Identification of 14-series sulfido-conjugatedmediators that promote resolution of infectionand organ protectionJesmond Dalli, Nan Chiang, and Charles N. Serhan1

Center for Experimental Therapeutics and Reperfusion Injury, Department of Anesthesiology, Perioperative and Pain Medicine, Harvard Institutes ofMedicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA 02115

Edited by Carl F. Nathan, Weill Cornell Medical College, New York, NY, and approved September 25, 2014 (received for review August 5, 2014)

Upon infection and inflammation, tissue repair and regenerationare essential in reestablishing function. Here we identified potentmolecules present in self-limited infectious murine exudates,regenerating planaria, and human milk as well as macrophagesthat stimulate tissue regeneration in planaria and are proresolv-ing. Characterization of their physical properties and isotopetracking indicated that the bioactive structures contained docosa-hexaenoic acid and sulfido-conjugate (SC) of triene double bondsthat proved to be 13-glutathionyl, 14-hydroxy-docosahexaenoicacid (SCI) and 13-cysteinylglycinyl, 14-hydroxy-docosahexaenoicacid (SCII). These molecules rescued Escherichia coli infection-mediated delay in tissue regeneration in planaria, improving re-generation intervals from ∼4.2 to ∼3.7 d. Administration of SCsprotected mice from second-organ reflow injury, promoting repairvia limiting neutrophil infiltration, up-regulating Ki67, and Roofplate-specific spondin 3. At nanomolar potencies these conjugatesalso resolved E. coli infections by limiting neutrophil infiltrationand stimulating bacterial phagocytosis and clearance as well asefferocytosis of apoptotic cells. Together, these findings identifypreviously undescribed conserved chemical signals and pathwaysin planaria, mouse, and human tissues that enhance host responsesto contain infections, stimulate resolution of inflammation, and pro-mote the restoration of function.

regeneration | omega 3 | leukocytes | inflammation | eicosanoids

Given the rise in antibiotic-resistant infections and the criticalrole that barrier breach plays in microbial invasion, identifi-

cation of new endogenous signals that promote pathogen clear-ance and tissue repair/regeneration is of wide interest (1). Whenself-limited, acute inflammation is a host-protective response(2, 3) that is orchestrated by chemical mediators and is an activeprocess generated by evolutionarily conserved biosynthetic path-ways (3, 4). During initiation of inflammation, potent mediatorsare locally produced that promote vascular leakage, leukocyterecruitment, and pain (5, 6). In disease, these pathways can bedysregulated, leading to heightened inflammatory responses andperpetuation of the disease state (6–9). One approach to regu-lating exuberant inflammatory responses is inhibition of initiatingmediators (e.g., eicosanoids, including prostaglandins) via bio-synthetic enzyme inhibitors (6, 10) and receptor antagonists (5). Inthe context of infection, this approach may be of limited clinicalutility and can have potential drawbacks, including immune sup-pression (7, 8).In self-limited inflammation, endogenous programs are acti-

vated at the onset that regulate the amplitude of the inflammatoryresponses and stimulate resolution (3, 9). Central to these host-protective responses are novel families of endogenous chemicalmediators termed specialized proresolving mediators (SPMs) (9).In sterile inflammation and injury, these mediators actively limitfurther neutrophil recruitment and promote macrophage clear-ance of apoptotic cells and tissue debris (7–9). In self-resolvinginfections, endogenous resolution programs are also activatedduring the early stages of inflammatory responses, with the

up-regulation of select SPMs [including Resolvin (Rv) D1, RvD5,and Protectin D1]. These mediators enhance bacterial killing andclearance along with regulating phagocyte recruitment (11).The levels of these potent leukocyte agonists decline during

the later phase of the self-limited inflammatory response (11),allowing the possibility that other signals may be produced thatregulate leukocyte responses to promote tissue repair and re-generation. Given the pivotal roles of chemical signals in infec-tions, we investigated whether mediators within self-resolvinginfections could regulate tissue repair and regeneration withoutimmunosuppression. Because maresin 1 (7R,14S-dihydroxy-docosa-4Z,8E,10E,12Z,16Z,19Z-hexaenoic acid; MaR1) displayspotent proresolving and tissue-regenerative actions (12, 13), weinvestigated whether previously undescribed signals are producedduring self-limited infections that regulate tissue regeneration. Inthis report, we identify a new pathway and mediators in planaria,mouse, and human tissues that promote repair and regenerationduring infection. Identification of two new sulfido-conjugate (SC)mediators provides the first evidence, to our knowledge, forautacoids produced during the resolution of infections that signalinnate host responses and accelerate repair.

Results and DiscussionTo obtain self-resolving infectious exudates and assess tissueregeneration, we used murine Escherichia coli peritonitis rele-vant to human infections and mapped leukocyte trafficking.E. coli inoculation at 105 colony-forming units (CFUs) per mousei.p. gave a self-limited host response that reached maximalneutrophil infiltration at 12 h and subsequently declined (Fig. 1A).

Significance

Inability of the body to contain infections may lead to collat-eral organ damage resulting from unchecked innate immuneresponses. Here we investigated the chemical signals producedby immune cells to expedite clearance of bacteria and promoteorgan repair and tissue regeneration. We identified moleculesproduced during self-limited infections and in human milk thatpromote clearance of bacteria as well as accelerate tissue re-generation. In addition, these molecules also protected organsfrom exuberant inflammatory responses by limiting selectwhite blood cell recruitment and up-regulating the expressionof proteins involved in tissue repair. Therefore, these resultsidentify new resolution moduli that regulate phagocytes toclear bacteria and activate the regeneration milieu.

Author contributions: C.N.S. conceived the overall research plan and experimental design;J.D., N.C., and C.N.S. designed research; J.D. and N.C. performed research; J.D. and N.C.analyzed data; and J.D., N.C., and C.N.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: cnserhan@zeus.bwh.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1415006111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1415006111 PNAS | Published online October 16, 2014 | E4753–E4761

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Monocyte/macrophage numbers increased between 4 and24 h, demarking onset of the resolution phase. We next isolatedproducts from resolving infectious exudates (i.e., 24 h) andassessed their ability, with planaria, to stimulate tissue regen-eration. Because MaR1 stimulates tissue regeneration (12) andelutes within methyl formate fractions from C18 solid-phaseextraction [referred to as solid-phase extraction chromatographic(SPE-C) isolate fraction 1], we sought evidence for signals indistinct chromatographic fractions. Here we assessed eluates inmethanol fractions (SPE-C isolate fraction 2) for previouslyundescribed signals that displayed tissue-regenerative properties.Planaria undergo both restorative and physiological regenerationvia evolutionarily conserved pathways, making this an ideal sys-tem to identify chemical signals involved in tissue regeneration(14). To this end, planaria (Dugesia japonica) were injured onday 0 and time-dependent head regeneration was monitored. Toquantitate regeneration, we calculated tissue-regeneration in-dices (TRIs) (12). Following head resection, regeneration en-sued, giving a TRImax (maximum tissue regeneration) at 6 d andT50 (the interval at which 50% regeneration, TRI50, occurred) at∼4.3 d (Fig. 1B). Isolates from 24-h infection-resolving exudates

dose-dependently accelerated head regeneration (r2 = 0.91) asearly as 2 d after surgery, shortening T50 to ∼3.3 d. For directcomparison, maresin 1 (Fig. 1B and Fig. S1A), the known dihy-droxy-containing potent macrophage-derived proresolving me-diator (12), accelerated this process to essentially the sameextent. Toward human translation, and because milk carriesnutrients and is appreciated to have products relevant to infantdevelopment and immune status (15), we also assessed tissue-regenerative properties of human milk isolates using the samechromatographic fractions (Fig. 1B). Incubation of planaria withhuman milk isolates dose-dependently accelerated regenerationand reduced T50 from ∼4.3 to ∼3.5 d (Fig. 1B and Fig. S1B).These results demonstrate that both mouse resolving exudatesand human milk possess tissue-regenerative properties that elutewithin the methanol fractions from solid-phase C18 extractions.Next, we investigated whether these bioactive molecules reg-

