Mutually Antagonistic Actions of S100A4 and S100A1 on Normal and Metastatic Phenotypes

Mutually antagonistic actions of S100A4 and S100A1 on normal and metastatic phenotypes

Guozheng Wang1, Shu Zhang1, David G Fernig1, Marisa Martin-Fernandez2, Philip S Rudland1 and Roger Barraclough*,1

1School of Biological Sciences, University of Liverpool, Crown Street, Liverpool L69 7ZB, UK; 2M Martin-Fernandez, Daresbury Laboratory, Daresbury, Warrington WA4 4AD, UK

Increased levels of the homodimeric calcium-binding 1% S100A4 positive carcinoma cells survived over the protein, S100A4, have been shown to cause a metastatic same period (Rudland et al., 2000). In model systems, phenotype in at least three independent model systems of elevated levels of S100A4 confer a metastatic phenotype breast cancer and its presence in carcinoma cells has been upon benign rat mammary tumour cells (Davies et al., shown to be associated with a reduction in the survival of 1993; Lloyd et al., 1998) and in transgenic mice bearing patients suffering from a range of different cancers. neu-oncogene- or MMTV-induced primary mammary S100A4 has been shown to interact in vitro with another tumours (Ambartsumian et al., 1996; Davies et al., member of the S100 family of proteins, S100A1. The 1996). The poor outlook for cancer patients with purpose of the present study was to find out whether elevated expression of S100A4 is likely to be due to S100A1 could affect S100A4 function. Fluorescence the ability of S100A4 to induce a metastatic phenotype; resonance energy transfer was used to show the interac- however, the mechanism of the strong metastasis- tion of S100A4 and S100A1 in living cells and the binding inducing properties of S100A4 is not known. S100A4 affinities between S100A4 and S100A1 were determined has been reported to interact with a number of protein using a biosensor. S100A1 reduced the S100A4 inhibition targets, at least in vitro, namely nonmuscle tropomyosin of nonmuscle myosin A self-association and phosphoryla- (Takenaga et al., 1994), nonmuscle myosin A heavy tion in vitro. S100A1 reduced S100A4 induced motility chain (Kriajevska et al., 1994; Ford and Zain, 1995), p53 and growth in soft agar and metastasis in vivo. The results (Grigorian et al., 2001), liprin b1 (Kriajevska et al., show for the first time that interactions between different 2002) and methionine aminopeptidase 2 (Endo et al., S100 proteins can affect cancer-related activity, and that 2002). S100A4 also self-associates to form a homodimer the presence of S100A1 protein in carcinoma cells might as shown by its NMR structure (Vallely et al., 2002). modulate the effect of S100A4 on their metastatic However, Wang et al. (2000) and Tarabykina et al. abilities. (2000) have shown that S100A4 associates with another Oncogene (2005) 24, 1445–1454. doi:10.1038/sj.onc.1208291 S100 protein, S100A1, using the yeast two-hybrid Published online 20 December 2004 system. It is now shown that S100A1 can antagonize functions of S100A4 in vitro and in vivo. Keywords: S100A1; S100A4; metastasis; motility

Results

Preferential association of S100A4 with S100A1 Introduction Previous experiments have shown that S100A4 interacts The presence of the EF-hand-containing calcium-bind- with S100A1 in the yeast two-hybrid system (Wang et al., ing protein S100A4 in carcinoma cells has been 2000). Thus, S100A4 was tested in the yeast two-hybrid associated with a poor prognosis in patients with system for self-association as well as for interaction with carcinomas of the bladder, colorectum, gall bladder, S100A1, S100P or S100A2. b-Galactosidase activities oesophagus and lung (Kimura et al., 2000; Ninomiya from the yeast cells expressing S100A4/S100A1 fusion et al., 2001; Davies et al., 2002; Gongoll et al., 2002; proteins (Figure 1) were signiﬁcantly higher than those Nakamura et al., 2002). In an archival set of 349 breast expressing S100A4/S100A4, S100A4/S100P and cancer patients, less than 20% of patients with S100A4 S100A4/S100A2 (all Po0.01, Student’s t test) in the in their carcinoma cells were alive after 20 years, yeast two-hybrid system, indicating that stronger inter- whereas greater than 90% of patients with less than action of S100A1 with S100A4 occurred in yeast cells and that S100A4 might preferentially bind to S100A1 in *Correspondence: R Barraclough; E-mail: [email protected] other cells. Received 17 January 2004; revised 1 October 2004; accepted 15 October Binding parameters for the interactions between 2004; published online 20 December 2004 recombinant S100A4 (rS100A4) and recombinant Association of S100 proteins G Wang et al 1446 10 from the kinetic binding parameters and which demon- strate the internal consistency of the measurements.

) Interaction of S100A4 with S100A1 in mammalian cells 7.5 revealed by ﬂuorescence resonance energy transfer

er units (FRET)

Mill That S100A4 and S100A1 interact in mammalian cells 5.0 was shown by cotransfecting HeLa cells with expression plasmids encoding enhanced cyan fluorescent protein sidase ( (ECFP)-S100A4 fusion protein and enhanced yellow fluorescent protein (EYFP)-S100A1. A statistically 2.5 significant reduction (P ¼ 0.0001) in the fluorescence

-galacto β lifetime of the ECFP-S100A4 by ﬂuorescence lifetime imaging was detected in the presence of the EYFP- 0 S100A1 acceptor, but not in the presence of EYFP (Figure 2). This result shows that ECFP-S100A4 and EYFP-S100A1 exhibit close contact (o5 nm) in HeLa

cells.

