Cluster analysis of networks generated through homology: automatic identification of important protein communities involved in cancer metastasis

BACKGROUND

Protein-protein interactions have traditionally been studied on a small scale, using classical biochemical methods to investigate the proteins of interest. More recently large-scale methods, such as two-hybrid screens, have been utilised to survey extensive portions of genomes. Current high-throughput approaches have a relatively high rate of errors, whereas in-depth biochemical studies are too expensive and time-consuming to be practical for extensive studies. As a result, there are gaps in our knowledge of many key biological networks, for which computational approaches are particularly suitable.

RESULTS

We constructed networks, or 'interactomes', of putative protein-protein interactions in the rat proteome--the rat being an organism extensively used for cancer studies. This was achieved by integrating experimental protein-protein interaction data from many species and translating this data into the reference frame of the rat. The putative rat protein interactions were given confidence scores based on their homology to proteins that have been experimentally observed to interact. The confidence score was furthermore weighted according to the extent of the experimental evidence, giving a higher weight to more frequently observed interactions. The scoring function was subsequently validated and networks constructed around key proteins, identified as being highly up- or down-regulated in rat cell lines of high metastatic potential. Using clustering methods on the networks, we have identified key protein communities involved in cancer metastasis.

CONCLUSION

The protein network generation and subsequent network analysis used here, were shown to be useful for highlighting key proteins involved in metastasis. This approach, in conjunction with microarray expression data, can be extended to other species, thereby suggesting possible pathways around proteins of interest.

Spatially distinct binding of Cdc42 to PAK1 and N-WASP in breast carcinoma cells

While a significant amount is known about the biochemical signaling pathways of the Rho family GTPase Cdc42, a better understanding of how these signaling networks are coordinated in cells is required. In particular, the predominant subcellular sites where GTP-bound Cdc42 binds to its effectors, such as p21-activated kinase 1 (PAK1) and N-WASP, a homolog of the Wiskott-Aldritch syndrome protein, are still undetermined. Recent fluorescence resonance energy transfer (FRET) imaging experiments using activity biosensors show inconsistencies between the site of local activity of PAK1 or N-WASP and the formation of specific membrane protrusion structures in the cell periphery. The data presented here demonstrate the localization of interactions by using multiphoton time-domain fluorescence lifetime imaging microscopy (FLIM). Our data here establish that activated Cdc42 interacts with PAK1 in a nucleotide-dependent manner in the cell periphery, leading to Thr-423 phosphorylation of PAK1, particularly along the lengths of cell protrusion structures. In contrast, the majority of GFP-N-WASP undergoing FRET with Cy3-Cdc42 is localized within a transferrin receptor- and Rab11-positive endosomal compartment in breast carcinoma cells. These data reveal for the first time distinct spatial association patterns between Cdc42 and its key effector proteins controlling cytoskeletal remodeling.

Connexin interaction patterns in keratinocytes revealed morphologically and by FRET analysis

Multiple connexins, the major proteins of gap junctions, have overlapping expression in the human epidermis and are postulated to have a key role in keratinocyte differentiation and homeostasis. The functional importance of connexins in the epidermis is emphasised by the association of mutations in four human connexins with various hyperproliferative skin disorders. As immunohistochemistry demonstrated overlapping expression of specific connexins in keratinocytes, we performed colocalisation analyses and applied a modified FRET methodology to assess possible heteromeric interactions between different combinations of four wild-type (wt) and mutant connexins. The data generated indicate that there is evidence for multiple connexin interactions at the plasma membrane between (wt)Cx26, (wt)Cx30 and (wt)Cx31 in keratinocytes and thus, the potential for the formation of a large number of different channel types each with different channel properties. In addition, we demonstrate that the inherent in vitro trafficking defect of the skin disease mutations (D50N)Cx26 and (G11R)Cx30 can be overcome partially by the coexpression of different wild-type connexins but this rescue does not result in large gap junction aggregates at the plasma membrane. These data indicate that skin disease associated Cx26 or Cx30 mutations are likely to disrupt a number of different channel types important in distinct aspects of keratinocyte biology.

