressed Myc-Kalirin-7-GEFKO in cortical neurons together with GFP (Figures 7C and 7D). In most neurons, Myc-Kalirin-7-GEFKO was expressed at lower levels than the other Myc-tagged proteins; we included in our study

Neuron 236

Figure 6. Kalirin-7 Overexpression in Cultured Cortical Neurons Affects the Morphology of Dendritic Spine-Like Structures One-week-old dissociated primary cortical neurons were transfected with GFP alone (C and D) or together with Myc-Kalirin-7 (A, B, and E) (indicated by vertical text). After 42 hr, neurons were stained with GFP polyclonal antibody and a Myc monoclonal antibody. The experiment was performed four times, each condition was done in duplicate, with essentially identical results. (A) Exogenous Kalirin-7 is targeted to dendritic spine-like structures. (B) High magnification from (A). Arrows, spine-like structures; asterisks, Myc-Kalirin-7 targeted to spine-like structures; pound sign, small puncta. (C) Control neurons transfected with GFP. (D) High magnification image from (C). (E) High magnification images of morphologies of spine-like structures induced by expression of Kalirin-7. Scale bars: 20 ␮m (A and C); 10 ␮m (B, D, and E).

only those neurons expressing Myc-Kalirin-7-GEFKO at levels comparable to other Myc-tagged Kalirin proteins. Myc-Kalirin-7-GEFKO was detected in dendrites, but dendritic staining for Myc was weaker than in Myc-Kalirin-7-expressing neurons, indicating a potentially impaired dendritic trafficking. Myc-Kalirin-7-GEFKO staining was often (Figure 7D, asterisks), but not always, present in the outer portion of the dendritic filopodia; however, the distribution of Myc-Kalirin-7-GEFKO was not as clearly punctate as the distribution of Kalirin-7. The morphology of the dendritic filopodia on Myc-Kalirin-GEFKO expressing neurons (Figure 7D, arrows) is similar to that of control neurons. Thus, the GEF activity of Kalirin-7 is important in its ability to alter dendritic morphology. To quantitatively evaluate the morphological features induced by Kalirin-7 and its mutants, we plotted the fraction of neurons displaying unusually shaped or odd spinelike structures (Figure 8A). A significantly higher fraction of

the Kalirin-7 transfected neurons expressed odd dendritic protrusions. Expression of Myc-Kalirin-7-⌬CT resulted in a significantly larger fraction of neurons bearing dense, aberrant neuritic filopodia on their cell bodies (Figure 8B). The sizes of dendritic filopodia were also measured (Figure 8C). Kalirin-7-expressing neurons had more spinelike structures with larger areas while neurons expressing Kal7-⌬CT or Kal7-GEFKO were similar to controls. We also measured the linear density of spine-like structures on dendrites (Figure 8D); neurons expressing Kalirin-7 had an increased density of spine-like structures on their dendrites (Figure 8D). In contrast, the linear density of spine-like structures on the dendrites of neurons expressing Kal7-GEFKO was even lower than control neurons. In conclusion, overexpression of Kalirin-7 resulted in larger and more spines, while expression of Kal7-GEFKO resulted in fewer spines, suggesting that Kal7-GEFKO may be interfering with endogenous processes regulating spine density.

Kalirin Regulates Spine Morphology 237

Figure 7. Overexpression of Kalirin-7 Mutants in Cultured Cortical Neurons One-week-old dissociated cortical primary neurons were transfected with GFP together with Myc-Kalirin-7-⌬CT (A and B), or Myc-Kalirin-7GEFKO (C and D) (indicated by vertical text). After 42 hr, neurons were stained for GFP and Myc. Scale bars represent 20 ␮m (A and C) or 10 ␮m (B and D); arrows indicate spine-like filopodia; an, aberrant neurites; ampersands, filopodia with little Myc-Kal7-⌬CT; asterisks, localization of Kal7-GEFKO to spine-like filopodia.

Discussion Kalirin-7 as a Regulator of the Postsynaptic Actin Cytoskeleton The structure and shape of dendritic spines is important for synaptic function (Harris and Kater, 1994; Harris, 1999). Several recent studies strongly support the hypothesis that activity-dependent changes in spine number contribute to variations in synaptic strength. Induction of LTP in the hippocampus results in an increase in the number of dendritic spines (Engert and Bonhoeffer, 1999), with the appearance of multiple spines contacting the same presynaptic terminal (Toni et al., 1999). Similarly, tetanus induces the sprouting of new filopodial protrusions on dendrites (Maletic-Savatic et al., 1999). Dendritic spines are mobile (Fischer et al., 1998), and remodeling of the actin cytoskeleton is believed to underlie changes in spine shape and number (Fischer et al., 1998; Halpain et al., 1998; Harris, 1999; Matus, 1999; Van Rossum and Hanisch, 1999). Postsynaptic filamentous actin also affects synaptic efficiency. For example, F-actin in spines is involved in the differential localization of NMDA and AMPA receptors (Allison et al., 1998) and affects AMPA-mediated transmission (Kim and Lisman, 1999). Changes in spine shape during synaptic activity are NMDA receptor–dependent (Maletic-Savatic et al., 1999) and treatment of cultured hippocampal neurons with NMDA results in rapid loss of spines concomitant with loss of F-actin (Halpain et al., 1998). Taken together, these data suggest that glutamate receptor activation modulates cytoskeletal dynamics in dendritic spines (Van Rossum and Hanisch, 1999). Early in postnatal de-

