Introduction

The six known mammalian ADP-ribosylation factor (Arf) proteins are myristylated small GTPases that regulate membrane trafficking and cytoskeletal organization. Arfs are regulated by Arf-GAPs because they require GEFs and GAPs for completion of a full GTP–GDP cycle (Jackson et al., 2000; Randazzo and Hirsch, 2004). Arf-GAP protein with SH3 domains, ankyrin repeats and pleckstrin homology domains (ASAP1), otherwise known as Centaurin β4, Def1, DDEF1, PAG2, and AMAP1, is a member of the Arf-GAP family.

ASAP1 is a multi-domain protein. Through its pleckstrin homology domain, ASAP1 is able to bind phosphoinositide phospholipids. This binding is required for its Arf-GAP activity (Randazzo and Hirsch, 2004). Furthermore, in addition to ankyrin repeats, ASAP1 contains an SH3 domain and proline-rich SH3-binding motifs. These domains allow it to interact with signaling and adaptor proteins such as c-src and Crk (Brown et al., 1998), CD2AP/CMS (Liu et al., 2005), CIN85 (Kowanetz et al., 2004), CrkL (Oda et al., 2003), POB1 and paxillin (Oshiro et al., 2002), cortactin (Onodera et al., 2005), FAK (Liu et al., 2002), and Pyk2 (Andreev et al., 1999; Kruljac-Letunic et al., 2003). A BAR domain at the N-terminus allows ASAP1 to bend membranes, important for the trafficking of EGFR (Nie et al., 2006). Subcellularly, ASAP1 localizes to focal adhesions, membrane ruffles, and is also found perinuclearly (Brown et al., 1998; Randazzo et al., 2000; Liu et al., 2002).

ASAP1 has been implicated in regulating cell motility and invasion (Furman et al., 2002; Oshiro et al., 2002; Liu et al., 2005; Onodera et al., 2005; Hashimoto et al., 2006; Nam et al., 2007; Lin et al., 2008). It is amplified and overexpressed in uveal melanomas and in colorectal, prostate, and breast carcinomas (Buffart et al., 2005; Ehlers et al., 2005; Onodera et al., 2005; Lin et al., 2008). ASAP1 has been functionally linked with breast cancer metastasis (Onodera et al., 2005).

We have earlier defined a panel of 268 genes that are upregulated in metastatic pancreatic and breast carcinoma cells (Nestl et al., 2001). Through further screening of this panel, our attention was drawn to ASAP1 as a gene that might be functionally involved in the dissemination of a broad range of tumor types. Consistently, we show that ASAP1 expression is upregulated in a broad range of tumor types, correlates with poor survival of human colorectal cancer patients, and promotes metastasis of pancreatic tumor cells in vivo.

Results

Refinement of earlier metastasis-associated gene expression profiles

We have earlier identified 268 different genes from two rat tumor progression systems (Bsp73 pancreatic and 13762NF mammary carcinoma models), whose expression is exclusive to or upregulated in metastatic cells (Nestl et al., 2001). To screen for genes that might have a general function in tumor progression, we created custom microarray chips spotted with the 268 genes, and differentially probed them with cDNA derived from a further rat tumor progression model (G and MatLyLu cells, non-metastatic and metastatic cell lines, respectively, of the Dunning prostate tumor progression model (Isaacs et al., 1986)). Thereby, we identified 68 genes that are >twofold higher expressed in the metastatic MatLyLu cells compared with the non-metastatic G cells (Supplementary Table 1). These 68 genes are thus upregulated in three different metastatic tumor models and are, therefore, candidates for genes that may have a function in the progression of several cancer types.

Expression of many of the 68 genes has earlier been correlated with tumor progression, and in some cases, a functional role has been shown. Around a third of the genes have no known function. One of the genes identified was ASAP1. Given its association with tumor progression, we chose to study it further.

ASAP1 expression correlates with progression and is found in a wide variety of human tumor types

To confirm the enhanced expression of ASAP1 in metastatic cells, we performed northern blot analysis using different cell lines from the Dunning model and, as a positive control, 1AS (poorly metastastic), and ASML (highly metastatic) pancreatic carcinoma cells from which we originally derived the ASAP1 cDNA clone. As shown in Figure 1, expression of ASAP1 correlated well with the metastatic proclivity of the cells.

Figure 1
figure 1

(a) Northern blot analysis of ASAP1 expression in cell lines of the Bsp73 pancreatic and Dunning prostate tumor models. The metastatic potential of the cells is indicated as a bar diagram, with increasing metastatic potential being indicated by increasing thickness of the bar. The northern blot was stained with methylene blue to visualize the RNA on the filter (RNA loading control), then it was probed with a rat ASAP1 probe. (b) Expression of ASAP1 protein in lysates of ASML and 1AS cells. ASAP1 was detected using western blot analysis and the 7B12 antibody. The position of the molecular weight markers is indicated (kDa). The western blot was subsequently stripped and probed with an anti-actin antibody as a loading control.

