Activated Macrophage Survival Is Coordinated by TAK1 Binding Proteins

Macrophages play diverse roles in tissue homeostasis and immunity, and canonically activated macrophages are critically associated with acute inflammatory responses. It is known that activated macrophages undergo cell death after transient activation in some settings, and the viability of macrophages impacts on inflammatory status. Here we report that TGFβ- activated kinase (TAK1) activators, TAK1-binding protein 1 (TAB1) and TAK1-binding protein 2 (TAB2), are critical molecules in the regulation of activated macrophage survival. While deletion of Tak1 induced cell death in bone marrow derived macrophages even without activation, Tab1 or Tab2 deletion alone did not profoundly affect survival of naïve macrophages. However, in lipopolysaccharide (LPS)-activated macrophages, even single deletion of Tab1 or Tab2 resulted in macrophage death with both necrotic and apoptotic features. We show that TAB1 and TAB2 were redundantly involved in LPS-induced TAK1 activation in macrophages. These results demonstrate that TAK1 activity is the key to activated macrophage survival. Finally, in an in vivo setting, Tab1 deficiency impaired increase of peritoneal macrophages upon LPS challenge, suggesting that TAK1 complex regulation of macrophages may participate in in vivo macrophage homeostasis. Our results demonstrate that TAB1 and TAB2 are required for activated macrophages, making TAB1 and TAB2 effective targets to control inflammation by modulating macrophage survival.


Introduction
Macrophages are characterized by phagocytic activity, and play diverse roles in different tissue types. While resident macrophages participate in morphogenesis and tissue homeostasis, resident and recruited macrophages also play a major role in acute inflammatory responses [1]. Upon tissue injury or invasion by microorganisms, circulating inflammatory monocytes are recruited and differentiated toward mature macrophages. These macrophages are canonically activated by necrotic debris and bacterial moieties through Toll-like receptor signaling pathway, developing into socalled M1 polarized macrophages [2]. Activated macrophages clean dead cells and microorganisms by phagocytosis and produce inflammatory cytokines resulting in amplification of inflammation. Subsequently, these activated macrophages are deactivated or killed to terminate inflammatory conditions. In some experimental settings, it is known that lipopolysaccharide (LPS)-induced activation of macrophages reduces macrophage viability [3][4][5]. However, the mechanism by which activated macrophages undergo cell death is still largely elusive.
TGFb-activated kinase (TAK1) is a member of the mitogenactivated protein kinase kinase kinase (MAPKKK) family, and is an indispensable intermediate of cytokine and Toll-like receptor pathways [6][7][8]. TAK1 is recruited to and activated by the receptor proximal complex of TNF, IL-1, and Toll-like receptors through a poly-ubiquitin chain-mediated mechanism [9]. TAK1-binding protein 2 (TAB2) and its closely related protein, TAK1binding protein 3 (TAB3), have ubiquitin binding domains and tether between TAK1 and the poly-ubiquitin chain resulting in activation of TAK1 [10][11][12][13][14]. TAB2 and TAB3 may redundantly function in innate immune pathways, but TAB2 plays an indispensable role at least during development [15]. Additionally, it has recently been shown that Tab3 deletion does not impair innate or adaptive immunity [16]. Thus, TAB2 is the major adaptor between TAK1 and activating poly-ubiquitin chains in immune cells. TAK1 is also activated through another binding partner, TAK1-binding protein 1 (TAB1), which is structurally unrelated to TAB2/3 and binds to TAK1 at a site different from the TAB2/3-binding site [17,18]. TAB1 is found to be constantly associated with TAK1, and we recently demonstrated that TAB1 is involved in stress-dependent TAK1 activation [19] and activity of TAK1 in epithelial tissues [20]. Major known downstream molecules of TAK1 are IkB-kinases (IKKs) and mitogen-activated protein kinases (MAPKs) including p38 and JNK, which in turn activate transcription factors NF-kB and AP-1, respectively.
In vivo, TAK1 signaling is found to be important for immune responses in T and B cells through regulation of NF-kB and MAPK pathways in a mouse model [21][22][23][24], which is anticipated from the results in the tissue culture system described above. However, unexpectedly, the most overt phenotype caused by Tak1 deficiency in vivo is tissue damage associated with cell death in the epidermis, intestinal epithelium and liver [25][26][27][28][29][30][31]. Since Tak1 deficiency does not cause cell death in primary culture fibroblasts or keratinocytes, the cell death must be induced depending on the in vivo environment. TAK1 has been found to be integral to prevent tissue-derived TNF-induced cell death in vivo, which is evidenced by the fact that Tnfr1 deletion can rescue cell death and tissue damage in these tissues [26,27,29]. Single deletion of Tab1 or Tab2 does not cause any abnormalities in the epidermis and intestinal epithelium but double deletion of Tab1 and Tab2 phenocopies Tak1 deficiency [20], suggesting that TAB1 and TAB2 redundantly function in TAK1 regulation in these tissues. However, the specific roles of TAB1 and TAB2 in adult tissues are still largely elusive.
Recent studies have demonstrated that Tak1 deficiency in myeloid cells results in hyper-proliferation of neutrophils and increased inflammatory conditions [32,33]. Bone marrow derived macrophages (BMDMs) generated from myeloid-specific Tak1deficient mice have been reported to undergo spontaneous cell death under normal culture conditions [32,34]. These studies have determined that TAK1 is required for proper differentiation of myeloid lineage and survival of macrophage precursors and/or mature macrophages. However, it is not clear whether TAK1 is important for maintenance of progenitors or mature macrophages and what the role of TAK1 is in activation of macrophages. Here we investigate the role of TAK1 and its binding partners, TAB1 and TAB2, in both mature naïve and activated macrophages, and have determined that, TAB1 and TAB2 are essential modulators of TAK1 activity and cell survival in LPS-activated macrophages.