PKC controls HGF-dependent c-Met traffic, signalling and cell migration

The growth factor/receptor pair HGF/c-Met exerts control on proliferation, morphogenesis and motility, and through overexpression and mutation is implicated in cancer. Here we have investigated the relationship between receptor signalling and traffic, and its control by specific PKC isotypes. It is shown that c-Met signalling to the ERK cascade occurs within endosomal compartments and that it is in this compartment that PKCepsilon specifically exerts its control on the pathway with the consequent accumulation of ERK in focal complexes. These events are clearly separated from the subsequent microtubule-dependent sorting of c-Met to its perinuclear destination, which is shown to be under the control of PKCalpha. Thus while it is shown that traffic to endosomes is essential for HGF/c-Met to trigger an ERK response, the subsequent traffic and signalling of c-Met controlled by these two PKC isotypes are unconnected events. The dynamic properties conferred by the PKCepsilon control are shown to be essential for a normal HGF-dependent migratory response. Thus PKCs are shown to control both receptor traffic and signal traffic to relay HGF/c-Met responses.

Quantitative fluorescence resonance energy transfer (FRET) measurement with acceptor photobleaching and spectral unmixing

Fluorescence resonance energy transfer (FRET) by acceptor photobleaching is a simple but effective tool for measurements of protein-protein interactions. Until recently, it has been restricted to qualitative or relative assessments owing to the spectral bleed-through contamination resulting from fluorescence overlap between the donor and the acceptor. In this paper, we report a quantitative algorithm that combines the spectral unmixing technique with FRET by acceptor photobleaching. By spectrally unmixing the emissions before and after photobleaching, it is possible to resolve the spectral bleed-through and retrieve the FRET efficiency/interaction distance quantitatively. Using a human keratinocyte cell line transfected with cyan fluorescent protein (CFP)- and yellow fluorescent protein (YFP)-tagged Cx26 connexins as an example, FRET information at homotypic gap junctions is measured and compared with well-established methods. Results indicate that the new approach is sensitive, flexible, instrument independent and solely FRET dependent. It can achieve FRET estimations similar to that from a sensitized emission FRET method. This approach has a great advantage in providing the relative concentrations of the donor and the acceptor; this is, for example, very important in the comparative study of cell populations with variable expression levels.

Two dynamic morphotypes of sarcoma cells, asymmetric stellate and triangle with leading lamella, are related to malignancy

A notion of the dynamic morphotype was developed as a conjunction between cell shape and migration. This enabled the investigation of the relationship between malignancy and patterns of dynamic morphology in neoplastic cells in vitro. Time-lapse cinemicroscopy was used to analyse the cell behaviour of three rat neoplastic cell lines (K2, T15, and A8), differing in metastatic potential, that were instrumental in revealing a coincidence between high migratory activity and appearance of the 3D structure of actin cables in high-malignant A8 cells (Pokorná et al., 1994). A set of criteria was established for visual classification of cell morphology. Matching the pattern of cell morphology with locomotory activity led to identification of four dynamic morphotypes. Cell speed was determined by tracking and the dynamic morphotypes assigned by the operator. All the three cell populations were studied for incidence of the dynamic morphotypes in culture media differing in pH: 6.6 simulating acid extracellular condition in tumours, physiological 7.4, and alkaline 8.2. The results showed that acid pH stimulated motile activity in the intermediate-malignant T15 and most malignant A8 cells. The T15 and A8 cells also manifested a prolonged continuation of fast locomotion in the early G1 phase and displayed a prevalence of two fast moving dynamic morphotypes: asymmetric stellate and triangle with leading lamella.

Protein kinase C controls microtubule-based traffic but not proteasomal degradation of c-Met

Upon hepatocyte growth factor stimulation, its receptor c-Met is rapidly internalized via clathrin-coated vesicles and traffics through an early endosomal compartment. We show here that c-Met accumulates progressively in perinuclear compartments, which in part include the Golgi. The c-Met content in the Golgi is principally the newly synthesized precursor form and, to a lesser extent, the internalized, recycling c-Met. By following the trafficking of c-Met inside the cell using a semi-automatic procedure and using inhibition or activation of protein kinase C (PKC) and microtubule depolymerizing agents, we show that PKC positively controls the trans-cytosolic movement of c-Met along microtubules. In parallel to its traffic, internalized c-Met is progressively degraded by a proteasome-sensitive mechanism; the lysosomal pathway does not play a substantial role. Inhibition or promotion of c-Met traffic to the perinuclear compartment does not alter the kinetics of proteasome-dependent c-Met degradation. Thus susceptibility to proteasomal degradation is not a consequence of post-endocytic traffic. The data define a PKC-controlled traffic pathway for c-Met that operates independently of its degradative pathway.

