In vivo two-photon imaging reveals monocyte- dependent … · Immune-mediated pulmonary diseases are a significant public health concern. Analysis of leukocyte behavior in the lung

In vivo two-photon imaging reveals monocyte-dependent neutrophil extravasation duringpulmonary inflammationDaniel Kreisela,b,1, Ruben G. Navab,1, Wenjun Lib,1, Bernd H. Zinselmeyera,c,1, Baomei Wanga, Jiaming Laib,Robert Plessd, Andrew E. Gelmana,b, Alexander S. Krupnickb, and Mark J. Millera,2

Departments of aPathology and Immunology, bSurgery, and dComputer Science and Engineering, Washington University, St. Louis, MO 63110; and cNationalInstitute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20824

Edited* by Michael D. Cahalan, University of California, Irvine, CA, and approved September 13, 2010 (received for review June 21, 2010)

Immune-mediated pulmonary diseases are a significant publichealth concern. Analysis of leukocyte behavior in the lung isessential for understanding cellular mechanisms that contribute tonormal and diseased states. Here, we used two-photon imaging tostudy neutrophil extravasation from pulmonary vessels and sub-sequent interstitial migration. We found that the lungs containeda significant pool of tissue-resident neutrophils in the steady state.In response to inflammation produced by bacterial challenge ortransplant-mediated, ischemia-reperfusion injury, neutrophils wererapidly recruited from the circulation and patrolled the interstitiumand airspaces of the lung. Motile neutrophils often aggregated indynamic clusters that formed and dispersed over tens of minutes.These clusters were associated with CD115+ F4/80+ Ly6C+ cells thathad recently entered the lung. The depletion of blood monocyteswith clodronate liposomes reducedneutrophil clustering in the lung,but acted by inhibiting neutrophil transendothelial migration up-stream of interstitial migration. Our results suggest that a subsetof monocytes serve as key regulators of neutrophil extravasationin the lung and may be an attractive target for the treatment ofinflammatory pulmonary diseases.

lung | two-photon microscopy | transendothelial migration | ischemia |transplant

Lungs are the site of many human diseases, which is in large partdue to their constant exposure to many infectious and noxious

agents. Specifically, neutrophil-mediated responses in the lung arecritical as a first line of defense against infections (1). To this end,a recent study demonstrated that respiratory influenza infection ofneutrophil-depleted mice is associated with enhanced viral repli-cation, severe pulmonary inflammation, and death (2). However,in other disease processes such as pulmonary ischemia reperfusioninjury, neutrophil activation can be deleterious (3). Inhibition ofneutrophil recruitment has proven beneficial in several experi-mental models of lung inflammation (4). Thus, a better un-derstanding of neutrophil trafficking in the lung is crucial for thedevelopment of new and effective therapeutic strategies.Two-photon (2P) microscopy has been widely adopted by

immunologists and microbiologists to study single-cell dynamics intissue explants and living mice (5–7). Recently, we and others haveused 2P microscopy to study cellular immune responses inexplanted lungs (8–11). Two-photon microscopy has superiorspatiotemporal resolution to positron emission tomography andmagnetic resonance imaging (12, 13), and greater tissue penetra-tion and less photodamage, than confocal microscopy. Despitethe potential advantages of using 2P microscopy in pulmonaryresearch, this approach has not been applied to study leukocytedynamics in vivo due to the technical difficulties of imaging a rap-idly ventilated lung.Here, we used 2P time-lapse imaging to study neutrophil traf-

ficking in the lungs of mechanically ventilated mice. We observeda significant pool of lung-resident neutrophils in the steady state.In response to inflammation, neutrophils were recruited rapidly

from the circulation and displayed robust interstitial migration inthe lung. Motile neutrophils formed dynamic clusters around re-cently emigrated quantum-dot (Q-dot) positive cells that expressmonocyte surface markers. Depleting blood monocytes withclodronate liposomes reduced neutrophil clustering in the lung byimpairing neutrophil transendothelial migration, which left largenumbers of neutrophils stranded along the vascular endothelium.Our results suggest that blood monocytes play a central role inregulating neutrophil trafficking in the lung.

ResultsIn Vivo Imaging of Leukocyte Trafficking in the Steady State. Weused 2Pmicroscopy to examine leukocyte trafficking in LysM-GFPmice (14), in which endogenous neutrophils are brightly labeledand monocytes and macrophages are labeled to a lesser extent(15). For imaging, mice were anesthetized, intubated, and placedon a ventilator. The left lung was exposed by performing a lateralthoracotomy, and the animal was placed in a custom 2P imagingchamber. To visualize the pulmonary vasculature and determinewhether neutrophils were extravascular, we injected quantum-dots(Q-dots; Invitrogen) (Fig. 1A). We first examined lungs in healthymice by using intravital 2P microscopy and found that many neu-trophils were sequestered in the pulmonary microcirculation(Movie S1) as reported by others (16). In addition to the circu-lating pool of neutrophils, we found a substantial number of ex-travascular neutrophils in the lung (Fig. 1A, yellow arrowheads).To address the possibility that the imaging preparation itself in-duced neutrophil extravasation, we examined freshly explantedlungs from healthy mice and found that they also containeda population of extravascular neutrophils (Fig. 1B). In this respect,the lung resembled secondary lymphoid organs such as lymphnodes (Fig. 1C), which contained tissue-resident neutrophils, butnot other nonlymphoid tissues including heart, brain, liver, kidney,small bowel, and footpad (Fig. 1 D–I), which were virtually free ofextravasated neutrophils.Intravital imaging and single-cell tracking revealed that extra-

