LETTERdoi:10.1038/nature12902

Sessile alveolar macrophages communicate withalveolar epithelium to modulate immunityKristin Westphalen1, Galina A. Gusarova1, Mohammad N. Islam1, Manikandan Subramanian2, Taylor S. Cohen3, Alice S. Prince3

& Jahar Bhattacharya1,4

The tissue-resident macrophages of barrier organs constitute thefirst line of defence against pathogens at the systemic interface withthe ambient environment. In the lung, resident alveolar macrophages(AMs) provide a sentinel function against inhaled pathogens1. Bacterialconstituents ligate Toll-like receptors (TLRs) on AMs2, causing AMsto secrete proinflammatory cytokines3 that activate alveolar epithe-lial receptors4, leading to recruitment of neutrophils that engulfpathogens5,6. Because the AM-induced response could itself causetissue injury, it is unclear how AMs modulate the response to preventinjury. Here, using real-time alveolar imaging in situ, we show that asubset of AMs attached to the alveolar wall form connexin 43 (Cx43)-containing gap junction channels with the epithelium. During lipo-polysaccharide-induced inflammation, the AMs remained sessileand attached to the alveoli, and they established intercommunicationthrough synchronized Ca21 waves, using the epithelium as the con-ducting pathway. The intercommunication was immunosuppres-sive, involving Ca21-dependent activation of Akt, because AM-specificknockout of Cx43 enhanced alveolar neutrophil recruitment andsecretion of proinflammatory cytokines in the bronchoalveolar lavage.A picture emerges of a novel immunomodulatory process in which asubset of alveolus-attached AMs intercommunicates immunosup-pressive signals to reduce endotoxin-induced lung inflammation.

As most of our understanding of AMs is based on studies in whichAMs were isolated by bronchoalveolar lavage (BAL) and studied in vitro,or in which they were depleted from the lungs5,7, we lack an understand-ing of the dynamic interactions between AMs and the alveolar epithe-lium that might critically modulate the lung’s inflammatory response.To determine these interactions in situ, we took advantage of the factthat AMs express CD11c8 and crossed Cd11c-cre mice9 with Rosa26-LSL-eYFP mice10 to obtain mice expressing enhanced yellow fluor-escence protein (eYFP) in CD11c-expressing cells.

Live confocal microscopy of isolated, blood-perfused mouse lungs11

derived from CD11c/eYFP mice revealed YFP1 cells in the subpleuralinterstitium and the alveolar lumen (Fig. 1a). To distinguish AMs fromlung dendritic cells that also express CD11c8,12, we gave mice intra-alveolar microinjections of the AM marker Siglec F (ref. 13) and anantibody for the major histocompatibility complex class II (MHC-II),which marks lung dendritic cells12. These microinjections revealed thatluminal YFP1 cells were AMs, because they stained strongly for SiglecF but weakly for MHC-II (Fig. 1b), whereas interstitial YFP1 cells thatwe stained by topical antibody applications to the saponin-permeabilizedpleura were MHC-II1/Siglec F2 dendritic cells (Fig. 1b). The alveolarepithelium inhibits transepithelial transit of reagents injected in thealveolar lumen4. Hence, alveolar microinjections of dye resulted in fluo-rescence uptake in AMs, but not in dendritic cells (Fig. 1c), suggestingthat dendritic cells did not communicate with the alveolar lumen. Infurther affirmation that luminal YFP1 cells were AMs, YFP1-MHC-IIlow cells recovered in the BAL failed to induce T-cell proliferation inantigen-presentation assays (Extended Data Fig. 1a). Together, these

findings indicate that YFP1 AMs localized to the alveolar lumen,whereas YFP1 dendritic cells were compartmentalized in the perial-veolar interstitium.

We detected a single AM for approximately every three alveoli (ExtendedData Fig. 1b), suggesting that AMs carry out pathogen surveillance bypatrolling the alveolar surface14. However, AMs remained stationary atfixed alveolar locations for the duration of our imaging studies, whichlasted up to 4 h (Extended Data Fig. 1c). Our attempts to dislodge AMsby BAL or by direct alveolar microinjection of buffer were unsuccessful(Extended Data Fig. 1c). To determine whether AMs might be inducedto migrate towards bacteria, we microinjected Staphylococcus aureus inthe alveolar lumen. Although AMs rapidly ingested the S. aureus thatlay within one AM diameter, they did not migrate towards the bacteria(Fig.1d and Extended Data Fig. 1c). The alveolar liquid flow15 appearedto wash bacteria towards the AMs (data not shown). Thus, contrary toexpectations, our findings indicate that AMs were sessile.

Macrophages express Cx43 (ref. 16), potentially enabling AMs toform gap junction channels (GJCs) with the alveolar epithelium. Todetermine the presence of GJCs, we applied photolytic uncaging to inducecell-specific increases in cytosolic Ca21 (ref. 17), and fluorescence recov-ery after photobleaching (FRAP) to quantify intercellular dye diffusion11.In 40% of AMs, uncaging-induced Ca21 waves travelled from the epi-thelium to AMs (Fig. 1e) and in the opposite direction (data not shown).Cx43 expression in AMs correlated directly with FRAP (Fig. 1f). A1 h treatment with GAP26 and 27, inhibitors of Cx43-based GJCsand hemichannels, blocked uncaging-induced Ca21 waves (Fig. 1e) aswell as FRAP (data not shown) between AMs and the epithelium. InCD11chigh MHC-IIlow AMs, which we obtained by BAL and by extrac-tion from lung tissue after BAL, respectively (Extended Data Fig. 1a, d),Cx43 protein and messenger RNA expression were higher in tissue thanin BAL AMs (Fig. 1g), suggesting that Cx43 was higher in alveolus-adherent than alveolus-non-adherent AMs. In mice with CD11c-specific Cx43 knockout (CD11c Cx432/2) (Extended Data Fig. 2a),AMs remained immobile even after alveolar microinjections of bac-teria or PBS buffer (Extended Data Fig. 2b, c). Hence, Cx43 was notresponsible for AM immobility. In lungs given intranasal Escherichia-coli-derived lipopolysaccharide (LPS) instillation, neutrophils enteredand migrated freely on the alveolar surface (Supplementary Fig. 1 andSupplementary Video 1), ruling out non-specific physical factors incausing AM immobility. Taken together, our findings reveal the pres-ence of Cx43high and Cx43low AM populations in the lung, in whichCx43high AMs formed GJCs with the alveolar epithelium.

