Neurotensin stimulates sortilin and mTOR in human …Neurotensin stimulates sortilin and mTOR in human microglia inhibitable by methoxyluteolin, a potential therapeutic target for

Neurotensin stimulates sortilin and mTOR in humanmicroglia inhibitable by methoxyluteolin, a potentialtherapeutic target for autismArti B. Patela,b, Irene Tsilionia, Susan E. Leemanc,1, and Theoharis C. Theoharidesa,b,d,e,1

aMolecular Immunopharmacology and Drug Discovery Laboratory, Department of Integrative Physiology and Pathobiology, Tufts University School ofMedicine, Boston, MA 02111; bGraduate Program in Cell, Molecular, and Developmental Biology, Sackler School of Graduate Biomedical Sciences, TuftsUniversity, Boston, MA 02111; cDepartment of Pharmacology, Boston University School of Medicine, Boston, MA 02118; dDepartment of Internal Medicine,Tufts University School of Medicine and Tufts Medical Center, Boston, MA 02111; and eDepartment of Psychiatry, Tufts University School of Medicine andTufts Medical Center, Boston, MA 02111

Contributed by Susan E. Leeman, August 9, 2016 (sent for review April 15, 2016; reviewed by Paul R. Dobner, Paul M. Hardy, and Stephen Skaper)

We had reported elevated serum levels of the peptide neurotensin(NT) in children with autism spectrum disorders (ASD). Here, we showthat NT stimulates primary humanmicroglia, the resident immune cellsof the brain, and the immortalized cell line of human microglia-SV40.NT (10 nM) increases the gene expression and release (P < 0.001) ofthe proinflammatory cytokine IL-1β and chemokine (C-X-C motif) li-gand 8 (CXCL8), chemokine (C-C motif) ligand 2 (CCL2), and CCL5 fromhuman microglia. NT also stimulates proliferation (P < 0.05) of micro-glia-SV40. Microglia express only the receptor 3 (NTR3)/sortilin and notthe NTR1 or NTR2. The use of siRNA to target sortilin reduces (P <0.001) the NT-stimulated cytokine and chemokine gene expressionand release from human microglia. Stimulation with NT (10 nM) in-creases the gene expression of sortilin (P < 0.0001) and causes thereceptor to be translocated from the cytoplasm to the cell surface,and to be secreted extracellularly. Our findings also show increasedlevels of sortilin (P< 0.0001) in the serum from childrenwith ASD (n=36), compared with healthy controls (n= 20). NT stimulation of micro-glia-SV40 causes activation of the mammalian target of rapamycin(mTOR) signaling kinase, as shown by phosphorylation of its sub-strates and inhibition of these responses by drugs that prevent mTORactivation. NT-stimulated responses are inhibited by the flavonoidmethoxyluteolin (0.1–1 μM). The data provide a link between sortilinand the pathological findings of microglia and inflammation of thebrain in ASD. Thus, inhibition of this pathway using methoxyluteolincould provide an effective treatment of ASD.

autism spectrum disorders | human microglia | methoxyluteolin | mTOR |sortilin

Autism spectrum disorders (ASD) are neurodevelopmentaldisorders (1, 2). The prevalence of ASD is now estimated to be

1 in 45 children (3). Unfortunately, there is still no distinct patho-genesis (4) even though a number of neuropathological defects havebeen reported in the brains of children with infantile autism (5).Microglia, the highly plastic resident immune cells of the brain (6,7), have been shown to be activated in the brains of patients withASD (8–11). Microglia activation and proliferation could lead tofocal inflammation of the brain and “choking” of normal synapticconnectivity (12, 13). Microglia express membrane receptors forseveral neuropeptides, allowing them to communicate with neurons,astrocytes (14), and mast cells (15), known to be involved in allergicand inflammatory processes (16).Various stimuli, such as the bacterial lipopolysaccharide (LPS)

(14, 17), have been shown to switch microglia into theM1 phenotype,denoted by the release of proinflammatory cytokines, interleukin(IL)-1β, IL-6, and tumor necrosis factor (TNF) (18), as well as thechemokines (C-C motif) ligand 2 (CCL2) and CCL5 (8, 19), alsofound to be increased in brains of deceased patients with ASD.Immune dysfunction (18, 20–22) and inflammation of the brain (23–25) are now invoked in the pathogenesis of ASD. However, the

stimuli that promote these inflammatory processes in the brain arepresently unknown.Our laboratory had reported increased serum levels of the

peptide neurotensin (NT), but not substance P or β-endorphin(26), in children with ASD (26, 27). NT is found in the brain (28,29) and is primarily secreted from neurons (29) and astrocytes(30). NT responses are mediated through three receptors: NTR1(31) and NTR2 (32, 33), which belong to the G protein-coupledseven-transmembrane receptor family (34), and NTR3, also knownas sortilin (35). NTR3/sortilin is a type I sorting protein [part of theVps10p domain single-transmembrane receptor family (31)], amultifaceted receptor mainly expressed in the CNS during em-bryonic development (36).NTR3/sortilin has been shown to be expressed in murine

microglia through which NT stimulates IL-1β, CCL2, and TNFgene expression (37). However, rodent microglia have majorbiochemical and pharmacological differences compared withprimary human microglia (38). Moreover, animal models do notreflect human inflammatory processes (39). A subset (1–5%) ofASD cases has gene mutations in regulatory proteins upstreamof the signaling complexes termed the “mammalian target of

Significance

Human microglia, the resident immune cells of the brain, expressonly the neurotensin (NT) receptor-3/sortilin. NT significantlyincreases microglia synthesis and release of proinflammatorycytokine IL-1β and chemokine (C-X-C motif) ligand 8 (CXCL8),chemokine (C-C motif) ligand 2 (CCL2), and CCL5 via NTR3/sortilin.A soluble form of this receptor is secreted from stimulatedmicroglia and is increased in the serum of children with autismspectrum disorders (ASD). These responses and the NT-stimulatedincreases in microglia numbers are mediated via mammalian tar-get of rapamycin (mTOR) activation and are inhibitable by thenatural flavonoids luteolin and methoxyluteolin.

Author contributions: A.B.P. and T.C.T. designed research; A.B.P. and I.T. performed re-search; A.B.P. contributed new reagents/analytic tools; A.B.P., I.T., S.E.L., and T.C.T. ana-lyzed data; and A.B.P., S.E.L., and T.C.T. wrote the paper.

Reviewers: P.R.D., University of Massachusetts Medical School; P.M.H., Autism ResearchInstitute; and S.S., University of Padua.

Conflict of interest statement: T.C.T. has been awarded, under an agreement with TuftsUniversity, the following patents: US 9,050,275 - Methods of Screening for and TreatingAutism Spectrum Disorders and Compositions for Same and US 9,176,146 - Methods ofTreating Autism Spectrum Disorders and Compositions for Same. A Provisional PatentApplication was also filed by Tufts University as US 62/396,546 - Compositions and Methodsof Autism Treatment. T.C.T. is also the Scientific Director of Algonot, LLC (Sarasota, FL) that hasdeveloped the flavonoid-containing trademarked dietary supplement NeuroProtek, for whichTufts receives royalties. No portion of the work described in this paper was funded by Algonot.1To whom correspondence may be addressed. Email: [email protected] [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1604992113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1604992113 PNAS | Published online September 23, 2016 | E7049–E7058

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0 D

ownl

oade

d by

gue

st o

n M

arch

1, 2

020

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

rapamycin” (mTOR) (4, 40). These mutations in mice lead to abehavioral phenotype resembling autism (41), and targeting themTOR pathway has been shown to reverse autism-like behavior(42, 43). The phosphoinositide 3-kinase (PI3K)/AKT/mTORsignaling pathway also regulates the activation of both microglia(44) and mast cells (45), which may cross-talk to exacerbateinflammation of the brain (15).There are no clinically available drugs addressing the core symp-

toms of ASD. The natural flavonoid luteolin has potent antioxidantand antiinflammatory properties (46). It also inhibits activation ofmicroglia (17, 47–49). Luteolin also reverses autism-like behaviorin mice (50). Two clinical studies further reported that a luteolin-containing dietary formulation significantly improved attentionand sociability in children with ASD (51, 52). Its structural analogmethoxyluteolin (3′,4′,5,7-tetramethoxyflavone) is a more potentmast cell inhibitor (53) and is more metabolically stable (54);hence, it is more likely to reach therapeutic levels in the brain.Here, we extend our previous finding of elevated serum NT

levels in children with ASD by investigating whether NT stimu-lates the activation of human microglia, the specific NT receptorinvolved, and pathways that can be targeted for inhibition of theseprocesses by methoxyluteolin.