vascular neutrophils were weaklymotile (mean velocity= 2.88 μm/min) (Fig. 2 A–D), and the number of motile cells varied widelyfrom mouse to mouse (Movie S1 and Movie S2). In contrast,neutrophils in lung explants were predominantly motile and mi-grated randomly through the tissue with a mean velocity of 8 μm/

Author contributions: D.K., R.G.N., W.L., B.H.Z., B.W., J.L., A.E.G., A.S.K., and M.J.M. de-signed research; D.K., R.G.N., W.L., B.H.Z., B.W., J.L., A.E.G., A.S.K., and M.J.M. performedresearch; R.P. contributed new reagents/analytic tools; R.G.N., B.H.Z., B.W., R.P., andA.E.G. analyzed data; and D.K., R.G.N., and M.J.M. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1D.K., R.G.N., W.L., and B.H.Z. contributed equally to this work.2To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1008737107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1008737107 PNAS | October 19, 2010 | vol. 107 | no. 42 | 18073–18078

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

min (Fig. 2 E–H and Movie S3), which is similar to interstitialvelocities reported for neutrophils at sites of inflammation (17).

Impact of Inflammation on Leukocyte Recruitment and Motility. Onepossible explanation for the variability of neutrophil motility invivo is that perhaps individual mice in our colony had occult re-spiratory infections or inflammation that could affect cell motility.Moreover, the robust neutrophil motility observed in the explan-ted lungs could be due to the trauma associated with surgicalremoval of the lung and ex vivo imaging. To test whether in-flammation influenced neutrophil motility in the lung, we admin-istered bacteria intratracheally and assessed pulmonary neutrophilmotility by intravital 2P microscopy. Within minutes of bacterial

challenge, we observed a dramatic influx of cells from the circu-lation and a significant increase in resident neutrophil motility(mean velocity = 9.68 μm/min) (Fig. 2 I–L and Movie S4). Similarresults were obtained after intratracheal administration ofEscherichia coli BioParticles (Movie S5). In response to bacterialinfection, neutrophils in the lung often formed dynamic clusters(Fig. 3 A and B and Movie S6) reminiscent of leukocyte behaviorobserved in other tissues after infection (15, 18, 19).To determine whether cell behaviors were specific to bacterial

challenge or representative of inflammation in general, we imagedneutrophil recruitment during lung transplant-mediated ischemiareperfusion injury (20), a process that contributes to high rates ofearly and late graft dysfunction in the clinics (21). This model is

Fig. 1. Neutrophil and macrophage distribu-tion in various tissues of LysM-GFP mice. Neu-trophils (bright green) and macrophages (dimgreen) are easily distinguishable based on theirdifferent brightness levels and distinct morpho-logical characteristics. Blood vessels (red) werelabeled by i.v. injection of nontargeted 655-nmQ-dots and the laser-induced second harmonicgeneration signal appears blue. In addition toresident macrophages (white arrowheads),there is a large number of extravascular tissue-resident neutrophils (yellow arrowheads) seen inlungs in vivo (A) (Movie S1) and explanted lungs(B), and in lymph nodes (C). In heart tissue (D),resident macrophages were observed, but neu-trophils were present only within blood vessels.(Scale bar: 80 μm.) Lower are zoomed views ofUpper. (Scale bar: 20 μm.) Tissue-resident mac-rophages (white arrowheads) are found in othertissues, but neutrophils (yellow arrowheads) aredetected primarily within the vasculature inbrain (E), liver (F), kidney (G), small intestine (H),or hind footpad (I). (Scale bar: 15 μm.)

Fig. 2. Time-lapse imaging of neutrophil behavior in lungtissue in vivo, ex vivo, and under inflammatory conditions.(A) Intravital 2P imaging of resident neutrophils (green) inthe parenchyma of the lung in vivo. Images are individualframes from a continuous time-lapse movie (Movie S2). Arare motile cell (yellow arrowheads) is shown migratingthrough the interstitial tissue. (B) Individual cells weretracked and cell displacement squared (μm2) vs. time (min)shows a strong linear correlation indicative of random cellmigration. Plots of average neutrophil track speed (C) andmeandering index (MI), n = 20 (D). The MI was calculatedby dividing the distance a cell traveled from its startingpoint by the track length. Values of >0.8 are commonlyassociated with chemotaxis, whereas values of <0.5 areconsistent with random cell migration. (E) Neutrophils(green) migrating in explanted lung tissue (Movie S3). Arepresentative neutrophil track is highlighted (yellow ar-rowhead). Neutrophil displacement squared vs. time plot(F), mean track speed (G), and MI (H) in lung explants, n =20. (I) Neutrophil (green) behavior in vivo 5 min afterintratracheal administration of bacteria (L. mono-cytogenes, EGD strain) (Movie S4). A representative neu-trophil track is highlighted (yellow arrowhead). Neutrophildisplacement squared vs. time plot (J), mean track speed(K), and MI (L) after bacterial challenge, n = 16. (Scale bars:10 μm.) Relative time is displayed in min:sec.