1. A1/A1 1. 4. A4/P 4.

5. A4/A2 A4/- 6. 3. A4/A4 2. A4/A1 Figure 1 Association of S100A4 with S100 proteins in the yeast S100A4, but not S100A1 interacts with a recombinant two-hybrid system. Yeast cells were transfected with plasmids fragment of heavy chain of nonmuscle myosin II LexA-S100A1 þ YESTrp2-S100A1 for self-association of S100A1 (A1/A1), with LexA-S100A4 þ YESTrp2-S100A1 for interaction isoform A between S100A4 and S100A1 (A4/A1); with LexA-S100A4 þ YESTrp2-S100A4 for self-association of S100A4 (A4/A4); with Using the optical biosensor, rS100A4 was shown to LexA-S100A4 þ YESTrp2-S100P for interaction between S100A4 interact with a 149 amino-acid recombinant fragment of and S100P (A4/P); with LexA-S100A4 þ YESTrp2-S100A2 for nonmuscle myosin heavy chain A (rNMMHCIIA) interaction between S100A4 and S100A2 (A4/A2); with LexA- (Figure 3) as described previously (Chen et al., 2001). S100A4 þ YESTrp2 vector as a negative control to check for absence of interaction between S1000A4 and the product of empty In contrast, rS100A1 exhibited barely detectable inter- YESTrp2 vector (A4/-). The b galactosidase activity, an approx- action with rNMMHCIIA (Figure 3). imate indicator of the strength of interaction, was determined as described previously (Wang et al., 2000) S100A1 reverses the inhibitory effect of S100A4 on the sedimentation and phosphorylation of rNMMHCIIA In low-salt buffer, rNMMHCIIA exhibits self-associa- S100A1 (rS100A1) in vitro in nonfusion form were tion, which can be detected by a loss of rNMMHCIIA determined using an optical biosensor, in a buffer from the supernatant following centrifugation, as containing 0.5 mM calcium. The binding of rS100A4 to described in Materials and methods. rS100A4 can immobilized rS100A4 and to immobilized rS100A1 was partially reverse this self-association, whereas rS100A1 always homogenous; there was no evidence for the has no effect upon self-association of rNMMHCIIA presence of more than one binding site for rS100A4 in (Figure 4). Preincubation of rS100A4 with increasing rS100A4 or rS100A1. Plots of kon against ligand concentrations of rS100A1, at least up to an S100A1/ concentrations yielded straight lines (r ¼ 0.94–0.98) S100A4 ratio of 0.85, reversed the inhibitory effect of (data not shown). The association rate constant (kass) rS100A4 on rNMMHCIIA association (Figure 5). for the interaction between rS100A4 and immobilized S100A4, but not S100A1, inhibited the phosphoryla- rS100A1 in the presence of 0.5 mM Ca2 þ was tion of rNMMHCIIA in vitro by protein kinase C 0.2070.03 Â 105 M/s, slightly faster than that for the (Figure 6). When rS100A4 and rS100A1 were preincu- 7 5 self-association of rS100A4 (kass ¼ 0.12 0.01 Â 10 M/s). bated together, the inhibitory effect of rS100A4 on the The dissociation rate constant of rS100A4 from phosphorylation was partially relieved and the overall immobilized rS100A4 calculated over 10 values at phosphorylation of rNMMHCIIA was increased different concentrations of rS100A4 (Figure 6, lanes 1–5). (kdiss ¼ 0.00870.0005) was similar to that of rS100A4 k 7 from immobilized rS100A1 ( diss ¼ 0.006 0.001). The S100A1 reduces the motility and anchorage-independent afﬁnity calculated from the kinetic parameters for the growth of a high S100A4-containing rat mammary cell interaction between rS100A1 and rS100A4 line (Kd ¼ 300760 nM) was higher than that for the self- association of rS100A4 (Kd ¼ 670780 nM). The equiva- To investigate the possible effect of S100A1 on the lent Kd values calculated from the extents of binding activities of S100A4 in vivo, the level of S100A1 was observed at equilibrium were 5007140 and upregulated in S100A1-negative rat mammary cell lines, 10007300 nM for the interaction between rS100A4 the low-S100A4 Rama 37 and the high-S100A4 KP1- and rS100A1 and for the self-association of rS100A4, Rama 37, by transfection of an S100A1-encoding respectively, values which are similar to those calculated expression vector, as described in Materials and

Oncogene Association of S100 proteins G Wang et al 1447

Figure 2 Fluorescence lifetime imaging of the interaction between S100A4 and S100A1. (a) Fluorescence lifetime imaging map of ECFP-S100A4 fusion protein expressed in HeLa cells that simultaneously coexpressed S100A1-EYFP fusion protein. (b) Distribution of the first-order average fluorescence lifetimes for (a). (c) Variation of the fluorescence lifetime distribution found in different cells in two independent transfections. (d–f)Asin(a–c), but in the absence of acceptor, S100A1-EYFP. (g) Time-resolved fluorescence lifetime decay of the pixels marked by the arrows in (a) and (d). (h) Average temporal distribution of the data in (c) and (f) after normalization

methods. Clones were picked from the S100A1-trans- was detected by Western blotting. Both the parental cell fected KP1-Rama 37 cells (designated KP1-S100A1-1 lines, Rama 37 and KP1-Rama 37, and the cells and KP1-S100A1-2) that expressed a level of S100A1 transfected with empty vector (designated Rama 37 V mRNA which was similar to that of S100A4 mRNA in and KP 1 V, respectively), showed undetectable levels of the same cells. Clones were also isolated from trans- S100A1 (Figure 7). fected Rama 37 cells (designated Rama 37-S100A1-1 Immunoﬂuorescent staining of the KP1-S100A1-1 cell and Rama 37-S100A1-2) that expressed similar levels of line showed that all the cells examined expressed both S100A1 to that of KP1-S100A1/2 cells, but no S100A4 S100A1 and S100A4 and about 90% of the cells

Oncogene Association of S100 proteins G Wang et al 1448 150 a S100A1 − + − S100A4 + − −

NMMHCIIA

100

123

b 1200

1000 50

Extent of binding (arc seconds) 800

600 0 12 400 Figure 3 Interaction of S100A1 and S100A4 with immobilized recombinant nonmuscle myosin heavy chain isoform IIA (NMMHCIIA). Recombinant NMMHCIIA was immobilized on 200 an aminosilane surface of a biosensor (Materials and methods).