Crystal violet assay
BMDMs were plated onto 12-well plates at a concentration of 2610 5 cells per well and treated with 0.3 mM 4-OHT. In some experiments, BMDMs were subsequently treated with LPS (1 mg/ ml) for 1 and 3 days. Cells were fixed using 10% formalin, and stained with 0.1% crystal violet. The dye was eluted and analyzed at 595 nm.

Flow cytometry
BMDMs were detached from culture dishes and incubated with annexin V-Pacific Blue (BioLegend) and Fixable Viability Dye eFluor 780 (eBiosicence) for cell death analysis. Stained cells were analyzed on flow cytometer (BD Biosciences LSR II), and data were analyzed using FlowJo software (Tree Star). Events were gated to exclude debris based on forward scatter (FSC) and side scatter (SSC) profile, then gated on Pacific Blue (annexin V) or APC-Cy7 (fixable viability dye) when compared to unstained control.

Western blotting
BMDMs were lysed in extraction buffer (20 mM HEPES [pH 7.4], 150 mM NaCl, 12.5 mM b-glycerophosphate, 1.5 mM MgCl 2 , 2 mM EGTA, 10 mM NaF, 2 mM DTT, 1 mM Na 3 VO 4 , 1 mM PMSF, 20 mM aprotinin, 0.5% Triton X-100) in ice for 15 minutes. Cells and debris were then pelleted by centrifugation at 20,000 G for 10 at 4uC. Cell extracts were resolved on SDS-PAGE and transferred to Hybond-P membranes (GE Healthcare). The membranes were immunoblotted with various antibodies, and the bound antibodies were visualized with horseradish peroxidase-conjugated antibodies against rabbit or mouse IgG using the ECL Western blotting system (GE Healthcare).

Mouse model (in vivo)
For in vivo experiments, deletion of Tab1 was achieved in Tab1 iKO mice by intraperitoneal injection of 50 mg per kg tamoxifen on 3 consecutive days. After a period of 3-5 weeks, to reduce effects of Cre toxicity, whole blood was isolated and gene deletion was verified by Western Blot. To exclude the effect of Cre toxicity, we included F+Cre (Rosa26.CreERT2 Tab1 flox/+ ) mice as controls. Mice were intraperitoneally injected with 8 mg/kg LPS, and were euthanized at 72 hours and dissected. Peritoneal leukocytes were collected by peritoneal lavage and collected in phosphate buffered saline (PBS), and splenocytes were harvested and prepared in a single cell suspension. Red blood cells were lysed by suspending cells in 0.83% NH 4 Cl lysis buffer and washed once with PBS. Cells were incubated with anti-CD16/32 in ice for 20 minutes to block FccRII/III, followed by incubation with fluorophore-conjugated monoclonal antibodies (CD11b, B220 CD3e, and F4/80) to evaluate cell type. Cells were washed in PBS and characterized on a BD LSRII flow cytometer (BD Biosciences). Data analysis was performed using FlowJo software (Tree Star).

