PRAM-1 Is Required for Optimal Integrin-Dependent Neutrophil Function

Cloning of murine PRAM-1.

In order to isolate the murine homolog of PRAM-1, mRNA was isolated from murine bone marrow-derived macrophages, and primers derived from human PRAM-1 were used to amplify a 400-bp fragment of the 3′ end of the cDNA via reverse transcription (RT)-PCR. The 400-bp fragment was sequenced and used as a probe to search the National Center for Biotechnology Information murine expressed sequence tag database using the BLASTn program. Overlapping, aligned expressed sequence tags were used to design primers for amplification and sequencing of 1,100 bp of the 3′ end. The complete coding sequence was generated by 5′ rapid amplification of cDNA ends using a murine spleen cDNA library (Clontech; catalog no. ML5011t). The complete sequence was confirmed by amplification of full-length cDNA from murine bone marrow-derived mRNA.

Generation of PRAM-1-deficient mice.

A 3-kb XbaI-HindIII genomic DNA fragment containing the PRAM-1 start codon and 1,000 bp of the second coding exon was replaced by a neomycin resistance gene cassette. The targeting construct was electroporated into 129/SVJ embryonic stem cells (R1; Genome Systems, St. Louis, Mo.) grown on feeder layers of murine embryonic fibroblasts. Targeted embryonic stem cells were positively selected with G418 (200 μg/ml) and negatively selected with ganciclovir. Specificity of targeting vector integration within the PRAM-1 locus was determined by PCR screening and Southern analysis of G418-resistant clones. Targeted clones were microinjected into blastocysts obtained from C57BL/6 mice, which were then transferred to pseudopregnant recipient mice. Two founder chimeric mice were mated with C57BL/6 mice, and heterozygous offspring were identified by Southern blotting of tail genomic DNA. Mice harboring the targeted allele were backcrossed to C57BL/6 four times; control mice were backcrossed and age-matched for all experiments. Mice were housed at the University of Pennsylvania Animal Care Facility under pathogen-free conditions and used in accordance with the National Institutes of Health and Institutional Animal Care and Use Committee guidelines.

Isolation of bone marrow neutrophils.

Bone marrow-derived neutrophils were prepared as previously described (39). Briefly, bone marrow was flushed from the femur and tibia of 2- to 6-month-old mice with Hanks' balanced salt solution (HBSS) Prep (5.4 mM KCl, 0.3 mM Na₂HPO₄, 0.8 mM KH₂PO₄, 4.2 mM NaHCO₃, 137 mM NaCl, 5.6 mM dextrose, 20 mM HEPES, 0.5% fetal calf serum [FCS]). Red blood cells were lysed in hypotonic NaCl, followed by addition of an equal volume of hypertonic NaCl solution to restore normal tonicity. Cells were resuspended in 5 ml of HBSS Prep, layered over a 62.5% Percoll gradient, and centrifuged at 2,200 rpm for 30 min. Mature neutrophils were harvested from the bottom layer of the gradient. The purity of the neutrophil preparation was assessed using flow cytometry; standard preparations yielded neutrophils of >90% purity determined by surface expression of Gr-1 (Pharmingen, San Diego, Calif.).

Antibodies.

Chicken antiserum recognizing a peptide (CZPQTELSEQPKKSSQSE) from the amino terminus of murine PRAM-1 was generated by Aves Labs Inc. (Tigard, Oreg.). Sheep antiserum directed against a glutathione S-transferase fusion protein containing the carboxyl-terminal 300 amino acids of murine PRAM-1 was generated by Elmira Biologicals Inc. (Iowa City, Iowa). The antiphosphotyrosine monoclonal 4G10 and anti-Vav antibodies were purchased from Upstate Biotechnologies Inc. (Waltham, Mass.). All other antibodies were purchased from Cell Signaling Inc. (Beverly, Mass.).

Immunoprecipitation and Western blotting.