Full size image

To study ASAP1 expression at the protein level, we generated anti-ASAP1 monoclonal antibodies using a human ASAP1–GST fusion protein as immunogen. One of these antibodies (7B12) cross-reacted with both rat and human ASAP1 in both western blot and immunohistochemistry assays (Figure 1b; Supplementary Figure S1 and data not shown). Western blot analysis of 1AS and ASML cells using the 7B12 antibody confirmed the augmented transcription of ASAP1 in ASML cells (Figure 1b). Multiple ASAP1 bands have been earlier reported in other cell types and are due at least in part to alternative splicing (Brown et al., 1998).

To determine the expression of ASAP1 in human tumors, we stained a panel of 18 different tumor types with the 7B12 antibody. In comparison with the corresponding non-neoplastic tissue, ASAP1 expression was upregulated in many types of tumors (Figure 2; Supplementary Table 2). Examples include carcinomas of the stomach, colon, gall bladder, breast, bladder and ovary, papillary carcinoma of the thyroid, and esophageal, head, and neck squamous cell carcinomas. For a few tumor types (for example endometrial cancers), ASAP1 expression was equivalent in the corresponding non-transformed tissue. Cervical carcinoma exhibited weaker staining in the tumors compared with normal cervical epithelium.

Figure 2
figure 2

Expression of ASAP1 in human tumors. Normal esophageal, colorectal, and thyroid tissues (a, c, and e, respectively) and their corresponding tumors (b, d, and f, respectively) were immunostained using the 7B12 antibody and AEC (red color), followed by counterstaining with hematoxylin (blue color). (g) ASAP1 expression correlates with metastasis formation and poor overall survival of patients with colorectal adenocarcinomas. ASAP1 expression was analyzed by qRT–PCR, and compared with the development of metastases, time to metastasis development, and survival of the patients. The Kaplan–Meier plots show the proportion of patients with and without metastasis against time (left lower panel) and the proportion of patients surviving against time (right lower panel).

Full size image

To determine whether ASAP1 expression correlates with tumor progression and metastasis in human cancer patients, we performed qRT–PCR analysis using microdissected tumor material obtained from 42 patients with colorectal adenocarcinomas. None of them had metastases at surgery, but 20 developed distant metastases metachronously. ASAP1 expression in the primary tumors of patients who developed metastases metachronously was significantly higher than in patients who did not develop metastases (P=0.0432, Mann–Whitney Rank Sum Test). Kaplan–Meier survival analysis of these patients also revealed that ASAP1 expression is associated with a short metastasis-free survival, as well as with a poor overall survival (Figure 2g).

rASAP1c is a novel alternatively spliced form of ASAP1

To facilitate functional studies, we cloned full-length rat ASAP1 using 5′-RACE and RT–PCR. Analysis of the resulting full-length ASAP1 cDNAs revealed three different alternatively spliced versions (Figure 3a) that we named rASAP1a, rASAP1b, and rASAP1c. The rASAP1a and rASAP1b sequences correspond to the already-described murine splice variants Shag1a and Shag1b. These variants differ in that one of the three proline-rich domains is spliced out of rASAP1b and Shag1b. Both have been shown to have ARF-GAP activity (Brown et al., 1998).

Figure 3
figure 3

(a) Schematic diagram showing different splice variants of rat ASAP1. The different domains labeled are the BAR domain (BAR), pleckstrin homology domain (PH), Arf-GAP domain (Arf-GAP), Ankyrin repeats (a), proline-rich domains 1, 2, and 3 (P1, P2, and P3) and the SH3 domain (SH3). The position of tyrosine 308 is indicated. (b) Arf-GAP assays show that rASAP1c can activate Arf1 GTPase activity. 1AS cells were transiently transfected with empty vector (pcDNA3.1), or expression constructs for rASAP1c, HA-tagged Arf1, and HA-tagged Arf6 as indicated. The transfected cells were incubated with [32P] orthophosphate. The Arf1 and Arf6 proteins were then immunoprecipitated using anti-HA antibodies, and their GDP/GTP loading status was analyzed by TLC. The upper panel shows the autoradiogram of the TLC plate. The origin and positions of GTP and GDP are indicated. The lower part of the figure shows a western blot of lysates of the transfected cells made in parallel and stained with 7B12 anti-ASAP1 antibodies to show that the transient transfection had worked. Quantification of the radiolabel signal on the autoradiogram was performed using the integrated density function of ImageJ.

