A6 Peptide Activates CD44 Adhesive Activity, Induces FAK and MEK Phosphorylation, and Inhibits the Migration and Metastasis of CD44-Expressing Cells

Short peptides with unique biological activities are emerging as potential modifiers of the biology of many types of cancer. A6 is an 8-amino acid peptide (acetyl-KPSSPPEE-amino) that has been shown to have anti-invasive, antimigratory, and antiangiogenic activities in a variety of in vitro and in vivo model systems (1–3). On the basis of these activities, A6 was advanced into clinical trials in which it was found to have an excellent safety profile with no systemic adverse events (4–6). In one study, conducted in ovarian cancer patients with early biochemical relapse, A6 treatment was associated with prolongation of progression-free survival (6).

CD44 is a heavily glycosylated membrane protein that functions in cell–cell and cell–matrix adhesion (reviewed in Goodison and colleagues; ref. 7). The major ligand for CD44 is hyaluronic acid (HA), which is an integral part of the extracellular matrix formed by tumor and stromal cells. The amino acid sequence of A6 exhibits marked homology with a linear sequence within CD44. The homologous sequence (120-NASAPPEE-127) resides within the HA-binding domain of CD44 in an exposed linker between 2 β strands (8). The A6-like sequence is present in all isoforms of CD44 and straddles the splice junction between standard exons 3 and 4 (9). The asparagine residue, a potential N-glycosylation site, is located 2 residues carboxyl from a cysteine residue that is part of a disulfide bond that is critical for conformational integrity of CD44 and its ability to bind HA (8). Thus, the A6-like sequence in CD44 seems to be located within a region of the ligand-binding domain that is structurally important and potentially conformationally dynamic.

Human CD44 gene is composed of 10 exons that are expressed in the ubiquitous standard isoform of CD44 (CD44s) and 10 variant exons that are alternatively spliced within CD44s at an insertion site after the fifth standard exon (9). In addition to HA, CD44 can bind several other extracellular matrix ligands including osteopontin (10, 11), fibronectin (12), and collagen (13), and it interacts with other membrane proteins, including HER2 (14). Three discrete regions of CD44 interact with HA (15–18). Whereas CD44 is constitutively active on some cells, ligand binding seems to be influenced in a cell-specific manner by N-glycosylation (19) and, in some instances, requires activation induced by the binding of anti-CD44 monoclonal antibodies or treatment with phorbol esters and interleukins (20–22). Several mechanisms have been implicated in CD44 activation including variant exon usage, receptor oligomerization, glycosylation, and sulfation (for a review, see Ponta and colleagues; ref. 23). The key conformational determinants of the activated form of CD44 remain unidentified; however, the interaction of HA with CD44 induces conformational changes that alter the orientation of a crucial HA-binding arginine residue (24).

This study identifies CD44 as a target for the antimigratory and antitumor effects of the A6 peptide. It was found that A6 inhibited chemotaxis, enhanced adhesion to HA, and induced focal adhesion kinase (FAK) and MAP/ERK kinase (MEK) phosphorylation in a manner dependent on CD44 expression. That A6 directly interacts with CD44 was shown by the observation that CD44 is affinity labeled by biotinylated A6 and that A6 perturbs the binding of a monoclonal anti-CD44 antibody (DF-1485). A6 significantly inhibited the generation of lung foci in an experimental metastasis model, suggesting that activation of CD44 by agents such as A6 may represent a novel strategy for preventing the formation of metastases.

The capped A6 peptide (acetyl-Lys-Pro-Ser-Ser-Pro-Pro-Glu-Glu-NH₂) was synthesized, purified, and prepared as a 100 mg/mL solution in sterile PBS by Altheas Technologies, Incorporated. A6C (acetyl-Lys-Pro-Ser-Ser-Pro-Pro-Glu-Glu-Cys-NH₂) was generated (Bio-Quant, Inc.) to introduce a free sulfhydryl moiety for the purpose of tagging A6 with maleimide-(polyethylene glycol)₁₁-biotin (Pierce Chemical Co.). The product of this coupling (A6C-biotin) was purified via reverse-phase high-performance liquid chromatography and subjected to mass spectrometry to confirm placement of the biotin moiety on the peptide. A6C and A6C-biotin bioactivity was confirmed in the chemotaxis assay.