For poly-RGD stimulation, tissue culture plates were coated with 15 μg of poly-RGD (Sigma Aldrich, St. Louis, Mo.)/ml in phosphate-buffered saline (PBS) for 2 h and then washed three times with PBS. Neutrophils were plated on poly-RGD or left in suspension at 37°C. Cells were lysed at indicated time points using a radioimmunoprecipitation assay-based buffer (40); insoluble material was removed through centrifugation at 14,000 rpm for 10 min in a tabletop microfuge. For Western blotting of whole cell lysates, 4× sample buffer (0.5 M Tris-HCl [pH 6.8], 277 mM sodium dodecyl sulfate, 40% glycerol, 20% 2-mercaptoethanol, 1% bromophenol blue) was added, and samples were boiled for 5 min. For immunoprecipitation, lysates containing approximately 10 million cells were precleared with GammaBind G-Sepharose (Amersham Pharmacia Biotech, Piscataway, N.J.). Two micrograms of anti-Vav antibody or 3 μl of PRAM-1 antiserum was added to the cell extracts and rotated at 4°C for 2 h. Twenty-five microliters of GammaBind beads were then added, and samples in tubes were rotated for an additional 4 h. Samples were washed three times and then boiled for 5 min in 2× sample buffer. Samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Bio-Rad, Hercules, Calif.) and transferred to nitrocellulose membranes (Bio-Rad). Membranes were blocked with 5% bovine serum albumin in Tris-buffered saline-Tween 20 for 30 min and then probed overnight with the appropriate antibody diluted in Tris-buffered saline-Tween20-5% bovine serum albumin. Antibody binding was detected by horseradish peroxidase-conjugated secondary antibody (Bio-Rad) and autoradiography.

Complete blood count.

Cardiac puncture was used to withdraw 250 μl of blood from carbon dioxide-asphyxiated mice. The blood was immediately transferred into EDTA-coated tubes (Becton-Dickinson, Franklin, N.J.). The complete blood count (CBC) was determined with a HEMA VET MASCOT 850 instrument (CDC Technologies Inc., Oxford, Conn.).

Measurement of FcγR-induced calcium flux and ROI production.

For labeling with the calcium-sensitive fluorochrome Indo-1, neutrophils were resuspended at 10⁷ cells/ml in HBSS Prep containing 3 μg of Indo-1 (Molecular Probes, Eugene, Oreg.)/ml and 4 mM probenecid and incubated at 37°C for 20 min. Cells were washed twice in PBS and then incubated with 2 μg of anti-FcγRII/III (Pharmingen)/ml in PBSg (125 mM sodium chloride, 8 mM sodium phosphate, 2 mM sodium phosphate monobasic, 5 mM potassium chloride, 5 mM glucose) for 15 min at 4°C. Following three washes in cold PBSg, cells were resuspended in ice-cold PBSg at 10⁷ cells/ml until initiation of the assay. Two minutes prior to stimulation, cells were diluted to 0.5 million cells/ml in 37°C KRP buffer (PBSg, 1 mM calcium chloride, 1.5 mM magnesium chloride). Cell stimulation was initiated by adding 15 to 30 μg of goat anti-rat immunoglobulin G (FcγR; Pharmingen)/ml or 2.5 μM formyl-MetLeuPhe (fMLP). Calcium levels were analyzed on an LSR flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, Calif.). Simultaneous immune complex-induced calcium flux and ROI production were detected by using the immune complex reagent Fc OxyBurst green (Molecular Probes), and data were collected on a FACStar Plus (Becton Dickinson Immunocytometry Systems) cytometer while the cell suspension was maintained at 37°C with continuous stirring.

Adhesion-dependent degranulation.

Immulon 4 HBX plates (Thermo Labsystems, Helsinki, Finland) were coated overnight at room temperature with 150 μg of sheep fibrinogen (Sigma Aldrich)/ml in PBS. The wells were washed three times, and 10 μl of a 10 mM solution of MgCl₂ with or without TNF-α (500 ng/ml) or phorbol myristate acetate (PMA; 1 μM) was added to the appropriate wells (four wells per condition per time point). Prewarmed neutrophils (100 μl) suspended at 3 × 10⁶ cells/ml in HBSS supplemented with 10 mM HEPES and 1 mM Ca²⁺ were added to each well. Wells were harvested after 0, 30, and 45 min, and cells were pelleted. Supernatant was diluted 4 to 20 times in carbonate buffer, and 100 μl per well was transferred to a fresh 96-well Immulon 4 HBX plate to allow lactoferrin binding overnight at 4°C. The enzyme-linked immunosorbent assay for lactoferrin was performed as previously described (38).