Next, we gave mice intranasal LPS to induce lung injury or PBS as acontrol. We removed lungs 1, 4 or 24 h after the instillations to estab-lish isolated perfused lungs for imaging studies11. Instillation of fluor-escent LPS confirmed LPS entry in AMs (Fig. 2a). Fluorescent LPS didnot enter interstitial dendritic cells, indicating that in alveoli LPS ligatedAMs, not dendritic cells. Fluorescent LPS uptake was markedly greaterin BAL-derived dendritic cells than in those recovered from the tissue

1Lung Biology Laboratory, Department of Medicine, Division of Pulmonary, Allergy and Critical Care, Columbia University Medical Center, New York, New York 10032, USA. 2Department of Medicine,Division of Molecular Medicine, Columbia University Medical Center, New York, New York 10032, USA. 3Department of Pediatrics, Columbia University Medical Center, New York, New York 10032, USA.4Department of Physiology & Cellular Biophysics, College of Physicians and Surgeons, Columbia University Medical Center, New York, New York 10032, USA.

2 7 F E B R U A R Y 2 0 1 4 | V O L 5 0 6 | N A T U R E | 5 0 3

Macmillan Publishers Limited. All rights reserved©2014

of post-lavage lungs, suggesting that access to the airway lumen may begreater for BAL- than tissue-derived dendritic cells (Extended DataFig. 3).

By contrast with PBS-treated lungs, in which there were no notableeffects, the AMs of LPS-treated lungs showed synchronous Ca21 spikes,all of which occurred within 1 min of one another in AM clusters. Spikeslasted 10–15 s, appearing every 10–20 min (Fig. 2b and SupplementaryVideo 2). They began 4 h after LPS and increased over 24 h (Fig. 2c).Concomitant Ca21 spikes occurred in the adjoining epithelium (Fig. 2d).The spikes travelled between different AMs, often separated by severalalveoli, across the intervening epithelium (Fig. 2b and Extended DataFig. 4a). We did not identify a specific, ‘pacemaker’ AM that initiatedsynchronous spiking. Intra-alveolar microinjection of GAP26/27 blockedsynchronous epithelial and AM spikes (Fig. 2d, e), suggesting that therewas interdependence between these responses. GAP26/27 did not blocknon-synchronous spikes (data not shown). Connexin hemichannels areimplicated in some forms of ATP-dependent purinergic signalling18.However, after LPS treatment, cytosolic dyes (fluo-4, calcein) that cantransit connexin channels did not leak from AMs or the epithelium(Extended Data Fig. 4b, c). The ATP receptor inhibitor pyridoxalphosphate-6-azo(benzene-2,4-disulphonic acid) tetrasodium salt hydrate(PPADS) did not modify spike formation (Fig. 2e). These findings ruleout a role for connexin hemichannels in inducing spike intercommun-ication. Alveolar microinfusion of the inositol-(1,4,5)-trisphosphate(Ins(1,4,5)P3) receptor inhibitor xestospongin C, but not depletionof extracellular Ca21, blocked the Ca21 spikes (Fig. 2e), indicating thatthe spikes resulted from Ca21 release from intracellular stores. Depletionof alveolar neutrophils, which we identified as CD11b1 cells of 6–8-mmdiameter, did not diminish spike formation (Extended Data Fig. 4d).The numbers of AMs per imaging field were identical at all time points(Extended Data Fig. 4e). Together, these findings indicate that LPSinduced intercommunicated Ca21 spikes between AMs lying in differentalveoli and that the communication occurred through the epithelium.

LPS ligation of TLR4 induces signalling through the adaptor proteinsmyeloid differentiation factor 88 (MyD88) and TIR-domain-containing

b

CalceinMHC-II Merge

c

Int

Alv

Microscope

2

Depth from pleura (μm)

x y

z

Alv20

Chang

e in c

ycle

thre

sho

ld

(Gapdh

vs Cx43

mR

NA

)

–15

–10

–5

0ND

T B

Cx43 fl

uo

rescence

(gre

y levels

)

T B0

50

100

150

*

BAL Tissue g

–

AM

s w

ith fl

uo

rescence

incre

ase (%

of

tota

l)

GAP26/27 +

25

50

*

Baseline

e Uncaging

YF

P+ c

ells

(%

)

0

25

50

75

100Intra-alveolarInterstitial

AM

DC

YFP MHC-II Siglec F

* *

*

f

Cx43low

Cx43high

Bleach: Pre

Post (min)

0 30 0 50 100 150

0

10

20

30

FR

AP

(g

rey levels

aft

er

fluo

rescence re

co

very

)

Cx43 expression (grey levels)

P < 0.01

Tota

l

Siglec

F

MHC-II

Tota

l

Siglec

F

MHC-II

a

Tim

e (m

in)

d

10

30

0

Figure 1 | Live confocal microscopy of AMsin situ. a, Interstitial (arrowhead) and luminal(arrow) YFP1 cells (yellow/green) inautofluorescent alveoli (red) (n 5 15). Sketch ofimaged field. b, Immunofluorescence andquantification of alveolar (top) and interstitial(bottom) YFP1 cells (n 5 4 or 5). DC, dendriticcell. c, Dye (green) injection in alveolus (Alv).Interstitial (Int) MHC-II1 dendritic cell (red,arrowhead) and luminal AM (arrow) (n 5 3).d, Merge of S. aureus (green, arrow in inset) withAMs (red; arrowhead in inset) (n 5 4). e, Ca21

uncaging in epithelium (dotted circle) spreadsCa21 (arrow) to AM (arrowhead) (n 5 4 or 5). Barsshow GAP26/27 effect. f, Alveoli with Cx43 (red)and calcein (green) show Cx43low (arrowhead) andCx43high AMs (arrow) expressing YFP (yellow/green). High-power images show calcein in AMsbefore (Pre) and after (Post) photobleaching. Linedrawn by linear regression. n 5 23 AMs, fourlungs. g, Cx43 (red), CD11c (green) and nuclei(blue) in AMs from BAL (B) or lung tissue (T)(n 5 4). Bars show protein (left) and mRNA (right)expressions in AMs. ND, not detected. Scale bars,15mm. Data show mean 6 standard error of themean (s.e.m.). *P , 0.05.

