FLIP: a novel regulator of macrophage differentiation and granulocyte homeostasis

Cell surface death receptors, including Fas, TRAIL and TNFR1, are capable of mediating programmed cell death through the activation of caspases 8 or 10. Cellular Flice-like inhibitory protein (FLIP) is critical in the protection against death receptor-mediated apoptosis.^1-3  After death receptor ligation caspases 8/10 are recruited to the receptor through adaptor molecules, FADD or TRADD,^4-6  resulting in homotypic interactions and autocatalytic caspase activation, which is suppressed by FLIP.⁷  FLIP is highly expressed in a variety of tumors, and the forced reduction of FLIP is capable of sensitizing certain tumors to death receptor-mediated apoptosis.^8-10  Our earlier studies with human monocytes demonstrated that FLIP was induced during monocyte to macrophage differentiation, which protected macrophages from Fas-FasL mediated apoptosis.¹¹  In addition, FLIP is highly expressed in the macrophages within the synovial tissue of patients with rheumatoid arthritis, and the forced reduction of FLIP sensitizes macrophages to Fas-mediated apoptosis.^11,12 

In addition to protecting against apoptosis, a variety of other functions of FLIP have been characterized. At low levels compared with caspase 8, FLIP may be proapoptotic by contributing to caspase 8 activation, promoting Fas-mediated apoptosis.¹³  The role of FLIP in T lymphocytes is complex, because it is necessary for the maturation of thymocytes by suppressing apoptosis¹⁴  and is capable of protecting against the extrinsic cell death of mature lymphocytes, but may also promote apoptosis and the activation of NF-κB (reviewed in Krammer et al¹⁵ ). In dendritic cells an N-terminal p22 fragment of FLIP induces NF-κB activation.¹⁶  The deletion of FLIP in B lymphocytes results in decreased B lymphocytes and increased sensitivity to Fas-mediated apoptosis and the increased activation of JNK and p38.¹⁷  Other nonapoptotic functions of FLIP include the suppression of TNFα-induced JNK activation¹⁸  and the impairment of the ubiquitin-proteasome system, resulting in the increased expression of genes regulated by β-catenin.¹⁹  However, the potential nonapoptotic functions of FLIP in myeloid cells are unknown.

To elucidate the in vivo role of FLIP in macrophages and neutrophils, we developed mice deficient in Flip in myeloid cells. These mice had significantly reduced body weight and died prematurely. There was a dramatic increase in circulating neutrophils and monocytes (> 10-fold), which was accompanied by multiorgan neutrophil infiltration. Isolated neutrophils deficient in Flip demonstrated a mature morphology and underwent normal apoptosis, oxidative burst, and degranulation. In contrast to monocytes, macrophages were greatly reduced in the spleen, lymph nodes, and peritoneal cavity. In ex vivo cultures, the deletion of Flip prevented macrophage differentiation, and mixed bone marrow chimeric mice failed to recapitulate the phenotype of the mice deficient in Flip in myeloid cells. Together, these observations identify a novel role for FLIP in myeloid homeostasis.

The Flip targeting vector, containing 3 loxP sequences inserted in tandem orientation and a neomycin resistance (neo^R) cassette, was transfected into the B6/Blu ES cells (ES Cell Core at Washington University, St Louis, MO). After removing the neo^R cassette, chimeras were generated by blastocyst injection and embryos reimplantation in C57BL/6 mouse establishing floxed Flip (Flip^f/+) mice, which were bred with the Zp3-cre mice to generate mice with one allele of Flip deleted (Flip^d/+). After further crosses, myeloid cells deficient in FLIP were generated by crossing with LysM^c/+ mice (Jackson Laboratory,²⁰ ), resulting in the genotypes identified in Table 1. Mating Flip^f/+, LysM^c/+ and Flip^+/d, LysM^c/+ mice resulted in Flip^f/d, LysM^c/+ mice, which were deficient in FLIP in myeloid cells.

Complete blood counts and differentials were performed using a Hemavet 950 (Drew Scientific).²¹  FLIP expression was determined by immunoblotting with an antibody to the N-terminal peptide of FLIP (Cell Signaling). For cultivation of microorganisms, tissues was collected under sterile conditions and cultured in Brain Heart Infusion Broth (Remel) at 37°C overnight. Circulating cytokines/chemokines were quantified by the Premixed 32 Plex kit (Millipore), and the data were acquired by LUMINEX 200.