Expression of cell surface markers.

Levels of receptor surface expression were measured by using flow cytometry. Cells were labeled with fluorescently conjugated antibodies (BD Pharmingen) for Ly6G (RB6-8C5), CD16/CD32 (2.4G2), CD18 (C71/16), Mac-1 (M1/70), and LFA-1 (M17/4). For the evaluation of upregulation of surface integrin expression following inflammatory stimulus, neutrophils were incubated with 50 ng of TNF-α (Peprotech, Princeton, N.J.)/ml for 30 min at 37°C.

Adherent respiratory burst.

Ninety-six-well Immulon 4 HBX plates (Thermo Labsystems) were coated with 150 μg of fibrinogen (Sigma; F-9754)/ml in PBS, 15 μg of poly-RGD (Sigma; F-5022)/ml in PBS, or 20 μg of anti-CD18 (C71/16) or isotype control antibody (Pharmingen)/ml in carbonate buffer. Antibody-coated wells were blocked with 10% FCS for 30 min. Prewarmed neutrophils (100 μl) suspended at 4 × 10⁶ cells/ml in HBSS-10 mM HEPES-100 mM ferricytochrome c (Sigma Aldrich)-1 mM Ca²⁺ were added to the wells. Adherent respiratory burst was measured through oxidation-induced changes in ferricytochrome C absorbance as previously described (34). When indicated, 50 ng of murine TNF-α/ml, 10 μM fMLP, or 100 nM PMA was added.

Staphylococcus aureus killing assay.

S. aureus, wood strain (American Type Culture Collection, Manassas, Va.), was grown overnight on brain heart infusion (BHI) agar (Difco, Sparks, Md.). The morning of the assay, five colonies were transferred into BHI broth and grown at 37°C with constant shaking. The bacterial concentration was measured by absorbance at 590 nm. Shaking was terminated when absorbance reached 0.1 to 0.4, and the bacterial concentration was calculated using an absorbance constant attained from previous serial dilutions of cultures. The S. aureus cells were opsonized with 1 μl of opsonizing antibody (Molecular Probes) in HBSS with 50% mouse serum for 15 min at 37°C and then washed twice with PBS. Isolated bone marrow-derived neutrophils suspended in HBSS-10 mM HEPES were warmed for 30 min at 37°C. Bacteria and neutrophils (20 × 10⁶ cells/ml) were combined at a 1:2 ratio in HBSS containing 10% mouse serum and rotated at 37°C. At designated time points, three 40-μl aliquots were removed and combined with equal volumes of 2% NP-40 on ice to lyse neutrophils. Serial dilutions of the lysates were plated upon BHI agar and grown overnight. The reported percent survival is represented as the percentage of colonies remaining at each time point in comparison to the number at time zero for each condition.

Lipopolysaccharide-induced endotoxic shock.

Cohorts of 8 to 10 age-, weight-, backcross-, and gender-matched mice were injected intraperitoneally with 0.5 or 1 mg of lipopolysaccharide (LPS; Sigma Aldrich) in sterile PBS. Mice were monitored for signs of endotoxic shock (diarrhea, eye and nose discharge, fur ruffling) every 6 h in the first 24 h and twice per day in the following days. Moribund mice were euthanized to minimize stress and suffering. All remaining mice were euthanized on day 7.

Murine PRAM-1 is expressed predominantly in neutrophils.