All cell lines were obtained from American Type Culture Collection with the exception of OVCAR4, OVCAR5, and OVCAR8, which were obtained from the Developmental Therapeutics Program of the National Cancer Institute, and A2780, which was obtained from the European Collection of Cell Cultures. Cell lines were cultured for less than 6 months before use in this study with the exception of A2780, which was cultured for more than 50 passages. All culture reagents were obtained from Cellgro Mediatech, Inc. All media were supplemented with 10% FBS, 1,000 U/mL penicillin, 100 μg/mL streptomycin, and 4 mmol/L l-glutamine. SKOV3 cells were cultured in McCoy's 5A medium. OVCAR3, OVCAR4, OVCAR5, OVCAR8, and A2780 cells were cultured in RPMI 1640. The murine melanoma B16-F10 line and all other cells used were maintained in Dulbecco's Modified Eagle's Medium (DMEM), to which 1 mmol/L sodium pyruvate was included with the above supplements.

Chemotaxis of cancer cell lines were done in Boyden chambers essentially as described (25). Cell migration was done through collagen I–coated filters containing 8-μm pores toward the conditioned media of growing NIH3T3 fibroblasts (serum-free DMEM), supplemented with 10 ng/mL vascular endothelial growth factor (CM/V) for 20 hours at 37°C. When assessing the effect of A6, cells were preincubated for 30 minutes at 37°C with 0 to 100 μmol/L A6, before addition to the Boyden chambers. In these instances, peptide was also included in the lower chamber as well. After migration, the membranes were stained with 30% Giemsa and the number of migrating cells determined in 4 microscopic fields (∼1 mm²/field) at a total magnification of ×200. Experiments were conducted in quadruplicate and data are presented as the mean ± SD.

The antimetastatic activity of A6 was tested in the murine B16-F10 lung metastasis model (26). Treatment with A6 (100 mg/kg, s.c. twice daily, n = 15) or DPBS (vehicle, n = 15) was initiated 2 hours postinoculation of 10⁶ cells per mouse and continued for 11 days. The mice were then sacrificed and the lungs harvested and stained with Bouin's fixative and staining solution (Sigma Chemical Co.). For each mouse, the number of black B16-F10 tumor nodules in the medial lobe of the right lung and the left superior lobe was determined and the data averaged within the groups.

A6C-biotin (100 μmol/L in cold PBS) was bound to SKOV3 cells and then cross-linked by the addition of 5 mmol/L bis(sulfosuccinimidyl) suberate (Pierce Chemical Co.), a homo-, bifunctional cross-linker reagent that contains an amine-reactive NHS ester at each end of an 11.4 angstrom (8-atom) spacer arm (27). Triton X-100 cell lysates were generated, cleared by centrifugation, and then subjected to immunoblot analyses after transfer to polyvinylidene difluoride (PVDF) membranes. Proteins cross-linked to A6C-biotin were detected by incubating the membranes with 10 ng/mL of streptavidin conjugated to horseradish peroxidase (HRP-SA; EMD Biochemicals). CD44 was detected on blots by staining with 1 μg/mL DF1485 and peroxidase-donkey anti-mouse IgG (Jakson ImmunoResearch). Detection was done with freshly prepared chemiluminescence reagent (ECL Plus; GE Healthcare) and exposure to Hypermax ECL film (GE Healthcare).

Triton X-100 lysates, generated from SKOV3 cells that had been bound and cross-linked to A6C-biotin, were precleared by incubating with Protein A–agarose (Sigma Chemical Co.) for 2 hours at room temperature. Aliquots representing 100 μg of the cleared lysates were then incubated overnight at room temperature with 20 μL of Protein A–agarose plus 2 μg of anti-CD44 (DF1485; Santa Cruz Biotech) or murine IgG1 (Sigma Chemical Co.). After washing 5 times with lysis buffer, the immunoprecipitated proteins were eluted from the Protein A–agarose by boiling in SDS-PAGE sample buffer. The eluted material and 5 μg aliquots of the starting material were then subjected to SDS-PAGE, transfer to PVDF membranes, and blotting with HRP-SA.

