Rac1 and Rac2 regulate macrophage morphology but are not essential for migration.

Rac GTPases are believed to contribute to migration in leukocytes by transducing signals from cell surface receptors to the actin and microtubule cytoskeletons. Mammals have three closely related Rac isoforms, Rac1, Rac2 and Rac3, and it is widely assumed that cell migration requires the activity of these Rac GTPases. We have previously shown that Rac1-null mouse macrophages have altered cell shape and reduced membrane ruffling but normal migration speed. Here we investigate the behaviour of macrophages lacking Rac2 (Rac2–/–) or Rac1 and Rac2 (Rac1/2–/–). Rac2–/– macrophages have reduced F-actin levels and lack podosomes, which are integrin-based adhesion sites, and their migration speed is similar to or slightly slower than wild-type macrophages, depending on the substrate. Unexpectedly, Rac1/2–/– macrophages, which do not express Rac1, Rac2 or Rac3, migrate at a similar speed to wild-type macrophages on a variety of substrates and perform chemotaxis normally, although their morphology and mode of migration is altered. However, Rac1–/– and Rac1/2–/– but not Rac2–/– macrophages are impaired in their ability to invade through Matrigel. Together, these data show that Rac1 and Rac2 have distinct roles in regulating cell morphology, migration and invasion, but are not essential for macrophage migration or chemotaxis.


Introduction
Rac proteins are Rho-family GTPases involved in transducing signals from cell surface receptors to the cytoskeleton, cell-cell and cell-substratum adhesions and cell cycle (Jaffe and Hall, 2005). They are molecular switches that cycle between a GTPbound active form and a GDP-bound inactive form. Mammals have three highly homologous Rac genes, Rac1, Rac2 and Rac3, of which Rac1 and Rac2 are expressed in leukocytes. Although the amino acid sequences of Rac1 and Rac2 are 90% identical, they have unique roles in regulating leukocyte function (Bokoch, 2005).
Expression of dominant-negative Rac1 (N17Rac1) and constitutively active Rac1 (V12Rac1) mutants has been used to implicate Rac GTPases in specific cellular responses. Constitutively active Rac GTPases promote formation of lamellipodia and membrane ruffles, whereas N17Rac1 inhibits cell migration, membrane ruffling and lamellipodial extension induced by growth factors, cytokines and adhesion in many cell types including macrophages (Di Marzio et al., 2005;Ridley et al., 2003). However, these studies do not differentiate between the different Rac isoforms. Further, data from dominant-negative studies can be misleading as it is not known whether N17Rac1 only inhibits Rac1 or also affects other closely related Rho GTPases such as Cdc42, TC10 and RhoG (Etienne-Manneville and Hall, 2002). Rac isoform-specific functions have been investigated in knockout mice and more recently using RNAi. Rac2 is expressed primarily in haematopoietic cells, and Rac2-null mice are viable and have been studied extensively (Bokoch, 2005;Weston and Stankovic, 2004). Similarly, Rac3-null mice are viable (Corbetta et al., 2005), whereas Rac1-null mice die early in embryogenesis (Sugihara et al., 1998). To compare the function of Rac1 and Rac2 in the same cell type, conditional knockouts of Rac1 have been generated in B cells, T cells, haematopoietic stem cells and neutrophils. Using these, Rac1 and Rac2 have been shown to share some redundant functions as well as unique roles in proliferation, motility, adhesion, phagocytosis and haematopoiesis (Bokoch, 2005;Dinauer, 2003;Pradip et al., 2003;Sun et al., 2004).