Organs were dissected and fixed in 10% neutral formalin, embedded in paraffin, and then 5-μm sections were stained with hematoxylin and eosin. Cryosections were processed for immunofluorescence staining of macrophages as described,²²  then incubated with primary antibodies to mouse F4/80 (B8; eBioscience), CD169 (AbD Serotec), or isotype control immunoglobulin G at 4°C overnight, and subsequently with AlexaFluor 594 conjugated anti-rat secondary antibody (Invitrogen) for 2 hours and counterstained with DAPI (4′,6-diamidino-2-phenylindole).

Cells were preincubated with anti-mouse CD16/CD32 antibody (BD Phamingen) to block cell surface Fc III/II receptors before antibody staining. Immunophenotyping was performed by multicolor fluorochrome-conjugated antibody cocktails, including antibodies to CD 45, CD11b, F4/80, Gr 1, CD115, CD62L, CD4, CD8, CD3, CD19, B220, and CD11c (eBioscience or BD Pharmingen). Data were acquired on a BD LSR II flow cytometer (BD FACSDIVA software) and analyzed by FlowJo Version 8.8.4 (TreeStar). Single cell populations were sorted by Moflo High Speed Sorter.

C57BL/6 recipients CD45.1⁺ congenic (B6.SJL-Ptprc^a Pepc^b/BoyJ, 6-week-old females; The Jackson Laboratory) were lethally irradiated (1000 rads), followed by the retroorbital administration of 5 × 10⁶ donor whole bone marrow cells. The donor bone marrow was collected from CD45.1⁺ (wild-type) and CD45.2⁺ (Flip^f/d, LysM^c/+) mice. The bone marrow was transferred to the recipients containing 0%, 25%, 50%, 75%, and 100% of cells from Flip^f/d, LysM^c/+ mice. Recipients received sulfamethoxazole (50 mg/mL) and trimethoprim (8 mg/mL) in the drinking water and were subjected to phenotypic analysis 8 weeks posttransplantation.

Mitochondrial transmembrane potential (Δψ_μ) was assessed by retention of Rh123 as described.²³  Cell membrane integrity was assessed by exclusion of DAPI. Apoptosis was assessed by detecting the expression of annexin V (BD Pharmingen).^11,23 

Mouse peripheral blood cells were incubated with dihydrorhodamine 123 (0.5μM), with or without phorbol 12-myristate 13-acetate (PMA; 10nM) at 37°C for 30 minutes. Thereafter, the samples were chilled on ice and incubated with fluorochrome-conjugated anti-CD11b and -Gr1, followed by erythrocyte lysis. Oxidative burst was determined by measuring the conversion of nonfluorescent dihydrorhodamine 123 into fluorescent rhodamine123²⁴  and degranulation by the increased expression of cell surface CD11b.²⁴  Myeloperoxidase staining was performed by incubating slides in a solution of 3.3′-diaminobenzidine and hydrogen peroxide.

Spleen (1 × 10⁴) or bone marrow cells (0.5 × 10⁵), combined from femurs and tibias, were seeded in methylcellulose media (Methocult: StemCell Technologies) in the presence of interleukin-3 (IL-3; 10 ng/mL), IL-6 (10 ng/mL), stem cell factor (50 ng/mL), and granulocyte macrophage colony-stimulating factor (GM-CSF; 10 ng/mL) for 7 days. Colony-forming unit of granulocyte and macrophage (CFU-GM) colonies were counted under light microscropy. For megakaryocyte-CFUs (CFU-Mk), cells (1 × 10⁵) were seed in a collagen-based media (Megacult-C; StemCell Technologies) supplemented with IL-3 (10 ng/mL), IL-6 (10 ng/mL), IL-11 (10 ng/mL), thrombopoietin (10 ng/mL) for 7 days. The Megacult-C cultures were dehydrated and stained for acetylcholinesterase activity to determine the CFU-MK colonies.²⁵ 

CD117⁺ (c-Kit) hematopoietic stem cells from the bone marrow of 8- to 12-week-old Flip^f/+ or Flip^f/d mice were isolated using antibody to the stem cell receptor CD117 (StemCell Technologies) according to the manufacturer's instructions. These freshly isolated cKit⁺ cells were cultured in the presence of stem cell factor (10 ng/mL), IL-3 (10 ng/mL), and IL-6 (10 ng/mL) overnight, followed by culture with recombinant retroviral vectors expressing green fluorescent protein (GFP) alone or GFP plus Cre for an additional 24 hours.²⁵  The cells were then incubated with 20% L929 conditioned medium containing macrophage colony-stimulating factor (M-CSF) to allow in vitro macrophage differentiation for 5 days.

