Variable workup calls for guideline development for type 2A hereditary haemochromatosis.

BACKGROUND: Type 2A hereditary haemochromatosis (type 2A HH) is a rare iron-loading disorder caused by mutations in the HFE2 gene, which encodes the HJV protein. We present characteristics, treatment and follow-up of subjects diagnosed with type 2A HH in the Netherlands to increase awareness of the disease and its treatment, and to define knowledge gaps. METHODS: We collected clinical, biochemical and genetic data from seven patients (two female; five probands) from six families genetically diagnosed with type 2A HH at the Expertise Center for Iron Disorders, Radboud University Medical Centre between 2006 and 2016. RESULTS: The five probands presented with heterogeneous complaints between the ages of 19 and 39. One of two patients with delayed clinical diagnosis developed hypogonadism and Y. enterocolitica sepsis. Diagnostic workup and follow-up varied. When assessed, elevated transferrin saturation (79-98%), ferritin (1400-6200 µg/l) and severely elevated liver iron levels were found, and in all subjects, phlebotomies were initiated. One subject was switched to erythrocytapheresis. Target ferritin levels varied. Despite long-term iron depletion, two subjects developed clinical complications. Sanger sequencing revealed two pathogenic HFE2 variants (homozygous or compound heterozygous) for the five families of Dutch descent and one new pathogenic variant in the family of non-Dutch descent. CONCLUSION: Three genetic variants caused type 2A HH in six families. Clinical diagnosis was delayed in two subjects. We observed variance in presentation, workup, follow-up and treatment. We found new complications in long-term iron-depleted patients. We recommend research and guidelines for optimal workup, follow-up and treatment of type 2A HH.

Optimal resource allocation in colonoscopy: timing of follow-up colonoscopies in relation to adenoma detection rates.

BACKGROUND AND STUDY AIMS: The assessment of indications for follow-up colonoscopy may help to improve the allocation of available endoscopy resources. The aim of this study was to assess the timing of early follow-up colonoscopy and surveillance utilization in relation to adenoma detection rate (ADR) at follow-up. METHODS: An assessment of the timing and yield of follow-up colonoscopies was performed in patients with non-inflammatory bowel disease (IBD) in a Dutch multicenter study. The primary outcome was the number of patients with a prior (index) colonoscopy. The necessity for follow-up procedures was assessed using the ADR. RESULTS: Of 4800 consecutive patients undergoing a colonoscopy, 1249 non-IBD patients had undergone an index colonoscopy. Of these, follow-up procedures were performed within 1 year in 27 % (331/1249). Excluding incomplete colonoscopy, incomplete polypectomy, or poor bowel preparation on index, the ADR on early follow-up was 4 % for symptomatic and 26 % for asymptomatic patients. Among the asymptomatic patients with a follow-up colonoscopy at > 1 year (n = 463), an ADR of 23 % (108/463) was found. In 27 % of these patients, the observed surveillance intervals were in accordance with American Gastroenterological Association (AGA) surveillance recommendations; 60 % were classified as over-utilization and 13 % as under-utilization according to the AGA. Optimal utilization follow-up colonoscopies had higher ADRs on follow-up compared with over-utilized procedures (31 % vs. 17 %; P < 0.001). CONCLUSIONS: Follow-up colonoscopy in symptomatic patients within a year has limited value in terms of adenoma detection. A considerable proportion of surveillance colonoscopies are performed too early according to current guidelines, resulting in low detection rates. Both aspects can be targeted for optimal usage in endoscopic capacity.

In vitro genetic screen identifies a cooperative role for LPA signaling and c-Myc in cell transformation.

c-Myc drives uncontrolled cell proliferation in various human cancers. However, in mouse embryo fibroblasts (MEFs), c-Myc also induces apoptosis by activating the p19Arf tumor suppressor pathway. Tbx2, a transcriptional repressor of p19Arf, can collaborate with c-Myc by suppressing apoptosis. MEFs overexpressing c-Myc and Tbx2 are immortal but not transformed. We have performed an unbiased genetic screen, which identified 12 oncogenes that collaborate with c-Myc and Tbx2 to transform MEFs in vitro. One of them encodes the LPA2 receptor for the lipid growth factor lysophosphatidic acid (LPA). We find that LPA1 and LPA4, but not LPA3, can reproduce the transforming effect of LPA2. Using pharmacological inhibitors, we show that the in vitro cell transformation induced by LPA receptors is dependent on the Gi-linked ERK and PI3K signaling pathways. The transforming ability of LPA1, LPA2 and LPA4 was confirmed by tumor formation assays in vivo and correlated with prolonged ERK1/2 activation in response to LPA. Our results reveal a direct role for LPA receptor signaling in cell transformation and tumorigenesis in conjunction with c-Myc and reduced p19Arf expression.

