key: cord-103592-lkngp2u6
authors: Bachmaier, Kurt; Stuart, Andrew; Hong, Zhigang; Tsukasaki, Yoshikazu; Singh, Abhalaxmi; Chakraborty, Sreeparna; Mukhopadhyay, Amitabha; Gao, Xiaopei; Maienschein-Cline, Mark; Kanteti, Prasad; Rehman, Jalees; Malik, Asrar B.
title: Selective Nanotherapeutic Targeting of the Neutrophil Subset Mediating Inflammatory Injury
date: 2020-07-02
journal: bioRxiv
DOI: 10.1101/2020.06.30.180927
sha: 
doc_id: 103592
cord_uid: lkngp2u6

Inflammatory tissue injury such as acute lung injury (ALI) is a disorder that leads to respiratory failure, a major cause of morbidity and mortality worldwide. Excessive neutrophil influx is a critical pathogenic factor in the development of ALI. Here, we identify the subset of neutrophils that is responsible for ALI and lethality in polymicrobial sepsis. The pro-inflammatory neutrophil subpopulation was characterized by its unique ability to endocytose albumin nanoparticles (ANP), upregulation of pro-inflammatory cytokines and chemokines as well as the excessive production of reactive oxygen species (ROS) in models of endotoxemia and septicemia. ANP delivery of the drug piceatannol, a spleen tyrosine kinase (Syk) inhibitor, to the susceptible subset of neutrophils, prevented ALI and mortality in mice subjected to polymicrobial infection. Targeted inhibition of Syk in ANP-susceptible neutrophils had no detrimental effect on neutrophil-dependent host defense because the subset of ANPlow neutrophils effectively controlled polymicrobial infection. The results show that neutrophil heterogeneity can be leveraged therapeutically to prevent ALI without compromising host defense.

Subsets of neutrophils differ markedly in their response to both homeostatic and inflammatory signals (1) . Neutrophil heterogeneity is apparent in the lungs of naïve mice, where a large proportion is marginated in the microvasculature, where they may function as immune sentinels, while other neutrophils circulate unimpeded (2) . Neutrophils are an essential component of the innate immune response to polymicrobial infection due to their ability to eliminate the infectious agents (3) . However, neutrophils can also become pathogenic in diseases by promoting excessive inflammation such as in the case of acute lung injury (ALI), a main cause of morbidity and mortality worldwide (4) . Excessive activation of neutrophils by bloodstream bacteria and their products, such as the bacterial endotoxin lipopolysaccharide (LPS), results in tissue damage and organ dysfunction (5) . Therapeutic efforts of curbing this excessive neutrophilic inflammation have been frustratingly ineffective (6) . Administration of nitric oxide, norepinephrine, low dose corticosteroids, prostaglandin E1, or recombinant activated protein C, when critically evaluated, did not significantly improve patient mortality (7) . Moreover, neutralizing key inflammatory mediators such as the cytokines TNF-α, and IL-1β, or reactive oxygen species (ROS), has also failed (8) . In experimental models, the elimination of neutrophils markedly decreases the severity of ALI (4) . On the other hand, there is the risk of compromising host defense in the setting of global neutrophil impairment and deterioration of pulmonary function during recovery from neutropenia (9) .

Targeting specific subsets of neutrophils could represent an optimal therapeutic approach if one could identify deleterious neutrophil subsets without compromising the subsets essential for host defense. The unique ability of neutrophils to rapidly change their phenotype and function according to changes in their microenvironment (10) (11) (12) (13) (14) is a manifestation of neutrophil heterogeneity, but the distinct roles of neutrophil subsets in the setting of sepsis or endotoxemia and not well understood.

We hypothesized that subsets of neutrophils are primarily responsible for the maladaptive hyperinflammatory response that causes ALI, multiple organ failure, and death. In the present study, we identified the subset of neutrophils that incorporated specially formulated albumin nanoparticles (ANP) as the subset that could be therapeutically targeted without impairing the elimination of bacteria in experimental septicemia.

Heterogeneous response of neutrophils to endotoxin and septicemia. After i.v. injection of albumin nanoparticles (ANP) to naive mice, we observed ANP-uptake in liver and spleen, whereas lungs, heart and kidney remained mostly free of ANP (Supplemental Figure 1) . In response to i.p.

challenge with the endotoxin of Gram-negative bacteria, lipopolysaccharide (LPS), uptake of i.v.injected ANP in heart, kidney, liver and spleen did not increase compared to naïve mice. In lungs, however, ANP uptake increased significantly after LPS challenge (Supplemental Figure 1 ). Ly6G + polymorphonuclear neutrophils (PMN) have the capacity to internalize ANP (15) . We next determined whether uptake of ANP in the lung was restricted to Ly6G + PMN and whether there was heterogeneity in the endocytosis of ANP among Ly6G + PMN. In response to i.p. LPS, only CD45 + leukocytes endocytosed i.v. injected ANP whereas parenchymal cells (CD45 neg ) did not ( Figure 1A ). ANP-endocytosis was restricted to Ly6G + PMN and largely absent in CD64 + monocytes/macrophages, NK1.1 + NK cells, or lymphocytes (data not shown). Pulmonary PMN endocytosed ANP in a bimodal manner, with one subset showing highly efficient uptake (ANP high ), and the other subset demonstrating minimal to no uptake (ANP low ) ( Figure 1A ). Bacterial endotoxins amplify the neutrophil activation in septicemic mice, leading to increased PMN sequestration in lungs where PMN release pro-inflammatory mediators and further enhance the recruitment of immune cells (16) . Using cecal ligation and puncture (CLP), a reproducible and clinically relevant mouse model of polymicrobial infection that causes ALI, we found that in naïve control mice after 2 sequential i.v. injections of ANP only ~4% of lung PMN endocytosed ANP as evidenced by ANP-specific fluorescence ( Figure 1B) . At 6h after a sham operation, laparotomy plus cecal ligation without puncture of the cecum, and 2 sequential i.v. injections of ANP, ANP high PMN increased to only ~11% ( Figure 1B) . Induction of severe polymicrobial sepsis by CLP, however, increased the frequency of ANP high lung cells 6-fold over baseline conditions to ~24%