ResultsNT Induces Expression of Proinflammatory Cytokines and Chemokinesin Human Microglia. To evaluate whether NT can switch humanmicroglia to the proinflammatory M1 phenotype, we first used ahuman cytokine and chemokine array blot to identify any cy-tokines and chemokines differentially expressed in control andNT-stimulated human microglia-SV40. NT increases expressionof numerous proinflammatory cytokines and chemokines (Fig.S1), which were then measured by specific ELISAs. NT stimu-lation at physiological doses (10 or 100 nM) for 24 h increasessecretion of the proinflammatory cytokine IL-1β and chemokine(C-X-C) motif ligand 8 (CXCL8), CCL2, and CCL5 (P < 0.001),compared with controls, from both primary human microgliaand microglia-SV40 (Fig. 1 A–D). LPS (10 or 100 ng/mL) usedas a positive control also increases (P < 0.0001) the secretion ofall mediators, including IL-6 and TNF, from both human microgliacell types after 24-h stimulation (Fig. S2). NT stimulation (10 or100 nM) significantly increases (P < 0.0001) the gene expression ofIL-1β, CXCL8, CCL2, and CCL5 after 12 h in both primary human

microglia and microglia-SV40 (Fig. 2 A–D). LPS (10 or 100 ng/mL;P < 0.0001) also increases the synthesis of these mediators fromhuman microglia after 12-h stimulation.

Human Microglia Express Only NTR3/Sortilin, Which Increases in Geneand Surface Protein Localization in Response to NT. We next in-vestigated the expression of the three types of NT receptors,NTR1, NTR2, and NTR3/sortilin, in both primary human microgliaand the human microglia-SV40 cell line. The high-affinity NTR1 andlow-affinity NTR2 are undetectable; however, NTR3/sortilin geneexpression is detectable, and significantly increases (P < 0.001) after12 or 24 h of NT treatment (10 or 100 nM) (Fig. 3A). Protein levelsof NTR1 and NTR2 are also undetectable by Western blot analysis.To determine changes in the cellular localization of NTR3/

sortilin after NT stimulation, differential interference contrast(DIC)/confocal immunofluorescence microscopy was carriedout. Immunodetectable NTR3/sortilin in control primary humanmicroglia and microglia-SV40 reveals a cytosolic distribution.Stimulation with NT (10 nM) for 24 h increases the surface-as-sociated NTR3/sortilin in primary human microglia (Fig. 3B) andin the microglia-SV40, which appears to be colocalized with thefilamentous actin-binding protein ionized calcium-binding adaptormolecule-1 (Iba-1) (Fig. 3C). We hypothesized that surfaceNTR3/sortilin may also be secreted extracellularly. Stimulation ofhuman microglia-SV40 with NT (10 nM) for 24 h (P < 0.0001) or48 h (P < 0.05) increases levels of soluble NTR3/sortilin (Fig. 3D).There is no apparent significant difference in the total cellularNTR3/sortilin levels in human microglia stimulated with NT (10 or100 nM) after 24 or 48 h, compared with control cells (Fig. 3E).

NT Stimulates Proinflammatory Cytokine and Chemokine Releasefrom Human Microglia via NTR3/Sortilin. Even though humanmicroglia do not express NTR1 and NTR2, we pretreatedmicroglia-SV40 with the NTR1 selective nonpeptide antagonistSR48692 or the dual NTR1/NTR2 antagonist SR142948A (10–1,000 nM for 1 h), before stimulation with NT (10 or 100 nM).Pretreatment with these receptor antagonists did not affect IL-1β, CXCL8, CCL2, and CCL5 mediator release from humanmicroglia-SV40. To determine whether NTR3/sortilin mediatesthe NT-stimulated proinflammatory cytokine and chemokine

Fig. 1. Proinflammatory mediator release from human microglia. Primaryhuman microglia (HM) (5 × 104 cells) and immortalized HM-SV40 (5 × 104 cells)were stimulated with NT (10 or 100 nM) for 24 h to measure release of IL-1β (A),CXCL8 (B), CCL2 (C), and CCL5 (D) by specific ELISAs. All conditions wereperformed in triplicate for each dataset and were repeated three times (n = 3).Significance of comparisons is denoted by *P < 0.05, **P < 0.001, or ***P <0.0001. Conc., concentration.

Fig. 2. Proinflammatory mediator gene expression in HM. Primary HM(2.5 × 105 cells) and immortalized HM-SV40 (2.5 × 105 cells) were stimulatedwith NT (10 or 100 nM) for 12 or 24 h to determine changes in gene ex-pression levels of IL-1β (A), CXCL8 (B), CCL2 (C), and CCL5 (D) by qRT-PCR. Allconditions were performed in triplicate for each dataset and were repeated threetimes (n = 3). Results were normalized against the endogenous gene, GAPDH, andare expressed relative to the mean of the gene of interest. Significance of com-parisons is denoted by *P < 0.05, **P < 0.001, or ***P < 0.0001.

E7050 | www.pnas.org/cgi/doi/10.1073/pnas.1604992113 Patel et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

release, human microglia were subjected to siRNA down-regulationof NTR3/sortilin levels. Human microglia-SV40 were transfectedwith two different scrambled and targeted NTR3/sortilin siRNAs.Gene expression analysis by quantitative real-time (qRT)-PCR

revealed >95% knockdown of NTR3/sortilin in siRNA-trans-fected cells after 48 h, compared with control siRNA-transfectedor unstimulated microglia (Fig. 4A). Protein levels of NTR3/sortilin in microglia-SV40 transfected with NTR3 siRNA areabolished, whereas the control cells treated with scrambledsiRNA retain normal expression levels after 48 or 72 h (Fig. 4B).Stimulation by NT of microglia in which NTR3/sortilin levelswere down-regulated significantly decreases (P < 0.001) IL-1β,CXCL8, CCL2, and CCL5 release, compared with control cells(Fig. 4 C–F).To ensure that the effect of down-regulated NTR3/sortilin was

not due to any involvement in intracellular mediator transport orrelease, we investigated the level of proinflammatory mediatorgene expression in NT-stimulated scrambled and targeted NTR3/sortilin siRNA-treated microglia SV40. Gene expression of IL-1β,CXCL8, CCL2, and CCL5 significantly decreases (P < 0.001) afterNT stimulation in microglia with down-regulated NTR/sortilinlevels, compared with scrambled siRNA-treated or unstimulatedmicroglia (Fig. S3).

NT Activates PI3K/mTOR Signaling in Human Microglia That Is Blockedby Luteolin and Methoxyluteolin. To investigate the signalingpathway involved in the stimulation of human microglia-SV40 inresponse to NT, a phosphoarray blot to detect the phosphory-lated (p) signaling proteins was used. Protein levels of phos-phorylated substrates that are up-regulated in microglia-SV40

after NT stimulation include the downstream mTOR substrates,p70 ribosomal 6 kinase (S6K) and eukaryotic initiation factor4E-binding protein 1 (4EBP1) proteins (Fig. S4A). Western blotanalysis was then performed to detect the total and p-levels ofmTOR, as well as p70S6K and 4EBP1 proteins after stimulationby NT (10 nM) from 0 to 60 min. NT increases the levels ofpmTOR Ser2448 and the downstream mTORC1 substrate pp70S6KThr389 within 30 min (Fig. 5A).We next investigated whether the natural flavonoids luteolin

and methoxyluteolin affect mTOR signaling by comparing theirinhibitory effect with the inhibitory effect of various mTOR inhib-itors in microglia stimulated by NT. We used the first-generationallosteric mTOR inhibitor, rapamycin; the small-molecule ATP-competitive kinase inhibitor of mTORC1 and mTORC2, KU-0063794 (KU); and the dual PI3K/mTOR inhibitor, PF-04691502(PF). Human microglia-SV40 were serum-starved overnight andpreincubated with rapamycin, 0.5 μM KU/PF, or luteolin andmethoxyluteolin (0.1–10 μM) for 12 h before NT stimulation(30 min), which significantly decreases levels of pmTORSer2448and p70S6KThr389 compared with those levels in microgliastimulated by NT alone (Fig. 5 B–D). NT did not increase thelevels of total or p4EBP1 proteins in stimulated microglia-SV40;however, the levels p4EBP1 in unstimulated microglia decreasein the presence of mTOR inhibitors, whereas the flavonoids hadno effect on these inhibitors (Fig. 5E).Phospho-ELISAs were also performed on microglia stimu-

lated with NT and/or pretreated with the PI3K/mTOR inhibitorsor luteolin and methoxyluteolin to quantify levels of pAKTSer473, pmTORSer2448, and pp70S6KThr389 to assess activa-tion of mTOR. Levels of pAKT or pmTOR or p70S6K proteins

Fig. 3. NTR3/sortilin gene expression and cellular localization in HM, and its release in response to NT. (A) Primary HM (2.5 × 105 cells) and immortalizedHM-SV40 were stimulated with NT for 12 or 24 h to measure gene expression levels of NTR3/sortilin in control and NT-stimulated (10 or 100 nM) microgliaby qRT-PCR. All conditions were performed in triplicate for each dataset and were repeated three times (n = 3). Results were normalized against theendogenous GAPDH and expressed relative to the mean of the control for the gene of interest, with significance of comparisons denoted by *P < 0.05,**P < 0.001, or ***P < 0.0001. Microglia-SV40 (5 × 103 cells per four-well chamber–coated slide) were stimulated with NT (10 nM) for 24 h, and then fixedand permeabilized to stain for nuclei using DAPI (blue), with specific antibodies for Alexa 488-NTR3/sortilin (green) or Alexa 594–Iba-1, a microglialmarker protein (red), whereas rabbit IgG was used for the negative control. The cell surface and cytosolic distribution of NTR3/sortilin protein is shown incontrol and NT-stimulated primary HM (B) and immortalized HM-SV40 (C ), where colocalization with Iba-1 is also apparent (white arrows). Qualitativeanalysis was done using images from triplicates, and representative images are shown. (D) HM-SV40 (1 × 106 cells) were stimulated with NT (10 nM) for24 or 48 h to measure release of soluble NTR3/sortilin in culture media by ELISA. All conditions were performed in triplicate for each dataset and re-peated three times (n = 3). Significance of comparisons is denoted by *P < 0.05 or **P < 0.001. (E ) HM-SV40 (1 × 106 cells) were stimulated with NT (10 or100 nM) for 24 or 48 h to measure total cellular NTR3/sortilin levels by Western blot analysis, where TREM-2 served as a microglial marker protein andβ-actin as the loading control. All conditions were performed in triplicate for each dataset and were repeated three times (n = 3), and a representativeimage is shown.