Autotaxin (NPP-2) in the brain: cell type-specific expression and regulation during development and after neurotrauma.

Autotaxin is a secreted cell motility-stimulating exo-phosphodiesterase with lysophospholipase D activity that generates bioactive lysophosphatidic acid. Lysophosphatidic acid has been implicated in various neural cell functions such as neurite remodeling, demyelination, survival and inhibition of axon growth. Here, we report on the in vivo expression of autotaxin in the brain during development and following neurotrauma. We found that autotaxin is expressed in the proliferating subventricular and choroid plexus epithelium during embryonic development. After birth, autotaxin is mainly found in white matter areas in the central nervous system. In the adult brain, autotaxin is solely expressed in leptomeningeal cells and oligodendrocyte precursor cells. Following neurotrauma, autotaxin is strongly up-regulated in reactive astrocytes adjacent to the lesion. The present study revealed the cellular distribution of autotaxin in the developing and lesioned brain and implies a function of autotaxin in oligodendrocyte precursor cells and brain injuries.

Lysophosphatidic acid: mitogen and motility factor.

LPA (lysophosphatidic acid), the simplest of al glycerophospholipids, is a potent inducer of cell proliferation, migration and survival. It does so by activating its cognate G-protein-coupled receptors, four of which have been identified. LPA receptors couple to at least three distinct G-proteins and thereby activate multiple signal transduction pathways, particularly those initiated by the small GTPases Ras, Rho and Rac. Our recent work has shown that LPA signals Rac activation via the Tiam1 GDP/GTP exchange factor and thereby stimulates cell migration. Here we discuss recent progress in our understanding of LPA action.

Regulating c-Ras function. cholesterol depletion affects caveolin association, GTP loading, and signaling.

Cholesterol-rich and caveolin-containing microdomains of the plasma membrane, termed "caveolae," have been implicated in signal transduction. However, the role of caveolae in regulating the Ras-MAP kinase cascade is incompletely understood. The mammalian Ras isoforms (H, N, and K) use different membrane anchors to attach to the plasma membrane and thereby may localize to functionally distinct microdomains, which might explain isoform-specific signaling. Here, we show that, in Cos epithelial cells, endogenous K-Ras colocalizes largely with caveolin, whereas N-Ras localizes to both caveolar and noncaveolar subdomains; H-Ras localization was below detection limits. We find that epidermal growth factor (EGF) activates N-Ras but fails to activate K-Ras in these cells. Extraction of cholesterol with methyl-beta-cyclodextrin disrupts complex formation between caveolin and K- and N-Ras and, strikingly, enables EGF to activate both K-Ras and N-Ras. While cholesterol depletion enhances GTP-loading on total c-Ras, activation of the downstream MEK-MAP kinase cascade by EGF and lysophosphatidic acid but not that by phorbol ester is inhibited. Thus, plasma membrane cholesterol is essential for negative regulation of c-Ras isoforms (complexed to caveolin), as well as for mitogenic signaling downstream of receptor-activated c-Ras.

Gap junction protein connexin-43 interacts directly with microtubules.

Gap junctions are specialized cell-cell junctions that mediate intercellular communication. They are composed of connexin proteins, which form transmembrane channels for small molecules [1, 2]. The C-terminal tail of connexin-43 (Cx43), the most widely expressed connexin member, has been implicated in the regulation of Cx43 channel gating by growth factors [3-5]. The Cx43 tail contains various protein interaction sites, but little is known about binding partners. To identify Cx43-interacting proteins, we performed pull-down experiments using the C-terminal tail of Cx43 fused to glutathione-S-transferase. We find that the Cx43 tail binds directly to tubulin and, like full-length Cx43, sediments with microtubules. Tubulin binding to Cx43 is specific in that it is not observed with three other connexins. We established that a 35-amino acid juxtamembrane region in the Cx43 tail, which contains a presumptive tubulin binding motif, is necessary and sufficient for microtubule binding. Immunofluorescence and immunoelectron microscopy studies reveal that microtubules extend to Cx43-based gap junctions in contacted cells. However, intact microtubules are dispensable for the regulation of Cx43 gap-junctional communication. Our findings suggest that, in addition to its well-established role as a channel-forming protein, Cx43 can anchor microtubule distal ends to gap junctions and thereby might influence the properties of microtubules in contacted cells.

