n e w s a n d v i e w s

cell transplantation3. Other effects of G-CSF treatment have remained less clear. One large retrospective study showed an increased risk of GVHD when G-CSF was given after marrow transplantation but not after mobilized blood cell transplantation4. Two other studies, how-ever, showed no association between treatment with G-CSF and risk of GVHD5,6.

The study by Morris et al.1 illuminates these conflicting results. The key observation is that TBI increases expression of G-CSF receptors by recipient antigen-presenting cells (Fig. 1). This effect was not observed when methods other than TBI were used for conditioning. Enhanced signaling through G-CSF receptors increases expression of CD1d, which functions as a ligand for antigen recognition by invariant natural killer T cells in the recipient. Activated invariant natural killer T cells produce interferon-γ, which stimulates recipient dendritic cells to increase their expression of CD40, which augments the activation of donor CD8 cells that initiate GVHD. The implication for human studies is that treatment with G-CSF would be expected to increase the risk of GVHD only in patients who receive TBI before the transplant.

The authors also showed that treatment with G-CSF after the transplant did not increase the severity of GVHD when donor cells came from mice treated with G-CSF1. Exposure of donor T cells to G-CSF biases their response to produce interleukin-4 and interleukin-10, rather than interferon-γ, reducing their ability

to cause GVHD. These results in mice corre-spond with clinical observations that treat-ment of the recipient with G-CSF is associated with an increased risk of GVHD after trans-plantation with bone marrow but not with G-CSF–mobilized blood cells4.

Additional studies will be needed to clarify whether treatment with G-CSF increases the risk of GVHD specifically in the subgroup of subjects who received TBI before a bone mar-row transplant.

A more important question is whether a simi-lar phenomenon occurs with transplantation of cord blood cells. These cells have the advantages of immediate availability and a less stringent requirement for tissue matching between the donor and the recipient7. On the other hand, the numbers of hematopoietic stem cells in cord blood grafts are considerably lower than in marrow grafts. The low number of stem cells delays engraftment, causing death in 10–20% of patients after cord blood transplantation8,9. As a way of accelerating granulocyte production after cord blood transplantation, many patients are treated with G-CSF. At least two large ret-rospective studies have shown no detectable effect of G-CSF on the risk of GVHD after cord blood transplantation8,9. The interpretation of these findings is limited, however, as many of the patients were not treated with TBI.

If further studies confirm that treatment with G-CSF increases the risk of GVHD in patients who have cord blood transplantation after TBI,

several approaches could be taken to circum-vent the choice between delayed engraftment and GVHD. If TBI is necessary, administration of G-CSF could be delayed until day 7 after the transplant. Recipient dendritic cells have either disappeared or lost their hypersensitiv-ity to G-CSF by day 7 after HCT, as shown by Morris et al.1, but donor hematopoietic cells are still able to respond10. Alternatively, it might be possible to use a conditioning regimen that does not contain TBI.

Other growth factors such as granulocyte- macrophage colony–stimulating factor (GM-CSF) have been used to accelerate engraftment3. Studies to date have not shown an increased risk of GVHD in individuals who received GM-CSF, but the number of patients was too small to provide a definitive conclusion. Studies of GM-CSF in mice could help answer this question.

1. Morris, E.S. et al. Nat. Med. 15, 436–441 (2009).2. Offner, F. et al. Blood 88, 4058–4062 (1996).3. Dekker, A. et al. J. Clin. Oncol. 24, 5207–5215

(2006). 4. Ringden, O. et al. J. Clin. Oncol. 22, 416–423 (2004). 5. Ho, V.T. et al. Bone Marrow Transplant. 32, 771–775

(2003). 6. Khoury, H.J. et al. Blood 107, 1712–1716 (2006). 7. Grewal, S.S., Barker, J.J., Davies, S.M. & Wagner, J.E.

Blood 101, 4233–4244 (2003). 8. Rocha, V. et al. N. Engl. J. Med. 342, 1846–1854

(2000). 9. Gluckman, E. et al. Exp. Hematol. 32, 397–407

(2004). 10. Hagglund, H. et al. Bone Marrow Transplant. 24,

831–836 (1999).

364 volume 15 | number 4 | april 2009 nature medicine

Neutrophil sandwiches injure the microcirculationMark R Looney & Michael A Matthay

Experiments in two mouse models of thromboinflammatory disease show how neutrophils stick to red blood cells and platelets—leading to reduced blood flow and damage to the microcirculation. Polarized expression of αMβ2 integrins on neutrophils helps set the process in motion (pages 384–391).