It is often assumed that because Rac1 induces formation of lamellipodia it will also play a role in cell migration. However, we have previously shown that although Rac1-deficient macrophages have defects in cell spreading and membrane ruffling they are able to migrate normally (Wells et al., 2004). Others have shown that Rac2-null macrophages have reduced superoxide production, phagocytosis and integrin-mediated migration in a Transwell assay (Pradip et al., 2003;Yamauchi et al., 2004). Rac3 is not expressed in macrophages, but it is Rac GTPases are believed to contribute to migration in leukocytes by transducing signals from cell surface receptors to the actin and microtubule cytoskeletons. Mammals have three closely related Rac isoforms, Rac1, Rac2 and Rac3, and it is widely assumed that cell migration requires the activity of these Rac GTPases. We have previously shown that Rac1-null mouse macrophages have altered cell shape and reduced membrane ruffling but normal migration speed. Here we investigate the behaviour of macrophages lacking Rac2 (Rac2 -/-) or Rac1 and Rac2 (Rac1/2 -/-). Rac2 -/macrophages have reduced F-actin levels and lack podosomes, which are integrin-based adhesion sites, and their migration speed is similar to or slightly slower than wild-type macrophages, depending on the substrate. Unexpectedly, Rac1/2 -/macrophages, which do not express Rac1, Rac2 or Rac3, migrate at a similar speed to wild-type macrophages on a variety of substrates and perform chemotaxis normally, although their morphology and mode of migration is altered. However, Rac1 -/and Rac1/2 -/but not Rac2 -/macrophages are impaired in their ability to invade through Matrigel. Together, these data show that Rac1 and Rac2 have distinct roles in regulating cell morphology, migration and invasion, but are not essential for macrophage migration or chemotaxis.
possible that Rac2 compensates for loss of Rac1. Here we report that Rac2 deletion has little effect on macrophage migration speed although it does induce loss of podosomes and reduce levels of polymerized actin. Deletion of Rac1 and Rac2 alters cell morphology but surprisingly does not prevent the migration or chemotaxis of macrophages, although it alters their mode of migration, and loss of Rac1 reduces invasion.
Rac1 -/-BMM have a clearly distinct shape to Wt BMM. They are more elongated and have a smaller spread area, but they still have some lamellipodia (Wells et al., 2004). The morphology of Rac2 -/-BMM was not as different: they were more elongated than Wt BMM ( Fig. 2A, Fig. 3B; supplementary material Movies 1 and 2), but their spread area was not affected (Fig. 3A). By contrast, Rac1/2 -/-BMMs had a distinctive morphology. Many of them had multiple thin protrusions, giving cells a stellate appearance ( Fig. 2A). These thin protrusions contained bundles of microtubules but little F-actin. Rac1/2 -/-BMMs had a significantly reduced spread area and were more elongated than Wt BMMs (Fig. 3A,B). To quantify the effects of Rac isoform deletion on cell morphology, cells were categorised into three groups: stellate cells with more than two microtubule-containing protrusions, Journal of Cell Science 119 (13) elongated cells with two microtubule-containing protrusions, and migratory cells with one or more F-actin-rich lamellae and a tail (Fig. 3A). The majority of Wt BMMs in growth medium adopted a migratory morphology, whereas Rac2 -/-BMMs mostly had an elongated morphology and Rac1/2 -/-BMMs either had a stellate morphology or elongated morphology (Fig. 3B,D; movies 1-3). Rac2 -/rarely had spread lamellae (supplementary material Movie 2), and the thin protrusions of Rac1/2 -/-BMMs did not spread out to form lamellae.

Podosome assembly is regulated differentially by Rac1 and Rac2