( Figure 1B ). We consistently found that CD11b expression levels on PMN in peripheral blood, lung, and liver, were greater on ANP high PMN than on ANP low PMN ( Figure 1E ), and lung ANP high PMN showed the highest CD11b expression ( Figure 1E) , indicating a higher level of inflammatory activation of the ANP high PMN subset. Moreover, in septicemic mice, the percentages of ANP high PMN was significantly greater than in sham controls in blood, lung, and liver ( Figure 1F ), consistent with an increased pro-inflammatory state as well as increased adhesiveness of the ANP high PMN subset. This heterogeneity in CD11b activation on PMN suggested that susceptibility to ANPendocytosis delineated distinct subsets of pulmonary PMN.

To define the differences between the PMN subsets, we next analyzed whether ANP high PMN had a transcriptomic profile different from ANP low PMN. We performed an unbiased analysis of lung PMN transcriptomic profiles using RNA-Seq.

We challenged mice with i.p. injections of LPS or saline and administered ANP i.v. 5h later. At 1h

after ANP injection, we euthanized mice and harvested PMN from single cell suspension of their lungs and sorted the Ly6G + PMN by flow cytometry according to their ANP uptake into ANP low and ANP high PMN. Immediately after sorting, we prepared PMN mRNA for RNA-Seq analysis. We generated a heat map and dendrogram ( Figure 2A ) to represent the normalized PMN gene expression data. We found that the biological replicates clustered into 4 groups with distinct transcriptomic profiles; i.e., the mRNA profiles defined PMN from LPS-challenged or saline-injected mice, and were distinct in ANP low and ANP high PMN ( Figure 2A ). Using MetaCore Pathway analysis to identify pathways that were different between ANP high PMN and ANP low PMN, we found that the pathways regulating immune response and immune cell migration were significantly overrepresented in the ANP high PMN (Supplemental Table) . Pathways containing chemokine receptors were significantly enriched in ANP high PMN, consistent with the concept that PMN heterogeneity is a function of differential PMN trafficking into tissue presumably facilitated by different chemoreceptor expression. We also found that chemokine receptors were over-represented 6. To identify the chemokine receptors for each PMN subset, we generated separate heatmaps for chemokine receptors, plotting all genes with CPM > 0.25 (10 reads at sequencing depth of 40M reads) regardless of differential expression levels. In naïve mice, ANP high PMN showed relative over-expression of chemokine receptors Cxcr3, Cxcr4, and Ccrl2 ( Figure 2B ). In LPSchallenged mice, ANP high PMN showed relative over-expression of chemokine receptors Ccr1, Ccr5, Ccr7, Ccr10, Ccrl2, Cxcr4 and Cxcr5 ( Figure 2C ). We next assessed the expression of chemokines in ANP high PMN and ANP low PMN. In saline injected control mice, ANP high PMN were significantly enriched for the expression of the chemokines Ccl3, Ccl4, and Cxcl3 ( Figure 2D ). In LPS-challenged mice, ANP high PMN demonstrated relative over-expression of the chemokines Ccl3, Ccl4, Ccl5, Ccl6, Ccl17, Cxcl1, Cxcl2, Cxcl3, Cxcl16 ( Figure 2E ). Of note, mice were only exposed to ANP for the last hour of the 6h LPS-challenge, and the differences in gene expression between ANP high PMN and ANP low PMN in LPS challenged mice were far greater than those in the naïve mice, indicating that the ANP uptake itself likely did not affect the gene expression profiles.