BL 1

Time after LPS (h)

4 24

TotalSync *

*

*

c

Ca

2+ s

pik

es

0

5

10

Alv

LPS Siglec F MHC-II Merge

Int

a

Ca

2+ (fluo

-4 g

rey level)

Time (min)0 10 20

0

100

200

LPS 0

100

200 PBS

e

* * Ca

2+ s

pik

es

0

5

10

15

Time (s)

50

b LPS

0

10

30

40

50

255

0

Grey level

80

100

120

PBS

d

Ca

2+ (%

of

mean

fluo

-4 fl

uo

rescence) AE

AM

0 10 20

LPS + GAP

Time (min)

0 10 200 10 20

LPS

CtlGAP PP

XeC Ca0

Figure 2 | Communicating Ca21 spikes in AMs. a, Fluorescent LPS (green),AM (red) and dendritic cell (blue) in interstitium (Int) and alveolar lumen (Alv)(n 5 3). b, YFP-expressing AMs (yellow; top), pseudocoloured sequentialimages show increased fluo-4 fluorescence (arrowheads) 24 h after LPS (n 5 4).Dashes sketch spike path between AMs. Tracings show synchronous AMresponses (arrows). c, Ca21 spikes normalized for ten AMs per imaging field.Sync, synchronous spikes; Total, all spikes; BL, baseline (n 5 8); 1 h, n 5 6; 4 h,n 5 4; 24 h, n 5 12. d, Ca21 oscillations and spikes (arrows) in AMs andadjoining alveolar epithelium (AE) 24 h after LPS or PBS (n 5 4). GAP, GAP26/27. e, Ca21 spikes are 24 h after LPS, normalized for ten AMs per imaging field.Ctl, LPS alone (n 5 12); PP, PPADS (n 5 4); XeC, xestospongin C (n 5 4); Ca0,Ca21 depletion (n 5 3). GAP, n 5 5. Scale bars, 20mm. Bars, mean 6 s.e.m.*P , 0.05 versus LPS alone or corresponding baseline.

RESEARCH LETTER

5 0 4 | N A T U R E | V O L 5 0 6 | 2 7 F E B R U A R Y 2 0 1 4

Macmillan Publishers Limited. All rights reserved©2014

adaptor-inducing interferon-b (TRIF; also known as Ticam-1)19.MyD88 and TRIF are implicated in lung injury20,21. We generated mice(CD11c Myd882/2) lacking MyD88 in CD11c-expressing cells22. InCD11c Myd882/2, but not wild-type mice, LPS-induced Ca21 spikesin AMs (Fig. 3a) and the epithelium (Fig. 3b, c) were lacking, andalveolar neutrophil entry at 24 h was diminished (Fig. 3d, e), althoughboth mice had similar Cx43 expression in AMs (Extended Data Fig. 4f).

We conclude that MyD88-dependent signalling was responsible forthe LPS-induced spike formation and lung inflammation, and thatAMs initiated the signalling.

Synchronous spikes in AMs and the epithelium were inhibited inCD11c Cx432/2 mice (Fig. 3a–c), although non-synchronous Ca21

spikes in AMs were similar to those of wild-type mice (Fig. 3a). Lunginflammation was markedly greater in CD11c Cx432/2 than wild-typemice, as indicated by increased LPS- or E.-coli-induced alveolar neu-trophil recruitment and BAL leukocyte counts (Fig. 3d, e and ExtendedData Fig. 5). As compared with Cx43flox/flox mice (littermate controls),BAL from CD11c Cx432/2 mice contained more proinflammatorycytokines (Fig. 4a). Cx43 knockdown in bone-marrow-derived macro-phages did not alter cytokine secretion (Extended Data Fig. 6), rulingout Cx43 depletion as a determinant of the response. LPS-inducedmortality was higher in CD11c Cx432/2 mice than in littermate controls(Fig. 4b). In CD11c Cx432/2 mice, LPS caused greater degradation ofI-kBa (Fig. 4c) and increased nuclear translocation of NF-kB (ExtendedData Fig. 7a). AM numbers did not differ between wild-type andCD11c Cx432/2 mice (Extended Data Fig. 7b). Together, these find-ings indicate that Cx43 knockout in AMs augmented LPS-inducedinflammation and lung injury, indicating that AM-epithelium GJCswere protective.

An increase in intracellular Ca21 activates the Ca21/calmodulin-dependent kinase kinase (CAMKK) and its downstream target, thepro-survival kinase Akt23,24. As these CAMKK-induced mechanismsare undetermined for lung inflammation, we immunoprecipitatedCAMKKa or Akt from wild-type lungs. In each case, LPS enhanced

b

Time after LPS (h)

24 72 120

Surv

ival p

rob

ab

ility

(%

)

0

25

50

75

100

LC

Cx–/–

*

a

TNF-α

0

25

50

LPSPBS

IL-6

0

0.5

1.0

1.5

2.0

ND ND

LPSPBS

*

IL-1β

0

25

50

75

100

125

LPSPBS

* MIP-1α

0

50

100

150

200

ND ND

LPSPBS

*

RANTES

0

250

500

750

1,000

LPSPBS

*

*

TARC

0

10

20

30

LPSPBS

*

*

LC

Cx–/–

e

p-Akt Thr 308

Akt

LC Cx–/–

0

200

400

600

800

1,000

LPSPBS

CXCL5

Co

ncentr

atio

n (p

g m

l–1)

0

50

100

150

200

250

300

LPSPBS

CXCL1

c

p-Akt Thr 308

Akt

LC LC

PBS LPS

Cx–/– Cx–/–

I-κBαp-Akt Thr 308

CAMKKα

Akt

Actin

I-κBα

Sc Sc

PBS LPS

Si Si

f

1

2

3

4

5

LPSPBS

BA

L leuko

cyte

s (10

6) Sc

Si

* g

*

IP: CAMKKα

p-Akt Thr 308

CAMKKα

LPS PBS

d IP: AKT

Akt

LPS PBS

Co

ncentr

atio

n (p

g m

l–1)

Co

ncentr

atio

n (p

g m

l–1)

Co

ncentr

atio

n (ng

ml–

1)

Co

ncentr

atio

n (ng

ml–

1)

Co

ncentr

atio

n (p

g m

l–1)

Co

ncentr

atio

n (p

g m

l–1)

Co

ncentr

atio

n (p

g m

l–1)

b

Time after LPS (h)

24 72 120

Surv

ivalp

rob

ab

ility

(%)

0

25

50

75

100

LC LC

Cx–/–

*

a

TNF-α

25

50

IL-6

0

0.5

1.0

1.5

2.0

ND ND

LPSPBS

*

IL-1β

0

25

50

75

100

125*

MIP-1α

0

50

100

150

200

ND ND

*

RANTES

0

250

500

750

1,000

LPSPBS

*

*

TARC

0

10

20

30

LPSPBS

*

*

LC

Cx–/–

e

p-Akt Thhr 308

Akt

LC Cx–/–

0

200

400

600

800

1,000

LPSPBS

CXCL5

Co

ncentr

atio

n (p

g m

l–1)