Patel et al. PNAS | Published online September 23, 2016 | E7051

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

(Fig. S4 B–D) in microglia pretreated with inhibitors before NTstimulation significantly decrease (P < 0.001), compared with NTstimulation. Noteworthy, methoxyluteolin (5 μM) shows greaterreduction of phosphorylated levels of pmTORSer2448 andpp70S6KThr389 compared with luteolin (5 μM) or any of themTOR inhibitors.

NT-Induced Proinflammatory Cytokine and Chemokine Expression inHuman Microglia Is Dependent on mTOR Activation, Which Is Inhibitedby Luteolin and Methoxyluteolin. Human microglia-SV40 wereserum-starved overnight, pretreated with PI3K and/or mTORinhibitors (0.1–1 μM, 2 h) or with luteolin and methoxyluteolin(0.1–10 μM, 2 h), and then stimulated by NT (10 nM) for 24 h tomeasure release of cytokines and chemokines in serum-freemedia. The release of all proinflammatory significantly decreases(P < 0.001) after treatment with the dual PI3K/mTOR inhibitorPF or the mTOR inhibitors, rapamycin and KU, at optimal in-hibitory concentrations of 0.5 μM. The flavonoids luteolin andmethoxyluteolin (0.5–10 μM) also inhibit release of these me-diators from NT-stimulated human microglia (Fig. 6). As apositive control, the release of proinflammatory cytokine IL-1βand the CXCL8, CCL2, and CCL5 mediators was measured inresponse to LPS (10 ng/mL), which also significantly decreases(P < 0.001) in the presence of PI3K and/or mTOR inhibitors,as well as luteolin or methoxyluteolin (Fig. S5). Methoxyluteolinis a more potent inhibitor than luteolin at equimolar flavonoidconcentrations for the release of proinflammatory cytokine andchemokines from either NT-stimulated microglia (Fig. 6) orLPS-stimulated microglia (Fig. S5), with maximal flavonoid in-hibition at 10 μM.We next investigated whether mTOR signaling is involved in

the production of proinflammatory cytokines and chemokines inhuman microglia after NT stimulation and whether luteolin andmethoxyluteolin can inhibit these responses. The gene expres-

sion of IL-1β, CXCL8, CCL2, and CCL5 in microglia pretreatedwith PI3K and/or mTOR inhibitors (0.1–1 μM) and luteolin ormethoxyluteolin (0.1–10 μM) for 2 h, before stimulation by NT(10 nM) for 12 h, was measured. Pretreatment with PI3K and/ormTOR inhibitors and luteolin or methoxyluteolin (0.1 μM) sig-nificantly decreases (P < 0.001) all cytokine and chemokine genelevels, even after stimulation by NT (Fig. 7) or LPS (Fig. S6).Methoxyluteolin and luteolin (0.1 μM) inhibit the gene expres-sion of CXCL8, CCL2, and CCL5 more potently than the PI3Kand/or mTOR inhibitors in human microglia-SV40 stimulatedwith NT.

NT Stimulated-Proliferation of Human Microglia Is Dependent onmTOR Signaling That Is Inhibited by Luteolin and Methoxyluteolin.Cellular proliferation of human microglia-SV40 stimulated withNT (10 or 100 nM) was assessed by the 3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay after 24 or 48 h.Proliferation of microglia increases (P < 0.001) after 48-hstimulation by NT (10 nM), compared with unstimulatedmicroglia (Fig. 8A), and not after 24 h. To evaluate the in-volvement of mTOR signaling in cellular proliferation, micro-glia-SV40 were pretreated with the PI3K and/or mTORinhibitors or luteolin and methoxyluteolin for 2 h before stim-ulation by NT (10 nM). The PI3K and/or mTOR inhibitors(0.5 μM) and luteolin or methoxyluteolin (5 μM) decrease (P <0.0001) NT-stimulated proliferation of microglia after 48 h(Fig. 8B).

Increased Levels of Serum NTR3/Sortilin Is Detected in the Serum ofChildren with ASD. In view of the increased NT in the serum ofchildren with ASD and the extracellular secretion of solubleNTR3/sortilin from NT-stimulated microglia, we measured sol-uble sortilin in the serum of children with ASD, compared withhealthy controls. Increased levels (P < 0.0001) of soluble NTR3/sortilin are detected in the serum of children with ASD (meanlevels of normal controls: 0.9063 ng/mL ± 0.895, n = 36), comparedwith age- and sex-matched healthy controls (0.1395 ng/mL ± 0.094,n = 20) (Fig. 9A). In an attempt to see if there was a correlationbetween serum-soluble NTR3/sortilin and circulating NT, levels ofthese proteins were measured in the same group of patients withASD and controls. There is a significant positive correlation be-tween serum NTR3/sortilin and NT levels (Spearman’s r = 0.3940,P = 0.0283) (Fig. 9A, Inset).

DiscussionThe impetus for this present study came from our previous re-ports that serum NT levels are increased in children with ASD(26, 27). Unlike the previous studies that used murine microglia(55), the present findings show that NT stimulates primarymicroglia obtained from human brains and also an immortalizedhuman cell line, microglia-SV40. Human microglia express onlyNTR3/sortilin and not the other known NT receptors, NTR1 orNTR2. NTR3/sortilin was previously shown to be expressed inthe immortalized human (37, 56) and murine (55) microglia celllines. Stimulation with NT (nM) increases gene expression andrelease of the proinflammatory cytokine IL-1β and chemokinesCXCL8, CCL2, and CCL5, relevant to ASD. Increased levels ofthese mediators have been reported in the brains (8, 19) andblood (57, 58) of patients with ASD, as well as in mice withautistic-like behavior (59).Stimulation with NT also increases proliferation of human

microglia-SV40 only after 48 h, which is in agreement with theprevious report that incorporation of 3H-thymidine in humanmicroglia C13NJ in response to NT remains unchanged after24 h (56). In addition, although the selective NTR1 antagonistSR48692 was previously shown to reduce binding of labeledNT (IC50 of 238 ± 46 nM) to this cell line (56), for our study,neither the NTR1 nor NTR2 antagonists had any effect on

Fig. 4. NT induces proinflammatory mediator release from human micro-glia via NTR3/sortilin. Immortalized HM-SV40 (1 × 106 cells) were transfectedwith two different predesigned and validated siRNAs targeting humanNTR3/sortilin (NTR3#1 and NTR3#2) or scrambled controls (Sc#1 and Sc#2)for 48 and 72 h before evaluation of NTR3/sortilin gene levels by qRT-PCR(A) and protein levels by Western blot analysis (B). Control and siRNA-transfected microglia-SV40 (5 × 104 cells) were stimulated with NT (10 nM)for 24 h to measure release of IL-1β (C), CXCL8 (D), CCL2 (E), and CCL5 (F) byELISA. All conditions were performed in triplicate for each dataset and wererepeated three times (n = 3). Significance of comparisons is denoted by***P < 0.0001.

