Most vertebrate genomes contain 3 TAGLN genes and many tissues and cells express more than one gene. In order to examine the function of single TAGLN gene products and to increase the chances of being able to deplete all TAGLN gene products in a given cell, we made an initial screen of cell lines from different tissue sources in order to determine what complement of TAGLN gene products were expressed. After analyzing cell lines of fibroblast, muscle, neuronal and epithelial origin, we selected the rat embryo fibroblast cell line REF52 as suitable for further study. This was not only due to its flattened morphology and well organized actin stress fibre morphology that would aid morphological characterization of effects on the actin cytoskeleton, but also based on western blotting showing the presence of only one of the TAGLN gene products namely SM22α (Figure 1A). In keeping with previous reports that SM22 is an F-actin binding/bundling protein, both the endogenous SM22α, revealed by specific antibody staining, and an exogenous SM22α-GFP construct localized strongly with the prominent F-actin containing stress fibres in these cells (Figure 1B-G).

In order to further dissect the role of SM22, we used siRNA-mediated SM22 depleted REF52 cells to investigate the contribution of SM22 to the actin morphology. Depletion of SM22 to between 20% and 40% of normal levels resulted in a dramatic and reproducible change in actin stress fibre morphology (Figure 2 and Additional file 1, Figure S1). In clones of REF52 cells depleted for SM22, using either REF52 stably expressing siRNA constructs directed against SM22 (Additional file 1, Figure S1) or transient RNAi expression driven from a pSiren-DNR-dsRed plasmid (Figure 2), there was a reproducible and qualitatively similar reduction in SM22 with concomitant loss of actin stress fibre organization. In greater than 70% of wild-type cells actin was organized in longer relatively parallel arrays of fibres, the typical morphology for this cell type (Figure 2A, B). However, in two independent clones (Knockdown A and B; Figure 2) of SM22 depleted cells there was a significant increase in the proportion of cells observed with shorter and orthogonal arrays of actin fibres. These cells also had a marked increase in the proportion of cells with no apparent stress fibres. See insets in Figure 2A for detail of the phenotypes. These observations demonstrate that SM22 has a significant role in the organization of the actin cytoskeleton, and that reduction in SM22 levels leads to a less bundled, less well organized and potentially more dynamic actin cytoskeleton.

Given the importance of the actin cytoskeleton in cell motility, and that SM22 levels appear altered in tumour cells that often have altered migratory properties, we next investigated the effect of reducing SM22 levels on various in vitro measures of cell migration, chemotaxis and invasion. For improved consistency of cellular response in these experiments REF52 cells stably expressing shRNA were used with a resulting 50% depletion of SM22 compared to sense control (Figure 3A). Single cell tracking of individual SM22 depleted cells revealed that the SM22 depleted cells showed impaired movement with considerably reduced trajectories compared to the sense control. This data was then used to calculate cellular velocities, and as expected revealed that both SM22 depleted clones had significantly reduced average velocity compared to that of the sense control (Figure 3D). Similar effects on cell migration have been noted for vascular smooth muscle cells depleted for SM22 17 .

We also analyzed whether there was a change in chemotactic response in SM22 depleted cells. Loss of actin organization can affect not only the mechanics of cell migration, but also cell polarity, which is required for productive directional migration. Cells exposed to a serum gradient in a Dunn chamber assay showed no significant defect in their chemotactic response, indeed Clone 1B appeared to have a higher persistence ratio than wildtype cells (Figure 4A) but nonetheless still migrated a shorter distance. However, motility assays carried out on rigid two-dimensional substrates do not always reflect the motility and invasion of cells in three dimensions which would be more analogous to the in vivo situation. We therefore exposed SM22 depleted cells to a chemotactic gradient in Matrigel coated Boyden chambers. Interestingly, in this environment the SM22 depleted cells were able to invade and migrate through the Matrigel much more readily than the control cells (Figure 4B). Therefore despite the inability to migrate and the relative lack of responsiveness to chemotactic stimuli on hard two-dimensional substrates, SM22 depleted cells appear to be much more effective at migrating through soft substrates in response to a similar chemotactic stimulus.

Podosomes, and the related structures invadopodia, are typically found in macrophages, osteoclasts and tumour cells and are not usually seen in other mesenchymal cells unless transformed for example with active Src 18 . Smooth muscle cells and myoblasts however, have been shown to form podosomes under certain non-transformed conditions, including following phorbol ester or TGF-β stimulation 19 20 21 22 . On examining the actin staining pattern of SM22 depleted cells in more detail, we noted in addition to the change from the prominent stress fibre phenotype seen in wild type and control REF52 cells, to the less organized actin morphology of SM22 depleted cells (Figure 2), peripheral bands of diffuse F-actin staining and dense puncta of F-actin were now observed with higher frequency (Figure 5A, B). These F-actin puncta and diffuse clouds of F-actin are reminiscent of podosomes and F-actin rosettes seen in other migratory and invasive cells. Quantification of these additional actin phenotypes demonstrated that there was a increase in both lamellipodial rosettes and/or spontaneous podosome formation in cells depleted for SM22 (Figure 5C) although only the increase in peripheral actin bands was significant. These findings demonstrate a further change in cellular actin structures in response to SM22 depletion. The relatively static stress fibre phenotype with associated focal adhesions changes to a more dynamic F-actin based adhesion system is seen by the appearance of more podosomes and rosettes in SM22 depleted cells.