0

50

100

150

200

250

300

LPSPBS

CXCL1

0

LPSPBS

0

LPSPBS

ND ND

LPSPBS

c

p-Akt Thr 30

Akt

LC LC

PBS LPS

Cx–/– Cx–/–

I-κBαt Thr 308p-Akt

CAMKKC α

Akt

Actin

I-κBα

Sc Sc

PBS LPS

Si Si

f

1

2

3

4

5

LPSPBS

BA

L leuko

cyte

s (10

6) Sc

Si

*g

*

α

p-Akt Thr 308

CAMKKα

08

IP: CAMKKα

LPS PBS

ddIP: AKT

Akt

LPS PBS

Co

ncentr

atio

n (p

g m

l–1)

Co

ncentr

atio

n (p

g m

l–1)

Co

ncentr

atio

n (ng

ml–

1)

Co

ncentr

atio

n (ng

ml–

1)

Co

ncentr

atio

n (p

g m

l–1)

Co

ncentr

atio

n (p

g m

lC

–1)

Co

ncentr

atio

n (p

g m

lC

–1)

Figure 4 | AM–epithelial signalling. a, BAL cytokine enzyme-linkedimmunosorbent assay (ELISA) in CD11c Cx432/2 (Cx2/2) and littermatecontrol (LC) mice (n 5 3 or 4). ND, not detectable. b, Kaplan–Meier plots.n 5 16 littermate control mice, 17 CD11c Cx432/2 mice. c, d, Data are from24 h after treatment (n 5 4). IP, immunoprecipitation; p, phospho. e, In situimmunofluorescence (red) of the alveolar epithelium and YFP1 AMs

(yellow/green) 24 h after LPS (n 5 4). Scale bar, 20mm. f, g, Lung lysate westernblots and BAL leukocyte counts 24 h after LPS in lungs given scrambled (Sc) orCAMKKa-specific (Si) siRNA. n 5 4 for PBS plus siRNA; n 5 5 for LPSplus siRNA; n 5 6 for other conditions. All blots are from the same sampleset (n 5 3). CAMKKa and actin were processed on different gels. Bars,mean 6 s.e.m. *P , 0.05 versus littermate control or scrambled RNA.

a SynNon

Num

ber

of

Ca

2+

sp

ikes p

er

10 A

Ms

0

2

4

6

WT

My–/

–

Cx–/

–

* * *

*

Ca

2+

(% o

f m

ean

fluo

-4 fl

uo

rescence)

Ep

ithelia

l cell

C

a2+ s

pik

es

0

1

2

PBS

WT

LPS

WT

* *

*

c

My–/

–

Cx–/

–

e

PBS

BA

L leuko

cyte

s (10

6)

0

1

2

3

4

5

LPS

* *

*

b

80

100

120WT

0 10 20Time (min)

0 10 200 10 20

My–/– Cx–/–

d Wild type CD11c Myd88–/– CD11c Cx43–/–

CD11b

Calcein

green

Ly6G

Merge

WT

WT

My–/

–

Cx–/

–

Figure 3 | AM MyD88 in inflammatory signalling. a–c, Synchronized (Syn)and non-synchronized (Non) Ca21 spikes (arrows) and oscillations in wild-type (WT, n 5 12), CD11c Myd882/2 (My2/2, n 5 5) and CD11c Cx432/2

(Cx2/2, n 5 5) mice. d, Alveoli (green), and Ly6G1 (red) and CD11b1

neutrophils (blue) 24 h after LPS (n 5 4). Scale bar, 30mm. e, Responses are 24 hafter treatments. LPS wild type, n 5 9; others, n 5 4. Bars, mean 6 s.e.m.*P , 0.05.

LETTER RESEARCH

2 7 F E B R U A R Y 2 0 1 4 | V O L 5 0 6 | N A T U R E | 5 0 5

Macmillan Publishers Limited. All rights reserved©2014

pull-down of the corresponding binding partner (Fig. 4d), and it enhancedAkt phosphorylation in wild-type, but not in CD11c Cx432/2 mice(Fig. 4c). Imaging indicated that the alveolar epithelium was the siteof the LPS-induced enhancement of phospho-Akt expression (Fig. 4e).This effect was inhibited in CD11c Cx432/2 mice (Fig. 4e) and bytreatment with the intracellular Ca21 chelator BAPTA-AM (ExtendedData Fig. 8a). Short interfering RNA (siRNA)-induced knockdownof CAMKKa decreased Akt phosphorylation, whereas it increasedI-kBadegradation and BAL leukocyte counts (Fig. 4f, g). We generatedSPC Cx432/2 mice lacking Cx43 in the alveolar epithelium25. InSPC Cx432/2 mice, LPS-induced responses were similar to those ofCD11c Cx432/2 mice, in that Akt phosphorylation decreased and BALleukocyte counts increased (Extended Data Fig. 8b). Thus, loss of Cx43on either face of AM-epithelial GJCs induced similar effects. Thesefindings indicate that Cx43-based AM GJCs suppressed inflammationthrough CAMKKa-induced phosphorylation of epithelial Akt.

Our studies highlight the importance of intercellular connectivity inlung immunity. AMs critically elicit lung inflammation. However, con-comitantly, sets of alveolus-attached AMs intercommunicate immuno-suppressive signals. Cx43 deletion in AMs increased the secretion ofcytokines that were likely to be predominantly of AM (MIP-1a) and ofepithelial (CXCL1 and 5) origin, suggesting the possibility that AMsand the epithelium might mutually suppress cytokine release. Previouslung studies implicated syncytial connectivity in the endothelium andepithelium in surfactant secretion17, leukocyte recruitment26 and hyp-oxic vasoconstriction27. Here we show that Cx43high AMs co-opt syncyt-ial communication to subvert lung inflammation. This communicationmight play a part in other forms of lung inflammation, such as thoseinvolving tolerogenic responses to antigen. Although future studies areneeded to elucidate further the roles of Ca21-regulatory mechanisms inthis process, especially regarding second messengers such as Ins(1,4,5)P3

that can diffuse through GJCs28, we propose that Cx43 expression inAMs might provide a drug-delivery focus for new therapeutics forinflammatory lung disease.

METHODS SUMMARYAll animal experiments were approved by the Institutional Animal Care and UseCommittee of Columbia University Medical Center. We imaged isolated, blood-perfused lungs by laser scanning microscopy (LSM 510 META; Zeiss)11. Alveoliwere imaged to a depth of 40mm from the pleura. We loaded alveolar cells withdyes and reagents by alveolar micropuncture11. LPS concentrations were 1 mg kgbody weight21 for all experiments, and 25 mg kg21 for survival studies. We infusedcalcein-stained S. aureus (1 3 108 bacteria ml21) by alveolar micropuncture. ForCa21 imaging (one image every 5 or 10 s), we microinfused alveoli with fluo-4. Forphotolytic Ca21 uncaging17, we targeted single cells co-loaded with fluo-4 and theultraviolet-radiation-sensitive Ca21 cage, o-Nitrophenyl EGTA, with high-intensityultraviolet illumination (,320 nm, 10 pulses s21) in 2-mm-diameter spots. In situCx43, NF-kB and Akt staining was carried out after fixation and permeabilizationof the alveolus. We quantified Cx43 mRNA by qPCR in AMs sorted from BAL andlung tissue samples (Influx Cell Sorter; BD Biosciences). BAL and cell culturesupernatant cytokines were analysed by ELISA. Western blot analyses and co-immunoprecipitations were performed as previously described29. siRNA was com-plexed with freshly extruded liposomes and intranasally instilled.