These RNA-Seq data unequivocally demonstrated the existence of lung PMN subsets with a distinct response to the inflammatory stimulus LPS. Based on our RNA-Seq data and MetaCore Pathway analysis, we selected a group of chemokine receptors and cytokines to determine the kinetics of their expression after LPS-stimulation. We performed this independent validation by quantitative PCR (qPCR) and flow cytometry. We found that mRNA expression of the chemokine receptor Ccr1 was significantly greater in ANP high PMN than in ANP low PMN at 3h, 6h, and 12h after LPS challenge ( Figure 3A ). Importantly, we found that CCR1 receptor cell surface expression, consistent with the mRNA data, was significantly greater on lung ANP high PMN than in ANP low PMN before and 3h, 6h, and 12h after LPS stimulation ( Figure 3B ). CXCR2 expression was reduced at 6h after LPS-stimulation compared to 3h stimulation, CXCR4 receptor cell surface expression increased at 3h after LPS-stimulation and was greater in ANP high than in ANP low PMN, and decreased to expression levela of unstimulated PMN thereafter ( Figure 3B ). These data demonstrate that the observed mRNA expression heterogeneity translated into cell surface protein expression heterogeneity. The mRNA levels of the chemokines Ccl4, Ccl3, Cxcl2, Cxcl3 ( Figure 3C -F) were significantly greater in ANP high PMN than in ANP low PMN 3h, 6h, and 12h after in vivo LPS challenge. Ccl4 ( Figure 3C ) and Ccl3 ( Figure 3D ) expression, in particular, was vastly greater in ANP high than in ANP low PMN, suggesting that ANP high PMN are specialized cells of inflammation, which recruit and activate additional PMN. In addition, heterodimers of CCL3 and CCL4 are known to attract monocytes/macrophages (17) . The cytokine IL-1β is essential for antibacterial function and expression of IL1β was induced ~21-fold in ANP low PMN 3h after LPS challenge compared to ANP low PMN from saline injected control mice ( Figure 3G ); in ANP high PMN, IL1β was induced ~78fold over ANP high PMN from saline injected control mice ( Figure 3H ). Expression of the pleiotropic cytokine IL-15 expression was significantly greater in the ANP high than in the ANP low PMN in lungs 3h, 6h, and 12h after LPS challenge ( Figure 3H ). These data demonstrate that pulmonary ANP high PMN can markedly amplify the inflammatory response.

We next determined whether adoptively transferring ANP high PMN from donors into syngeneic recipient mice would induce inflammation in these mice. Donor BALB/c mice were challenged with a lethal dose of LPS [30mg/kg] and injected with two doses of ANP labeled with the stable fluorochrome AF647 at 1h and 2h after LPS challenge ( Figure 4A ). At 3h after LPS challenge, donor mice were euthanized and lung single cell suspensions were prepared for flow cytometric sorting into ANP high and ANP low neutrophils ( Figure 4B ). Syngeneic recipient mice were injected i.v. with 8x10 5 pulmonary ANP high PMN or, as controls, with an equal number of pulmonary ANP low PMN from the same donors (three donors per recipient mouse were required to achieve the necessary cell number). At 2h prior to transfer of donor cells, recipient mice were treated with a sublethal dose of LPS [1mg/kg] to activate their endothelium, a prerequisite for initiating neutrophilic lung inflammation (18) . At 20h after the adoptive transfer, we assessed lung inflammation in recipient mice ( Figure 4A ). We found ANP + Ly6G + PMN in lungs of recipient mice, confirming successful transfer of donor cells to recipient mice and their homing to the lung ( Figure 4C ).

Transfer of donor ANP high PMN significantly increased lung inflammation in the recipient mice when compared to mice that received ANP low cells ( Figure 4C ). Moreover, after transfer of ANP high PMN, lung Ly6G + PMN produced more ROS when compared to controls receiving ANP low PMN ( Figure 3D ). Because ROS induce tissue inflammation (12, 19) , and ANP high cells are carriers of large amounts of mRNA for inflammatory cytokines and chemokines ( Figure 3 ), we measured the inflammatory mediators IL-1β and CXCL2 in lung tissue extracts. IL-1β, released during activation of NLRP3 inflammasome, mediates tissue injury, whereas the chemokine CXCL2 amplifies the inflammatory cycle by attracting additional pro-inflammatory neutrophils (20) (21) (22) . Mice receiving ANP high PMN had significantly greater concentrations of IL-1β ( Figure 4E ) and CXCL2 ( Figure 4F ) in their lungs than recipients of ANP low PMN. These data demonstrated the intrinsic ability of ANP high PMN to promote lung inflammation.

Firm PMN adhesion on microvascular endothelium, induced by endotoxin or bacteremia, upregulates Mac-1 (a heterodimer of CD11b and β2-integrin CD18), contributing to maximal activation of PMN (23). Syk activity is required for β2-integrin-mediated neutrophil activation (24) . Given our data above, we reasoned that inhibiting integrin signaling specifically in the subset of ANP high PMN would reduce lung inflammation in the polymicrobial sepsis model. We thus used the drug piceatannol, a Syk inhibitor (25, 26) , that is readily incorporated into ANP due to its poor water solubility (15, 27) , to inhibit Syk-mediated β2-integrin-dependent neutrophil adhesion. We found that therapeutic administration of piceatannol loaded ANP high PMN protected CD1 mice from lethal polymicrobial sepsis ( Figure 5A ). Treatment with two i.v. injections of piceatannol incorporated into ANP (PANP), 2h and 4h after CLP, significantly reduced mortality of mice when compared to control groups treated with ANP without any drug after challenge with CLP ( Figure 5A ). Treatment using PANP reduced CLP lethality to the rate of sham-operated (laparotomy plus cecal ligation without puncture of the cecum) mice ( Figure 5A ). Injections of PANP alone had no effect on the survival rate compared to saline injected controls ( Figure 5A ). CLP challenged mice treated with ANP, without any drug, had the same mortality rate as saline-injected controls ( Figure 5A ), demonstrating that targeting specifically the ANP high subset of PMN is sufficient to prevent CLP-induced mortality.