Podosomes are F-actin-rich integrin-based adhesions observed in a variety of cell types including macrophages, osteoclasts and endothelial cells (Linder and Aepfelbacher, 2003). They localize primarily in protrusive regions of macrophages (Evans et al., 2003), and are thought to contribute to cell migration. Podosomes have an F-actin core surrounded by a ring of integrins and associated proteins such as paxillin and vinculin (Linder and Aepfelbacher, 2003;Luxenburg et al., 2006). N17Rac1 inhibits podosome assembly (Moreau et al., 2003), but whether Rac1 or Rac2 contribute to podosome formation is not known. Podosomes were quantified in BMMs by comparing F-actin and paxillin localization. Podosomes were present in 45% of Wt BMMs, but no podosomes were detected in Rac2 -/or Rac1/2 -/-BMMs (Table 1). Rac1 -/-BMMs contained F-actin foci localised within lamellae of BMMs (Fig. 4), similar to Wt BMMs. However, paxillin was not present in around 50% of these foci and in the remainder paxillin was diffusely localized around the foci and not in a clear ring, thus these structures were not normal podosomes ( Fig. 4 and Table 1). CSF-1 stimulates podosome assembly in CSF-1-starved Wt BMMs (Wheeler et al., 2006), but no podosomes were observed in CSF-1stimulated Rac2 -/or Rac1/2 -/-BMMs (Fig. 2D). CSF-1 increased the number of Rac1 -/-BMMs with F-actin foci by 30%, but these were still not real podosomes, as determined by paxillin localization (Fig. 4; data not shown). Rac2 is therefore essential for podosome assembly, whereas in the absence of Rac1 F-actin foci are formed that could be podosome precursors, but many of these lack correctly associated paxillin. To determine the expression profile of Rac isoforms in BMMs, RT-PCR was performed with isoform-specific primers to Rac1, Rac2 and Rac3 on cDNA derived from Wt, Rac2 -/and Rac1/2 -/-BMM mRNA. PCR was carried out for 30 cycles. (B) As a positive control for mRNA extraction and cDNA synthesis, RT-PCR was performed with primers to CDK4 on BMMderived cDNA. PCR was carried out for 30 cycles. Rac1 and Rac2 in macrophage migration Rac1 and Rac2 are not required for macrophage migration Rac is considered to be essential for cell migration (Etienne-Manneville and Hall, 2002;Ridley et al., 2003). Inhibition of Rac activity in the Bac1 macrophage cell line by microinjecting dominant-negative Rac1 dramatically reduced cell migration speed and consequently chemotaxis towards CSF-1 (Allen et al., 1998). Surprisingly, in murine BMMs Rac1 deletion has no effect on macrophage migration speed or CSF-1-induced chemotaxis (Wells et al., 2004). This suggests that Rac2 could be the major Rac isoform regulating macrophage migration, and indeed Rac2 -/macrophages have reduced transmigration across Transwell filters (Pradip et al., 2003). However, there was no significant difference in the migration speed of Rac2 -/-BMMs compared with Wt BMMs (Table 2, supplementary material Movies 1 and 2). Surprisingly, Rac1/2 -/-BMMs migrated faster than Wt BMMs on tissue culture plastic (Table 2, supplementary material  Movie 3).
Determining the net displacement of cells from their origin provides an indication of their persistence. Wt BMMs tended to shuttle their nucleus back and forth and most do not translocate more than 30 m from their origin (supplementary material Movie 1 and Fig. 5A,B). A higher proportion of Rac1/2 -/-BMMs migrated further from their origin than Wt cells, whereas Rac2 -/-BMMs had a lower displacement (Fig.  5B, Table 2). This indicated that the reduction in nuclear shuttling and/or turning combined with the increase in speed of migration of Rac1/2 -/-BMMs causes them to move at a BMMs were stimulated with CSF-1 and stained as in (B). Cells were considered to be ruffling if more than one dorsal ruffle was present. *P<0.05 and **P<0.01 compared with levels in Wt BMMs, n=30 (Student's t-test). (D) Dorsal ruffling in response to CSF-1 stimulation in Wt, Rac2 -/and Rac1/2 -/-BMMs. BMMs were starved of CSF-1, then stimulated with CSF-1 for 30 minutes, fixed and stained with TRITC-phalloidin to show F-actin. Confocal images were taken through the medial plane of cells to show membrane ruffles; arrows indicate membrane ruffles. higher velocity than Wt BMMs within the same time frame.
To determine the persistence of each population of BMMs, the ratio of net displacement to total path-length was calculated (see Materials and Methods). Interestingly, Rac1/2 -/-BMMs had a higher persistence, indicating that these cells take a more persistent path when they migrate (Fig. 5C). Rac2 -/-BMMs also had a higher persistence, but this probably reflects the fact that a higher proportion of these cells have both a low net displacement and short path length, and thus the persistence of these individual cells is close to 1 (  (Wells and Ridley, 2005), showed that both Rac2 -/and Rac1/2 -/-BMMs were able to migrate towards CSF-1 (n=25 cells) (data not shown).