ulated signaling pathways that trigger head-to-tail differentiationin D. japonica (16). Two days postinjury, in regenerating blaste-mas, isolates from both mouse resolving exudates and humanmilk significantly increased expression of the fibroblast growthfactor receptor-like gene nou-darake (Djndk) and mitogen-acti-vated protein kinase phosphatase gene (Djmpka; Fig. 1C) anddown-regulated DjAdb-Ba (Abdominal-B–like gene), a target Hoxgene of posterior Wnt/β-catenin signaling in neoblast progeny(14, 16). Because extracellular signal-regulated kinase (ERK)regulates expression of these genes (16), we tested whether ERKsignaling was responsible for the actions of molecules from re-solving exudates. ERK inhibition abrogated regenerative actionsof these infectious resolving exudates (Fig. S1C). Together, theseresults suggest that ERK signaling is involved in the regenerativeactions of molecules from resolving infectious exudates.Macrophages are key in regulating host responses during in-

flammation and infections, coordinating both the onset andresolution of inflammatory responses (8, 17, 18). During sterileinflammation, the resolution phase is characterized by increasesin resolution-phase macrophages (rMs) (18) and production ofmaresins (13). Using lipid mediator metabololipidomics, weidentified MaR1 and related isomers within SPE-C isolate frac-tion 1 as well as previously undescribed signals in SPE-C isolatefraction 2 in infectious resolving exudates (n = 3 mice exudates).Hence, SPE-C isolate fraction 2 obtained from both mouse re-solving exudates and human milk carried regenerative properties(Fig. 1 B and C) that were unique and distinct from MaR1 (12),which elutes within SPE-C isolate fraction 1 (Table S1).Based on these findings, we next assessed the role of human

macrophage 12-lipoxygenase (LOX) (19) in the biosynthesis ofthese products. Isolates obtained from human macrophages(HMϕs) carried tissue-regenerative actions with D. japonica thatwere lost in cells transfected with shRNA targeting 12-LOX (Fig.1D). In HMϕs, docosahexaenoic acid (DHA) is a 12-LOX sub-strate in maresin biosynthesis (19, 20); therefore, we investigatedwhether the bioactive products in SPE-C isolate fraction 2 werealso from DHA. Incubation of infectious exudates obtained fromself-resolving peritonitis with radiolabeled DHA demonstratedaccumulation of labeled material in SPE-C isolate fraction 2 atboth 12 h (peak of inflammation) and 24 h (resolution) (Fig.S2A). In these fractions, significantly higher radioactivity fromDHA was present in exudates from mice with self-resolvingperitonitis compared with those with delayed resolving perito-nitis. The presence of four DHA-derived products in these iso-lates was confirmed using liquid chromatography tandem massspectrometry (LC-MS-MS) (Fig. S2B).Peaks I and II gave essentially identical MS-MS fragmentation

patterns, with the parent ion (M+H) displayingm/z at 521. Spectrafrom peaks III and IV also gave identical MS-MS fragmentationwith m/z for the parent ion at 650, suggesting that I and II werelikely related to each other and III and IV were related to eachother (Fig. 2 A and B). Similar results were also obtained with

B

D

A

Time (h)

Monocytes/ Macrophages

Surgical Injury

+Human Milk+E.coli RE

+MaR1

0 1 2 3 4 5 6

100

200

300

400

500

600

Days post surgery

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

*

*

TRI50

T50

C

Surgical Injury

+E.coli RE+Human Milk

Djsfpr-A Djndk DjmpkA DjAdb-Ba

**

**

Rel

ativ

e m

RN

A ex

pres

sion

0

10

20

30

**

* *

**

PMN

Num

ber o

f inf

iltra

ted

Leuk

ocyt

es

(x10

6 /mou

se)

0 102 103 104 105

CD11b

0

102

103

104

105

Ly6G

40.9 0.6

10.947.7

12 24 36 480

2

4

6

E.coli RE Products HM Products

*

*

* *

Djsfpr-ADjndk

Djmpk-ADjAdb-Ba

Initiation Resolution

0 1 2 3 4 5 6

100

200

300

400

500

600

Days post surgery

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

Surgical Injury

+HM 12-LOX shRNA+HM CS shRNA

TRI50

T50

**#

*

TRImax

Fig. 1. New pathway promotes tissue regeneration. (A) Leukocyte re-cruitment following E. coli (105 CFUs per mouse, i.p.) inoculation (Materialsand Methods). Results are mean ± SEM for n = 4 mice per time point. (B)Surgically injured planaria were incubated with SPE-C isolate fraction 2 fromE. coli resolving infectious exudates (REs), human milk, maresin 1 (100 nM), orvehicle (surgical injury; water containing 0.01% EtOH). Tissue regenerationindices were determined. T50, time interval corresponding to 50% of maximaltissue regeneration (TRI50). (C) Gene expression in regenerating blastemas(Bottom). Results are mean ± SEM (n = 3 per group pooled from blastemas of9 animals). (D) Injured planaria were incubated with SPE-C isolate fraction2 from human macrophages transfected with shRNA for 12-lipoxygenase,control scrambled sequence (CS shRNA), or vehicle. (B and D) Results aremean ± SEM for n = 9 planaria per incubation. *P < 0.05, **P < 0.0001 vs.surgical injury group. #P < 0.01 vs. 12-LOX shRNA at day 4.

E4754 | www.pnas.org/cgi/doi/10.1073/pnas.1415006111 Dalli et al.

human milk, regenerating planaria, and human macrophages (Fig.2 C–F and Fig. S2C). In macrophages transfected with 12-LOXshRNA compared with mock transfected cells, quantification ofproducts in SPE-C isolate fraction 2 using LC-MS-MS demon-strated a significant reduction in the levels of products identifiedunder peaks I–IV (Fig. 2G). These results are in line with theobserved reduction in biological activity of these isolates (Fig. 1D),implicating HMϕ 12-LOX in the initiation of these signals.Because 14S-hydro(peroxy)-docosahexaenoic acid (14S-

HpDHA) is the product of the HMϕ 12-LOX biosynthetic pre-cursor to maresins, we tested whether 14S-HpDHA was a pre-cursor of these regenerative molecules. Incubation of HMϕs with14S-HpDHA also gave products III and IV (Fig. 3A) as wellas I and II (Fig. 3B). Products beneath III and IV each gaveUV chromophores with maximum absorbance at 280 nm andshoulders at 270 and 295 nm in the reverse-phase high-pressureliquid chromatography mobile phase (Fig. S3A), characteristicof a conjugated triene double-bond system coupled to an auxo-chrome allylic to the triene such as sulfur (6, 21). This was cor-roborated in incubations with a desulfurization reagent, Raneynickel (6, 21), that gave 14-hydroxy-docosahexaenoic acid (14-HDHA) (Fig. S3 B and C). These results together with MS-MSfragmentation indicated a 13-glutathionyl,14-hydroxy-, and 22-carbon backbone that originated from DHA (Fig. 3A), andtherefore was coined sulfido-conjugated product I (SCI). To gainfurther evidence for this deduced structure, we assessed deu-terium incorporation in SCI with HMϕs and d5-14S-HpDHA(Fig. S4A). d5-SCI gave the expected 5-Da shift in the parent ionmass, from m/z 650 to m/z 655, as well as in the m/z of fragmentscontaining carbons 21 and 22, including that resulting from a di-agnostic 14- to 15-carbon break, which increased in mass fromm/z 109 to m/z 114 (Fig. S4C). The proposed glutathione conju-gate structure was further corroborated by treating SCI (Fig. S4 Band C) with diazomethane. This approach was also used to elu-cidate the structure of the products beneath I and II, namely 13-cysteinylglycinyl, 14-hydroxy-docosahexaenoic acid (SCII; Fig.3B and Figs. S5 and S6).We next confirmed that these molecules carried tissue-