Statistical analysis
Cell counts were normalized to control and compared using a two-tailed Student's t-test. Values shown are means 6 standard deviation with results considered significant if a probability of Type I error was ,.05.

Deletion of Tak1 or double deletion of Tab1 or Tab2 spontaneously kills bone marrow derived macrophages
We characterized bone marrow derived macrophages (BMDMs) in adult mice having deletions of Tak1, Tab1 or Tab2 single or Tab1 and Tab2 double gene using the ubiquitouslyexpressed inducible Cre recombinase system, Rosa26.CreERT deleter mice. Rosa26.CreERT Tak1 flox/flox (referred to as Tak1 iKO ), Rosa26.CreERT Tab1 flox/flox (referred to as Tab1 iKO ), Rosa26.CreERT Tab2 flox/flox (referred to as Tab2 iKO ), and Rosa26.CreERT Tab1 flox/flox Tab2 flox/flox (referred to as d iKO ) were compared with littermate or age matched controls, no-Cre flox/flox (referred to as WT or control). In some experiments, mice having heterozygous gene deletion, Rosa26.CreERT flox/+ were also used as a control, which did not show any abnormality nor did WT. Bone marrow cells were differentiated to macrophages and gene deletion was subsequently induced by treatment of 4-hydroxytamoxifen (4-OHT) in macrophage culture medium, and vehicle (ethanol) was treated as control. The amounts of TAK1, TAB1 and TAB2 proteins were determined by immunoblotting at 4 days with 4-OHT or vehicle treatment. For Tak1 iKO and d iKO macrophages, we additionally treated with necrostatin-1 (Nec-1), an inhibitor of receptor interacting protein 1 (RIP1), which is known to block Tak1-deficient macrophage death [34]. Nec-1 blocked cell death and allowed us to recover sufficient protein for Western blotting.
We observed fluctuations in the amount of proteins among different litters and with Nec-1 treatment. Nonetheless, TAK1, TAB1, and TAB2 were greatly reduced within 4 days with 4-OHT treatment in Tak1 iKO macrophages ( Figure 1A). Reduction of TAB1 and TAB2 in Tak1 iKO macrophages is presumably due to destabilization of unbound TAB1 and TAB2. In contrast, TAB1 or TAB2 but not TAK1 was reduced in Tab1 iKO or Tab2 iKO macrophages, respectively. Both TAB1 and TAB2 but not TAK1 were reduced in d iKO macrophages. These indicate that TAK1 is intact in Tab1 iKO , Tab2 iKO , and d iKO macrophages, which allow us to investigate the specific roles of TAB1 and TAB2 without altering the protein level of TAK1 in these macrophages. The number of macrophages declined at 8 days with 4-OHT treatment in Tak1 iKO macrophages, whereas the number of control macrophages, including WT with 4-OHT and Tak1 iKO with vehicle treatment, were not altered during the period of experiment ( Figure 1B). Thus, TAK1 is essential for cultured macrophage integrity. These results are consistent with the previously reported results in myeloid-specific deletion of Tak1 [34]. Thus, TAK1 seems to be important for survival of not only precursors but also mature naïve macrophages. In contrast to Tak1 deletion, either Tab1 or Tab2 single deletion only moderately or slightly decreased the number of macrophages ( Figure 1B). This indicates that TAB1 or TAB2 may partly contribute to but is not required for naïve macrophage survival. Importantly, we found that double deletion of Tab1 and Tab2 reduced the number of macrophages to a level similar to Tak1 deletion ( Figure 1B). Collectively, these results suggest that TAB1 and TAB2 may redundantly function to maintain TAK1 activity in naïve macrophages and TAK1 basal activity is required for survival of naïve macrophages.