To investigate the cause of the increased velocity of Rac1/2 -/-BMMs, the mechanism of migration of Rac1/2 -/-BMMs, Rac2 -/-BMMs and Wt BMMs was compared. Wt, Rac2 -/and Rac1/2 -/-BMMs had very different morphologies when migrating ( Fig. 5D-F, supplementary material Movies 1-3). Some Wt BMMs had a classical type of migration cycle , consisting of generation and forward movement of a lamella, movement of the cell body in the direction of the lamella and retraction of the tail (Fig. 5D, movie 1). Many Wt BMMs extended protrusions transiently in alternate directions, and did not translocate far (supplementary material Movie 1). Rac2 -/-BMMs were generally more elongated and their protrusions were not as fully spread as Wt BMM lamellae. Rac2 -/-BMMs then followed a similar pattern to Wt BMMs, moving their nuclei in the direction of the lamellae and then retracting their tails (Fig. 5E, supplementary material Movie 2). The method of Rac1/2 -/-BMM migration was clearly different: Rac1/2 -/-BMMs alternated between a rounded and stellate morphology, where they extended protrusions in several directions (Fig. 5F, supplementary material Movie 3). One or two of these protrusions then persisted and elongated and finally the cell snapped back into a rounded morphology in a new position.

Effect of substrate on macrophage migration speed
To determine whether the substrate affected the relative migration speeds of cells with and without Rac1 and/or Rac2, BMMs were plated on fibronectin, laminin or uncoated glass. To reduce the influence of matrix produced by the BMMs themselves, cell migration was monitored immediately after plating. We have previously shown that BMMs migrate on glass at Journal of Cell Science 119 (13) Wt, Rac1 -/-, Rac2 -/and Rac1/2 -/-BMMs were seeded onto glass in macrophage growth medium. After 24 hours they were fixed and stained with anti-paxillin antibodies and TRITC-phalloidin. A podosome was identified as a focus of F-actin that colocalised with paxillin on the basal surface of a cell (Fig. 4). A cell with one or more podosomes was counted as 'containing podosomes'. Data shown are pooled from six independent experiments (mean ± s.e.m.). *Some Rac1 -/cells only contained F-actin foci without associated paxillin so by this analysis they were considered not to contain podosomes.  approximately half the speed they do on plastic (Wells et al., 2004). Wt BMMs migrated slowest on fibronectin, and fastest on laminin (Table 3). This may be a consequence of differences in adhesion of the cells to each substrate, because they did not spread well on laminin, whereas on fibronectin they were well spread (data not shown). Rac1/2 -/-BMMs had similar migration speeds to Wt BMMs on fibronectin and laminin, although they were slightly slower on glass (Table 3). Interestingly, Rac2 -/-BMMs migrated much slower than Wt BMMs on laminin and slightly slower on glass. Others have shown that Rac2 -/-BMMs have reduced haptotactic migration on vitronectin, which is the main matrix protein that coats glass (Pradip et al., 2003), but migration on laminin has not been reported. We noticed that Rac2 -/-BMMs spread more on laminin than Wt or Rac1/2 -/-BMMs (data not shown), which could explain their reduced migration speed. Loss of Rac2 alone therefore does reduce migration under certain conditions, whereas additional loss of Rac1 compensates for this defect and allows the cells to migrate at similar speed to Wt cells.

Rac1 is required for invasion through Matrigel
Invasion through 3-D extracellular matrix may involve different signalling pathways to those that regulate 2-D migration (Ridley, 2004). To test whether Rac1 or Rac2 was required for macrophage invasion, BMMs were seeded onto Matrigel layered on Transwell filters. After 24 hours, the number of macrophages on the bottom of the Transwell filters was determined. For comparison, migration through uncoated filters was determined. Consistent with previous observations (Pradip et al., 2003), Rac2 -/-BMMs showed reduced migration through uncoated Transwell filters (Fig. 6A), although the migration of Rac1 -/and Rac1/2 -/-BMMs was not impaired. By contrast, invasion of Rac1 -/and Rac1/2 -/but not Rac2 -/-BMMs through Matrigel was reduced compared with Wt BMMs (Fig. 6B). These results clearly show that Rac1 and Rac2 make different contributions to migration, and indicate that Rac1 contributes specifically to invasion, although it does not affect migration speed on 2-D substrates.

Discussion
In this study we characterise the morphology and migratory behaviour of Rac2 -/and Rac1/2 -/macrophages. We show that Rac1 and Rac2 play distinct roles in regulating macrophage morphology, cytoskeletal organization and invasion, but that loss of Rac1 and Rac2 does not prevent BMM migration. Instead, Rac1/2 -/-BMMs can translocate further, faster and more persistently than Wt cells under some conditions. Formation of the characteristic spread lamella, dorsal ruffles and podosomes of BMMs, however, is dependent on Rac expression and requires both Rac1 and Rac2 activity. Interestingly, the morphology of Rac1/2 -/macrophages, with multiple microtubule-rich protrusions, is similar to that of Drosophila SR + cells treated with siRNA to all Rac proteins (Kunda et al., 2003), suggesting that the role of Rac proteins in regulating morphology is conserved in evolution.