velopment, dendrites emit long filopodia, which grow and retract rapidly, finally making connections with axons (Ziv and Smith, 1996; Fiala et al., 1998; Harris, 1999). These filopodia are believed to be the developmental precursors of adult spines (Luscher et al., 2000). During adulthood, newly emerged filopodia may form new synaptic contacts with preexisting presynaptic terminals. An important role for Rho-like GTPases in regulating dendritic and spine morphology is suggested by several studies (Threadgill et al., 1997; Li et al., 2000; Nakayama et al., 2000). Rac1 affects the size, shape, and number of dendritic spines, as well as the number of synapses located on each spine (Luo et al., 1996). Overexpression of dominant-negative Rac1 results in a reduction in the number of spines, while activated Rac1 alters the morphology of spines (Nakayama et al., 2000). Our experiments demonstrate that Kalirin-7 is targeted to dendritic spine-like structures through its C terminus, the same region that interacts with PDZ domain–containing proteins. The C terminus of Kalirin-7 may interact with PDZ proteins already localized to these sites. Excessive activation of Rac1 may induce exaggerated alterations of the shape of these structures. Some of the morphologies of these dendritic protrusions are detected in wild-type neurons as well, but others are very unusual. The spine-like structures induced by Kalirin-7 resemble those induced by activated Rac1 (Nakayama et al., 2000). Controlled activation of endogenous Kalirin-7 is presumably used to regulate spine morphogenesis. The targeting of wild-type Kalirin-7 to dendritic spines requires the C terminus; elimination of the PDZ binding

Neuron 238

Figure 8. Quantitation of the Morphological Features of Dendrites of Transfected Neurons (A) Plot of the ratio of neurons with unusual spine morphologies. The differences between Kalirin-7 versus GFP control (asterisk, p ⬍ 0.0112) were statistically significant by Student’s t test. (B) Plot of the ratio of neurons with aberrant neurite (an) formation on the cell soma. Over 50 neurons were counted in two separate experiments. The differences between Kal7⌬CT versus GFP control (asterisk, p ⬍ 0.034) were statistically significant by Student’s t test. (C) Histograms of the areas of spine-like filopodia. Over 200 spines were measured from 10–15 neurons in two different experiments. (D) Histograms of the linear spine density; over 50 neurons were counted in two separate experiments. Bars represent averages, and error bars are standard deviations.

motif yields a soluble molecule, mostly resident in the cell soma. The essential role of the C terminus of Kalirin-7 in synaptic targeting strongly suggests a role for PDZ domain proteins. The C-terminally truncated molecule induces formation of abundant filopodia on the cell body; in contrast, neurons expressing only GFP have smooth cell bodies. Overexpression of activated Rac1 also yielded aberrant filopodia on the cell soma (Nakayama et al., 2000). The phenotype of the GEF-inactive Kalirin-7 mutant suggests that GEF activity is essential for the function of Kalirin-7 in spine morphogenesis. Regulation of Kalirin-7 function may involve recruitment to specific sites as well as alterations in its GEF activity. Kalirin-7 activity at the PSD may be increased through recruitment by PDZ proteins. The local concentrations of F-actin observed in NIH3T3 cells at sites of coclustering of PSD-95 and Kalirin-7 may reflect regulation by recruitment. The absence of the C terminus of Kalirin-7 does not totally eliminate the effect of PSD-95 on Kalirin-7, suggesting that the interaction between the two proteins may be more extensive. Further work is needed to clarify the molecular details of regulation of Kalirin GEF activity by PSD-95. Whether PSD-95 plays a role in regulating spine morphology is not clear, and to our knowledge, no such connection has been described. The PSD-95 knockout mouse did not show alterations

in spine or PSD morphology. In contrast, defects in postsynaptic structure of the neuromuscular synapse in Dlg mutant Drosophila suggest a role for SAPs in the morphology of postsynaptic structures (Budnik et al., 1996). While the downstream effects of Kalirin clearly include activation of Rac1, the other domains of Kalirin-7 are likely to play important roles. In this case, PSD-95 may inhibit Rac1 activation by Kalirin-7, without interfering with interactions mediated by the spectrin-like region, the Sec14p-like region or the PH domain. The reduction of Rac1 activation by Kalirin-7 in the presence of PSD-95 in HEK293 cells may be a result of sequestration of Kalirin-7 by PSD-95, away from Rac1. Kalirin-7 as a Common Signaling Partner to Multiple Divergent PDZ Domain Proteins In our yeast two-hybrid screen, Kalirin-7 interacted with a large number of proteins with diverse functions and subcellular localizations. Further work is required to determine which of these interactions occur in vivo and their functional significance. Physiological relevance for these interactions is suggested by the fact that Kalirin-7 and its putative interactors have similar tissue distributions and subcellular localizations. In neurons, many of these putative binding partners are synaptically localized and are concentrated at the PSDs. Most of these