Ras-MAP kinase signaling by lysophosphatidic acid and other G protein-coupled receptor agonists.

Lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) are extracellular lipid mediators that signal through distinct members of the Edg/LP subfamily of G protein-coupled receptors (GPCRs). LPA and S1P receptors are expressed in almost every cell type and can couple to multiple G proteins (G(i), G(q) and G(12/13)) to mediate a great variety of responses, ranging from rapid morphological changes to long-term stimulation of cell proliferation. LPA serves as the prototypic GPCR agonist that activates the small GTPases Ras (via G(i)) and RhoA (via G(12/13)), leading to activation of the mitogen-activated protein kinase (MAPK) cascade and reorganization of the actin cytoskeleton, respectively. This review focuses on our current insights into how Ras-MAPK signaling is regulated by GPCR agonists in general, and by LPA in particular.

Src and Pyk2 mediate G-protein-coupled receptor activation of epidermal growth factor receptor (EGFR) but are not required for coupling to the mitogen-activated protein (MAP) kinase signaling cascade.

The epidermal growth factor receptor (EGFR) and the non-receptor protein tyrosine kinases Src and Pyk2 have been implicated in linking a variety of G-protein-coupled receptors (GPCR) to the mitogen-activated protein (MAP) kinase signaling cascade. In this report we apply a genetic strategy using cells isolated from Src-, Pyk2-, or EGFR-deficient mice to explore the roles played by these protein tyrosine kinases in GPCR-induced activation of EGFR, Pyk2, and MAP kinase. We show that Src kinases are critical for activation of Pyk2 in response to GPCR-stimulation and that Pyk2 and Src are essential for GPCR-induced tyrosine phosphorylation of EGFR. By contrast, Pyk2, Src, and EGFR are dispensable for GPCR-induced activation of MAP kinase. Moreover, GPCR-induced MAP kinase activation is normal in fibroblasts deficient in both Src and Pyk2 (Src-/-Pyk2-/- cells) as well as in fibroblasts deficient in all three Src kinases expressed in these cells (Src-/-Yes-/-Fyn-/- cells). Finally, experiments are presented demonstrating that, upon stimulation of GPCR, activated Pyk2 forms a complex with Src, which in turn phosphorylates EGFR directly. These experiments reveal a role for Src kinases in Pyk2 activation and a role for Pyk2 and Src in tyrosine phosphorylation of EGFR following GPCR stimulation. In addition, EGFR, Src family kinases, and Pyk2 are not required for linking GPCRs with the MAP kinase signaling cascade.

Galpha(13) mediates activation of a depolarizing chloride current that accompanies RhoA activation in both neuronal and nonneuronal cells.

Loss of membrane potential (membrane depolarization) is one of the earliest and most striking responses of quiescent cells to stimulation with serum or G protein-coupled receptor (GPCR) agonists such as lysophosphatidic acid and thrombin. Membrane depolarization is due to the activation of a chloride conductance. While this response has received relatively little attention in the past, it is clear that the acute loss of membrane potential may have important physiological consequences. However, the dissection of the underlying G protein pathway and the establishment of cause-effect relationships have remained elusive to date. Here we report that, in neuronal cells, the depolarizing chloride current invariably accompanies GPCR-induced activation of RhoA and subsequent neurite retraction, and neither of these events requires phosphoinositide hydrolysis or Ca2+ mobilization. Through antibody microinjections and a genetic approach, we demonstrate that activation of the chloride conductance is mediated by Galpha(13) in a RhoA-independent manner in both neuronal cells and fibroblasts. We further show that, in neuronal cells, this newly described Galpha(13) pathway may profoundly modulate membrane excitability during RhoA-regulated neurite remodeling.

Characterization of p190RhoGEF, a RhoA-specific guanine nucleotide exchange factor that interacts with microtubules.