Previous studies of human tissue and myeloid leukemic cell lines demonstrated that human PRAM-1 is expressed during myeloid differentiation and is upregulated during granulocytic (retinoid-induced) differentiation of the promyelocytic cell line NB4 (41). To develop a murine model system, we first isolated cDNA for murine PRAM-1 (GenBank AY665714). Analysis of the full-length amino acid sequence revealed that murine PRAM-1 is 53% identical and 57% similar to human PRAM-1 (Fig. 1A), with greatest homology in the carboxy-terminal SH3-like domain (98% identical). Human PRAM-1 also contains two tyrosine-based motifs that conform to the consensus sequence recognized by group I SH2 domains (52). One of these motifs is wholly conserved in murine PRAM-1, while the other contains a single amino acid substitution at position Y + 2 (D→A), which likely alters its SH2 domain binding affinity. A BLAST search with murine PRAM-1 identified two mRNA sequences generated by Riken (Kanagawa, Japan). The first sequence was derived using automated computational analysis (GenBank XM128537). The murine PRAM-1 sequence generated by RT-PCR in our laboratory is identical to this derived sequence except for a 45-amino-acid region (residues 525 to 662) near the carboxy terminus, which is not present in the Riken sequence. Interestingly, the PRAM-1 homolog ADAP exhibits 120- and 130-kDa forms that result from alternative splicing of 138 bp (46 amino acids), also within the carboxy terminus (57). Analysis of the murine PRAM-1 locus on chromosome 17 revealed that exons 3 and 4 encode the inserted 45 amino acids, suggesting that the Riken sequence may represent a splice variant of PRAM-1. However, multiple independent derivations of PRAM-1 cDNA by RT-PCR from murine bone marrow yielded only the insert-containing sequence (data not shown). The second Riken sequence was derived using a cDNA library from mouse neonatal skin (GenBank AK076291). This sequence includes the 45 amino acids representing exons 3 and 4 but differs from both our sequence and the computationally derived sequence at several nucleotide positions.

Like human PRAM-1, the murine ortholog is found predominately in murine bone marrow and in purified mature neutrophils. Although PRAM-1 mRNA was detectable by RT-PCR in bone marrow-derived macrophages and mast cells (data not shown), immunoprecipitation and Western blotting for PRAM-1 protein indicated only very weak expression in macrophages and undetectable levels in mast cells (Fig. 1B). Together, these data identify murine PRAM-1 as a largely neutrophil-specific adaptor protein.

Genetic targeting of PRAM-1.

To evaluate the importance of PRAM-1 for neutrophil differentiation and function, we generated mice with targeted disruption of the PRAM-1 gene. A neomycin resistance cassette was inserted into the PRAM-1 locus by homologous recombination, replacing the first exon and most of the second exon (Fig. 2A). Locus-specific integration was confirmed by Southern analysis of EcoRI-digested genomic DNA (Fig. 2B). Immunoprecipitation and Western analysis of bone marrow cell lysates confirmed that PRAM-1 protein is not expressed in mice homozygous for the disrupted gene (Fig. 2C). PRAM-1-deficient mice are viable, fertile, and born at the expected frequency from heterozygote-by-heterozygote matings.

PRAM-1^−/− mice have normal numbers of neutrophils.

The expression pattern and regulation of PRAM-1 are similar to that of the transcription factor CCAAT/enhancer binding protein epsilon (C/EBPɛ). Both proteins are predominantly expressed in myeloid tissue and are transcriptionally repressed by the oncogenic fusion protein PML-RARα in human acute promyelocytic leukemia blasts (26, 41, 60, 61). C/EBPɛ is required for granulocytic differentiation in both mice and humans (27); therefore, we asked if PRAM-1 might facilitate signaling pathways regulating neutrophil differentiation or maturation. However, analysis of PRAM-1-deficient mice revealed no gross abnormalities in the neutrophil compartment or any other hematopoietic lineage. As measured by CBC, wild-type (WT) and PRAM-1^−/− mice had comparable numbers of peripheral blood neutrophils as well as total leukocytes (Fig. 3A). Microscopically, PRAM-1^−/− bone marrow was indistinguishable from that of the wild type, demonstrating normal cellularity and morphology of developing neutrophils and other hematopoietic precursors (Fig. 3B, upper panel). Bone marrow-purified PRAM-1^−/− neutrophils also demonstrate normal nuclear morphology and stain for chloroacetate esterase, a neutrophil-specific marker (2, 29) (Fig. 3B, lower panels). Additionally, cell surface expressions of the granulocyte markers Gr-1, FcγRII/III, and β₂ (CD18) and β₃ (CD61) integrins were equivalent in the WT and PRAM-1^−/− neutrophils (Fig. 3C). Collectively, these data indicate that PRAM-1 is not essential for normal neutrophil differentiation or maturation.