Full size image

The rASAP1c sequence is a novel splice variant that, similar to rASAP1b, lacks one proline-rich domain, but additionally lacks a short sequence encoding 15 amino acids at the N-terminus of the protein between the BAR and pleckstrin homology domains (amino-acids 304–318, sequence VGGLYVASRANSSRR, see Figure 3a). This stretch of amino acids contains Tyr308, whose phosphorylation by c-src and Pyk2 inhibits the Arf-GAP activity of ASAP1 (Kruljac-Letunic et al., 2003). To determine whether deletion of this amino-acid stretch affects the ARF-GAP activity of rASAP1c, we performed ARF-GAP activity assays in which we transiently transfected 1AS cells with either rASAP1c or empty vector, together with HA-tagged Arf1 and Arf6 (Figure 3b). The transfected cells were incubated with [32P] orthophosphate. The Arf1 and Arf6 proteins were then immunoprecipitated using anti-HA antibodies, and their GDP/GTP loading status was analyzed by TLC. The GDP:GTP ratio of Arf6 did not change in response rASAP1c. However, expression of ASAP1 strongly shifted the loading of Arf1 from mainly GTP bound to GDP bound (Figure 3b). These data show that rASAP1c has Arf-GAP activity for Arf1, as it promotes hydrolysis of GTP bound to Arf1, but not for Arf6.

We next determined the expression profile of different ASAP1 isoforms in a panel of rat tissues (Supplementary Figure S3). Although the relative expression levels of the different ASAP1 isoforms vary from tissue to tissue, we observed ASAP1c expression in all tissues analyzed. Therefore, expression of ASAP1c in tumor cells is unlikely to represent a cancer-specific escape mechanism to avoid negative regulation of its Arf-GAP activity.

ASAP1 promotes the metastasis of experimental tumors

The cell line 1AS expresses low endogenous levels of ASAP1 and is weakly metastastic (Figure 1). To determine whether ASAP1 has an important function in tumor metastasis, 1AS cells were stably transfected with either rASAP1c, empty vector, or a mutant form of ASAP1 containing a mutation (R811A) in the P1 proline-rich domain (kind gift from Paul Randazzo). This mutant exhibits significantly reduced binding to the SH3 domains of proteins such as c-src and Crk (Brown et al., 1998). Three independent clones for each construct were selected on the basis of ASAP1 protein expression in western blot analyses (Supplementary Figure S2). These clones were injected subcutaneously into syngeneic rats in a spontaneous metastasis assay, and tumor growth and metastasis formation were assessed. No significant difference in take rate, tumor growth, or size of draining lymph nodes was observed between the different groups (data not shown). None of the animals injected with 1AS clones transfected with empty vector or dominant-negative ASAP1 developed lung metastases. However, lung metastases were observed in significant numbers of animals injected with clones expressing rASAP1c (Figure 4a; P<0.002, Fisher's exact test). These data suggest that ASAP1 promotes metastasis, and that binding of ASAP1 to SH3-domain-containing proteins is likely to have a function in its metastasis-stimulating activity.

Figure 4
figure 4

(a) Stable overexpression of rASAP1c in 1AS tumor cells promotes metastasis to the lungs. Independent clones of 1AS cells transfected with rASAP1c (#4, #31, #38), R811A ASAP1 (#36, #52, #55), or with empty vector (#1, #2, #9) were injected subcutaneously into groups of eight syngeneic rats. Once the size of the tumors had reached the legal limit, the animals were killed and the presence or absence of lung metastases was scored. The mean and standard error of the three independent clones is shown. Statistical significance is indicated (Fischer's exact test). (b) Expression of dominant-negative ASAP1 in ASML cells suppresses lung metastasis formation in rats. ASML cells were stably transfected with R811A ASAP1 or with pcDNA3.1 as an empty vector control (see Supplementary Figure S2). Two independent clones from each transfection were injected subcutaneously into groups of eight rats each. When the tumors reached the legal limit, the animals were killed and their lungs metastases measured. The mean and standard error of the average size of the lung metastases is shown for each clone. Statistical significance between the dominant-negative ASAP1 and empty vector control groups was calculated using the Student's t-test.

Full size image

To further support the notion that ASAP1 expression can promote metastatic dissemination, we stably transfected the R811A ASAP1 construct or empty vector into the highly metastatic ASML cells that strongly express endogenous ASAP1 (Figure 1b) and selected stable clones (Supplementary Figure S2). These clones were injected into experimental animals. At autopsy, while there was no significant difference in the numbers of lung metastases when the experimental groups were compared (data not shown), the size of the metastases was smaller in animals injected with R811A ASAP1-expressing ASML cells compared with the controls (Figure 4b; P=0.03, Student's t-test). Expression of R811A ASAP1 had neither effect on tumor take rate or tumor growth in vivo nor on ASML proliferation in cell culture (data not shown). It is, therefore, likely that the lung metastases in the rats bearing R811A ASAP1 tumors were smaller because metastatic dissemination was slower compared with control animals. These data suggest that the R811A mutant acted in a partial dominant-negative manner in these experiments, and substantiate the idea that ASAP1 can promote metastasis.