E7052 | www.pnas.org/cgi/doi/10.1073/pnas.1604992113 Patel et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

proinflammatory cytokine/chemokine release from NT-stimulatedhuman microglia-SV40, implying that these NT responses aremediated via NTR3/sortilin.The difference in the localization of immunodetectable NTR3/

sortilin from the cytosol in nonstimulated microglia to the cellsurface after stimulation with NT led us to wonder whethersortilin may be secreted extracellularly. The colocalization ofNTR3/sortilin with Iba-1, which interacts with actin and is oftenused to detect microglia (60), allowed us to speculate that Iba-1may participate in movement of NTR3/sortilin to the cellsurface. The extracellular domain of NTR3/sortilin, known assoluble sortilin (61), is shown here to be secreted from NT-stimulated cultured human microglia. Soluble sortilin is also in-creased in the serum of children with ASD, compared withnormal healthy children. Because there is a strong positive cor-relation of the soluble receptor with serum NT levels in thesesame patients with ASD, serum NTR3/sortilin may bind to serumNT and limit its biological activity much like the soluble IL-1receptor binds to circulating IL-1β (62).Our experiments next explored the signaling pathways in-

volved following NT stimulation of human microglia mediated byNTR3/sortilin by the use of a kinase array blot that detectsphosphorylated substrates; this array shows increased mTORsignaling, following which it was determined that activation ofthe mTOR pathway is necessary for proinflammatory mediatorexpression in NT-stimulated human microglia. These findingsprovide evidence why some patients with ASD who have genemutations in one negative regulatory protein of mTOR, thephosphatase and tensin homolog (63, 64), develop inflammationof the brain that is linked to ASD pathogenesis (65, 66).Our findings also suggest that mTOR signaling is involved in

the transcriptional regulation of proinflammatory cytokine andchemokine synthesis in human microglia. This effect may bemediated via the activation of nuclear factor-kappa B (NF-ĸB)

(67) and the signal transducer and activator of transcription (STAT)pathways (68), critical for transcription of proinflammatory cy-tokines and chemokines (69). An important finding is that theflavonoids luteolin and methoxyluteolin significantly inhibit geneexpression of all the proinflammatory mediators, as well as theactivation of mTOR, after stimulation by NT.Not only did the flavonoids inhibit NT-stimulated microglial

responses but they also inhibited LPS-stimulated proinflamma-tory cytokine and chemokine synthesis in microglia, as previouslyshown in murine microglia (48, 70). Greater concentrations offlavonoids are required to inhibit cytokine or chemokine proteinrelease, compared with gene expression. This apparent discrep-ancy may be due to some initial secretion before gene expressionis fully inhibited by the lower flavonoid concentrations. Alter-natively, there may be differential inhibition of gene expression,involving the inhibition of nuclear transcription targets (NF-KBor STAT), compared with cytokine or chemokine protein se-cretion. Instead, mediator trafficking and secretion may involveinhibition of specific target proteins involved in vesicle fusion,such as soluble N-ethylmaleimide-sensitive factor attachmentprotein complexes (71, 72).The source of the increased levels of serum NT in children

with ASD is not known. The highest levels of serum NT we hadmeasured before were present in those children with ASD whohad gastrointestinal symptoms (27). NT is also found in the gut(29, 73) and increases permeability of the intestinal lumen (74).NT may enter the blood and reach the brain by stimulatingperivascular mast cells (75, 76), which disrupt both the gut–bloodbarrier (77, 78) and the blood–brain barrier (BBB) (79–81) (Fig.9B). Microglia-derived IL-1β, CXCL8, CCL2, and CCL5 in re-sponse to NT can augment the activation of perivascular mastcells, further disrupting the BBB (82–84), perhaps initiatinga feedback mechanism. Microglia are stimulated by mast cell-derived histamine (85) and tryptase (86). Hence, communication

Fig. 5. NT stimulates human microglia via activation of mTOR signaling, inhibited by luteolin (Lut) and methoxyluteolin (Methlut). (A) HM-SV40 (1 × 106 cells)were stimulated with NT (10 nM) for 0–60 min. The total and phosphorylated levels of mTOR and mTOR substrates p70S6K (increase after 30 min of NTstimulation, red box) and 4EBP1 were analyzed by Western blot analysis, for which β-actin served as the loading control. (B) HM-SV40 were preincubated withthe dual PI3K/mTOR inhibitor (PF, 0.5 μM) and the mTOR [rapamycin (Rap) or KU, 0.5 μM, 24 h] inhibitors or flavonoids (Lut and Methlut, 5 μM) in serum-freemedia overnight and, and then stimulated with NT (10 nM) for 30 min. The total and phosphorylated levels of mTOR and mTOR substrates were assessed byWestern blot analysis. Results from densitometric analysis are presented as normalized phosphorylated-to-total protein levels of pmTORSer2448 and mTOR(C), pp70S6KThr389 and p70S6K (D), and p4EBP1Thr37/46 and 4EBP1 (E). All conditions were performed in triplicate for each dataset and were repeated threetimes (n = 3). Significance of comparisons was determined for stimulated cells and for those cells with inhibitors/flavonoids, as denoted by the horizontal lines(P < 0.001 or P < 0.0001), and also among each of the inhibitor/flavonoid treatments shown by the horizontal brackets and by the corresponding *P < 0.05,**P < 0.001, or ***P < 0.0001.

Patel et al. PNAS | Published online September 23, 2016 | E7053

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

between mast cells and microglia (15, 87), found to be activatedin brains of patients with ASD (8–11, 88), is implicated in in-flammation of the brain (89).Luteolin had previously been shown to inhibit only LPS-

induced IL-6 release from both primary and immortalizedmurine microglia (17). In vivo, luteolin reversed autism-likebehavior in the maternal immune activation mouse model (50,90, 91). Moreover, two pilot, open-label, clinical studies usinga luteolin-containing dietary formulation reported significantimprovement in attention and sociability in children with ASD(51, 52). Methoxyluteolin is a more potent inhibitor thanluteolin or rapamycin (92) and other mTOR inhibitors (93),indicating that it could be developed into an effective treatmentfor ASD.The several findings presented in this report increase our

understanding of the mechanistic pathway by which the peptideNT may play an important causal role in the pathogenesis ofASD. Presently, there is no clinically effective drug to treat thepathophysiology of ASD (94). Targeting NTR3/sortilin and/orusing methoxyluteolin may provide important novel therapeuticapproaches for ASD.

MethodsHuman Microglia Cell Culture. The immortalized human microglia-SV40 cellline derived from primary human microglia was purchased from AppliedBiological Materials, Inc. (ABM, Inc.) and cultured in Prigrow III medium

supplemented with 10% (vol/vol) FBS and 1% penicillin/streptomycin in type Icollagen-coated T25-flasks (ABM, Inc.). Microglia-SV40 maintained specificphenotype and proliferation rates for over 10 passages, during which allexperiments were performed using multiple microglia thaws and sub-cultured cells. Experiments were carried out in type I collagen-coated platesor four-well chamber slides (BD PureCoat ECM Mimetic Cultureware Colla-gen I peptide plates; Becton Dickinson). Primary human microglia isolatedfrom human brain tissue were purchased from ScienCell Research Labora-tories and cultured in microglia medium supplemented with 5% (vol/vol)FBS, 1% penicillin/streptomycin, and 5% (vol/vol) microglia growth supple-ment in poly-L-lysine–coated T-25 flasks (ScienCell Research Laboratories) orChamber Slide Lab-Tek II CC2 chamber slides (Thermo Fisher Scientific). Pri-mary human microglia were not subcultured and were used within 7 d aftercultures were initiated. Cell viability was determined by trypan blue(0.4%) exclusion.

Microglia Treatments. Huma microglia were stimulated by NT or LPS, and/orpretreated with NTR1 selective nonpeptide antagonist SR48692 or non-selective antagonist NTR1/2 (10–1,000 nM; Sigma–Aldrich), the dual PI3K/mTOR inhibitor PF (10 nM–1,000 nM, 2 h; TOCRIS Biosciences) or the mTORinhibitor rapamycin (10 nM–1,000 nM, 24 h) or KU (10–1,000 nM, 2 h); andthe flavonoids (luteolin or methoxyluteolin, 0.1–10 μM; 2, 12, or 24 h;PharmaScience Nutrients). All inhibitors were dissolved in water or DMSO(final concentration of <0.1%). For siRNA knockdown experiments, two differentpredesigned and validated Silencer Select siRNAs targeting human NTR3/sortilinand control siRNAs were purchased from Life Technologies. siRNA (10–100 nM)transfection was carried out using Lipofectamine RNAiMAX in Opti-MEMreduced serum and antibiotic-free medium (Life Technologies) for 48 h be-fore evaluation of gene knockdown qRT-PCR and protein analysis byWestern blot analysis.

Detection of NT Receptor Expression. The presence of gene expression ofNTR3/sortilin was determined by quantitative real-time PCR (qRT-PCR) usingspecific primers and antibodies to distinguish between the three known NTreceptor subtypes NTR1, NTR2, and NTR3/sortilin. Receptor gene expressionof NTR1, NTR2, and NTR3/sortilin was measured by qRT-PCR using Taqmangene expression assays and validated primers (Applied Biosystems). TotalRNA from control and NT-stimulated (NT full length of the active frag-ment residues 8–13) microglia (2.5 × 105 cells per six-well type I collagen

Fig. 6. Microglia proinflammatory mediator release in response to NT isattenuated by the PI3K/mTOR inhibitors and the flavonoids Lut and Methlut.HM-SV40 (5 × 104 cells) were pretreated with the dual PI3K/mTOR (PF,0.1 μM) and the mTOR (Rap and KU, 0.1 μM) inhibitors or flavonoids [Lut andMethlut, 5 μM (Upper) or 0.1–10 μM (Lower)] for 2 h, and then stimulatedwith NT (10 nM) for 24 h in serum-free medium to measure release of IL-1β(A), CXCL8 (B), CCL2 (C), and CCL5 (D) by ELISA. All inhibitors were dissolvedin water or DMSO with a final concentration <0.1%. All conditions wereperformed in triplicate for each dataset and were repeated three times(n = 3). Significance of comparisons was determined for stimulated cells andfor those cells with inhibitors/flavonoids, as denoted by the horizontal lines(P < 0.001 or P < 0.0001), and also among each of the inhibitor/flavonoidtreatments shown by the horizontal brackets and by corresponding *P <0.05, **P < 0.001, and ***P < 0.0001.