Kalirin Regulates Spine Morphology 239

Figure 9. Model of Kalirin-7 Function at the Postsynaptic Density For simplicity, other known PSD molecules are not represented. Kalirin-7, as well as NMDA receptor subunits, may interact with multiple PDZ domains of PSD-95.

proteins are also present in the neuronal soma, and a subset appear to be associated with certain membrane trafficking pathways. Although concentrated at the PSD, Kalirin-7 is found in diverse subcellular locations (Penzes et al., 2000) and could bind different interactors in different locations. With a substantial cytosolic pool of Kalirin-7, it may be recruited to various specific sites of action by interaction with different PDZ domain proteins. Proteins containing PDZ domains, in addition to their role in clustering receptors (Sheng and Pak, 1999), are crucial in generation of signaling complexes at cellular membranes (Fanning and Anderson, 1999). Recruitment of Kalirin-7 to particular sites may result in the formation of signaling microdomains, where, depending upon the state of activation of its DH-PH domain, Kalirin-7 could activate Rho GTPases, and hence control local actin filament rearrangements. Kalirin-7, through its interaction with proteins like G␤␥, iNOS, HAP-1, and PAM could recruit additional proteins to these complexes (Figure 9). In addition, a role for Kalirin-7 in trafficking of membrane proteins is suggested by its interaction with X11␣/ Mint-1, GIPC, syntenin, and AF-6/afadin. X11␣/Mint-1 is involved in the trafficking of ␤-amyloid (Borg et al., 1998). GIPC binds to clathrin-coated vesicles (De Vries et al., 1998), and syntenin is involved in pro-TGF trafficking (Fernandez-Larrea et al., 1999). AF-6/afadin contains kinesin-like and myosin-like domains, suggesting a role in vesicle transport (Ponting, 1995). This possibility is also supported by the interaction of Kalirin with trafficking-related proteins, such as peptidylglycine ␣–amidating monooxygenase (Alam et al., 1996, 1997), a neuropeptide processing enzyme, HAP-1 (Colomer et al., 1997), and inducible nitric oxide synthase (Ratovitski et al., 1999), a regulator of secretion. These interactions are mediated by the spectrin-like region of Kalirin, leaving its C terminus available for other interactions.

Experimental Procedures Reagents Antibodies: Kalirin-7 and Kalirin-spectrin antibodies were described previously (Penzes et al., 2000); affinity-purified Kalirin-7 antibody was used at a dilution of 1:100. PSD-95 monoclonal antibodies were from Upstate Biotechnology and polyclonal antibody was a generous gift from Dr. Min Li (Johns Hopkins University). Plasmids: pEAK.Myc.His.Kalirin-7 was described previously (Penzes et al., 2000); psCEP-Kalirin-7-GEFKO was generated by site-directed mutagenesis with the Quick Change kit (Stratagene) of the psCEPKalirin-7 template, generated by replacing the NsiI–NotI fragment of psCEP Kalirin-8 (Alam et al., 1997) with the NsiI–NotI fragment of pEAK.His.Myc.Kalirin-7; psCEP-Kalirin-7-⌬CT was generated by replacement of the NsiI–NotI fragment of psCEP-Kalirin-7 with a fragment truncated after nt 4728 of Kalirin-7, thus encoding a protein lacking the C-terminal 60 amino acids; pCW-PSD-95 was a generous gift of Dr. Morgan Sheng (Harvard University); pEGFP-N2 was from Clontech. Oligonucleotides were synthesized at the Johns Hopkins University Molecular Biology Core. FITC-phalloidin was from Sigma.