Mark R. Looney is in the Cardiovascular Research

Institute and the Departments of Medicine and

Laboratory Medicine and Michael A. Matthay is

in the Cardiovascular Research Institute and the

Departments of Medicine and Anesthesia at the

University of California–San Francisco,

San Francisco, California, USA.

e-mail: michael.matthay@ucsf.edu

The systemic and pulmonary microcircula-tions are the targets of injury in sepsis and acute lung injury, disorders that are major causes of morbidity and mortality in criti-cally ill individuals. Damage to the smallest blood vessels, including arterioles, capillaries

and venules, leads to shock, acute respiratory failure and acute renal failure.

Individuals who have sickle cell disease develop occlusion and hypoxia of the micro-circulation when challenged with inflamma-tory insults1,2. The microcirculation is also the target of injury from the transfusion of blood products that can produce severe damage to the lung, known as transfusion-related acute lung injury (TRALI). TRALI is estimated to occur in approximately 1 in 5,000 blood transfusions, although it is probably more common, with the lower numbers due to underreporting and other factors3.

Mouse models of sickle cell disease and TRALI have provided insights into how damage to the microcirculation occurs in response to inflam-mation. In two such clinically relevant animal models, Hidalgo et al.4 now provide insight into how neutrophils respond to acute inflammation and contribute to damage. In this issue of Nature Medicine, the authors show how sequestered neu-trophils coordinate tissue injury by interacting with the endothelium, circulating red blood cells and platelets4. These interactions are mediated through the actions of endothelial E-selectin, neutrophil E-selectin ligand-1 (ESL-1) and the promiscuous leukocyte integrin αMβ2 (also known as CD11b/CD18, Mac-1 and CR3).

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

n e w s a n d v i e w s

nature medicine volume 15 | number 4 | april 2009 365

Neutrophils are key sentinel cells of the innate immune system and are the premier cellular responders to acute inflammation. For example, in models of acute lung injury ranging from acid aspiration5 to ischemia- reperfusion6 to TRALI7, depletion of neu-trophils before the injury stimulus protects rabbits, rats and mice, respectively, from lung injury. These noninfectious models differ from the most common clinical causes of acute lung injury, pneumonia and sepsis, in which neu-trophils are needed to control the infection. Nonetheless, animal work has demonstrated that the proinflammatory response of the neu-trophil can lead to an increase in endothelial and epithelial permeability and, in the case of sepsis, shock and global organ injury8. Directing the neutrophil responses to com-bat acute infections without injuring the host remains a lofty goal of modern medicine.

As intrepid responders to acute inflamma-tion, neutrophils are uniquely equipped to mingle with a variety of cells in the microcir-culation. To facilitate delivery of neutrophils in the bloodstream to areas of injury and infection, circulating neutrophils marginate along endothelial cells via endothelial E- and P-selectins and their corresponding neutrophil ligands ESL-1, P-selectin glycoprotein ligand-1 and CD44.

Hidalgo et al.4 provide evidence for neutrophil- platelet or neutrophil–red blood cell interac-tions in the inflamed systemic microcirculation in the absence of organ injury, findings that imply a possible functional role for the hetero-typic aggregates. However, neutrophil-platelet aggregates detected in a mouse model of acid-induced acute lung injury were responsible for organ injury, as blocking these interactions via the P-selectin pathway protected the mice from lung endothelial injury9.

Using elegant in vivo multichannel fluo-rescent imaging of the systemic circulation, the authors studied how endothelial cells communicate with neutrophils to stimulate heterotypic interactions with red blood cells and platelets4. The authors report that ESL-1 on neutrophils, by binding to E-selectin on endothelium, signals the polarized activation of surface αMβ2 integrin on the neutrophil4.

In a mouse model of sickle cell disease, they show that αMβ2 integrin in turn medi-ates the heterotypic associations of neutro-phils with normal and sickle red blood cells4; in the TRALI model, αMβ2 integrin mediates the interaction of neutrophils with platelets. The αMβ2 integrin is constitutively present on the neutrophil surface, but inflammation and ESL-1 signaling activates and polarizes it to the leading edge of crawling neutrophils, where the heterophile aggregates form. In the

sickle cell model, these events lead to capture of sickle red blood cells and vascular occlu-sion, and, in the TRALI model, they lead to the capture of platelets and endothelial injury mediated by reactive oxygen species (Fig. 1).