ASAP1 interacts with h-prune and binds to a variety of SH3 domains

How might ASAP1 promote metastasis at the molecular and cellular levels? Like ASAP1, h-prune directly interacts with cortactin (Zollo, unpublished). It also stimulates metastasis formation, and promotes cell motility through its phosphodiesterase activity (Marino and Zollo, 2007); h-prune also negatively regulates the metastasis suppressor protein nm23-H1 (D’Angelo et al., 2004) that interacts with Arf protein (Palacios et al., 2002). Given these observations, we reasoned that ASAP1 may interact with h-prune. As shown in Figure 5a, h-prune and ASAP1 indeed co-immunoprecipitate. Furthermore, expression of ASAP1 stimulates the phosphodiesterase activity of h-prune (Figure 5b), suggesting a potential mechanism by which ASAP1 might promote metastasis.

Figure 5
figure 5

Protein interaction partners of ASAP1. (a) An expression construct for ASAP1 with an HA epitope at the NH2-terminal was transfected transiently into 293 cells stably transfected with Flag-tagged h-prune. Immunoprecipitation was performed using anti-Flag Abs to pull-down h-prune protein that was western blotted and probed with anti-HA antibody (left). Lanes 2 and 5 show the input HA-ASAP1 proteins. Similar experiments were performed on the right panel, except that pull downs were performed with anti-HA antibody and the western blot was probed with a Flag antibody. The results show that h-prune co-immunoprecipitates with ASAP1 protein in both assays. (b) ASAP1 promotes h-prune phosphodiesterase activity. Cells were cotransfected with either empty vector or ASAP1. The h-prune was then immunoprecipitated from lysates of the cells and its phosphodiesterase activity was determined. (c) ASAP1 interacts with a number of SH3 domains. Purified SH3-domain-containing proteins arrayed and immobilized on a filter (SH3-Domain Array I) were incubated with a lysate from ASML cells (ASAP1). After washing the filter, ASAP1 from the lysate that bound to the immobilized proteins was detected on the filter as for a western blot using the 7B12 antibody and ECL. The control array was treated identically (Control), except that it was incubated with non-specific antibody rather than with anti-ASAP1 antibodies. Alternatively arrays were incubated with lysates from cells transfected with Flag-tagged ASAP1b or ASAP1c and probed with anti-flag antibodies. The proteins arrayed on the filters were as follows: A1, 2: Amphiphysin; B1, 2: Dlg2; C1, 2: VAV-D1; D1, 2: BLK; A3, 4: Lck; B3, 4: EMP55; C3, 4: NCK1-D3; D3, 4: Abl; A5, 6: SPCN; B5, 6: FGR; C5, 6: Y124; D5, 6: PLCgamma; A7, 8: Cortactin; B7, 8: SLK; C7, 8: PEXD; D7, 8: Riz; A9, 10: MLPK3; B9, 10: Nebulin; C9, 10: BTK; D9, 10: PI3beta; A11, 12: Yes1; B11, 12: c-src; C11, 12: RasGAP; D11, 12: ITSN-D1; A13, 14: Abl2; B13, 14: FYB-D1; C13, 14: PSD95; D13, 14: ITSN-D2; A15, 16: SJHUA; B15, 16: Hck; C15, 16: Tin; D15, 16: TXK; A17, 18: Itk; B17, 18: VAV-D2; C17, 18: HS1; D17, 18 GST control; A19, 20; CRK-D2; B19, 20: NOF2-D1; C19, 20: Stam. Full details can be found at http://www.panomics.com. (d) ASAP1b and ASAP1c co-immunoprecipitate with SLK. Expression constructs for Flag-tagged ASAP1b or ASAP1c were transfected into HEK293 cells. Lysates were immunoprecipitated in the presence or absence of anti-Flag M2 antibodies, anti-TNP control antibodies, and Protein G–agarose (Resin) as indicated. Immunoprecipitates were western blotted together with total lysate as a positive control. Blots were probed with anti-SLK antibodies (IB SLK), then stripped and probed with 7B12 (IB ASAP1).

Full size image

The R811A ASAP1 mutant is unable to promote metastasis (Figure 4), suggesting that binding to SH3-domain-containing proteins may be required for ASAP1 to promote metastasis. We, therefore, examined the ability of endogenous ASAP1 protein from ASML cells to interact with a panel of 38 purified SH3-domain-containing proteins (Figure 5c). In comparison with controls, endogenous ASAP1 binds strongly and specifically to the src family members c-Src (B 11, 12), SLK (B 7, 8) Hck (B 15, 16), and Yes1 (A 11, 12). It also binds less strongly to cortactin (A 7, 8) and PAK-interacting protein exchange factor β (betaPIX; C 5, 6). Yes1, cortactin, c-Src, and SLK were all found by RT–PCR to be robustly expressed in ASML cells (data not shown). In addition, we found that Flag-tagged ASAP1b and ASAP1c isoforms bind to a common subset of SH3-domain proteins, including those listed above, and weakly to others such as Lck (A 3, 4). Furthermore, SLK co-immunoprecipitated with both ASAP1b and ASAP1c (Figure 5d). These findings identify potential ASAP1 interaction partners that may functionally contribute to the tumor-relevant functions of ASAP1, and suggest that the ASAP1c isoform has similar SH3-domain-binding activity compared with the ASAP1b isoform.