Fig. 7. Microglia proinflammatory mediator gene expression in responseto NT is attenuated by the PI3K/mTOR inhibitors and the flavonoids Lut andMethlut. HM-SV40 (5 × 104 cells) were pretreated with the dual PI3K/mTOR(PF, 0.1 μM) and the mTOR (Rap and KU, 0.1 μM) inhibitors or the flavonoids(Lut and Methlut, 0.1 μM) for 2 h, and then stimulated with NT (10 nM) for12 h in serum-free media to determine changes in gene levels of IL-1β (A),CXCL8 (B), CCL2 (C), and CCL5 (D) by qRT-PCR. All inhibitors were dissolvedin water or DMSO with a final concentration <0.1%. Results were nor-malized against the endogenous gene GAPDH and are expressed relative tothe mean of the gene of interest. All conditions were performed in tripli-cate for each dataset and were repeated three times (n = 3). Significance ofcomparisons was determined for stimulated cells and for those cells withinhibitors/flavonoids denoted by the horizontal lines (P < 0.001 or P <0.0001), and also among each of the inhibitor/flavonoid treatments shownby the horizontal brackets and by corresponding *P < 0.05 and ***P <0.0001.

E7054 | www.pnas.org/cgi/doi/10.1073/pnas.1604992113 Patel et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

or poly-L-lysine–coated plates; Becton Dickinson) for 24 h before stimulationwith NT (1–1,000 nM) or LPS (10–1,000 ng/mL) (Sigma–Aldrich) was carried outwas isolated after 6, 12, or 24 h using an RNeasy Mini Kit (Qiagen) according tothe manufacturer’s instructions. Reverse transcription was performed with300 ng of total RNA using the iScript cDNA Synthesis Kit (Bio-Rad). For qRT-PCR,samples were run at 45 cycles using an Applied Biosystems 7300 Real-Time PCRSystem. Relative mRNA abundance was determined from standard curves runwith each experiment. Gene expression was normalized to GAPDH, which wasused as an endogenous control. The gene expression of receptors of NT (NTR1,NTR2, and NTR3) and the cell type-specific antigens for the microglial lineage,CD11b and CD86, were also determined.

Protein levels of NTR1, NTR2, and NTR3/sortilin were determined byWestern blot analysis on cellular lysates harvested from microglia (1 × 106

cells) stimulated with NT (1–100 nM) for 24 or 48 h. The membranes wereprobed with the following primary antibodies: NTR1 G9 (Santa Cruz Bio-

technology); NTR2 (EMD Millipore Corporation); SORT1 (Sigma–Aldrich);NTR3 C20, triggering receptor expressed on myeloid cells-2 (TREM-2), andIba-1 (Santa Cruz Biotechnology); and β-actin for the loading control (CellSignaling Technology). All proteins were visualized with horseradish-per-oxidase–conjugated secondary antibodies and then by enhanced Super-Signal West Pico chemiluminescence (Fisher Scientific).

Cellular localization studies on NTR3/sortilin in basal and NT-stimulatedmicroglia were done using confocal microscopy. Microglia (5 × 103 cells perfour-well chamber–coated slide) were stimulated with NT (10 or 100 nM, 12or 24 h). Cells were fixed in 4% (wt/vol) paraformaldehyde in PBS for 10 minon ice and/or permeabilized using a solution of 0.1% Triton X-100 and0.05% saponin. Free aldehydes were quenched with 50 mM glycine/lysine inPBS. Blocking was carried out with 1% (wt/vol) BSA in PBS containing 10%(vol/vol) goat or donkey serum (EMD Millipore) for 30 min at room tem-perature. Cells were incubated overnight with the following monoclonalprimary antibodies: NTR3/sortilin C-20, and TREM-2 (from Santa Cruz Bio-technology) at a dilution of 1:500 and CD11b/integrin αM, Iba-1, and NTR3/sortilin (from Abcam). For immunofluorescence detection, the followingsecondary antibodies were used: anti-rabbit IgG-FITC (Santa Cruz Bio-technology) or anti-rabbit Alexa 488 and/or anti-goat Alexa 594 (Life Tech-nologies) at a dilution of 1:1,000 for 2 h. Nuclei were stained with DAPI orHoechst (50 ng/mL), and slides were then mounted using ProLong DiamondAntifade Mountant (Life Technologies). Imaging was carried out using aNikon A1R inverted confocal microscope, with an automated stage using aconventional point scanner (1,020 × 1,020 pixel field view) and equippedwith a spectral detector for confocal scanning laser imaging and a trans-mitted-laser DIC detector (three solid-state lasers and a dual-line argon gaslaser, producing excitation lines at 403, 457/476, 488/514, 560, and 640 nm),and NIS-Element Software for image analysis (Nikon Instruments, Inc.).

Protein levels of soluble NTR3/sortilin in culture medium from humanmicroglia after stimulation by NT (10 nM) for 24 or 48 h were measured usingthe SORT1 ELISA (LifeSpan BioSciences, Inc.). Initially, human microglia wereseeded (2 × 106 cells per type I collagen-coated T25-flask) for 24 h and serum-starved for 12 h, before stimulation with NT (10 nM). After 24 or 48 h, su-pernatant fluids were collected (10 mL) and concentrated (100 μL) using 10-Kcellulose membrane centrifugal filter units (Merck Millipore). For all exper-iments, the control cells were treated with an equal volume of culture me-dium, and the minimum detectable level for soluble NTR3/sortilin by ELISAwas 0.157 ng/mL.

Proinflammatory Mediator Gene Expression. Microglia (1 ×105 cells per well)were seeded in six-well, type I collagen- or poly-L-lysine–coated plates for24 h before stimulation with NT (1–1,000 nM) or LPS (10–1,000 ng/mL) wascarried out for 12 or 24 h. Total RNA was isolated as described above, andqRT-PCR was performed using Taqman gene expression assays to assess theexpression of IL-1β, CXCL8, CCL2, and CCL5 in microglia after NT stimulation

Fig. 9. Increased serum NTR3/sortilin levels in ASD and the proposed scheme by which NT may contribute to inflammation of the brain. (A) Levels of solubleNTR3/sortilin and circulating NT were measured by ELISA in the serum of children with ASD, compared with age- and sex-matched healthy controls. Sig-nificance of comparisons is denoted by P < 0.0001. (Inset) Positive correlation between serum sortilin and NT levels is shown using the Spearman rankcorrelation test (Spearman’s r = 0.3940, P = 0.0283). (B) Diagrammatic representation of how serum NT could derive primarily from the gut and increasepermeability of the intestinal lumen and the BBB by stimulating perivascular mast cells. NT in the brain could then stimulate microglia via NTR3/sortilin, whichis elevated in the serum of children with ASD. Proinflammatory mediator release from microglia through activation of mTOR signaling kinase thus maycontribute to inflammation of the brain and the pathogenesis of ASD. The flavonoid Methlut inhibits these processes and could be a novel treatment of ASD.

Fig. 8. NT-stimulated proliferation of human microglia depends on mTORsignaling and is inhibited by Lut and Methlut. (A) HM-SV40 (5 × 103 cells perwell for 48 h or 10 × 103 cells per well for 24 h) were stimulated with NT (10or 100 nM) in phenol-free medium, and proliferation was measured usingthe MTT-based assay. Absorbance of converted dye was measured at awavelength of 570 nm with background subtraction at 690 nm; cell pro-liferation was calculated with control cells set at 100%. (B) Microglia (5 × 103

cells per well) were pretreated with the dual PI3K/mTOR (PF, 0.5 μM) and themTOR (Rap and KU, 0.5 μM) inhibitors or the flavonoids (Lut and Methlut,5 μM) for 2 h and then stimulated with NT (10 nM) for 48 h, and an MTTassay was performed. All conditions were done in triplicate for each datasetand were repeated three times (n = 3). Results are expressed as the per-centage (%) of cell proliferation relative to the control cells, with signifi-cance of comparisons determined for control and stimulated cells, asdenoted by *P < 0.05 or **P < 0.001. Multiple comparisons were also madefor stimulated cells and for those cells with inhibitors/flavonoids, as denotedby the horizontal lines (P < 0.001 or P < 0.0001), and also among each of theinhibitor/flavonoid treatments shown by the horizontal brackets and bycorresponding *P < 0.05 and **P < 0.001.

Patel et al. PNAS | Published online September 23, 2016 | E7055

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

for 6, 12, or 24 h. For select experiments, microglia were pretreated withPI3K/mTOR inhibitors before stimulation with NT or LPS for 12 h in serum-free media before harvesting cell lysates. For all qRT-PCR studies, the sug-gested best-coverage Taqman probes were selected and performed. Thegene levels of IL-1β, TNF, CXCL8, CCL2, and CCL5 were measured, and ex-pression was normalized to GAPDH endogenous control.