TAB1 and TAB2 are required for survival of LPSstimulated macrophages
To determine the role of TAK1 complex in activated macrophages, we treated Tab1 iKO or Tab2 iKO macrophages with bacterial moiety, lipopolysaccharide (LPS). Wild type macrophages treated with LPS did not exhibit reduced cell viability in our experimental setting ( Figure S1). Even in this condition, we found  Figure 2C). Apoptotic cells characterized by annexin V binding-positive but permeability dye-negative were also increased by LPS treatments ( Figure 2D). Thus, LPS-activated macrophages require TAB1 for their survival, and Tab1 deficiency causes cell death having both necrotic and apoptotic features. Similarly, we found that Tab2 iKO macrophages were significantly declined upon LPS treatment ( Figure 3A). In contrast to Tab1 deletion, Tab2 iKO macrophage underwent cell death only one day after LPS treatment. LPS-treatment of Tab2 iKO macrophages increased the frequency of annexin V-positive and cell permeability dye-positive population ( Figure 3B). Similar to Tab1 iKO , both necrotic and apoptotic cells were increased by LPS treatment in Tab2 iKO macrophages ( Figure 3C and 3D). These results demonstrate that, in contrast to naïve macrophages, both TAB1 and TAB2 are important for activated macrophage survival.  We hypothesize that TAB1 and TAB2 mediate LPS-induced TAK1 activation, which may be required for LPS-activated macrophage survival. To test this, we examined the levels of TAK1 activation and subsequent downstream events, activation of NF-kB and p38, following LPS stimulation. We note that, since Tak1 or Tab1 and Tab2 double deletion spontaneously kills macrophages, we treated macrophages with Nec-1 to obtain live macrophages with Tak1 or Tab1 and Tab2 double deletion. It is known that while RIP1 participates in NF-kB and p38 pathways, RIP1 catalytic activity is dispensable [39]. Consistent with this notion, NF-kB and p38 were activated in wild type macrophages even with Nec-1 treatment. We found that LPS activated TAK1 and its downstream IKK and p38 in macrophages ( Figure 4A, middle 4 lanes), and Tak1 deficiency reduced activation of both IKK and p38 ( Figure 4A, left 4 lanes). TAK1 activity was monitored by phosphorylation of Thr 187 ( Figure 4A, top panel), which is known to be associated with activation of TAK1 [40] However, non-specific bands were detected around the phosphorylated TAK1 in macrophages protein extracts, which were seen even in unstimulated macrophagse ( Figure 4A, asterisks). Thus, we also utilized retardation of TAK1 band on SDS-PAGE to monitor TAK1 activation, which is caused by phosphorylation of several sites associated with TAK1 activation [41,42]. While TAK1 then treated with 1 mg/ml LPS for the indicated period of time. Whole cell extracts were analyzed by Western blotting using the indicated antibodies. Asterisks indicate non-specific bands. (C) Tab2 iKO or control BMDMs were cultured with 0.3 mM 4-OHT for 3 days then treated with 1 mg/ml LPS for the indicated period of time. Whole cell extracts were analyzed by Western blotting using the indicated antibodies. Asterisks indicate non-specific bands. doi:10.1371/journal.pone.0094982.g004  Figure 4A, second panel), and reduced activation of IKK and p38 ( Figure 4A, right 3 lanes) suggesting that LPSinduced activation of IKK and p38 is largely mediated by TAK1 and that TAB1 and TAB2 are essential for TAK1 activation in response to LPS. Deletion of either Tab1 or Tab2 impaired LPSinduced migration shift of TAK1, suggesting some impairment of TAK1 activation in response to LPS (Figures 4B and C, top  panels). However, a single deletion of either Tab1 or Tab2 had a marginal effect on the LPS-induced degradation of IkB and phosphorylation and p38, suggesting that one of these proteins can activate TAK1 sufficiently at least in the pathways leading to activation of IKK and p38. These results collectively suggest that LPS-induced activation of TAK1 is mediated by TAB1 and TAB2, and that either deletion of Tab1 or Tab2 reduces TAK1 activation. However, LPS-induced activation of NF-kB and p38 seems to require at a minimum TAB1 or TAB2.

