Interferon-gamma coordinates CCL3-mediated neutrophil recruitment in vivo

Background We have shown previously that acute infection with the respiratory pathogen, pneumonia virus of mice (PVM), results in local production of the proinflammatory chemokine, CCL3, and that neutrophil recruitment in response to PVM infection is reduced dramatically in CCL3 -/- mice. Results In this work, we demonstrate that CCL3-mediated neutrophil recruitment is coordinated by interferon-gamma (IFNγ). Neutrophil recruitment in response to PVM infection was diminished five-fold in IFNγ receptor gene-deleted mice, although neutrophils from IFNγR -/- mice expressed transcripts for the CCL3 receptor, CCR1 and responded functionally to CCL3 ex vivo. Similarly, in the absence of PVM infection, CCL3 overexpression alone could not elicit neutrophil recruitment in the absence of IFNγ. Interestingly, although supplemental IFNγ restored neutrophil recruitment and resulted in a sustained weight loss among CCL3-overexpressing IFNγ -/- mice, CCL3-mediated neutrophil recruitment alone did not result in the pulmonary edema or respiratory failure characteristic of severe viral infection, suggesting that CCL3 and IFN-γ together are sufficient to promote neutrophil recruitment but not pathologic activation. Conclusion Our findings reveal a heretofore unrecognized hierarchical interaction between the IFNγ and CCL3, which demonstrate that IFNγ is crucial for CCL3-mediated neutrophil recruitment in vivo.


Background
Most respiratory virus infections are relatively benign and self-limited events. However, infection with highly pathogenic viruses can result in more severe sequelae, in which disease progresses to respiratory failure due to uncontrolled inflammation, pulmonary edema, and damage to lung tissue [1][2][3][4][5].
As part of an ongoing effort to understand inflammatory responses during severe respiratory virus infection, we have developed an inhalation model using the natural rodent pathogen, pneumonia virus of mice (PVM). Originally identified by Horsfall and colleagues [6,7], PVM is a pneumovirus (family Paramyxoviridae) that is closely related to respiratory syncytial virus (RSV), and is among the few characterized mouse models of virus-induced acute respiratory distress syndrome (ARDS) [7][8][9]. Among the prominent features of this infection, a minimal intranasal inoculum (30 -100 pfu) results in robust virus replication within bronchial epithelial cells that is accompanied by profound granulocyte recruitment. In the absence of pharmacologic intervention, PVM infection progresses to pulmonary edema and respiratory compromise, similar to the more severe forms of RSV infection experienced by human infants [10,11]. In our earlier studies, we identified the chemokine CCL3 (MIP-1α) as a crucial component of this inflammatory response. PVM not only elicits production of CCL3 by infected bronchial epithelial cells [12], mice devoid of CCL3 or its receptor, CCR1, recruit dramatically fewer neutrophils to airways [13]. Blockade of the CCL3/CCR1 proinflammatory signaling pathway in conjunction with antiviral therapy resulted in improved survival in response to an otherwise lethal virus inoculum [14,15]. As CCL3 is only one of several major pro-inflammatory signaling pathways activated by PVM infection [12], there is certainly the possibility of additive, synergistic, or hierarchical means to promote and to amplify the ongoing inflammatory response.
Although first identified as a component of the antiviral response to Sindbis virus [16], the role of the Th1 cytokine, interferon-γ (IFNγ) in pneumovirus infection remains uncertain. IFNγ is readily detected in bronchoalveolar lavage fluid and nasal washings from RSV-infected infants [17,18], and minimal or absent response has been correlated with poor clinical outcome [19][20][21][22][23][24]. IFNγ is also detected in BAL fluid of BALB/c mice in response to challenge with RSV virions [25,26] and plays a role in limiting the inflammatory response to secondary challenge and in generating the allergic histopathology in response to formalin-fixed RSV vaccine antigens and virion components [27,28]. Likewise, local production of IFNγ is a prominent response to PVM infection [12,29,30], although its role in modulating the primary inflammatory response has not yet been fully explored.
In this manuscript, we explore the role of IFNγ in modulating the inflammatory response to PVM infection, and utilize overexpression analysis to begin a dissection of the independent and interdependent contributions of both IFN-γ and CCL3 to the process of neutrophil recruitment in vivo.