All quantitative data are presented as mean ± SEM. Statistical analysis between groups was done with 2-tailed Student t test. Nonparametric data were analyzed by the Mann-Whitney rank sum test. One-way analysis of variance (Tukey pairwise mean comparison) was performed for multigroup analysis. Significance levels were set at .05.

To generate mice with Flip deleted in myeloid cells, loxP sequences were inserted flanking exons 2 and 3 of Flip (Figure 1A), which was used to generate Flip floxed (Flip^f) embryonic stem cells (supplemental Figure 1A, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Deletion of one floxed allele in the whole genome (Flip^d/+) was obtained by crossing heterozygous Flip^f/+ mice with Zp3-Cre mice that express Cre in the female germ line (supplemental Figure 1B). Myeloid linage-specific deletion of the flip^f was carried out by crossing Flip^f/+ or Flip^f/d mice with those expressing Cre driven by the LysM promoter (LysM^c/+; Figure 1B). When Flip^f/+ mice were crossed with the LysM^c/+ mice, the floxed allele was deleted in granulocytes and monocytes, demonstrating the efficiency of Cre in these cells. Crossing Flip^f/d mice with LysM^c/+ mice resulted in the deletion of both alleles of Flip in bone marrow and peripheral blood granulocytes (Figure 1B). However, complete deletion of the floxed Flip allele was not observed in monocyte bone marrow precursors or peripheral blood monocytes (Figure 1B). As expected, neither allele of Flip was deleted in circulating T or B lymphocytes when Flip^f/d or Flip^f/+ mice were crossed with LysM^c/+ mice (Figure 1B).

The deletion of Flip in myeloid cells resulted in the expected Mendalian birth rate with no indication of embryonic lethality (data not shown). However, the Flip^f/d, LysM^c/+ mice displayed a runted appearance and a 50% reduction in size (Figure 1C). None of the littermates of the Flip^f/d, LysM^c/+ mice (defined in Table 1) demonstrated reduced size (Figure 1C) or increased mortality (data not shown) and were therefore collectively used as the littermate controls in further experiments. Approximately 50% of the Flip^f/d, LysM^c/+ died before 32 weeks of age, whereas > 80% of Flip^f/d, LysM^c/c mice died by 3-5 weeks of age, most likely because of a more complete deletion of Flip (Figure 1D), and not because of the deletion of both LysM alleles, because there was no effect on phenotype, including body weight (Figure 1C) or blood count (Table 1), when one or both alleles of Flip were still present. These observations demonstrate that the reduction of FLIP in myeloid cells is critical for postnatal survival.

Examination of peripheral blood demonstrated a 5-fold increase of leukocytes in the Flip^f/d, LysM^c/+ mice compared with the littermate controls (58.8 ± 5.9 vs. 11.0 ± 0.7 × 10³/μL, P < .001). Circulating neutrophils and monocytes were increased > 10-fold compared with the controls (P < .001; Figure 2A and Table 1). Eosinophila, thrombocytosis, and anemia were also present in the Flip^f/d, LysM^c/+ mice, but not in the littermate controls (Table 1). Examination of stained peripheral blood smears from the Flip^f/d, LysM^c/+ mice demonstrated mature neutrophils with appropriate nuclear morphology and cytoplasmic granularity. The neutrophils from the Flip^f/d, LysM^c/+ mice were CD45⁺, CD11b⁺, CD115⁻, Gr1^hi, CD62L^hi, similar to the littermate controls (Figure 2C). The circulating monocytes appeared normal by morphology; however, the Flip^f/d, LysM^c/+ mice demonstrated an expanded population of monocytes that was CD45⁺, CD11b⁺, CD115⁺, Gr1⁺, CD62L⁺, identifying them as inflammatory monocytes (Figure 2C). In contrast, the expected distribution of resident (Gr-1⁻, CD62L⁻) and inflammatory (Gr-1⁺, CD62L⁺) monocytes were observed in the controls (Figure 2C). However, although the absolute number of inflammatory monocytes was increased in the Flip^f/d, LysM^c/+ mice, the number of resident monocytes was not different from the controls (Figure 2D). Analysis of peripheral blood also demonstrated a decreased percentage of B cells and CD4⁺ and CD8⁺ T cells, which was due to the marked increase of granulocytes. However, the absolute number of CD8⁺ cells was actually increased in the Flip^f/d, LysM^c/+ mice, whereas B cells and CD4⁺ lymphocytes were not different from the controls (supplemental Figure 2A-C).