Yeast Two-Hybrid Screen The yeast two-hybrid screen was performed essentially as described (Alam et al., 1997) (Figure 1B). Yeast (CG 1945) were transformed first with pPC97-K7-CT, encoding the GAL4 DNA binding domain K7-CT fusion protein (GAL4-K7-CT; Kalirin-7a amino acids 1607–1636) and then with a cDNA library prepared from the hippocampus/cortex of 3-week-postnatal rat pups subjected to a single maximal electroconvulsive shock-induced seizure (gift of Dr. A. Lanahan and Dr. P. Worley) (Li et al., 1995). Double transformants were grown on synthetic dextrose medium deficient in Trp, Leu, and His containing 10 mM 3-amino triazole. A screen of 6 ⫻ 106 transformants yielded 360 clones that survived on SC-Leu, -Trp, -His plates. Of these, 150 colonies showed strong ␤-galactosidase activity in a filter assay. We picked 20 colonies and sequenced the 5⬘ end of each cDNA. Seven different inserts were identified at this step. These clones were SAP102, syntenin, ZO-2, Chapsyn-110, spinophilin, MAGI-1, and PSD-95 (Table 1). The inserts were amplified by PCR, and a probe consisting of a mixture of the seven DNAs was generated. To eliminate redundant clones, we used this probe for hybridization with Southern blots of the DNAs of the remaining 130 yeast clones. We isolated non-cross-reactive clones, amplified

Neuron 240

the inserts by PCR, and sequenced the 5⬘ end of each PCR product. This approach allowed rapid screening of the large number of positives but does not provide an accurate estimate of the frequency of each individual positive. Tissue Preparation and Subcellular Fractionation Parietal cortex dissected from adult female Holtzman rats (Harlan) was homogenized in 10 volumes of buffer A: 4 mM HEPES, pH 7.5, 1 mM MgCl2, 0.5 mM CaCl2, and 0.32 M sucrose, containing 0.3 mg/ml PMSF, and a protease inhibitor cocktail [LBTI (50 ␮g/ml), leupeptin (2 ␮g/ml), benzamidine (16 ␮g/ml), and pepstatin (2 ␮g/ ml)]. Postsynaptic densities were purified using the protocol described by Carlin et al. (1980), with the exception that homogenization buffer A contained 4 mM HEPES, pH 7.5. Immunoprecipitations The crude synaptosomal fraction (P2) (Carlin et al., 1980) (0.5 mg protein) was incubated with 2% SDS, and 1 mg/ml K7-CT peptide for 10 min at 4⬚C. Extract was cleared by centrifugation at 200,000 ⫻ g for 10 min and supernatants were incubated with 10 ␮l Kalirinspectrin antibody, preimmune serum, or Kalirin-spectrin antibody preincubated with antigenic fusion protein (GST fusion protein consisting of Kalirin-spectrin repeats 4–7, 0.36 mg/ml). After 3 hr tumbling at 4⬚C, 60 ␮l protein A Sepharose (30% slurry) was added and incubation was continued for 1.5 hr at 4⬚C. The mixture was centrifuged at 4000 ⫻ g for 1 min, and the resin pellets were washed 4 times with 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 1% Triton X-100. Bound proteins were eluted by boiling the resin in 60 ␮l 1⫻ SDS–PAGE loading buffer and analyzed by SDS–PAGE and Western blotting. GST-Affinity Assays Rat cerebral cortex postsynaptic density fraction Triton/Sarcosyl extract supernatant (25 ␮g protein) was diluted 20-fold with binding buffer (1% Triton X-100 in 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, and protease inhibitors). Each GST fusion protein (5 ␮g) was bound to glutathione agarose resin (27 ␮l resin bed) for 30 min at room temperature and resin was washed twice with the same buffer. The fusion proteins used were the C-terminal 30 residues of Kalirin-7 and a mutant version with the final residue changed from Val to Asp (K7-V/D). For competition, K7-CT peptide (0.35 mg/ml) was included during the incubation with GST-K7-CT. After tumbling for 1 hr at 4⬚C, mixtures were centrifuged at 5000 ⫻ g for 1 min, and the resin pellets were washed 4 times with binding buffer. Bound proteins were eluted by boiling in SDS–PAGE loading buffer, resolved by SDS–PAGE, and analyzed by Western blotting with PSD-95-specific monoclonal antibodies.