Microarray profiling of IFN expression in response to PVM infection
Transcript encoding the cytokine IFNγ was detected in mouse lung tissue at various time points in response to PVM infection [12]. In response to a non-lethal inoculum of PVM strain J3666, IFNγ mRNA was detected above baseline levels beginning on day 5. IFN-γ mRNA levels peak at day 7 after inoculation, and fall rapidly to baseline between days 7 -14. Shown in Figure 1A are profiles of the 203 transcripts (of total 45,101 transcripts on the 430_2 mouse chip) that display kinetic expression correlations of 0.900 or greater with the IFN-γ profile, as per the 'find similar' algorithm of Genespring GX 7.3. Selected transcripts, categorized by function, are listed in Table 1. Among the transcripts that correlate with the IFNγ profile are 17 characterized interferon-response genes. Most intriguing is the close correlation (0.965) between the expression patterns of IFNγ and CCL3 (MIP-1α). CCL3 is essential for granulocyte recruitment in response to PVM infection [13]. As shown in Figure 1B, there is a significant correlation between levels of immunoreactive IFNγ and CCL3 in lung tissue from individual PVM-infected mice.

Detection IFN + NK and T cells in PVM infected lung tissue
Both total and IFNγ + subsets of NK cells, CD4 + and CD8 + T cells were enumerated in single cell suspensions of lung tissue from PVM-infected BALB/c mice evaluated at day 6 after inoculation with 10 pfu PVM strain J3666 [ Figure  1C]. Only a small fraction (<2%) of the CD4 + T cells detected at this time point stained positively for IFNγ, in contrast to the larger fraction of IFNγ + CD8 + T cells detected (9.9 ± 0.6 × 10 3 cells, 11% of total CD8 + T cells). Interestingly, 23% of the total NK cells (3.4 ± 0.9 × 10 3 cells) stained positively for IFNγ, an increase from 0.3 ± 0.08 × 10 3 cells, or 4% of the total NK cells detected in a single lung from uninfected mice (data not shown).

IFN -dependent responses to PVM infection
Wild type and IFNγ receptor gene deleted (IFNγR -/-) mice were infected with PVM and various parameters relating to the inflammatory response were assessed. Neutrophil recruitment to the airways was markedly diminished in IFNγR -/-mice [ Figure 2A], reduced from 54 ± 11 per 10 hpfs among wild type to 10 ± 1.3 hpfs among IFNγR -/mice, as determined on cytospin preparations of cells in BAL fluid (p < 0.001). These findings are consistent with those of Frey and colleagues [30], who described reduced   inflammation in association with reduced IFNγ production in the lungs of PVM infected, T-cell deficient mice. Given our earlier studies on the essential role of CCL3 in eliciting neutrophil recruitment, it is interesting to note that the absence of IFNγ signaling had no impact on local production of this chemokine in response to PVM infection [ Figure 2B]. IFNγ was also detected in response to PVM infection in both wild type and in IFNγR-/-mice, albeit at higher levels among the latter group, most likely due to the absence of feedback inhibition (data not shown). The diminished neutrophil recruitment, while significant, was not as profound as that observed in mice subjected to complete blockade of CCL3-mediated signaling, in which we observed 10 4 -10 5 fold-diminished neutrophil recruitment [14,15]. As might be anticipated from the diminished inflammatory response, we observe a sta-tistically significant increase in virus titer among the IFNγR-/-mice [ Figure 2C], although this difference is likewise not as dramatic as that observed in response to complete blockade of CCL3 signaling.