regenerative properties. Planaria incubated with SCI and SCIIgave accelerated tissue regeneration (T50 from ∼4 to ∼3 d; Fig.3C). To investigate the role of endogenous SCs in planaria tissueregeneration, D. japonica was incubated with a LOX inhibitor(baicalein) that reduced levels of both SCI and SCII measured3 d postsurgery (Fig. 3D) and significantly delayed head regrowth(Fig. 3E). This was rescued when planaria were incubated withSCI and SCII (Fig. 3E). We next investigated the ability of SCsto promote tissue regeneration during infection. Incubation ofD. japonica with E. coli gave a delay in regeneration, T50 from∼3.5 to ∼4.2 d, that was rescued by SCI and SCII addition (Fig.3E). Having established that together, SCI and SCII displayedtissue-regenerative actions, we next sought evidence for theirindividual biological actions. Incubation of injured planaria witheither SCI or SCII accelerated head regrowth. This proved to beconcentration-dependent, reducing T50 from ∼5 to ∼3.5 d (Fig. 4A and B and Fig. S7), with an r2 value of 0.85 for SCI and 0.97 forSCII (Fig. S7). In addition, SCI and SCII each up-regulatedgenes involved in head regeneration (Fig. 4C).Bioactive lipid mediators that are peptide conjugates such as

cysteinyl leukotrienes involve GST enzymes in their biosynthesis(10, 22). Therefore, we assessed the role of D. japonica GSTin SC biosynthesis. Whole-mount in situ hybridization (WISH)of uninjured planaria confirmed the expression of D. japonicaGST (DjGst), which was temporally regulated following injuryin regenerating blastemas (Fig. 5A). Double-stranded RNAknockdown ofDjGst (Fig. 5B) delayed tissue regeneration (Fig. 5C)and reduced SC levels (>80%; Fig. 5D) in regenerating planaria.Alignment of deduced D. japonica GST amino acid sequencewith human and mouse GST-mu4 (Table S2) demonstrated >70%

191

100 200 300 400 500 600m/z, Da

0

100 650

632618521

503

418400

343325

308

233

179

187162

109

Rel

ativ

e In

tens

ity (%

)

x3

100 200 300 400 500m/z, Da

0

100

109

173179

191

187205

217 325

343459

503

521

Rel

ativ

e In

tens

ity (%

)

x8

0

1

2

3

4

5

0

2

4

6CS shRNA12-LOX shRNA

Ion

coun

ts x

104

* *

100 200 300 400 500m/z, Da

147109

173

179187191

205217

235

521

503459

343Rel

ativ

e In

tens

ity (%

)

0

100

100 200 300 400 500 600m/z, Da

109162

179191

205

233 308

343325 400

418521

503618

632

650

Rel

ativ

e In

tens

ity (%

)0

100

A B

C D

E F

G

100 200 300 400 500 600m/z, Da

0

100

Rel

ativ

e In

tens

ity (%

)

109

162179

191233 308

343325

400418

521

503 618632

650

x5

Infectious resolving exudates

Regenerating planaria

Human macrophages

100 200 300 400 5000

100

x5

521

109173179

191187

217325343

459

503

m/z, Da

Fig. 2. Identification of previously undescribed signals in SPE-C isolatefraction 2. (A and B) Infectious exudates were obtained at 24 h after in-oculation with E. coli (105 CFUs per mouse). Exudates were then in-cubated with DHA (1 μg/mL, 37 °C, 45 min), products were extracted andsignals investigated by LC-MS-MS. (A) MS-MS spectrum for signals underpeaks III and IV. (B) MS-MS spectrum for signals under peaks I and II.Results are representative of n = 4 mice. (C and D) Planaria were surgicallyinjured; after 3 d, products were extracted and signals were investigatedby LC-MS-MS. Results are representative of n = 20 planaria. (E–G) Humanmacrophages were transfected with shRNA for 12-LOX or CS sequence;products were then extracted and signals were investigated by LC-MS-MS. (G) MRM quantification for products under peaks III and IV (Left) andpeaks I and II (Right). Results for E and F are representative of n = 3macrophage preparations. (G) Mean ± SEM expressed as peak area ioncounts. n = 3 macrophage preparations. *P < 0.05 vs. CS shRNA trans-fected macrophages.

Dalli et al. PNAS | Published online October 16, 2014 | E4755

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

sequence homology between the planarian and mammalian en-zymes, which is also expressed in HMϕs (n = 4). These resultsindicate that a GST enzyme(s) is involved in SC biosynthesis andis involved in tissue regeneration in planaria.With human macrophages, we investigated product–precursor

relationships for DHA with SCI and SCII. DHA was rapidlyconverted by activated human macrophages to SCI, with levelsreaching a maximum at 15 min. SCII was also rapidly producedin these incubations, with levels that remained elevated between15 and 60 min (Fig. 6A). Essentially similar relationships wereobtained with activated human macrophages and 14S-HpDHA(n = 3 separate incubations). To assess whether SCI was a precursorof SCII, we next incubated human macrophages with a γ-glutamyltransferase inhibitor (Acivicin) known to inhibit leukotriene (LT)D4 formation (23). Incubation of activated human macrophageswith this inhibitor gave a significant increase in SCI and a re-duction in SCII levels (Fig. 6B). Similar results were obtained withregenerating planaria, where addition of the inhibitor increasedSCI levels and decreased SCII levels 2 d postinjury (Fig. 6C).Because SCs were biosynthesized in both resolving infectious

exudates (Fig. 2 and Fig. S2) and regenerating planaria (Figs. 2,5, and 6), we determined their ability to stimulate resolutionduring infection in mice. For quantitative assessment of resolu-tion components with E. coli infection, we used resolution in-dices (11). Inoculation of mice with 105 CFUs E. coli (self-limited inoculum) gave Tmax ∼12 h, T50 ∼32 h, and resolutioninterval (Ri) ∼20 h (Fig. 7A). SCI and SCII (50 ng each permouse) significantly reduced neutrophil numbers at 24 h,shortening Ri to ∼10 h (Fig. 7A), and promoted an exudatemacrophage phenotype switch toward an rM phenotype, in-creasing rM markers including T-cell immunoglobulin andmucin domain containing 4 (TIMD4) (∼34%), interleukin-10(IL-10) (∼15%), and Arginase 1 (∼76%) (n = 4 mice per group).SCs also gave significant increases in leukocyte phagocytosis ofE. coli in vivo in mice (Fig. 7B). For human translation, we de-termined the actions of SCI and SCII on human phagocytes,where each dose-dependently increased macrophage effer-ocytosis of apoptotic cells, a key proresolving action (3), toa similar extent as MaR1 (Fig. 7C). Of note, SCI and SCII eachenhanced bacterial phagocytosis and intracellular reactive oxy-gen species (Fig. S8) used in bacterial killing (11, 17).