LPS-induced cell death in Tab1-or Tab2-deficient macrophages is partially rescued by inhibition of RIP1
Tak1 deletion is reported to cause RIP1-dependent cell death upon TNF treatment in several cell types [26,27,29]. Earlier study demonstrates that Tak1-deficient naïve macrophages die in a RIP1-dependent manner [34]. We hypothesize that activated macrophages die with the mechanism similar to that in naïve macrophages due to insufficient activity of TAK1. To test this, we examined the involvement of RIP1 kinase activity in Tab1 iKO and Tab2 iKO macrophage death by using Nec-1. In Tab1 iKO macrophages, Nec-1 significantly and marginally reduced LPS-induced necrotic and apoptotic cells, respectively ( Figure 5A and 5B). Nec-1 treatment exhibited some trends of reduction of cell death in Tab2 iKO macrophages, although these did not reach to a statistic significance ( Figure 5C and 5D). Both necrotic and apoptotic cells were decreased. These results suggest that RIP1 kinase activity may be at least partially involved in LPS-induced cell death in Tab1 iKO and Tab2 iKO macrophages. Figure 6. LPS treated Tab1 iKO macrophages are reduced in vivo. (A) Tab1 iKO or control including Tab1 F+ Cre mice were intraperitoneally injected with 50 mg/kg tamoxifen for 3 consecutive days. After 3-5 weeks, peripheral blood was collected and leukocyte extracts were tested for Tab1 deletion by Western blotting. (B) Tab1 iKO (n = 6) and control mice (n = 9) were intraperitoneally injected with 8 mg/kg LPS. Peritoneal leukocytes were collected at 72 hours and stained with fluorophore-conjugated antibodies. Shown is percent positive, excluding dead cells and debris, for the indicated markers. Percentages of CD11b+ F4/80+, CD11b+, CD3e+ or B220+ cells of total cells +/2 SD is shown. doi:10.1371/journal.pone.0094982.g006

Tab1-deficient macrophages were reduced upon LPS injection in vivo
We next examined the role of TAK1 complex in macrophages in vivo. We anticipate LPS injection to recruit and activate monocytes from the bone marrow and circulation to the peritoneal cavity. Germline deletion of Tak1 gene causes multiple defects in embryogenesis [43,44], and deficiency of Tak1 in adult mice also causes acute severe liver injury and mortality [25,26]. We were not able to analyze Tak1 deficient macrophages in adult mice due to these multiple acute effects. Tab2 iKO mice did not exhibit overt abnormality upon gene deletion by tamoxifen injection; however, LPS stimulation caused acute liver dysfunction within 6 h (unpublished observations). Tab2-deficient macrophages were difficult to analyze in LPS-challenged mice for this reason. We then focused on Tab1 iKO mice. Tab1 gene was deleted in Tab1 iKO mice by tamoxifen injection for 3 consecutive days. Although germline deletion of Tab1 causes embryonic lethal phenotype [35,45], we found that deletion of Tab1 gene in adult mice did not cause overt abnormalities even though TAB1 proteins were greatly diminished ( Figure 6A). To rule out the effect of Cre toxicity described previously [46,47], the experiments were performed after more than 3 weeks post-tamoxifen injections. The number of peritoneal macrophages was determined in Tab1-deficient adult mice following LPS injection ( Figure 6B). The frequency of CD11b+ F4/80+ macrophages in peritoneal fluid was lower in Tab1 iKO mice compared to littermate controls following LPS stimulation. T-cells, B-cells and CD11b single positive cells were not significantly changed by Tab1 deficiency. Tab1 deficiency alone did not cause changes in macrophage population ( Figure  S2A). Importantly, the frequency of CD11b+ F4/80+ macro-phages in peritoneal fluid was increased by LPS to around 30% from 10-15% under unstimulated conditions in control mice, but such increase was not observed in Tab1 iKO mice. We noted that splenic CD11b+ F4/80+ macrophages were not significantly altered by LPS stimulation, and Tab1 deletion did not cause any alteration of splenic macrophages ( Figure S2B), supporting the notion that participation of TAB1-dependent pathway in macrophage survival varies depending on macrophage type or activation state [48]. These suggest that TAB1 is important for activated macrophage maintenance in vivo.