These findings are noteworthy and clini-cally relevant for several reasons. They establish a link between E-selectin–induced regional activation of the neutrophil αMβ2 integrin through ESL-1, a ligand previously not known to transduce signals. The results also expand the function of E-selectin to include the induction of distinct signaling events in neutrophils, adding to its classical role in leukocyte rolling. The E-selectin–ESL-1–αMβ2 integrin pathway has the potential for pharmacologic intervention. Because of its importance in both thrombotic and inflammatory experimental models, this pathway also has the potential for broad rel-evance in microcirculatory diseases.

Future work is needed to identify the spe-cific ligands on red blood cells and platelets

that participate in the heterotypic neutrophil interactions. Targeting platelets in throm-boinflammatory diseases is potentially an attractive area of investigation, but it must be carefully tested given the role of platelets in maintaining normal systemic and pulmo-nary vascular permeability10. Also needed is an investigation of the dynamic neutro-phil–red blood cell and neutrophil-platelet interactions—a circulatory bed that differs substantially from the systemic microcir-culation11. In the lung microcirculation, for example, neutrophils are sequestered by mechanical forces as well as by selectin-mediated mechanisms.

Although pharmacologically disrupting the E-selectin–ESL-1–αMβ2 integrin pathway has the potential to protect against injury to the microcirculation, timing may be key. Lung injury in TRALI, for example, usually manifests maximally within 30–60 minutes after infusion of a blood product and then resolves over the next few days12. With complications of sickle cell disease, such as acute chest syndrome and

Figure 1 Heterotypic interactions of neutrophils in the inflamed microvasculature. With inflammation, E-selectin is expressed in the microvasculature leading to rolling of neutrophils on the endothelium via ESL-1. In the sickle cell disease model (top), inflammation is combined with circulating sickle red blood cells. E-selectin engagement of ESL-1 on neutrophils produces outside-in signaling and clustered activation of αMβ2 integrin on the neutrophil leading edge. Activated αMβ2 integrin binds sickle red blood cells, possibly through complement component 3 deposited on the sickle red blood cell surface—leading to vascular occlusion and tissue hypoxia, which could produce the clinical syndrome of a vaso-occlusive crisis. In the TRALI model (bottom), a two-event model of injury is produced by combining inflammation with a blood transfusion containing major histocompatibility complex class I–specific antibody (MHC I Ab). The transfused antibody binds to endothelial MHC I antigen leading to the capture of neutrophils via FcγR. As in the sickle cell disease model, E-selectin–ESL-1 engagement also polarizes activated αMβ2 integrin at the neutrophil leading edge—resulting in capture of circulating platelets, although the platelet antigen has not been identified. Neutrophils ultimately produce reactive oxygen species (ROS), which leads to endothelial injury, endothelial permeability and extravasation of plasma into the interstitium and air spaces of the lung. RBC, red blood cell.

Vascular occlusionTissue hypoxia

Sickle cellRBC

Endothelialpermeability

RBC

E selectin

NeutrophilPlatelet

ESL-1

ROS

FcγRengagement

MHC I AbMHC Iantigen

Plateletligand?

RBCligand?

αMβ2

PlatePlateellettellettligannddd??

bI AbI ACMHCMMMM C MHCHC IC IMH

FcFcFF γγγγγγRRageemeengggaa

ne

HCC IMHtiggenant

lectinlectin L-1L 111ESLEESLE

Sickle cell diseaseInflammation

trNeutropophilhilphphhhteletteletPlatPlat

Blood transfusion(MHC I Ab)

Inflammation

Kim

Cae

sar

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

n e w s a n d v i e w s

366 volume 15 | number 4 | april 2009 nature medicine

vaso-occlusive pain crises, microcirculatory injury is dynamic and will challenge attempts to interrupt the pathway. Investigating whether heterotypic neutrophil interactions are impor-tant in subacute or resolving inflammation will be a major goal of future investigations.