Either rS100A1 (column 1) or rS100A4 (column 2) was added to a NMMHCIIA in supernatant Arbitrary Units final concentration of 0.5 mm in the presence of 0.5 mM CaCl2. The 0 extent of binding of each binding reaction was calculated using Fastfit software (Materials and methods) 123 Figure 4 Effect of S100 proteins on rNMMHCIIA association. rNMMHCIIA (5 mM) was incubated either in the presence of 5 mM rS100A4 (lane 1 and column 1), or 5 mM rS100A1 (lane 2 and column 2) or in the absence of rS100A4 or rS100A1 (lane 3 and examined expressed both S100A1 and S100A4 at similar column 3) in Bundling Buffer (see Materials and methods) levels (data not shown). About 5% of cells expressed overnight at 41C. The mixture was centrifuged at 13 600 g for 30 min. The resultant supernatants were analysed by SDS–PAGE. more S100A1 than S100A4 and about 5% expressed The rNMMHCIIA retained in the supernatant was detected by more S100A4 than S100A1 (data not shown). At higher Western blotting with an anti-NMMHCIIA serum. The upper magnification, the immunofluorescence showed that the panel is an example of a Western blot and the lower panel S100A1 and S100A4 colocalized, mainly in the peri- summarizes the quantitative results of four individual experiments. 7 nuclear region of the cells and smaller amounts of the The mean s.d. is shown proteins colocalized with one another in the cytoplasm (not shown). Using Boyden chamber assays, the S100A4-expressing KP1-Rama 37 cell line exhibited a higher rate of motility than its parental cell line, Rama 37 (P ¼ 0.0001, Student’s t test), whereas the two Rama 37-S100A1 S100A4 ++++ + + − transfectant cell lines, Rama 37-S100A1-1 and Rama 37- S100A1 − ++ + + + + S100A1-2, showed similar rates of motility to that of the parental Rama 37 cell line (P ¼ 0.82 and 0.6, respec- NMMHCIIA tively, Fisher Exact test, Table 1) and vector control cell line, Rama 37V (P ¼ 0.3 and 0.2, respectively, Fisher 1234 5 6 7 Exact test, Table 1). This suggests that a high level of S100A4, but not S100A1, promotes cell motility. Both Figure 5 Effects of rS100A1 on the rS100A4-induced inhibition of KP1-S100A1 cell lines, which express S100A1 and rNMMHCIIA association. rS100A4 (5 mM) was preincubated at 41C overnight in a buffer of 10 mM Tris-HCl, pH 7.0, 100 mM S100A4, showed a significantly lower rate of motility NaCl, 0.5 mM CaCl2 with 0 nM (lane 1), 43 nM (lane 2), 128 nM than their parental cell line, KP1-Rama 37 (P ¼ 0.0001) (lane 3), 428 nM (lane 4), 1.28 mM (lane 5), 4.28 mM (lane 6) and vector control cell line, KP1V (P ¼ 0.0001), both of rS100A1. The mixtures were incubated with 5 mM rNMMHCIIA in which expressed a high level of S100A4, but no S100A1 Bundling Buffer overnight at 41C and then centrifuged at 13 600 g for 30 min. The resultant supernatants were analysed by SDS– (Table 1). These results indicate that upregulated PAGE. The rNMMHCIIA retained in the supernatant was S100A1 reduces the motility associated with upregulated detected by Western blotting with an anti-NMMHCIIA serum. S100A4. The arrow points to the band of rNMMHCIIA

Oncogene Association of S100 proteins G Wang et al 1449 NMMHCIIA + + + + + + + + S100A4 + + + + + + − −

S100A1 + + + + + − − +

12345678 KP1-S100A1-1 KP1-Rama 37 KP1-vector Rama 37-S100A1-1 Rama 37 Rama 37-vector

1000 a S100A1

800 b S100A4

600 c Actin 400

(Arbitrary units) d 200 S100A1

NMMHCIIA phosphorylation 0 e S100A4 12345678 Figure 7 Upregulation of S100A1 and S100A4 in rat mammary Figure 6 Effects of rS100A1 and rS100A4 on the phosphorylation tumour cell lines. In all, 10 mg of total RNA extracted from of rNMMHCIIA. Prior to phosphorylation reactions, 5 mM parental and transfected cell lines were analysed by Northern rS100A4 was preincubated with different amounts of S100A1, blotting (a–c) with 32P dCTP-labelled probes. Filters were 0.1 mM (lane 1), 0.3 mM (lane 2), 1 mM (lane 3), 3 mM (lane 4), 10 mM incubated with probes for S100A1 (a), S100A4 (b), b-actin (c). (lane 5) or with 0 mM (lane 6) in the presence of 0.5 mM CaCl2 For Western blotting (d–e), 40 mg of total protein from each cell overnight at 41C or with no S100 protein (lane 7). rS100A1 (1 mM) lysate was analysed by SDS 15% PAGE. A mouse anti-human was also preincubated (lane 8). The preincubated mixtures were S100A1 monoclonal antibody (DAKO) and rabbit anti-human then incubated with 5 mM rNMMHCIIA for 40 min at room S100A4 polyclonal antibody (DAKO) were used to detect S100A1 temperature and the phosphorylation reaction was carried out as (d) and S100A4 (e) proteins, respectively. Cell lines were KP1- described in Materials and methods. The phosphorylated Rama 37, a metastatic S100A4-expressing derivative of Rama 37; rNMMHCIIA was separated by SDS–PAGE and detected by KP1-S100A1-1, KP1 cells transfected with an expression vector autoradiography (upper panel). The arrow indicates the phos- containing S100A1 cDNA; KP1 vector, KP1 cells transfected with 7 phorylated rNMMHCIIA band. A bar chart s.d. of an average of empty expression vector; Rama 37, a benign nonmetastatic cell line three individual experiments is shown (lower panel) which does not express S100A1 or S100A4; Rama 37-S100A1-1, Rama 37 cells transfected with an expression vector containing S100A1 cDNA; Rama 37 vector, Rama 37 cells transfected with empty expression vector The colony-forming abilities in soft agar of the transfected cells and their parental cells were examined. The Rama 37 and Rama 37-S100A1 cells grew poorly in a soft agar assay (not shown). However, KP1-Rama 37 mammary tumours with incidences ranging from 49 to cells showed a much increased capability for colony 71% (Table 1), values that were not significantly growth in soft agar than that of its parental cells, Rama different from one another (P ¼ 0.05–1.0, Fisher Exact 37. This result suggests that upregulation of S100A4 test). Immunocytochemical staining confirmed that the promotes the anchorage-independent growth of Rama patterns of expression of S100A1 and S100A4 in 37 cells. The two clones of KP1-S100A1 cells, expressing primary tumours were the same as those of the cell both S100A1 and S100A4, showed a lower ability to lines injected (data not shown). Animals injected with grow in soft agar than the KP1-Rama 37 or KP1V cells, Rama 37V or Rama 37-S100A1-1 cells produced no which expressed a high level of S100A4 alone (all metastases, while Rama 37-S100A1-2 cells produced P ¼ 0.001, Student’s t test), but had a higher ability of metastases in only 8% of rats injected. However, the rats anchorage-independent growth (Table 1) than that of injected with KP1-Rama 37 cells or KP1 cells trans- Rama 37 cells or Rama 37-S100A1 cells (P ¼ 0.012 and fected with empty expression vector, KP1V, produced 0.017, respectively). This suggests that upregulation of metastases in the lungs in 80% of the rats with tumours S100A1 reduced the anchorage-independent growth of (Table 1), highly significantly higher than Rama 37V, the S100A4-containing cells. Rama 37-S100A1-1 and Rama 37-S100A1-2 (each P ¼ 0.0001, Fisher Exact test). The two clones of S100A1 reduces the incidence of metastases induced by S100A4- and S100A1-containing cells, KP1-S100A1-1 S100A4 in a rat mammary tumour model and KP1-S100A1-2, produced metastases in the lungs in 21% and 18% of the rats with tumours, respectively, Subcutaneous injections of cell lines, Rama 37V, Rama both highly significantly lower than the incidence with 37-S100A1-1 and Rama 37-S100A1-2, KP1-Rama 37 the S100A4-positive, S100A1-negative, KP1-Rama 37 and KP1-S100A1-1 and KP1-S100A1-2 into the mam- cells (both Po0.0001, Fisher Exact test) or KP1V cells mary fat pads of female syngeneic rats yielded primary (both 0.0001, Fisher Exact test). These results show that