Washed cultured cells were suspended in ice cold, serum-free McCoy's 5A media containing 0.5% bovine serum albumin [fluorescence-activated cytometry (FAC) buffer] at 10⁶/mL. Aliquots (0.5 mL) were incubated with 1 μg/mL anti-CD44 antibodies DF1485, B-F24, F-4 (Santa Cruz Biotechnology) or IM7 (Biolegend) for 1 hour at 4°C. When assessing the effect of A6 on antibody-binding, cell aliquots were preincubated with A6 for 30 minutes at 4°C before the addition of primary antibody. After 2 washes, the cells were suspended in cold FAC buffer containing 10 μg/mL fluorescein isothiocyanate–secondary antibody (Jackson ImmunoResearch) and incubated for 30 minutes at 4°C. After a single wash, cell staining was assessed with a Beckman Coulter Epics XL-MCL cytofluorimeter, collecting 10,000 nongated events for each sample using Coulter Epic software for data analysis. The mean fluorescence intensity of the negative control sample (murine IgG₁) for each cell type was adjusted during data collection to a value lesser than 0.5 before analysis of the other samples.

SKOV3 cells were cultured overnight with titrating concentrations (0–100 μmol/L). The cells were then detached, washed, and fluorescently labeled by incubating with 20 μg/mL CFDA-SE (Invitrogen, Inc.) in Hank's buffered saline (Cellgro, Inc.) for 30 minutes at 37°C. After washing, the cells were suspended in DPBS containing 10% FBS at 10⁶/mL and incubated for 30 minutes at room temperature with no additions, 0.001 to 100 μmol/L A6 or with 1 μg/mL IM7, a blocking anti-CD44 rat monoclonal antibody (BioLegend). The cells were then placed into quadruplicate microtiter wells that had been coated with 50 μL of 1 mg/mL HA (Calbiochem #385902) overnight and blocked for 30 minutes with 10% FBS in DPBS. The cells were allowed to adhere at room temperature for 30 minutes, after which the wells were decanted and washed 3 times. The fluorescence in each well was assessed (excitation/emission: 460/536 nm) with an F_max plate fluorimeter (Molecular Device Corp.) and the mean ± SD fluorescence values of the quadruplicate determinations calculated.

Cells were detached from their culture vessels by brief trypsinization, washed 3 times in DPBS, and resuspended at 3.33E5/mL in CM/V. The cells were then treated with or without 100 μg/mL soluble HA (Calbiochem #385902) for 30 minutes at 37°C under 5% CO₂. Aliquots were then plated into 10-cm culture dishes and A6 added to yield final concentrations of 0, 10, 100, and 1000 nmol/L. The cells were cultured for 18 hours at 37°C under 5% CO₂.

After 18 hours of treatment, Triton X-100 lysates were generated in cold radioimmunoprecipitation assay (RIPA) buffer (Thermo Scientific #89901) containing cocktails of broad spectrum phosphatase inhibitors (Calbiochem #524625) and protease inhibitors (Sigma #P8340). For each test condition, the lysates of the nonadherent and adherent populations were generated and combined. Samples representing 20 μg of total protein of each combined lysate were subjected to immunoblot analyses, with rabbit polyclonal antibodies specific for FAK, FAK-pY397, FAK-pY576/577, FAK-pY925, MEK-pS217/221, or a monoclonal anti-MEK (41GP; Cell Signaling Technology).

The integrated densities of the bands of interest were measured using ImageJ software (NIH). The densities of the stained bands of each RIPA lysate were normalized to the integrated density obtained for the mouse–anti-glyceraldehyde-3-phosphate dehydrogenase signal of the lysate. For each antigen and lysate, the density obtained for the anti-phospho–antigen stained band was normalized to that of the anti-panantigen stained band.