Rho family GTPases control numerous cellular processes including cytoskeletal reorganization and transcriptional activation. Rho GTPases are activated by guanine nucleotide exchange factors (GEFs) which stimulate the exchange of bound GDP for GTP. We recently isolated a putative GEF, termed p190RhoGEF that binds to RhoA and, when overexpressed in neuronal cells, induces cell rounding and inhibits neurite outgrowth. Here we show that the isolated tandem Dbl homology/pleckstrin homology domain of p190RhoGEF activates RhoA in vitro, but not Rac1 or Cdc42, as determined by GDP release and protein binding assays. In contrast, full-length p190RhoGEF fails to activate RhoA in vitro. When overexpressed in intact cells, however, p190RhoGEF does activate RhoA with subsequent F-actin reorganization and serum response factor-mediated transcription. Immunofluorescence studies show that endogenous p190RhoGEF localizes to distinct RhoA-containing regions at the plasma membrane, to the cytosol and along microtubules. In vitro and in vivo binding experiments show that p190RhoGEF directly interacts with microtubules via its C-terminal region adjacent to the catalytic Dbl homology/pleckstrin homology domain. Our results indicate that p190RhoGEF is a specific activator of RhoA that requires as yet unknown binding partners to unmask its GDP/GTP exchange activity in vivo, and they suggest that p190RhoGEF may provide a link between microtubule dynamics and RhoA signaling.

Interaction of c-Src with gap junction protein connexin-43. Role in the regulation of cell-cell communication.

Cell-cell communication via connexin-43 (Cx43)-based gap junctions is transiently inhibited by certain mitogens, but the underlying regulatory mechanisms are incompletely understood. Our previous studies have implicated the c-Src tyrosine kinase in mediating transient closure of Cx43-based gap junctions in normal fibroblasts. Here we show that activated c-Src (c-SrcK(+)) phosphorylates the COOH-terminal tail of Cx43, both in vitro and in intact cells. Coimmunoprecipitation experiments reveal that Cx43 associates with c-SrcK(+) and, to a lesser extent, with wild-type c-Src, but not with kinase-dead c-Src. Mutation of residue Cx43 Tyr(265) (Cx43-Y265F mutant) abolishes both tyrosine phosphorylation of Cx43 and its coprecipitation with c-Src. Expression of c-SrcK(+) in Rat-1 cells disrupts gap junctional communication. Strikingly, the communication-defective phenotype is bypassed after coexpression of the Cx43-Y265F mutant or a COOH-terminally truncated version of Cx43 (Cx43Delta263) that lacks residue Tyr(265). Our results support a model in which activated c-Src phosphorylates the COOH-terminal tail of Cx43 on residue Tyr(265), resulting in a stable interaction between both proteins leading to inhibition of gap junctional communication.

Connexin-43 interactions with ZO-1 and alpha- and beta-tubulin.

Gap junctions are composed of connexins that form transmembrane channels between adjacent cells. The C-terminal tail of connexin-43 (Cx43), the most widely expressed connexin member, has been implicated in the regulation of Cx43 channel gating. Interestingly, channel-independent processes regulated by Cx43 have also been postulated. In our studies to elucidate the mechanism of Cx43 channel gating by growth factors and to explore additional functions of gap junctions, we have identified three interacting partners of the C-terminal tail of Cx43 (Cx43CT). (i) the c-Src tyrosine kinase, which phosphorylates Cx43CT and is involved in G protein-mediated inhibition of Cx43 gap junctional communication. (ii) the ZO-1 'scaffold' protein, which might recruit signaling proteins into Cx43-based gap junctions. (iii) microtubules (consisting of alpha/beta-tubulin dimers), which extend with their distal ends to Cx43-based gap junctions, suggesting that Cx43 gap junctions may play a novel role in regulating microtubule stability in contacted cells. Here we show that Cx43 binds alpha-tubulin equally well as beta-tubulin. In addition, we show that the second, but not the first, PDZ domain of ZO-1 binds directly to Cx43, and we confirm that the very C-terminal isoleucine residue of Cx43 is critical for ZO-1 binding.

Feasibility and pharmacokinetics of intraperitoneal suramin in advanced malignancy.