The Drosophila hairy RNA localization signal modulates the kinetics of cytoplasmic mRNA transport

In several Drosophila cell types, mRNA transport depends on microtubules, the molecular motor dynein and trans-acting factors including Egalitarian and Bicaudal-D. However, the molecular basis of transcript recognition by the localization machinery is poorly understood. Here, we characterize the features of hairy pair-rule RNA transcripts that mediate their apical localization, using in vivo injection of fluorescently labelled mRNAs into syncytial blastoderm embryos. We show that a 121-nucleotide element within the 3'-untranslated region is necessary and sufficient to mediate apical transport. The signal comprises two essential stem-loop structures, in which double-stranded stems are crucial for localization. Base-pair identities within the stems are not essential, but can contribute to the efficiency of localization, suggesting that specificity is mediated by higher-order structure. Using time-lapse microscopy, we measure the kinetics of localization and show that impaired localization of mutant signals is due to delayed formation of active motor complexes and, unexpectedly, to slower movement. These findings, and those from co-injecting wild-type and mutant RNAs, suggest that the efficiency of molecular motors is modulated by the character of their cargoes.

Rapid actin transport during cell protrusion

Transformed rat fibroblasts expressing two variants of green fluorescent protein, each fused to beta-actin, were used to study actin dynamics during cell protrusion. The recently developed FLAP (fluorescence localization after photobleaching) method permits the tracking of one fluorophore after localized photobleaching by using the other as a colocalized reference. Here, by visualizing the ratio of bleached to total molecules, we found that actin was delivered to protruding zones of the leading edge of the cell at speeds that exceeded 5 micrometers per second. Monte Carlo modeling confirmed that this flow cannot be explained by diffusion and may involve active transport.

Localization of DNA polymerases eta and iota to the replication machinery is tightly co-ordinated in human cells

Y-family DNA polymerases can replicate past a variety of damaged bases in vitro but, with the exception of DNA polymerase eta (poleta), which is defective in xeroderma pigmentosum variants, there is little information on the functions of these polymerases in vivo. Here, we show that DNA polymerase iota (poliota), like poleta, associates with the replication machinery and accumulates at stalled replication forks following DNA-damaging treatment. We show that poleta and poliota foci form with identical kinetics and spatial distributions, suggesting that localization of these two polymerases is tightly co-ordinated within the nucleus. Furthermore, localization of poliota in replication foci is largely dependent on the presence of poleta. Using several different approaches, we demonstrate that poleta and poliota interact with each other physically and that the C-terminal 224 amino acids of poliota are sufficient for both the interaction with poleta and accumulation in replication foci. Our results provide strong evidence that poleta targets poliota to the replication machinery, where it may play a general role in maintaining genome integrity as well as participating in translesion DNA synthesis.

Localization of DNA polymerases eta and iota to the replication machinery is tightly co-ordinated in human cells

Y-family DNA polymerases can replicate past a variety of damaged bases in vitro but, with the exception of DNA polymerase eta (poleta), which is defective in xeroderma pigmentosum variants, there is little information on the functions of these polymerases in vivo. Here, we show that DNA polymerase iota (poliota), like poleta, associates with the replication machinery and accumulates at stalled replication forks following DNA-damaging treatment. We show that poleta and poliota foci form with identical kinetics and spatial distributions, suggesting that localization of these two polymerases is tightly co-ordinated within the nucleus. Furthermore, localization of poliota in replication foci is largely dependent on the presence of poleta. Using several different approaches, we demonstrate that poleta and poliota interact with each other physically and that the C-terminal 224 amino acids of poliota are sufficient for both the interaction with poleta and accumulation in replication foci. Our results provide strong evidence that poleta targets poliota to the replication machinery, where it may play a general role in maintaining genome integrity as well as participating in translesion DNA synthesis.