Similarly, in the absence of polymicrobial infection but after i.p. challenge with a lethal dose of the endotoxin LPS (LD80), mice treated with 2 sequential i.v. injections of PANP 1h and 2h after LPS challenge, showed significantly reduced mortality when compared to control mice injected with ANP alone ( Figure 5B ). Reduced mortality after PANP treatment was correlated with the presence of significantly fewer highly inflammatory CD11b high CD45 high PMN ( Figure 5C ), and reduced CD11b expression on lung PMN when compared to ANP-treated controls ( Figure 5D ). Furthermore, while CD11b expression on PMN was reduced by PANP treatment in the lungs, in peripheral blood PMN CD11b surface-expression was higher when compared to PMN from ANP-treated mice ( Figure   5D ). To test whether augmented CD11b expression on PMN in peripheral blood was a consequence of PANP-induced PMN trafficking from the lung to the peripheral blood, we used two-photon microscopy to visualize PMN trafficking in the lung in vivo (28) . We determined the number of Ly6G + PMN in microvasculature and velocity of their migration through the microvasculature in lungs (Video, Figure 5E ). In LPS challenged CD1 mice, the number of PMN increased, as measured by PMN-specific fluorescence (Video). Treatment of PMN with PANP, however, significantly increased the velocity of Ly6G + PMN in the lung microvasculature and reduced the number of Ly6G + PMN as compared to ANP treated controls (Movie, Figure 5E ,F). Cell targeted treatment of ANP high PMN by inhibiting β2-integrin signaling, and accelerated the transit of PMN through lungs, and thus reduced exposure time of lung tissue to noxious PMN-derived mediators.

ROS production is a potent mediator of tissue damage (12, 19) . We found that ANP high cells were characterized by high ROS production ( Figure 4D ). Syk, whose enzymatic activity is the cellular target of piceatannol (25), is required for integrin-mediated neutrophilic superoxide production (24).

We measured ROS production by bone marrow Ly6G + PMN in vitro (Supplemental Figure 2 ). Bone marrow PMN responded to stimulation with the bacterial peptide fMLP with ROS production (Supplemental Figure 2) . PMN with higher uptake of PANP showed greater reduction in ROS production (Supplemental Figure 2) . Moreover, the delivery of piceatannol via PANP increased drug efficacy by orders of magnitude when compared to free drug because of its incorporation in the toxic PMN subset (Supplemental Figure 2) . We therefore examined whether PANP treatment reduced superoxide production by lung PMN of endotoxemic mice. We challenged mice with a lethal dose of LPS and analyzed the production of ROS by the lung Ly6G + PMN ex vivo. We found that ANP high PMN had significantly higher intracellular ROS levels than ANP low PMN ( Figure 6A ).

PANP treatment, however, almost completely blocked ROS production in these cells ( Figure 6A ).

These data demonstrated that ANP high PMN are largely responsible for ROS production by lung inflammatory cell in endotoxemia because targeted treatment via PANP markedly reduced ROS production.

Mice doubly deficient for NADPH oxidase and iNOS (gp91phox −/− nos2 −/− ) develop spontaneous infections (29) . ROS production and complementary NO production by PMN are essential to control host microbial diversity and microbial infection (29) , and functional lung PMN are essential for the task of clearing bloodstream bacteria because resident macrophages in liver and spleen alone are insufficient (30, 31) . We therefore determined the effects of PANP treatment on the bacterial burden of CD1 mice in the CLP model of polymicrobial infection. We found no exacerbation or amelioration of the bacterial burden as a result of PANP treatment when compared to ANP treatment ( Figure   6B ), suggesting that selective inhibition of integrin signaling in the ANP high PMN subset did not compromise bacterial elimination. Two consecutive i.v. injections of PANP, given 2h and 4h after CLP, did not increase bacteremia when compared to ANP injected controls ( Figure 6B ). The bacterial burden of lungs, livers, and spleens of bacteremic mice was similar between PANPtreated mice and ANP-treated controls ( Figure 6B ). PMN-dependent antimicrobial function was fully preserved after PANP treatment, suggesting that antimicrobial functions, ingestion and elimination of bacteria are mainly performed by ANP low PMN (Supplemental Figure 3 ).

Because PANP treatment did not weaken anti-microbial resistance, we analyzed parameters of tissue inflammation and damage We measured crucial inflammatory mediators, IL-1β, and CXCL2.

In lung tissue extracts of mice subjected to CLP. We found a substantial reduction in the concentration of IL-1β and CXCL2 after PANP treatment when compared to ANP treated controls ( Figure 6C ). We next measured nitrotyrosine formation in lungs and livers of septicemic mice. In the experimental mice as well as in septicemic patients, activated lung myeloid cells, inflammatory or resident, and epithelial type II cells, release both NO and superoxide which react to form peroxynitrite, a potent oxidant causing tissue damage (32) . Peroxynitrite (ONOO − ), but not NO or superoxide alone, nitrates tyrosine residues (32) . We observed that nitrotyrosine-specific staining in inflammatory and parenchymal cells was significantly reduced in lungs and livers of mice treated with PANP when compared to ANP-treated controls ( Figure 6D ,E). These data suggest that antimicrobial function and tissue damaging function are performed by distinct subsets of PMN.