Because of the changes in leukocytes, lymphoid organs were subjected to pathologic analysis. A dramatic infiltration of neutrophils was observed not only in the bone marrow, but also in spleen, lymph nodes, and thymus (Figure 3). Within the spleen and lymph nodes, the architecture was disrupted, with loss of lymphoid follicles. The thymus demonstrated loss of a distinct medulla and cortex, an increase of neutrophils, and a reduction of thymocytes (Figure 3). Morphologically, the bone marrow revealed an expansion of neutrophils and a reduction of erythropoiesis, without morphologic evidence of myeloid dysplasia (Figure 3). No histologic abnormalities of bone were noted in the Flip-deficient mice. The spleen demonstrated a reduction in the percentage of B lymphocytes, and of CD8⁺ and CD4⁺ lymphocytes in the Flip^f/d, LysM^c/+ mice, although only the absolute number of B cells was reduced (P < .01; supplemental Figure 3). By immunophenotyping, a significant (P < .05-.01) increase in the absolute number of Gr-1⁺ neutrophils was identified in the spleen and lymph nodes, but not the bone marrow or thymus, even though the percentage of neutrophils was increased in each of the organs, from the Flip^f/d, LysM^c/+ mice (supplemental Figure 4). Nonlymphoid tissues were also affected. Accumulation of neutrophils was demonstrated in the lungs in the interalveolar spaces and around large blood vessels and in the small intestine, especially the jejunum (supplemental Figure 5). Neutrophilic infiltration was less severe in the liver and heart (supplemental Figure 5), as well as the stomach, large intestine, kidney, and brain (data not shown). Overall, pathologic analysis demonstrated an expansion and a multiorgan infiltration of neutrophils.

To determine whether the increased neutrophils might be due to an underlying infection, organs from Flip^f/d, LysM^c/+ mice were cultured for microorganisms. Cultures of the spleen, liver, lymph nodes, or peritoneal fluid were positive for bacteria in 30% of the mice, whereas none of the littermate controls was positive. There was no difference in the neutrophil counts between the Flip^f/d, LysM^c/+ mice that were culture positive and negative, although the culture positive mice were smaller (< 10 g) compared with those that were culture negative (supplemental Figure 6A-B). The cultures revealed several commensal intestinal organisms including Streptococcus sanguins, Enterococcus faecalis, Proteus mirabilis, and Escherichia coli. Based on susceptibility testing, Flip^f/d, LysM^c/+ mice were treated with antibiotics using ampicillin in drinking water and/or gentamicin subcutaneously. Treatment reduced the positive cultures by 50% but there was no reduction of circulating neutrophils (supplemental Figure 6C) and no increase of weight. These results suggest that infection was not the cause of the neutrophil expansion or the increased mortality.

Because of the increased numbers of neutrophils and infections, studies were performed to determine whether there was an abnormality of survival or function of neutrophils. Control peripheral blood neutrophils exhibited spontaneous apoptosis with the loss of mitochondrial transmembrane potential and uptake of DAPI that was limited immediately after isolation and increased at 4 hours, with the marked loss of membrane integrity (DAPI⁺) by 24 hours (Figure 4A). The percentage of cells that had lost mitochondrial transmembrane potential or membrane integrity at 0 and 4 hours was somewhat less in the Flip^f/d, LysM^c/+ mice (Figure 4A). However, neutrophil cell death was essentially complete by 24 hours using circulating granulocytes from the Flip^f/d, LysM^c/+ mice, similar to the controls (Figure 4B). Because the Flip^f/d, LysM^c/+ mice experienced infections, studies were performed to examine neutrophil function. Peripheral blood neutrophils from Flip^f/d, LysM^c/+ mice demonstrated normal oxidative burst (Figure 4C) and degranulation after activation with PMA (Figure 4D). In response to the TLR4 ligand lipopolysaccharide, after 1 hour no neutrophil oxidative burst was detected, although degranulation was observed and was comparable in the Flip-deficient and control mice (data not shown). In addition, myeloperoxidase activity of peripheral blood neutrophils in the Flip^f/d, LysM^c/+ mice was comparable to the controls (Figure 4E). These observations do not suggest that an intrinsic abnormality of apoptosis or function was responsible for the neutrophilia or infection.

