Neutrophil recruitment to sites of injury or infection is essential for host defense, but it needs to be tightly regulated to prevent tissue damage. Phosphoinositide 3-kinase (PI3K), which generates the phosphatidylinositol (3,4,5)-trisphosphate [PI(3,4,5)P3], is necessary for neutrophil motility in vivo; however, the role of SH2-domain-containing 5-inositol phosphatase (SHIP) enzymes, which hydrolyze PI(3,4,5)P3 to phosphatidylinositol 3,4-bisphosphate [PI(3,4)P2], is not well understood. Here we show that SHIP phosphatases limit neutrophil motility in live zebrafish. Using real-time imaging of bioprobes specific for PI(3,4,5)P3 and PI(3,4)P2 in neutrophils, we found that PI(3,4,5)P3 and PI(3,4)P2 accumulate at the leading edge while PI(3,4)P2 also localizes to the trailing edge of migrating neutrophils in vivo. Depletion of SHIP phosphatases using morpholino oligonucleotides led to increased neutrophil 3D motility and neutrophil infiltration into wounds. The increase in neutrophil wound recruitment in SHIP morphants was rescued by treatment with low dose PI3Kγ inhibitor, suggesting that SHIP limits neutrophil motility by modulating PI3K signaling. Moreover, overexpression of the SHIP phosphatase domain in neutrophils impaired neutrophil 3D migration. Taken together, our findings suggest that SHIP phosphatases control neutrophil inflammation by limiting neutrophil motility in vivo.
Neutrophil infiltration into sites of infection or tissue injury is essential for host defense. On the other hand, tight control of neutrophil accumulation and activation is necessary to prevent tissue damage and chronic inflammation. Despite substantial progress in identifying mechanisms that initiate neutrophil wound attraction (Deng et al., 2012; McDonald et al., 2010; Niethammer et al., 2009; Yoo et al., 2010; Yoo et al., 2011), how neutrophil infiltration is regulated in live animals is not well defined.
It is well established that chemoattractant induced G protein-coupled receptor signaling induces phosphatidylinositol (3,4,5)-trisphosphate [PI(3,4,5)P3]/phosphatidylinositol 3,4-bisphosphate [PI(3,4)P2] concentration at the leading edge during single cell chemotaxis in vitro (Parent, 2004; Servant et al., 2000) and neutrophil motility in vivo (Yoo et al., 2010). Leading edge activation of phosphoinositide 3-kinase (PI3K) results in the rapid accumulation of PI(3,4,5)P3/PI(3,4)P2, polarized F-actin polymerization and directed motility (Barberis and Hirsch, 2008; Hawkins and Stephens, 2007; Yoo et al., 2010). The PI3K product PI(3,4,5)P3 can be hydrolyzed to PI(3,4)P2 by SH2-domain-containing inositol 5-phosphatase (SHIP) (Rohrschneider et al., 2000). In support of an important role for SHIP in controlling inflammation, knockout mice show robust leukocyte infiltration into lungs (Helgason et al., 1998). Moreover, bone marrow-derived mast cells from SHIP-deficient mice, show increased cytokine induced PI(3,4,5)P3 levels and Akt activation (Liu et al., 1999). By contrast, SHIP1-null neutrophils in vitro have impaired polarization and motility, suggesting that SHIP1 is necessary for neutrophil motility (Nishio et al., 2007). These reports suggest opposite roles for SHIP in regulating neutrophil motility in vitro and inflammation in vivo, indicating that further studies are needed to delineate the in vivo role of SHIP in regulating neutrophil motility.
The zebrafish, Danio rerio, has emerged as a powerful vertebrate model to study cell motility and immunity (Huttenlocher and Poznansky, 2008; Kanther and Rawls, 2010). The optical transparency of zebrafish embryos together with tissue specific promoter driven gene expression allow for high-resolution real time imaging at both the subcellular and single cell level in live animals. Here, we show that SHIP plays an important role in limiting leukocyte motility and infiltration in live zebrafish embryos. Depletion of SHIP 5′-phosphatases increased neutrophil wound attraction and random motility through a PI3K-dependent pathway. Moreover, ectopic expression of the SHIP1 phosphatase domain impaired neutrophil migration. Taken together, our findings suggest that SHIP phosphatases serve as a brake that limit neutrophil motility and inflammation, at least in part through their effects on PI3K signaling.