18074 | www.pnas.org/cgi/doi/10.1073/pnas.1008737107 Kreisel et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

advantageous because ischemic injury in lung grafts is associatedwith robust neutrophil recruitment (3). Furthermore, by usingLysM-GFP mice as recipients, but not donors, we could imageexclusively neutrophils originating from the circulation. We ob-served a substantial recruitment of GFPhigh cells to the lung graft 2h after transplantation into LysM-GFP recipients (Fig. 3C andD).Flow cytometric analysis demonstrated that ≈95% of the GFPhigh

lung-infiltrating cells, with a high degree of side scatter, expresshigh levels of Gr1, Ly6G, and CD11b, and do not express CD115,indicating that these cells are primarily neutrophils (Fig. S1). Is-chemia reperfusion injury induced vigorous neutrophil extravasa-tion as well as cell arrest in the subpleural capillary network (Fig.S2). Extravascular neutrophils in the graft migrated with robustmotility similar to neutrophils responding to bacterial challenge orthose tracked in explanted lungs (Fig. 2). Excluding stationary cellsin the clusters, the mean neutrophil velocity (8.85 μm/min) wassignificantly higher after ischemia reperfusion injury than inbaseline lungs (2.88 μm/min). Similar to lungs challenged withbacteria, transplanted lung grafts contained large neutrophilclusters (Fig. 3C andD andMovie S7). Both the frequency and sizeof neutrophil clusters were significantly increased over intravitalbaseline lungs (Fig. 3E). Moreover, many neutrophil clustersin transplanted lungs were dynamic, increasing and decreasing insize during our 30- to 60-min imaging window (Movie S7).

In Vivo Tracking of Blood Monocytes with Nontargeted Q-Dots. Inaddition to GFP-labeled neutrophils, we also saw a population ofred cells in the circulation that had the same emission character-istics as the 655-nm Q-dots used to label the blood vessels (Fig.

S3A). Using flow cytometry, we found that the Q-dot-positive cellsexpressed F4/80, the M-CSF receptor (CD115), the myeloidmarker CD11b, Ly-6C, and low levels of Gr1 (Fig. S3B), which isconsistent with these cells being peripheral blood monocytes (22,23). Within the lung, we typically observed Q-dot-positive mono-cytes rocketing through large vessels and moving in a jerky in-termittent pattern in alveolar capillaries and the subpleuralcapillary network. In the absence of inflammation, these mono-cytes rarely arrested inside vessels or entered the tissue. However,under inflammatory conditions, Q-dot-positive monocytes werefound rolling in the vessels and occasionally entering the tissues at2–3 h after transplantation (Movie S8).

Neutrophil Clustering Is Dynamic, Transient, and Associated withQ-Dot-Positive Cells. The observation that transplanted lungs con-tained dynamic neutrophil clusters is surprising, because the graftwould be expected to be more uniform in terms of ischemia-induced neutrophil recruitment signals than infected lungs wherebacteria are restricted to specific regions of the tissue. This findingsuggested that neutrophils were responding to important localmigration cues, perhaps originating from a cellular source in theengrafted lung. In support of this hypothesis, we observed in-dividual Q-dot-labeled monocytes leaving the circulation (Fig. 4Aand Movie S8) and colocalizing with large neutrophil clusters(50–100 μm below the pleura). Notably, we found that 60.9% ofneutrophil clusters are closely associated with at least one Q-dot-labeled monocyte (Fig. S4). Of the neutrophil clusters that areassociated with Q-dot-labeled cells, 78.6% are associated with onemonocyte, 14.3% with two monocytes, and 7.1% with threemonocytes. We created a “heatmap” of cell density and speed tovisualize cell migration behavior over time in the lung graft. Thisapproach generated a spatiotemporal map of cell chemotaxis invivo and allowed us to test whether neutrophil distribution wasregulated by stable global signals or by transient local signals.Using this analysis tool, we found that the migration behavior ofneutrophils was nonuniform and consistent with cells respondingto spatiotemporally restricted chemotactic signals in the graft (Fig.4 B and C and Movie S9), rather than randomly distributingthrough the tissue as expected for global recruitment signals. Atlocal regions where clusters began to form, neutrophil velocity inthe vicinity increased and the cell tracks were relatively straight.Neutrophils joining clusters showed a significant forward bias (Fig.4D) with increased velocity as they approached within 50 μmof thecluster center (8.59 μm/min) and cell tracks displayed highmeandering coefficients (>0.8) consistent with chemotaxis (Fig.4 E and F). By contrast, cells at a distance of tens of microns fromthe center of the clusters moved more slowly (4.15 μm/min) andhad lower meandering coefficients as expected for random mi-gration (Fig. 4 G and H). Chemoattraction was short-lived, andneutrophils could be found leaving mature clusters after 10–20min. These cells were often recruited to nascent clusters formingnearby, indicating that they retained their chemotactic potential.In most cases, clusters did not completely disassociate over ourimaging time window and a number of neutrophils typicallyremained at the cluster center as it decreased in size.