Because there were increased granulocytes, monocytes, and platelets, studies were performed to determine the effect of Flip deletion on myelopoiesis. Examination of the bone marrow revealed no difference of GM-CFU between the Flip^f/d, LysM^c/+, and the littermate controls (Table 2). In contrast, using cells from the spleen, a significant (P < .01) increase of CFU-GM was observed in the Flip^f/d, LysM^c/+ mice (Table 2). The changes were not restricted to the GM, because the CFU-Mk were also increased (P < .001) in the spleens of the Flip^f/d, LysM^c/+ mice, but not the bone marrow (Table 2). Serum from the peripheral blood was used to determine whether growth factors might contribute to the extramedullary hematopoiesis. G-CSF, GM-CSF, M-CSF, IL-6, and IL-17 were all increased in the Flip^f/d, LysM^c/+ compared with the control mice (Table 3). These observations document a marked increase of splenic extramedullary hematopoiesis, which was associated with an increase of myeloid growth factors.

Splenomegaly (Figure 5A) with loss of lymphoid follicles and infiltration with granulocytes (Figure 3) was observed in the Flip^f/d, LysM^c/+ mice. However, the total number of cells in the Flip^f/d, LysM^c/+ and control spleens was not different (Figure 5B). This is most likely due to an increase of immature and myeloid cells that are larger than lymphoid cells, which were reduced. An increased size of cells in the spleens for the Flip^f/d, LysM^c/+ mice was documented by the right shift in forward scatter determined by flow cytometry (Figure 5B). No RBC sequestration was apparent (Figure 3). Immunofluorescence staining revealed a marked reduction of macrophages in both the red pulp (anti-F4/80; Figure 5C) and the marginal zone (anti-CD169; Figure 5D) in the Flip^f/d, LysM^c/+ mice. Examination of lymph nodes also showed a marked reduction of F4/80⁺ macrophages (Figure 5E). In contrast there was no reduction of macrophages in the liver (data not shown). By immunoblot analysis, FLIP was reduced but still present in lysates of the spleen (Figure 5F) and lymph nodes (Figure 5G) of the Flip^f/d, LysM^c/+ mice, most likely because of a marked increase of neutrophils that do not express FLIP and a marked reduction of B cells in spleen (supplemental Figure 3). Therefore, the deletion of Flip in myeloid cells results in a reduction of macrophages in the spleen and lymph nodes, with the loss of lymphoid follicles.

F4/80⁺, Gr1⁻ resident macrophages were abundant in the peritoneal cavity of the littermate control mice expressing FLIP (Figure 6A). However, in the Flip^f/d, LysM^c/+ mice, the increased total cell count (Figure 6A-B) was due to a marked (P < .01) increase of Gr1⁺, F4/80⁻ neutrophils, whereas F4/80⁺,Gr1⁻ macrophages were significantly (P < .001) reduced (Figure 6A-B). There was a highly significant inverse relationship between the percentage of macrophages and granulocytes (R = -0.75, P < .001), that was more clearly described by the trend line (Figure 6C). Analysis of the peritoneal space of wild-type and the Flip^f/d, LysM^c/+ mice revealed that the percentage of granulocytes increased to > 10% only when the percentage of macrophages was reduced to < 10% of the total peritoneal cells (Figure 6C). The macrophages that could be isolated from the peritoneum of Flip^f/d, LysM^c/+ mice did not demonstrate deletion of Flip (Figure 6D). In addition, the Flip^f/d, LysM^c/+ mice demonstrated an increase percentage of cells expressing intermediate levels of F4/80 and Gr1 (F4/80^mid, Gr1^mid), which were also CD62L⁺ (data not shown) and may represent inflammatory monocytes. Further, there was a reduction of resident CD19⁺ B cells but not CD3⁺ T cells in the Flip^f/d, LysM^c/+ mice (supplemental Figure 7). To examine the role of FLIP in elicited macrophages, thioglycollate was injected into the peritoneum. At 20 hours, there was a marked increase of neutrophils in the controls, whereas neutrophil dominance persisted in the Flip^f/d, LysM^c/+ mice (Figure 6E). In both the Flip^f/d, LysM^c/+ mice and the controls there was a population of F4/80^mid, Gr1^mid cells at 24 hours (Figure 6E). At 72 hours, the controls demonstrated a marked increase of F4/80^mid, Gr1⁻ elicited macrophages. In contrast, in the Flip^f/d, LysM^c/+ mice, there was no increase of elicited macrophages, however, the marked increase of neutrophils persisted (Figure 6E-F). Therefore, the deletion of Flip in myeloid cells, not only results in an increase of peritoneal neutrophils, but also a marked reduction of resident and elicited macrophages.