Receptor expression and responses of neutrophils from IFN R gene-deleted mice
As part of an initial attempt to determine whether neutrophils from IFNγR -/-mice were capable of responding to CCL3, we explored receptor expression and ligandmediated calcium flux in neutrophils isolated from both gene-deleted and wild type mice. As shown in Figure 3A, both wild type and IFNγR-/-neutrophils express transcripts encoding CCR1, the major receptor for CCL3; no significant difference in absolute copy number was determined. Likewise, CCL3 induced dose-dependent intracel- The microarray analysis software package, Genespring GX 7.3 'find similar' function was used to inspect all transcript profiles for patterns related to that displayed by IFN-γ. The minimum correlation considered to be similar was set at 0.900 (see Figure 1A). lular calcium flux in both gene-deleted and wild type neutrophils [ Figure 3B], demonstrating that neutrophils from IFNγR-/-mice have the innate capacity to respond to this chemoattractant ligand; the EC50s and maximum calcium fluxes detected were indistinguishable between the wild type and gene-deleted strains.

Overexpression of CCL3
In order to examine the independent and interdependent contributions of CCL3 and IFNγ to the process of neutrophil recruitment in vivo, we generated a method for overexpression of CCL3 in vivo. CCL3 was detected in lung tissue homogenates [ Figure 4A], reaching levels similar to those detected in lung tissue of mice in response to PVM infection [12]. Immunoreactive CCL3 was detected in bronchial epithelial cells [ Figure 4B]. No CCL3-positive cells were detected in lung tissue from mice challenged with control vector (vctrl) [ Figure 4C].

Inflammatory responses to IFN and CCL3
We examined neutrophil recruitment in response to CCL3 overexpression in IFNγ gene-deleted mice (IFNγ -/-) with and without IFNγ supplementation. As shown in Figure  5A, few neutrophils are detected in BAL fluid at baseline (vctrl) and no recruitment over baseline is observed in response to IFNγ alone. Likewise, overexpression of CCL3 in the absence of IFNγ does not elicit neutrophil recruitment. Neutrophil recruitment (~10 -fold over baseline) was observed in response to CCL3 expression only in the presence of IFNγ. At the microscopic level, no inflammation was observed in lung tissue of IFNγ -/-mice in response to CCL3 overexpression alone [ Figure 5B]. In contrast, significant pathology was observed in lung tissue of IFNγ -/-mice expressing CCL3 and supplemented with exogenous IFNγ. Findings include moderate peribronchiolar granulocytic infiltration and substantial parenchymal involvement but minimal edema fluid within the bronchioles and in the parenchymal tissue [ Figure 5C]. Interestingly, weight loss is sustained among the mice overexpressing CCL3 while receiving supplemental IFNγ over the 9 day examination period [ Figure 5D], but, despite the substantial inflammatory response, we observe no progression to respiratory failure up to and including t = 14 days.