Discussion
Our current studies identify an essential role for TAK1 modulator proteins, TAB1 and TAB2, in LPS-activated macrophages, in BMDMs, and in a mouse model of inflammation (Figure 7). The roles of TAB1 and TAB2 in macrophages have not been well characterized in the published literature to date. We demonstrate that TAB1 and TAB2 are essential for LPS-activated macrophage survival but are dispensable for survival of naïve macrophages. Bacterial moieties including LPS are strong activators of macrophages, which polarize macrophages toward an inflammatory, M1 phenotype. M1 macrophages mediate acute inflammation by secreting cytokines and chemokines, which play a major role in initiation of inflammatory responses [49]. These classically activated macrophages are prone to cell death, which contributes to the termination of inflammatory responses [3][4][5]. Chronically activated macrophages are associated with many disease conditions including obesity, autoimmune diseases, atherosclerosis and asthma [50]. In contrast, naïve macrophages in tissues, so-called resident macrophages such as microglia, Kupffer cells and Langerhans cells, play indispensable roles during development and in tissue repair [51][52][53]. Resident unstimulated macrophages are also important for prompt responses to protect tissues from insults. Therefore, the presence of unstimulated resident macrophages is beneficial to maintain tissue integrity. Thus, limiting inflammatory conditions by controlling only an activated sub-set of macrophages could be a useful tool in treating disease. Our results demonstrate that deletion of either Tab1 or Tab2 effectively kills only LPS-activated macrophages in vitro, and that Tab1 deletion prevents increase of peritoneal macrophages upon LPS stimulation. These results indicate that TAB1 and TAB2 are potentially useful targets to selectively control the activated fraction of macrophages.
It is interesting that the Tab1 Tab2 diKO shows diminished activation of IKK and p38 following LPS stimulation, which is similar to the Tak1 iKO phenotype. When these genes are individually deleted, LPS-induced IKK and p38 activation are nearly normal and naïve macrophages persist. However, macrophages having deletion of either Tab1 or Tab2 could not withstand LPS-activation and Tab1 iKO macrophages underwent necrosis after 3-4 days, while Tab2 iKO die after 1 day, despite showing early LPSinduced TAK1 activation at or near WT levels ( Figures 4B and C). In light of the nearly normal activation of NF-kB and p38 in Tab1 iKO and Tab2 iKO in response to LPS treatment, TAB1-and TAB2-dependent activated macrophage survival may occur through signaling pathway(s) independently of NF-kB and p38, which is in contrast to previous studies [4,5].
Recapitulating our in vitro results, we found that the number of peritoneal macrophages was lesser in LPS-treated Tab1 iKO mice when compared to control. Based on our in vitro data, we anticipate this disease model to produce a net increase in LPSactivated peritoneal macrophages in control but not Tab1 iKO mice, however it cannot be conclusively excluded that the observed reduction is due to a defect in recruitment or expansion independent of cell death. Importantly, other hematopoietic cell types, including T cells, B cells and granulocytes, were found to be unaffected, suggesting TAB1-dependent survival signaling that is potentially unique to macrophages. We note here that liver was found to be undamaged in Tab1-deficient mice upon LPS injection under our experimental conditions, although higher doses of LPS are known to cause liver damage. Inferences based on our studies are limited by our mouse model, which has Tab1 deleted in all cells, such that one cannot rule out the effects from other cell types in vivo. Future studies analyzing macrophage-specific conditional knockout mice for Tab1 and Tab2 could bring insight to the roles of these genes in macrophage cell death. Further research focused on this important mechanism in macrophages could inform inflammatory disease models, particularly diseases in which microorganisms target macrophages. The data supporting that TAB1 is essential for macrophage survival in LPS-treated mice could be used to improve our understanding of the control of inflammation. Figure S1 Wild-type macrophages treated with LPS do not have reduced viability under experimental conditions. Viability of LPStreated control macrophages. Tab2 flox/flox BMDMs were cultured for 8 days with 0.3 mM 4-OHT followed by 3 days 1 mg/ml LPS, and viability was measured by Crystal Violet Assay. Shown are mean percentages of attached macrophages compared to 8 days treated with vehicle +/2 SD of 3 independently performed experiments. (TIF) Figure S2 TAB1-dependent survival depends on type of macrophage. (A) Peritoneal leukocytes were collected from Tab1deficient mice treated with vehicle (PBS) at 72 hours and stained with fluorophore-conjugated antibodies. Shown is percent positive of 2 control and 3 Tab1 iKO for the indicated markers. Percentages of CD11b+ F4/80+, CD11b+, CD3e+ or B220+ cells of total cells 6SD is shown. (B) Tab1 iKO and control mice were intraperitoneally injected with 8 mg/kg LPS. Splenocytes were collected and stained with fluorophore-conjugated antibodies. Shown is percent positive of 6 control and 4 Tab1 iKO for CD11b+ F4/80+, CD11b+, CD3e+ or B220+ as a percentage of total cells 6SD. (TIF)