We next determined the effect of PANP treatment on pulmonary edema which is a characteristic feature of inflammatory lung injury. A marked increase in lung wet-to-dry weight ratio is indicative of breakdown of the alveolar capillary barriers, the hallmark of ALI. Pneumonia is the most common cause of ALI in patients (33) and also the most common cause of sepsis (8) . In the model of pneumonia induced by i.t. instillation of live P. aeruginosa bacteria, PANP treatment significantly reduced pulmonary edema when compared to treatment with the control ANP ( Figure 6F ).

Photomicrographs of lung or liver sections from septicemic mice treated with ANP or PANP showing nitrotyrosine formation. Paraffin embedded sections were stained with specific Ab to nitrotyrosine (red staining), and with DAPI to visualize nuclei (blue staining). Polymicrobial sepsis was induced by CLP, mice were treated with PANP or ANP 2h and 4h after challenge and were sacrificed for tissue processing and staining 18h after challenge. Bar Furthermore, treatment with PANP significantly reduced pulmonary edema in endotoxemic or septicemic mice when compared to lungs from ANP treated controls ( Figure 6F) . A reduction of tissue damage, because of reduced lung inflammation, could be the proximate cause of reduced inflammatory lung injury after treatment. Measuring a markers of overall cell damage, lactic dehydrogenase (LDH) (34) , revealed that the polymicrobial sepsis-induced increased serum activity of LDH was significantly reduced by PANP treatment when compared to ANP treated controls ( Figure 6G ). In addition, hepatocyte-specific sorbitol dehydrogenase (SDH) activity, a marker of hepatocyte damage (35) , was also significantly reduced by PANP treatment of septicemic mice ( Figure 6G ). Endocytosis of ANP delineates two distinct subsets of PMN. ANP high PMN cause tissue damage whereas ANP low PMN control microbial infection ( Figure 6H ).

Targeting neutrophilic inflammation that is pathogenic in ALI remains an important unmet clinical need. Our observation that neutrophils primed by bacterial infections or bacterial derived products exhibit heterogeneity in their capacity to endocytose ANP lead to the discovery of two distinct neutrophilic subsets, one that causes inflammatory injury and one that controls microbial infection.

Functional and phenotypic PMN heterogeneity (36) (37) (38) (39) (40) led us to test the hypothesis that immunopathology and severe tissue inflammation of endotoxemia and septicemia can be treated by targeting a distinct neutrophilic subset. Increased expression of chemokines and chemokine receptors in ANP high PMN was consistent with their role in promoting tissue inflammation. Several of the chemokines such as CCL3 and CCL4 or CXCL2 and CXCL3 are members of the macrophage inflammatory protein family and are typically thought to be released by macrophages to increase the influx of pro-inflammatory cells such as neutrophils (41) . Our data suggest that a subset of neutrophils might be a substantial source of these inflammatory proteins in models of ALI.

All cells need to express genes that are required for their intrinsic functions, whereas production of secreted factors can be delegated to subsets (42) . Our results indicate that a specific subset of phenotypic and functionally distinct neutrophils is responsible for lethality in experimental polymicrobial sepsis. ANP high PMN were characterized by higher expression of the chemokine receptors for the ligands released following LPS activation such as CCR1 and CCR5 (the receptors for CCL3) which could point to a possible positive feedback loop in which inflammatory ANP high PMN attract additional inflammatory PMN, and thus actively promote a vicious cycle of hyperinflammation and tissue injury (43) . Given their phenotypic and functional profile, ANP high PMN might play a pathogenic role in COVID-19, the disease caused by coronavirus SARS-Cov-2 (44) .

The main cause of COVID-19-mortality is acute respiratory failure (45) . In patients with severe COVID-19, activated neutrophils, recruited to the pulmonary microvasculature, produce histotoxic mediators including ROS (46) . Activation of neutrophils might contribute to cytokine release syndrome ("cytokine storm") that characterizes severe COVID-19 disease (47) . Therapy targeting ANP high PMN might prevent a patient's hyperinflammatory response to SARS-Cov-2 without weakening the antiviral response.

It has been shown that the incorporation of denatured albumin beads depends on Mac-1 expression (48) . ANP-incorporation, by contrast, is Mac-1-independent (15) suggesting a distinct molecular mechanism of ANP-endocytosis. "Aged neutrophils", were first described in vitro as functionally deficient (49) and have been subsequently shown to promote disease in vivo in models of sickle-cell disease or endotoxin-induced septic shock (50) . While "Aged neutrophils" home to the bone marrow under steady-state conditions ANP high PMN do not home to the bone marrow under steady-state conditions and markers of aged peripheral blood neutrophils, CXCR4, CXCL2, CD62L, and TLR4, do not distinguish ANP high from ANP low PMN under steady-state or inflammatory conditions in the lung (Figure 2 ). These findings thus show that the previously described subset of aged neutrophils is distinct from the ANP high PMN subset we identified.

Administration of ANP carrying piceatannol, a Syk inhibitor, dramatically improved survival in polymicrobial sepsis but, critically, did not increase the host's bacterial burden. Syk function is required for the essential antibacterial functions of neutrophils (51) . We found that ANP low PMN are more efficient in ingestion and elimination of E. coli bacteria in vitro than ANP high PMN and inhibition of Syk in ANP high PMN does not impair control of polymicrobial infection in vivo. By therapeutically leveraging the preferential ANP uptake of the inflammatory PMN subset, we succeeded in limiting immunopathology caused by polymicrobial infection. Pathogens adapt to host resistance mechanisms, but not host tolerance mechanisms (52) . It is conceivable that an ANP-based approach of targeting toxic subset of neutrophils, by improving host tolerance, might assist the host in combating antibiotic-resistant microbial infections (53) . ANP could also be used to deliver compounds other than piceatannol or to specifically deliver siRNAs or microRNAs. Furthermore, ANP-based therapy might be useful in other forms of neutrophilic injury; e.g., ischemia reperfusion injury of the myocardium or the kidney (54) .