Proinflammatory Mediator Release. The detection of the various cytokines/chemokines within human microglia cell-conditioned culture medium/supernatant fluid was carried out using the Human Cytokine Array Panel A(R&D Systems). Thereafter, specific mediator release in human microglia-conditioned culture medium was quantified using commercially availableELISA kits (R&D Systems) as per the manufacturer’s instructions. Microglia(0.5 × 105 cells per well) were seeded in 12- or 24-well type I collagen- orpoly-L-lysine–coated plates for 24 h before stimulation with NT (1–1,000 nM)or LPS (10–1,000 ng/mL) was carried out. For select experiments, microgliawere pretreated with PI3K/mTOR for 30 min before stimulation with NT orLPS for 24 h in serum-free media. After 12 or 24 h, supernatant fluids werecollected and IL-1, IL-6, TNF, CXCL8, CCL2, and CCL5 release was measured.For all experiments, the control cells were treated with an equal volume ofculture medium, and the minimum detectable level for all mediators byELISA was 5 pg/mL.

Assessing mTOR Activation. The detection of the various signaling proteinswithin human microglia was carried out using the Human Phospho-KinaseArray (R&D Systems). The activation of mTORC1 was assessed by phos-phorylation of pmTOR Ser2448 and the downstream mTORC1 substrates,pp70S6K and p4EBP1, whereas mTORC2 activation was determined by pAKTSer473 levels, using Western blot analysis. Microglia (1 × 106 cells) wereseeded in 100-mm type I collagen-coated dishes for 24 h, serum-starvedovernight (or pretreated with inhibitors), and then stimulated with NT (10–100 nM) for 0–60 min before cell lysates were harvested in radioimmuno-precipitation assay buffer (Sigma–Aldrich) containing Halt Protease andPhosphatase Inhibitor Mixtures (Thermo Fisher Scientific). The total proteinconcentration was determined by the bicinchoninic acid assay (ThermoFisher Scientific) using BSA protein as a standard. The total cellular proteins(20 or 40 μg) were separated using 4–20% Mini-PROTEAN TGX precast gels(BioRad) under SDS denaturing conditions and electrotransferred onto PVDFmembranes (EMD Millipore). Blocking was carried out with 5% (wt/vol) BSAin Tris-buffered saline containing 0.1% Tween-20. The membranes wereprobed with the following primary antibodies: mTOR (7C10); pmTORSer2448; mTORC1 substrates p70S6K, pp70SK Thr389, 4EBP1, and p4EBP1Thr37/46; and β-actin as the loading control. All proteins were visualizedwith horseradish peroxidase-conjugated secondary antibodies and then bySuperSignal West Pico enhanced chemiluminescence (Thermo Fisher Scientific).In parallel experiments, using Pathscan Phospho-ELISA Kits (R&D Systems)for the detection of pAKTSer473, pmTORSer2448, and pp70S6K proteins, thelevels of phosphorylated proteins were measured in microglia after thetreatments described.

Microglia Proliferation. Microglia (1 × 104 cells per well for 24 h or 5 × 103

cells per well for 48 h) were seeded in type I collagen-coated, 96- or 48-wellflat-bottomed plates (Becton Dickinson) for 24 h before pretreatment withPI3K/mTOR inhibitors for 30 min and/or stimulation with NT (10–100 nM). Allexperiments were conducted in phenol-free PriGrow III media (ABM, Inc.).Proliferation was measured using the MTT-based in vitro toxicology assay kit(Sigma–Aldrich); according to the manufacturer’s instructions, MTT stocksolution (5 mg/mL) was added to each culture being assayed to equal one-tenth of the original culture volume and incubated for 3–4 h, after which theconverted dye was solubilized with acidic isopropanol (HCl, 0.04–0.1 N, inabsolute isopropanol). Absorbance of converted dye was measured at awavelength of 570 nm with background subtraction at 690 nm, and thepercentage of cell proliferation was calculated as percent total with controlcells as 100%.

Human Subjects. Fasting blood was obtained from Caucasian male children(n = 36, range: 4–13 y of age), whose blood was obtained as part of their di-agnostic workup at the Attikon General Hospital in Athens, Greece. Children

were diagnosed with ASD based on clinical assessment and corroborated bymeeting the cutoff scores on both the Diagnostic and Statistical Manual ofMental Disorders, Fifth Edition (DSM-5) symptom list and the autism di-agnostic observation schedule algorithm. They were medication-free beforeblood draw for at least 2 wk for all psychotropic medications and 4 wk forfluoxetine or depot neuroleptics. The exclusion criteria were as follows: (i )any genetic condition linked to ASD (e.g., Rett syndrome, fragile Xsyndrome, tuberous sclerosis, focal epilepsy); (ii ) any genetic syndromeinvolving the CNS, even if the link with ASD was uncertain; (iii ) anyneurological disorder involving pathology above the brainstem, other thanuncomplicated nonfocal epilepsy; (iv) contemporaneous evidence, orunequivocal retrospective evidence, of probable neonatal brain damage;(v) clinically significant visual or auditory impairment, even after correction;(vi) any severe nutritional or psychological deprivation; (vii) cutaneous orsystemic mastocytosis; (viii) history of allergies and upper airway diseases;and (ix) history of inflammatory diseases (e.g., juvenile rheumatoid arthritis,inflammatory bowel disease). Informed consent was obtained from all par-ents. Serum was also collected from normally developing, healthy malechildren (n = 20, range: 4–13 y of age) unrelated to subjects with ASD, whowere seen for routine health visits at the Pediatric Department of the SocialSecurity Administration polyclinic. Serum samples were labeled only with acode number and with the age and sex of the subjects. All ASDs and controlblood samples were prepared immediately, and serum was stored at −80 °C.All samples were de-identified except for age and sex, and were consideredunder Exemption number 4 as per institutional review board (IRB) guide-lines. This study was approved by the IRB decision 8/20–7-11 of the AttikonGeneral Hospital and confirmed by the Tufts Medical Center IRB. Sampleswere then transported on dry ice to Boston for analysis. The detection ofsoluble NTR3/sortilin in serum was done using the human SORT1 ELISA(LifeSpan BioSciences, Inc.).

Statistical Analysis. All conditions were performed in triplicate, and all ex-periments were repeated at least three times (n= 3). Results from cultured cellsare presented as mean ± SD. Comparisons were made between (i) controland stimulated cells and (ii) stimulated cells with and without siRNA pre-treatment using the unpaired, two-tailed, Student’s t test, with significance ofcomparisons denoted by the horizontal lines and by *P < 0.05, **P < 0.001,and ***P < 0.0001. Comparisons were also made between (i) all conditionswith stimulated cells and with inhibitors using one-way ANOVA followed bypost hoc analysis by Dunnett’s multiple comparison test, for which significanceis denoted by horizontal lines and indicated values of P < 0.001 or P < 0.0001and (ii) all of the inhibitors/flavonoids among themselves using one-wayANOVA followed by post hoc analysis by Tukey’s multiple comparison test,with those conditions for which there is significance denoted by the hori-zontal brackets and by the corresponding probability value (*P < 0.05, **P <0.001, and ***P < 0.0001). Analysis of human serum samples is presented asa scattergram with symbols representing individual data points and horizontallines representing the mean for each group. Normality of distribution waschecked with the Shapiro–Wilk’s test. Comparison between the healthy con-trol and ASD groups was performed using the Mann–Whitney U test and theWilcoxon matched pair test. Correlations between serum sortilin and NT levelswere examined using the Spearman rank correlation test. Significance ofcomparisons is denoted by P < 0.0001.The analysis was performed usingGraphPad Prism version 5.0 software (GraphPad Software). Representa-tive images for Western blots were scanned and analyzed using ImageJ(NIH; https://imagej.nih.gov/ij/) and confocal images were analyzed usingFiji ImageJ.

ACKNOWLEDGMENTS. We thank Christopher Talbot (Tufts University Schoolof Medicine) for assistance with troubleshooting experimental methodolo-gies. We also thank Jerold Harmatz (Tufts University School of Medicine) foradvice with the statistical analysis. We thank Drs. William Bachovchin, DavidSanford, and Yuhong Zhou (Tufts University School of Medicine) forchecking the purity of methoxyluteolin using nuclear magnetic resonanceand mass spectroscopy, as well as Dr. Chia-Ling Tsai (Kainan University) forproviding us with a chemically synthesized methoxyluteolin standard. Thiswork was supported, in part, by a grant from the Jane Botsford JohnsonFoundation (to T.C.T.) and a Predoctoral Fellowship from the Nancy LurieMarks Family Foundation (to A.B.P.).

1. Lai MC, Lombardo MV, Baron-Cohen S (2014) Autism. Lancet 383(9920):896–910.2. Volkmar FR, McPartland JC (2014) From Kanner to DSM-5: Autism as an evolving di-

agnostic concept. Annu Rev Clin Psychol 10:193–212.3. Zablotsky B, Black LI, Maenner MJ, Schieve LA, Blumberg SJ (2015) Estimated preva-

lence of autism and other developmental disabilities following questionnaire changesin 2014 national health interview survey. Natl Health Stat Report 87:1–20.

4. Willsey AJ, State MW (2015) Autism spectrum disorders: From genes to neurobiology.Curr Opin Neurobiol 30:92–99.

5. Kemper TL, Bauman ML (2002) Neuropathology of infantile autism. Mol Psychiatry 7(Suppl 2):S12–S13.

6. Paolicelli RC, et al. (2011) Synaptic pruning by microglia is necessary for normal braindevelopment. Science 333(6048):1456–1458.

