CORRECTION

Correction: TIMP-1 promotes hypermigration of Toxoplasma-infected primary dendritic cells via CD63–ITGB1–FAK signaling(doi:10.1242/jcs.225193)Einar B. Ólafsson, Emily C. Ross, Manuel Varas-Godoy and Antonio Barragan

There was an error in J. Cell Sci. (2019) 132, jcs225193.

The same GAPDH loading control blots were used in Fig. 4A,B without an explanation. The correct legend is shown below clarifying thatresults are from a single experiment.

Fig. 4. Phosphorylation of FAK and SRC upon challenge with Toxoplasma and inhibition of hypermotility by gene silencing of Ptk2(FAK) or by antagonism of FAK–SRC–PI3K. (A,B) Western blot analysis of total and phosphorylated protein expression of FAK (A)and SRC (B) in DCs unchallenged or challenged with T. gondii tachyzoites for 1 h. Graphs show total protein (FAK and SRC) andphosphorylated protein (p-397: FAK and p-416: SRC), relative to unchallenged DCs, normalized to GAPDH expression. ADU, arbitrarydensitometry unit. n=4 biological replicates from independent blots. Representative blots from one experiment are shown, with sameloading control (GAPDH) for total FAK/p-397 and total SRC/p-416, respectively.

This legend has been corrected in the online and pdf versions of the paper. The authors apologise for any inconvenience caused.

1

© 2019. Published by The Company of Biologists Ltd | Journal of Cell Science (2019) 132, jcs230920. doi:10.1242/jcs.230920

Journal

ofCe

llScience

RESEARCH ARTICLE

TIMP-1 promotes hypermigration of Toxoplasma-infected primarydendritic cells via CD63–ITGB1–FAK signalingEinar B. Ólafsson1, Emily C. Ross1, Manuel Varas-Godoy2 and Antonio Barragan1,*

ABSTRACTTissue inhibitor of metalloproteinases-1 (TIMP-1) exerts pleiotropiceffects on cells including conferring metastatic properties to cancercells. As for metastatic cells, recent paradigms of leukocyte migrationattribute important roles to the amoeboid migration mode of dendriticcells (DCs) for rapid locomotion in tissues. However, the role of TIMP-1in immune cell migration and in the context of infection has not beenaddressed. We report that, upon challenge with the obligateintracellular parasite Toxoplasma gondii, primary DCs secrete TIMP-1 with implications for their migratory properties. Using a short hairpinRNA (shRNA) gene silencing approach, we demonstrate that secretedTIMP-1 and its ligand CD63 are required for the onset of hypermotilityin DCs challenged with T. gondii. Further, gene silencing and antibodyblockade of the β1-integrin CD29 (ITGB1) inhibited DC hypermotility,indicating that signal transduction occurred via ITGB1. Finally, genesilencing of the ITGB1-associated focal adhesion kinase (FAK, alsoknown as PTK2), as well as pharmacological antagonism of FAK andassociated kinases SRC and PI3K, abrogated hypermotility. Thepresent study identifies a TIMP-1–CD63–ITGB1–FAK signaling axis inprimary DCs, which T. gondii hijacks to drive high-speed amoeboidmigration of the vehicle cells that facilitate its systemic dissemination.

KEY WORDS: Apicomplexa, Leukocyte motility, Amoeboidmigration, CD29, LAMP-3

INTRODUCTIONResident dendritic cells (DCs) in the intestinal lamina propria areamong the first leukocytes to encounter pathogens. The obligateintracellular parasite Toxoplasma gondii infects virtually all warm-blooded vertebrates, with an estimated 30% of humans chronicallyinfected (Pappas et al., 2009). From the point of entry in theintestinal tract, T. gondii rapidly achieves systemic disseminationand establishes chronic infection of the central nervous system.Reactivated and acute infection in immunosuppressed orimmunocompromised individuals can lead to lethal toxoplasmicencephalitis (Montoya and Liesenfeld, 2004).Upon T. gondii infection, DCs play a determinant role in

pathogenesis by driving protective T helper 1 (Th1) responsesthrough interleukin-12 (IL-12) production (Liu et al., 2006).Paradoxically, the T. gondii tachyzoite stage exploits the inherentmigratory ability of DCs for dissemination via a ‘Trojan horse’mechanism (Lambert et al., 2006, 2009b; Courret et al., 2006; Bierly

et al., 2008). Within minutes of active invasion by tachyzoites, DCsadopt a hypermigratory phenotype (reviewed in Weidner andBarragan, 2014), which mediates rapid systemic dissemination ofthe parasite in mice (Lambert et al., 2006; Kanatani et al., 2017). Thisdramatic migratory activation requires the discharge of parasiticsecretory organelles into the host cell cytoplasm (Weidner et al.,2013) and is triggered by GABAA receptor-mediated (Fuks et al.,2012) activation of voltage-dependent calcium channels (VDCC) onthe host DC plasma membrane (Kanatani et al., 2017). Tachyzoite-infected DCs undergo rapid cytoskeletal reorganization, expressenhanced locomotion on 2-dimensional (2D) surfaces (termedhypermotility) (Weidner et al., 2013), enhanced migration in 3Dcollagen matrix (Kanatani et al., 2015) and enhanced transmigrationin vitro (Lambert et al., 2006).

After encountering a foreign antigen, DCs undergo a complexmaturation process, encompassing upregulation of co-stimulatorymolecules and MHC class II. A shift in chemokine receptorexpression facilitates homing to secondary lymphoid organs (Ohlet al., 2004). Another hallmark of DC maturation is the transitionfrom mesenchymal to amoeboid (MAT) migration mode, whichenables high-speed locomotion through interstitial tissues and acrossbiological barriers (Alvarez et al., 2008; Lämmermann et al., 2008;Friedl and Wolf, 2003). In contrast to mesenchymal migration,amoeboid migration is not dependent on firm adhesion to anddegradation of extracellular matrix (ECM). Instead, the protrusiveflow of the actin cytoskeleton primarily drives locomotion(Lämmermann et al., 2008). It was recently reported that T. gondiitachyzoites induceMAT in infectedDCs but the signaling behind thisdramatic migratory activation remained unknown (Weidner et al.,2013; Kanatani et al., 2015; Olafsson et al., 2018).

The tissue inhibitors of matrix metalloproteinases (TIMPs)comprise four members (TIMP-1, -2, -3 and -4), which tightlyregulate the turnover of ECM proteins as inhibitors of matrixmetalloproteinases (MMPs) (reviewed in Visse and Nagase, 2003).However, numerous MMP-independent roles for the TIMPs havebeen identified, including pro-metastatic roles (reviewed in Stetler-Stevenson, 2008). Specifically, previous studies identified CD63 asa receptor for TIMP-1 in human breast epithelial cells (Jung et al.,2006). CD63, a ubiquitously expressed tetraspanin, is localized atthe plasma membrane and within the endosomal system. Like othertetraspanins, CD63 interacts with numerous proteins, includingintegrins (Radford et al., 1996). Upon binding TIMP-1, CD63activates the β1-subunit of integrins (ITGB1) (Jung et al., 2006; Leeet al., 2014). In normal and cancer cells, activated ITGB1 recruitsfocal adhesion kinase (FAK), which auto-phosphorylates Tyr-397and subsequently recruits SRC kinase (Ando et al., 2018) andphosphatidylinositol-4, 5-bisphosphate 3-kinase (PI3K) (Toricelliet al., 2013), leading to MAPK activation and cytoskeletalrearrangements (Jung et al., 2006; Ando et al., 2018). Mountingevidence suggests TIMP-1–CD63 signaling functions as a keyregulator of cancer cell survival (Jung et al., 2006), metastasisReceived 14 September 2018; Accepted 27 December 2018

1Department of Molecular Biosciences, The Wenner-Gren Institute, StockholmUniversity, 106 09 Stockholm, Sweden. 2Centro de Investigacion Biomedica,Faculty of Medicine, Universidad de los Andes, 7620001 Santiago, Chile.