Experiments were performed to determine whether the reduction of macrophages upon Flip excision was due to a defect in differentiation or to cell death after differentiation. c-Kit⁺ hematopoietic stem cells from floxed Flip mice were infected with a recombinant GFP-Cre retrovirus to promote Flip excision in vitro. After infection, the cells were differentiated into macrophages for 5 days. There was no difference in the numbers of GFP⁺ cells (Figure 7A) or GFP⁺ cells that were CD11b⁻, F4/80⁻, or CD11b⁺, F4/80⁻, or CD11b⁺, F4/80⁺ mature macrophages (Figure 7B) when Flip^f/+ and Flip^f/d c-Kit⁺ cells were infected with a control GFP-expressing retrovirus. In contrast, there was a reduction of GFP⁺ cells (Figure 7C) and GFP⁺ cells that were CD11b⁻, F4/80⁻, or CD11b⁺, F4/80⁻, or CD11b⁺, F4/80⁺ mature macrophages (P < .01-.001; Figure 7D) on days 1, 2, and 5 after infection of Flip^f/d c-Kit⁺ cells with the GFP, Cre-expressing retrovirus. This reduction does not appear to be due to increased apoptosis in response to Flip deletion, because there was no difference in the percentage of annexinV positivity for any of the cell types examined (data not shown). Further, when total bone marrow cells were used to generate macrophages, the caspase inhibitors IETD-fmk or zVAD.fmk did not rescue the GFP⁺, F4/80⁺ macrophages generated from the Flip^f/d mice ex vivo after the deletion of Flip (data not shown). Together these observations suggest that FLIP is necessary for macrophage differentiation.

To determine whether the neutrophilia was intrinsic to the loss of Flip in myeloid cells, bone marrow reconstitution experiments were performed using CD45.1⁺ wild-type and CD45.2⁺Flip^f/d, LysM^c/+ bone marrow cells injected into CD45.1⁺ lethally irradiated hosts all on a C57BL/6 background. In mixed chimera experiments, when 25%, 50%, or 75% of the bone marrow cells injected into the lethally irradiated recipients were from the Flip^f/d, LysM^c/+ mice, no increase of circulating neutrophils or monocytes was observed (Table 4). Further, decreased body weight, increased spleen size, and decreased peritoneal macrophages were not observed in the mixed chimeras. However, the CD45.2⁺Flip^f/d, LysM^c/+ bone marrow cells were unable to compete with the wild-type CD45.1⁺ cells (supplemental Figure 8). It was only when 100% of the bone marrow cells were from the Flip^f/d, LysM^c/+ mice that the inflammatory phenotype, including increased circulating neutrophils (Table 4), was observed. All of the mice that received 100% Flip^f/d, LysM^c/+ bone marrow cells demonstrated an increase of CD45.2⁺ neutrophils compared with all the other groups (supplemental Figure 8C). Of interest, the one mouse with the greatest increase of CD45.2⁺ neutrophils also demonstrated an increase of recipient CD45.1⁺ neutrophils. G-CSF, GM-CSF, and IL-6 were increased only in this mouse and the 2 other mice with the greatest increase of neutrophils (data not shown), supporting the role of growth factors/cytokines in the neutrophilia. These observations suggest that the neutrophilia observed in the Flip^f/d, LysM^c/+ mice is due to a cell autonomous defect in the bone marrow progenitors which results in increased cytokines/growth factors that contribute to the granulopoiesis.