Oncogene Association of S100 proteins G Wang et al 1450 abthat S100A4 interacts with another S100 protein, S100A1. In the present experiments, the binding affinity of this interaction in vitro, determined using a biosensor M T assay, shows Kd values in the range for natural M biological interactions (nM) and an interaction between M S100A4 and S100A1 has been demonstrated in living T mammalian cells using FRET techniques. More im- portantly, a completely novel finding is now made that the presence of S100A1 antagonizes a major biological function of S100A4, that of the induction of metastasis c d in vivo and associated properties in cultured cells. S100A4 interacts with, and inhibits the self-association of rNMMHCIIA (Kriajevska et al., 1994; Ford and Zain, 1995), and it has been suggested that S100A4 affects the phosphorylation of a NMMHCIIA fragment T by protein kinase C and casein kinase 2 in vitro (Kriajevska et al., 1998, 2000). When tested in vitro, T S100A1 significantly and specifically reversed the inhibitory effects of S100A4 on both the self-assembly of a recombinant fragment of NMMHCIIA and its Figure 8 Interaction of tumour with muscle in vivo.(a) A primary phosphorylation by protein kinase C. S100A1 itself tumour (T) of vector-transfected Rama 37 cells (Rama 37V) shows no local invasion of muscle cells (M). (b) Primary tumour of Rama showed no effect on the self-assembly and phosphoryla- 37 cells transfected with S100A4 (KP1-Rama 37) showing tumour tion by PKC of the NMMHCIIA in vitro, suggesting cells (T) invading muscle cells (M). (c) Primary tumour of Rama 37 strongly that the effect of S100A1 is not a direct one, but cells transfected with S100A1 (Rama 37-S100A1) shows the acts by affecting S100A4 function. tumour (T) associated with proliferating muscle cells (arrows). (d) S100A1 also inhibited biological effects of S100A4 in Primary tumour of KP1 cells transfected with S100A1 (KP1- S100A1) showing small myoblast-like cells (arrows) close to the vivo. Coexpression of S100A1 in cells expressing high tumour cells (T). All panels show staining with antimyoglobin levels of S100A4 significantly reduced their rate of antibody as described in Materials and methods. Bars ¼ 200 mM motility and invasion. High expression of S100A4 conferred on the Rama 37 cells the transformation- associated property of anchorage-independent growth, which is thought to be important for tumour cell the presence of S100A1 reduces the metastasis-forming survival in distant organs (Nakanishi et al., 2002). ability of S100A4. Coexpression of S100A1 in the S100A4-expressing Some of the primary tumours exhibited invasion of Rama 37 cells significantly reduced their rate of colony peripheral blocks of striated muscle (Figure 8; Table 1). formation in soft agar. The combined incidence of muscle invasion by tumours The above results not only show that S100A4 may from the two S100A1- and S100A4-expressing cell lines, exert its effect on several major steps in metastasis, KP1-S100A1-1 and KP1-S100A1-2, was significantly thereby greatly enhancing the metastatic abilities of less than muscle invasion by the S100A4-positive, tumour cells, but also that S100A1 has an antagonistic S100A1-negative, KP1-Rama 37 cells (P ¼ 0.03, Fisher effect on these S100A4 abilities associated with metas- Exact test) or KP1V cells (P ¼ 0.007, Fisher Exact test) tasis, contributing to the overall significant reduction by showing that the presence of S100A1 reduces the S100A1 of the incidence of S100A4-induced metastasis invasive potential of the S100A4-containing cells. in the syngeneic rat model. It is important to note that Normal host muscle cell proliferation, including the S100A1 by itself did not reduce cell motility, invasion or proliferation of smaller myoblast-like cells which were anchorage-independent growth in vivo but only reduces identified by staining for myoglobin (Rudland et al., the effects exerted by S100A4. The absence of a direct 1984), was found around the Rama 37-S100A1- and inhibitory effect of S100A1 on metastasis itself was not KP1-S100A1-derived tumours, but not around Rama shown, however, such an effect is considered unlikely in 37V- and KP1-Rama 37-derived tumours (Figure 8). view of the absence of a direct inhibitory effect of The incidence of muscle proliferation in the tumours S100A1 on those in vitro processes associated with derived from the two clones of KP1-S100A1 cells (40 metastasis in vivo. and 45%) was significantly higher than in tumours Unexpectedly in these experiments, marked muscle derived from KP1-Rama 37 cells (0%) (P ¼ 0.02 and cell proliferation was observed around the primary 0.009, respectively, Fisher Exact test). tumours derived from the S100A1-expressing cells (Rama 37-S100A1 and KP1-S100A1), but was not observed in primary tumours derived from control cell Discussion lines, nor in the S100A4-expressing KP1-Rama 37 cells. Although, S100A1 has been found in replicating It has been shown previously (Tarabykina et al., 2000; myoblasts (Sorci et al., 1999), its presence in muscle Wang et al., 2000) using the yeast two-hybrid system cells is associated with the regulation of enhanced

Oncogene Association of S100 proteins G Wang et al 1451 Table 1 Cell motility, growth in soft agar, metastasis, muscle invasion and muscle proliferation by cells expressing S100 proteins Cell lines A1a A4b Motility (%) Colonies/ﬁeld Tumour incidence Incidence of me- Invasion of mus- Incidence of mus- 7s.d.c 7s.d.d (%)e tastases (%)f cle (%)g cle proliferation (%)h