*Author for correspondence (antonio.barragan@su.se)

A.B., 0000-0001-7746-9964

1

© 2019. Published by The Company of Biologists Ltd | Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

(Forte et al., 2017) and stem cell migration (Wilk et al., 2013;Lee et al., 2014). However, the role of TIMP-1–CD63 signaling inleukocyte biology remains enigmatic.Previously, we reported that autocrine TIMP-1 secretion upon

Toxoplasma challenge impedes the ability of DCs to degrade ECM(Olafsson et al., 2018). Here, in an infection model with thetachyzoite stage of T. gondii, we address the impact of TIMP-1 inthe migratory activation of DCs and define a TIMP-1–CD63–ITGB1–FAK signaling axis in primary DCs.

RESULTSToxoplasma-challenged DCs secrete TIMP-1, and Timp-1gene silencing inhibits DC hypermotilityWe recently reported that in Toxoplasma-challenged DCs, autocrineTIMP-1 secretion reduced degradation of the ECM (Olafsson et al.,2018). Because TIMP-1 has also been associated with the invasiveproperties of metastatic cancer cells, we investigated the role ofTIMP-1 in the hypermigratory phenotype exhibited by tachyzoite-infected DCs derived from C57BL/6 mice. Shortly after challengewith Toxoplasma, DCs upregulated Timp-1 mRNA expression(Fig. 1A) and secreted TIMP-1 protein (Fig. 1B). To investigatewhether secreted TIMP-1 impacted on Toxoplasma-induced DChypermotility, we applied a shRNA gene silencing approach. First,the transduction and knockdown efficiency of lentivirus targetingTimp-1mRNA (shTIMP-1) and control virus (shLuc) was quantified.ShTIMP-1-transduced DCs exhibited significantly reduced levels ofTimp-1 mRNA expression (Fig. 1C) and secreted TIMP-1 protein(Fig. 1D). Next, the motility of Toxoplasma-challenged transducedDCs was analyzed (Fig. 1E,F). Importantly, transduction withshTIMP-1 significantly inhibited the motility and mean velocities ofToxoplasma-infected DCs, while the baseline motility and meanvelocities of non-challenged DCs were non-significantly affected byshTIMP-1 transduction (Fig. 1F,G). Further, recombinant mouseTIMP-1 (rmTIMP-1) partially restored hypermotility in Toxoplasma-challenged TIMP-1-silenced DCs (Fig. 1F,H). In contrast, treatmentwith rmTIMP-1 impacted non-significantly on the motility of mock-treated DCs or shLuc-transduced unchallenged (Fig. S1A,B) andToxoplasma-challenged DCs (Fig. S1A,C). The finding that TIMP-1modulates the hypermigratory phenotype of Toxoplasma-infectedDCs prompted us to investigate the signaling pathways that mediatethis infection-related enhanced migration of DCs.

Silencing of the Cd63 gene suppresses DC hypermotilityBecause CD63 is the only described non-MMP binding partner forTIMP-1 (Jung et al., 2006) and the CD63–TIMP-1 complex hasbeen associated with cytoskeletal rearrangements (Lee et al., 2014),we investigated the putative implication of CD63 in DChypermotility. First, we assessed Cd63 mRNA and CD63 proteinexpression in unchallenged and Toxoplasma-infected DCs. Cd63mRNA expression was slightly downregulated upon infection(Fig. S2A) and CD63 antibodies stained unchallenged andToxoplasma-infected DCs (Fig. 2A,B). Flow cytometry analysesdetected non-significant differences in total and plasma membraneCD63 protein expression between unchallenged DCs, Toxoplasma-infected DCs and bystander DCs (Fig. 2C; Fig. S2B,C). Next, wesilenced Cd63 mRNA expression to investigate its role inhypermotility. DCs transduced with shCD63 (Fig. S2D) exhibitedsignificantly reducedCd63mRNA (Fig. 2D) and protein expression(Fig. 2E; Fig. S2E). Consonant with motility data from Timp-1-silenced DCs (Fig. 1), gene silencing of Cd63 significantly reducedthe motility and the mean velocities of infected DCs, which reachedvelocities non-significantly different from unchallenged DCs

(Fig. 2F,G). Taken together, these data situate CD63 as acomponent in the signaling cascade that facilitates hypermotilityand support the involvement of TIMP-1–CD63 signaling in themigratory activation of Toxoplasma-infected DCs.

ITGB1 signaling is implicated in DC hypermotilityBecause the TIMP-1–CD63 complex interacts with and activatesITGB1 integrin signaling (Jung et al., 2006; Lee et al., 2014), weexamined ITGB1 signaling in hypermotility. Antibody stainingindicated ITGB1 protein expression in both unchallenged DCs,Toxoplasma-infected DCs and bystander DCs (Fig. 3A,B), withoverall similar expression levels to those determined using flowcytometry (Fig. 3C; Fig. S3A). Similar results were obtained whenITGB1 levels were examined in cells gated for CD63 expression(Fig. S3B). To assess the role of ITGB1 signaling in Toxoplasma-induced hypermotility, we blocked ITGB1 activation with twoseparate monoclonal antibodies (Sangaletti et al., 2008). First, wetested the ITGB1-blocking efficiency of the antibodies byexamining their impact on DC adhesion to fibronectin (theprincipal ECM ligand of ITGB1). DC binding to fibronectinwas significantly reduced in the presence of ITGB1-blockingantibodies, while non-significant effects were exerted by the isotypecontrol (Fig. S3C,D). Next, we investigated the impact of ITGB1-blockade on Toxoplasma-induced DC hypermotility. A significantdecrease in the mean velocity of Toxoplasma-infected DCs wasobserved upon ITGB1 blockade, while non-significant effects wererecorded for baseline motility or by treatment with isotype control(Fig. 3D,E). Finally, Itgb1 mRNA expression was silenced, with asignificant reduction in ITGB1 mRNA (Fig. 3F) and proteinexpression (Fig. 3G; Fig. S3E) in shITGB1-transduced cells. Themotility and mean velocities of Toxoplasma-infected DCs transducedwith shITGB1 were significantly inhibited, while the baselinemotility and mean velocities of non-challenged DCs were non-significantly affected by shITGB1 treatment (Fig. 3H,I; Fig. S3F,H).Because ITGB1 binds to fibronectin, we additionally assessed themotility of shITGB1-transduced DCs in presence of fibronectin(Fig. S3G,H). Consonant with data in collagen, the motility andmeanvelocities of shITGB1-transduced Toxoplasma-infected DCs weresignificantly inhibited (Fig. S3I). Taken together, these data support arole for ITGB1 signaling in Toxoplasma-induced hypermotility.Further, as ITGB1 activation is necessary for TIMP-1–CD63 signaltransduction, the data reinforce the involvement of TIMP-1–CD63signaling in Toxoplasma-induced DC hypermotility.

FAK (Ptk2) silencing impedes DC hypermotilityUpon ITGB1 activation, FAK auto-phosphorylates Tyr-397 andrecruits the kinases SRC and PI3K (Ando et al., 2018; Toricelliet al., 2013). Because FAK is a central node in the ITGB1signaling cascade, we hypothesized that FAK, and substratekinases SRC and PI3K, impacted on hypermotility. First, weassessed total protein expression and phosphorylation of FAK andSRC in DCs challenged with Toxoplasma. We detected non-significant differences in total protein expression of both kinaseswhen compared with uninfected cells, while the relativephosphorylated protein forms were significantly elevated shortlyafter challenge with Toxoplasma (Fig. 4A,B). Next, we assessedthe impact of pharmacological antagonism of FAK, SRC, PI3Kand phosphatase and tensin homolog (PTEN, a PI3K substratephosphatase) on hypermotility (Fig. S4B). First, non-significanteffects were observed on Toxoplasma invasion and replication inthe presence of inhibitors (Fig. S4A). Antagonism of PTEN hadnon-significant effects on DC hypermotility. In sharp contrast,

2

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

antagonism of FAK, SRC and PI3K phosphorylation significantlyinhibited the hypermotility of Toxoplasma-infected DCs, withnon-significant effects on the baseline motility of unchallengedDCs (Fig. 4C,D).As gene silencing confers minimal off-target effects compared with

pharmacological inhibition, we investigated the motility of DCs

transduced with lentivirus targeting Ptk2mRNA (shFAK) (Fig. S4C).Ptk2mRNA and FAK protein expression were significantly reduced inshFAK-transduced DCs (Fig. 4E,F). Importantly, motility and meanvelocities were significantly reduced in shFAK-transducedToxoplasma-infected DCs, while non-significant effects wereobserved on the baseline motility of unchallenged DCs (Fig. 4G,H).