Discussion
In previous work, we demonstrated that the actions of the chemokine, CCL3, signaling via its receptor CCR1, were crucial for granulocyte recruitment to the lungs in response to PVM infection [13][14][15]; CCL3 has also been shown to be a crucial mediator of granulocyte recruitment in mouse models of influenza [31]. Paradoxically, CCL3 gene-deletion results in augmented neutrophil and eosinophil recruitment in response to Cryptococcus neoformans infection [32]. Here we show that CCL3-mediated neutrophil recruitment depends directly on IFNγ signaling, both in the setting of acute virus infection and in response to heterologous CCL3 expression in the respiratory epithelium.
Granulocyte recruitment is a primary finding in severe respiratory virus infection; activation of granulocytes can result in the release of proinflammatory cytokines and proteolytic enzymes that can contribute to the ongoing lung damage [33][34][35][36][37]. Interestingly, although neutrophils are recruited to the lung parenchyma in response to CCL3 via coordination by IFNγ, these cytokines alone clearly are not sufficient to induce the inflammatory state that ultimately promotes lung damage and respiratory failure. Thus, despite our findings demonstrating improved survival from PVM infection with CCR1 blockade [15], and Neutrophil recruitment in response to PVM infection is diminished in IFN-γR gene-deleted mice those of He and colleagues [38], who likewise demonstrated that CCR1 antagonism provided protection against neutrophil-mediated lung injury in a mouse model of acute pancreatitis, the results presented here, in which we observe neutrophil recruitment but minimal clinical disease, suggest that neutrophil recruitment and neutrophil activation are to some extent distinct and discrete signaling events. It will be crucial to identify the proinflammatory mediators that activate and well as those that recruit neutrophils in order to have a complete picture of the proinflammatory state characteristic of PVM infection.
The experimental studies performed in this manuscript utilize both IFNγ and IFNγR gene-deleted mice, which are in BALB/c and C57BL/6 background strains, respectively. PVM infection has been explored systematically in several inbred strains of mice by Anh and colleagues [39] who determined that the C57BL/6 strain is somewhat more resistant to infection than BALB/c, but that both of these inbred strains can ultimately succumb to the sequelae of severe disease. We have used both of these strains extensively for our studies (reviewed in [7][8][9]) and both respond to PVM infection with robust virus replication in lung tissue, granulocyte recruitment and local production of proinflammatory cytokines, including CCL3 and IFNγ; no systematic differences, other than the aforementioned susceptibility to infection, have been detected.
Both CCL3 and IFNγ have been detected in human studies and in mouse models of other severe respiratory virus infections, including avian influenza, SARS coronavirus, and human respiratory syncytial virus [17,18,[40][41][42][43][44][45][46][47], although the potential for interplay between these specific signaling pathways has not been considered previously. Our data suggest that that IFNγ and CCL3 signaling pathways, both crucial features of the response to pneumovirus infection, interact in a hierarchical fashion, as IFNγ does not elicit neutrophil recruitment on its own [ Figure  5A], but is crucial for CCL3 to function effectively. Interactions between IFNγ and CCL3 may occur at the level of signal transduction, or via alterations to the neutrophil itself. As has been documented clearly, CCL3 can function alone to induce changes in calcium concentration and chemotactic responses in mouse neutrophils in vitro [48]. The current literature on interactions of IFNγ with granulocytes was recently reviewed [49]. Among the possibilities that may address our findings, Hansen and Finbloom [50] reported that human neutrophils express IFNγ receptors and Bonecchi and colleagues [51] have shown that human neutrophils respond to IFNγ with increased expression of a variety of mediators and receptors, including the primary CCL3 receptor, CCR1. It is unclear whether mouse neutrophils respond in a similar fashion, and whether or not these defined molecular responses take place in vivo, although we have shown here that neutrophils from IFNγR gene-deleted mice express transcripts for CCR1 and mobilize intracellular calcium in response to CCL3 when examined ex vivo. We have not yet explored the possibility that the IFNγ coordinates neutrophil recruitment in response to CCL3 in a more indirect fashion, possibly via one or more intermediary cytokines. An example of this phenomenon was reported by Khader and colleagues [52], who demonstrated that Mycobacterium tuberculosis-infected dendritic cells from IL-12p40 genedeleted mice that were unresponsive to a CCL19 gradient were also overproducing the cytokine IL-10. Most intriguing, addition of IL-10 to wild-type dendritic cells reproduced the inhibited chemotaxis response.

Conclusion
In summary, we demonstrate here that CCL3, a proinflammatory mediator produced in response to RSV and shown to be a crucial in recruiting neutrophils in response Heterologous expression of CCL3 in mouse lungs Neutrophil recruitment in response to CCL3 is ablated in IFN-γ gene-deleted mice to the mouse pneumovirus, PVM, functions via a hierarchical relationship with IFNγ. Specifically, CCL3 recruits neutrophils to the lung in vivo only in coordination with IFNγ-mediated signaling pathways. The mechanism via which IFNγ modulates neutrophil responses to CCL3 is an intriguing subject for future exploration.