were transfected with pEGFP-N2 alone or together with pEAK. Myc.his.Kalirin-7, psCEP.Myc.Kalirin-7-GEFKO, or psCEP.Myc. Kalirin-7-⌬CT (14 ␮g each DNA/30 mm well), with the Biolistic PDS1000/He gene delivery system or by the calcium phosphate method (Threadgill et al., 1997). After 42 hr, cells were fixed in 3.7% paraformaldehyde for 20 min, permeabilized, and blocked with 2% normal goat serum with 0.1% Triton X-100. Primary antibodies (Myc monoclonal and GFP ployclonal) were added in 2% normal goat serum for 2 hr and secondary antibodies for 1 hr. Rac1 Activation Assays PEAK-Rapid HEK293 cells (Edge Biosystems) were transiently transfected with pEAK-10.His.Myc.Kalirin-7 (8 ␮g), pEAK-10.His.Myc. Kalirin-7-⌬CT (8 ␮g), pCW-PSD-95 (4 ␮g), alone or in combinations, using Lipofectamine (GIBCO-BRL). All transfections contained 12 ␮g of total DNA, pGEX-5X2 plasmid being used as control DNA or carrier. Cells were serum starved overnight and activation of Rac1 was measured with the Rac1 activation assay kit (Upstate Biotechnology). Western blots were analyzed with the Scion Image software. For each experiment, the optical density value for Rac1 activation in the control sample was considered the reference, and the value for Rac1 activation in each sample was divided by the value for the control, yielding fold activation. For the samples expressing Myctagged proteins, the value for Rac1 activation was multiplied by the ratio between the expression level of the Myc-tagged protein and the expression level of Myc-Kalirin-7. Since only a small fraction of the total Rac1 is activated by transfection, compared to the total activatable Rac1 (Figure 5A, lane 1), it is reasonable to assume that Rac1 activation is proportional to the amount of GEF protein expressed. Acknowledgments Dr. Anthony Lanahan and Dr. Paul Worley (Department of Neuroscience, The Johns Hopkins University) generously provided the rat hippocampus/cortex cDNA library. We thank Dr. Dezhi Liau and Bing Ye for providing cultures of cortical neurons. We thank Jee Hae Kim (Department of Neuroscience, The Johns Hopkins University) for useful advice. We thank Kirsten Barnes for her help with the yeast two-hybrid screen. We thank Dr. Henry Keutmann (Endocrine Unit, Massachusetts General Hospital) for the synthesis of the antigenic peptide. We thank Marie Bell, Kate Deanehan, and Lixian Jin for general laboratory assistance. This work was supported by NIH grants DA-00266 and DK-32948. Received March 22, 2000; revised November 3, 2000. References

Transient Transfections NIH 3T3 and HEK293 cells were cultured in Dulbecco’s modified Eagle medium (DMEM)/F-12 containing 10% fetal bovine serum (Hyclone) and 10% NuSerum (Collaborative Research). Briefly, cells grown on glass slides for 3 days to 40%–60% confluence were transfected with 1 ␮g plasmid DNA/4 cm2 [pEAK10.His.Myc.Kalirin-7, pCW-PSD-95, or both] and 4 ␮l lipofectamine (GIBCO-BRL) in 1 ml complete serum-free medium for 5 hr, after which they were fed with growth medium. After 1 day, the cells were serum starved by replacing the medium with DMEM/F-12 containing no serum for 16 hr. Cells were fixed in 4% formaldehyde in PBS (50 mM NaPi, pH 7.5, 150 mM NaCl) for 30 min at room temperature, permeabilized, and stained as described (Penzes et al., 2000). Cell morphology was evaluated by analyzing ⵑ30 transfected cells and counting the fraction of cells displaying a particular phenotype. Primary Cultures of Rat Hippocampal Neurons Low density hippocampal neuronal cultures were prepared from E18 Sprague–Dawley rat pups (Goslin and Banker, 1991) and were fixed and stained as described (Xia et al., 1999). Transfection of Cortical Neurons Cortical neurons were cultured at high density on polylysine-coated coverslips in MEM containing 5% heat inactivated horse serum, 2% glutamine, penicillin, and streptomycin. Neurons at 1 week in culture

Alam, M.R., Caldwell, B.D., Johnson, R.C., Darlington, D.N., Mains, R.E., and Eipper, B.A. (1996). Novel proteins that interact with the COOH-terminal cytosolic routing determinants of and integral membrane peptide-processing. Enzyme J. Biol. Chem. 271, 28636– 28640. Alam, M.R., Johnson, R.C., Darlington, D.N., Hand, T.A., Mains, R.E., and Eipper, B.A. (1997). Kalirin, a cytosolic protein with spectrinlike and GDP/GTP exchange factor-like domains that interacts with peptidylglycine ␣-amidating monooxygenase, an integral membrane peptide-processing. Enzyme J. Biol. Chem. 272, 12667– 12675. Allen, P.B., Ouimet, C.C., and Greengard, P. (1997). Spinophilin, a novel protein phosphatase 1 binding protein localized to dendritic spines. Proc. Natl. Acad. Sci. USA 94, 9956–9961. Allison, D.W., Gelfand, V.I., Spector, I., and Craig, A.M. (1998). Role of actin in anchoring postsynaptic receptors in cultured hippocampal neurons: differential attachment of NMDA versus AMPA receptors. J. Neurosci. 18, 2423–2436. Awasaki, T., Saito, M., Sone, M., Suzuki, E., Sakai, R., Ito, K., and Hama, C. (2000). The Drosophila trio plays an essential role in patterning of axons by regulating their directional extension. Neuron 26, 119–131. Bateman, J., Shu, H., and Van Vactor, D. (2000). The guanine nucleo-