Based predominantly on results with dominant-negative Rac mutants, Rac signalling is believed to be essential for regulation of cell migration . Although dominant-negative N17Rac1 inhibits macrophage migration (Allen et al., 1997;Allen et al., 1998), there is no evidence that it specifically inhibits the activity of Rac1 alone. Indeed, N17Rac1 induces similar morphological changes in both Wt and Rac1 -/-BMMs (Wells et al., 2004). Dominant-negative Rac mutants are believed to inhibit Rac activity by sequestering exchange factors (Feig, 1999), but several exchange factors are known to activate more than one Rho GTPase (Bishop and Hall, 2000;Karnoub et al., 2004). It is thus likely that N17Rac1 sequesters exchange factors for other Rho GTPases in addition to Rac1, thereby inhibiting their function as well as Rac1 in macrophages.
One of the major functions of Rac is to induce actin polymerization in lamellipodia and membrane ruffles through WAVE/SCAR and the Arp2/3 complex (Miki et al., 1998). Rac is also involved in the formation and turnover of integrinbased adhesions (del Pozo et al., 2000). Membrane ruffling was decreased in macrophages lacking Rac1 (Allen et al., 1997;Wells et al., 2004) and/or Rac2, indicative of a reduced ability to stimulate actin polymerization at the plasma membrane. Rac1 -/or Rac2 -/macrophages also had a more elongated morphology, probably reflecting a reduced ability to spread on the substratum through impaired adhesion formation and/or lamellipodium extension. Cdc42 may compensate in part for lack of Rac2 activity and contribute to migration, because it activates the Arp2/3 complex through WASPs . Consistent with a role for Cdc42 in BMM migration, dominant-negative Cdc42 reduces dorsal ruffling in Wt and Rac1 -/-BMMs (Wells et al., 2004). Indeed, dominant-negative Cdc42 can decrease migration speed in a variety of cell types, including endothelial cells (Wojciak-Stothard and Ridley, 2003), and increased Cdc42 activity has been proposed to account for the increased migration speed of Rac2 -/hematopoietic stem/progenitor (HSC/P) cells (Yang et al., 2001). Other Rac-or Cdc42-related GTPases, such as RhoG, TCL and TC10, which interact with overlapping downstream targets (Wherlock and Mellor, 2002), could also contribute to migration in the absence of Rac.
Our data together with those from other groups indicate that Rac1 and Rac2 play different roles in regulating cell motility. Rac2 deletion alone results in decreased F-actin levels in neutrophils (Roberts et al., 1999) and macrophages. The fact that Rac2 expression is restricted to haematopoietic cells may reflect the increased need for such cells to move rapidly during inflammatory and immune responses, compared with the relatively slow migration of other cell types such as epithelial cells and fibroblasts during development and wound healing. Rac2 but not Rac1 also appears to be selectively required for superoxide production in neutrophils (Bokoch, 2005;Glogauer et al., 2003). In neutrophils, Rac2 may play a general role in migration by regulating F-actin levels, whereas Rac1 does not affect migration speed but is required for chemotaxis (Sun et al., 2004). However, knockdown of Rac1 by siRNA in fibroblasts, which do not express Rac2, reduced migration speed (Pankov et al., 2005). Sequence divergence between Rac1 and Rac2 in the C-terminal polybasic domain is likely to underlie their distinct functions in leukocytes, in part by affecting their intracellular localization and also their interactions with effectors (Filippi et al., 2004;Yamauchi et al., 2004).