Microarray analysis
Generation of gene microrarray data was as described previously [12]. Data collected were evaluated using the microarray software analysis package Genespring GX 7.3. The 'find similar' function was used to inspect all 45,101 transcript profiles in order to detect kinetic profiles similar to that of IFNγ. The minimum correlation to be considered a similar profile was set at 0.900. The higher the correlation coefficient (maximum 1.000 for complete overlap), the more similar the gene expression profiles.

Mouse, virus and vector stocks
BALB/c and C57BL/6 mice were purchased from Taconic Laboratories (Germantown, NY and Rockville, MD). Homozygous IFNγ gene-deleted (IFNγ -/-) mice [53] on a BALB/c background and IFNγ receptor gene-deleted (IFNγR -/-) mice [54] on a C57BL/6 background were purchased from Jackson Laboratories, Bar Harbor, ME. All animal studies were performed as per approved protocols CHUA #634 (SUNY Upstate) or LAD 8E (NIAID). PVM strain J3666 was passaged, stored and quantitated as described previously [13]. Mice were anaesthetized and inoculated by intranasal challenge with 30 -100 plaque forming units (pfu) PVM also as previously described. For challenge with recombinant vectors (described as follows), dilutions of secondary stock aliquots of vCCL3 and vctrl (described in the section to follow) were prepared in RPMI cell culture medium. Under brief anaesthesia, mice were inoculated with 150 μl of stock (50 μl/dose × 3 doses) to achieve challenges of 1.0 -1.5 × 10 11 pfu per mouse. On days indicated, mice in each challenge group were sacrificed by cervical dislocation and bronchoalveolar lavage (BAL) fluid, total lung protein and total lung RNA were harvested. For some experiments, mice received 15 μg recombinant murine IFNγ (R&D Systems, Minneapolis, MN) diluted in tissue culture medium (RPMI + 10% fetal calf serum) or tissue culture medium (vehicle) via intraperitoneal injection one day prior to intranasal challenge with the vCCL3 or vctrl which yielded 323 ± 28 pg IFNγ/mg lung on day 4 post-inoculation.

Flow cytometric determination of IFN + NK and T cells in mouse lung tissue
Whole lungs of BALB/c mice (uninfected or day 6 after inoculation with 10 pfu PVM, n = 5 per datapoint) were cut into ~3 mm 3 pieces in HBSS buffer (Invitrogen) and pressed through a 100 micron cell strainer (BD Bio-sciences, San Jose, CA) to obtain single cell suspensions. Cells were suspended in RPMI-1640 medium supplemented with 10% fetal calf serum, 2 mM glutamine, 100 U/mL penicillin, 100 U/mL streptomycin, 50 μM 2-mercaptoethanol, 1 mM sodium pyruvate, and nonessential amino acids (all from Invitrogen) and incubated for 6 hrs at 37°C at a density of 1 × 10 6 cells/ml with 1 μM ionomycin, 20 ng/ml phorbol-12-myristate acetate (EMD Biosciences, San Diego, CA) and 10 μg/ml brefeldin A (Sigma-Aldrich Co., St. Louis, MO). DNAse I (Sigma) was added for 5 minutes and then cells were washed once and stained with violet LIVE/DEAD Fixable Dead Cell stain (Invitrogen) for 30 minutes on ice, washed in PBS, fixed in 4% PFA, and stored at -80°C until analysis. Intracellular cytokine staining was performed as described previously [55]. Cells were stained with I-Ad FITC, DX5-PE, CD3-PE-Cy5, CD4 PerCP/Cy5.5, IFNγ PE-Cy7, and CD8 APC-Cy7 (BD Biosciences) in PBS with 0.1% BSA, 0.1% saponin (Sigma) and 5% nonfat dry skim milk. Controls were stained with isotype matched antibodies. Samples were acquired with a 4-laser LSR II flow cytometer (BD Biosciences) and analyzed on FlowJo software (Tree Star, Inc., San Carlos, CA). Viable lymphocytes were identified by typical forward and side scatter and negative staining for LIVE/DEAD violet. T cells were identified as I-A -, CD3+ and either CD4 + or CD8 + as indicated; NK cells were identified as I-A -, CD3 -, DX5+. Quadrant statistical markers were based on corresponding isotype matched controls. Samples consisted of a known fraction (typically 1/4) of the cells obtained from a whole lung, and the entire sample was analyzed (typically 2 -4 × 10 5 events) yielding the absolute number of cells per lung.