The cellular mechanisms responsible for maintaining homeostatic balance in myeloid cells has not been fully characterized. Here, we demonstrate that the deletion of Flip in myeloid cells resulted in a marked increase of circulating neutrophils, associated with multiorgan neutrophil infiltration, including spleen, lymph nodes, thymus, intestine, and lung. There was no evidence of myeloid dysplasia in the bone marrow, spleen, or other organs. Increased granulopoiesis with an increase of CFU-GM was observed in the spleens of the Flip^f/d, LysM^c/+ mice, although not in the bone marrow. Under homeostatic conditions, the half life of circulating neutrophils is 8 hours or less, and neutrophilia may occur by increased survival. Although there was a slight reduction of neutrophil apoptosis at 0 and 4 hours, which may be due to the increased circulating growth factors, there appeared to be no intrinsic defect in neutrophil apoptosis, because at 24 hours neutrophil apoptosis in vitro was nearly complete using neutrophils from the Flip^f/d, LysM^c/+ mice. Because FLIP protects against death receptor-mediated apoptosis, a reduction of FLIP might be expected to lead to decreased cell survival. However, Flip mRNA was not expressed in circulating wild-type neutrophils (data not shown), further supporting our observations that suggest FLIP does not play a role in neutrophil survival. This interpretation is consistent with known role of the intrinsic mitrochondrial pathway involving Mcl-1, A1, and Bcl-x_L in neutrophil apoptosis.²⁶  In contrast to the results observed with FLIP, the deletion of Mcl-1 using LysM-Cre results in the neutropenia with no reduction of macrophages.^27,28 

Neutrophilia may also occur by increased production or increased release from the bone marrow in response to stress such as infection.²⁹  The Flip^f/d, LysM^c/+ mice demonstrated extramedullary hematopoiesis associated with an increase of growth factors in the circulation that promote myelopoiesis. However, the data do not support the underlying infection as the cause for the increased neutrophils because 70% of the Flip^f/d, LysM^c/+ mice were culture negative. Further, although antibiotic treatment reduced the rate of infections, it had no effect on the circulating neutrophils or the phenotype. These observations suggest that the persistent acute inflammation contributed to the stunted growth and premature death in the Flip^f/d, LysM^c/+ mice.

The marked increase of circulating neutrophils and the multiorgan infiltration with neutrophils was not expected. Prior studies have demonstrated that mice lacking the adhesion molecule CD18 exhibit neutrophilia with increased IL-17 and G-CSF, which contributed to the increased neutrophil counts.³⁰  Further, in these mice, the phagocytosis of apoptotic neutrophils by macrophages and dendritic cells suppressed granulopoiesis, which was mediated by a reduction of IL-23 and subsequently IL-17 and G-CSF.³¹  However, the Flip^f/d, LysM^c/+ mice demonstrated a reduction of splenic, lymph node, and peritoneal macrophages, suggesting that increased IL-23 was not responsible. In fact, circulating p40, which is common to both IL-12 and IL-23, was not elevated (data not shown). In contrast, IL-17 and G-CSF, in addition to IL-6, GM-CSF, and M-CSF, were all increased in the Flip^f/d, LysM^c/+ mice, suggesting that these growth factors may have contributed to the increased granulopoiesis. However, the results of the bone marrow transfer experiments suggest that the primary defect was a cell autonomous defect in myeloid development of FLIP-deficient bone marrow–derived cells, because these cells were capable of transferring the disease in the absence of competitor wild-type cells. The defect in myeloid development resulted in increased growth factors and cytokines, which resulted in the neutrophilia. Further, the mixed chimera experiments demonstrate that Flip^f/d, LysM^c/+ bone marrow cells were not able to compete with the wild-type cells, suggesting an impaired hematopoietic stem/progenitor cell phenotype. This may be due to excessive cytokines and growth factor production in the Flip^f/d, LysM^c/+ mice, which may change their ability in homing, repopulating, or potentially exhaust the stem/progenitor cells.^32,33  It is unlikely that FLIP deletion in the hematopoietic stem cells was directly responsible, because LysM is weakly expressed and only in a minority of these cells.³⁴ 

Resident peritoneal macrophages were markedly reduced in the Flip^f/d, LysM^c/+ mice. When peritoneal macrophages were reduced to < 10%, peritoneal granulocytes increased, suggesting that the reduction of macrophages contributes to the increase of granulocytes. The few Gr1⁻, F4/80⁺ macrophages that could be isolated for the peritoneum of the Flip^f/d, LysM^c/+ mice, or differentiated from bone marrow or spleen, were not deleted for Flip. Therefore, we were unable to examine the nonapoptotic functions of FLIP in macrophages, because any macrophages that could be isolated expressed FLIP. Prior studies demonstrated that under resting conditions neither Gr1⁺ nor Gr1⁻ monocytes homed to the peritoneum, whereas only Gr1⁺ monocytes migrated to the peritoneum after thioglycollate.^35,36  In the Flip^f/d, LysM^c/+ mice, resident F4/80^mid, Gr1^mid cells, consistent with inflammatory monocytes, were increased in the peritoneum. The recruitment of these cells may have been due to the acute inflammation with neutrophils. After the injection of thioglycollate at 20 hours, the inflammatory monocytes and neutrophils were comparable in the Flip^f/d, LysM^c/+ and control mice. However, in the Flip^f/d, LysM^c/+ mice, no increase of F4/80⁺, Gr1⁻ elicited macrophages was observed at 72 hours, consistent with a failure of the inflammatory monocytes to differentiate into macrophages, resulting in a persistent acute inflammatory response. We cannot exclude the possibility that following macrophage differentiation, the cells died. However, even when examined at 48 hours (data not shown), no increase of Gr1⁻ macrophages was observed in the Flip^f/d, LysM^c/+ mice, suggesting that Flip is essential for macrophage differentiation.