Online Content Any additional Methods, Extended Data display items and SourceData are available in the online version of the paper; references unique to thesesections appear only in the online paper.

Received 8 April; accepted 20 November 2013.

Published online 19 January 2014.

1. Laskin, D. L., Sunil, V. R., Gardner, C. R. & Laskin, J. D. Macrophages and tissueinjury: agents of defense or destruction? Annu. Rev. Pharmacol. Toxicol. 51,267–288 (2011).

2. Medzhitov, R. Toll-like receptors and innate immunity. Nature Rev. Immunol. 1,135–145 (2001).

3. Thorley, A. J. et al. Differential regulation of cytokine release and leukocytemigration by lipopolysaccharide-stimulated primary human lung alveolar type IIepithelial cells and macrophages. J. Immunol. 178, 463–473 (2007).

4. Kuebler, W. M., Parthasarathi, K., Wang, P. M. & Bhattacharya, J. A novel signalingmechanism between gas and blood compartments of the lung. J. Clin. Invest. 105,905–913 (2000).

5. Maus, U. A. et al. Role of resident alveolar macrophages in leukocyte traffic into thealveolar air space of intact mice. Am. J. Physiol. Lung Cell. Mol. Physiol. 282,L1245–L1252 (2002).

6. Zhang, P., Summer, W. R., Bagby, G. J. & Nelson, S. Innate immunity andpulmonary host defense. Immunol. Rev. 173, 39–51 (2000).

7. Cohen, T. S. & Prince, A. S. Activation of inflammasome signaling mediatespathologyof acuteP. aeruginosapneumonia. J.Clin. Invest. 123,1630–1637 (2013).

8. Guth, A. M. et al. Lung environment determines unique phenotype of alveolarmacrophages. Am. J. Physiol. Lung Cell. Mol. Physiol. 296, L936–L946 (2009).

9. Caton, M. L., Smith-Raska, M. R. & Reizis, B. Notch–RBP-J signaling controls thehomeostasis of CD82 dendritic cells in the spleen. J. Exp. Med. 204, 1653–1664(2007).

10. Srinivas, S. et al. Cre reporter strains produced by targeted insertion of EYFP andECFP into the ROSA26 locus. BMC Dev. Biol. 1, 4 (2001).

11. Islam, M. N. et al. Mitochondrial transfer from bone-marrow-derived stromal cellsto pulmonary alveoli protects against acute lung injury. Nature Med. 18, 759–765(2012).

12. Miller, J. C. et al. Deciphering the transcriptional network of the dendritic celllineage. Nature Immunol. 13, 888–899 (2012).

13. Thornton, E. E. et al. Spatiotemporally separated antigen uptake by alveolardendritic cells and airway presentation to T cells in the lung. J. Exp. Med. 209,1183–1199 (2012).

14. Kirby, A.C., Coles,M.C.& Kaye,P.M.Alveolarmacrophages transportpathogens tolung draining lymph nodes. J. Immunol. 183, 1983–1989 (2009).

15. Lindert, J., Perlman, C. E., Parthasarathi, K. & Bhattacharya, J. Chloride-dependentsecretion of alveolar wall liquid determined by optical-sectioning microscopy. Am.J. Respir. Cell Mol. Biol. 36, 688–696 (2007).

16. Pfenniger, A., Chanson, M. & Kwak, B. R. Connexins in atherosclerosis. Biochim.Biophys. Acta 1828, 157–166 (2013).

17. Ichimura, H., Parthasarathi, K., Lindert, J. & Bhattacharya, J. Lung surfactantsecretion by interalveolarCa21 signaling.Am. J. Physiol. LungCell.Mol. Physiol. 291,L596–L601 (2006).

18. Wong, C. W. et al. Connexin37 protects against atherosclerosis by regulatingmonocyte adhesion. Nature Med. 12, 950–954 (2006).

19. Akira, S.&Takeda,K. Toll-like receptor signalling.NatureRev. Immunol. 4,499–511(2004).

20. Li, H. et al. Toll-like receptor 4-myeloid differentiation factor88signaling contributesto ventilator-induced lung injury in mice. Anesthesiology 113, 619–629 (2010).

21. Imai, Y. et al. Identification of oxidative stress and Toll-like receptor 4 signaling as akey pathway of acute lung injury. Cell 133, 235–249 (2008).

22. Subramanian, M., Thorp, E., Hansson, G. K. & Tabas, I. Treg-mediated suppressionof atherosclerosis requires MYD88 signaling in DCs. J. Clin. Invest. 123, 179–188(2013).

23. Yano, S., Tokumitsu, H. & Soderling, T. R. Calcium promotes cell survival throughCaM-K kinase activation of the protein-kinase-B pathway. Nature 396, 584–587(1998).

24. Chen, B. C., Wu, W. T., Ho, F. M. & Lin, W. W. Inhibition of interleukin-1b-inducedNF-kB activation by calcium/calmodulin-dependent protein kinase kinase occursthrough Akt activation associated with interleukin-1 receptor-associated kinasephosphorylation and uncoupling of MyD88. J. Biol. Chem. 277, 24169–24179(2002).

25. Perl, A. K., Wert, S. E., Nagy, A., Lobe, C. G. & Whitsett, J. A. Early restriction ofperipheral and proximal cell lineages during formation of the lung. Proc. Natl Acad.Sci. USA 99, 10482–10487 (2002).

26. Parthasarathi, K. et al. Connexin 43 mediates spread of Ca21-dependentproinflammatoryresponses in lungcapillaries. J.Clin. Invest.116,2193–2200(2006).

27. Wang, L. et al. Hypoxicpulmonaryvasoconstrictionrequires connexin 40-mediatedendothelial signal conduction. J. Clin. Invest. 122, 4218–4230 (2012).

28. Decrock, E. et al. IP3, a small molecule with a powerful message. Biochim. Biophys.Acta 1833, 1772–1786 (2013).

29. Huang, B. X. & Kim, H. Y. Effective identification of Akt interacting proteins by two-step chemical crosslinking, co-immunoprecipitation and mass spectrometry.PLoS ONE 8, e61430 (2013).