Isolation of neutrophils from wild type and IFN R-/-mice
Neutrophils were isolated from wild type and IFNγR-/mice as described [56]. Briefly, 2 mL intraperitoneal injection of thioglycollate was administered and 4 hours later, mice were sacrificed and cells were harvested by peritoneal flush with 10 mL PBS (without calcium or magnesium). Cells were washed, red blood cells lysed with distilled water, and viability determined at >95% by trypan blue exclusion. Further isolation via Ficoll/ Hypaque density gradient centrifugation yielded neutrophil purities of 85 -99% as determined by modified Giemsa staining of cytospin preparations.

Construction vCCL3 and control (vctrl) overexpression vectors
Generation of overexpression vectors was accomplished using the commercially available AdEasy XL vector system (Stratgene, La Jolla CA) according to the manufacturer's instructions. Briefly, murine CCL3 (GenBank Accesion No. NM 011337) was ligated into the multiple cloning site of the shuttle vector, pShuttleCMV; the corresponding control plasmid, containing the β-galactosidase gene, was supplied with the kit. The constructs were linearized with Pme I and transformed into an E. coli strain, (BJ5183) which contains the replication-incompetent pAD-1 backbone. Transformants were selected for kanamycin resistance, and recombinants subsequently identified by restriction digestion. Once recombinants were identified, they were produced in bulk using the recombination-deficient bacterial strain, XL-10 Gold. Purified recombinant plasmid DNA was digested with Pac I to expose inverted terminal repeats and used to transfect AD-293 cells in which the deleted viral assembly genes are complemented in vivo. The resulting constructs, vctrl and vCCL3 were harvested from the transfected AD-293 cells when more than 90% of the monolayer exhibited cytopathic effects. Secondary stocks were produced in a similar fashion. Titration was performed by standard plaque assay. The concentration of secondary stocks reached titers of ~10 12 pfu/ml for each construct. Viral stocks were stored at -80°C prior to use.

Bronchoalveolar lavage (BAL) and differential cell counts
At time points indicated, BAL fluids were harvested from 5 mice by trans-tracheal instillation and removal of prechilled phosphate-buffered saline with 0.25% bovine serum albumin (BSA; 0.80 ml instillation with recovery of 0.5 to 0.6 ml per mouse). Neutrophil counts were determined by visual inspection of methanol-fixed cytospin preparations stained with modified Giemsa (DiffQuik, Fisher Scientific, Pittsburgh PA).

Chemokine and cytokine determinations
Concentrations of CCL3 and IFNγ were determined in BAL fluid (pg/ml) isolated as previously described [13] from five mice per datapoint, using commercially available ELISA kits (R&D Systems, Minneapolis, MN). Total protein was determined the Bradford colorimetric assay using bovine serum albumin standards

Gross and microscopic pathology and immunohistochemical detection of CCL3
Paraffin blocks of formalin-fixed lung tissue from mice challenged with vctrl or vCCL3 were paraffin-embedded and sectioned. Standard hematoxylin and eosin staining of formalin-fixed tissue was performed by American Histolabs (Gaithersburg, MD). To detect CCL3 protein expression in situ, slides were incubated with a 1:50 dilution of goat anti-CCL3 (R&D Systems, Minneapolis, MN) followed by a 1:400 dilution of biotinylated rabbit antigoat Ig and developing reagents (performed by Histoserv, Inc., Germantown, MD).

Virus titer
Quantitative reverse transcriptase PCR to document PVM titer in mouse lung tissue was as described previously [57]. Datapoints are presented as copies of PVM SH gene per 10 9 copies GAPDH.

Authors' contributions
CB contributed to the initial design of the CCL3 overexpression studies and carried out the experimental work.