Metastasis, the spread of malignant tumor cells from a primary site to distant sites, is the most life-threatening complication of cancer and a major problem of cancer treatment.1, 2 The metastatic process consists of multiple steps: (i) dissociation of tumor cell(s) from the primary site with a concomitant loss of cell-cell and cell-extracellular matrix (ECM) adhesions; (ii) tumor-cell adhesion to and subsequent local digestion of basement membrane; (iii) retraction of endothelial cells and subsequent intravasation; (iv) survival within the vasculature; (v) extravasation from vasculature at a distant site and (vi) growth in a “foreign” or ectopic organ environment.3, 4 Much attention has been paid to the interaction of cell-cell or cell-ECM adhesion5–7 and to proteolysis by a variety of classes of degradative enzymes and their inhibitors,8, 9 even though tumor cell migration is also an essential step for establishment of cancer metastasis. It is well known that the actin cytoskeleton is a major component of the cell motility machinery.10, 11 Evidence is accumulating that changes in actin cytoskeletal organization, adhesiveness and motility are important not only for tumor development and progression but also may be critical for determining invasion and metastatic potential of tumor cells.12–14 However, the molecular mechanisms underlying the migration of tumor cells are not yet well understood.
Gelsolin is a representative actin-regulatory protein with an 82 kDa mass and is present in most vertebrate tissues.15, 16 Gelsolin controls the length of actin polymers in vitro by a variety of mechanisms. At least 3 different activities (severing, capping and nucleating through interaction with both filamentous (F-) and monomeric (G-) actins) are responsible for reorganization of the actin cytoskeleton. These functions are tightly regulated by calcium ions, pH and polyphosphoinositides.17 Since actin filament reorganization is important for cell shape and motility, gelsolin has crucial roles in the control of these cellular functions.18, 19 The amino acid sequence of the gelsolin molecule has 6 homologous repeats (designated G1–G6), and extensive studies using proteolytic fragments and recombinant truncates of gelsolin indicate that the various functions of gelsolin involve the cooperative interaction of the domains encoded by repeated sequences. About 100 carboxyl-terminal amino acids confer calcium regulation upon the gelsolin function.20–24
We have reported previously a tumor-suppressive function of His321, a mutant gelsolin with proline to histidine amino acid substitution at residue 321 in ras-transformed NIH/3T3 cells.25 His321 protein has higher actin monomer binding ability, reduced severing and increased nucleating activities in vitro. In addition, His321 can bind PIP2 more strongly and inhibit phospholipase Cγ activity more efficiently than wild-type gelsolin does.26 Furthermore, we have demonstrated that gelsolin expression is frequently downregulated in several types (gastric, bladder, colon and lung) of human cancers.27–30 Ectopic expression of wild-type gelsolin resulted in suppression of tumorigenicity of both bladder and colon carcinoma cell lines.28, 29 In addition to our studies, other groups have also reported downregulation of gelsolin expression in breast,31, 32 endometrial and ovarian33 and prostate carcinomas.34
Our study was designed to assess the effects of gelsolin overexpression on metastasis and to determine the importance of a carboxyl-terminus that confers Ca2+ dependency on gelsolin for its effects. We transfected expression vectors with cDNA encoding either full-length wild-type, His321 mutant form or carboxyl-terminal truncate (C-del) of gelsolin into a highly metastatic murine melanoma cell line, B16-BL6, and examined phenotypes of transfectants, such as growth rate, colony formation in soft agar, cell motility in vitro and metastasis formation in vivo.
MATERIAL AND METHODS
The mouse melanoma cell line, B16-BL6,35, 36 and NIH3T3 cells were grown in DMEM (Nissui, Tokyo, Japan) supplemented with 10% FCS (GIBCO-BRL, Life Technologies, Rockville, MD) and 0.03% L-glutamine (Nissui). Culture was maintained at 37°C in a moist atmosphere of 95% air and 5% CO2.