A6 perturbed the chemotaxis of SKOV3 ovarian cancer cells toward NIH3T3 conditioned medium supplemented with VEGF (CM/V). A6 blocked the migration of SKOV3 cells in a concentration-dependent manner. In the representative experiment shown in Fig. 1A, the IC₅₀ was 92.4 nmol/L. The mean IC₅₀ of 3 experiments was 40.0 ± 26.7 nmol/L (SEM). To assess the breadth of its antimigratory activity, the ability of A6 to inhibit chemotaxis was tested against a panel of additional cell lines. As shown in Table 1, A6 was effective at blocking the migration of the OVCAR8, OVCAR3, ES2, IGROV-1, MDA-MB-468, and MDA-MB-361 cells with IC₅₀ values in the range of 10 to 100 nmol/L. The MDA-MB-231 breast cancer line exhibited an intermediate IC₅₀ value of 288 nmol/L. A6 was less effective at inhibiting migration of MDA-MB-435 and CaOV3 cells, exhibiting IC₅₀ values in the μmol/L range. For 3 cell lines, OVCAR4, OVCAR5, and a late passage population of A2780, migration was not inhibited by A6 at any concentration tested. Thus, although A6 was very potent against some cell lines, it lacked any activity against others, suggesting a mechanism dependent upon expression of a specific receptor or signal transduction pathway. A6 was not cytotoxic at any concentration used in this study (data not shown).

To determine whether A6 inhibits the colonization of secondary tissues by circulating cancer cells, its activity was tested in the well-characterized B16-F10 lung metastatic model. In vitro, A6 was as effective in blocking the chemotactic migration of B16-F10 cells as it was for SKOV3 cells. In the representative experiment whose results are shown in Fig. 1B, the IC₅₀ was 28.8 nmol/L; the mean IC₅₀ of 3 experiments was 21.3 ± 7.6 (SEM) nmol/L. In the in vivo model C57Bl/6 mice were injected intravenously with 10⁶ cells and then treated with either vehicle alone or A6 at a dose of 100 mg/kg s.c. twice daily for 11 days before sacrifice and enumeration of the number of nodules in the lungs. As shown in Figs. 1C and D, treatment with A6 reduced the number of lung nodules. In the representative experiment shown in Fig. 1C, A6 reduced the number of lung metastases to 50% of control. In 3 independent experiments, the number was decreased to just 49.4 ± 16.0 (SEM)% of those in the control mice (P = 0.0029 in an unpaired t test). Body weights of both groups were stable throughout the short study, and no gross changes other than those associated with metastatic lung disease were noted at postmortem. Thus, A6 exhibited substantial antimetastatic activity in this aggressive in vivo model.

The panel of ovarian cancer cell lines was screened for association of the expression of cell surface molecules known to be required for migration and sensitivity to the antimigratory effect of A6. As shown in Fig. 2, a correlation was found between the expression of CD44 and sensitivity to A6, as tested by flow cytometric analysis with 4 different anti-CD44 antibodies. All 4 antibodies were made against intact cells or partially purified protein. The epitopes to which these antibodies bind have not been mapped, with the exception of IM7. A6 greatly inhibited the migration of SKOV3 and OVCAR3 cells but failed to inhibit OVCAR4 and late passage A2780 cells. A6 slightly inhibited the migration of OVCAR5 cells; however, the effect was not clearly concentration dependent over the range tested. Flow cytometric analyses (Fig. 2B) showed the same relative pattern in CD44 expression levels for each of the anti-CD44 antibodies tested: SKOV3>OVCAR3>OVCAR4, OVCAR5>A2780. Thus, the 2 cell lines most affected by A6 in the migration assay (SKOV3 and OVCAR3) expressed the greatest levels of CD44. Conversely, the lines expressing the lowest level of CD44, A2780, and OVCAR4 were least affected.

The correlation between sensitivity to the antimigratory activity of A6 and expression of CD44 suggested that A6 may directly interact with CD44. To explore this possibility, SKOV3 cells were bound and cross-linked to biotinylated A6 peptide (A6C-biotin). Lysates prepared from the cross-linked cells were subjected to immunoprecipitation using an anti-CD44 monoclonal antibody (DF1485; Santa Cruz Biotech.) or its isotype control. The eluted proteins, as well as starting lysates, were subjected to PAGE/blotting and then stained for the presence of the biotin tag and CD44 with HRP-SA and the DF1485 antibody, respectively.