Fig. 1. TIMP-1 secreted by Toxoplasma-infected DCs is required for DC hypermotility. (A) qPCR analysis of Timp-1mRNA fromDCs challenged with freshlyegressed T. gondii tachyzoites for 1, 2 and 4 h, related to unchallenged DCs in completemedium (CM) (n=3 independent experiments). (B) TIMP-1 polypeptide insupernatants from DCs incubated with and without T. gondii tachyzoites for 1, 2 and 4 h, quantified using ELISA (n=3 independent experiments). (C) Timp-1mRNA expression (2−ΔCt) in control shLuc- and shTIMP-1-transduced DCs, relative to mock-treated (n=3 independent experiments). (D) TIMP-1 protein insupernatants from T. gondii-challenged DCs transduced with control shLuc or shTIMP-1 lentivirus relative to mock-treated DCs, quantified by ELISA (n=3independent experiments). (E) Representative micrographs of unchallenged DCs and DC challenged with RFP-expressing T. gondii tachyzoites. Cells weremock-treated, transduced with lentiviral vectors (expressing eGFP) targeting a non-expressed target (shLuc) or Timp-1 mRNA (shTIMP-1), with and withouttreatment with rmTIMP-1. Arrowheads indicate T. gondii-infected cells expressing an eGFP-reporter. Scale bars: 200 µm. Micrographs are representative of 3independent experiments. (F) Representative motility plots of DC as described and represented in E (n=3 independent experiments). (G) Mean velocity of mock-treated, shLuc- and shTIMP-1-transduced DCs, with and without T. gondii challenge, performed as in E (n=3 independent experiments). (H) Mean velocity ofT. gondii-infected mock-treated DCs, and shTIMP-1-transduced DCs with and without T. gondii challenge and rmTIMP-1 treatment, performed as in E (n=3independent experiments). Data are presented as mean±s.e.m. *P<0.05; ns, non-significant by Student’s t-test (C,D) or repeated measures one-way ANOVA,Tukey’s HSD post-hoc test (G,H).

3

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

Taken together, these data demonstrate that FAK signaling plays adeterminant role in Toxoplasma-induced DC hypermotility. BecauseFAK is phosphorylated upon challenge with Toxoplasma and is

necessary for ITGB1 signal transduction, these data support theactivation of TIMP-1–CD63 signaling via ITGB1 in DCs uponToxoplasma infection.

Fig. 2. See next page for legend.

4

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

DISCUSSIONBecause T. gondii tachyzoites utilize infected DCs fordissemination, and TIMP-1 is highly upregulated in parasitizedDCs, we investigated putative roles for TIMP-1 in DC migration.We report a novel role of TIMP-1 as a signaling molecule for theinduction of amoeboid high-speed migration in infected DCs andextend the TIMP-1–CD63–ITGB1–FAK signaling axis from cancercell migration to leukocyte migration.We demonstrate that silencing of the genes encoding TIMP-1 and

its receptor CD63 inhibit the hypermigratory phenotype exhibitedby parasitized DCs. Further, we show that inhibition of TIMP-1 and/or CD63 downstream signaling elements (ITGB1, FAK, SRC andPI3K) inhibits hypermotility. CD63 is the only known non-proteaseligand for TIMP-1, with an impact on intracellular signaling (Junget al., 2006). Additionally, a ligand–receptor interaction linked toapoptosis has been described for TIMP-1 and pro-MMP-9–CD44(Lambert et al., 2009a). However, we previously showed thatMmp9mRNA expression is strongly downregulated in Toxoplasma-challenged DCs (Olafsson et al., 2018) and it is therefore unlikelythat this signaling significantly impacts on hypermotility. Further,several ligand–receptor functions have been described for TIMP-2,-3 and -4 (Stetler-Stevenson, 2008). However, in sharp contrast tothe elevated mRNA expression and protein secretion of TIMP-1, wepreviously reported downregulation of Timp-2, -3 and -4 mRNA inDCs upon challenge with T. gondii (Olafsson et al., 2018). Takentogether, our data outline a primary role for the TIMP-1–CD63signaling axis in the motility of Toxoplasma-infected DCs.Toxoplasma-induced DC hypermigration is initiated by autocrine

γ-aminobutyric acid (GABA) secretion from infected DCs (Fukset al., 2012), which activates ionotropic GABAA receptor channels,triggering VDCC activation and the influx of extracellular Ca2+

(Kanatani et al., 2017). Ca2+ acts as a ubiquitous secondarymessenger and governs TIMP-1 exocytosis (Dranoff et al., 2013).Thus, oscillating Ca2+ in Toxoplasma-infected DCs (Kanatani et al.,

2017) plausibly stimulates the secretion of TIMP-1. Because DCsdisplay the hypermigratory phenotype throughout infection(Lambert et al., 2006) and TIMP-1 secretion by infected DCs ismaintained up to 24 h post infection (Olafsson et al., 2018), TIMP-1is a prime candidate for a feedback loop that maintainshypermotility of infected DCs over time. Yet, the mechanisms bywhich T. gondii upregulates Timp-1 gene expression of the infectedhost cell await further investigation.

Upon T. gondii infection, dramatic morphological changes takeplace in DCs, with rounding-up (amoeboid) morphology, loss ofpodosomes, decreased adhesion with redistribution of integrins, andhigh-velocity locomotion on 2D and in 3D confinements (Weidneret al., 2013; Kanatani et al., 2015; Olafsson et al., 2018). All this isconsistent with the cell undergoing a mesenchymal to amoeboidtransition (MAT), reminiscent of that described in cancer cells andrapidly moving T lymphocytes (Friedl and Wolf, 2010). Further,amoeboid cell motility is characterized by non-proteolytic migrationand reduced dependence on integrin-ECM binding (Lämmermannet al., 2008), which is consistent with the reduced adhesion ofToxoplasma-infected DCs to ECM components (Weidner et al.,2013; Kanatani et al., 2015). Jointly, because an inhibitory effect onhypermotility was observed upon antibody blockade or genesilencing of Itgb1, this indicates that hypermotility-related ITGB1signaling is likely more dependent on CD63 activation than ontactile adhesion-dependent interactions with ECM components.Additionally, a dysregulation of focal adhesion-related integrinsignaling was reported in Toxoplasma-infected monocytic cells(Cook et al., 2018). With the dual effects of TIMP-1 in mind,namely inhibition of MMP-dependent ECM remodeling andpromoting locomotion through CD63–ITGB1 signaling, theelevated TIMP-1 secretion by Toxoplasma-infected DCs iscompatible with the amoeboid migration mode of leukocytes(Lämmermann et al., 2008). In this context, our data show thatT. gondii infection induces amoeboid hypermotility in DCs, withdependence on secreted TIMP-1. Conversely, treatment withrmTIMP-1 significantly rescued hypermotility in TIMP-1-silenced infected DCs but was insufficient to inducehypermotility in naïve DCs. This may indicate that additionalearly activation of the DC occurs upon challenge with Toxoplasma(Kanatani et al., 2017), which is also in line with the redistributionand altered post-translational modifications of CD63 upon DCmaturation (Engering et al., 2003) and the redistribution of integrinsin Toxoplasma-infected DCs (Weidner et al., 2013). Nonetheless, inthe absence of TIMP-1 or CD63–ITGB1–FAK signaling, amoeboidhigh-speed migration did not take place, situating TIMP-1 as aprime mediator of hypermotility in Toxoplasma-infected DCs.