Supplementary Information is available in the online version of the paper.

Acknowledgements WethankB.Reizis forprovidingtheCd11c-cremiceandJ.Whitsettfor providing the Spc-cre mice. We thank I. Tabas for discussions. This study wassupportedbyUSNational InstitutesofHealthgrantsHL78645,HL57556andHL64896to J.B., HL73989 to A.S.P., and Parker B. Francis Fellowships to T.S.C. and M.N.I.

Author Contributions K.W. designed and carried out the experiments, prepared thefigures and wrote the initial manuscript. G.A.G. contributed to western blot,immunoprecipitation and FRAP experiments. M.N.I. carried out the NF-kB in situstainings, andcontributed toBAL cell countingand survival studies.M.S. performed theantigen-presentation assay and provided the CD11c Myd882/2 mice. T.S.C. provided S.aureus and contributed to ELISA studies. A.S.P. contributed to the experimental design.J.B. was responsible for the overall project, designed the experiments and wrote theinitial manuscript. All authors edited the manuscript.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. The authors declare no competing financial interests.Readers are welcome to comment on the online version of the paper. Correspondenceand requests for materials should be addressed to J.B. (jb39@columbia.edu).

RESEARCH LETTER

5 0 6 | N A T U R E | V O L 5 0 6 | 2 7 F E B R U A R Y 2 0 1 4

Macmillan Publishers Limited. All rights reserved©2014

METHODSFluorophores. We purchased fluo-4 AM, calcein AM, calcein red AM and TO-PRO-3 Iodide (642/661) from Invitrogen.Antibodies. We purchased fluorescence-tagged antibodies against CD11c (cata-logue no. 11-0114), MHC-II (catalogue no. 17-5321), CD11b (catalogue no. 12-0112,17-0112), CD4 (catalogue no. 11-0043-85), CD8 (catalogue no. 17-0081-81), CD3(catalogue no. 17-0032-82) and Ly6G (catalogue no. 12-5931) from eBiosciences;Alexa Fluor 633 goat anti-rabbit IgG antibody (catalogue no. A-21071) and AlexaFluor 488 goat anti-rabbit IgG antibody (catalogue no. A-11008) from Invitrogen;and Siglec F antibody (catalogue no. 552126) from BD Pharmingen. AppropriateIgG controls were purchased from eBiosciences. We purchased I-kBa antibody(catalogue no. sc-371-G), NF-kB p65 antibody (catalogue no. sc-109), Cx43 anti-body (catalogue no. sc-9059) and CAMKKa antibody (catalogue no. sc-11370)from Santa Cruz Biotechnology, and phospho-Akt (Thr 308) (catalogue no. 2965),Akt (pan) (catalogue no. 4691) and Cx43 antibody (catalogue no. 3512) from CellSignaling. As secondary antibodies for western blot, we purchased goat anti-rabbitIgG-HRP (catalogue no. sc-2030), donkey anti-goat IgG-HRP (catalogue no. sc-2033) from Santa Cruz.Reagents. We purchased o-Nitrophenyl EGTA (NP-EGTA) (100mM) and BAPTA(100mM) from Invitrogen, GAP27 (500mM) from Tocris, GAP26 (200mM) fromAlpha Diagnostics, xestospongin C (25mM) from Calbiochem, saponin (0.01%)from ICN Biomedicals, Tween 20 (0.5%) from BIO-RAD and LPS (E. coli 0111:B4),fluorescent LPS (E. coli 0111:B4), Triton X-100 (0.2%), PPADS (100mM) anddoxycycline from Sigma. We purchased IL-6 ELISA kit from BioLegend andCXCL5 and MIP-1a ELISA kits from R&D systems.Solutions. Agents were dissolved in a HEPES-buffered vehicle of pH 7.4 andosmolarity of 295 mOsm containing 150 mM Na1, 5 mM K1, 1 mM Ca21,1 mM Mg21 and 10 mM glucose.Animals. All animal procedures were approved by the Institutional Animal Careand Use Committee of Columbia University Medical Center. Animals were between2–6-months old and were age and sex matched. All mice were on a C57BL/6background, except Balb/c mice (Jackson Laboratory), which we used for antigen-presentation assays, and SPC Cx432/2 mice, which were on a B6FVBF2 back-ground. Cd11c-cre mice were provided by B. Reizis, and Spc-rtTA/TetO-cre (Spc-cre)mice were provided by J. Whitsett, Cx43flox/flox (stock no. 008039) and C57BL/6wild-type mice were purchased from Jackson Laboratory. To achieve pan-epithelialknockout of Cx43 in Spc Cx432/2 mice, we maintained pregnant females on doxy-cycline (1 mg ml21 in drinking water) for 48 h from embryonic (E) day E6.5–8.5(ref. 25). Gestation was dated from the day of vaginal plug formation.Acute lung injury. We intranasally instilled LPS in sterile PBS at concentrations of1 mg kg body weight21 for all experiments, and 25 mg kg21 for survival studies.We instilled control mice with PBS.Bacteria. S. aureus LAC USA300 (ref. 30) and E. coli (clinical isolate) grew onLuria–Bertani (LB) agar at 37 uC. For infection, we inoculated LB broth with asingle colony that grew overnight at 37 uC. We stained stationary phase S. aureuswith calcein (5mM) and microinfused them at concentrations of 1 3 108 cellsby alveolar micropuncture. E. coli was intranasally instilled at concentrations of1 3 106 bacteria ml21 in 1.5 ml kg body weight21 PBS.Lung imaging and in situ immunofluorescence. We used our previouslyreported methods to establish isolated blood-perfused mouse lungs for imagingexperiments11. We imaged the lungs with a laser scanning microscopy system(LSM 510 META; Zeiss)11. All dyes and reagents were microinfused by alveolarmicropuncture11. In all experiments in which we infused more than one dye, weconfirmed the absence of bleed-through between fluorescence emission channels.We microinfused fluo-4 (10mM) for 45 min, all other dyes (10mM) and fluorescence-tagged antibodies (4mg ml21) for 20 min. Antibody infusions were followed bywashout. For administrations through the permeabilized pleura, we topicallyapplied saponin (0.01%) and then antibodies. We performed in situ staining forCx43, NF-kB and Akt after we fixed the lungs with paraformaldehyde (4%) for20 min. After permeabilizing the alveolar epithelium with Tween 20 (0.5%) for Cx43and Akt staining, or with Triton X-100 (0.2%) for NF-kB staining, we blocked thetissue for 20 min with 1% fetal bovine serum (FBS) solution, then microinfusedprimary antibody (45 min) and secondary antibody (30 min) in 1% FBS-containingHEPES buffer. Nuclear staining was established with 1mM TO-PRO-3 iodide insolution with the secondary antibody. Each step was followed by a 15 min washoutwith HEPES buffer containing 1% FBS and 0.01% Tween 20. All images wererecorded as single images and processed using MetaMorph imaging software orImage J. Brightness and contrast adjustments were applied to individual colourchannels of entire images and equally to all experimental groups. We did not applyfurther downstream processing or averaging.Ca21 determinations and uncaging. We detected cytosolic Ca21 using fluo-4.We recorded Ca21 responses at one image per every 5 or 10 s for 20 min. For photo-lytic Ca21 uncaging, we loaded alveoli with fluo-4 and the ultraviolet-light-sensitive