PURPOSE: Bioactive lipids have been causally linked to intraabdominal malignancies such as ovarian cancer. In advanced tumors confined to the peritoneal cavity. inhibition of lipid growth factors present in ascites might induce tumor remissions. The systemic toxicity of the growth factor inhibitor suramin has so far hampered its use in standard oncologic practice, but this could be alleviated by intraperitoneal administration. In this study the feasibility, toxicity and pharmacokinetics of intraperitoneal suramin administration are described. METHODS: Patients with histologically verified cancer confined to the abdominal cavity, for which no effective therapy was available, were treated with intraperitoneal suramin through a Tenckhoff catheter. Patients with ascites were treated with low-volume continuous i.p. infusions of 500 mg/24 h, and patients without ascites were treated with intermittent large-volume i.p. infusions of 1000 mg three times a week. Regular pharmacokinetic sampling of plasma and ascites fluid was carried out. Patients were treated for 6 weeks or until development of progressive disease or until plasma suramin levels exceeded 250 mg/l. RESULTS: Nine patients were treated in ten periods, three with intermittent i.p. suramin, and seven with continuous i.p. suramin, for a median of 28.5 days (16-42 days), with a median suramin dose of 12 g (range 9 21 g ). Treatment was discontinued because of high systemic suramin levels in three patients (all in the intermittent schedule), progressive disease (five patients) or completion of planned treatment (one patient). Toxicity was mild, without any of the systemic side effects commonly associated with suramin. Intraperitoneal suramin levels were consistently higher than plasma levels in all patients, but this effect was most marked in the continuous infusion schedule. CONCLUSIONS: Intraperitoneal suramin infusion in patients with advanced peritoneal cancers is feasible and well-tolerated. Continuous low volume i.p. infusion in patients with ascites confers the largest pharmacokinetic advantage.

Development of our current understanding of bioactive lysophospholipids.

Lysophosphatidic acid (LPA) serves as the prototypic lysophospholipid mediator that acts through G-protein-coupled receptors to evoke a host of responses in numerous target cells. The hormone- and growth-factor-like activities of LPA, mediated by distinct G proteins, were discovered about 10 years ago. Since then, considerable progress has been made in our understanding of LPA receptor signaling, culminating in the recent identification of a growing family of heptahelical receptors specific for LPA and the structurally related lysolipid, sphingosine-1-phosphate (S1P). In addition to stimulating Gi-Ras-mediated cell proliferation, LPA and S1P induce rapid G alpha 12/13-RhoA-mediated cytoskeletal changes underlying such diverse responses as neurite retraction, cell rounding, and enhanced tumor cell invasiveness. LPA also triggers inhibition of gap-junctional communication. This overview focuses on how our understanding of LPA as an intercellular lipid mediator has developed during the last decade.

Intramolecular interactions between the juxtamembrane domain and phosphatase domains of receptor protein-tyrosine phosphatase RPTPmu. Regulation of catalytic activity.

RPTPmu is a receptor-like protein-tyrosine phosphatase (RPTP) whose ectodomain mediates homotypic cell-cell interactions. The intracellular part of RPTPmu contains a relatively long juxtamembrane domain (158 amino acids; aa) and two conserved phosphatase domains (C1 and C2). The membrane-proximal C1 domain is responsible for the catalytic activity of RPTPmu, whereas the membrane-distal C2 domain serves an unknown function. The regulation of RPTP activity remains poorly understood, although dimerization has been proposed as a general mechanism of inactivation. Using the yeast two-hybrid system, we find that the C1 domain binds to an N-terminal noncatalytic region in RPTPmu, termed JM (aa 803-955), consisting of a large part of the juxtamembrane domain (120 aa) and a small part of the C1 domain (33 aa). When co-expressed in COS cells, the JM polypeptide binds to both the C1 and the C2 domain. Strikingly, the isolated JM polypeptide fails to interact with either full-length RPTPmu or with truncated versions of RPTPmu that contain the JM region, consistent with the JM-C1 and JM-C2 interactions being intramolecular rather than intermolecular. Furthermore, we find that large part of the juxtamembrane domain (aa 814-922) is essential for C1 to be catalytically active. Our findings suggest a model in which RPTPmu activity is regulated by the juxtamembrane domain undergoing intramolecular interactions with both the C1 and C2 domain.

Receptor protein-tyrosine phosphatase RPTPmu binds to and dephosphorylates the catenin p120(ctn).