Expression vectors and DNA transfection
The expression vectors carrying mouse wild-type and His321 mutant full-length gelsolin cDNAs were described previously.25 The carboxyl-terminus-truncated gelsolin (C-del) cDNA was created by PCR-mediated mutagenesis as described previously37 and subcloned into multicloning sites of LK444. As a selection marker, 1 μg of pMiwhph carrying the hygromycin B-resistant (hyg Br) gene (deposited by Dr. Atsushi Kawakami, Nagoya University), obtained from Health Science Research Resources Bank (HSRRB) in Japan, was cotransfected with the expression vectors (10 μg) into B16-BL6 cells (1 × 106 cells/10 cm dish) using lipofectin reagent (GIBCO-BRL, Life Technologies). Transfected cells were selected by culturing them in a selection medium containing 400 μg/ml hygromycin B (Wako Chemicals, Tokyo, Japan) for 14–20 days and individual colonies were picked randomly and then expanded.
Polymerase chain reaction (PCR) and reverse transcription-polymerase chain reaction (RT-PCR)
Genomic DNA of hygromycin B-resistant transfectants was extracted with 1× PCR buffer with nonionic detergents and Proteinase K (Boehringer Mannheim, Mannheim, Germany) (10 mM Tris-HCl, pH 8.3, 2.5 mM MgCl2, 50 mM KCl, 0.1 mg/ml gelatin, 0.45% NP-40, 0.45% Tween 20 and 60 μg/ml Proteinase K). PCR was carried out using the expression vector-specific LK2 primer (5′-GCC AGG ATC AGT CGA C-3′) and AS1 primer (5′-AAA GGC ACT GAT TGG TGA-3′) to confirm the integration of full-length gelsolin cDNA. Total RNA was isolated with RNAzol (Cinna/Biotex Laboratories, Houston, TX) or TRIzol reagent (GIBCO-BRL, LifeTechnologies) according to the manufacturer's instructions. After DNase I treatment, cDNA were reverse transcribed with the AS7 primer (5′-CCG CCA GTT CTT GAA GAA-3′) using Superscript II (GIBCO-BRL, Life Technologies) and PCR was carried out with LK2 for detection of expression vector-derived gelsolin mRNA or S1 (5′-TAT TGG CTG GGCAAT GAA-3′) for endogenous gelsolin mRNA.
Two-dimensional polyacrylamide gel electrophoresis (2D-PAGE)
Two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) was performed as described previously.38 Polypeptides were visualized with a silver-staining kit (2D-Silver stain II; Daiichi Pure Chemicals, Tokyo, Japan).
Antibodies and Western blot analysis
Antimouse gelsolin antisera were prepared as described previously.39 Briefly, 3 rabbits were immunized with purified mouse gelsolin produced in E. coli26 in Freund's complete adjuvant subcutaneously in their backs. After 2 boosts with purified gelsolin in incomplete adjuvant, rabbits were bled. Production of antimouse gelsolin sera were checked by Western blot analysis using preimmune sera as a negative control. Cells were lysed with 1× SDS sample buffer (62.5 mM Tris-HCl pH 6.8, 2% SDS and 5% 2-mercaptoethanol) and boiled for 5 min. In some cases, protein was extracted with lysis buffer for 2D-PAGE as described previously,38 and protein concentration was determined by Bio-Rad protein assay (Bio-Rad Laboratories, Richmond, CA) with BSA as a standard. Twenty micrograms of total protein were separated with 10% of SDS acrylamide gel and electrotransferred to nitrocellulose membrane. After blocking with 5% nonfat dry milk in TBST (Tris-buffered saline; 10 mM Tris-HCl, pH 8.0, 150 mM NaCl containing 0.05% Tween-20), they were probed with rabbit antigelsolin antibody (Ra#3/3rd) or mouse antiactin monoclonal antibody (Clone C4; Boehringer Mannheim), followed by incubating with antirabbit (Biosource, Camarillo, CA) or antimouse immunoglobulin antibody (Jackson Immuno-Research Laboratories, West Grove, PA) and then visualized with the ECL system (Amersham Pharmacia, Buckinghampshire, UK). Band images were scanned with a GT-6000 Scanner (Epson-SEIKO, Tokyo, Japan) and densitometric analyses were performed using NIH Image, version 1.56.
Growth curve and soft agar assay
Cells were seeded at a density of 5 × 104 cells per well in 6-well plates in DMEM with either 10% or 5% FCS, respectively. Cell number was determined by counting with a hemocytometer after trypsinization. For the assay of colony-forming efficiency in soft agar, 5 × 103 cells in 1 ml growth medium containing 0.33% Noble agar (DIFCO Laboratories, Detroit, MI) were incubated in 60 mm dishes with lids (Coster Scientific, Corning, NY) overlaid on 4 ml of 0.5% agar medium. Cells were incubated at 37°C in a moist atmosphere of 95% air and 5% CO2, and the number of colonies were counted after 3 weeks.