E7056 | www.pnas.org/cgi/doi/10.1073/pnas.1604992113 Patel et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

7. Shemer A, Erny D, Jung S, Prinz M (2015) Microglia plasticity during health and dis-ease: An immunological perspective. Trends Immunol 36(10):614–624.

8. Vargas DL, Nascimbene C, Krishnan C, Zimmerman AW, Pardo CA (2005) Neuroglialactivation and neuroinflammation in the brain of patients with autism. Ann Neurol57(1):67–81.

9. Morgan JT, et al. (2010) Microglial activation and increased microglial density ob-served in the dorsolateral prefrontal cortex in autism. Biol Psychiatry 68(4):368–376.

10. Rodriguez JI, Kern JK (2011) Evidence of microglial activation in autism and its pos-sible role in brain underconnectivity. Neuron Glia Biol 7(2-4):205–213.

11. Morgan JT, Barger N, Amaral DG, Schumann CM (2014) Stereological study ofamygdala glial populations in adolescents and adults with autism spectrum disorder.PLoS One 9(10):e110356.

12. Suzuki K, et al. (2013) Microglial activation in young adults with autism spectrumdisorder. JAMA Psychiatry 70(1):49–58.

13. Edmonson C, Ziats MN, Rennert OM (2014) Altered glial marker expression in autisticpost-mortem prefrontal cortex and cerebellum. Mol Autism 5(1):3.

14. Pannell M, Szulzewsky F, Matyash V, Wolf SA, Kettenmann H (2014) The sub-population of microglia sensitive to neurotransmitters/neurohormones is modulatedby stimulation with LPS, interferon-γ, and IL-4. Glia 62(5):667–679.

15. Skaper SD, Giusti P, Facci L (2012) Microglia and mast cells: Two tracks on the road toneuroinflammation. FASEB J 26(8):3103–3117.

16. Theoharides TC, Valent P, Akin C (2015) Mast cells, mastocytosis, and related disor-ders. N Engl J Med 373(2):163–172.

17. Jang S, Kelley KW, Johnson RW (2008) Luteolin reduces IL-6 production in microgliaby inhibiting JNK phosphorylation and activation of AP-1. Proc Natl Acad Sci USA105(21):7534–7539.

18. Li X, et al. (2009) Elevated immune response in the brain of autistic patients.J Neuroimmunol 207(1-2):111–116.

19. Ashwood P, et al. (2011) Elevated plasma cytokines in autism spectrum disordersprovide evidence of immune dysfunction and are associated with impaired behavioraloutcome. Brain Behav Immun 25(1):40–45.

20. Zimmerman AW, et al. (2005) Cerebrospinal fluid and serum markers of inflammationin autism. Pediatr Neurol 33(3):195–201.

21. Estes ML, McAllister AK (2015) Immune mediators in the brain and peripheral tissuesin autism spectrum disorder. Nat Rev Neurosci 16(8):469–486.

22. Young AM, et al. (2016) From molecules to neural morphology: Understandingneuroinflammation in autism spectrum condition. Mol Autism 7:9.

23. Hagberg H, Gressens P, Mallard C (2012) Inflammation during fetal and neonatal life:Implications for neurologic and neuropsychiatric disease in children and adults. AnnNeurol 71(4):444–457.

24. Theoharides TC, Asadi S, Patel AB (2013) Focal brain inflammation and autism.J Neuroinflammation 10:46.

25. Theoharides TC, Stewart JM, Panagiotidou S, Melamed I (2015) Mast cells, brain in-flammation and autism. Eur J Pharmacol 778:96–102.

26. Angelidou A, et al. (2010) Neurotensin is increased in serum of young children withautistic disorder. J Neuroinflammation 7:48.

27. Tsilioni I, et al. (2014) Elevated serum neurotensin and CRH levels in children withautistic spectrum disorders and tail-chasing Bull Terriers with a phenotype similar toautism. Transl Psychiatry 4:e466.

28. Carraway R, Leeman SE (1973) The isolation of a new hypotensive peptide, neuro-tensin, from bovine hypothalami. J Biol Chem 248(19):6854–6861.

29. Dobner PR, Barber DL, Villa-Komaroff L, McKiernan C (1987) Cloning and sequenceanalysis of cDNA for the canine neurotensin/neuromedin N precursor. Proc Natl AcadSci USA 84(10):3516–3520.

30. Vincent B, Vincent JP, Checler F (1994) Neurotensin and neuromedin N undergodistinct catabolic processes in murine astrocytes and primary cultured neurons. Eur JBiochem 221(1):297–306.

31. Navarro V, et al. (2001) Pharmacological properties of the mouse neurotensin re-ceptor 3. Maintenance of cell surface receptor during internalization of neurotensin.FEBS Lett 495(1-2):100–105.

32. Chalon P, et al. (1996) Molecular cloning of a levocabastine-sensitive neurotensinbinding site. FEBS Lett 386(2-3):91–94.

33. Mazella J, et al. (1996) Structure, functional expression, and cerebral localization ofthe levocabastine-sensitive neurotensin/neuromedin N receptor from mouse brain.J Neurosci 16(18):5613–5620.

34. Vincent JP, Mazella J, Kitabgi P (1999) Neurotensin and neurotensin receptors. TrendsPharmacol Sci 20(7):302–309.

35. Petersen CM, et al. (1997) Molecular identification of a novel candidate sorting re-ceptor purified from human brain by receptor-associated protein affinity chroma-tography. J Biol Chem 272(6):3599–3605.

36. Hermans-Borgmeyer I, Hermey G, Nykjaer A, Schaller C (1999) Expression of the100-kDa neurotensin receptor sortilin during mouse embryonal development. BrainRes Mol Brain Res 65(2):216–219.

37. Martin S, Dicou E, Vincent JP, Mazella J (2005) Neurotensin and the neurotensin re-ceptor-3 in microglial cells. J Neurosci Res 81(3):322–326.

38. Smith AM, Dragunow M (2014) The human side of microglia. Trends Neurosci 37(3):125–135.

39. Seok J, et al.; Inflammation and Host Response to Injury, Large Scale CollaborativeResearch Program (2013) Genomic responses in mouse models poorly mimic humaninflammatory diseases. Proc Natl Acad Sci USA 110(9):3507–3512.

40. Lee Y (2015) Roles of mTOR signaling in brain development. Exp Neurobiol 24(3):177–185.

41. Kwon CH, et al. (2006) Pten regulates neuronal arborization and social interaction inmice. Neuron 50(3):377–388.

42. Wang H, Doering LC (2013) Reversing autism by targeting downstream mTOR sig-naling. Front Cell Neurosci 7:28.

43. Huber KM, Klann E, Costa-Mattioli M, Zukin RS (2015) Dysregulation of mammaliantarget of rapamycin signaling in mouse models of autism. J Neurosci 35(41):13836–13842.

44. Dello Russo C, Lisi L, Tringali G, Navarra P (2009) Involvement of mTOR kinase incytokine-dependent microglial activation and cell proliferation. Biochem Pharmacol 78(9):1242–1251.

45. Kim MS, Kuehn HS, Metcalfe DD, Gilfillan AM (2008) Activation and function of themTORC1 pathway in mast cells. J Immunol 180(7):4586–4595.

46. Middleton E, Jr, Kandaswami C, Theoharides TC (2000) The effects of plant flavonoidson mammalian cells: Implications for inflammation, heart disease, and cancer.Pharmacol Rev 52(4):673–751.

47. Dirscherl K, et al. (2010) Luteolin triggers global changes in the microglial tran-scriptome leading to a unique anti-inflammatory and neuroprotective phenotype.J Neuroinflammation 7:3.

48. Kao TK, et al. (2011) Luteolin inhibits cytokine expression in endotoxin/cytokine-stimulated microglia. J Nutr Biochem 22(7):612–624.

49. Zhu L, et al. (2014) Luteolin inhibits SH-SY5Y cell apoptosis through suppression of thenuclear transcription factor-κB, mitogen-activated protein kinase and protein kinaseB pathways in lipopolysaccharide-stimulated cocultured BV2 cells. Exp Ther Med 7(5):1065–1070.

50. Parker-Athill E, et al. (2009) Flavonoids, a prenatal prophylaxis via targeting JAK2/STAT3 signaling to oppose IL-6/MIA associated autism. J Neuroimmunol 217(1-2):20–27.

51. Theoharides TC, Asadi S, Panagiotidou S (2012) A case series of a luteolin formulation(NeuroProtek�) in children with autism spectrum disorders. Int J ImmunopatholPharmacol 25(2):317–323.

52. Taliou A, Zintzaras E, Lykouras L, Francis K (2013) An open-label pilot study of aformulation containing the anti-inflammatory flavonoid luteolin and its effects onbehavior in children with autism spectrum disorders. Clin Ther 35(5):592–602.

53. Weng Z, Patel AB, Panagiotidou S, Theoharides TC (2015) The novel flavone tetra-methoxyluteolin is a potent inhibitor of human mast cells. J Allergy Clin Immunol135(4):1044–52.e5.

54. Walle T (2007) Methylation of dietary flavones greatly improves their hepatic meta-bolic stability and intestinal absorption. Mol Pharm 4(6):826–832.