Interestingly, Rac1 and Rac2 had different affects on the assembly of podosomes in macrophages, which are integrinbased adhesions thought to be involved in matrix degradation Journal of Cell Science 119 (13) BMMs were plated in macrophage growth medium onto glass coverslips coated with 10 g/ml fibronectin, 5 g/ml laminin or uncoated. Cell migration was recorded by phase-contrast time-lapse microscopy from 30 minutes to 16 hours after plating. Results shown are mean migration speeds ± s.e.m., n=50 cells for each condition from three independent experiments. *P<0.05 vs values obtained on fibronectin. † P<0.05 vs Wt BMMs on the same substrate, Student's t-test. Fig. 6. Rac1 is required for invasion through Matrigel. BMMs were plated onto Transwell filters (A) or Matrigel-coated Transwell filters (B). The number of cells on the lower side of the filters was counted after 24 hours. Results are from three independent experiments, data are means ± s.e.m. and thereby migration (Linder and Aepfelbacher, 2003). Podosomes are primarily localized to lamellae in polarized macrophages (Evans et al., 2003;Wheeler et al., 2006). Dominant-negative Rac1 has previously been shown to inhibit podosome assembly in dendritic cells (Burns et al., 2001). We show that Rac2 -/-BMMs completely lack podosomes, whereas Rac1 -/-BMMs have F-actin foci that resemble podosomes but have reduced association of paxillin. This decrease in paxillin around F-actin foci in Rac1 -/-BMMs may reflect the fact that the podosomes are less mature; for example podosomes in osteoclasts appear to recruit more paxillin, F-actin and other proteins as they cluster and form rings (Luxenburg et al., 2006). Rac1 could affect paxillin recruitment through its interaction with the PIX-GIT-paxillin complex (Matafora et al., 2001). Rac2, on the other hand, appears essential for the formation of F-actin foci, and may therefore contribute directly to podosome actin polymerization. Podosomes are most abundant in haematopoietic cells that express Rac2, such as dendritic cells and macrophages (Linder and Aepfelbacher, 2003), but can be induced by certain stimuli in cells that probably lack Rac2, such as endothelial cells (Moreau et al., 2003;Tatin et al., 2006), where Cdc42 rather than Rac may be the major regulator of podosome assembly.
Absence of podosomes in Rac2 -/and Rac1/2 -/-BMMs did not affect migration speed on 2-D substrates. Podosomes are thought to be related to invadopodia, which contribute to matrix degradation in invading cancer cells (McNiven et al., 2004;Seals et al., 2005). However, Rac2 -/-BMMs were not impaired in invasion through Matrigel, whereas macrophages lacking Rac1 had decreased invasion. It is possible that Rac1 affects metalloprotease production or activation in macrophages, as has been previously suggested for fibroblastic cells (Kheradmand et al., 1998;Zhuge and Xu, 2001), and that this is responsible for the reduced invasive capacity. In vivo, accumulation of exudate macrophages during peritoneal inflammation was reduced in Rac2 -/mice (Dinauer, 2003;Pradip et al., 2003), and it will be interesting to know which stage of this multi-step process is impaired by lack of Rac2, and whether Rac1 contributes to macrophage recruitment in vivo.
Rac2 -/and Rac1/2 -/-BMM migrate at a similar speed to Wt cells, whereas the migration of Rac2 -/and Rac1/2 -/neutrophils, haematopoietic stem cells and T cells is dramatically impaired (Croker et al., 2002;Dinauer, 2003;Gu and Williams, 2002;Yang et al., 2001). This difference between macrophages and these haematopoietic cell types is likely to reflect the speed of migration: in vitro, neutrophil and T cells migrate up to 20 times faster than macrophages, and thus are likely to be more dependent on rapid Rac-mediated actin polymerization for their migration. However, Rac1/2 -/-BMMs turn less and thus have a greater directional persistence than Wt BMMs. RNAi of Rac1 in fibroblasts also increases directional persistence (Pankov et al., 2005). In BMMs this effect could be due to a decrease in lamellar protrusion: Wt BMMs often reverse their direction by extending a lamella in the opposite direction to their migration path, whereas Rac1/2 -/-BMMs rarely do this. Together our results show that although Rac proteins are not essential for macrophage migration, they play distinct roles in regulating cell shape, cytoskeletal organization, directional persistence and invasion through 3-D matrices.

Reverse transcriptase PCR
To elucidate the expression profile of Rac isoforms in BMMs, reverse transcriptase PCR (RT-PCR) was performed with isoform-specific primers derived from mouse cDNA sequences as described previously (Wells et al., 2004). As a positive control for mRNA extraction and cDNA production, RT-PCR was performed with primers specific to mouse CDK4. In both cases PCR was carried out for 30 cycles.