Cell spreading assay
Cell spreading on a tissue culture plate was examined by observation under a phase contrast microscope. The percentage of spread cells was determined by counting the number of cells showing decreased cell refractility, formation of cell polarity and projections around the cell periphery divided by the total number of cells (at least 300 cells) in the fields.
Cell motility assay
Chemotaxis assays were performed basically as described previously with some modifications.40 An amount of 20 mg/ml of Fibronectin (Boehringer Mannheim) was used as a chemoattractant in the lower compartment of Tranwell™ (Costar Scientific) chambers (600 μl/well). Cells (2 × 104 cells in 100 μl) were added into the upper compartment and incubated for 4 hr. Then the cells that had migrated were counted after filter fixation with 3% glutaraldehyde in PBS and Giemsa staining.
Flow cytometry (FCM)
Cells were harvested with EDTA, washed with ice-cold PBS containing 0.1% BSA and 0.05% NaN3 (BSA buffer) and then resuspended in the staining solution containing a primary antibody and incubated for 60 min on ice. After washing with BSA buffer twice, a second incubation with FITC-conjugated antirat or antirabbit immunoglobulin monoclonal antibody (Jackson ImmunoResearch Laboratories) was performed. After washing with BSA buffer twice, cells were analyzed on a Becton-Dickinson FACS Calibur using CellQuest software. A population of at least 10,000 cells were analyzed in each experiment. The monoclonal antibodies used for analyses of cell surface adhesion molecules were rat antimouse β1 (Chemicon, Temecula, CA) and integrin α4 (Coulter, Marseille, France).
Tumorigenicity and metastasis assay
Cells were trypsinized from monolayer cultures, counted and spun down at 1,200 rpm for 5 min and resuspened with DMEM. Five C57BL/6 syngenic 6-week-old male mice (Japan SLC, Shizuoka, Japan) per clone were injected with 5 × 104 cells of wild-type, His321 mutant or C-del gelsolin transfectants, the neocontrols or parental B16-BL6 cells in a final volume of 50 μl. Mice received intrafootpad (i.f.p.) injections to investigate local tumor growth in the foreign microenvironment of footpad muscles and to monitor postsurgical spontaneous metastasis in the lungs. Tumor size was measured regularly using hand calipers every 2 days. Volumes were calculated by the formula 1/2 × L × W2 where L and W are length and width, respectively, of the tumor measured in 2 dimensions. About 3 weeks after inoculation, the mice were lightly anesthetized with ether, and tumor-bearing feet were amputated. About 40 days after cell inoculation in experiment 1, 3 of 5 mice were sacrificed by overdose of ether and their lungs were examined for metastatic colonies. Two mice were kept until they died spontaneously to assess survival. In experiment 2, 3 of 6 mice were sacrificed on day 60 after cell inoculation and the remains were observed for survival. In vivo experiments were conducted under guidelines for use of experimental animals by the Hokkaido University School of Medicine, and all manipulations with the animals that may have caused pain or distress were done under anesthesia.
Statistical significance of difference was assessed using Student's t-test or Welch's t-test. All p-values are 2-sided and considered statistically significant for p < 0.05.