55. Dicou E, Vincent JP, Mazella J (2004) Neurotensin receptor-3/sortilin mediates neu-rotensin-induced cytokine/chemokine expression in a murine microglial cell line.J Neurosci Res 78(1):92–99.

56. Martin S, Vincent JP, Mazella J (2003) Involvement of the neurotensin receptor-3 inthe neurotensin-induced migration of human microglia. J Neurosci 23(4):1198–1205.

57. Masi A, et al. (2015) Cytokine aberrations in autism spectrum disorder: A systematicreview and meta-analysis. Mol Psychiatry 20(4):440–446.

58. Tsilioni I, Taliou A, Francis K, Theoharides TC (2015) Children with autism spectrumdisorders, who improved with a luteolin-containing dietary formulation, show re-duced serum levels of TNF and IL-6. Transl Psychiatry 5:e647.

59. Choi GB, et al. (2016) The maternal interleukin-17a pathway in mice promotes autism-like phenotypes in offspring. Science 351(6276):933–939.

60. Ito D, et al. (1998) Microglia-specific localisation of a novel calcium binding protein,Iba1. Brain Res Mol Brain Res 57(1):1–9.

61. Navarro V, Vincent JP, Mazella J (2002) Shedding of the luminal domain of theneurotensin receptor-3/sortilin in the HT29 cell line. Biochem Biophys Res Commun298(5):760–764.

62. Hannum CH, et al. (1990) Interleukin-1 receptor antagonist activity of a human in-terleukin-1 inhibitor. Nature 343(6256):336–340.

63. Butler MG, et al. (2005) Subset of individuals with autism spectrum disorders andextreme macrocephaly associated with germline PTEN tumour suppressor gene mu-tations. J Med Genet 42(4):318–321.

64. Varga EA, Pastore M, Prior T, Herman GE, McBride KL (2009) The prevalence of PTENmutations in a clinical pediatric cohort with autism spectrum disorders, develop-mental delay, and macrocephaly. Genet Med 11(2):111–117.

65. Costa-Mattioli M, Monteggia LM (2013) mTOR complexes in neurodevelopmental andneuropsychiatric disorders. Nat Neurosci 16(11):1537–1543.

66. Kassai H, et al. (2014) Selective activation of mTORC1 signaling recapitulates micro-cephaly, tuberous sclerosis, and neurodegenerative diseases. Cell Reports 7(5):1626–1639.

67. Dan HC, et al. (2008) Akt-dependent regulation of NF-kappaB is controlled by mTORand Raptor in association with IKK. Genes Dev 22(11):1490–1500.

68. Saleiro D, Platanias LC (2015) Intersection of mTOR and STAT signaling in immunity.Trends Immunol 36(1):21–29.

69. Weichhart T, et al. (2008) The TSC-mTOR signaling pathway regulates the innate in-flammatory response. Immunity 29(4):565–577.

70. Park JS, et al. (2007) Anti-inflammatory mechanisms of isoflavone metabolites inlipopolysaccharide-stimulated microglial cells. J Pharmacol Exp Ther 320(3):1237–1245.

71. Yang Y, et al. (2013) Polyphenols differentially inhibit degranulation of distinctsubsets of vesicles in mast cells by specific interaction with granule-type-dependentSNARE complexes. Biochem J 450(3):537–546.

72. Yang Y, et al. (2015) Dynamic light scattering analysis of SNARE-driven membranefusion and the effects of SNARE-binding flavonoids. Biochem Biophys Res Commun465(4):864–870.

73. Carraway RE, Mitra SP, Evers BM, Townsend CM, Jr (1994) BON cells display the in-testinal pattern of neurotensin/neuromedin N precursor processing. Regul Pept 53(1):17–29.

74. Castagliuolo I, et al. (1999) Neurotensin is a proinflammatory neuropeptide in colonicinflammation. J Clin Invest 103(6):843–849.

Patel et al. PNAS | Published online September 23, 2016 | E7057

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

75. Miller LA, Cochrane DE, Carraway RE, Feldberg RS (1995) Blockade of mast cell his-

tamine secretion in response to neurotensin by SR 48692, a nonpeptide antagonist ofthe neurotensin brain receptor. Br J Pharmacol 114(7):1466–1470.

76. Donelan J, et al. (2006) Corticotropin-releasing hormone induces skin vascular per-

meability through a neurotensin-dependent process. Proc Natl Acad Sci USA 103(20):7759–7764.

77. Wallon C, et al. (2008) Corticotropin-releasing hormone (CRH) regulates macromo-lecular permeability via mast cells in normal human colonic biopsies in vitro. Gut

57(1):50–58.78. Theoharides TC, Doyle R (2008) Autism, gut-blood-brain barrier, and mast cells. J Clin

Psychopharmacol 28(5):479–483.79. Theoharides TC (1990) Mast cells: The immune gate to the brain. Life Sci 46(9):

607–617.80. Esposito P, et al. (2002) Corticotropin-releasing hormone and brain mast cells regulate

blood-brain-barrier permeability induced by acute stress. J Pharmacol Exp Ther 303(3):1061–1066.

81. Ribatti D (2015) The crucial role of mast cells in blood-brain barrier alterations. ExpCell Res 338(1):119–125.

82. Theoharides TC, Weinkauf C, Conti P (2004) Brain cytokines and neuropsychiatric

disorders. J Clin Psychopharmacol 24(6):577–581.83. Takeshita Y, Ransohoff RM (2012) Inflammatory cell trafficking across the blood-brain

barrier: Chemokine regulation and in vitro models. Immunol Rev 248(1):228–239.84. Rochfort KD, Cummins PM (2015) The blood-brain barrier endothelium: A target for

pro-inflammatory cytokines. Biochem Soc Trans 43(4):702–706.

85. Dong H, et al. (2014) Histamine induces upregulated expression of histamine recep-tors and increases release of inflammatory mediators from microglia. Mol Neurobiol49(3):1487–1500.

86. Zhang S, Zeng X, Yang H, Hu G, He S (2012) Mast cell tryptase induces microglia activationvia protease-activated receptor 2 signaling. Cell Physiol Biochem 29(5-6):931–940.

87. Skaper SD, Facci L, Giusti P (2014) Mast cells, glia and neuroinflammation: Partners incrime? Immunology 141(3):314–327.

88. Gupta S, et al. (2014) Transcriptome analysis reveals dysregulation of innate immune re-sponse genes and neuronal activity-dependent genes in autism. Nat Commun 5:5748.

89. Theoharides TC, Tsilioni I, Patel AB, Doyle R (2016) Atopic diseases and inflammationof the brain in the pathogenesis of autism spectrum disorders. Transl Psychiatry 6(6):e844.

90. Hsiao EY, McBride SW, Chow J, Mazmanian SK, Patterson PH (2012) Modeling anautism risk factor in mice leads to permanent immune dysregulation. Proc Natl AcadSci USA 109(31):12776–12781.

91. Garay PA, Hsiao EY, Patterson PH, McAllister AK (2013) Maternal immune activationcauses age- and region-specific changes in brain cytokines in offspring throughoutdevelopment. Brain Behav Immun 31:54–68.

92. Li J, Kim SG, Blenis J (2014) Rapamycin: One drug, many effects. Cell Metab 19(3):373–379.

93. Wander SA, Hennessy BT, Slingerland JM (2011) Next-generation mTOR inhibitors inclinical oncology: How pathway complexity informs therapeutic strategy. J Clin Invest121(4):1231–1241.

94. Ghosh A, Michalon A, Lindemann L, Fontoura P, Santarelli L (2013) Drug discovery forautism spectrum disorder: challenges and opportunities. Nat Rev Drug Discov 12(10):777–790.

E7058 | www.pnas.org/cgi/doi/10.1073/pnas.1604992113 Patel et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

, 202

0

Correction

IMMUNOLOGY AND INFLAMMATIONCorrection for “Neurotensin stimulates sortilin and mTOR inhuman microglia inhibitable by methoxyluteolin, a potential thera-peutic target for autism,” by Arti B. Patel, Irene Tsilioni, Susan E.Leeman, and Theoharis C. Theoharides, which appears in issue 45,November 8, 2016, of Proc Natl Acad Sci USA (113:E7049–E7058;first published September 23, 2016; 10.1073/pnas.1604992113).The authors note that their conflict of interest statement was

omitted during publication. The authors declare the following:“T.C.T. has been awarded, under an agreement with Tufts Uni-versity, the following patents: US 9,050,275 - Methods of Screeningfor and Treating Autism Spectrum Disorders and Compositions forSame and US 9,176,146 - Methods of Treating Autism SpectrumDisorders and Compositions for Same. A Provisional Patent Ap-plication was also filed by Tufts University as US 62/396,546 -Compositions and Methods of Autism Treatment. T.C.T. is also theScientific Director of Algonot, LLC (Sarasota, FL) that has de-veloped the flavonoid-containing trademarked dietary supplementNeuroProtek, for which Tufts receives royalties. No portion of thework described in this paper was funded by Algonot.”

www.pnas.org/cgi/doi/10.1073/pnas.1616587113

E7138 | PNAS | November 8, 2016 | vol. 113 | no. 45 www.pnas.org