Defective trafficking and cell death is characteristic of skin disease-associated connexin 31 mutations

Distinct germline mutations in the gene (GJB3) encoding connexin 31 (Cx31) underlie the skin disease erythrokeratoderma variabilis (EKV) or sensorineural hearing loss with/without peripheral neuropathy. Here we describe a number of functional analyses to investigate the effect of these different disease-associated Cx31 mutants on connexon trafficking and intercellular communication. Immunostaining of a biopsy taken from an EKV patient harbouring the R42P mutation revealed sparse epidermal staining of Cx31, and, when present, it had a perinuclear localization. Transfection and microinjection studies in both keratinocytes and fibroblast cell lines also demonstrated that R42P and four other EKV-associated mutant Cx31 proteins displayed defective trafficking to the plasma membrane. The deafness/neuropathy only mutant 66delD had primarily a cytoplasmic localization, but some protein was visualized at the plasma membrane in a few transfected cells. Both 66delD- and R32W-Cx31/EGFP proteins had significantly impaired dye transfer rates compared to wild-type Cx31/EGFP protein. A striking characteristic feature observed with the dominant skin disease Cx31 mutations was a high incidence of cell death. This was not observed with wild-type, R32W 66delD Cx31 proteins. In conclusion, we have identified some key cellular phenotypic differences with respect to disease-associated Cx31 mutations.

Restoration of podosomes and chemotaxis in Wiskott-Aldrich syndrome macrophages following induced expression of WASp

We used a direct-viewing (Dunn) chemotaxis chamber to analyse the chemotactic responses of human normal and Wiskott-Aldrich syndrome (WAS) macrophages to the cytokine colony stimulating factor-1 (CSF-1). In five patients with classic WAS, where specialised adhesion complexes called podosomes are absent, chemotaxis of macrophages was abolished. The deficient chemotactic responses of WAS macrophages following cytokine stimulation could be correlated with abnormalities in cell polarisation and actin organisation. In a series of cell microinjection studies we found that normal chemotactic responses were restored in WASp macrophages transfected with a full-length human WAS construct. Expression of exogenous WAS protein (WASp) in these cells also restored normal polarised cell morphology and the ability to form podosomes.

Fluorescence localization after photobleaching (FLAP): a new method for studying protein dynamics in living cells

FLAP is a new method for localized photo-labelling and subsequent tracking of specific molecules within living cells. It is simple in principle, easy to implement and has a wide potential application. The molecule to be located carries two fluorophores: one to be photobleached and the other to act as a reference label. Unlike the related methods of fluorescence recovery after photobleaching (FRAP) and fluorescence loss in photobleaching (FLIP), the use of a reference fluorophore permits the distribution of the photo-labelled molecules themselves to be tracked by simple image differencing. In effect, FLAP is therefore comparable with methods of photoactivation. Its chief advantage over the method of caged fluorescent probes is that it can be used to track chimaeric fluorescent proteins directly expressed by the cells. Although methods are being developed to track fluorescent proteins by direct photoactivation, these still have serious drawbacks. In order to demonstrate FLAP, we have used nuclear microinjection of cDNA fusion constructs of beta-actin with yellow (YFP) and cyan (CFP) fluorescent proteins to follow both the fast relocation dynamics of monomeric (globular) G-actin and the much slower dynamics of filamentous F-actin simultaneously in living cells.

Ezrin is a downstream effector of trafficking PKC-integrin complexes involved in the control of cell motility