ASAP1 regulates the adhesive and motile properties of tumor cells

To further investigate the metastasis-promoting function of ASAP1, we transiently transfected 1AS cells with rASAP1c or R811A ASAP1 (Supplementary Figure S2) and analyzed their behavior in a number of assays.

ASAP1 expression had no effect on the proliferation or apoptosis rates of the cells (data not shown). However, ASAP1 had a significant effect on the adhesive properties of the cells, as rASAP1c-transfected cells bound better to all substrates tested except for the BSA control, whereas R811A ASAP1-transfected cells bound worse (Figure 6), implying a dominant-negative activity of this mutant in these assays.

Figure 6
figure 6

ASAP1 promotes cell binding to a variety of ECM substrates. 1AS cells were transiently transfected with either R811A ASAP1 (1AS-R811A), rASAP1c (1AS-rASAP1c), or empty vector (1AS-pcDNA3.1). The cells were then allowed to adhere to plastic coated with the indicated ECM substances and non-bound cells were then washed away. The number of cells binding to each substrate was then quantified colorimetrically (OD595) by staining with crystal violet. The mean and standard error of triplicate samples are plotted. Statistical significance relative to the empty vector control group is indicated (Student's t-test).

Full size image

To determine the effect of ASAP1 on cell motility, we wounded monolayers of 1AS cells transiently transfected with either wild-type ASAP1, R811A ASAP1, or with empty vector, then analyzed migration of cells into the wounded area. As shown in Figure 7a, the expression of wild –type, but not R811A ASAP1, significantly increased the migration of cells into the wounded area. To substantiate these data, we performed Boyden Chamber migration assays using 1AS cells transiently transfected with either rASAP1c, R811A ASAP1, or the control vector. The assays were performed either in serum-free medium or in the presence of an FCS gradient. As shown in Figure 7b, rASAP1c-transfected cells exhibited a substantially higher motility than the control-transfected and R811A ASAP1 cells. Significantly, similar levels of motility were observed regardless of the presence or absence of an FCS gradient, suggesting that ASAP1 expression increases random cell motility.

Figure 7
figure 7

(a) ASAP1 expression promotes closure of a monolayer wound. 1AS cells were transiently transfected with either R811A ASAP1, rASAP1c, or empty vector (pcDNA3.1). The cells were grown as monolayers, then a scratch through the monolayer was made. The reduction of the wounded area 24 h later was calculated. The mean and standard error of three independent experiments are presented. (b) ASAP1 promotes random cell motility. 1AS cells were transiently transfected with either R811A ASAP1, rASAP1c, or empty vector (pcDNA3.1). The cells were plated onto the surface of the transwell chambers in the presence (+ FCS) or absence (− FCS) of an FCS gradient. After 6 h, the cells that had migrated were stained with crystal violet. The number of cells migrated was then measured (OD595) as described in Materials and methods. The mean and standard error of triplicate assays is presented. (c) ASAP1 promotes invasiveness. 1AS cells were transiently transfected with either R811A ASAP1 (1AS-R811A), rASAP1c (1AS-rASAP1c), or empty vector (1AS-pcDNA3.1). The cells were then plated onto the surface of matrigel-coated transwell chambers. After 6 h, the cells that had migrated were stained with crystal violet. The number of cells migrated was then measured (OD595) as described in Materials and methods. The mean and standard error of triplicate assays are presented.

Full size image

To examine the effect of ASAP1 expression on the invasive properties of tumor cells, we performed Boyden Chamber invasion assays using filters coated with matrigel. 1AS cells were transiently transfected with either ASAP1c, R811A ASAP1, or the control vector, and their ability to invade across the matrigel layer was assessed. The rASAP1c-transfected cells showed a significantly higher invasive capacity than the control and R811A ASAP1-transfected cells (Figure 7c). Together, these data show that ASAP1 expression promotes tumor cell motility and invasion, and suggest that binding to SH3-domain-containing proteins is required for this activity.

Discussion

Prompted by secondary analysis of an unbiased screen for metastasis-associated gene expression, we show that expression of ASAP1 is upregulated in a variety of human tumor types and correlates with poor metastasis-free survival and overall survival in human colorectal cancer patients. In a syngeneic animal model of pancreatic cancer, it promotes cell adhesiveness, motility, and invasion, and it stimulates metastasis formation in vivo.