Ca21 cage, NP-EGTA. Then, we targeted a high-intensity ultraviolet beam (,320 nm,10 pulses s21 for 20–30 s) that was 2mm in diameter to single epithelial cells orAMs.Cell isolation and immunofluorescence. We did all procedures on ice withCa21-free PBS containing 2 mM EDTA and 1% FBS, hereafter called ‘PBS’.Centrifugations were at 500g for 5–10 min. To isolate BAL AMs, we lavaged fivemice per experiment with 33 1 ml PBS. For lung tissue AM isolation, we perfusedlungs with 5 ml PBS through vascular cannulas to clear blood. We carefully passedlung tissue and spleens and through a 40-mm cell strainer (BD Biosciences). Westained splenic T cells with CD4 and CD8, splenic dendritic cells with CD11c, andBAL and lung tissue AMs with CD11c and MHC-II antibodies. Incubations werefor 1 h at a concentration of 2mg ml21. We sorted the cells using an Influx CellSorter (BD Biosciences) or analysed cells by flow cytometry (LSR II, BD Biosciences).We fixed and permeabilized AMs for Cx43 protein immunofluorescence. Cx43 orisotype-specific IgG, and then secondary antibody incubations, were for 1 h each,after which we performed a cytospin (Thermo Scientific).Antigen presentation assay. We assayed antigen-presentation activity by mixedleukocyte reaction. We used lung AMs or splenic dendritic cells as antigen-presentingcells and splenic T cells from C57BL/6 or Balb/c mice as responder cells. All cellswere isolated by FACS-based sorting using specific antibodies. We confirmed cellviability of .95% as estimated by Trypan blue dye (Invitrogen) exclusion. Welabelled splenic T cells with Cell Trace CFSE (5-(and 6)-carboxyfluorescein diacetatesuccinimidyl ester) Proliferation Kit (Invitrogen). Per well of a 96-well plate, weincubated 15,000 T cells with 5,000 antigen-presenting cells in DMEM supple-mented with 10% FBS, penicillin and streptomycin. Incubation of C57BL/6 T cellswith isogenic splenic dendritic cells served as a negative control; Balb/c T cells withC57BL/6 splenic dendritic cells served as a positive control. C57BL/6 AMs fromBAL were incubated with Balb/c T cells. After 3 days of co-culture, we measuredT-cell proliferation by the CFSE dye dilution technique on CD31 gated cells byflow cytometry (FACScalibur; BD Biosciences). We set gating conditions of theflow cytometer using isotype- and fluorophore-matched IgGs. We analysed prim-ary data using standard software (FlowJO; Tree Star).RNA extraction and quantitative RT–PCR. We extracted total RNA from 100–200,000 AMs per sample using the RNAqueous Micro Kit (Invitrogen). We con-verted equal amounts of RNA in cDNA using SuperScript III First-strand Systemfor RT–PCR (Invitrogen). We performed TaqMan gene expression assays (AppliedBiosystems) for Cx43 (Gja1, Mm00439105_m1) and for Gapdh as a housekeepinggene. We performed PCRs in triplicates using an Applied Biosystems 7300 Real-Time PCR system. Data had identical cycle thresholds for Gapdh.ELISA. We lavaged mice with 1 ml Ca21-free PBS, centrifuged the samples at 500gfor 15 min and preserved the supernatant for cytokine determinations. Sampletesting was carried out by Quansys Biosciences using a multiplex chemilumin-escence assay (Q-plex) for the detection of mouse cytokines and chemokines.ELISAs for IL-6, MIP-1a and CXCL5 were performed according to the manufac-turer’s manual.Immunoprecipitation. We cleared blood and BAL leukocytes from lungs andBAL leukocytes by perfusion of the vasculature with 5 ml ice-cold PBS and by BALwith 33 1 ml ice-cold PBS. We then homogenized the lungs in IP Lysis buffer(Pierce). We immunoprecipitated Akt or CAMKKa from 750mg protein obtainedfrom freshly isolated whole lung homogenates using Protein A/G PLUS-Agarosebeads (Santa Cruz)29. For immunoprecipitation, we covalently linked the antibodyto Protein A/G beads as previously described29.Western blot analysis. We homogenized freshly dissected lungs that were clearedfrom blood and BAL leukocytes in protein lysis buffer, and subjected 50–75mg ofthe lung lysates to western blot analysis29. We used primary and secondary anti-bodies following the manufacturer’s instructions. We blotted for pan-Akt, actin orCAMKKa to assess for equal protein loading. We imaged the blots using a Kodakmolecular imaging station (IS4000MM).siRNA experiments. We purchased Camkka siRNA (SMARTpool siGENOMECamkk1 siRNA, catalogue no. M-049735-01-0050) and Cx43 siRNA (SMARTpoolsiGENOME Gja1 siRNA, catalogue no. M-051694-01-0005) (Thermo Scientific).For CAMKKa knockdown experiments, we intranasally instilled mice with 50mgsiRNA complexed with freshly extruded liposomes as previously described31. Bone-marrow-derived macrophages were incubated in media containing Cx43 siRNA(2mg ml21 in lipofectamine/Opti-MEM solution) for 72 h, then treated with1mg ml21 LPS for 24 h. We determined cytokine secretion in cell culture super-natants by ELISA. We confirmed Cx43 and CAMKKa knockdown by western blot.Statistical analysis. The animals were inbred mice, hence no randomization wasrequired. Other than survival experiments and BAL leukocyte counts, in which theinvestigator was blinded to mouse genotype and/or treatment, there was no blind-ing as animals were marked as having received LPS or buffer. All values areexpressed as mean 6 s.e.m. Data are for $10 AMs per imaging field in each

LETTER RESEARCH

Macmillan Publishers Limited. All rights reserved©2014

experiment. We compared paired observations using paired two-tailed Student’st-test or Wilcoxon rank test. We applied an analysis of variance with Bonferroni’spost-hoc analysis for multiple comparisons. Mouse survival was compared usingKaplan–Meier statistics. There was no pre-specified effect size. Each experimentwas replicated at least three times as independent biological replicates, as indicatedin the figure legends. Differences were considered significant at P , 0.05.