Kalirin Regulates Spine Morphology 241

tide exchange factor trio mediates axonal development in the Drosophila embryo. Neuron 26, 93–106. Benard, V., Bohl, B.P., and Bokoch, G.M. (1999). Characterization of rac and cdc42 activation in chemoattractant-stimulated human neutrophils using a novel assay for active GTPases. J. Biol. Chem. 274, 13198–13204. Borg, J.P., Yang, Y., De Taddeo-Borg, M., Margolis, B., and Turner, R.S. (1998). The X11alpha protein slows cellular amyloid precursor protein processing and reduces Abeta40 and Abeta42 secretion. J. Biol. Chem. 273, 14761–14766. Budnik, V., Koh, Y.-H., Guan, B., Hartmann, B., Hough, C., Woods, D., and Gorczyca, M. (1996). Regulation of synapse structure and function by the Drosophila tumor suppressor gene dlg. Neuron 17, 627–640. Carlin, R.K., Grab, D.J., Cohen, R.S., and Siekevitz, P. (1980). Isolation and characterization of postsynaptic densities from various brain regions: enrichment of different types of postsynaptic densities. J. Cell Biol. 86, 831–845. Cho, K.O., Hunt, C.A., and Kennedy, M.B. (1992). The rat brain postsynaptic density fraction contains a homolog of the Drosophila discs-large tumor suppressor protein. Neuron 9, 929–942. Colomer, V., Engelender, S., Sharp, A.H., Duan, K., Cooper, J.K., Lanahan, A., Lyford, G., Worley, P., and Ross, C.A. (1997). Huntingtin-associated protein 1 (HAP1) binds to a Trio-like polypeptide, with a rac1 guanine nucleotide exchange factor domain. Hum. Mol. Genet. 6, 1519–1525. Debant, A., Serra-Pages, C., Seipel, K., O’Brien, S., Tang, M., Park, S.H., and Streuli, M. (1996). The multidomain protein Trio binds the LAR transmembrane tyrosine phosphatase, contains a protein kinase domain, and has separate rac-specific and rho-specific guanine nucleotide exchange factor domains. Proc. Natl. Acad. Sci. USA 93, 5466–5471. De Vries, L., Lou, X., Zhao, G., Zheng, B., and Farquhar, M.G. (1998). GIPC, a PDZ domain containing protein, interacts specifically with the C terminus of RGS-GAIP. Proc. Natl. Acad. Sci. USA 95, 12340– 12345.

Harris, K.M., and Kater, S.B. (1994). Dendritic spines: cellular specializations imparting both stability and flexibility to synaptic function. Annu. Rev. Neurosci. 17, 341–371. Hirao, K., Hata, Y., Ide, N., Takeuchi, M., Irie, M., Yao, I., Deguchi, M., Toyoda, A., Sudhof, T.C., and Takai, Y. (1999). A novel multiple PDZ domain-containing molecule interacting with N-methyl-D-aspartate receptors and neuronal cell adhesion proteins. J. Biol. Chem. 273, 21105–21110. Itoh, M., Nagafuchi, A., Yonemura, S., Kitani-Yasuda, T., Tsukita, S., and Tsukita, S. (1993). The 220-kD protein colocalizing with cadherins in non-epithelial cells is identical to ZO-1, a tight junction-associated protein in epithelial cells: cDNA cloning and immunoelectron microscopy. J. Cell. Biol. 121, 491–502. Jesaitis, L.A., and Goodenough, D.A. (1994). Molecular characterization and tissue distribution of ZO-2, a tight junction protein homologous to ZO-1 and the Drosophila discs-large tumor suppressor protein. J. Cell. Biol. 124, 949–961. Johnson, R.C., Penzes, P., Eipper, B.A., and Mains, R.E. (2000). Isoforms of kalirin, a neuronal Dbl family member, generated through use of different 5⬘- and 3⬘-ends along with an internal translational initiation site. J. Biol. Chem. 275, 19324–19333. Kaibuchi, K., Kuroda, S., and Amano, M. (1999). Regulation of the cytoskeleton and cell adhesion by the rho family GTPases in mammalian cells. Annu. Rev. Biochem. 68, 459–486. Kim, C.H., and Lisman, J.E. (1999). A role of actin filament in synaptic transmission and long-term potentiation. J. Neurosci. 19, 4314– 4324. Kim, E., Cho, K.O., Rothschild, A., and Sheng, M. (1996). Heteromultimerization and NMDA receptor-clustering activity of Chapsyn-110, a member of the PSD-95 family of proteins. Neuron 17, 103–113. Kim, J.H., Liao, D., Lau, L.F., and Huganir, R.L. (1998). SynGAP: a synaptic RasGAP that associates with the PSD-95/SAP90 protein family. Neuron 20, 683–691. Kornau, H.C., Schenker, L.T., Kennedy, M.B., and Seeburg, P.H. (1995). Domain interaction between NMDA receptor subunits and the postsynaptic density protein PSD-95. Science 269, 1737–1740.

Dobrosotskaya, I., Guy, R.K., and James, G.L. (1997). MAGI-1, a membrane-associated guanylate kinase with a unique arrangement of protein-protein interaction domains. J. Biol. Chem. 272, 31589– 31597.