We demonstrate that gene silencing of Ptk2 and pharmacologicalantagonism of FAK inhibit DC hypermotility. Additionally, FAKand SRC phosphorylation was rapidly amplified in Toxoplasma-infected DCs. Upon integrin activation, FAK auto-phosphorylatesTyr-397, then recruits and phosphorylates SRC and PI3K, leadingto reorganization of the actin cytoskeleton (Xing et al., 1994; Chenet al., 1996). Because FAK is a central node in bridging signalingcues to the actin cytoskeleton via integrins (Guan, 1997), it likelyrepresents a key step in the cytoskeletal changes and migratoryactivation of Toxoplasma-infected DCs. Importantly, the rapidonset of hypermotility is well in line with FAK and SRCphosphorylation shortly after parasite invasion and likely precedestranscriptional regulation of hypermotility (Weidner et al., 2013).

We report that the TIMP-1–CD63–ITGB1–FAK signaling axiscan be activated in primary DCs with an impact on their migratoryproperties. To our knowledge, the data also provide the first

Fig. 2. CD63 is present on the plasmamembrane of Toxoplasma-infectedDCs and gene silencing of Cd63 impedes DC hypermotility.(A) Micrographs of DCs with and without T. gondii challenge stained forT. gondii surface antigen-1 (SAG-1, blue), CD63 (red) and F-actin (green) 4 hpost-infection. Scale bars: 20 µm. (B) Z-stack maximum-intensity projection(MIP) of T. gondii-infected DCs stained for SAG-1 (blue), CD63 (red) andF-actin (green) 4 h post-infection. Scale bars: 10 µm. Micrographs in A,B arerepresentative of 3 independent experiments. (C) Flow cytometry analysis oftotal and plasma membrane (PM) CD63 protein expression in unchallenged(green) and T. gondii-infected (red) DCs (as identified by expression of DCsurface marker CD11c). Histograms show gates used to define CD11c+ cells(left), GFP+ cells (center) and CD63+ cells (right). Dashed and continuous linesshow total and plasma membrane CD63 expression, respectively. Black lineand grey fill represent unstained control and isotype control, respectively.Graphs show the percentage of CD63+ cells and CD63 median fluorescenceintensity (MFI) for each population (n=6 independent experiments). (D) Cd63mRNA expression in control shLuc- and shCD63-transduced DCs, relative tomock-treated (n=4 independent experiments). (E) CD63 protein expression ofDCs (CD11c+) mock-treated transduced with control shLuc or shCD63lentivirus. Gating performed as in C. Histogram shows gate used to defineCD63+ cells. Grey fill represents isotype control. Graphs show the percentageof CD63+ cells and CD63 MFI for each population (n=5 independentexperiments). (F) Representative motility plots of mock-treated DCs andmRuby-expressing DCs transduced with lentiviral vectors targeting Cd63mRNA (shCD63) or a non-expressed target (shLuc), unchallenged orchallenged with GFP-expressing T. gondii tachyzoites. Data are representativeof 3 independent experiments. (G) Mean velocity of mock-treated, shLuc-and shCD63-transduced DCs unchallenged or challenged with T. gondii,performed as in F (n=3 independent experiments). Data are presented asmean±s.e.m.). *P<0.05; ns, non-significant by Student’s t-test (C,D,E) orrepeated measures one-way ANOVA, Tukey’s HSD post-hoc test (G).

5

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

evidence of an intracellular pathogen manipulating this signalingaxis to promote its dissemination. Based on the data at hand, wepropose a model for the migratory activation (Fig. 5): T. gondiielicits TIMP-1 secretion by parasitized DCs, with TIMP-1 bindingto CD63 then triggering ITGB1–FAK signaling, which promotesthe protrusive flow of actin and amoeboid motility.In line with the paradox of matrix proteolysis in infectious

diseases (Elkington et al., 2005), TIMP-1 dysregulation may haveimplications for immune cell responses and inflammation during

toxoplasmosis. From the perspective of the host, under conditions ofT. gondii infection, TIMP-1 upregulation may confer enhanceddissemination in shuttling DCs, and also reduce tissue pathology,which may be especially advantageous in dampening the effects ofencephalitic infection. In Toxoplasma-infected mice, systemicTIMP-1 is elevated (Tomasik et al., 2016) and reduced parasiticloads in the central nervous system have been reported in TIMP-1-deficient mice (Clark et al., 2011). Further, elevated serum TIMP-1is associated with severe Plasmodium falciparum malaria

Fig. 3. See next page for legend.

6

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

(Dietmann et al., 2008), and elevated TIMP-1 in cerebrospinal fluidis associated with viral and bacterial meningitis (Kolb et al., 1998;Tsai et al., 2011). Because MMPs can have a role in immunity butcan also contribute to immunopathology in bacterial, viral andparasitic infections (Elkington et al., 2005), our data highlightpossible pathophysiological effects of TIMP-1: pro-disseminatoryeffects by activating shuttling leukocytes, while reducing tissueproteolysis (Olafsson et al., 2018) and thereby dampeninginflammation systemically or locally in the perivascularmicroenvironment. Along these lines, TIMP-1 upregulation isconsistently associated with cancer progression and pro-metastaticeffects (Jackson et al., 2017), mainly, but not exclusively, mediatedby TIMP-1–CD63–ITGB1 activation of FAK signaling (Grunwaldet al., 2016). However, this pathway has remained unexplored inimmune cells and in the context of infection. This work alsohighlights possible conserved signaling mechanisms betweenmetastasizing cells and immune cells, to be taken intoconsideration when designing cancer therapies that targetTIMP-1–CD63–ITGB1–FAK signaling.

MATERIALS AND METHODSEthics statementThe Regional Animal Research Ethical Board, Stockholm, Sweden,approved experimental procedures and protocols involving extraction ofcells from mice (N135/15, N78/16), following proceedings described in EUlegislation (Council Directive 2010/63/EU).

Cells and parasitesMurine bone marrow-derived DCs (DCs) were generated as previouslydescribed (Fuks et al., 2012). Briefly, cells from bone marrow of 6–10-week-old male or female C57BL/6 mice (Charles River) were cultivated in

RPMI 1640 with 10% fetal bovine serum (FBS), gentamicin (20 μg/ml),glutamine (2 mM) and HEPES (0.01 M), referred to as complete medium(CM; all reagents from Life Technologies), and supplemented with 10 ng/ml recombinant mouse GM-CSF (Peprotech). Loosely adherent cells wereharvested after 6 or 10 days of maturation. Toxoplasma gondii tachyzoites ofthe wild-type and RFP-expressing Prugniaud strain (Pru and Pru-RFP, typeII) (Pepper et al., 2008) or GFP-expressing Ptg strain (Ptg-GFP, type II)(Kim et al., 2001; Hitziger et al., 2005) lines were maintained by serial 2-daypassages in human foreskin fibroblast (HFF-1 SCRC-1041, American TypeCulture Collection) monolayers. The hypermigratory phenotype induced inDCs by type II strains has been previously characterized (Lambert et al.,2009b).

ReagentsThe following soluble reagents were used in functional assays; rmTIMP-1(1 µg/ml, R&D systems), LEAF-purified anti-CD29 (clone HMβ1-1,102209, BioLegend), LEAF-purified IgG isotype control (clone HTK888,400969, BioLegend), purified NA/LE anti-CD29 (clone Ha2/5, 555002,BD Pharmingen), PF-573228 (PF-228), Wortmanin, SrcI1, SF-1670. Allinhibitors were acquired from Tocris and used at 1 µM. Blocking antibodiesand isotype control were used at 10 µg/ml.