Expression of exogenous wild-type, His321 mutant and carboxyl-terminus-truncated (C-del) gelsolin in transfectants of B16-BL6 mouse melanoma cells
To obtain clones expressing wild-type, His321 mutant and truncate of gelsolin, expression vectors with cDNA encoding either full-length wild-type, His321 mutant or the carboxyl-terminal truncate (C-del) of gelsolin were transfected into B16-BL6 cells and selected with hygromycin B for 14–20 days. Expression of the transfected wild-type gelsolin in 12 clones from 2 separated dishes (6 clones each) was assessed by Western blot analysis with antimouse gelsolin antibody prepared as described in Material and Methods, and the densitometric analysis was carried out. About 1.5-fold of gelsolin polypeptides were detected in 319S-A5 and -B5 compared to parental B16-BL6 cells (Fig. 1a), and they were used for further analyses. Since His321 mutant gelsolin has more basic pI value due to an amino acid substitution, proline to histidine,25 it can be distinguished from the endogenous wild-type gelsolin by 2D-PAGE. Two representative clones expressing the His321 polypeptide are shown in Figure 1b. One- to 1.5-fold of the His321 spots compared to endogenous gelsolin were detected in His321 transfectants, L2S-A1 and -B2. In addition, the integration of transfected cDNA was confirmed by PCR analysis using genomic DNA from each hygromycin B-resistant clone, including the neo-controls (444-1 and 444-3) as a template. DNA fragments with an expected size (2,237 bp) were detected in both wild-type and His321 gelsolin full-length cDNA transfectants (data not shown). As shown in Figure 1c, DNA fragments derived from transfected cDNA were detected in both wild-type and His321 gelsolin full-length cDNA transfectants using the LK2 sense primer, specific for the β-actin promoter sequence, and the AS7 gelsolin antisense primer, but were not detected in the neocontrols and parental B16-BL6 cells. DNA fragments amplified with the S1 and AS7 primers containing both endogenous and exogenous gelsolin were detected in all clones by RT-PCR, demonstrating intactivity of RNA samples. Polypeptide of the carboxyl-terminal truncate C-del gelsolin in transfectants was detected by Western blot analysis with antimouse gelsolin antibody as described above for transfectants of wild-type gelsolin (Fig. 1d).
Characterization of transfectants expressing wild-type, His321 mutant and C-del gelsolin in vitro
Cell morphology of the representative transfectants expressing wild-type (319S-A5), His321 mutant (L2S-A1) and C-del (Cd-D3) gelsolin, as well as the neocontrols (444-1) and parental B16-BL6 cells are shown in Figure 2a. At the initial phase of cell attachment to the culture dish about 4 hr after plating, wild-type and His321 mutant gelsolin transfectants were still rounded and less attached on dishes, whereas C-del gelsolin transfectants, the neocontrols and parental B16-BL6 cells exhibited irregular shapes at their periphery, sickled and polarized with active membrane ruffles. Figure 2b shows that cell spreading of clones expressing full-length (wild-type and His321 mutant) gelsolin was significantly reduced (p < 0.01) compared to that of clones expressing C-del gelsolin, the neocontrol clones and parental B16-BL6 cells. Growth rates of transfectants expressing wild-type, His321 mutant and C-del gelsolin on culture dishes in DMEM containing either 5% or 10% FCS were similar to the neocontrols and parental cells (data not shown). In contrast to growth properties in liquid culture conditions, colony-forming ability of transfectants expressing wild-type, His321 mutant and C-del gelsolin in soft agar was significantly (p < 0.05) suppressed (Fig. 3).
Since gelsolin is a representative of actin-regulatory proteins having at least 3 different activities (severing, capping and nucleating) and is critical for actin reorganization, we next carried out chemotaxis assay using wild-type, His321 mutant and C-del gelsolin transfectants, as well as the neocontrols and parental B16-BL6 cells. As shown in Figure 4, chemotactic migration of wild-type, His321 mutant but not C-del gelsolin transfectants toward fibronectin was significantly reduced (p < 0.01) compared to the neocontrols and parental B16-BL6 cells. The expression of integrin subunits (α4 and β1) were analyzed by FCM, since fibronectin is a ligand of integrin receptors. In addition, cell attachment to the extracellular matrix is mainly mediated by β1-integrin, and integrin α4 subunit was previously shown to suppress cell invasion and metastasis of mouse melanoma cells.41 Their expression on the cell surface was found to be similar among the transfectants and parental B16-BL6 cells (data not shown). These results indicate that ectopic expression of gelsolin in B16-BL6 mouse melanoma cells resulted in reduction of chemotactic migration without affecting integrin expression.