1. Bunn, H.F. N. Engl. J. Med. 337, 762–769 (1997).

2. Gladwin, M.T. & Vichinsky, E. N. Engl. J. Med. 359, 2254–2265 (2008).

3. Toy, P. et al. Crit. Care Med. 33, 721–726 (2005). 4. Hidalgo, A. et al. Nat. Med. 15, 384–391 (2009).5. Folkesson, H.G., Matthay, M.A., Hebert, C.A. &

Broaddus, V.C. J. Clin. Invest. 96, 107–116 (1995). 6. Eppinger, M.J., Deeb, G.M., Bolling, S.F. & Ward, P.A.

Am. J. Pathol. 150, 1773–1784 (1997).7. Looney, M.R., Su, X., Van Ziffle, J.A., Lowell, C.A.

& Matthay, M.A. J. Clin. Invest. 116, 1615–1623 (2006).

8. Ware, L.B. & Matthay, M.A. N. Engl. J. Med. 342, 1334–1349 (2000).

9. Zarbock, A., Singbartl, K. & Ley, K. J. Clin. Invest. 116, 3211–3219 (2006).

10. Bozza, F.A., Shah, A.M., Weyrich, A.S. & Zimmerman, G.A. Am. J. Respir. Cell Mol. Biol. 40, 123–134 (2009).

11. Burns, A.R., Smith, C.W. & Walker, D.C. Physiol. Rev. 83, 309–336 (2003).

12. Bux, J. & Sachs, U.J. Br. J. Haematol. 136, 788–799 (2007).

Dust mites’ dirty dealings in the lungClare M Lloyd

Toll-like receptors on lung epithelia recognize allergens and help provoke asthma. The findings put new emphasis on innate immunity as a driver of allergic responses (pages 410–416).

Clare M. Lloyd is in the Leukocyte Biology Section,

UK National Heart and Lung Institute, Imperial

College, London, UK.

e-mail: c.lloyd@imperial.ac.uk

The normal pulmonary response to harmless airborne particles such as pollen, animal dander and house dust mites is tolerance, achieved by a complex network of cells and molecules within the lung. In contrast, asthmatic individuals respond with inflammatory reactions, leading to cellular infiltration of the lungs coupled with changes to lung function, airway hyperrespon-siveness and bronchospasm, or ‘wheezing’.

A combination of genetic and environmen-tal factors is thought to influence whether an inflammatory reaction is initiated, and the pul-monary epithelium is increasingly implicated as a key player in this process1. However, the molecular mechanisms underlying interactions between the immune system and the epithelium are not well understood. In this issue of Nature Medicine, Hammad et al.2 show that triggering of Toll-like receptor 4 (TLR4), a receptor that recognizes conserved components of microbes, on epithelial cells helps drive the development of allergic reactions to a common household antigen.

The pulmonary epithelium provides a bar-rier between the outside environment and the internal tissues. This barrier is maintained by tight junctions located at the apical surface of epithelial cells, which enable the cells to adhere together. Apart from providing a physical bar-rier, the pulmonary epithelium actively contrib-utes to pulmonary immune responses. Lung epithelial cells secrete a range of cytokines and chemokines and are in intimate contact with the immune system. Epithelial cells sense microbes via pattern recognition receptors, which include the TLRs that recognize pathogen-associated molecular patterns from viruses, bacteria, fungi,

TLR4

GM-CSFTSLPIL-25IL-33

IL-4IL-5

IL-13

EosinophiliaAHRIgE

Airway remodeling

Dust miteallergen

CCL2CCL20

DC recruitmentand maturation

Immature DC

Mature DC

Naive T cell TH2 cell

TH2 polarization

↑ MHC II

↑ CD40

↑ CD86

Lung epithelial cells

Figure 1 Direct route for dust mite. Endotoxin-containing allergens such as house dust mite extract activate TLR4 on lung structural cells such as epithelial cells—thereby promoting the secretion of cytokines and chemokines (such as C-C motif chemokine ligand-2 (CCL2) and CCL20) that promote dendritic cell (DC) recruitment and maturation2. Activated dendritic cells have high surface expression of major histocompatibility complex class II (MHC II), CD86 and CD40 and promote generation of allergen-specific TH2 cells. These TH2 cells secrete the classical cytokines associated with allergic reactions—IL-4, IL-5 and IL-13, which are responsible for the symptoms such as eosinophilia and airway hyperreactivity (AHR). Hammad et al.2 found that HDM triggering of TLR4 on epithelial cells also induced the production of the innate cytokines TSLP, IL-25 and IL-33. Each of these cytokines is also associated with generation of allergic immunopathology and airway hyperresponsiveness.

Kim

Cae

sar

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.