Motility assayMotility assays were performed as previously described (Weidner et al., 2013)with slight modification as in Olafsson et al. (2018). Briefly, DCs werecultured in 96-well plates in CM with and without freshly egressed T. gondiitachyzoites (Pru-RFP or Ptg-GFP, MOI 3, 4 h) and soluble reagents (asindicated). Bovine collagen I (1 mg/ml, Life Technologies) or fibronectin(100 µg/ml, Life Technologies), as indicated, were then added and live cellimaging was performed for 1 h, 1 frame/min, at 10× magnification (Z1Observer with Zen 2 Blue v. 4.0.3, Zeiss). Time-lapse images wereconsolidated into stacks and motility data was obtained from 30 cells/condition (Manual Tracking, ImageJ) yielding mean velocities (Chemotaxisandmigration tool v2.0, Ibidi). Infected cells were defined byRFPorGFP cellco-localization. Transduced cells were defined by eGFP, tGFP or mRubyreporter expression.

Replication assayT. gondii tachyzoite (Pru-RFP, MOI 1) replication in DCs was assessed 24 hpost-infection by vacuole counts as previously described (Dellacasa-Lindberg et al., 2011) and analyzed using immunocytochemistry (100vacuoles were quantified per condition).

Adhesion assayLab-tek chambers (VWR) were coated with 5 μg/cm2 fibronectin (Gibco) for2 h at 37°C. Wells were washed twice with RPMI (Thermo Fisher scientific)and the cell suspension was allowed to attach at 37°C, 5% CO2 for 15 min.Unbound cells were removed by washing twice with RPMI and once withPBS.Attached cells were fixed (4%PFA), stained (DAPI) and 5 fields of viewper condition were imaged with a 10× objective on a Z1 Observer (Zeiss)microscope with Zen 2 Blue software (v. 4.0.3, Zeiss). DAPI stained nucleiwere quantified with Cellprofiler v2.1.1(Broad Institute).

Polymerase chain reaction (PCR)DCs were cultured in CM with and without freshly egressed T. gondiitachyzoites (Pru-RFP, MOI 3) for 1, 2, 3 or 4 h. Total RNA was extractedusing TRIzol reagent (Ambion). First-strand cDNA was synthesized withSuperscript III Reverse Transcriptase (Invitrogen). Real-time quantitativePCR (qPCR) was performed using SYBR Green PCR master mix (Kapabiosystems), forward and reverse primers (200 nM) and cDNA (100 ng)with a Rotor Gene 6000 system (Corbett). qPCR was run for 45 cycles.Products were analyzed with RG-6000 application software (v1.7, Corbett).Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and β-actin (ACTB)were used as housekeeping genes to generate ΔCt values. 2−ΔCt values wereused to calculate relative knockdown efficiency and 2−ΔΔCt values were usedto calculate expression fold-change upon infection. All primers (Invitrogen)were designed using Get-prime or Primer-BLAST software (Table S1).

Fig. 3. Expression of ITGB1 in Toxoplasma-infected DCs anddependence of hypermotility on ITGB1. (A) Micrographs of DCs with andwithout T. gondii challenge stained for SAG-1 (blue), ITGB1 (red) and F-actin(green) 4 h post-infection. Scale bars: 20 µm. (B) Z-stack MIP of T. gondii-infected DCs stained for SAG-1 (blue), ITGB1 (red) and F-actin (green) 4 hpost-infection. Micrographs in A,B are representative of 3 independentexperiments. Scale bars: 10 µm. (C) Flow cytometry analysis of total andplasma membrane (PM) ITGB1 protein expression in unchallenged (green)and T. gondii-infected (red) DCs (CD11c+). Histogram shows gates used todefine ITGB1+ cells. CD11c+ cells and GFP+ cells were gated as in Fig. 2C.Dashed and continuous lines show total and plasma membrane ITGB1expression, respectively. Grey fill represents isotype control. Graphs showthe percentage of ITGB1+ cells and ITGB1 MFI for each population (n=3independent experiments). (D) Representative motility plots of DCsunchallenged or challenged with RFP-expressing T. gondii tachyzoites, anduntreated (CM) or treated with ITGB1-blocking antibodies (anti-CD29 clonesHa2/5 or HMβ1-1) or isotype control (clone HTK888). Data are representativeof 3 independent experiments. (E) Mean velocity of DCs treated with Ha2/5,HMβ1-1 or HTK888, with and without T. gondii challenge, performed as in D(n=3 independent experiments). (F) Itgb1 mRNA expression in control shLuc-and shITGB1-transduced DCs, relative to mock-treated (n=3 independentexperiments). (G) ITGB1 protein expression of DCs (CD11c+) mock-treated(black) and transduced with control shLuc (green) or shITGB1 (red) lentivirus.Gating performed as in C. Histogram shows gate used to define ITGB1+ cells.Grey fill represents isotype control. Graphs show the percentage of ITGB1+

cells and ITGB1 MFI for each population (n=3 independent experiments).(H) Representative motility plots of mock-treated DCs and eGFP-expressingDCs transduced with lentiviral vectors targeting Itgb1 mRNA (shITGB1)or a non-expressed target (shLuc), unchallenged or challenged with RFP-expressing T. gondii tachyzoites. Data are representative of 3 independentexperiments. (I) Mean velocity of mock-treated, shLuc- and shITGB1-transduced DCs unchallenged or challenged with T. gondii, performed as in H(n=3 independent experiments). Data are presented as mean±s.e.m. *P<0.05;ns, non-significant by Student’s t-test (C,F,G) or repeated measures one-wayANOVA, Tukey’s HSD post-hoc test (E,I).

7

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

TIMP-1 ELISASupernatants from DCs cultured in CM with and without freshly egressedT. gondii tachyzoites (Pru-RFP, MOI 3) were collected 1, 2, 3 or 4 h postinfection. Supernatants from mock-treated and shRNA-transduced DCschallenged with freshly egressed T. gondii tachyzoites (Pru-RFP, MOI 3)were collected 5 days post transduction and 4 h post infection. The sampleswere analyzed using ELISA (Mouse TIMP-1 Quantikine ELISA Kit, R&DSystems) according to the manufacturer’s instructions. Results werenormalized to a CM control.

Flow cytometryDCs were cultured in CM with and without freshly egressed T. gondiitachyzoites (Ptg-GFP, MOI 3, 4 h). Following Fc receptor blockade (1:100;clone 2.4G2, 101301, BD Pharmingen), cells were stained with anti-CD11c–PE–Cy7 (clone N418, 25-0114-82, eBiosciences). After fixation (PFA 4%)cells were directly stained (plasma membrane CD63 and ITGB1 expression)with anti-CD63–APC (clone NVG-2, 17-0631-82, eBiosciences) and/or anti-ITGB1–Pacific Blue (clone HMB1-1, 102224, BioLegend), rat isotypecontrol–APC (clone eBR2a, 12-4321-42, eBiosciences) or hamster isotype