The ability to tolerate pathogens in experimental polymicrobial sepsis was greatly strengthened by targeting ANP high PMN with a drug that accelerates the velocity of pulmonary PMN and abrogates their ROS production. PANP did not target β2-integrins directly (15) but mitigated downstream β2-integrin signaling, and are thus more likely to be effective when β2-integrin are already engaged; i.e., in the lung microvasculature of septic patients (55) . Further potential improvement of ANP-based therapies over the current standard treatments of septic patients (56) lies in the precision targeting of the relevant pathogenic subset of neutrophils. Earlier efforts to neutralize reactive oxygen species , to use antibodies against key inflammatory cytokines such as TNF-α, IL-1β, or to inhibit the endotoxin LPS (8, (57) (58) (59) have failed to reduce mortality associated with ALI/ARDS possibly because they did not discriminate between PMN subsets and may have compromised both host defense (resistance) and tissue repair (tolerance) (33) . Compared with the therapeutic administration of exogenous cells, generated from donors or from the patients themselves, for example, mesenchymal stromal/stem cells (MSCs) (60), ANP-based therapy has the advantage of targeting the host's endogenous cells dependent on their pathogenic activation.

One limitation of our approach is that we cannot distinguish whether the specification into ANP high PMN occurs in the bone marrow prior to egress of PMN into circulation or whether it occurs in the tissue itself. It is also possible that the ANP high state characterized by upregulation of chemokine receptors and chemokine ligands is reversible and that PMN can transition between these states, in response to environmental cues even during the short life-span of neutrophils (61) .

The present data support the concept that the generation of heterogeneous PMN subpopulations evolved as a host defense mechanism to avoid an indiscriminate response by all PMN to septicemia (42) . (1, 62) . We demonstrate here that neutrophil heterogeneity can be effectively leveraged for a nanoparticle-based therapy.

Mice. We used outbred male CD 1 mice, at a body weight ranging from 34g to 38g and, for adoptive transfer experiments, male inbred BALB/c mice between 8 and 10 wk of age. Mice were treated in accordance with the NIH Guide for the Care and Use of Laboratory Animals and UIC animal care committee's regulations. All procedures were approved by the UIC IACUC.

Synthesis of uniform-sized spheric albumin nanoparticles. ANP and PANP, synthesized as described (15) , were of consistent hydrodynamic size (120nm ± 28nm diameter and zeta potential (-27± 5.48mV) distribution. We injected i.v. 8.3mg/kg body weight of ANP, or of ANP loaded with 8.9µM piceatannol (PANP) 1h and 2h after challenge with LPS or 2h and 4h after CLP.

Flow cytometry and cell sorting. Single cell suspensions were prepared as described (63) Tissue damage markers. The activity, LDH and SDH was determined using commercial kits according to manufacturers' instructions. Histopathology was evaluated in sections from paraffin embedded or frozen tissues using specific antibodies to nitrotyrosine as described (64) .

Quantification of hydrogen peroxide production. We measured hydrogen peroxide production using the Amplex Red Hydrogen Peroxide Kit (Invitrogen) following the manufacturer's instructions. 

Heterogeneity of neutrophils

Neutrophil kinetics in health and disease

The role of neutrophils in immune dysfunction during severe inflammation

Neutrophils and acute lung injury

Sepsis: Current Dogma and New Perspectives

Surviving Sepsis Campaign: International Guidelines for Management of Severe Sepsis and Septic Shock: 2012

Drotrecogin Alfa (Activated) in Adults with Septic Shock

Sepsis: pathophysiology and clinical management

Respiratory status deterioration during G-CSF-induced neutropenia recovery

Phenotypic and functional change of cytokine-activated neutrophils: inflammatory neutrophils are heterogeneous and enhance adaptive immune responses

Getting to the site of inflammation: the leukocyte adhesion cascade updated

Neutrophil heterogeneity: implications for homeostasis and pathogenesis

Neutrophils in Homeostasis

Prevention of vascular inflammation by nanoparticle targeting of adherent neutrophils

Identification of Human Macrophage Inflammatory Proteins 1α and 1β as a Native Secreted Heterodimer

Endothelium-derived toll-like receptor-4 is the key molecule in LPS-induced neutrophil sequestration into lungs

Reactive Oxygen Species in Inflammation and Tissue Injury

Critical role for CXCR2 and CXCR2 ligands during the pathogenesis of ventilator-induced lung injury

Critical role of endothelial CXCR2 in LPS-induced neutrophil migration into the lung

Caspase-11-mediated endothelial pyroptosis underlies endotoxemia-induced lung injury

Annual Review of Immunology

Syk is required for integrin signaling in neutrophils

Piceatannol, a Syk-selective tyrosine kinase inhibitor, attenuated antigen challenge of guinea pig airways in vitro

PICEATANNOL (3,4,3',5'-TETRAHYDROXY-TRANS-STILBENE) IS A NATURALLY-OCCURRING PROTEIN-TYROSINE KINASE INHIBITOR

Stabilized imaging of immune surveillance in the mouse lung

Phenotype of mice and macrophages deficient in both phagocyte oxidase and inducible nitric oxide synthase

The lung is a host defense niche for immediate neutrophil-mediated vascular protection

Outcomes of bacteremia in patients with cancer and neutropenia: observations from two decades of epidemiological and clinical trials

The acute respiratory distress syndrome

Prognostic Significance of Elevated Serum Lactate Dehydrogenase (LDH) in Patients with Severe Sepsis

Pneumotoxicity and hepatotoxicity of styrene and styrene oxide

The mercurial nature of neutrophils: still an enigma in ARDS?