PRAM-1^−/− neutrophils flux calcium and produce ROI after FcγR stimulation.

The adaptor SLP-76, a binding partner of PRAM-1 (41), modulates signaling pathways downstream of tyrosine kinase-associated receptors in several hematopoietic lineages, including the T-cell receptor in T cells (9), FcɛR in mast cells (46), and GPVI-collagen receptor and α_IIbβ₃ integrin in platelets (8, 24). Recently, studies of neutrophils demonstrated that SLP-76 is also required for signaling through Fcγ receptors and integrins (43). As PRAM-1 can associate with SLP-76 in myeloid cell lines, we hypothesized that PRAM-1 might also modulate the signaling cascades of these receptors in neutrophils. In response to cross-linking of FcγR, WT neutrophils mobilize calcium and produce ROI (11, 25). To test the ability of PRAM-1-deficient neutrophils to generate a calcium flux, Indo-1-labeled PRAM-1^−/− and WT neutrophils were incubated with antibodies to FcγRII/III. When stimulated by hyper-cross-linking, the FcγR PRAM-1^−/− neutrophils produced a calcium flux equivalent to that of WT neutrophils (Fig. 4A). To assess FcγR-induced ROI production, neutrophils were incubated with Fc OxyBurst, a synthetic immune complex reagent conjugated to the oxidation-sensitive compound dichlorodihydrofluorescein diacetate (DCHF). When Fc OxyBurst is phagocytosed and oxidized, DCHF fluorescence can be quantified by flow cytometry (17, 23). Figure 4B shows ROI production by WT and PRAM-1^−/− neutrophils following immune complex internalization. Though in this experiment PRAM-1^−/− neutrophils produce slightly fewer ROI, this difference is not reproducible. Together, these studies suggest that FcγR-dependent signaling remains intact in the absence of PRAM-1.

PRAM-1^−/− neutrophils have defects in integrin-mediated responses.

In neutrophils, integrins are essential for several functions, including adhesion, spreading, migration, and adhesion-dependent ROI production and degranulation (4, 32). The PRAM-1 homolog ADAP has been shown to be required for integrin-mediated T-cell adhesion (19, 45). We therefore assessed each of these responses following integrin stimulation of PRAM-1^−/− neutrophils. Mature, unstimulated neutrophils do not spread on fibrinogen-coated surfaces; however, stimulants such as TNF-α activate integrins, allowing for cell spreading. During activation with TNF-α, both the percentage of cells that spread (WT, 29.8% ± 8.8%; knockout, 34.1% ± 4.4%) and the area of the spread cells (WT, 800.6 ± 58 pixels; knockout, 745.5 ± 40.2 pixels) were equivalent in WT and PRAM-1^−/− neutrophils (Fig. 5A, upper panels). Spreading of PRAM-1^−/− and WT neutrophils was also equivalent on slides coated with poly-RGD (Fig. 5A, lower panels), an engineered polymer of the fibronectin-binding motif (Arg-Glu-Asp) that directly stimulates integrins without an additional activation signal.

Studies suggest that integrin-induced spreading is a prerequisite for adhesion-dependent inflammatory responses, such as ROI production and degranulation (42). However, despite normal spreading in response to integrin ligands, we observed that adhesion-dependent ROI production in the PRAM-1^−/− neutrophils was significantly decreased following stimulation with fibrinogen in concert with the inflammatory agent TNF-α or the bacterial peptide fMLP (Fig. 5B). Adhesion-dependent degranulation, as measured by release of the granule enzyme lactoferrin, was also significantly decreased in the PRAM-1^−/− neutrophils (Fig. 5C). The defects in both ROI production and degranulation were rescued by addition of PMA, demonstrating that the machinery required for each of these processes is functional in PRAM-1^−/− neutrophils. To ensure that the observed defects are restricted to integrin-dependent signaling events, we next examined the effect of stimulating WT and PRAM-1-deficient neutrophils with fMLP on plates coated with FCS. These experiments were performed in magnesium-free medium to ensure that integrin ligands present within the FCS would not interact with their receptors. While ROI production in response to fMLP alone is quantitatively less robust than the adhesion-dependent response, both WT and PRAM-1-deficient neutrophils generate ROI under these conditions, confirming that the phenotype of the PRAM-1-null cells is restricted to the adhesion-dependent response (data not shown).