Fig. 4. Phosphorylation of FAK and SRC upon challenge with Toxoplasma and inhibition of hypermotility by gene silencing of Ptk2 (FAK) or byantagonism of FAK–SRC–PI3K. (A,B) Western blot analysis of total and phosphorylated protein expression of FAK (A) and SRC (B) in DCs unchallenged orchallenged with T. gondii tachyzoites for 1 h. Graphs show total protein (FAK and SRC) and phosphorylated protein (p-397: FAK and p-416: SRC), relativeto unchallenged DCs, normalized to GAPDH expression. ADU, arbitrary densitometry unit. n=4 biological replicates from independent blots. Representativeblots from one experiment are shown, with same loading control (GAPDH) for total FAK/p-397 and total SRC/p-416, respectively. (C) Representativemotility plots of DCs unchallenged or challenged with RFP-expressing T. gondii tachyzoites, untreated (CM) or treated with inhibitors (1 µM) as indicated(PF-573228: FAK; SrcI1: SRC; Wortmanin: PI3K; SF1670: PTEN). Data are representative of 3 independent experiments. (D) Mean velocity of DCs treated withPF-573228, SrcI1, Wortmanin or SF1670, with and without T. gondii challenge, performed as in C (n=3 independent experiments). (E) Ptk2 mRNA expressionin DCs transduced with shFAK or control shLuc lentivirus, relative to mock-treated (n=3 independent experiments). (F) FAK protein expression in DCstransduced with control shLuc or shFAK lentivirus, relative to mock-treated DCs, normalized to GAPDH expression. (n=4 biological replicates from independentblots). (G) Representative motility plots of mock-treated DCs and DCs transduced with control shLuc or shFAK lentivirus, unchallenged or challenged withRFP-expressing T. gondii tachyzoites. Data are representative of 3 independent experiments. (H) Mean velocity of mock-treated, shLuc- and shFAK-transducedDCs with and without T. gondii challenge, performed as in G (n=3 independent experiments). Data are presented as mean±s.e.m. *P<0.05; ns, non-significant byStudent’s t-test (A,B,E,F) or repeated measures one-way ANOVA, Tukey’s HSD post-hoc test (D,H).

8

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

control–Pacific Blue (clone HTK888, 400925, BioLegend). Alternatively, cellswere permeabilized (IntraPrep Permeabilization kit, Beckman Coulter) beforeCD63 or ITGB1 staining (total CD63 and ITGB1 expression). All antibodieswere used at a dilution of 1:100. Samples were run on the LSRFortessa CellAnalyzer (BD Biosciences). Data was analyzed in FlowJo (Tree Star).

ImmunocytochemistryDCs were plated on poly-L-lysine (Sigma-Aldrich)-coated glass coverslips andchallenged with freshly egressed T. gondii tachyzoites (Pru, MOI 2, 4 h). Afterfixation (4% PFA, Sigma-Aldrich) and blockade (5% BSA, Sigma-Aldrich)cells were incubated with anti-CD63 (clone 446703, MAB5417, R&DSystems) or anti-ITGB1 (MAB1997,Millipore). Cells were then permeabilized(0.2% Triton X-100, Sigma-Aldrich) and stained with phalloidin Alexa Fluor488 (Invitrogen) and anti-SAG-1 (clone P30/3, MA1-83499, Invitrogen).Primary antibodieswere used at 1:1000. Anti-mouseAlexa Fluor 350 (A21049,Invitrogen) and anti-rat Alexa Fluor 594 (A21471, Invitrogen) were used assecondary antibodies at 1:1000. Coverslips were mounted with Lab VisionPermaFluo Aqueous Mounting Medium (Thermo Scientific). Micrographswere generated using a 63× objective on a laser scanning confocal microscope(LSM 800, Zeiss). Maximum intensity projections (MIP) were produced inImageJ from 5 µm thick z-stacks generated using a 63× objective of a laserscanning confocal microscope (LSM 800, Zeiss).

Western blotDCs were cultured in CM with and without freshly egressed T. gondiitachyzoites (Pru-RFP, MOI 3). Cells were homogenized in RIPA buffer

(150 mM NaCl, 50 mM Tris, 0.1% Triton, 0.5% deoxycholic acid, 0.1%SDS) with cOmplete mini protease and phosphatase inhibitors (Roche).Proteins were separated using 8% SDS-PAGE (Life Technologies), blottedonto a PVDFmembrane (Millipore) and blocked (5% BSA, Sigma-Aldrich)followed by incubation with anti-FAK (3285, Cell Signaling Technology),anti-Tyr-397-FAK (3283, Cell Signaling Technology), anti-SRC (2108,Cell Signaling Technology), anti-Tyr-416-SRC (clone D49G4, 6943, CellSignaling Technology) or anti-GAPDH (ABS16, Millipore) followed byanti-rabbit or anti-mouse HRP (7074 and 7076, Cell Signaling Technology).All antibodies were used at 1:1000, except anti-GAPDH (1:3000). Proteinswere revealed by mean of enhanced chemiluminescence (GE Healthcare) ina BioRad ChemiDoc XRS+. Densitometry analysis was performed usingImageJ (NIH, MD, USA).

Lentiviral vector production and in vitro transductionSelf-complementary hairpin DNA oligos targeting Timp-1 (shTIMP-1) andCd63 (shCD63) mRNA and a non-related sequence (luciferase, Luc) werechemically synthesized (DNA Technology, Denmark), aligned and ligatedin a self-inactivating lentiviral vector (pLL3.7, 11795, Addgene) containinga CMV-driven eGFP (shTIMP-1, shLuc) or mRuby (shCD63, shLuc)reporter and a U6 promoter upstream of the cloning restriction sites (HpaIand XhoI) (Table S2). Restriction enzyme analysis and directDNA sequencing confirmed the correct insertion of short hairpin RNA(shRNA) sequences. shRNA targeting Ptk2 (shFAK, TRCN0000023485,Sigma-Aldrich) or Itgb1 (shITGB1, TRCN0000066645, Genscript) mRNAwere on self-inactivating lentiviral vectors (pLKO.1 and pLL3.7,

Fig. 5. Proposed model of the migratory activation of Toxoplasma-infected DCs through the TIMP-1–CD63–ITGB1–FAK signaling axis. (I) T. gondii (Tg)actively invades DCs and resides in a non-fusogenic parasitophorous vacuole (PV). (II) Shortly after parasite invasion, T. gondii-infected DCs upregulate Timp-1mRNA expression (but not Timp-2, -3 or -4 mRNA expression) (Olafsson et al., 2018). (III) T. gondii-infected DCs secrete TIMP-1 protein, which reducespericellular proteolysis of extracellular matrix (ECM) via inhibition of secreted metalloproteinases (MMP) and membrane-bound MMPs (MT-MMP or ADAM)(Olafsson et al., 2018). (IV) Secreted TIMP-1 binds to and activates CD63 (Jung et al., 2006), which in turn interacts with ITGB1 at the plasma membrane (Junget al., 2006). Gene silencing of Timp-1 (shTIMP-1), Cd63 (shCD63) and Itgb1 (shITGB1) and Ab blockade of ITGB1 (Ha2/5 and HMβ1-1) inhibits hypermotilitywhile recombinant mouse TIMP-1 (rmTIMP-1) rescues hypermotility of shTIMP-1-silenced T. gondii-infected DCs. (V) Activated ITGB1 induces FAK Tyr-397auto-phosphorylation, which in turn activates SRC and PI3K signaling (Ando et al., 2018; Toricelli et al., 2013). FAK and SRC are phosphorylated shortly afterchallenge with T. gondii. Gene silencing of Ptk2 (shFAK) and pharmacological antagonism of FAK (PF-573228), SRC (SrcI1) or PI3K (Wortmanin) inhibitshypermotility. (VI-VII) ITGB1–FAK signaling via SRC and PI3K induces cytoskeletal rearrangements (Weidner et al., 2013) and shifts the cell into a hypermotilestate that facilitates dissemination. Continuous and dotted barred lines indicate direct and indirect negative regulation, respectively. Dashed arrows indicate thechronology of events. ’C’ tethered to TIMP-1 indicates the C-terminus of the protein.