We tested SC isolates from regenerating planaria to deter-mine whether they displayed actions in mammalian species.Here they significantly reduced neutrophil recruitment in murineperitonitis (Fig. S9A) and exudate eicosanoid levels includingLTB4 and thromboxane (Tx) B2, actions that by direct compar-ison were shared with MaR1 (Fig. S9 B and C). Planarian SCisolates also stimulated human macrophage efferocytosis of ap-optotic neutrophils (Fig. S9D). Hence, these results indicatedthat SCs are antiinflammatory and proresolving, and thattheir structure functions are conserved from planaria, mice,and humans.Vessel occlusion is a common consequence of many inflam-

matory conditions, leading to local ischemia that, upon re-flow, can result in second-organ injury by activated leukocytesand, in extreme cases, organ failure (24). SCI and SCII gavesignificant protection from leukocyte-mediated tissue damage(Fig. 8A), decreasing leukocyte infiltration into lungs andspleens, actions comparable to RvD1 (Fig. S10A). SCs also re-duced plasma eicosanoid levels, including LTB4, LTC4, and TxB2(Fig. S10 B and C). In addition, immunofluorescence analysis oflung sections from mice given SCI and SCII demonstrated an up-regulation of antigen Ki67, which plays a role in cell proliferation(25), and Roof plate-specific Spondin 3 (RSPO3; Fig. 8B), whichdisplays tissue-regenerative actions (26).Several lines of evidence support the proposed biosynthetic

scheme for the formation of SC signals in Fig. 9. They are asfollows: (i) DHA is converted by HMϕ 12-LOX to 14S-HpDHAand 13,14-epoxide intermediate, demonstrated using acid meth-anol trapping with human cells and recombinant enzyme (13, 19).Knockdown of macrophage 12-LOX reduced both SCI andSCII (Fig. 2) as well as loss of the tissue-regenerative actionsof SPE-C isolate fraction 2 (Fig. 1). (ii) Radiolabel from pre-cursor [14C]DHA was recovered in SPE-C isolate fraction 2 thatwas distinct from MaR1-containing fractions (Fig. S2). (iii) Reci-procity existed in product–precursor relationships between DHA,SCI, and SCII (Fig. 6) as well as between 14-HpDHA and SCIand SCII. (iv) Inhibition of planaria LOX reduced both SCI andSCII in regenerating planaria and delayed tissue regenerationthat was rescued by addition of SCI and SCII (Fig. 3). (v)Knockdown of planaria GST reduces both SCI and SCII levels inregenerating planaria and delays tissue regeneration (Fig. 5). (vi)Incubation of human macrophages and regenerating planaria

A

1 2 3 4 5 60

100

200

300

400

500

600

700

TRI50

T50

* *

*

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

Days post surgery

Surgical Injury

+SC

1 2 3 4 5 60

100

200

300

400

500

600

700UninfectedInfected

* *

** 100

200

300

400

500

600

700

1 2 3 4 5 60

*

+SCInfected

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

TRI50 TRI50

T50 T50

Days post surgery Days post surgery 1 2 3 4 5 60

100

200

300

400

500

600

Days post surgery

Surgical Injury

+LI+LI+SC

**

*

* * **

##

##

##

#

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

*

Surgery+LI

*

Are

a be

neat

h th

e cu

rve

(ion

coun

ts x

105 )

0

5

10

15

20

25

B

100 200 300 400 500 600m/z, Da

650

521503

308 343233179162

575418109

325

400191 6145930

100

235

217

618

205187

632

Rel

ativ

e In

tens

ity (%

)

100 150 200 250 300 350 400 450 500 550m/z, Da

Rel

ativ

e In

tens

ity (%

)

0

100

235

191

179

205

109

521

503477

504 459

329325 343

281

147 173

217

C

D E F

173COOH

OH

SNH

NH3O

HOOC

191

179+H

343

145

235

109

+H

205+H

+H

329

-H

115

131-H

5COOH

OH

S

HN NH

O

OHOOC

NH3HOOC

593

191

308+2H

343

+2H521+2H

418

235

109

+H

205+H

+H

Fig. 3. New sulfido conjugates promote regen-eration. Human macrophages (1 × 107 cells per mL)were incubated with 14-HpDHA (1 μM, PBS+/+) andE. coli (1 × 108 CFUs per mL, 30 min, 37 °C), andproducts were isolated by RP-UV-HPLC and assessedby LC-MS-MS. Representative MS-MS spectra used inthe identification of (A) SCI from peaks III and IV and(B) SCII from peaks I and II. Results represent n = 10macrophage preparations. (C) After surgical injury, pla-naria were incubated with SCI plus SCII (10 nM) or ve-hicle (surgical injury; water containing 0.01% EtOH) andregeneration indices determined. (D) After surgical in-jury, planaria were incubated with lipoxygenase in-hibitor (L.I.; 100 μM) or vehicle (surgical injury; watercontaining 0.01% EtOH). SCI (Left) and SCII (Right) werequantified by LC-MS-MS. Results are mean ± SEM. n = 3representative of 40 planaria per group. (E) Surgicallyinjured planaria, were incubated with lipoxygenase in-hibitor (L.I.; 100 μM), L.I. plus SCI plus SCII (SC; 100 nM) orvehicle. (F) After surgical injury, planaria were incubatedwith E. coli (108 CFUs), E. coli plus SCI and SCII (100 nM),or vehicle (surgical injury; water containing 0.01%EtOH), and regeneration indices were determined.Results are mean ± SEM for n = 9 planaria per group.*P < 0.05, **P < 0.01 vs. the respective control group.#P < 0.01, ##P < 0.001 vs. the respective L.I. group.

E4756 | www.pnas.org/cgi/doi/10.1073/pnas.1415006111 Dalli et al.

γ-glutamyl transferase inhibitor led to increased SCI and de-creased SCII (Fig. 6). (vii) SCI and SCII carry distinct structuresfrom that of MaR1 (12), as demonstrated by LC-MS-MS resultsof the free acids (Figs. 2 and 3 and Fig. S2) and trimethyl anddimethyl ester derivatives, as well as deuterium incorporationfrom DHA (Figs. S4 and S6). (viii) Identification of 14-HDHAfrom incubations of SCI and SCII with Raney nickel as well as theSCI and SCII respective UV chromophores supports the pres-ence of a sulfido group allylic to a triene double-bond system ineach of these molecules (Figs. S3 and S5). (ix) SCI and SCIIisolated using RP-UV-HPLC each separately regulated humanphagocyte responses and promoted tissue regeneration in pla-naria in a dose-dependent manner (Figs. 4 and 7 and Figs. S7 andS8). (x) When administered together in vivo, SCI and SCII reg-ulated host mouse responses to E. coli infections promotingclearance of bacteria and resolution of infections. (xi) SCI andSCII together protected against second-organ reflow injury (col-lateral tissue damage) and promoted tissue repair (Figs. 7 and 8,Fig. S10, and Table S1).In summary, using a systematic approach, we identified con-

served chemical signals from planaria, mouse, and human tissuesthat are antiinflammatory, proresolving, and tissue-regenerative.These new peptide–lipid conjugate molecules accelerate resolu-

tion of E. coli infection and stimulate phagocytic functions, tissueregeneration in planaria, and tissue repair in mice, thereby ful-filling criteria as immunoresolvents, namely agents that stimulateresolution of inflammation (12). The proposed biosynthesis ofthese mediators occurs via lipoxygenation of DHA, producing14-hydro(peroxy)-docosahexaenoic acid and an epoxide intermedi-ate (Fig. 9) that is converted to SC. This step in planaria relies ona GST enzyme(s), giving 13-glutathionyl, 14-hydroxy-docosahex-aenoic acid and 13-cysteinylglycinyl, 14-hydroxy-docosahexaenoicacid (Fig. 5), and opens the characterization of other peptide–lipidconjugates in the SPM genus. These specific peptide–lipid con-jugates carry a carbon-14–position alcohol and are biosynthesizedvia the maresin-epoxide intermediate; thus, they belong to themaresin (macrophage mediators in resolving inflammation) family(13). Given that these previously undescribed signals regulatethe cardinal signs of resolution, namely clearance of debris andinfections by phagocytes, tissue regeneration, and regulation ofproinflammatory chemical mediators, we coin these SC moleculesmaresin conjugates in tissue regeneration. Together, these find-ings provide new signals and pathways in host responses to injury,acute inflammation, and infectious exudates that are resolutionmoduli promoting homeostasis.