Clodronate-Liposome Depletion Inhibits Neutrophil ExtravasationDownstream of Endothelial Arrest. The physical association of Q-dot-positive cells with neutrophil clusters suggested that monocytesmight regulate neutrophil chemotaxis transiently and locally ateffector sites in the lung tissue. To assess the role of monocytes inneutrophil chemotaxis and cluster formation, we used repeated i.v.injections of clodronate liposomes (22) to deplete monocytes inrecipient mice before lung transplantation. Compared with controlconditions, clodronate-liposome treatment resulted in reducedneutrophil cluster formation in lung tissue at 2 h after trans-plantation (Fig. 5 A and B). Unexpectedly, we also observed a cor-responding decrease in the number of extravascular neutrophils

Fig. 3. Neutrophils cluster after intratracheal bacterial challenge. Two-photon images of neutrophil (green) distribution 5 (A) and 30 min (B) afterintratracheal administration of bacteria (Movie S6). Nontargeted Q-dots (red)were injected i.v. 30min before bacterial challenge. Neutrophils in the clusterswere often nonmotile. (Scale bar: 60 μm.) Two-photon images of neutrophil(green) distribution in lung grafts 2 h after transplant (C) and 2.5 h aftertransplant (D) (Movie S7). Time stamp is shown in min:sec. Yellow arrowheadsshow clusters that are forming or remain similar in size; white arrowheadsshow clusters that appear to dissociate. (E) The number of neutrophils percluster in steady-state lungs (gray squares), lung explants (red triangles), andlungs after transplantation (blue diamonds) (*, 0.0371; **, 0.0088).

Kreisel et al. PNAS | October 19, 2010 | vol. 107 | no. 42 | 18075

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

at this time (Fig. 5B). Upon closer examination, 2P micros-copy revealed a striking defect in neutrophil transendothelialmigration. Neutrophils accumulated inside vessels, clogging arteri-oles as well as capillaries. Despite the fact that numerous neu-trophils firmly arrested along the endothelium, the majority of cellsfailed to show evidence of diapedesis during our 2-h imagingwindow. In control transplant recipients <5% of neutrophils wereintravascular at 2-h after engraftment, in contrast to clodronate-liposome–treated mice, where >90% of neutrophils remainedtrapped within vessels (Fig. 5C).

DiscussionPrevious in vivo imaging studies of the lung using the adoptivetransfer of fluorescently or radioactively labeled neutrophils pro-vided important anatomical and physiological insight (16). How-ever, in these studies, the normal ventilation of the lungwas alteredand for the most part, imaging was restricted to superficial vessels.Moreover, there is concern that the adoptively transferred cellpopulations might traffic abnormally due to their ex vivo prepara-tion. Importantly, our approach allowed the trafficking of endog-enous neutrophils and monocytes to be analyzed quantitativelyboth in the microcirculation and as they migrate through the pa-renchymal tissue of a living lung. Our work provides a fundamentalin vivo description of neutrophil migration in pulmonary tissue.Our approach took advantage of LysM-GFP mice (14), in whichendogenous neutrophils are brightly labeled and monocytes andmacrophages are labeled to a lesser extent (15). Others have

reported that the expression ofGFP is increased onmonocytes andmacrophages after lymphocytic choriomeningitis virus infection(24). However, similar to the conclusions of others (15), our flowcytometric analysis of the lung confirmed that the brightest in-filtrating GFP+ cells were neutrophils (Fig. S1).We found that the lungs contain a significant pool of tissue-

resident neutrophils in the steady state. This was similar to lym-phoid organs and contrasted with other solid organs that containedfew if any extravascular neutrophils. Lung-resident neutrophilsmight provide a first line of defense to protect the epithelial surfaceof the lung against infection or alternatively these cells couldcontribute to the lung’s susceptibility to chronic inflammatorydiseases. Lung infiltrating neutrophils are likely to contribute tothe comparatively high rates of ischemia reperfusion injury-me-diated lung graft dysfunction (25). The presence of neutrophils inthe lung parenchyma might represent an evolutionary tradeoff infavor of rapid pathogen clearance at the expense of lowering of thethreshold for inflammatory disease.Through serendipity, we found that i.v.-injected nontargeted Q-

dots fluorescently labeled monocytes in the peripheral blood. Themost likely explanation for this phenomenon is that monocytes arelabeled by ingesting Q-dots in the circulation similar to establishedprotocols using fluorescent latex beads (23, 26). The Q-dot–labeling approach could provide a facile method to study mono-cytes in vivo in a wide range of experimental animal diseasemodelssuch as atherosclerosis or arthritis. Q-dot labeling has the advan-tages that cells are labeled in vivo (i.e., without ex vivo manipu-