Quantification of F-actin levels, membrane ruffling, cell adhesive area and elongation ratio
To analyse F-actin levels, BMM were fixed and stained with TRITC-phalloidin for 30 minutes. Digital images of the basal plane of 30 cells from each genotype (ten cells/experiment, three separate experiments) were generated by confocal laserscanning microscopy, using the same gain and offset settings for all sections. The average pixel intensity of each cell was determined using Metamorph 3.1 (Universal Imaging Systems), and the background pixel intensity was subtracted. Membrane ruffles were analysed on the dorsal surface of TRITC-phalloidin-stained cells (Wells et al., 2004).
To analyse adhesive area and elongation of BMMs, digital images of the basal plane of BMMs were acquired by confocal microscopy. Images were pre-processed using Adobe Photoshop 6.0 and ImageJ (NIH) and then quantified using Metamorph 3.1. Briefly, images were passed through a medium filter using a 3ϫ3 kernel to remove interference from background light. Each image was then converted into a binary threshold image, and cell area, the longest chord through this area (length), and the widest chord perpendicular to this chord (width) was measured. The numerical data from analysis of cell area, length and width were exported from Metamorph to Microsoft Excel. The elongation of the cell was calculated by dividing the length by the width.

Analysis of cell migration
For migration on plastic, BMMs (4ϫ10 4 ) were seeded onto 2.5-cm tissue culture plastic dishes (Nunc) and incubated in macrophage growth medium for at least 18 hours and no more than 72 hours at 37°C in 10% CO 2 before time-lapse microscopy. For migration on glass, coverslips were either left uncoated or coated with fibronectin (10 g/ml) or laminin (5 g/ml) for 2 hours, then BMMs (2ϫ10 4 cells/ml) were seeded on each coverslip, and incubated for 30 minutes before timelapse microscopy. During the time-lapse experiments cells were incubated at 37°C in a humidified chamber at 10% CO 2 . For chemotaxis analysis, BMMs were seeded onto glass coverslips, starved of CSF-1 for 18 hours and exposed to a gradient of CSF-1 in a Dunn chemotaxis chamber as previously described (Wells and Ridley, 2005). Phase-contrast micrographs of live cells were taken using a KPM1E/K-S10 Hitachi Denshi 768ϫ576 pixel, 8-bit CCD camera using a 10ϫ Plan/Neofluar objective (Zeiss). Data sets were collected using kinetic imaging motion analysis software (Andor Technology, Belfast, UK). Images were collected every 5 minutes for 6.5 hours. Analysis of cell speed and directionality was then carried out using Mathematica 6.0 (Wolfram Research Institute) workbooks and Microsoft Excel (Allen et al., 1998;Wells et al., 2004). Student's t-tests were used to compare the speeds of migration of different populations of BMM. Statistical significance was accepted for P<0.05.
To provide an indication of BMMs turning during migration, the directional persistence was calculated. The persistence (P) is the ratio of the final displacement of a cell (n) from its origin (displacement) to the total path length of a cell during the time-lapse film: P n = (x ff ,y ffx f1 ,y f1 ) / [(␦x,y f1 ) + (␦x,y f2) +...(␦x,y f3 )] n , where n is an individual cell, x f, y f are coordinates of the cell for a particular frame and x ff ,y ff are the final coordinates of the cell. P n =1 if a cell moves in a straight line and does not turn. P pop = (⌺ Pn f1 +n f2 ..n ff ) / n pop , where pop is the cell population.

Matrigel invasion and Transwell migration assays
Transwell filters (6.5-mm diameter, 5-m pore size) (Corning) were uncoated or coated with 50 l of 1 mg/ml Matrigel (BD Biosciences). BMMs (1ϫ10 5 ) were added to the upper chamber in RPMI 1640 containing 10% FCS, and RPMI 1640 containing 10% FCS and 33 ng/ml CSF-1 was added to the lower chamber. Each assay was performed in triplicate. After 24 hours, cells and Matrigel in the upper chamber were removed, and cells on the bottom of the Transwell filter were fixed and stained with REASTAIN Quick-Diff kit (Reagena). Cells were counted on an epifluorescence microscope.