0 1 2 3 4 5 6

100

200

300

400

500

600Surgical Injury

+SCII

Days post surgery

*

* TRI50

T50

*

TRI50

T50

Surgical Injury

+SCI

0 1 2 3 4 5 6

100

200

300

400

500

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

Days post surgery

* * *

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

A B

+SCI+SCIISurgical Injury

***

Rel

ativ

e m

RN

A ex

pres

sion

Djsfpr-A Djndk DjmpkA DjAdb-Ba0

5

10

15

20

*** ***

*** ***

***

** ***

C

Fig. 4. SCI and SCII promote tissue regeneration and regulate key signalingpathways in planaria. After surgical injury, planaria were incubated with (A)SCI (100 nM), (B) SCII (100 nM), or vehicle (surgical injury; water containing0.01% EtOH), and regeneration indices were determined. Results representtwo independent experiments and are mean ± SEM for n = 9 planaria pergroup. (C) After surgical injury, planaria were incubated with SCI (100 nM),SCII (100 nM), or vehicle, and gene expression was assessed 2 d postinjury inregenerating head blastemas. Results are mean ± SEM for n = 3 per in-cubation pooled from blastemas of 9 animals. *P < 0.05, **P < 0.01, ***P <0.001 vs. surgical injury group.

+DjGstdsRNA

*0

100

200

300

400

500

MC

TR le

vels

(p

g/10

0mg

prot

ein)

Surgical Injury

Tiss

ue R

egen

erat

ion

Inde

x (m

m)

1 2 3 4 5 60

200

400

600 DjGst dsRNASurgical Injury

TRI50

Days post surgery

*

**

T50

A

Days post surgery

DjGst

0 1 2 3 4 5

1

2

3

4

5

Rel

ativ

e up

regu

latio

n

DjGst

DjGst dsRNA

WT

B

C D

Fig. 5. Regulation of tissue regeneration and SC production by planariaGST. (A) DjGst expression by WISH in uninjured animals (Left; represen-tative of n = 9) and time course for DjGst expression in regeneratingblastemas (Right; mean ± SEM for n = 4 per interval pooled from blastemasof 12 animals). (B–D) Planaria were fed homogenized beef liver containingDjGst dsRNA or beef liver (WT). (B and C) After 8 d, DjGst expression wasassessed by WISH (B) and tissue regeneration kinetics were determined (C ).(D) SC levels 3 d postsurgery. Results are mean ± SEM for n = 13 planariaper group. *P < 0.05, **P < 0.01 vs. surgical injury group. MCTR, maresinconjugates in tissue regeneration.

Dalli et al. PNAS | Published online October 16, 2014 | E4757

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Materials and MethodsMurine Peritonitis. Mouse experiments were conducted in accordance withguidelines from the Harvard Medical Area Standing Committee on Animals(protocol 02570). Mice were inoculated with E. coli (serotype 06:K2:H1) as inref. 11 and peritoneal exudates were collected by lavaging the peritonealcavity with 4 mL of PBS (without calcium and magnesium) at the indicatedintervals. Leukocytes were enumerated by light microscopy using nuclearmorphology staining with Turk’s solution and flow cytometry probing forCD11b-PerCP/Cy5.5 (eBioscience; clone M1/70), F4/80-PE (eBioscience; cloneBM8), and Ly6G-FITC (eBioscience; clone 1A8) as described (11). For productisolation, 24-h peritoneal exudates were collected and two volumes of ice-

cold methanol were added. Isolates were obtained as detailed in LipidMediator Metabololipidomics and Isolation of Bioactive Fractions.

In select experiments 12 h after E. coli inoculation, mice were administeredeither 100 ng of SCI and SCII (50 ng each per mouse) or vehicle [saline con-taining 0.1% ethanol (EtOH)] via i.p. injection. Peritoneal exudates were thencollected and leukocyte counts were determined as above. Assessment ofleukocyte phagocytosis of E. coli in inflammatory exudates was conducted asdescribed (11). Briefly, cells were incubated with anti–CD11b-PerCP/Cy5.5–conjugated antibody; subsequently, the cells were permeabilized using a BDCytofix/Cytoperm Fixation/Permeabilization Kit following the manufacturer’sinstructions (BD Biosciences); cells were then incubated with a FITC-conju-gated anti-E. coli antibody (GeneTex; GTX40856), and phagocytosis wasassessed as mean fluorescence in the CD11b-positive cell population.

Resolution indices were calculated as described (11), where Ψmax is themaximal neutrophil (PMN) number in the exudates; Tmax is the time pointwhen PMN numbers reach the maximum; R50 is 50% of the maximal PMNnumber; T50 is the time point when PMN numbers reduce to 50% of themaximum; and Ri = T50 − Tmax, the time period when 50% of PMNs are lostfrom the exudates.

Micewere administered either 105 (self-resolving) or 107 (delayed resolving)CFUs E. coli, and exudates were harvested after 12 or 24 h. These were thenincubated with either DHA (Cayman Chemical) or 14C-labeled DHA (AmericanRadiolabeled Chemicals; 1 μM, 37 °C, pH 7.45). The incubations were thenstopped with two volumes of ice-cold methanol and products were extractedusing solid phase extraction (SPE) columns as detailed below. Radioactivitywas measured using a scintillation counter, and DHA-derived products wereassessed using LC-MS-MS and product ion scan targeting m/z 343 (Fig. S2B).

In determined experiments, mice were administered SC isolates from regen-erating planaria i.p., obtained as described in LipidmediatorMetabololipidomicsand Isolation of Bioactive Fractions, immediately before zymosan adminis-tration (i.p., 1 mg per mouse). After 4 h, peritoneal exudates were obtained,leukocytes were enumerated as detailed above, and eicosanoid levels wereassessed by lipid mediator metabololipidomics as described below.

Ischemia Reperfusion. Mice were anesthetized by i.p. injection of xylazine(80 mg/kg) and ketamine (10 mg/kg). To initiate hind-limb ischemia, tourni-quets consisting of a rubber bandwere placed on each hind limb. Tenminutesbefore the initiation of reperfusion, vehicle (saline containing 0.1% EtOH),SCI (50 ng) plus SCII (50 ng), or Resolvin D1 (500 ng) were administered by i.v.injection. At the end of reperfusion (3 h), mice were euthanized, blood wascollected via cardiac puncture, and plasma was isolated for lipid mediatormetabololipidomics. Lungs were harvested; left lungs were frozen in liquidnitrogen and stored at −80 °C, and right lungs were stored in 10% (vol/vol)buffered formalin and processed for histology and hematoxylin and eosin(H&E) staining by the Children’s Hospital Boston, Department of Pathology,imaged using a Keyence BZ-9000 microscope and BZ II imaging software(Keyence). Expression of Ki67 and RSPO3 was assessed by immunofluores-cence staining. Briefly, sections were deparaffinized in xylene and rehy-drated and then blocked in 10% (vol/vol) horse serum and stained witheither rat anti-mouse RSPON3 antibody (R&D Systems) for 1 h at roomtemperature and then with sheep anti-rat Alexa 594-conjugated antibody(BioLegend) or with rabbit anti-mouse antibody (Abcam) and then withdonkey anti-rabbit Alexa 488 antibody (BioLegend). Slides were mounted inVECTASHIELD mounting solution with DAPI (Vector Laboratories) and im-munofluorescence was assessed using a Zeiss LSM 510 Meta confocal micro-scope. Images were processed using ImageJ software (National Institutes ofHealth) and Adobe Photoshop CS6 (Adobe Systems) software. Frozen lungs

0

1

2

3

4

5

0

50

100

150DHA

SC

SC

5 10 15 45 60Time (min)

SC

and SC

levels (pg/10

7 macrophages)

DH

A le

vels

(g/

107 m

acro

phag

es)

0

20

40

60

80 SC

E.coli +Acivicin

SC

0

50

100

150

pg/1

07 m

acro

phag

es

** ** **

**

**

5 1015 4560 5 1015 4560Time (min) Time (min)

0

10

20

40

60

* *

pg/1

00m

g pr

otei

n

+AcivicinSurgical injuryA B C

0

10

20

30SC SC

Fig. 6. Endogenous SCI is converted to SCII withhuman macrophages and planaria. (A) Humanmacrophages (3 × 107 cells) were incubated withDHA (37 °C, pH 7.45) and E. coli (1.5 × 108 CFUs),and product levels were assessed using LC-MS-MS.Results are mean for n = 3 separate incubations. (B)Human macrophages were incubated with or with-out γ-glutamyl transferase inhibitor (GTI, Acivicin;2.5 mM, 37 °C, pH 7.45, 30 min) and then DHA (37 °C,pH 7.45) and E. coli (1.5 × 108 CFUs), and precursorand product levels were assessed by LC-MS-MS.Results are mean ± SEM for n = 3 distinct incu-bations. (C ) Planaria were surgically injured andthen incubated with or without GTI (2.5 mM). SCI and SCII were assessed 3 d after injury using LC-MS-MS. Results are mean ± SEM for n = 20 planariaper group. *P < 0.05 vs. surgical injury group, **P < 0.01 vs. macrophages plus E. coli.