Fig. 4. Leukocyte dynamics during transplant mediated is-chemia reperfusion injury. (A) Blood vessels (red) were labeledby i.v. injection of nontargeted 655-nm Q-dots. A time-lapse2P image sequence shows a Q-dot–positive cell leaving thepulmonary vasculature through a small branch of a medium-sized vessel (yellow track). Extravasation of the Q-dot–positivecell is associated with neutrophil extravasation and sub-sequent cluster formation (Movie S8). (Scale bar: 20 μm.) Aheatmap visualization was generated to show the spatio-temporal changes in neutrophil velocity (B) and density (orintegrated intensity) (C) (Movie S9). For speed, the color scaleranges from <2 μm/min (blue) to >10 μm/min (red). A localincrease in cell velocity (B) (white arrowheads) precedes a 4-fold increase in cell density (yellow arrows) (C). (Scale bar: 60μm.) (D) Neutrophil tracks (yellow arrows) show a symmetricalshort-range migration bias toward the cluster. (Scale bar: 15μm.) Time stamps in A–D show relative time in min:sec. Neu-trophils were divided in two groups based on distance fromthe center of the cluster, and the migration of each group wasanalyzed separately. Mean track speed (E) and MI (F) ofneutrophils approaching within 50 μm of clusters, n = 13.Mean track speed (G) and MI (H) of neutrophils distal toclusters (>50 μm from clusters), n = 13.

18076 | www.pnas.org/cgi/doi/10.1073/pnas.1008737107 Kreisel et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

lation), they are brightly fluorescent and that they appear to ad-here to vessels and traffic into inflamed tissues, although long-termcytotoxic effects have not been rigorously assessed.Previous reports have demonstrated that the i.v. administration

of clodronate liposomes can be used to deplete blood monocytes(22, 27). In our hands, clodronate-liposome treatment also depletedQ-dot–positive cells from the circulation, consistent with the flowcytometry data that suggests that these cells are monocytes. Ourevidence that bloodmonocytes mediate neutrophil extravasation inthe lung is based on the observations that Q-dot–positive cells oftencolocalize in vessels near sites of neutrophil extravasation, Q-dot–positive cells express the monocyte surface markers CD115, F4/80,and L6C, and finally that the clodronate-liposome-mediated de-pletion of monocytes dramatically impairs neutrophil trans-endothelialmigration.A caveat with our approach is that we cannotexclude the possibility that clodronate liposomes might depleteother rare cell types. For example, others have described the exis-tence of pulmonary intravascular macrophages, which—similar toKupffer cells in the liver—play a role in the phagocytosis of largeparticles such as cellular debris (28, 29). Although these cells are farless prevalent in mice than in other species, they could potentiallycapture clodronate liposomes from the circulation. However, it isimportant to emphasize that in our transplant experiments, onlyrecipient mice were treated with i.v. clodronate liposomes. Basedon phagocyte depletion rates in other tissues, we would not expectresidual clodronate liposomes in the recipient animal to efficientlydeplete graft-associated macrophages in the 2–3 h window aftertransplantation. Therefore, the effect of clodronate-liposome de-pletion on neutrophil extravasation is unlikely to be due to thedepletion of tissue resident macrophages. Notably, clodronate-li-posome treatment has been shown not to deplete or directly affectneutrophil behavior in vitro and in vivo (30, 31).Intravital 2P imaging showed that the depletion of blood

monocytes inhibited neutrophil recruitment to the lung specificallyat the transendothelial migration step. These findings extend onprevious reports, which suggested that monocyte and neutrophiltrafficking were interdependent during pulmonary inflammation.In response to intratracheal instillation of CCL2 and LPS, alveolarneutrophil accumulation was markedly inhibited when animalswere treated with anti-CCR2 antibody or were genetically de-ficient in CCR2 (32). Based on experiments using bone marrowchimeras, the authors concluded that recruitment of neutrophils tothe airspaces depended on CCR2-expressing blood monocytes(33). Of note, this study relied on quantitative analysis of neu-trophils within the bronchoalveolar lavage fluid and could there-fore not provide mechanistic insight into precisely at what stepmonocytes influence neutrophil recruitment. Moreover, the ex-pression of CCR2 on other cell types, including dendritic cells, NKcells, mast cells, T cells and, in particular, neutrophils, complicatesthe interpretation of the studies described above (34, 35). None-theless, our results using clodronate-liposome depletion confirmthe conclusions of Maus and colleagues (33) and extend on theirfindings by showing that monocytes play a specific and previouslyunrecognized role in facilitating neutrophil transendothelial mi-gration out of pulmonary vessels.