In the course of this work, we identified rASAP1c as a novel splice variant of ASAP1 that lacks a 15 amino-acid sequence at the N-terminus containing tyrosine residue Y308. Phosphorylation of Y308 by c-src and Pyk2 negatively regulates the ARF-GAP activity of ASAP1 (Kruljac-Letunic et al., 2003). A point mutant form of ASAP1 in which Y308 can no longer be phosphorylated still retains Arf-GAP activity (Kruljac-Letunic et al., 2003). However, the rASAP1c isoform retains Arf-GAP activity (Figure 3b). This activity should, therefore, be refractory to negative regulation by c-src and Pyk2 through phosphorylation of Y308. Our data show that rASAP1c is able to promote tumor cell motility, invasiveness, and metastasis, suggesting that Y308 phosphorylation is not required for these activities.

How could ASAP1 expression promote invasion and metastasis? We observed that a mutant form of ASAP1 (R811A), which has impaired binding to SH3-domain-binding proteins, was unable to promote metastasis of 1AS cells, suppressed metastasis of ASML cells, impaired the adhesive properties, and did not stimulate the motility or invasiveness of 1AS cells. Thus, protein–protein interactions mediated by the proline-rich domains of ASAP1 should have a decisive function. Consistently, we show that ASAP1 can bind to the SH3 domain of several members of the Src family, and more weakly to cortactin and betaPIX. Additional SH3-domain-containing proteins may also interact with ASAP1: >100 SH3-domain-containing proteins are known; here we screened 38. Importantly, ASAP1b and ASAP1c bind similarly to a common subset of SH3-domain proteins in the panel tested (Figure 5c), indicating that lack of regulation through Y308 does not affect the ability of ASAP1c to bind to SH3-domain proteins. Our data confirm the binding of ASAP1 to c-src and cortactin (Brown et al., 1998; Onodera et al., 2005). The interaction between ASAP1 and cortactin has earlier been implicated in the invasiveness and metastasis of breast cancer cells (Onodera et al., 2005; Hashimoto et al., 2006). The data also identify SLK, hck, lck, Yes1, and betaPIX as novel potential ASAP1-binding proteins, and we found that SLK co-precipitates with both ASAP1b and ASAP1c. Future experiments will focus on confirming the other putative interactions and identifying the SH3-domain protein(s) that critically interact with ASAP1 in the process of metastasis.

Using co-immunoprecipitation, we also show here that ASAP1 interacts with h-prune, a protein known to promote metastasis (Marino and Zollo, 2007). Furthermore, ASAP1 stimulates the phosphodiesterase activity of h-prune, an activity required for its motility-promoting properties (D’Angelo et al., 2004). These observations provide further insight into potential mechanisms by which ASAP1 promotes metastasis.

The Arf-GAP activity of rASAP1c might also contribute to the pro-metastatic effects we report here. In human breast cancer cells, ASAP1 has been shown to promote invasion and metastasis through regulation of Arf6 (Sabe et al., 2009). However, consistent with other reports (Brown et al., 1998; Furman et al., 2002), we show here that rASAP1c regulates Arf1, but not Arf6. This may reflect cell type-specific differences in the spectrum of Arfs regulated by ASAP1. It remains to be shown whether Arf1 functions similarly to Arf6 in promoting metastasis.

ASAP1 probably exerts its pro-metastatic influence through the enhanced cell motility and invasiveness we observed. The expression of ASAP1 has been associated with the invasiveness of breast and prostate cancer cells (Onodera et al., 2005; Lin et al., 2008). Overexpression of ASAP1 has also been reported to promote the migration of cells in response to PDGF and IGF-1, and to increase the general motility of cells, dependent on the Arf-GAP activity of ASAP1 (Furman et al., 2002). Mislocalization or siRNA-mediated knockdown of ASAP1 inhibits cell migration and impairs EGF-dependent chemotaxis (Liu et al., 2005).

A large number of studies implicate ASAP1 in the regulation of cellular structures such as focal adhesions and membrane ruffles that have a pivotal function in cell motility and invasiveness (reviewed in Randazzo and Hirsch, 2004). The ASAP1-interacting proteins c-src, SLK, Hck, cortactin, and betaPIX have also all been shown to be involved in the regulation of podosome formation, actin-rich dynamic adhesion structures that are found mainly in motile cells, and contribute to tissue invasion and matrix remodeling (Wagner et al., 2002; Frame, 2004; Cougoule et al., 2005; Webb et al., 2005; Zhou et al., 2006; Bharti et al., 2007).