Tumorigenicity and metastatic ability of transfectants expressing full-length (wild-type and His321 mutant) and C-del gelsolins
To examine in vivo growth and metastatic abilities, 5 × 104 cells of wild-type, His321 mutant and C-del gelsolin transfectants as well as the neocontrols and parental B16-BL6 cells were injected in C57BL/6 syngeneic mice. Although reduced growth ability of transfectants expressing wild-type, His321 mutant and C-del gelsolin in soft agar was observed, these cells formed tumors in the footpads of mice. Nevertheless, the latency of both wild-type and His321 mutant gelsolin transfectants was slightly longer than the neocontrols and parental B16-BL6 cells (5–7 days for either wild-type or His321 mutant gelsolin transfectants vs. 3 days for the neocontrols and parental B16-BL6 cells as judged by the appearance of black nodules in mouse footpads). Despite these different rates of tumor formation, tumor growth was observed in all mice after 3 weeks following inoculation (Fig. 5a). The tumor-bearing feet with similar tumor sizes were amputated for spontaneous metastasis assay to exclude the possible growth-suppressive effect of gelsolin on metastasis. RT-PCR analysis with total RNAs extracted from amputated tumor-bearing footpads using the primer combination shown in Figure 1c revealed that the introduced gelsolin cDNAs were retained and expressed in the tumors (Fig. 5b).
Forty days (experiment 1) and 60 days (experiment 2) after the cells' inoculation, 3 mice out of 5 or 6 were sacrificed and their lungs were examined for metastatic colonies. As shown in Table I, the number of metastatic colonies in mice injected with either wild-type or His321 mutant gelsolin transfectants were significantly (p < 0.05) reduced than the neocontrols or B16-BL6 parent cells. In addition, incidences of tumor recurrence on the foot remained after amputation, reflecting that local invasion in mice with transfectants expressing full-length gelsolin (wild-type and His321 mutant) was lower than in the neocontrol clones and B16-BL6 parent cells. In experiment 1, no mouse inoculated with transfectants expressing full-length gelsolin died before the day 40 sacrifice, while 6 of 16 mice with the neocontrols or parental B16-BL6 cells died before the sacrifice (Table I). Survival days of mice inoculated with transfectants of full-length gelsolin were clearly longer than mice with the neocontrols and parental B16-BL6 cells. In contrast to full-length gelsolin, C-del gelsolin exhibited reduced metastasis-suppressive function in mice. Moreover, incidences of tumor recurrence, as well as survival days of mice inoculated with transfectants expressing C-del gelsolin were similar to the neocontrols and parental B16-BL6 cells (Table I), although colony-forming ability of transfectants expressing C-del gelsolin in soft agar was significantly suppressed (Fig. 3). These results indicate that suppression of metastasis by full-length gelsolins is not simply due to the reduced growth of the cell clones and that the carboxy-terminal 86 amino acids are critical for suppression of chemotaxis to fibronectin and metastasis in mice.
Cell transformation is often accompanied by a variety of phenotypic alterations including changes in cell shape and motility, loss of anchorage dependency, loss of cell-cell contact and loss of cell growth contact inhibition. Many of these changes are caused by deregulation in the microfilament cytoskeleton system. It has been reported that the expression of several actin-binding proteins including gelsolin often decreased in tumor cells and their reexpression by gene transfer causes suppression of tumorigenicity;42 therefore, they belong to class II tumor suppressor genes.43, 44 These observations indicate that appropriate cell architecture, such as maintenance of cell polarity and well-organized stress fibers, is crucial for normal cell regulation.
In our study, we sought to determine the effects of gelsolin overexpression on B16-BL6 murine melanoma cells with tumorigenic and highly metastatic abilities. In addition, the carboxyl-terminal truncated form (C-del) of gelsolin was used to determine whether the carboxyl-terminus that confers Ca2+ dependency on gelsolin functions could be important for the effects of gelsolin. We demonstrated here that overexpression of wild-type, His321 mutant and C-del gelsolin in B16-BL6 melanoma cells resulted in the reduction of colony formation in soft agar. However, full-length gelsolins (wild-type and His321 mutant) but not C-del gelsolin reduced chemotactic cell migration to fibronectin in B16-BL6 cells. Furthermore, retardation of tumor growth and suppression of lung metastasis in vivo were observed with clones expressing full-length (wild-type and His321 mutant) gelsolin. Reduced tumorigenic phenotypes of melanoma cells by full-length gelsolins (wild-type and His321 mutant) in vitro and in vivo are consistent with our previous results, demonstrating frequent downregulation of gelsolin expression in several types of human cancers and suppression of tumorigenicity by ectopic expression of wild-type gelsolin in cancer cells.45 The transfectants expressing full-length gelsolins, both wild-type and the His321 mutant, exhibited the reduced cell motility in vitro, reflecting less cell spreading compared to the neocontrols and parental B16-BL6 cells at the initial phase of cell attachment after replating (Fig. 2b). As cancer cell migration is essential for the establishment of metastasis, including detachment from primary site, invasion of surrounding connective tissue and invasion at metastatic sites, the suppression of cell motility could be responsible for the reduction of lung metastasis of transfectants expressing full-length gelsolins in vivo.