Li, X.J., Li, S.H., Sharp, A.H., Nucifora, F.C., Schilling, G., Lanahan, A., Worley, P., Snyder, S.H., and Ross, C.A. (1995). A huntingtinassociated protein enriched in brain with implications for pathology. Nature 378, 398–402.

Engert, F., and Bonhoeffer, T. (1999). Dendritic spine changes associated with hippocampal long-term synaptic plasticity. Nature 399, 66–70.

Li, Z., Van Aelst, L., and Cline, H.T. (2000). Rho GTPases regulate distinct aspects of dendritic arbor growth in Xenopus central neurons in vivo. Nat. Neurosci. 3, 217–225.

Fanning, A.S., and Anderson, J.M. (1999). PDZ domains: fundamental building blocks in the organization of protein complexes at the plasma membrane. J. Clin. Invest. 103, 767–772.

Liebl, E.C., Forsthoefel, D.J., Franco, L.S., Sample, S.H., Hess, J.E., Cowger, J.A., Chandler, M.P., Shupert, A.M., and Seeger, M.A. (2000). Dosage-sensitive, reciprocal genetic interactions between the Abl tyrosine kinase and the putative GEF trio reveal trio’s role in axon pathfinding. Neuron 26, 107–118.

Fernandez-Larrea, J., Merlos-Suarez, A., Urena, J.M., Baselga, J., and Arribas, J. (1999). A role for a PDZ protein in the early secretory pathway for the targeting of proTGF-alpha to the cell surface. Mol. Cell 3, 423–433. Fiala, J.C., Feinberg, M., Popov, V., and Harris, K.M. (1998). Synaptogenesis via dendritic filopodia in developing hippocampal area CA1. J. Neurosci. 18, 8900–8911. Fischer, M., Kaech, S., Knutti, D., and Matus, A. (1998). Rapid actinbased plasticity in dendritic spines. Neuron 20, 847–854. Goslin, K., and Banker, G. (1991). Culturing nerve cells, K.G.A.G. Banker, ed. (Cambridge, MA: Massachusetts Institute of Technology Press). Grootjans, J.J., Zimmermann, P., Reekmans, G., Smets, A., Degeest, G., Durr, J., and David, G. (1997). Syntenin, a PDZ protein that binds syndecan cytoplasmic domains. Proc. Natl. Acad. Sci. USA 94, 13683–13688.

Liu, X., Wang, H., Eberstadt, M., Schnuchel, A., Olejniczak, E.T., Meadows, R.P., Schkeryantz, J.M., Janowick, D.A., Harlan, J.E., Harris, E.A., et al. (1998). NMR structure and mutagenesis of the N-terminal Dbl homology domain of the nucleotide exchange factor Trio. Cell. 95, 269–277. Luo, L., Hensch, T.K., Ackerman, L., Barbel, S., Jan, L.Y., and Jan, Y.N. (1996). Differential effects of the Rac GTPase on Purkinje cell axons and dendritic trunks and spines. Nature 379, 837–840. Luscher, C., Nicoll, R.A., Malenka, R.C., and Muller, D. (2000). Synaptic plasticity and dynamic modulation of the postsynaptic membrane. Nat. Neurosci. 3, 545–550.

Halpain, S. (2000). Actin and the agile spine: how and why do dendritic spines dance? Trends Neurosci. 23, 141–146.

Mains, R.E., Alam, M.R., Johnson, R.C., Darlington, D.N., Back, N., Hand, T.A., and Eipper, B.A. (1999). Kalirin, a multifunctional PAM COOH-terminal domain interactor protein, affects cytoskeletal organization and ACTH secretion from AtT-20 cells. J. Biol. Chem. 274, 2929–2937.

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.

Maletic-Savatic, M., Malinow, R., and Svoboda, K. (1999). Rapid dendritic morphogenesis in CA1 hippocampal dendrites induced by synaptic activity. Science 283, 1923–1927.

Harris, K.M. (1999). Structure, development, and plasticity of dendritic spines. Curr. Opin. Neurobiol. 9, 343–348.

Mandai, K., Nakanishi, H., Satoh, A., Obaishi, H., Wada, M., Nishioka, H., Itoh, M., Mizoguchi, A., Aoki, T., Fujimoto, T., et al. (1997). Afadin:

Neuron 242

a novel actin filament-binding protein with one PDZ domain localized at cadherin-based cell-to-cell adherens junction. J. Cell. Biol. 139, 517–528.

van Rossum, D., and Hanisch, U.K. (1999). Cytoskeletal dynamics in dendritic spines: direct modulation by glutamate receptors? Trends Neurosci. 22, 290–295.

Matus, A. (1999). Postsynaptic actin and neuronal plasticity. Curr. Opin. Neurobiol. 9, 561–565.