Rama 37 ÀÀ 15.871.00NDNDNDND Rama 37V ÀÀ 16.071.7 0 25/35 (71) 0/25 (0) 2/10 (20) 0/10 (0) Rama 37- + À 15.671.0 0 22/32 (69) 0/22 (0) 3/13 (23) 11/13 (85) S100A1-1 Rama 37- + À 16.371.2 0 25/36 (69) 2/25 (8) 7/23 (30) 17/23 (74) S100A1-2 KP1-Rama 37 À + 19.570.7 9.273.0 30/61 (49) 24/30 (80) 12/14 (86) 0/14 (0) KP1V À + 18.672.1 8.772.9 25/38 (66) 20/25 (80) 17/19 (90) 0/19 (0) KP1-S100A1-1 + + 14.471.6 2.171.5 24/49 (49) 5/24 (21) 4/10 (40) 4/10 (40) KP1-S100A1–2 + + 13.871.7 2.071.7 17/34 (50) 3/17 (18) 6/11 (55) 5/11 (45) aS100A1 and bS100A4 mRNA and protein detected by Northern and Western blotting; À, undetectable; +, positive. cMotility assay using the Boyden Chamber. Each cell line was subjected to at least three independent experiments. Motility is the mean percentage of cells that moved through the filter in a period of 20 h against the total number of cells seeded in a 24-well plate and cultured in the same medium under the same conditions as that used in the upper chamber (Materials and methods). The mean %7standard deviation (s.d.) are shown. KP1-Rama 37 and KP1V possess similar rates of motility (P ¼ 0.2, Student’s t test), but significantly higher than that of Rama 37 or Rama 37V (P ¼ 0.0001 and 0.0002, respectively, Student’s t test). KP1-S100A1-1, KP1-S100A1-2, Rama 37-S100A1-1 and Rama 37-S100A1-2 have similar motility rates to those of Rama 37 (P ¼ 0.27, 0.15, 0.82, 0.6, respectively, Student’s t test) or Rama 37V (P ¼ 0.07, 0.09, 0.3 and 0.2, respectively, Student’s t test), but significantly lower than those of KP1-Rama 37 or KP1V (both P ¼ 0.0001). dThe colony-forming ability in soft agar. Cells (2 Â 104) were seeded in a 35 mm culture dish. Each cell line was subjected to at least three independent experiments. After 4 weeks of routine culture, colonies with diameter larger than 0.06 mm were counted under the microscope. Rama 37, Rama 37V and Rama 37-S100A1 cells did not form any colonies in soft agar. The two cell lines of KP1-S100A1-1 and -2 formed similar numbers of colonies/field (P ¼ 0.35, Student’s t test) but significantly fewer colonies/field than that of KP1-Rama 37 or KP1V cells (both P ¼ 0.001, Student’s t test). eNumber of tumours/number of animals inoculated. Tumour incidence of the transfected cells was not significantly different from the Rama 37V cells (P ¼ 0.05–1.0, Fisher Exact test). fNumber of animals with lung metastases/number of animals with tumours. There was no significant difference in the incidence of metastasis between the two separate clones of Rama 37-S100A1 (P ¼ 0.39, Fisher Exact test) or between the two clones of KP1-S100A1 (P ¼ 1.0) or between KP1-Rama 37 and KP1V (P ¼ 1.0). However, KP1-S100A1-1 and KP1-S100A1-2 have a significantly lower incidence of metastases than KP1-Rama 37 (both Po0.0001, Fisher Exact test) or KP1V (Po0.0001 and P ¼ 0.0001, respectively, Fisher Exact test). gNumbers of animals with muscle invasion/numbers of animals with tumours surrounded by muscle. No difference was found in the incidence of muscle invasion between tumours from the two Rama 37-S100A1 clones (P ¼ 0.72, Fisher Exact test) or between tumours from the two KP1-S100A1 clones (P ¼ 0.67, Fisher Exact test) or between tumours from KP1-Rama 37 and KP1V (P ¼ 1.0, Fisher Exact test). Tumours from Rama 37-S100A1-1 and Rama 37-S100A1-2 exhibited no significant differences in invasion of muscle from Rama 37V (P ¼ 1.0 and 0.68, respectively, Fisher Exact test). KP1-S100A1-1 and KP1-S100A1-2 tumours combined were significantly different from KP1-Rama 37 (P ¼ 0.03, Fisher Exact test), and KP1V (P ¼ 0.007, Fisher Exact test). hNumber of animals with muscle cell proliferation/number of animals with tumours surrounded by muscle (samples without any muscle tissue evident on the sections were excluded). In those tumours that exhibited muscle associated with the tumour, there was significantly more muscle cell proliferation in Rama 37-S100A1-1 or Rama 37-S100A1-2 than in Rama 37V (both Po0.0001, Fisher Exact test). There was significantly more muscle cell proliferation in tumours derived from the two KP1-S100A1 clones than in tumours derived from KP1-Rama 37 (P ¼ 0.02 and 0.009, respectively, Fisher Exact test) or KP1V (P ¼ 0.009 and 0.003, respectively, Fisher Exact test). There were no significant differences between the two Rama 37- S100A1 clones (P ¼ 0.68) or the two KP1-S100A1 clones (P ¼ 1.0). ND, not determined

contractile performance of the muscle (Most et al., 2001, cells in vivo. Thus, S100A1 and S100A4 are mutually 2003). The present results suggest that S100A1, but not antagonistic towards each others’ activities in vivo. S100A4, produced by the tumour cells could stimulate How might S100A1 and S100A4 exert their mutually muscle proliferation and there are several possible antagonistic effects on one another? The inhibitory explanations for this effect. However, the previously effects of S100A1 on S100A4 activity are unlikely to be reported extracellular activity of some S100 proteins, mediated by the interaction of S100A1 with an S100A4 notably the neuronal S100B, which is neurotrophic for target, such as NMMHCIIA, since it has been shown cultured glial cells and astrocytes (Selinfreund et al., using the optical biosensor that S100A1 does not 1991), raises the possibility that S100A1 might exert its interact directly with the S100A4-binding, effect extracellularly on host muscle cells in vivo. rNMMHCIIA fragment (Figure 4). Previous studies Recently, extracellular S100A1 has been reported to (Tarabykina et al., 2000; Wang et al., 2000) and this inhibit apoptosis in ventricular cardiomyocytes (Most paper show that S100A1 interacts with S100A4 in vitro et al., 2003). In the present system in vivo, S100A1, but and in vivo. It is not clear whether the interaction not S100A4, stimulates host muscle cell proliferation in between S100A1 and S100A4 is due to the formation of the vicinity of the primary tumours. Interestingly, the heterodimers. On the one hand, determination of the incidence of host muscle proliferation around KP1- three-dimensional structure of S100A4 (Vallely et al., S100A1-derived tumours was signiﬁcantly reduced 2002) and S100A1 (Rustandi et al., 2002) has shown that relative to that around Rama 37-S100A1-derived both of these proteins form homodimers. On the other tumours (P ¼ 0.01, Fisher Exact test, mean of duplicate hand, it is well known that S100A1 forms heterodimers clones of each), suggesting that the presence of S100A4 with some other S100 proteins, for example with S100B might antagonize the effect of S100A1 on host muscle (Donato, 2001). Tarabykina et al. (2000) have detected