We observed dramatic neutrophil clustering in the lung afterintratracheal administration of bacteria, similar to the neutrophilswarming behavior observed by Chtanova and colleagues in thelymph node after s.c. infection with Toxoplasma gondii (15). In re-sponse to infection, neutrophil clustering is not surprising becauseneutrophilsmustmigrate to foci of infection to phagocytose and killbacteria. Moreover, others have observed focal neutrophil re-cruitment in response to traumatic tissue damage alone, such as atan injection site (15, 18, 19). However, ischemia reperfusion injuryafterorgan transplantationhasbeen considered a relatively “global”inflammatory event. Therefore, our observation that neutrophilsform focal clusters in the lung after transplant was unexpected.Analyzing neutrophil migration using a heatmap visualizationtechnique revealed that neutrophil chemokine gradients were typ-ically local and short-lived in vivo, resembling “chemokine bombs.”Robust neutrophil chemotaxis could be in response to necrotic cellsreleasing chemotactic substances such asATPor, perhaps, from therapid secretion of preexisting chemokine stores into the vicinity.Chtanovaandcolleagues suggested that a few “pioneer”neutrophilsresponding to T. gondii escape from an infected cell might providesignals that recruit late arriving neutrophils into swarms at foci ofinfection (15). In an analogous fashion, we observed Q-dot–labeledmonocytes within the majority of neutrophil clusters after trans-plant-mediated ischemia reperfusion injury. This observation sug-gests that monocytes might serve as a focal source of chemokinesacting on nearby neutrophils to recruit them to sites of effectorfunction. The dissolution of neutrophil clusters could representchemokine receptor desensitization, dissipation of the chemokinegradient, or perhaps chemorepulsion. The observation that neu-trophilsmigrating away fromclusterswereattractedminutes later tonearby cell clusters ismore consistent with a rapid dissipation of thechemokine gradient. In most cases, clusters did not completelydisassociate over our imaging time window; a small number ofneutrophils typically remained in the center of a cluster. Becauseneutrophils arrest during their oxidative burst, it is possible that thenonmotile cells represent a population of activated neutrophils.

ConclusionWe show that 2P microscopy is a promising approach for the invivo study of cellular immune mechanisms that operate duringinflammation and infection in the lung. We found that the de-pletion of blood monocytes impairs neutrophil recruitment to thelung specifically during transendothelial migration. Time-lapseimaging of neutrophil migration in vivo suggests that neutrophilrecruitment to effector sites and clustering is regulated by short-range transient chemokine gradients. The association of Q-dot–positive cells with these dynamic clusters suggests that monocytesmay play an important role in regulating the interstitial distributionof neutrophils and have important implications for the design oftherapeutics to treat inflammatory lung diseases.

MethodsMice and Monocyte Depletion. BALB/c mice were purchased from The JacksonLaboratories. LysM-GFP mice were obtained from Klaus Ley (La Jolla Institutefor Allergy and Immunology, La Jolla, CA) and maintained at our facility.Escherichia coli (K-12 strain) BioParticles conjugated with tetramethylrhod-

Fig. 5. Clodronate-liposome depletion impairs neutrophiltransendothelial migration after transplantation. Pulmonarybloodvessels (red)were labeledby i.v. injectionofnontargeted655-nmQ-dots. (A) Two-photon image of neutrophils (green)extravasating from a medium-size vessel (white lines) in acontrol lung graft at 2 h after transplantation. (B) Clodronate-liposome (CL) treatment of the transplant recipient results inneutrophil accumulation in medium-sized vessels and a re-ducednumberofextravasatedneutrophils. (C)Thepercentageof intravascular neutrophils observed at 2 h after engraftmentin untreated recipients (control, <5%) and in clodronate-liposome treated recipients (CL, >90%). (Scale bar: 60 μm.)

Kreisel et al. PNAS | October 19, 2010 | vol. 107 | no. 42 | 18077

IMMUNOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

amine (Invitrogen) (500 μg/mL) or 100,000 cfu of L. monocytogenes were ad-ministered intratracheally after dilution in 50 μL of PBS. Clodronate liposomesuspensions were prepared as described (31). Monocytes were depleted bytreatingmicewith three doses of i.v. clodronate-containing liposomesgiven 48(200 μL), 24 (100 μL), and 6 h (100 μL) before lung transplantation.

Lung Transplantation. After 18-h storage in low-potassium dextran glucose at4 °C, left BALB/c lungs were transplanted orthotopically into LysM-GFP miceas described (20, 36).