Protein kinase C (PKC) alpha has been implicated in beta1 integrin-mediated cell migration. Stable expression of PKCalpha is shown here to enhance wound closure. This PKC-driven migratory response directly correlates with increased C-terminal threonine phosphorylation of ezrin/moesin/radixin (ERM) at the wound edge. Both the wound migratory response and ERM phosphorylation are dependent upon the catalytic function of PKC and are susceptible to inhibition by phosphatidylinositol 3-kinase blockade. Upon phorbol 12,13-dibutyrate stimulation, green fluorescent protein-PKCalpha and beta1 integrins co-sediment with ERM proteins in low-density sucrose gradient fractions that are enriched in transferrin receptors. Using fluorescence lifetime imaging microscopy, PKCalpha is shown to form a molecular complex with ezrin, and the PKC-co-precipitated endogenous ERM is hyperphosphorylated at the C-terminal threonine residue, i.e. activated. Electron microscopy showed an enrichment of both proteins in plasma membrane protrusions. Finally, overexpression of the C-terminal threonine phosphorylation site mutant of ezrin has a dominant inhibitory effect on PKCalpha-induced cell migration. We provide the first evidence that PKCalpha or a PKCalpha-associated serine/threonine kinase can phosphorylate the ERM C-terminal threonine residue within a kinase-ezrin molecular complex in vivo.

Specific changes to the mechanism of cell locomotion induced by overexpression of (β)-actin

Overexpression of (β)-actin is known to alter cell morphology, though its effect on cell motility has not been documented previously. Here we show that overexpressing (β)-actin in myoblasts has striking effects on motility, increasing cell speed to almost double that of control cells. This occurs by increasing the areas of protrusion and retraction and is accompanied by raised levels of (β)-actin in the newly protruded regions. These regions of the cell margin, however, show decreased levels of polymerised actin, indicating that protrusion can outpace the rate of actin polymerisation in these cells. Moreover, the expression of (β)*-actin (a G244D mutant, which shows defective polymerisation in vitro) is equally effective at increasing speed and protrusion. Concomitant changes in actin binding proteins show no evidence of a consistent mechanism for increasing the rate of actin polymerisation in these actin overexpressing cells. The increase in motility is confined to poorly spread cells in both cases and the excess motility can be abolished by blocking myosin function with butanedione monoxime (BDM). Our observations on normal myoblasts are consistent with the view that they protrude by the assembly and cross linking of actin filaments. In contrast, the additional motility shown by cells overexpressing (β)-actin appears not to result from an increase in the rate of actin polymerisation but to depend on myosin function. This suggests that the additional protrusion arises from a different mechanism. We discuss the possibility that it is related to retraction-induced protrusion in fibroblasts. In this phenomenon, a wave of increased protrusion follows a sudden collapse in cell spreading. This view could explain why it is only the additional motility that depends on spreading, and has implications for understanding the differences in locomotion that distinguish tissue cells from highly invasive cell types such as leucocytes and malignant cells.

Effect of pH on proteinase secretion by transformed fibroblast populations

Effect of pH on secretion of proteolytic enzymes in cell cultures of three clonal lines of transformed fibroblasts (K2, T15 and K4) was studied by using 14C-labelled denatured proteins as substrate. One line of malignant macrophages derived from mouse reticulum cell sarcoma (J774.1) was used for comparison. The relative motility index of all cell lines was derived by computer analysis of quantitative estimations of cell dispersion in single-cell-derived colonies. Cultivation at pH 6.5 decreased the growth rate in most experiments as compared with that at pH 7.4, and stimulated cell motility to a different extent. The population of mouse malignant macrophages produced several-fold higher extracellular proteolytic activity than the fibroblast lines. Secretion of proteinases by the malignant macrophages was significantly stimulated by the lower pH. Enzyme secretion by two of the three fibroblast derivatives was also stimulated by acidic pH but to a lesser extent than the secretion of the malignant macrophages. The assessment of motility done by measurement of dispersion of cells in colony proved a positive correlation between motility and proteinase secretion in J774.1 cells and one transformed fibroblast clone (T15) but not in the two other clonal lines.