Our data suggest that enhanced ASAP1 activity modestly increases the overall adhesiveness of pancreatic carcinoma cells to a wide range of substrates in a manner that depends on the ASAP1 proline-rich repeats (Figure 6). Overexpression of ASAP1 inhibits cell spreading and membrane ruffling induced by growth factors such as PDGF (Randazzo et al., 2000; Furman et al., 2002; Liu et al., 2002; Nie et al., 2006). However, mislocalization or siRNA-mediated knockdown of ASAP1 also inhibits cell spreading (Liu et al., 2005). The function of the ASAP1 Arf-GAP activity in cell spreading is also unclear (Randazzo et al., 2000; Furman et al., 2002; Liu et al., 2002). These observations presumably reflect the complex multiple activities of ASAP1 and need to be further investigated to understand the underlying molecular biology.

In conclusion, our data implicate ASAP1 as being an important regulator of tumor invasiveness and metastasis in a broad range of tumor types. As a multi-domain adaptor protein, there are many ways in which ASAP1 could influence these events, some of which have already been delineated (reviewed in Sabe et al., 2009). The challenge in the future will be to dissect fully the molecular mechanisms involved. In turn, this should identify novel ways for therapeutically interfering with ASAP1 function in metastatic cancer cells.

Materials and methods

Cell lines

The cell lines AT1, AT2.1, AT3.1, AT6.1, MatLu, MatLyLu, G, ASML, 10AS, 1AS, SW1116, and LS174T have been described earlier (Nestl et al., 2001; Mengwasser et al., 2004).

Tumor experiments in vivo

These studies were approved by the local regulatory authority (Regierungspräsidium Karlsruhe). Syngeneic rats were injected subcutaneously with 5 × 105 cells. Animals were monitored until their tumors grew to 50 mm in one dimension, or until they became moribund, at which time they were killed and an autopsy was performed. At autopsy, the size of the primary tumor was measured and the number, size, and location of metastases in the draining lymph nodes and other organs was assessed. Tumor was snap frozen or fixed in 4% paraformaldehyde for paraffin embedding. Metastasis was defined as lymph nodes >10 mm in diameter, lung nodules >1 mm in diameter, or outgrowth of tumor cells from disaggregated lymph nodes or lungs cultured in selection medium. Lymph nodes and lungs were analyzed histologically to confirm the presence of metastases.

Northern blots

Northern blots were performed as described earlier (Hofmann et al., 1998). Blots were probed sequentially at high stringency using a rat ASAP1 probe (nucleotides 1322–1430 of GenBank Accesion number AF075462) and a GAPDH probe (PstI fragment of plasmid pGAPDH (Fort et al., 1985)).

Cloning of rat ASAP1 homologs

Sequence information from cDNA fragments of rat ASAP1 was used to design primers for 5′-RACE. Modified rat Marathon Spleen cDNA (Clontech-Takara Bio Europe, Saint-Germain-en-Laye, France) was used as a template. Products were cloned into the pCRII-Topo vector (Invitrogen, Eugene, OR, USA). Sequencing allowed the 5′ end of the rat ASAP1 sequence to be defined and primers to be designed for PCR amplification of full-length rat ASAP1 cDNAs—rASAP1F: 5′-CTAGCTAGCTAGTGTGAGACATGAGATCTTCAGCCT-3′ and rASAP1R: 5′-CTAGCTAGCTAGGACAGAAGGTTCTGCTTTTTTGCTAGT-3′. Three independent splice variants of rASAP1 were thereby cloned and sequenced. The rat ASAP1a, b, and c sequences have been submitted to GenBank (accession numbers bankit745585 DQ238622, bankit758444 DQ238623, and bankit745591 DQ238624, respectively). An expression construct for rASAP1c containing an N-terminal FLAG-tag was cloned into the pcDNA3.1 expression vector.

Transfections

1AS and ASML cells were stably or transiently transfected using Tfx50 (Promega, Mannheim, Germany). Transient transfection efficiencies of >90% were routinely achieved. Stably transfected clones were checked for expression of the transfected expression construct by western blotting. Clones showing the highest levels of expression were selected for further experiments (Supplementary Figure S2).

ARF-GAP activity assays

ARF-GAP assays were performed as described earlier (Furman et al., 2002). Briefly, cells were cotransfected with either HA-tagged ARF1 or ARF6 expression plasmids together with rASAP1c expression plasmid or the corresponding empty vector. The cells were subsequently labeled with 325 μCi/ml [32P] orthophosphate (Perkin Elmer Life Science, Waltham, MA, USA), lysed and immunoprecipitated using 2 μg anti-HA antibodies (Abcam, Cambridge, UK) and protein A beads (Oncogene Research Products, La Jolla, CA, USA). Precipitated proteins were eluted, spotted onto PEI cellulose TLC plates (Merck biosciences, Darmstadt, Germany), and resolved for 1–1.5 h using 0.65 M KH2PO4 buffer (pH 3.4).