9

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

respectively) with CMV-driven tGFP (pLKO.1) or eGFP (pLL3.7) reporterexpression (Table S2). Transfer plasmid (shRNA targeting TIMP-1, CD63,ITGB1, FAK or Luc) was co-transfected with psPAX2 (12260, Addgene)packaging vector and pCMV-VSVg (8454, Addgene) envelope vector intoLenti-X 293T cells (Clontech) using Lipofectamine 2000 (Invitrogen). Theresulting supernatant was harvested 24 h and 48 h post-transfection.Recovered lentiviral particles were centrifuged to eliminate cell debris,filtered through 0.45-mm cellulose acetate filters and concentrated bymeansof ultracentrifugation at 50,000 g. Titers were determined by transducingLenti-X 293T cells with serial dilutions of concentrated lentivirus. DCs(3 days post-bone marrow extraction) were transduced by means ofspinoculation at 1000 g for 30 min in the presence of hexadimethrinebromide (Polybrene, 8 µg/ml; Sigma-Aldrich). 5–7 days post-transduction,reporter (eGFP, mRuby or tGFP) expression was verified usingepifluorescence microscopy before cells were used in experiments. Thetransduction efficiency of cells used in all assays, defined as the number ofreporter-expressing cells by ocular inspection related to the total numbers ofcells in five representative fields of view, was consistently >40% (Kanataniet al., 2017; Olafsson et al., 2018). mRNA knockdown efficiency wasquantified 5–7 days post transduction using qPCR. Knockdown of TIMP-1,FAK, CD63 and ITGB1 protein was quantified from: supernatants fromshTIMP-1-transduced T. gondii-challenged DCs (Pru-RFP, MOI 3, 4 h)using ELISA; cell lysates of shFAK-transduced DCs using western blotting;and from DCs transduced with shCD63 (5 days post-transduction) orshITGB1 (7 days post-transduction) lentivirus using FACS.

Statistical analysisAll statistics were performed with Prism (v7, GraphPad). Normality wastested by the Shapiro–Wilks test. In all statistical tests P≥0.05 were definedas non-significant and P<0.05 were defined as significant.

AcknowledgementsWe thank all members of the Barragan lab for critical input.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: E.B.O., A.B.; Methodology: E.B.O., E.C.R., M.V.-G.; Validation:A.B.; Formal analysis: E.B.O., E.C.R.; Investigation: E.B.O., E.C.R.; Resources:M.V.-G.; Data curation: E.B.O.; Writing - original draft: E.B.O., A.B.; Visualization:E.B.O., E.C.R.; Project administration: A.B.; Funding acquisition: A.B.

FundingThe research was supported by grants from Vetenskapsrådet (Swedish ResearchCouncil) (K2014-56X-15133-11-6 and 2018-02411 to A.B.) and the EuropeanCommunity ERA-NET NEURON network (VR/2014-7533 to A.B.).

Supplementary informationSupplementary information available online athttp://jcs.biologists.org/lookup/doi/10.1242/jcs.225193.supplemental

ReferencesAlvarez, D., Vollmann, E. H. and von Andrian, U. H. (2008). Mechanisms andconsequences of dendritic cell migration. Immunity 29, 325-342.

Ando, T., Charindra, D., Shrestha, M., Umehara, H., Ogawa, I., Miyauchi, M. andTakata, T. (2018). Tissue inhibitor of metalloproteinase-1 promotes cellproliferation through YAP/TAZ activation in cancer. Oncogene 37, 263-270.

Bierly, A. L., Shufesky,W. J., Sukhumavasi,W., Morelli, A. E. and Denkers, E. Y.(2008). Dendritic cells expressing plasmacytoid marker PDCA-1 are Trojanhorses during Toxoplasma gondii infection. J. Immunol. 181, 8485-8491.

Chen, H.-C., Appeddu, P. A., Isoda, H. and Guan, J.-L. (1996). Phosphorylation oftyrosine 397 in focal adhesion kinase is required for binding phosphatidylinositol3-kinase. J. Biol. Chem. 271, 26329-26334.

Clark, R. T., Nance, J. P., Noor, S. and Wilson, E. H. (2011). T-cell production ofmatrix metalloproteinases and inhibition of parasite clearance by TIMP-1 duringchronic Toxoplasma infection in the brain. ASN Neuro 3, e00049.

Cook, J. H., Ueno, N. and Lodoen, M. B. (2018). Toxoplasma gondii disrupts beta1integrin signaling and focal adhesion formation during monocyte hypermotility.J. Biol. Chem. 293, 3374-3385.

Courret, N., Darche, S., Sonigo, P., Milon, G., Buzoni-Gatel, D. and Tardieux, I.(2006). CD11c- and CD11b-expressing mouse leukocytes transport singleToxoplasma gondii tachyzoites to the brain. Blood 107, 309-316.

Dellacasa-Lindberg, I., Fuks, J. M., Arrighi, R. B. G., Lambert, H., Wallin,R. P. A., Chambers, B. J. and Barragan, A. (2011). Migratory activation ofprimary cortical microglia upon infection with Toxoplasma gondii. Infect. Immun.79, 3046-3052.

Dietmann, A., Helbok, R., Lackner, P., Issifou, S., Lell, B., Matsiegui, P. B.,Reindl, M., Schmutzhard, E. and Kremsner, P. G. (2008). Matrixmetalloproteinases and their tissue inhibitors (TIMPs) in Plasmodiumfalciparum malaria: serum levels of TIMP-1 are associated with diseaseseverity. J. Infect. Dis. 197, 1614-1620.

Dranoff, J. A., Bhatia, N., Fausther, M., Lavoie, E. G., Granell, S., Baldini, G.,Hickman, D. S. A. and Sheung, N. (2013). Posttranslational regulation of tissueinhibitor of metalloproteinase-1 by calcium-dependent vesicular exocytosis.Physiol. Rep. 1, e00125.

Elkington, P. T. G., O’kane, C. M. and Friedland, J. S. (2005). The paradox ofmatrix metalloproteinases in infectious disease. Clin. Exp. Immunol. 142, 12-20.

Engering, A., Kuhn, L., Fluitsma, D., Hoefsmit, E. and Pieters, J. (2003).Differential post-translational modification of CD63molecules during maturation ofhuman dendritic cells. Eur. J. Biochem. 270, 2412-2420.

Forte, D., Salvestrini, V., Corradi, G., Rossi, L., Catani, L., Lemoli, R. M., Cavo,M. and Curti, A. (2017). The tissue inhibitor of metalloproteinases-1 (TIMP-1)promotes survival and migration of acute myeloid leukemia cells through CD63/PI3K/Akt/p21 signaling. Oncotarget 8, 2261-2274.

Friedl, P. and Wolf, K. (2003). Tumour-cell invasion and migration: diversity andescape mechanisms. Nat. Rev. Cancer 3, 362-374.

Friedl, P. andWolf, K. (2010). Plasticity of cell migration: a multiscale tuning model.J. Cell Biol. 188, 11-19.

Fuks, J. M., Arrighi, R. B. G., Weidner, J. M., Kumar Mendu, S., Jin, Z., Wallin,R. P. A., Rethi, B., Birnir, B. and Barragan, A. (2012). GABAergic signaling islinked to a hypermigratory phenotype in dendritic cells infected by Toxoplasmagondii. PLoS Pathog. 8, e1003051.

Grunwald, B., Harant, V., Schaten, S., Fruhschutz, M., Spallek, R., Hochst, B.,Stutzer, K., Berchtold, S., Erkan, M., Prokopchuk, O. et al. (2016). Pancreaticpremalignant lesions secrete tissue inhibitor of metalloproteinases-1, whichactivates hepatic stellate cells via CD63 signaling to create a premetastatic nichein the liver. Gastroenterology 151, 1011-1024.e7.

Guan, J.-L. (1997). Role of focal adhesion kinase in integrin signaling.Int. J. Biochem. Cell Biol. 29, 1085-1096.

Hitziger, N., Dellacasa, I., Albiger, B. and Barragan, A. (2005). Dissemination ofToxoplasma gondii to immunoprivileged organs and role of Toll/interleukin-1receptor signalling for host resistance assessed by in vivo bioluminescenceimaging. Cell. Microbiol. 7, 837-848.

Jackson, H. W., Defamie, V., Waterhouse, P. and Khokha, R. (2017). TIMPs:versatile extracellular regulators in cancer. Nat. Rev. Cancer 17, 38-53.