As shown in Fig. 3A, A6C-biotin labeled a number of cellular proteins in the starting material (first lane of the left panel). HRP-SA stained only 3 bands in the lysates of cells that had been cross-linked in the absence of A6C-biotin, indicating that, with these exceptions, HRP-SA staining of the A6C-biotin-labeled proteins in the samples cross-linked in the presence of A6C-biotin was specific. DF1485 specifically precipitated a large band representing protein of approximately 85 kDa that had been tagged with A6C-biotin. This band was stained to a significantly greater extent in the precipitate compared with the starting material, indicating that this protein was greatly enriched by the anti-CD44 immunoprecipitation. DF1485 stained a protein with this molecular weight in the starting material and precipitates generated from both the A6C-biotin-labeled and control cells, confirming the identity of CD44 on these blots and the presence of CD44 in all samples. These results indicate that A6 interacts with CD44 in a configuration that is intimate and stable enough to undergo cross-linking with bis(sulfosuccinimidyl) suberate.

Further evidence for an interaction between A6 and CD44 was provided by flow cytometric analysis of the ability of A6 to interfere with the binding of the anti-CD44 DF1485 antibody to SKOV3 cells. As shown in Fig. 3B, treatment of the SKOV3 cells with 10 μmol/L A6 for 30 minutes before incubation with 1 μg/mL DF1485 reduced the mean fluorescence intensity obtained. Maximal inhibition was observed with 10 μmol/L A6 which reduced the MFI by 38.4 ± 15% (SEM, n = 4). A6 treatment did not significantly reduce the binding of any of 4 other anti-CD44 antibodies tested, including the ligand blocking IM7 rat monoclonal antibody (data not shown). This result indicates that A6 does not nonspecifically change the entire extracellular structure of CD44 but rather produces a localized change in a single epitope of CD44. Unfortunately, the epitope to which DF1485 binds has not been mapped; nevertheless, the result is consistent with the idea that A6 induces a relatively subtle change in the structure of one part of the CD44 receptor.

HA is a principal ligand for CD44. The ability of A6 to affect this interaction was investigated with an assay measuring the adhesion of fluorescently labeled cells to HA-coated microtiter wells. In this assay, approximately 10% of the 50,000 SKOV3 cells added to each HA-coated well adhered after 30 minutes. Inclusion of a blocking monoclonal antibody (IM7) at a concentration of 1 μg/mL inhibited adhesion of SKOV cells to HA by 61.3 ± 10.2% (SEM, n = 4), showing the degree of specificity of this assay for CD44 activity.

Figure 4A shows that A6 produced a concentration-dependent increase in the number of SKOV3 cells adhering to HA-coated plates. The EC₅₀ was determined by nonlinear regression analysis to be 1.08 μmol/L. EC₅₀ values determined from a series of 5 experiments in which adhesion times (15 or 30 minutes), the number of cells added to each well (50,000 to 100,000), and washing techniques were altered, ranged from 0.4 to 5.03 μmol/L.

The ability of A6 to augment adhesion to HA of 3 cell lines expressing different levels of CD44 was assessed (Fig. 4B). Fewer than 10% of the A2780 cells, which do not express CD44, adhered to HA-coated plates. This low level adhesion was not blocked by IM7, indicating that it was not CD44 dependent. As expected, A6 did not enhance the adhesion of A2780 cells. In contrast, the adhesion of SKOV3 and OVCAR3 cells, which expressed relatively greater and lesser amounts of CD44, respectively, was blocked by IM7 indicating CD44 dependence. Importantly, treatment with 10 μmol/L A6 augmented the adhesion of the CD44-positive SKOV3 and OVCAR3 cells but did not enhance the CD44-independent adhesion of the A2780 cells. These results indicate that A6 not only binds to CD44, it also alters at least one of the functions of CD44.