Oncogene Association of S100 proteins G Wang et al 1452 S100A4/S100A1 heterodimers, as well as the homo- Recombinant S100A1 and S100A4 proteins were purified dimers, using electrospray ionization Fourier transform from E. coli cells as described for S100A4 (Gibbs et al., 1994). ion cyclotron resonance mass spectrometry and subse- For use as ligates in biosensor assays, the recombinant S100A1 quently have shown that S100A4 exists in solution as an and S100A4 monomeric proteins were further purified using equilibrium between monomers and homodimers (Tar- Superdex 75 gel filtration columns. The fractions containing the lower molecular weight proteins were pooled and abykina et al., 2001). Furthermore, previous studies immediately stored at À801C. (Wang et al., 2000) have shown that point mutations in the dimer interfaces of either S100A4 or S100A1 Optical biosensor assay interrupt both homodimerization and the heterodimeric interaction between S100A1 and S100A4, as determined The purified rS100A4, rS100A1 and rNMMHCIIA proteins by the yeast two-hybrid system. In the biosensor were immobilized on aminosilane surfaces (Thermo Electron, experiments, it is possible that the observed antagonistic Basingstoke, UK) and binding assays were carried out as et al et al effect of S100A1 protein on S100A4 is due to interac- described previously (Rahmoune ., 1998a, b; Chen ., 2001; Wang et al., 2004). Binding parameters, kon and koff,were tions between homodimeric forms, as no special calculated using FastFit software (Affinity Sensors) as pre- measures were taken to dissociate the recombinant viously described (Rahmoune et al., 1998a, b; Chen et al., 2001). S100 proteins into their individual subunits, once the recombinant proteins had been mixed. However, the FRET present results report the novel finding that S100 The S100A1 coding sequence was amplified by PCR using a 50 proteins can modulate the activity of others, and this 0 could be a general regulatory mechanism among this primer: 5 -CCGCAAGCTTCGATGGGCTCTGAGCTG, containing a HindIII site (underlined) and a 30 primer: 50- family of proteins. The finding that such a regulatory CGCGGATCCTCAACTGTTCTCCCAGAAG, containing a mechanism can affect the function of an S100 protein is BamHI site using YESTrp2-S100A1 DNA as a template. The likely to be of particular importance to those S100 S100A4 coding sequence was amplified by PCR using 50 proteins, such as S100A4 (Gongoll et al., 2002) that primer: 50-CTGTGAAGCTTCGATGGCGAGACCCTTG, seem to be involved in human cancer. containing a HindIII site, and 30 primer: 50-CGGGATCCT- CACTTCTTCCGGGGCTC, containing a BamHI site using pJLA602-S100A4 DNA (Gibbs et al., 1994) as a template. The Materials and methods expression vectors, pEYFP-S100A1 and pECFP-S100A4, were constructed by inserting the restriction-enzyme-cut, PCR DNA constructs and yeast two-hybrid assay products of S100A1 and S100A4 coding sequences into pEYFP-C1 and pECFP-C1 vectors (Clontech, Palo Alto, LexA-S100A4 (bait construct) and YESTrp2-S100A4, USA) between the HindIII and BamHI sites and checked by YESTrp2-S100A1 and YESTrp2-S100P (prey constructs) were automated DNA sequencing. HeLa cells were transiently described previously (Wang et al., 2000). The YESTrp2-S100A2 transfected with the expression plasmids for EYFP-S100A1 (bait) construct was constructed as follows. The coding and ECFP-S100A4 fusion proteins. Fluorescence lifetime sequence of S100A2 cDNA was amplified by RT–PCR using microscopy (FLIM) images in the time domain from single a50 primer containing a HindIII restriction enzyme site: cells were collected using time-correlated single photon GCTAAGCTTATGTGCAGTTCTCTGGAG, and a 30 primer counting (TCSPC) (Pepperkok et al., 1999) as described containing a BamHI restriction enzyme site: GCAGGATCCT previously (Zhang et al., 2004). CAGGGTCGGTCTGGGCAG. The resultant PCR product was cloned into the HindIII and BamHI sites of YESTrp2 Phosphorylation of recombinant fragment of nonmuscle myosin vector. For the yeast two-hybrid system, each pair of bait and heavy chain by protein kinase C prey vectors was delivered into L40 yeast cells by cotransforma- tion and the b-galactosidase activities of the transformants were Purified protein kinase C (PKC) (having a-, b-, g-PKC determined as described before (Becker and Fikes, 1993; Wang activities) was purchased from Upstate Biotechnology (Char- et al., 2000). Results for various pairs of bait and prey were lottesville, USA) and phosphorylation reactions were carried compared by the Student’s t test using SPSS software. out according to the manufacturer’s instructions. In total, 15 ml samples of the reaction mixture were subjected to SDS– Preparation of recombinant proteins polyacrylamide gel electrophoresis (SDS–PAGE). The gel was dried and radioactive phosphorylation signals were recorded Recombinant S100A4 (rS100A4) and the C-terminal recombi- on X-ray film overnight. nant fragment of nonmuscle myosin II isoform A heavy chain To test the effect of rS100A1 on the interaction between (rNMMHCIIA) were prepared as described previously (Gibbs rS100A4 and rNMMHCIIA phosphorylation, 5 mM rS100A4 et al., 1994; Chen et al., 2001). Recombinant S100A1 protein was preincubated overnight with 0–10 mM rS100A1 in buffer (S100A1) was expressed in Escherichia coli BL-21 DE3 containing 0.5 mM CaCl2 at 41C. The mixture was then (Novagen, MD, USA), using a pET11-S100A1 expression incubated with rNMMHCIIA at room temperature for plasmid. The plasmid was constructed as follows. The coding 40 min before carrying out phosphorylation reactions and gel sequence of human S100A1 mRNA was amplified by PCR separation as above. using a 50 primer: CTGCGTTTGCATATGGGCTCT- GAGCTGG containing an NdeI site and a 30 primer: Differential sedimentation of rNMMHCIIA in vitro CGAGGATCCTCAACTGTTCTCGGAGAAG containing a BamH1 site. The resulting PCR product was cut with the A semiquantitative centrifugation method to detect the self- restriction enzymes and subcloned into pET11a vector assembly of rNMMHCIIA was adapted from Murakami et al. (Novagen, MD, USA). The expression construct was verified (1995). rNMMHCIIA (5 mM) was incubated at 41Covernight by DNA sequencing. in imidazole-HCl (pH 7.5), 100 mM NaCl, 2.5 mM MgCl2