Two-Photon Microscopy. Time-lapse imaging was performed with a custombuilt 2P microscope running ImageWarp acquisition software (A&B Software).Mice were anesthetized with an i.p. injection of ketamine (50 mg/kg) andxylazine (10 mg/kg) and maintained with halved doses administered everyhour. Mice were intubated orotracheally with a 20 G angiocatheter andventilated with room air at a rate of 120 breaths/min and with a tidal volumeof 0.5 mL. The left lung was exposed through a left thoracotomy, and thelung was imaged by using a custom built chamber maintained at 37 °C. Asmall ring of VetBond was used to attach the lung tissue to the bottom of thecover glass without exerting pressure directly on the lung. For time-lapseimaging of leukocyte migration in the tissue parenchyma, we averaged 15video-rate frames (0.5 s per slice) during the acquisition to match the venti-lator rate and minimize movement artifacts. Each plane represents an imageof 220 × 240 μm in the x and y dimensions. Twenty-one to 31 sequentialplanes were acquired in the z dimension (2.5 μm each) to form a z stack. Tovisualize blood vessels, 20 μL of 655-nm nontargeted Q-dots in 100 μL of PBSwere injected i.v. Two-photon excitation produces a second harmonic signalfrom collagen (9) around alveoli, thus providing a useful landmark for the airspaces. Explanted tissue was examined as described (8, 9).

Flow Cytometry. Whole blood for analysis of Q-dot–positive cells and lungdigests for analysis of GFPhigh cells were prepared as described (37). Cells

were first incubated with anti-mouse CD16/CD32 for 15 min at 4 °C to blockFc receptors (Fc γ III/II receptor, 2.4G2; BD Pharmingen). Cells were stainedwith fluorochrome-labeled anti-F4/80 (clone BM8), anti-Ly-6C (clone AL-21),anti-CD11b (clone M1/70), anti-CD115 (clone AFS98), anti-Ly-6G (clone 1A8),and anti-Gr-1 (clone RB6-8C5) antibodies or isotype controls (BD PharMin-gen). Analysis was performed on a FACSCalibur (BD Bioscience) equipped forfour-color flow cytometry.

Data Analysis. Multidimensional rendering was done with Imaris (Bitplane),whereas manual cell tracking was done by using Volocity (Improvision). Datawere transferred and plotted in GraphPad Prism 5.0 (Sun Microsystems) forthe creation of the graphs. The neutrophil cluster analysis was performed byusing the cluster analysis function of the T cell Analysis program (TCA; JohnDempster, University of Strathclyde, Glasgow, Scotland) using a 25-μm radiusthreshold for cell-to-cell distances. Pseudocolored heatmaps were created tovisualize neutrophil speed [0 μm·min−1 (purple) to 12 μm·min−1 (red), anddensity (0–256 gray from purple to white)] within the tissue volume. Neu-trophil speed and density were considered to be continuous functions andthe fraction of neutrophils, which are imaged, as samples of this underlyingdistribution. These neutrophils were used to create a kernel estimate of thecomplete distribution, and the final video is color-coded based on this speedor density estimate. The kernel speed and density estimate uses the Parzenwindow approach with a Gaussian Kernel (38). The Gaussian Kernel waschosen by hand. The unpaired two-tailed Student’s t test was used forstatistical analysis.

ACKNOWLEDGMENTS. We thank A. P. Gieselman for animal care. D.K. andA.E.G. are supported by National Heart, Lung, and Blood Institute Grant1R01HL094601 and D.K. by Grant 1K08HL083983, jointly sponsored by theNational Heart, Lung, and Blood Institute and the Thoracic SurgeryFoundation for Research and Education. M.J.M. is supported by the NationalInstitute of Allergy and Infectious Diseases Grant AI077600.

1. Tsai KS, Grayson MH (2008) Pulmonary defense mechanisms against pneumonia andsepsis. Curr Opin Pulm Med 14:260–265.

2. Tate MD, et al. (2009) Neutrophils ameliorate lung injury and the development ofsevere disease during influenza infection. J Immunol 183:7441–7450.

3. Belperio JA, et al. (2005) CXCR2/CXCR2 ligand biology during lung transplantischemia-reperfusion injury. J Immunol 175:6931–6939.

4. Zarbock A, Allegretti M, Ley K (2008) Therapeutic inhibition of CXCR2 by Reparixinattenuates acute lung injury in mice. Br J Pharmacol 155:357–364.

5. Miller MJ, Wei SH, Parker I, Cahalan MD (2002) Two-photon imaging of lymphocytemotility and antigen response in intact lymph node. Science 296:1869–1873.

6. Cahalan MD, Parker I, Wei SH, Miller MJ (2003) Real-time imaging of lymphocytes invivo. Curr Opin Immunol 15:372–377.

7. Cahalan MD, Parker I, Wei SH, Miller MJ (2002) Two-photon tissue imaging: Seeingthe immune system in a fresh light. Nat Rev Immunol 2:872–880.

8. Halle S, et al. (2009) Induced bronchus-associated lymphoid tissue serves as a generalpriming site for T cells and is maintained by dendritic cells. J Exp Med 206:2593–2601.

9. Gelman AE, et al. (2009) Cutting edge: Acute lung allograft rejection is independentof secondary lymphoid organs. J Immunol 182:3969–3973.

10. St Croix CM, Leelavanichkul K, Watkins SC (2006) Intravital fluorescence microscopy inpulmonary research. Adv Drug Deliv Rev 58:834–840.