0

5

10

15

20

Leuk

ocyt

e E

.col

i pha

gocy

tosi

s (M

FI U

nits

x10

2 )

E.coli +SCI +SCII

*

A

C

-11-10-9 -8 -70

10

20

30

40

50

*

* **

*

**

* *

log [M]

SCII MaR1

log [M]

SCI MaR1

-11-10-9 -8 -70

10

20

30

40

50

* * *

**

*

*

*

max

Tmax T50

50Ri

Time post E.coli inoculation (h)

Per

itone

al P

MN

cou

nts

(x10

6 /mou

se)

** Ri ~20h

Ri ~10h

E.coli+SCI and SCII

0

1

2

3

4

5

12 24 36 48

SCI and SCII

B

Fig. 7. SCs resolve infection and stimulate efferocytosis. (A and B) Mice wereinoculated with E. coli (105 CFUs per mouse, i.p.) followed by either SCI plusSCII (50 ng per mouse each, i.p.) or vehicle (saline containing 0.1% EtOH) 12 hlater (A) Peritoneal leukocyte counts and resolution indices were determined(Materials and Methods). (B) In vivo E. coli phagocytosis. Results are mean ±SEM for n = 4 mice per interval. *P < 0.05, **P < 0.01 vs. E. coli. (C) Humanmacrophages (5 × 104 cells per well) were incubated with SCI (Left), SCII(Right), or MaR1, and fluorescently labeled apoptotic PMN and efferocytosiswere assessed. Results are mean ± SEM for n = 4 macrophage preparations.*P < 0.05, **P < 0.01 vs. vehicle group. MFI, mean fluorescence intensity.

E4758 | www.pnas.org/cgi/doi/10.1073/pnas.1415006111 Dalli et al.

were gently dispersed and centrifuged (2,000 × g), and tissue myeloperox-idase (MPO) levels were determined using mouse MPO ELISA (R&D Systems).

Leukocyte Phagocytosis. Macrophages were prepared from peripheral bloodmononuclear cells purchased from Children’s Hospital Blood Bank, and phago-cytosis was assessed as described (11). Briefly, macrophages (5 × 104 cells per well)were incubated with SCI (10 pM to 100 nM), SCII (10 pM to 100 nM), MaR1(10 pM to 100 nM), or vehicle [0.1% EtOH in Dulbecco’s phosphate bufferedsaline (DPBS)] for 15 min at 37 °C, and then fluorescently labeled apoptotic cellswere added and cells were incubated for 45 min at 37 °C. Extracellular fluo-rescence was quenched using Trypan blue (1:15 dilution) and phagocytosis wasassessed using an M3 SpectraMax plate reader (Molecular Devices). In selectexperiments, macrophages (5 × 104 cells per well) or neutrophils (1 × 105 cells perwell) were obtained from human healthy volunteers’ peripheral blood afterobtaining informed consent as described (11), in accordance with the PartnersHuman Research Committee Protocol (1999P001297), and incubated with 2′,7′-dichlorodihydrofluorescein (H2DCFDA) (5 μM, 30 min, 37 °C) and then with SCI(10 pM to 100 nM), SCII (10 pM to 100 nM), or vehicle (0.1% EtOH in DPBS,

15 min, 37 °C) and E. coli (1:50 leukocytes to E. coli, 45 min, 37 °C). Intracellularreactive oxygen species were determined by measuring fluorescence using anM3 SpectraMax plate reader. To assess bacterial phagocytosis, macrophages (5 ×104 cells per well) or neutrophils (1 × 105 cells per well) were incubated with SCI(10 pM to 100 nM), SCII (10 pM to 100 nM), or vehicle (0.1% EtOH in DPBS,15min, 37 °C) and then incubatedwith BacLight Green (Molecular Probes)-labeledE. coli (1:50 leukocytes to E. coli, 45 min, 37 °C). Extracellular fluorescence wasquenched using Trypan blue (1:15 dilution), and phagocytosis was assessedusing an M3 SpectraMax plate reader.

In determined experiments, human macrophages were incubated with SCisolates from regenerating planaria for 15 min at 37 °C, and then fluo-rescently labeled apoptotic cells were added and efferocytosis was assessedas detailed above.

Whole-Mount in Situ Hybridization, Real-Time PCR, and dsRNA Synthesis.Regenerating blastemas were obtained from planaria after surgical injuryand placed in TRIzol (Ambion). RNA was isolated following the manufacturer’sinstructions and cDNA was synthesized essentially as described (16) using

DAPI Ki67 RSPO3 Merge A Sham

R.I.+SCI+SCII

B Sham

R.I.

R.I.+SCI+SCII

R.I. Fig. 8. SCs are organ-protective. Mice were sub-jected to hind-limb ischemia (1 h) followed byreperfusion (3 h). Ten minutes before reperfusion,vehicle (saline containing 0.1% EtOH; R.I., reflowinjury), SCI plus SCII (50 ng each) were administeredintravenously. Lungs were then collected. (A) TissueH&E staining. (Scale bars, 100 μm.) Black arrowsindicate leukocyte-mediated tissue damage; bluearrows indicate intact alveolar regions. (B) Immuno-fluorescence staining. Nuclear material (DAPI,blue), Ki67 (Alexa 488, green), RSPO3 (Alexa 594,red). (Scale bars, 100 μm.) Results are representa-tive of n = 4 mice per group.

COOH

OH

SNH

NH2O

HOOC

COOH

O

COOH

OOH

COOH

OHOH

DHA

LOX

LOX

13,14-epoxy-Maresin

MaR1

COOH

OH

S

HN NH

O

OHOOC

NH2HOOC

Reduces neutrophil recruitment Stimulates efferocytosis Promotes tissue regeneration

Promote resolution of infections Stimulate bacterial phagocytosis and killing Promote tissue regeneration Protect organs by limiting neutrophil recruitment and stimulating repair

Human milk Human macrophages Mouse infectious resolving exudates Regenerating planaria

SCI

SCII

Fig. 9. Proposed SC biosynthetic scheme. Struc-tures are depicted in likely conformations based onbiosynthetic evidence (see main text and Table S1for further details). The stereochemistry of Maresin1 and the maresin-epoxide intermediate are estab-lished (21).

Dalli et al. PNAS | Published online October 16, 2014 | E4759

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

random hexamers, Oligo(dT)20, and SuperScript III (Invitrogen). Relative quanti-tative analysis for each gene product was carried out using an ABI 7900HT FastReal-Time PCR machine (Applied Biosystems). DjGapdh: forward, 5′-ACCAC-CAACTGTTTAGCTCCCTTA-3′; reverse, 5′-GATGGTCCATCAACAGTCTTTTGC-3′.Djsfpr-A: forward, 5′-TTGCTCTCTTTACGCTCCGGT-3′; reverse 5′-CGCATAGTTCC-CTGCATGGT-3′. Djndk: forward, 5′-TCACAAACTCCACCGCAGTACTTT-3′; reverse,5′-GGTATGGATTAGCATTATTGAATTGTG-3′. DjmpkA: forward, 5′-CACTGATAT-CTACTTCACGAAAGCCAG-3′; reverse, 5′-AAGGCATCCAGTTCATTTCCTAAAT-3′.DjAdb-Ba: forward, 5′-CGTAGGCAATACTTACATCACTAGACAAA-3′; reverse, 5′-TGTCTCTCCGACAAATGCAATTT-3′. DjGst: forward, 5′-TGTTGGCTGAAGA-AGTGCAAG-3′; reverse, 5′-TTCACCCATAAGCCAATGCT-3′. The constitutivelytranscribed housekeeping gene GAPDH was used to normalize target geneexpression levels as described (16).