Generation of anti-ASAP1 antibodies

A fragment of human ASAP1 cDNA (nucleotide 977–1532 of KIAA1249) was cloned downstream of and in frame with the GST gene in the pGEX-1 expression vector. Escherichia coli strain BL-21 was transformed with this construct and the expressed soluble GST–ASAP1 fusion protein purified with glutathione agarose (Sigma, Munich, Germany). The immunization of Balb/C-mice with the fusion protein, production and screening of hybridomas, and purification of the monoclonal anti-human ASAP1 antibodies were performed as described earlier (Harlow and Lane, 1988; Niebuhr et al., 1998).

Immunostaining

Sections of paraffin-embedded human tumor samples (Biocat, Heidelberg, Germany) were deparaffinized and immunostained with 5 μg/ml anti-ASAP1 antibodies followed by biotinylated secondary antibody and StreptABComplex (Dako, Glostrup, Denmark) according to standard procedures. Staining was visualized with AEC. Sections were counterstained with hematoxylin.

Western blots

Cells were lysed, fractionated using 8% SDS–PAGE and blotted onto PVDF membranes (Immobilion-P, Millipore, Bedford, MA, USA). The blots were probed with 1 μg/ml anti-ASAP1 antibody and HRP-conjugated secondary antibodies. Protein signals were detected using ECL (Amersham, Uppsala, Sweden) and subsequent exposure of the blots to X-ray film.

Screening of SH3 protein array

Cell lysates were incubated with SH3-Domain Array I arrays (http://www.panomics.com). Anti-ASAP1 monoclonal antibodies or anti-Flag antibodies were used to detect binding of ASAP1 in the lysates to the SH3 domains on the blots using western blot methodology. Control arrays were incubated with isotype control antibodies.

Co-immunoprecipitation

Hek293 cells were transfected with an expression plasmid for HA-tagged h-prune (D’Angelo et al., 2004) and/or with Flag-tagged ASAP as indicated. Cell lysates in 20 mM Tris–HCl, 2 mM MgAc, 0.3 mM CaCl2, 1mMDTT, 2 μg/ml pepstatin, 2 μg/ml aprotinin, and 2 μg/ml leupeptin were immunoprecipitated using anti-HA or anti-FLAG M2 monoclonal antibodies (Sigma and Roche, Basel, Switzerland) and western blotted. For SLK co-immunoprecipitations, a buffer containing 25 mM HEPES pH7.4, 150 mM NaCl, 10 mM MgCl, 1% lubrol, 10 mM Na3VO4, 1 mM PMSF, and Protease Inhibitor Tablets (Roche) was used, and western blots were probed with anti-SLK antibody ab65113 from Abcam.

h-prune PDE assay

Phosphodiesterase activity assays were performed as described earlier (D’Angelo et al., 2004).

qRT–PCR

Tissue specimens from 42 colon cancer patients were obtained with written consent. All patients were earlier untreated, did not have a history of familial colon cancer, and underwent surgical resection at the Robert-Rössle Cancer Hospital, Berlin. None of them had metastases at the time point of surgery, but 20 developed distant metastases metachronously. Tumor specimens (all adenocarcinomas) were snap frozen in liquid nitrogen and blinded for analysis. Tumor cell populations were microdissected from serial cryosections, total RNA was isolated, and used for quantitative real-time RT–PCR as described earlier (Stein et al., 2002). Primers and probes used to amplify ASAP1 were as follows:

forward 5′-tgactagcaaaacgcagaacc-3′, reverse 5′-acacacattatatccccctcc-3′, FITC 5′-cagtgtgtatatagctgctgttacagagta-3′; LCRed640 5′-gaaactcatggaagggccacctc-3′. Statistical significance of Kaplan–Meier curves was evaluated with the log-rank test.

Adhesion assays

Cells were seeded in serum-free medium onto substrate-coated 24-well plates (2 × 105/well) and incubated for 1 h at 37 °C. After washing, bound cells were fixed with 70% ethanol, stained with 0.1% crystal violet, then destained with 10% acetic acid. The optical density of the solution at 595 nm was determined.

Monolayer wounding (scratch) assay

Cells were plated on single-well chamber slides (Corning, Amsterdam, The Netherlands). On confluence, the monolayer was wounded with the edge of a plastic cell scraper (Renner, Darmstadt, Germany). Photographs of the wound were taken using a Zeiss Axiovert 25 microscope immediately after wounding and 24 h later, and reduction of the wounded area was calculated.

Motility and invasion assays

Cells were seeded in serum-free medium either on top of 50 μl Matrigel (BD Biosciences, San Jose, CA, USA, 1:3 diluted with serum-free medium) in the upper well of a Boyden chamber or on top of the filter of the Boyden chamber (pore size: 8 μm, Corning). Medium containing 10% calf serum was applied to the lower chamber. The cells were incubated for 6 h at 37 °C, fixed with 70% ethanol, stained with 0.1% crystal violet, then destained with 10% acetic acid. The optical density of the solution at 595 nm was determined.