Xia, J., Zhang, X., Staudinger, J., and Huganir, R.L. (1999). Clustering of AMPA receptors by the synaptic PDZ domain-containing protein PICK1. Neuron 22, 179–187.

Muller, B.M., Kistner, U., Veh, R.W., Cases-Langhoff, C., Becker, B., Gundelfinger, E.D., and Garner, C.C. (1995). Molecular characterization and spatial distribution of SAP97, a novel presynaptic protein homologous to SAP90 and the Drosophila discs-large tumor suppressor protein. J. Neurosci. 15, 2354–2366. Muller, B.M., Kistner, U., Kindler, S., Chung, W.J., Kuhlendahl, S., Fenster, S.D., Lau, L.F., Veh, R.W., Huganir, R.L., Gundelfinger, E.D., and Garner, C.C. (1996). SAP102, a novel postsynaptic protein that interacts with NMDA receptor complexes in vivo. Neuron 17, 255–265. Nakanishi, H., Obaishi, H., Satoh, A., Wada, M., Mandai, K., Satoh, K., Nishioka, H., Matsuura, Y., Mizoguchi, A., and Takai, Y. (1997). Neurabin: a novel neural tissue-specific actin filament-binding protein involved in neurite formation. J. Cell. Biol. 139, 951–961. Nakayama, A.Y., Harms, M.B., and Luo, L. (2000). Small GTPases Rac and Rho in the maintenance of dendritic spines and branches in hippocampal pyramidal neurons. J. Neurosci. 20, 5329–5338. Newsome, T.P., Schmidt, S., Dietzl, G., Keleman, K., A˚sling, B., Debant, A., and Dickson, B.J. (2000). Trio combines with dock to regulate pak activity during photoreceptor axon pathfinding in Drosophila. Cell 101, 283–294. Okamoto, M., and Sudhof, T.C. (1997). Mints, Munc18-interacting proteins in synaptic vesicle exocytosis. J. Biol. Chem. 272, 31459– 31464. Papa, M., Bundman, M.C., Greenberger, V., and Segal, M. (1995). Morphological analysis of dendritic spine development in primary cultures of hippocampal neurons. J. Neurosci. 15, 1–11. Penzes, P., Johnson, R.C., Alam, M.R., Kambampati, V., Mains, R.E., and Eipper, B.A. (2000). An isoform of kalirin, a brain-specific GDP/ GTP exchange factor, is enriched in the postsynaptic density fraction. J. Biol. Chem. 275, 6395–6403. Ponting, C.P. (1995). AF-6/cno: neither a kinesin nor a myosin, but a bit of both. Trends Biochem. Sci. 20, 265–266. Ratovitski, E.A., Alam, M.R., Quick, R.A., McMillan, A., Bao, C., Kozlovsky, C., Hand, T.A., Johnson, R.C., Mains, R.E., Eipper, B.A., and Lowenstein, C.J. (1998). Kalirin inhibition of inducible nitric oxide synthase. J. Biol. Chem. 274, 993–999. Schultz, J., Milpetz, F., Bork, P., and Ponting, C.P. (1998). SMART, a simple modular architecture research tool: identification of signaling domains. Proc. Natl. Acad. Sci. USA 95, 5857–5864. Sheng, M., and Pak, D.T. (1999). Glutamate receptor anchoring proteins and the molecular organization of excitatory synapses. Ann. NY Acad. Sci. 868, 483–493. Songyang, Z., Fanning, A.S., Fu, C., Xu, J., Marfatia, S.M., Chishti, A.H., Crompton, A., Chan, A.C., Anderson, J.M., and Cantley, L.C. (1997). Recognition of unique carboxyl-terminal motifs by distinct PDZ domains. Science 275, 73–77. Staudinger, J., Zhou, J., Burgess, R., Elledge, S.J., and Olson, E.N. (1995). PICK1: a perinuclear binding protein and substrate for protein kinase C isolated by the yeast two-hybrid system. J. Cell Biol. 128, 263–271. Steven, R., Kubiseski, T.J., Zheng, H., Kulkarni, S., Mancillas, J., Ruiz Morales, A., Hogue, C.W., Pawson, T., and Culotti, J. (1998). UNC-73 activates the Rac GTPase and is required for cell and growth cone migrations in C. elegans. Cell 92, 785–795. Threadgill, R., Bobb, K., and Ghosh, A. (1997). Regulation of dendritic growth and remodeling by Rho, Rac, and Cdc42. Neuron 19, 625–634. Toni, N., Buchs, P.A., Nikonenko, I., Bron, C.R., and Muller, D. (1999). LTP promotes formation of multiple spine synapses between a single axon terminal and a dendrite. Nature 402, 421–425. Ullmer, C., Schmuck, K., Figge, A., and Lubbert, H. (1998). Cloning and characterization of MUPP1, a novel PDZ domain protein. FEBS Lett. 424, 63–68.

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