Two separate PCR products containing T7 promoter sequences on the 5′ends of either sense or antisense strand DNA were used as templates totranscribe antisense and sense RNA using the T7 RiboMAX Express RNAiSystem (Promega) (DjGst with T7 promoter sequence: forward, 5′-GGATCC-TAATACGACTCACTATAGGAGTGGCAATCACCACCAAAT-3′; reverse, 5′-GGA-TCCTAATACGACTCACTATAGGAGTGGCAATCACCACCAAAT-3′. DjGst primers:forward, 5′-GGCTCCTTTGTTAGGTTACTGGA-3′; reverse, 5′-AGTGGCAATCAC-CACCAAAT-3′). These two complementary single-stranded RNA (ssRNA) weremixed and incubated at 70 °C (10 min) and then slowly cooled to roomtemperature (∼20 min) to allow annealing of the dsRNA. The remainingtemplate DNA and ssRNA were removed by treatment with DNase and RNaseA (37 °C, 30 min) essentially as described (27). The dsRNA was then pre-cipitated (0.1 volume of 3 M sodium acetate, pH 5.2, and 1 volume of iso-propanol) and further purified using MicroSpin G-25 Columns (GE Healthcare)to remove residual nucleotides.

PCR products containing T7 promoter sequences on the 5′ end of the sensestrand DNA were used to produce in situ hybridization probes to DjGst usingFISH Tag Kits (Invitrogen) following the manufacturer’s instructions. Planariawere treated with 2% hydrochloric acid in 5/8 Holtfreter’s solution for 5 minat 4 °C and fixed in 5/8 Holtfreter’s solution containing 4% paraformaldehydeand 5% methanol for less than 2 h at 4 °C. Planaria were dehydrated andrehydrated and then bleached, and probes were hybridized as described (28).

Planaria Regeneration. Planaria (D. japonica; Dj) were kept in water (PolandSpring) at 18 °C. All animals were starved for at least 7 d before theexperiments. Tissue regeneration was assessed as described (12). Briefly,planaria were subjected to head resection postoccularly (surgical injury). Theposterior portions of the planaria were then placed in spring water con-taining 0.01% EtOH, SPE-C isolate fraction 2 from resolving exudates or milkat the indicated dilutions, U0126 [ERK inhibitor; Cell Signaling Technology;25 μM (16)], U0126 plus resolving exudates SPE-C isolate fraction 2, SCI(100 nM), SCII (100 nM), lipoxygenase inhibitor (baicalein; 10 μM), or lipox-ygenase inhibitor plus SCI and SCII (100 nM). The extent of tissue re-generation during a 6-d period for D. japonica was determined usingcaptured images of the regenerating blastemas at regular intervals (24 h).These images were analyzed using ImageJ software. A tissue regenerationindex (TRI) was used that took into consideration the size of the regeneratedtissue total area (A) and the postocular width (W) of the animal, where TRI =A/W. In determined experiments, planaria were injured and incubatedin water containing 108 CFUs E. coli (serotype O6:K2:H1) or with E. coli plusSCI and SCII (100 nM), and tissue regeneration indices were assessed asdescribed above.

In select experiments, planaria were fed homogenized beef liver or beefliver containing dsRNA for DjGst as described (27). After 8 d, planaria wereeither taken for whole-mount in situ hybridization or subjected to head re-section and tissue regeneration assessed every day for 7 d. SCs were identifiedand quantified 3 d postinjury using lipid mediator metabololipidomics.

Planaria were surgically injured and incubated with or without Acivicin(2.5 mM). After 3 d, products were extracted by solid-phase extraction (seebelow), and SCI and SCII levels were investigated by LC-MS-MS.

Lipid Mediator Metabololipidomics and Isolation of Bioactive Fractions. Peri-toneal exudates and exudate cell incubations were immediately placedin two volumes of methanol. For lipid mediator profiling, 500 pg each ofdeuterium-labeled internal standards d8-5S-HETE, d4-LTB4, d5-LXA4, d4-PGE2,and d5-LTC4 was added to facilitate quantification in each respective chro-matographic region and sample recovery. Samples were then held at −20 °Cfor 45 min to allow for protein precipitation and centrifuged (1,200 × g, 4 °C,10 min). Products were then extracted using solid-phase extraction as de-scribed (29) and eluted using methyl formate (SPE-C isolate fraction 1) andmethanol (SPE-C isolate fraction 2). Eluted isolates were then brought todryness under nitrogen and suspended in methanol/water (50:50) for lipid

mediator metabololipidomics or ethanol for biological evaluation. For lipidmediator metabololipidomics of known mediators and pathway products,the LC-MS-MS system was operated as described (29). For identification andquantification of SCs, a Shimadzu LC-20AD HPLC and a Shimadzu SIL-20ACautoinjector paired with a QTrap 6500 (AB SCIEX) were used. An EclipsePlus C18 column (50 mm × 4.6 mm × 1.8 μm; Agilent) was kept in a column ovenmaintained at 50 °C (ThermaSphere; TS-130), and lipid mediators were elutedwith a mobile phase consisting of methanol/water/acetic acid at 55:45:0.01 (vol:vol:vol) that was ramped to 85:15:0.01 (vol:vol:vol) over 0.1 min, to 86:14:0.01(vol:vol:vol) for the next 3 min, to 90:10:0.01 (vol:vol:vol) for the next 1 min,and to 99.9:0:0.01 (vol:vol:vol) for the next 6 min. This was subsequentlymaintained at 99.9:0:0.01 (vol:vol:vol) for 2 min, and the flow rate wasmaintained at 0.65 mL/min. The QTrap 6500 was operated in positive ioni-zation mode using scheduled multiple reaction monitoring (MRM) coupledwith information-dependent acquisition and enhanced product ion scan.

Human milk (Biological Specialty) was placed in 2.5 volumes of ice-cold meth-anol; proteinswere thenallowed to precipitate for 30min at 4 °C and supernatantswere collected. These were then acidified to pH ∼3.5 and extracted using diethylether. Samples were then brought to dryness under vacuum on a rotary evapo-rator (Buchi); products were suspended in 1 mL of methanol and extracted asdetailed above using C18 SPE columns to obtain SPE-C isolate fractions.

Macrophages (1 × 107 cells in 175-cm2 flasks) were transfected withALOX12 human shRNA in pRS vector (20 μg; OriGene) or mock vector (pRSalone) using jetPEI transfection reagent (40 μL; following the manufacturer’sinstructions; Polyplus Transfection). Seventy-two hours later, 12-LOX ex-pression was assessed using immunofluorescence staining with human12-LOX antibody (Novus Biologicals) or relevant isotype control, and expres-sion was determined by flow cytometry. Transfected cells (1 × 107 per mL)were also incubated with E. coli (5 × 108 CFUs per mL, RPMI supplementedwith 0.1% human serum, 37 °C) for 1 h. These incubations were stoppedwith two volumes of ice-cold methanol, and isolates (SPE-C isolate fractions1 and 2) were obtained as described above.

Planaria (∼200 animals) were surgically injured as described above. After3 d, they were placed in two volumes of ice-cold methanol and tissues weregently dispersed using a glass dounce. Homogenates were placed at −20 °Cto allow for protein precipitation, and SPE-C isolate fraction 2 was obtainedas described above.