Neutrophils: Between Host Defence, Immune Modulation, and Tissue Injury

Social networking of human neutrophils within the immune system

Neutrophil heterogeneity in health and disease: a revitalized avenue in inflammation and immunity

Neutrophil recruitment and function in health and inflammation

Role of inflammatory mediators in the pathophysiology of acute respiratory distress syndrome

Desynchronization of the molecular clock contributes to the heterogeneity of the inflammatory response

Monocyte recruitment during infection and inflammation

The SARS-CoV-2 outbreak: What we know

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China

Cytokine release syndrome in severe COVID-19

The cytokine storm in COVID-19: An overview of the involvement of the chemokine/chemokine-receptor system

Heterotypic interactions enabled by polarized neutrophil microdomains mediate thromboinflammatory injury

Impairment of function in aging neutrophils is associated with apoptosis

Neutrophil ageing is regulated by the microbiome

Neutrophil-specific deletion of Syk kinase results in reduced host defense to bacterial infection

Disease Tolerance as a Defense Strategy

Two ways to survive infection: what resistance and tolerance can teach us about treating infectious diseases

Involvement of neutrophils in the pathogenesis of lethal myocardial reperfusion injury

SELECTED TREATMENT STRATEGIES FOR SEPTIC SHOCK BASED ON PROPOSED MECHANISMS OF PATHOGENESIS

Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock

Effect of eritoran, an antagonist of MD2-TLR4, on mortality in patients with severe sepsis: the ACCESS randomized trial

A randomized, double-blind, placebo-controlled trial of TAK-242 for the treatment of severe sepsis

Anti-tumor necrosis factor therapy in sepsis: Update on clinical trials and lessons learned

Mesenchymal stem (stromal) cells for treatment of ARDS: a phase 1 clinical trial

Selective roles for tolllike receptor (TLR)2 and TLR4 in the regulation of neutrophil activation and life span

Neutrophil Diversity in Health and Disease

E3 ubiquitin ligase Cblb regulates the acute inflammatory response underlying lung injury

iNOS expression and nitrotyrosine formation in the myocardium in response to inflammation is controlled by the interferon regulatory transcription factor 1

Measurement of Oxidative Burst in Neutrophils

FLOW CYTOMETRIC ANALYSIS OF THE GRANULOCYTE RESPIRATORY BURST -A COMPARISON STUDY OF FLUORESCENT-PROBES

STAR: ultrafast universal RNA-seq aligner

featureCounts: an efficient general purpose program for assigning sequence reads to genomic features

Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation

edgeR: a Bioconductor package for differential expression analysis of digital gene expression data

Controlling the false discovery rate: a practical and powerful approach to multiple testing

Analyzing real-time PCR data by the comparative CT method

CD40(TNFRSF5), c-F 2

Pyk2(FAK2), I-kB, MHC class II

TGF-beta 1, L-selectin

Ubiquitin, TAP1 (PSF1), Beta-catenin, AP-1, IRF9, RSAD2, Lck, PKC-theta

IL-1 beta, Fc gamma RI, TNF-R2, HLA-DRB1, TGF-beta, MHC class II 2.147E-09 2

Fc gamma RI, L-selectin, PECAM1, MHC class II, CCR5, CD 2.070E-08 2

CCL5, IL-15, IL-2R beta chain, IL-1 beta, AP-1, IL7RA, CCL17, CCR7, FKHR 6.744E-08 5

094E-08 5.814E-06 13 MHC class II alpha chain, IL-1 beta, I-kB

CCL5, IL-1 beta, TGF-beta 1, IL-1 alpha, Caspase-7, PD-L2, IL1RN, IL1R2, 1.341E-07 9

IL-1 beta, I-kB, TGF-beta 1, MHC class II

PSMB9, TAP1 (PSF1)

IL-1 beta, MHC class II, CD40(TNFRSF5), CCR7, IL-18, p38 MAPK, C 2.987E-07 1.535E-05 10 IL-1 beta, TNF-R2, MHC class II, IL-18

IL-1 beta, I-kB, TPL2(MAP3K8), MHC class II

c-FLIP(Long)

MHC class II, PD-L2, SLAM, CD40(TNFRSF5), CD137(TNFRSF9)

COX-2 (PTGS2), JAK2, MHC class II, Lck, CD40(TNFRSF5)

MHC class II, CD40(TNFRSF5)

093E-07 2.922E-05 12 TGF-beta 1, c-IAP2, c-FLIP(Short)

GRO-2, IL-1 beta, I-kB

IL-2R beta chain, I-kB, sIL-15RA, IL-15RA, Lck, NIK(MAP3K14)