Deletion of the SM22 homologue in yeast leads to increased actin dynamics and reduced levels of reactive oxygen species 13 . Here we describe SM22 depletion having an effect on actin organization by reducing prominent stress fibres to a more disorganized array (Figure 2) also suggestive of an increase in actin dynamics, with further evidence of an increase in actin dynamics revealed by the appearance of podosomes (Figure 5). In addition, invasive motility is also increased. A concomitant reduction in reactive oxygen species would potentially allow a cell depleted for SM22 to have an increased survival advantage in a hypoxic tumour environment. In order to test whether deleted SM22 levels would also lead to changes in cell's levels of reactive oxygen species, we subjected REF52 cells, either normal controls or SM22 depleted, to complete serum starvation for either 24 or 48 hours. Serum starvation is a recognized cellular stress that leads to elevated ROS in normal and tumour cells ( 23 24 and references therein). As can be seen from Figure 6A, both 24 and 48 h serum starvation induced a greater than 15-fold increase in ROS levels in control REF52 cells. However, analysis of ROS levels in SM22 depleted REF52 cells (Figure 6B), revealed considerably reduced ROS levels at 24 h and 48 h (4-fold and 11-fold lower respectively than controls). Moreover untreated SM22-depleted cells had a baseline ROS level one third of wildtype cells. SM22 depletion per se appears to reduce ROS levels in REF52 cells, as was seen in yeast when the yeast orthologue Scp1 was deleted 13 , but in addition SM22 depletion also appears to be protective against stressful stimuli that lead to ROS accumulation.

Given that SM22 depletion in REF52 cells had a significant effect on actin morphology and increased Matrigel invasion, and that SM22 is also reduced in many tumour cells 5 25 , we investigated the effect of manipulating SM22 levels in PC3 prostate cancer cells, a cell type that forms podosomes spontaneously 26 . As SM22 is not detectable by western blotting in PC3 cells (data not shown and see 27 ), we therefore compared the ability of normal PC3, and PC3 overexpressing SM22 to form podosomes and to migrate through Matrigel coated Boyden chambers. As can be seen in Figure 7A, SM22 overexpression reduced the ability of PC3 cells to migrate through Matrigel in response to a serum gradient. Podosomes were detected using antibodies to a podosome component cortactin. PC3 cells can be seen to exhibit prominent ruffling membranes and podosomes (Figure 7B-G), as indicated by co-staining of F-actin and cortactin. However, SM22 expressing PC3 cells had reduced ruffling membranes and were devoid of podosomes (Figure 7H-J).

SDS-PAGE and western blotting were carried out as described previously 41 . Antisera against SM22α (ab10135, Abcam plc) NP22, NP25 42 , tubulin (Sigma) and actin (Sigma) were used for the detection of SM22 isoforms and for quantization of shRNA and RNAi knockdown.

REF52 43 and PT67 (Clontech) cells were cultured in Dulbecco's modified Eagle's medium (DMEM) and PC3 cells 44 in RPMI 1640, each supplemented with 10% foetal calf serum (FCS) (Invitrogen). All transfections were performed using Lipofectamine (Invitrogen) at 70% confluency with 2 μg of plasmid of interest per 35 mm dish. For immunofluorescence microscopy, cells were seeded on coverslips, and following treatment of transfection were fixed with 3.7% paraformaldehyde for 10 minutes and permeabilised with 0.05% Triton X-100 in PBS for 1 minute. Fixed cells were then incubated with Alexa 488-conjugation Phalloidin (Molecular Probes) to visualize F-actin, anti-SM22α antibody (ab10135, Abcam plc), cortactin (4F11, Upstate). Images were captured digitally by cooled CCD camera on a Leica DMIRE2 microscope. For the generation of stable cell lines, stable transfected cells were first selected with relevant antibiotic 48 hours post transfection. Stable clones were then selected.

For cell migration assays, cells were allowed to adhere to 35 mm dishes overnight, then transferred onto a 37°C heated microscope stage within 5% CO₂ chamber. Cell movements were captured on 2.5 minute time-lapse using a Leica DMIRBE microscope and Volocity software. The migratory paths of individual cells were tracked using ImageJ (version 1.38x) with Manual Tracking PlugIns. Movement and velocity from the initial point where calculated for each cell using Excel. 10 cells were analyzed for each triplicate independent experiment per cell line. Matrigel-coated Boyden chambers were purchased from BD Biosciences and cell invasion assays were performed and analyzed according to the manufacturers instructions, and using 10% serum in the lower chamber as chemoattractant.

Potential 21mer sequences to be used for RNAi of SM22α were identified using the BBSRC Chick EST database RNAi target sequence prediction tool http://www.chick.umist.ac.uk/. The selected oligonucleotide sequences GCGTGATTCTGAGCAAGTTTTCAAGAGAAACTTGCTCAGAATCACGCTTTTTTACGCGT and ACGCGTAAAAAAGCGTGATTCTGAGCAAGTTTCTCTTGAAAACTTGCTCAGAATCACGC were cloned into pSIREN retroviral vector (Clontech). This vector was then transfected into the packaging PT67 cell line for production of retroviral SM22α siRNA vector. Target cells were infected twice with retrovirus before selection with puromycin. Sense control was previously generated in Ref52 cells using sense DG siRNA sequence 45 . Transient RNAi knockdown was achieved using pSiren-DNR-dsRed (Clontech).

Control REF52, or SM22 depleted REF52 cells were grown at subconfluent levels in serum containing media as described above, or in media without serum for 24 and 48 hours. Following the indicated treatments, cells were incubated with 2',7'-dichlorodihydrofluorescein diacetate (H₂DCFDA, Invitrogen, Karlsruhe, Germany) at 200 μg/μl for 1 hour at 37°C, washed in PBS, trypsinised and incubated with 100 μg/ml Propidium iodide (PI) for 15 minutes to visualize and exclude apoptotic/necrotic cells. Treated cells were then collected in PBS and analyzed using a Cyan Flow Cytometer (DakoCytomation) with Summit4.3 software.