11. Pena AM, et al. (2007) Three-dimensional investigation and scoring of extracellularmatrix remodeling during lung fibrosis using multiphoton microscopy. Microsc ResTech 70:162–170.

12. Ambrosini V, et al. (2007) Assessment of a chemically induced model of lungsquamous cell carcinoma in mice by 18F-FDG small-animal PET. Nucl Med Commun 28:647–652.

13. Garbow JR, Zhang Z, You M (2004) Detection of primary lung tumors in rodents bymagnetic resonance imaging. Cancer Res 64:2740–2742.

14. Faust N, Varas F, Kelly LM, Heck S, Graf T (2000) Insertion of enhanced greenfluorescent protein into the lysozyme gene creates mice with green fluorescentgranulocytes and macrophages. Blood 96:719–726.

15. Chtanova T, et al. (2008) Dynamics of neutrophil migration in lymph nodes duringinfection. Immunity 29:487–496.

16. Lien DC, et al. (1987) Physiological neutrophil sequestration in the lung: Visualevidence for localization in capillaries. J Appl Physiol 62:1236–1243.

17. Graham DB, et al. (2009) ITAM signaling by Vav family Rho guanine nucleotideexchange factors regulates interstitial transit rates of neutrophils in vivo. PLoS ONE 4:e4652.

18. Egen JG, et al. (2008) Macrophage and T cell dynamics during the development anddisintegration of mycobacterial granulomas. Immunity 28:271–284.

19. Peters NC, et al. (2008) In vivo imaging reveals an essential role for neutrophils inleishmaniasis transmitted by sand flies. Science 321:970–974.

20. Okazaki M, et al. (2007) A mouse model of orthotopic vascularized aerated lungtransplantation. Am J Transplant 7:1672–1679.

21. Lee JC, Christie JD (2009) Primary graft dysfunction. Proc Am Thorac Soc 6:39–46.22. Sunderkötter C, et al. (2004) Subpopulations of mouse blood monocytes differ in

maturation stage and inflammatory response. J Immunol 172:4410–4417.23. Tacke F, et al. (2006) Immature monocytes acquire antigens from other cells in the

bone marrow and present them to T cells after maturing in the periphery. J Exp Med203:583–597.

24. Kim JV, Kang SS, Dustin ML, McGavern DB (2009) Myelomonocytic cell recruitmentcauses fatal CNS vascular injury during acute viral meningitis. Nature 457:191–195.

25. Kuntz CL, et al. (2009) Risk factors for early primary graft dysfunction after lungtransplantation: A registry study. Clin Transplant 23:819–830.

26. Jakubzick C, Tacke F, Llodra J, van Rooijen N, Randolph GJ (2006) Modulation ofdendritic cell trafficking to and from the airways. J Immunol 176:3578–3584.

27. Wakim LM, Waithman J, van Rooijen N, Heath WR, Carbone FR (2008) Dendritic cell-induced memory T cell activation in nonlymphoid tissues. Science 319:198–202.

28. Brain JD, Molina RM, DeCamp MM, Warner AE (1999) Pulmonary intravascularmacrophages: Their contribution to the mononuclear phagocyte system in 13 species.Am J Physiol 276:L146–L154.

29. Winkler GC (1989) Review of the significance of pulmonary intravascularmacrophages with respect to animal species and age. Exp Cell Biol 57:281–286.

30. Qian Q, Jutila MA, Van Rooijen N, Cutler JE (1994) Elimination of mouse splenicmacrophages correlates with increased susceptibility to experimental disseminatedcandidiasis. J Immunol 152:5000–5008.

31. Van Rooijen N, Sanders A (1994) Liposome mediated depletion of macrophages:Mechanism of action, preparation of liposomes and applications. J Immunol Methods174:83–93.

32. Maus U, et al. (2002) The role of CC chemokine receptor 2 in alveolar monocyte andneutrophil immigration in intact mice. Am J Respir Crit Care Med 166:268–273.

33. Maus UA, et al. (2003) Monocytes are potent facilitators of alveolar neutrophilemigration during lung inflammation: Role of the CCL2-CCR2 axis. J Immunol 170:3273–3278.

34. Terwey TH, et al. (2005) CCR2 is required for CD8-induced graft-versus-host disease.Blood 106:3322–3330.

35. Reichel CA, et al. (2006) Chemokine receptors Ccr1, Ccr2, and Ccr5 mediate neutrophilmigration to postischemic tissue. J Leukoc Biol 79:114–122.

36. Krupnick AS, et al. (2009) Orthotopic mouse lung transplantation as experimentalmethodology to study transplant and tumor biology. Nat Protoc 4:86–93.

37. Gelman AE, et al. (2010) CCR2 regulates monocyte recruitment as well as CD4 T1allorecognition after lung transplantation. Am J Transplant 10:1189–1199.

38. Parzen E (1962) On the estimation of a probability density function and the mode.Ann Math Stat 33:1065–1076.

18078 | www.pnas.org/cgi/doi/10.1073/pnas.1008737107 Kreisel et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0