To identify signaling events induced or modulated by A6, a survey was made of the expression and phosphorylation of selected signaling proteins in SKOV3, B16-F10, and A2780 cells cultured in the presence of chemotactic stimuli and the presence or absence of A6 and soluble HA. The proteins surveyed included FAK, Rho, the p85 subunit of phosphoinositide-3-kinase (PI3K), protein kinase Cα (PKCα), Akt, phospholipase D (PLD), Erk1/2, and MEK. The most consistent and striking results obtained from this survey were that A6 enhanced phosphorylation of tyrosine residues Y397, 576/577, and Y925 of FAK and serine residues S217/S221 of MEK in the A6-responsive SKOV3 and B16F10 lines, but not in the nonresponsive A2780 line (Fig. 5). Densitometric analyses of the blots presented in Fig. 5A showed 30-, 6.7-, and 7-fold increases in the integrated densities of the FAK pY397, pY576/577, and pY925 signals, respectively, when SKOV3 cells were cultured in the presence of 100 nmol/L A6. The signals obtained with the anti-pY576/577 and pY925 antibodies for the B16-F10 cells cultured in the presence of A6 were approximately 2.2-fold greater than that of those obtained when the cells were cultured in its absence. Interestingly, the pY397 signal obtained with the B16-F10 cells did not respond to A6. The percent of total FAK that became phosphorylated could not be determined as the pan anti-FAK antibody failed to detect the phosphorylated forms. Treatment with A6 did not increase the FAK pY397, pY576/577, and pY925 signals of the non-CD44–expressing A2780 cells. In fact, a decrease in the pY925 signal was observed in the experiment presented in Fig. 5A. A6 treatment of SKOV3 and B16-F10 cells, but not A2780 cells, enhanced MEK phosphorylation (Fig. 5B). Densitometric analyses showed a 2.1- to 2.5-fold increase in the integrated densities of the pS217/221 signal generated for the CD44-positive cell lines, when cultured in the presence of 100 nmol/L A6. These results indicate that in the cells that expressed CD44, but not in those that did not, A6 activated elements of a signal transduction cascade known to be linked to CD44.

The effect of soluble HA on the ability of A6 to enhance FAK phosphorylation in SKOV3 cells was examined. SKOV3 cells were incubated with or without 100 μg/mL soluble HA for 30 minutes before being cultured in CM/V in the presence of 0, 10, 100, or 1,000 nmol/L A6. Lysates were prepared after 18 hours of culture and subjected to Western blot analysis for the levels of FAK, FAK pY397, pY576/577, and pY925. As shown in Fig. 5C and D, pretreatment with soluble HA and its inclusion in the assay antagonized the A6-induced phosphorylation of FAK. A clear dose-dependent induction of the pY397, pY56/577, and pY925 signals was observed in the absence of HA, with the response to 100 nmol/L A6 being similar in magnitude to those generated in the experiment presented in Fig. 5A. The response was biphasic; the signals induced by culture with 1 μmol/L A6 were reduced compared with those induced with 100 nmol/L A6. For all pY antigens and all concentrations of A6 tested, the inclusion of soluble HA reduced the anti-pY signal intensities. Although the nature of this antagonism is unclear, the demonstration that a CD44 ligand modulates the signaling response to A6 provides additional evidence for a functional relationship between A6 and CD44.

In this study, the A6 peptide was found to inhibit the chemotaxis of a subset of the ovarian and breast cancer cell lines (Table 1). The antimigratory activity of A6 correlated with CD44 expression (Fig. 2), suggesting that A6 may modulate CD44-dependent activities leading to the inhibition of cancer cell motility. That A6 directly interacts with CD44 is indicated by the demonstration that A6C-biotin affinity labeled SKOV3 CD44 and perturbed the binding of DF1485, a murine anti-CD44 monoclonal antibody, to intact SKOV3 cells (Fig. 3). This partial inhibition does not seem to be the result of a competition of A6 and CD44 for the antibody; the antibody failed to recognize the peptide or inhibit the binding of an anti-A6 antibody in an ELISA assay when the peptide was absorbed onto microtiter plates (data not shown). Therefore, A6 may induce conformational or other changes in the receptor that either lower the affinity of the epitope for the antibody or exclude the antibody from binding. Notably, A6 did not affect the binding of other anti-CD44 antibodies (i.e., B-F24, F-4, and IM7 antibodies; data not shown), showing that A6 treatment did not reduce CD44 levels over the time course of the staining required for the assay and suggesting the A6-induced changes in CD44 are localized. Importantly, the changes induced in CD44 by A6 represent an activation of the receptor: A6 potentiated cell adhesion to HA, a principal CD44 ligand (Fig. 4). This ability of A6, like its antichemotactic activity, was restricted to CD44-positive cells.