Oncogene Association of S100 proteins G Wang et al 1453 (Bundling Buffer) in a total volume of 100 ml. After centrifuga- following modifications. The number of cells seeded in the tion at 13 600 g for 30 min, 20 ml of the supernatant was analysed chamber was 1 Â 105, instead of 2 Â 105 and their viabilities by SDS–PAGE and the unassembled rNMMHCIIA molecules were monitored by seeding the same number of cells in a 24- remaining in the supernatant were detected by Western blotting well plate. The incubation time was 20 h instead of 24 h. The using a polyclonal rabbit antiserum to the rNMMHCIIA C- total viability is expressed as a percentage ¼ (total number terminal fragment. To test the effect of S100 proteins on S100A4 cells migrated to the lower side of the filter in 20 h/total in the rNMMHCIIA sedimentation assay, 5 mM rS100A4 was number of viable cells) Â 100%. Statistical analysis was by preincubated overnight at 41Cwith0–5mM rS100A1 in the Student’s t test. presence of 0.5 mM CaCl2. The mixture was incubated with rNMMHCIIA for 8 h at 41C. Bundling Buffer was added and Soft agar assay the mixture incubated overnight at 41C, as described above. Each well (35 mm) of a six-well culture dish was coated with Cell culture and transfection 2 ml bottom agar mixture (DMEM, 10% (v/v) FCS, 0.6% (w/ v) agar). After the bottom layer had solidified, 2 ml top agar The rat mammary (Rama) 37 nonmetastatic benign tumour- mixture (DMEM, 10% (v/v) FCS, 0.3% (w/v) agar) was added derived cell line (Dunnington et al., 1983) expresses undetect- containing the cells under test and the dishes incubated at 371C able levels of S100A1 mRNA and protein and barely in 10% (v/v) CO2 for 4 weeks. Twice a week, two drops of detectable levels of S100A4 mRNA and protein by Northern routine culture medium were added to each dish. The resultant and Western blotting, respectively. The KP1-Rama 37 cell line colonies, >0.06 mm in diameter, were counted using a had been derived from the Rama 37 cell line by transfection of microscope. A cell mass o0.06 mm was dying and too small cloned S100A4 genomic DNA (Davies et al., 1993) and to be counted as a colony that survived in the semisolid expresses a high level of S100A4 and undetectable level of medium. A total of 10 microscopic fields were randomly S100A1 proteins. Rama cells were cultured as described before selected and the average number of clones in each field was (Dunnington et al., 1983). For generating stably transfected calculated. Statistical analysis was by Student’s t test. cell lines, the Rama 37 (Dunnington et al., 1983) and KP1- Rama 37 (Davies et al., 1993) cell lines were transfected with pCDNA-S100A1 or empty pCDNA vectors using lipofecta- Tumorigenicity and metastasis mine (Invitrogen, Paisley, Scotland), and the transfected cells Metastasis assays using a rat model for breast cancer and were selected with 750 mg/ml Zeocin. Surviving clones were histology of tissues were carried out as described previously isolated and transferred to a multiwell dish using a ring cloning (Dunnington et al., 1983; Davies et al., 1993). Animals technique (Dunnington et al., 1983) and grown under selective containing microscopically visible metastases of malignant conditions for 7 days. The resulting transfected cell lines were cells in the lungs and blocks of striated muscle infiltrated by designated as follows: Rama 37 cells transfected with empty malignant cells at the periphery of the primary tumours were expression vector, R37V; Rama 37 cells transfected with scored positive for metastasis and invasion, respectively. S100A1 expression vector, Rama 37-S100A1-1 and Rama 37- Animals that contained no visible muscle blocks bordering S100A1-2; KP1-Rama 37 cells transfected with empty expres- the primary tumours were eliminated from the analysis of sion vector, KP1V; KP1-Rama 37 cells transfected with muscle invasion and muscle proliferation. For analysis of S100A1 expression vector, KP1-S100A1-1 and KP1-S100A1-2. muscle proliferation, the sections containing visible muscle Northern hybridization for rat b-actin, S100A1 or S100A4 blocks bordering the primary tumours were immunohisto- mRNAs was carried out as described previously (Sambrook chemically stained with antimyoglobin antibody (Sigma, et al., 1989). For Western blotting, 50 mg of total protein, lysed Dorset, UK) as described previously (Rudland et al., 1984). using Laemmli buffer (Laemmli, 1970) were separated by SDS Sections containing numbers of small myoblast-like cells 15% PAGE and then electrically blotted onto a PVDF expressing myoglobin were scored positive. Statistical analysis membrane (Millipore, Billerica, USA). The S100A1 and was by Fisher Exact test. S100A4 proteins were probed using a mouse anti-S100A1 Animals were maintained according to UKCCCR guidelines monoclonal antibody (DAKO, Glosrup, Denmark) and a under UK Home Office Project Licence No. 40/2395 to rabbit anti-S100A4 polyclonal antibody (DAKO), respectively. Professor PS Rudland.

Migration assay Acknowledgements Migration assays for Rama 37, KP1-Rama 37 and cells We thank Angela Platt-Higgins for histology and Joe Carroll transfected with pCDNA-S100A1 were carried out using for excellent technical assistance. This work was supported by Boyden Chamber units with 6.5 mm diameter polycarbonate the North West Cancer Research Fund, The Cancer and Polio membrane inserts with 8 mm pores (Corning Costar, Acton, Research Fund and the Biotechnology and Biological Sciences USA), as described previously (Jenkinson et al., 2004) with the Research Council of the UK.

References

Ambartsumian N, Grigorian M, Larsen F, Karlstrom O, Davies BR, Davies MPA, Gibbs FEM, Barraclough R and Sidenius N, Rygaard J, Georgiev G and Lukanidin E. Rudland PS. (1993). Oncogene, 8, 999–1008. (1996). Oncogene, 13, 1621–1630. Davies BR, O’Donnell M, Durkan GC, Rudland PS, Becker D and Fikes J. (1993). Gene Transcription, A Practical Barraclough R, Neal DE and Mellon JK. (2002). J. Pathol., Approach: Hames B and Higgins S (eds). IRL Press: Oxford, 196, 292–299. pp 295–319. Davies MPA, Rudland PS, Robertson L, Parry EW, Chen H-L, Fernig DG, Rudland PS, Sparks A, Wilkinson MC Jolicoeur PS and Barraclough R. (1996). Oncogene, 13, and Barraclough R. (2001). Biochem. Biophys. Res. Com- 1631–1637. mun., 286, 1212–1217. Donato R. (2001). Int. J. Biochem. Cell. Biol., 33, 637–668.