Adhesion-dependent neutrophil functions require two stimuli. First, an “inside-out” proinflammatory stimulus serves to activate the integrins by increasing receptor affinity (conformational change), avidity (membrane relocalization), and level of surface expression. Second, an “outside-in” signal is initiated by binding of the activated integrin to its ligand, yielding adhesion-dependent effector functions. To determine whether the adhesion-dependent deficiencies in PRAM-1^−/− neutrophils resulted from a defect in inside-out or outside-in signaling, we assayed each of these pathways in isolation. Stimulation of suspended cells with chemokines or cytokines induces several inside-out responses, including upregulation of integrin surface expression via membrane fusion of primary granules (32). We found that upregulation of Mac-1 (α_Mβ₂) surface expression in response to TNF-α was equivalent in PRAM-1^−/− and WT neutrophils. In addition, activation of cells with the chemoattractant fMLP produced a normal calcium flux and adhesion-independent ROI production in PRAM-1^−/− neutrophils (data not shown). Collectively, these data suggest that these inside-out signals are intact in the absence of PRAM-1.

We assessed outside-in signaling by measuring ROI production following direct integrin stimulation with either plate-bound anti-CD18 (β₂ integrin) antibody or poly-RGD. Both of these stimuli cross-link integrins and initiate integrin signaling without prior inside-out integrin activation. ROI production following either anti-CD18 or poly-RGD stimulation was decreased by 60 to 80% in the PRAM-1^−/− neutrophils (Fig. 5D). Together these data suggest that PRAM-1 is necessary for the integrin-mediated outside-in signal but not for cytokine-induced integrin activation.

To address the mechanism behind the defect in adhesion-dependent ROI production and degranulation, we undertook a biochemical survey of several signaling molecules known to be active in integrin signaling pathways. Vav is a guanine nucleotide exchange factor that regulates actin reorganization following integrin ligation (13, 37, 49, 63). Pyk2 is a member of the FAK family kinases, and a recent study has implicated this molecule in adhesion-dependent ROI production in neutrophils (18, 22). The mitogen-activated protein (MAP) kinases ERK2 and p38 are also activated following integrin stimulation and are thought to be necessary for activation of effector functions, although their precise role in neutrophils is not well defined (15, 32). Vav, MAPK, and Pyk2 family members all undergo phosphorylation following integrin stimulation. As shown in Fig. 5E, PRAM-1^−/− and WT neutrophils demonstrate equivalent phosphorylation of these molecules in response to the stimulus poly-RGD. These results suggest that PRAM-1 is not required for key phosphorylation events initiated by integrin stimulation.

PRAM^−/− neutrophils kill bacteria in vitro, and PRAM^−/− mice show normal susceptibility to LPS-induced endotoxic shock.

ROI production and the release of toxic granule contents are essential processes that enable neutrophils to kill invading pathogens, yet integrin-dependent production of these substances has also been implicated in tissue damage associated with endotoxic shock (7, 33). Thus, we asked next whether the observed reduction in ROI production and degranulation impairs the ability of PRAM-1^−/− neutrophils to kill bacteria or, conversely, attenuates neutrophil-dependent pathology in an in vivo model of endotoxic shock. To measure bacterial killing, purified neutrophils were incubated with S. aureus cells, and bacterial survival was assessed at several time points. In this assay, WT and PRAM-1-deficient neutrophils killed bacteria with similar efficiencies (Fig. 6A). In addition, no survival benefit was conferred by loss of PRAM-1 in LPS-induced endotoxic shock, suggesting that residual integrin-dependent ROI generation and degranulation capabilities observed in PRAM-1 mutant mice are sufficient to mediate tissue damage leading to lethality (Fig. 6B).