The most striking results obtained from the initial survey of the effects of A6 on the expression and phosphorylation of cancer cell signaling proteins was the demonstration that A6 elevates pFAK (pY397, pY576/577, and pY925) levels in the A6-responsive SKOV3 and B16F10 lines but failed to do so in the nonresponsive A2780 line. FAK has been implicated in diverse cellular roles including cell locomotion, mitogen response, and cell survival, primarily through its role in modulating the cytoskeleton (reviewed in Schaller and colleagues; ref. 28). FAK associates with the cytoplasmic domains of integrin adhesion receptors in focal adhesions and is activated by autophosphorylation of Y397 in response to integrin clustering (29, 30) This site serves a binding site for members of the Src family of tyrosine kinases which phosphorylate additional FAK tyrosine residues, including Y576/77 and Y925 (29, 30). Src binds with high affinity to a single site on the cytoplasmic domain of CD44 and Src activity is stimulated as a result of the HA-CD44 interaction (31, 32). Phospho-Y925 serves as a docking site for Grb and triggers Ras-dependent activation of the MAP kinase cascade, including MEK phosphorylation (33), as was observed in this study (Fig. 5B). Thus, A6-induced activation of CD44 seems to initiate signaling through CD44-mediated Src activation of FAK. Importantly, inclusion of soluble HA in the culture of SKOV3 cells with A6 abrogated the A6-induced FAK phosphorylation (Fig. 5, panels C and D). The precise mechanism for this antagonism remains to be determined.

The stimulatory effect of A6 on the adhesion of SKOV3 cells exhibited an IC₅₀ of 1 μmol/L, 2 orders of magnitude greater than that observed for its inhibitory effect on cell migration. The reason for this discrepancy is unclear, given that both phenomena seem to correlate with CD44 expression. It is possible that the discrepancy is due to A6 absorbing onto the HA-coated plates, thus lowering the effective concentration of soluble A6 in the adhesion assay. Alternatively, the 2 phenomena may have different thresholds as to the number of activated CD44 molecules required to initiate a measurable effect. It should also be noted that it is likely that the cellular events taking place in each assay may render CD44 conformationally different, thus presenting different affinities for A6. Lastly, it can not be excluded that A6 may exert an effect on cell adhesion to HA through a CD44-independent mechanism and that the positively effected cell lines tested to date surreptitiously express this receptor along with CD44 and that the negative A2780 line lacks both receptors.

It is likely that the sequence within the ligand-binding domain of CD44 sharing homology with A6 (CD44: 120-NASAPPEE-127) will be central to the mechanism by which A6 activates the adhesive activity of CD44. A6 may mimic the CD44 sequence and engage in a homotypic interaction that induces conformational changes in CD44 and/or CD44 dimerization leading to greater affinity and/or avidity for ligand. As mentioned, the perturbation of DF1485 binding by A6 does indeed suggest that A6 is inducing conformational changes in the receptor. Alternatively, the ability of A6 to mimic the CD44 sequence may permit it to interact with a CD44-binding partner resulting in release of repression of CD44 activity. Lastly, the possibility that A6 interacts with another membrane protein independent of CD44 and induces signals that secondarily activate CD44 cannot be excluded at this time.

From a clinical point of view, the most important observation to emerge from these studies is that A6 inhibits the formation of metastases in vivo. Because CD44 is a marker of cancer stem cells and has been linked to the expression of drug-resistant phenotypes (reviewed in Toole and colleagues; ref. 34), an intriguing issue currently being addressed is whether A6 modulates drug resistance. The ability of A6 to activate CD44 suggests that the peptide may modulate this critical determinant of therapeutic efficacy as well.

No potential conflicts of interest were disclosed.

The work was support by Ångstrom Pharmaceuticals, Inc., Solana Beach, CA.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.