The effects on cell motility in vitro are opposite from previous studies using fibroblast systems, indicating that gelsolin functions as a positive factor for cell motility.46–48 Furthermore, the chemotactic cell migration of Ras-transformed NIH3T3 cells expressing the His321 mutant or human cytoplasmic gelsolin25 toward fibronectin was also increased (Fujita et al., unpublished observation). Recently, Shieh et al.49 reported that high expression of gelsolin was a great risk factor for cancer recurrence in human nonsmall cell lung cancer. We observed, however, less recurrence of gelsolin transfectants in mice as judged by reappearance of tumor after an amputation. Taken together, it is likely that the effect of gelsolin overexpression on cell migration would be dependent on the cell types and their origin. In this respect, it is interesting to note that the expression of integrins can result in altered behavior of tumor cells, i.e., reduction of tumorigenicity in transformed chinese hamster ovary cells, suppression of metastasis in B16a murine melanoma cells or enhancement of metastasis in a rhabdomyosarcoma cell line, indicating that the effect of integrins are indeed cell type-dependent,42 similar to results reported for gelsolin. The fact that no significant difference on metastasis suppression was seen between wild-type and His321 mutant gelsolin, although the mutant has decreased severing and increased nucleating activities with higher phosphoinositide binding capacity,26 suggests that known functions may not determine cell type dependency.
In addition, we demonstrated that truncation of the carboxyl-terminus, required for the regulation of gelsolin functions by the Ca2+ ion, resulted in the reduction of the suppressive effect of gelsolin on chemotaxis to fibronectin and metastasis. This truncate served as a good control for evaluating the effects of full-length gelsolins because transfectants expressing C-del gelsolin metastasized to lung in mice similar to the neocontrols and parental B16-BL6 cells, even though they exhibited reduced colony-forming ability in soft agar assay. Therefore, it is unlikely that suppression of metastasis by full-length gelsolins is due to the reduced growth of the cell clones. Furthermore, these results suggest that the C-terminus of gelsolin could have a critical role in the suppression of chemotactic cell migration and metastasis, although details of the molecular mechanism(s) for metastasis suppression by full-length gelsolin remain to be elucidated. The fact that the truncation of the C-terminus reduced the metastasis-suppressive function of gelsolin suggests another possibility that distinct factor(s) that would be able to interact with the C-terminus of gelsolin may be responsible for different effects of gelsolin overexpression in different cell types.
To form metastases, tumor cells must pass through several stressful and highly selective steps including invasion, microcirculation, arrest in the capillary bed and resumption of proliferation in the distant organs. During this strong biological selection in vivo, it is believed that many tumor cells may undergo apoptosis.50 Recently, we and another group demonstrated that gelsolin is associated with apoptotic cell death.51–53 The transfectants expressing gelsolin were more sensitive to staurosporine, a potential apoptotic cell-death inducer than the neocontrol transfectants and parental cells as judged by condensed nuclear morphology (Fujita et al., unpublished observations). Therefore, an additional possible mechanism by which pulmonary metastasis in vivo, could be an enhancement of apoptotic cell death of gelsolin transfectants and rapid elimination of tumor cells during circulation.
In summary, our results demonstrated that gelsolin could affect cell motility that contributes to the establishment of metastasis of B16-BL6 mouse melanoma cells and that the C-terminus of gelsolin is critical for suppression of metastasis. The finding that gelsolin overexpression can inhibit tumor cell migration and suppress pulmonary metastasis suggests that manipulation of the level of the gelsolin protein might help regulate tumor metastasis in the B16-BL6 and perhaps other cancer cell models.
Supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture, Japan (to N. K. and H.F.) and the Uehara Memorial Foundation, Japan. Hokkaido Foundation for the Promotion of Scientific and Industrial Technology, the Kanae Foundation and Kanehara Memorial Foundation (to H.F.). We thank Drs. K. Riabowol (University of Calgary) and Paul A. Janmey (University of Pennsylvania) for critical reading of the manuscript.