Further documenting the importance of FLIP for macrophage differentiation, infection of c-Kit⁺ bone marrow cells with a recombinant GFP-Cre retrovirus resulted in a reduction of differentiated CD11b⁺, F4/80⁺ macrophages derived from bone marrow of Flip^f/d mice. When Flip was deleted, there was also a reduction of CD11b⁻, F4/80⁻ and CD11b⁺, F4/80⁻ cells, demonstrating that the defect was not limited to the terminal stage of differentiation. There was no evidence of increased apoptosis to account for the reduction of macrophages. Relevant to this observation, another member of the death receptor pathway, caspase 8, is also necessary for normal macrophage differentiation,³⁷  and the deletion of caspase 8 in myeloid cells prevents macrophage differentiation.³⁸  In contrast, overexpression of FADD, another member of the Fas-mediated death pathway, accelerates the differentiation of macrophages, independent of any death receptor.³⁹  Our observations are consistent with those that identified the role of a complex of FADD, caspase 8, RIP1, and FLIP, which suppressed NF-κB activation and was necessary for M-CSF–mediated macrophage differentiation.³⁹  These observations are also reminiscent of the observation that the germ line deletion of either FLIP, caspase 8, or FADD is embryonic lethal, suggesting that these molecules also cooperate during embryonic development.⁴⁰ 

Together, these observations suggest the deletion of FLIP in macrophages resulted in the development of neutrophilia and the inflammatory phenotype. Of interest, a recent study demonstrated that deletion of dendritic cells also resulted in neutrophilia and a multiorgan infiltration with neutrophils.⁴¹  Although these mice also exhibited increased Th17 cells, in contrast to the Flip^f/d, LysM^c/+ mice, they also exhibited an infiltration of CD4⁺ T cells in intestine and kidney, as well as increased numbers of F4/80⁺ macrophages.⁴¹  Therefore, an intact mononuclear phagocytic system appears necessary for normal granulopoiesis and that deletion of either macrophages or dendritic cells results in neutrophilia. In summary, this study demonstrates that the expression of FLIP in myeloid cells is critical for the differentiation of macrophages, which is a characteristic feature of chronic inflammation, and that the acute inflammatory phenotype is secondary to the reduction of macrophages and an increase of cytokines and growth factors.

The authors thank Dr Guang-Yu Yang, Northwestern University Department of Pathology, for assistance with interpreting the histology of the gastrointestinal tract. We thank Rudina Sobkoviak and Jeff Kraynak for their contributions helping to establish and maintain the Flip conditional knockout mice and Dr Chiang-Ching Huang, Northwestern University Department of Preventive Medicine, for assistance with statistical analysis.

This work was supported in part by National Institutes of Health (NIH) grant nos. AR048269 and AR055240 to R.M.P. and AR050250, AR054796, and AI06759 to H.P., and a Within Our Reach Grant from the American College of Rheumatology to R.M.P. J.D.C. is a Scholar of the Leukemia & Lymphoma Society.

National Institutes of Health

Contribution: Q.-Q.H. and R.M.P. conceived the project, designed experiments, performed data analysis, and wrote the manuscript; R.M.P. supervised the project; Q.-Q.H. and R.B. generated and maintained the KO mice and performed the major experiments in genotypic and phenotypic analysis and data acquisition; H.P. and J.D.C. designed experiments, interpreted results, and assisted in manuscript preparation; Z.H. designed and performed CFU and bone marrow infection experiments; L.K. performed immunohistochemistry analysis; S.G. performed the histologic and pathologic identification; H.A. and A.M. assisted with the mixed chimeras and their analysis; and all authors discussed the results and implications and contributed to the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Richard M. Pope, 240 E Huron, Ste M-300, Chicago, IL 60611; e-mail: rmp158@northwestern.edu.