TGFbeta1 induces a cell-cycle-dependent increase in motility of epithelial cells

We have previously shown that addition of type 1 transforming growth factor-beta (TGFbeta1) to an exponentially growing population of mink lung CCl64 cells increases their average intermitotic time from 14.4 to 20.3 hours, predominantly by extending G1 from 7.5 to 13.5 hours. Here we have used the DRIMAPS system (digitally recorded interference microscopy with automatic phase-shifting) for obtaining data on cellular mass distribution, cell motility and morphology. We found no significant change in the cells' rate of mass increase following TGFbeta1 treatment, which implies that the treated cells attained a higher mass during their extended cell cycle and this was confirmed by direct measurement of cell size. However, the cells showed a dramatic motile response to treatment: TGFbeta1-treated cells had a significantly higher time-averaged speed of 36.2 microm hour-1 compared to 14.5 microm hour-1 for the control cells. The time course of the response was gradual, reaching a maximum mean speed of 52.6 microm hour-1 after 15 hours exposure. We found that the gradual onset of the response was probably not due to a slow accumulation of a secondary factor but because cells were dividing throughout the experiment and most of the response to TGFbeta1 occurred only after the first cell division in its presence. Thus, taking only those cells that had not yet divided, the time-averaged speed of treated cells (26.1 micrometer hour-1) was only moderately higher than that of untreated cells (14.9 micrometer hour-1) whereas, for those cells that had divided, the difference in speed between treated cells (45.1 micrometer hour-1) and untreated cells (14.1 microm hour-1) was much greater. Increased speed was a consequence of enhanced protrusion and retraction of the cell margin coupled with an increase in cell polarity. TGFbeta1 also increased the mean spreading of the cells, measured as area-to-mass ratio, from 3.2 to 4.4 micrometer2 pg-1, and the intracellular mass distribution became more asymmetric. The observations indicate that a G2 signal may be necessary to reach maximal motility in the presence of TGFbeta1.

Using molecular genetics as a tool in understanding crawling cell locomotion in myoblasts

We have used digitally recorded interference microscopy with automatic phase shifting (DRIMAPS) to investigate the crawling locomotion of normal and mutant mouse myoblasts. Contraction forces that give rise to cell body movement, tail retraction and cell adhesion to the substrate in myoblasts and other locomoting tissue cells arise from the interactions of actin and non-muscle myosin II. The activity of non-muscle myosin II is regulated differently from that of skeletal myosin. Using DRIMAPS, we found that crawling locomotion was altered in myoblasts that heterologously expressed human beta-cardiac myosin heavy chain (MHC); the cells moved more slowly and had reduced rates of protrusion and retraction. Immunolocalization demonstrated that MHC and non-muscle myosin II were not co-localized, suggesting that MHC does not compete directly with myosin II, but interferes with cell locomotion by binding inappropriately to actin filaments and possibly cross-linking them. Myosin I may be involved in protrusion of the lamellipodia. However, using DRIMAPS, we found that crawling locomotion was unaltered in myoblasts that heterologously expressed a truncated myosin I which lacked the membrane-binding tail domain. This suggests that, if endogenous myosin I is important for cell locomotion, this mutant was unable to interfere with its action. We conclude that the effects on locomotion of expressing foreign or mutant proteins of the cytoskeleton in vertebrate cells can be subtle and can be swamped by the intrinsic variability of the cells. Their characterization requires automated methods of acquiring data, such as DRIMAPS, and careful statistical analysis in order to take account of other sources of variation.

Disruption of the talin gene compromises focal adhesion assembly in undifferentiated but not differentiated embryonic stem cells

We have used gene disruption to isolate two talin (-/-) ES cell mutants that contain no intact talin. The undifferentiated cells (a) were unable to spread on gelatin or laminin and grew as rounded colonies, although they were able to spread on fibronectin (b) showed reduced adhesion to laminin, but not fibronectin (c) expressed much reduced levels of beta1 integrin, although levels of alpha5 and alphaV were wild-type (d) were less polarized with increased membrane protrusions compared with a vinculin (-/-) ES cell mutant (e) were unable to assemble vinculin or paxillin-containing focal adhesions or actin stress fibers on fibronectin, whereas vinculin (-/-) ES cells were able to assemble talin-containing focal adhesions. Both talin (-/-) ES cell mutants formed embryoid bodies, but differentiation was restricted to two morphologically distinct cell types. Interestingly, these differentiated talin (-/-) ES cells were able to spread and form focal adhesion-like structures containing vinculin and paxillin on fibronectin. Moreover, the levels of the beta1 integrin subunit were comparable to those in wild-type ES cells. We conclude that talin is essential for beta1 integrin expression and focal adhesion assembly in undifferentiated ES cells, but that a subset of differentiated cells are talin independent for both characteristics.