RPTPmu is a prototypic receptor-like protein-tyrosine phosphatase (RPTP) that mediates homotypic cell-cell interactions. Intracellularly, RPTPmu consists of a relatively large juxtamembrane region and two phosphatase domains, but little is still known about its substrate(s). Here we show that RPTPmu associates with the catenin p120(ctn), a tyrosine kinase substrate and an interacting partner of cadherins. No interaction is detectable between RPTPmu and beta-catenin. Furthermore, we show that tyrosine-phosphorylated p120(ctn) is dephosphorylated by RPTPmu both in vitro and in intact cells. Complex formation between RPTPmu and p120(ctn) does not require tyrosine phosphorylation of p120(ctn). Mutational analysis reveals that both the juxtamembrane region and the second phosphatase domain of RPTPmu are involved in p120(ctn) binding. The RPTPmu-interacting domain of p120(ctn) maps to its unique N terminus, a region distinct from the cadherin-interacting domain. A mutant form of p120(ctn) that fails to bind cadherins can still associate with RPTPmu. Our findings indicate that RPTPmu interacts with p120(ctn) independently of cadherins, and they suggest that this interaction may serve to control the tyrosine phosphorylation state of p120(ctn) at sites of cell-cell contact.

Dynamin is required for the activation of mitogen-activated protein (MAP) kinase by MAP kinase kinase.

Internalization of activated receptors from the plasma membrane has been implicated in the activation of mitogen-activated protein (MAP) kinase. However, the mechanism whereby membrane trafficking may regulate mitogenic signaling remains unclear. Here we report that dominant-negative dynamin (K44A), an inhibitor of endocytic vesicle formation, abrogates MAP kinase activation in response to epidermal growth factor, lysophosphatidic acid, and protein kinase C-activating phorbol ester. In contrast, dynamin-K44A does not affect the activation of Ras, Raf, and MAP kinase kinase (MEK) by either agonist. Through immunofluorescence and subcellular fractionation studies, we find that activated MEK is present both at the plasma membrane and in intracellular vesicles but not in the cytosol. Our findings suggest that dynamin-regulated endocytosis of activated MEK, rather than activated receptors, is a critical event in the MAP kinase activation cascade.

Bioactive lysophospholipids and their G protein-coupled receptors.

Lysophosphatidic acid (LPA) and sphingosine 1-phosphate (S1P) are serum-borne lysophospholipids that signal through their cognate G protein-coupled receptors to evoke a great variety of responses in numerous cell types. In addition to stimulating cell proliferation and survival, LPA and S1P induce profound cytoskeletal changes through Rho-mediated signaling pathways, leading to such diverse responses as cell rounding, neurite retraction, and modulation of tumor cell invasiveness (transcellular migration). A major recent advance is the identification of a subfamily of heptahelical receptors for LPA and S1P.

Activation of RhoA by lysophosphatidic acid and Galpha12/13 subunits in neuronal cells: induction of neurite retraction.

Neuronal cells undergo rapid growth cone collapse, neurite retraction, and cell rounding in response to certain G protein-coupled receptor agonists such as lysophosphatidic acid (LPA). These shape changes are driven by Rho-mediated contraction of the actomyosin-based cytoskeleton. To date, however, detection of Rho activation has been hampered by the lack of a suitable assay. Furthermore, the nature of the G protein(s) mediating LPA-induced neurite retraction remains unknown. We have developed a Rho activation assay that is based on the specific binding of active RhoA to its downstream effector Rho-kinase (ROK). A fusion protein of GST and the Rho-binding domain of ROK pulls down activated but not inactive RhoA from cell lysates. Using GST-ROK, we show that in N1E-115 neuronal cells LPA activates endogenous RhoA within 30 s, concomitant with growth cone collapse. Maximal activation occurs after 3 min when neurite retraction is complete and the actin cytoskeleton is fully contracted. LPA-induced RhoA activation is completely inhibited by tyrosine kinase inhibitors (tyrphostin 47 and genistein). Activated Galpha12 and Galpha13 subunits mimic LPA both in activating RhoA and in inducing RhoA-mediated cytoskeletal contraction, thereby preventing neurite outgrowth. We conclude that in neuronal cells, LPA activates RhoA to induce growth cone collapse and neurite retraction through a G12/13-initiated pathway that involves protein-tyrosine kinase activity.