Cyclin D2, BFL1, NIK(MAP3K14), CD137(TNFRSF9), p38 MAPK

IL-1 beta, I-kB

S1P1 receptor, TGF-beta 1, CD94, TIE2, PD-L2

IKK-gamma, IL-1 beta

IL-2R beta chain, JAK2, c-Myc, IL-2R alpha chain, Granzyme 2.285E-06 6

IL-18, Caspas 2.406E-06 6

IL-2R beta chain, IL-1 beta, sIL-15RA, MHC class II

COX-2 (PTGS2), PGE2R2, IKK-gamma, IL-1 beta, I-kB, TGF-beta 1, MHC cl PTPN22, CD40(TNFRSF5), c-FLIP

NF-AT2(NFATC1)

Lck, CD40(TNFRSF5), PKC-theta, NIK(MAP3K14)

RIG--DRB1, TGF-beta, MHC class II, PTPN22, CD40(TNFRSF5), CSF1, IL-18, MHC class II beta chain, IL-2R alpha chain

lass II, KLF2, SLAM, CD40(TNFRSF5), PKC-theta, NIK(MAP3K14), FKHR, Bcl-6, IRF4, MALT1, CD86, IFN-gamma, CD4, BLIMP1 (PR -1, IL7RA, CCL17, CCR7, FKHR, PSGL-1, CCR10, NF-AT2(NFATC1), IFN-gamma 0, CD3, LAT, ITK, p38 MAPK, NF-AT2(NFATC1)

CD40(TNFRSF5), CCR7, MHC class II beta chain, p38 MAPK, ASK1 (MAP3K5), CD86, c-Jun spase-7, PD-L2, IL1RN, IL1R2, CSF1, IL-18, Endothelin-1, Caspase-1, MSR1, p38 MAPK, PD-L1, AML1 (RUNX1) IL-2R alpha chain, CD3 zeta, NF-AT2(NFATC1)

MHC class II beta chain, ZAP70, CD3, LAT, ETS1, Aiolos, CD3 zeta, NF-AT2(NFATC1)

RAR-alpha/RXR-alpha

NIK(MAP3K14), FKHR, Bcl-6, CD3, MALT1, IL-2R alpha chain, NF-AT2(NFATC1), CXCR4, IFN-gamma

IL-18, p38 MAPK, CD86, CXCR4, IFN-gamma ase-1, Granzyme B, CD86, IFN-gamma, FasR(CD95)

MHC class II, IRAK1/2, AP-1, CD40(TNFRSF5), HSP70, p38 MAPK, CD86, c-Jun

CD244, CD30(TNFRSF8), CD86, PD-L1 40(TNFRSF5), IL13RA1, ZAP70, CD3, LAT, ITK, NF-AT2(NFATC1), NF-AT, FasR(CD95) , Caspase-7, DR4(TNFRSF10A), c-IAP1, tBid, Bcl-2, Bid -beta 1, MHC class II, CCR5, CD40(TNFRSF5), CCR7, CSF1, CMKLR1, M-CSF receptor

D40(TNFRSF5), SMAD3, TIEG1, IL-2R alpha chain, NF-AT2(NFATC1)

HGF receptor (Met), c-IAP1, tBid

NFKBIA, IFN-gamma, Bcl-2, c-Jun

CD3, ITK, NF-AT2(NFATC1), VAV-1, IFN-gamma

Tcf(Lef), Dsh, Frizzled c, IL-2R alpha chain, Granzyme B, Bcl-2

USP18 reg class IB (p101), HIP1, ITK, p38 MAPK, CXCR4, c-Jun, VIL2 (ezrin) VAV-1, CD4 sl2, c-Jun, FasR(CD95)

VAV-1, IFN-gamma, Bcl-2 p38 MAPK

T2(NFATC1), CXCR4, c-Jun, PD-L1 , p38 MAPK, M-CSF receptor, CD86, SOCS3, IFN-gamma

Supplemental Figure 1 . ANP Biodistribution Measured by IVIS In naïve mice, i.v.-injected ANPs were mostly found to liver and spleen. After i.p. administration of the endotoxin LPS [30mg/kg], ANP appeared prominently in lungs, and remained visible in liver and in spleen. Albumin was labeled with vivotag 800 (PerkinElmer) and then formed into ANPs. Organs were harvested 4 h after 150µl tail vein injection of 2 mg/ml vivotag 800 labeled ANPs. Epi-fluorescence radiance scale corresponding to images. Fluorescence was measured by a Xenogen IVIS Spectrum (Caliper Life Sciences) and images were processed with Living Image software (ver. 4.3.1). An excitation filter of 785nm and emission filter of 820nm with 120 sec exposure times were used for all experiments. Representative data obtained from at least 3 mice per treatment group.

Flow cytometry histograms of bone marrow PMN stimulated with LPS [1mg/ml] and incubated with PANP, at the indicated doses, for 30 min at 37°C, and then processed for flow cytometry. Representative data obtained from at least 3 mice per treatment group PANP treatment reduces superoxide production by PMN in response to fMLP stimulation. PAMP reduced fMLP stimulated ROS production in a dose dependent manner. Bone marrow PMN were incubated with the indicated dises of PANP and stimulated with fMLP for the indicated time. ROS production was measured using an Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit. Representative data obtained from at least 3 mice per treatment group.