SC Biosynthesis, Isolation, and Derivatives. Human macrophages (1 × 107 cellsper mL) were suspended in PBS+/+ incubated with 14-HpDHA (1 μM), and E.coli (1 × 108 CFUs per mL, 37 °C, pH 7.45, 30 min). The 14-HpDHA was pro-duced by incubation of DHA with human macrophage 12-lipoxygenase, andisolated as described (13). Two volumes of methanol were then added andproducts were extracted using C18 columns as outlined above.

In select experiments, human macrophages (1 × 107 cells per mL) weresuspended in PBS+/+ incubated with d5-14-HpDHA (1 μM) and E. coli (1 × 108

CFUs per mL, 37 °C, pH 7.45, 30 min). Two volumes of methanol were addedand products were extracted using SPE columns as outlined above.

SCs obtained as detailed above were isolated using online UV-RP-HPLC(1100 Series; Agilent Technologies) and an Agilent Poroshell 120 C18 column(100 mm × 4.6 mm × 2.7 μm) with the mobile phase consisting of methanol/water at 55:45 (vol:vol) that was ramped to 85:15 (vol:vol) over 0.1 min, to86:14 (vol:vol) for the next 3 min, to 90:10 (vol:vol) for the next 1 min, and to100:0 (vol:vol) for the next 6 min. This was subsequently maintained at100:0 (vol:vol) for 2 min, and the flow rate was maintained at 0.65 mL/min.In select experiments, isolated SCI and SCII were incubated with diazo-methane in diethyl ether for 30 min at room temperature. Samples werethen brought to dryness and products were assessed by LC-MS-MS usingMRM monitoring of the following ion pairs: 549>193 and 692>336. SCI andSCII were incubated with activated Raney nickel catalyst for 20 min at roomtemperature. The resulting products were then assessed by LC-MS-MS usingMRM monitoring: 343>205 (Figs. S3 and S5).

Human macrophages (3 × 107 cells per mL) were incubated with DHA(15 μg, 37 °C, PBS, pH 7.45) and E. coli (1.5 × 108 CFUs). Incubations werestopped with two volumes of ice-cold methanol, products were extracted,and levels were assessed by LC-MS-MS. In select experiments, macrophageswere incubated with Acivicin (2.5 mM, 37 °C, PBS, pH 7.45) before addition ofDHA (15 μg) and Acivicin (2.5 mM) and products were assessed by LC-MS-MS.

Statistics. All results are expressed as means ± SEM. Differences betweengroups were compared using Student t test (two groups), one-way ANOVA(multiple groups) followed by post hoc Bonferroni test, or two-way ANOVA(multiple groups, multiple time points) followed by post hoc Bonferroni orSidak tests. The criterion for statistical significance was P < 0.05.

E4760 | www.pnas.org/cgi/doi/10.1073/pnas.1415006111 Dalli et al.

ACKNOWLEDGMENTS. The authors thank Mary Halm Small for expertassistance in manuscript preparation, Dr. Michael Levin and Dr. JunjiMokmura (Tufts University) for providing D. japonica seed colonies and for

helpful discussions, and Dr. Romain Colas for assistance with material prep-aration. This work was supported in part by the National Institutes of Health(Grant P01GM095467).

1. Nathan C (2012) Fresh approaches to anti-infective therapies. Sci Transl Med 4(140):140sr2.

2. Ward PA (2012) New approaches to the study of sepsis. EMBO Mol Med 4(12):1234–1243.

3. Serhan CN, Savill J (2005) Resolution of inflammation: The beginning programs theend. Nat Immunol 6(12):1191–1197.

4. Bannenberg GL, Aliberti J, Hong S, Sher A, Serhan C (2004) Exogenous pathogen andplant 15-lipoxygenase initiate endogenous lipoxin A4 biosynthesis. J Exp Med 199(4):515–523.

5. Nakamura M, Shimizu T (2011) Leukotriene receptors. Chem Rev 111(10):6231–6298.6. Samuelsson B (2012) Role of basic science in the development of new medicines:

Examples from the eicosanoid field. J Biol Chem 287(13):10070–10080.7. Fullerton JN, O’Brien AJ, Gilroy DW (2014) Lipid mediators in immune dysfunction

after severe inflammation. Trends Immunol 35(1):12–21.8. Tabas I, Glass CK (2013) Anti-inflammatory therapy in chronic disease: Challenges and

opportunities. Science 339(6116):166–172.9. Serhan CN (2014) Pro-resolving lipid mediators are leads for resolution physiology.

Nature 510(7503):92–101.10. Haeggström JZ, Funk CD (2011) Lipoxygenase and leukotriene pathways: Bio-

chemistry, biology, and roles in disease. Chem Rev 111(10):5866–5898.11. Chiang N, et al. (2012) Infection regulates pro-resolving mediators that lower antibiotic

requirements. Nature 484(7395):524–528.12. Serhan CN, et al. (2012) Macrophage proresolving mediator maresin 1 stimulates

tissue regeneration and controls pain. FASEB J 26(4):1755–1765.13. Serhan CN, et al. (2009) Maresins: Novel macrophage mediators with potent antiin-

flammatory and proresolving actions. J Exp Med 206(1):15–23.14. Sánchez Alvarado A (2006) Planarian regeneration: Its end is its beginning. Cell

124(2):241–245.15. Calder PC, et al. (2006) Early nutrition and immunity— progress and perspectives. Br J

Nutr 96(4):774–790.

16. Umesono Y, et al. (2013) The molecular logic for planarian regeneration along theanterior-posterior axis. Nature 500(7460):73–76.

17. Gordon S, Mantovani A (2011) Diversity and plasticity of mononuclear phagocytes.Eur J Immunol 41(9):2470–2472.

18. Stables MJ, et al. (2011) Transcriptomic analyses of murine resolution-phase macro-phages. Blood 118(26):e192–e208.

19. Deng B, et al. (2014) Maresin biosynthesis and identification of maresin 2, a new anti-inflammatory and pro-resolving mediator from human macrophages. PLoS ONE 9(7):e102362.

20. Dalli J, et al. (2013) The novel 13S,14S-epoxy-maresin is converted by human macro-phages to maresin 1 (MaR1), inhibits leukotriene A4 hydrolase (LTA4H), and shiftsmacrophage phenotype. FASEB J 27(7):2573–2583.

21. Samuelsson B (1983) Leukotrienes: Mediators of immediate hypersensitivity reactionsand inflammation. Science 220(4597):568–575.

22. Martinez Molina D, et al. (2007) Structural basis for synthesis of inflammatory me-diators by human leukotriene C4 synthase. Nature 448(7153):613–616.

23. Orning L, Bernstrom K, Hammarstrom S (1981) Formation of leukotrienes E3, E4 andE5 in rat basophilic leukemia cells. Eur J Biochem 120(1):41–45.

24. Majno G, Joris I (2004) Cells, Tissues, and Disease: Principles of General Pathology(Oxford Univ Press, New York), 2nd Ed.

25. Pipparelli A, et al. (2013) ROCK inhibitor enhances adhesion and wound healing ofhuman corneal endothelial cells. PLoS ONE 8(4):e62095.

26. Sato T, Clevers H (2013) Growing self-organizing mini-guts from a single intestinalstem cell: Mechanism and applications. Science 340(6137):1190–1194.

27. Rouhana L, et al. (2013) RNA interference by feeding in vitro-synthesized double-stranded RNA to planarians: Methodology and dynamics. Dev Dyn 242(6):718–730.

28. King RS, Newmark PA (2013) In situ hybridization protocol for enhanced detection ofgene expression in the planarian Schmidtea mediterranea. BMC Dev Biol 13:8.

29. Dalli J, Serhan CN (2012) Specific lipid mediator signatures of human phagocytes:Microparticles stimulate macrophage efferocytosis and pro-resolving mediators.Blood 120(15):e60–e72.

Dalli et al. PNAS | Published online October 16, 2014 | E4761

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US