Jung, K.-K., Liu, X.-W., Chirco, R., Fridman, R. and Kim, H.-R. C. (2006).Identification of CD63 as a tissue inhibitor of metalloproteinase-1 interacting cellsurface protein. EMBO J. 25, 3934-3942.

Kanatani, S., Uhlen, P. and Barragan, A. (2015). Infection by toxoplasma gondiiinduces amoeboid-like migration of dendritic cells in a three-dimensional collagenmatrix. PLoS ONE 10, e0139104.

Kanatani, S., Fuks, J. M., Olafsson, E. B., Westermark, L., Chambers, B., Varas-Godoy, M., Uhlen, P. and Barragan, A. (2017). Voltage-dependent calciumchannel signaling mediates GABAA receptor-induced migratory activation ofdendritic cells infected by Toxoplasma gondii. PLoS Pathog. 13, e1006739.

Kim, K., Eaton, M. S., Schubert, W., Wu, S. and Tang, J. (2001). Optimizedexpression of green flourescent protein in Toxoplasma gondii using thermostablegreen fluorescent protein mutants. Mol. Biochem. Parasitol. 113, 309-313.

Kolb, S. A., Lahrtz, F., Paul, R., Leppert, D., Nadal, D., Pfister, H.-W. andFontana, A. (1998). Matrix metalloproteinases and tissue inhibitors ofmetalloproteinases in viral meningitis: upregulation of MMP-9 and TIMP-1 incerebrospinal fluid. J. Neuroimmunol. 84, 143-150.

Lambert, H., Hitziger, N., Dellacasa, I., Svensson, M. and Barragan, A. (2006).Induction of dendritic cell migration upon Toxoplasma gondii infection potentiatesparasite dissemination. Cell. Microbiol. 8, 1611-1623.

Lambert, E., Bridoux, L., Devy, J., Dasse, E., Sowa, M.-L., Duca, L., Hornebeck,W., Martiny, L. and Petitfrere-Charpentier, E. (2009a). TIMP-1 binding toproMMP-9/CD44 complex localized at the cell surface promotes erythroid cellsurvival. Int. J. Biochem. Cell Biol. 41, 1102-1115.

Lambert, H., Vutova, P. P., Adams,W. C., Lore, K. and Barragan, A. (2009b). TheToxoplasma gondii-shuttling function of dendritic cells is linked to the parasitegenotype. Infect. Immun. 77, 1679-1688.

Lammermann, T., Bader, B. L., Monkley, S. J., Worbs, T., Wedlich-Soldner, R.,Hirsch, K., Keller, M., Forster, R., Critchley, D. R., Fassler, R. et al. (2008).Rapid leukocyte migration by integrin-independent flowing and squeezing.Nature453, 51-55.

Lee, S. Y., Kim, J. M., Cho, S. Y., Kim, H. S., Shin, H. S., Jeon, J. Y., Kausar, R.,Jeong, S. Y., Lee, Y. S. and Lee, M. A. (2014). TIMP-1 modulates chemotaxis of

10

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience

human neural stem cells through CD63 and integrin signalling. Biochem. J. 459,565-576.

Liu, C.-H., Fan, Y.-T., Dias, A., Esper, L., Corn, R. A., Bafica, A., Machado, F. S.and Aliberti, J. (2006). Cutting edge: dendritic cells are essential for in vivo IL-12production and development of resistance against Toxoplasma gondii infection inmice. J. Immunol. 177, 31-35.

Montoya, J. G. and Liesenfeld, O. (2004). Toxoplasmosis. Lancet 363, 1965-1976.Ohl, L., Mohaupt, M., Czeloth, N., Hintzen, G., Kiafard, Z., Zwirner, J.,Blankenstein, T., Henning, G. and Forster, R. (2004). CCR7 governs skindendritic cell migration under inflammatory and steady-state conditions. Immunity21, 279-288.

Olafsson, E. B., Varas-Godoy, M. and Barragan, A. (2018). Toxoplasma gondiiinfection shifts dendritic cells into an amoeboid rapid migration modeencompassing podosome dissolution, secretion of TIMP-1, and reducedproteolysis of extracellular matrix. Cell. Microbiol. 20, e12808.

Pappas, G., Roussos, N. and Falagas, M. E. (2009). Toxoplasmosis snapshots:global status of Toxoplasma gondii seroprevalence and implications forpregnancy and congenital toxoplasmosis. Int. J. Parasitol. 39, 1385-1394.

Pepper, M., Dzierszinski, F.,Wilson, E., Tait, E., Fang, Q., Yarovinsky, F., Laufer,T. M., Roos, D. and Hunter, C. A. (2008). Plasmacytoid dendritic cells areactivated by Toxoplasma gondii to present antigen and produce cytokines.J. Immunol. 180, 6229-6236.

Radford, K. J., Thorne, R. F. and Hersey, P. (1996). CD63 associates withtransmembrane 4 superfamily members, CD9 and CD81, and with beta1 integrinsin human melanoma. Biochem. Biophys. Res. Commun. 222, 13-18.

Sangaletti, S., DI Carlo, E., Gariboldi, S., Miotti, S., Cappetti, B., Parenza, M.,Rumio, C., Brekken, R. A., Chiodoni, C. and Colombo, M. P. (2008).Macrophage-derived SPARC bridges tumor cell-extracellular matrix interactionstoward metastasis. Cancer Res. 68, 9050-9059.

Stetler-Stevenson, W. G. (2008). Tissue inhibitors of metalloproteinases in cellsignaling: metalloproteinase-independent biological activities. Sci. Signal. 1, re6.

Tomasik, J., Schultz, T. L., Kluge, W., Yolken, R. H., Bahn, S. and Carruthers,V. B. (2016). Shared immune and repair markers during experimental toxoplasmachronic brain infection and schizophrenia. Schizophr. Bull. 42, 386-395.

Toricelli, M., Melo, F. H. M., Peres, G. B., Silva, D. C. P. and Jasiulionis, M. G.(2013). Timp1 interacts with beta-1 integrin and CD63 along melanoma genesisand confers anoikis resistance by activating PI3-K signaling pathwayindependently of Akt phosphorylation. Mol. Cancer 12, 22.

Tsai, H. C., Shi, M. H., Lee, S. S. J., Wann, S. R., Tai, M. H. and Chen, Y. S. (2011).Expression of matrix metalloproteinases and their tissue inhibitors in the serumand cerebrospinal fluid of patients with meningitis. Clin. Microbiol. Infect. 17,780-784.

Visse, R. and Nagase, H. (2003). Matrix metalloproteinases and tissue inhibitors ofmetalloproteinases: structure, function, and biochemistry. Circ. Res. 92, 827-839.

Weidner, J. M. and Barragan, A. (2014). Tightly regulated migratory subversion ofimmune cells promotes the dissemination of Toxoplasma gondii. Int. J. Parasitol.44, 85-90.

Weidner, J. M., Kanatani, S., Hernandez-Castaneda, M. A., Fuks, J. M., Rethi, B.,Wallin, R. P. and Barragan, A. (2013). Rapid cytoskeleton remodelling indendritic cells following invasion by Toxoplasma gondii coincides with the onset ofa hypermigratory phenotype. Cell. Microbiol. 15, 1735-1752.

Wilk, C. M., Schildberg, F. A., Lauterbach, M. A., Cadeddu, R. P., Frobel, J.,Westphal, V., Tolba, R. H., Hell, S. W., Czibere, A., Bruns, I. et al. (2013). Thetissue inhibitor of metalloproteinases-1 improves migration and adhesion ofhematopoietic stem and progenitor cells. Exp. Hematol. 41, 823-831.e2.

Xing, Z., Chen, H. C., Nowlen, J. K., Taylor, S. J., Shalloway, D. and Guan, J. L.(1994). Direct interaction of v-Src with the focal adhesion kinase mediated by theSrc SH2 domain. Mol. Biol. Cell 5, 413-421.

11

RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs225193. doi:10.1242/jcs.225193

Journal

ofCe

llScience