Oncogene Association of S100 proteins G Wang et al 1454 Dunnington DJ, Monaghan P, Hughes CM and Rudland PS. Nakanishi K, Sakamoto M, Yasuda J, Takamura M, Fujita N, (1983). J. Natl. Cancer Inst., 71, 1227–1240. Tsuruo T, Todo S and Hirohashi S. (2002). Cancer Res., 62, Endo H, Takenaga K, Kanno T, Satoh H and Mori S. (2002). 2971–2975. J. Biol. Chem., 277, 26396–26402. Ninomiya I, Ohta T, Fushida S, Endo Y, Hashimoto T, Yagi Ford H and Zain S. (1995). Oncogene, 10, 1597–1605. M, Fujimura T, Nishimura G, Tani T, Shimizu K, Gibbs FEM, Wilkinson MC, Rudland PS and Barraclough R. Yonemura Y, Heizmann CW, Scha¨ fer B, Sasaki T and (1994). J. Biol. Chem., 269, 18992–18999. Miwa K. (2001). Int. J. Oncol., 18, 715–720. Gongoll S, Peters G, Mengel M, Piso P, Klempnauer J, Kreipe Pepperkok R, Squire A, Geley S and Bastiaens PI. (1999). H and von Wasielewski R. (2002). Gastroenterology, 123, Curr. Biol., 9, 269–272. 1478–1484. Rahmoune H, Chen HL, Gallagher JT, Rudland PS and Grigorian M, Andresen S, Tulchinsky E, Kriajevska M, Fernig DG. (1998a). J. Biol. Chem., 273, 7303–7310. Carlberg C, Kruse C, Cohn M, Ambartsumian N, Chris- Rahmoune H, Rudland PS, Gallagher JT and Fernig DG. tensen A, Selivanova G and Lukanidin E. (2001). J. Biol. (1998b). Biochemistry, 37, 6003–6008. Chem., 276, 22699–22708. Rudland PS, Dunnington DJ, Gusterson B, Monaghan P and Jenkinson SR, Barraclough R, West CR and Rudland PS. Hughes CM. (1984). Cancer Res., 44, 2089–2101. (2004). Br. J. Cancer, 90, 253–262. Rudland PS, Platt-Higgins A, Renshaw C, West CR, Kimura K, Endo Y, Yonemura Y, Heizmann C, Scha¨ fer B, Winstanley JHR, Robertson L and Barraclough R. (2000). Watanabe Y and Sasaki T. (2000). Int. J. Oncol., 16, Cancer Res., 60, 1595–1603. 1125–1131. Rustandi RR, Baldisseri DM, Inman KG, Nizner P, Hamilton Kriajevska M, Bronstein IB, Scott DJ, Tarabykina S, Fischer- SM, Landar A, Zimmer DB and Weber DJ. (2002). Larsen M, Issinger O and Lukanidin E. (2000). Biochim. Biochemistry, 41, 788–796. Biophys. Acta, 1498, 252–263. Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular Kriajevska M, Cardenas M, Grigorian M, Ambartsumian N, Cloning: A Laboratory Manual, 2nd edn. Cold Spring Georgiev G and Lukanidin E. (1994). J. Biol. Chem., 269, Harbor Laboratory Press: Cold Spring Harbor, NY. 19679–19682. Selinfreund RH, Barger SW, Pledger WJ and Van Eldik LJ. Kriajevska M, Fischer-Larsen M, Moertz E, Vorm O, (1991). Proc. Natl. Acad. Sci. USA, 88, 3554–3558. Tulchinsky E, Grigorian M, Ambartsumian N and Lukani- Sorci G, Bianchi R, Giambanco I, Rambotti MG and Donato din E. (2002). J. Biol. Chem., 277, 5229–5235. R. (1999). Cell Calcium, 25, 93–106. Kriajevska M, Tarabykina S, Bronstein I, Maitland N, Takenaga K, Nakamura Y, Sakiyama S, Hasegawa Y, Sato K Lomonosov M, Hansen K, Georgiev G and Lukanidin E. and Endo H. (1994). J. Cell Biol., 124, 757–768. (1998). J. Biol. Chem., 273, 9852–9856. Tarabykina S, Kriajevska M, Scott DJ, Hill TJ, Lafitte D, Laemmli UK. (1970). Nature, 227, 680–685. Derrick PJ, Dodson GG, Lukanidin E and Bronstein I. Lloyd B, Platt-Higgins A, Rudland PS and Barraclough R. (2000). FEBS Lett., 475, 187–191. (1998). Oncogene, 17, 465–473. Tarabykina S, Scott DJ, Herzyk P, Hill TJ, Tame JR, Most P, Bernotat J, Ehlermann P, Pleger ST, Reppel M, Kriajevska M, Lafitte D, Derrick PJ, Dodson GG, Maitland Borries M, Niroomand F, Pieske B, Janssen PM, Eschenha- NJ, Lukanidin EM and Bronstein IB. (2001). J. Biol. Chem., gen T, Karczewski P, Smith GL, Koch WJ, Katus HA 276, 24212–24222. and Remppis A. (2001). Proc. Natl Acad. Sci. USA, 98, Vallely KM, Rustandi RR, Ellis KC, Varlamova O, Bresnick 13889–13894. AR and Weber DJ. (2002). Biochemistry, 41, 12670–12680. Most P, Remppis A, Pleger ST, Loffler E, Ehlermann P, Wang G, Rudland PS, White MRH and Barraclough R. Bernotat J, Kleuss C, Heierhorst J, Ruiz P, Witt H, (2000). J. Biol. Chem., 275, 11141–11146. Karczewski P, Mao L, Rockman HA, Duncan SJ, Katus Wang G, Zhang S, Fernig DG, Spiller D, Martin-Fernandez HA and Koch WJ. (2003). J. Biol. Chem., 278, 33809–33817. M, Zhang H, Ding Y, Rao Z, Rudland PS and Barraclough Murakami N, Singh S, Chauhan V and Elzinga M. (1995). R. (2004). Biochem. J., 382, 375–383. Biochemistry, 34, 16046–16055. Zhang S, Wang G, Fernig DG, Rudland PS, Webb SE, Nakamura T, Ajiki T, Murao S, Kamigaki T, Maeda S, Ku Y Barraclough R and Martin-Fernandez M. (2004). Eur. and Kuroda Y. (2002). Int. J. Oncol., 20, 937–941. Biophys. J., DOI: 10.1007/s00249-004-0428-x (in press).

Oncogene