30. Centers for Disease Conrol and Prevention. Methicillin-resistantStaphylococcus aureus infections in correctional facilities—Georgia,California, and Texas, 2001–2003. MMWR Morb. Mortal. Wkly. Rep. 52, 992–996(2003).

31. Rowlands, D. J. et al.Activation of TNFR1 ectodomain shedding by mitochondrialCa21 determines the severity of inflammation inmouse lungmicrovessels. J. Clin.Invest. 121, 1986–1999 (2011).

RESEARCH LETTER

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 1 | AM function and distribution a, Fluorescence-activated cell sorting (FACS) plots and group data for antigen-presentationassays on AMs from BAL samples. First box shows gating scheme forCD11chigh MHC-IIlow AMs (black box). n 5 3; 2, negative control; 1, positivecontrol. *P , 0.05 versus negative control. b, Quantifications of AMs andalveoli (Alv). *P , 0.05 versus left bar; n 5 20 imaging fields. c, Images and

quantifications of AM numbers before (top) and after (bottom) indicatedtreatments (n 5 4). Arrowheads indicate AMs. 4 h, 4 h imaging; BAL, 33 1 mlBAL; Bu, 20 min injection of buffer containing calcein; SA, S. aureus injection.Scale bar, 30mm. *P , 0.05 versus before. d, Gating scheme for sorting ofCD11chigh MHC-IIlow AMs (black box) from lung tissue samples. Bars aremean 6 s.e.m.

LETTER RESEARCH

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 2 | Cx43 knockout in AMs. a, Images show CD11c(green) and Cx43 (red) immunofluorescence and nuclei (blue) in lung-tissue-adherent AMs of wild-type (top) and CD11c Cx432/2 (Cx2/2; middle) mice.Bottom image is IgG control for Cx43 antibody. Bars quantify Cx43 expression.Ctl, control. Scale bar, 5mm. *P , 0.05 versus control; n 5 4. b, Sequential

images show progressive colocalization (yellow) of AMs (arrowheads) andS. aureus (green). Scale bar, 15mm. Replicated three times. c, Bars quantify AMnumbers 1 h after bacterial challenge. n 5 4 for wild type, n 5 3 forCD11c Cx432/2. Bars are mean 6 s.e.m.

RESEARCH LETTER

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 3 | Spatiotemporal uptake of fluorescent LPS in AMsand lung dendritic cells. a, Gating scheme for sorting of CD11chigh MHC-IIlow

AMs and CD11chigh MHC-IIlhigh dendritic cells (DCs; black boxes) from BAL(top) and lung tissue (bottom) samples. b, Representative FACS plots forfluorescent LPS uptake (green box) in AMs (black line) and dendritic cells (red

line) recovered from BAL (left) and lung tissue samples after BAL (right).c, Group data quantify percentages of dendritic cells and AMs among CD11c-expressing cells in BAL (B) and lung tissue after BAL (T). d, Quantification ofLPS uptake in cells of AM and dendritic-cell populations shown in c. n 5 4 fortissue, n 5 5 for BAL samples. Bars are mean 6 s.e.m. *P , 0.05.

LETTER RESEARCH

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 4 | AM distribution and intercommunication ofCa21 spikes. a, Images and sketch show the alveolar trajectory (dashed line) ofa Ca21 spike that originates in one AM (left arrowhead) and passes throughalveolar type II cells (arrows) to another AM (right arrowhead). Scale bar,20mm. b, c, Bars quantify fluo-4 fluorescence in baseline (b) and 24 h LPS-treated (c) lungs at the beginning (0) and after 20 min of imaging (20) (n 5 4).

d, Quantification of AM Ca21 spikes 4 h after LPS in control (Ctl, n 5 4)and leukocyte-depleted (Leu, n 5 3) lungs. e, Bar diagram shows AMs perimaging field in untreated lungs (Ctl) and at different time points after LPS.Ctl, n 5 8; 1 h, n 5 6; 4 h, n 5 4; 24 h, n 5 11. f, Bars quantify Cx43 expression.My2/2, CD11c Myd882/2. Bars are mean 6 s.e.m. n 5 4.

RESEARCH LETTER

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 5 | Cellular responses to E. coli. a, Bars quantify BALleukocytes 24 h after E. coli instillation in littermate control (Ctl) andCD11c Cx432/2 (Cx2/2) mice. Bars are mean 6 s.e.m. *P , 0.05 versuscontrol, n 5 8. b, Western blots for phosphorylated (p-Akt) and total Akt inwhole lung homogenates (n 5 4 lungs).

LETTER RESEARCH

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 6 | Cytokine secretion in response to Cx43knockdown in bone-marrow-derived macrophages. Cells were treated withscrambled or siRNA against Cx43. Differences were significant between LPSand PBS (P , 001), but not within groups (n 5 4). Bars are mean 6 s.e.m.ND, not detected.

RESEARCH LETTER

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 7 | NF-kB immunostaining in the alveolarepithelium and AM distribution in knockout mice. a, Images showimmunofluorescence for NF-kB (green) and alveolar nuclei (red) forCD11c Cx432/2 mice (Cx2/2) and littermate controls (Ctl) (n 5 4). Rectangle

in merged images depicts magnified nuclei. b, Bars quantify AMs per imagingfield in optically viewed lungs 24 h after LPS instillation. Ctl, wild-type mice(n 5 11). Cx2/2, CD11c Cx432/2 mice (n 5 6); My2/2, CD11c Myd882/2

mice (n 5 6). Scale bar, 10mm. Bars are mean 6 s.e.m.

LETTER RESEARCH

Macmillan Publishers Limited. All rights reserved©2014

Extended Data Figure 8 | Ca21 dependence of Akt phosphorylation andBAL leukocyte counts and p-Akt expression in SPC Cx432/2 mice.a, Western blots show phosphorylated Akt (p-Akt) and total Akt afterinstillations of LPS or LPS plus BAPTA (100mM) 4 h before lung isolation(n 5 4). b, Bars quantify BAL leukocytes 24 h after LPS instillation in littermatecontrol (Ctl) and SPC Cx432/2 (SCx2/2) mice. Bars are mean 6 s.e.m.*P , 0.05 versus control, n 5 4. Gels are representative western blots forphosphorylated Akt, Cx43 and total Akt in whole lung homogenates (n 5 4).

RESEARCH LETTER

Macmillan Publishers Limited. All rights reserved©2014