Results and Discussion
PI(3,4,5)P3/PI(3,4)P2 signaling in neutrophils in vivo
PI(3,4,5)P3/PI(3,4,)P2 localizes to the leading edge of neutrophils in vitro (Servant et al., 2000). We have also demonstrated, using the PH domain of Akt/PKB (PH-Akt), that PI(3,4,5)P3/PI(3,4)P2 localizes to the leading edge and occasionally the tail of neutrophils migrating in vivo (Yoo et al., 2010). The PH-Akt probe detects both PI(3,4,5)P3 and PI(3,4)P2 (Frech et al., 1997; Gray et al., 1999). To further define the endogenous dynamics and inter-conversion of these phosphoinositides in vivo, we used the mpx promoter to transiently express the PI(3,4,5)P3-specific bioprobe GRP1-PH and the PI(3,4)P2-specific bioprobe TAPP1-PH (Furutani et al., 2006) in zebrafish neutrophils. Using ratiometric imaging of neutrophils randomly migrating in the head, we found that TAPP1-PH [detecting PI(3,4)P2] concentrated both at the leading edge and the tail of motile neutrophils (Fig. 1A; supplementary material Movie 1). By contrast, GRP1-PH [detecting PI(3,4,5)P3] accumulated only at the leading edge of neutrophils (Fig. 1B; supplementary material Movie 2). Accordingly, ratiometric imaging comparing the accumulation of the PI(3,4,5)P3 to PI(3,4)P2 probes revealed that more PI(3,4,5)P3 than PI(3,4)P2 accumulated at the neutrophil leading edge (Fig. 1C; supplementary material Movie 3). We have previously shown by ratiometric imaging of the farnesylated membrane-bound probe EGFP-F, membrane signal periodically accumulated at the tail but not at the front of neutrophils (Yoo et al., 2010). Therefore, the cell front signal observed is unlikely due to membrane accumulation. These findings indicate that there are local sub-cellular pools of phospholipids that can be distinguished temporally and spatially during neutrophil migration in vivo.
SHIP expression and localization in zebrafish leukocytes
The dynamic accumulation of PI(3,4)P2 during neutrophil motility prompted us to investigate the function of SHIP phosphatases during neutrophil motility. In mammals, SHIP1 expression is restricted to the hematopoietic lineage (Liu et al., 1998) and SHIP2 is expressed broadly (Muraille et al., 1999). According to the PubMed database, zebrafish expresses one copy of SHIP1 (accession no. XP_001923007) that shares 61% identify with both human SHIP1a and SHIP1b (supplementary material Fig. S1). There are two SHIP2 paralogs in zebrafish, SHIP2a and SHIP2b. SHIP2a is most closely related to human SHIP2 (65% overall identity) and is broadly expressed in zebrafish (Jurynec and Grunwald, 2010). RT-PCR analysis using mRNA extracted from FACS-sorted leukocytes or whole zebrafish at 3 days post fertilization (dpf) using Tg(mpx:dendra2) (Mathias et al., 2009; Yoo and Huttenlocher, 2011), demonstrated that zebrafish myeloid cells express SHIP1 and SHIP2a, but not SHIP2b (Fig. 2A). Purity of the FACS-sorted myeloid cell population was confirmed by RT-PCR (Fig. 2B).
Previous studies have reported that SHIP1 and SHIP2 are localized at the leading edge in MDCK cells (Koch et al., 2005). To determine where SHIP phosphatases localize in motile neutrophils, we transiently expressed human EGFP-SHIP1 in Tg(mpx:mCherry) zebrafish embryos (Yoo et al., 2010), using the neutrophil specific lysozyme C (lyz) promoter (Kitaguchi et al., 2009). Human SHIP1, like PI(3,4)P2, localized both at the front and tail of motile neutrophils (Fig. 2C; supplementary material Movie 4). We further characterized the subcellular localization of zebrafish EGFP-SHIP1 and EGFP-SHIP2a in neutrophils in vivo. In accordance with our findings with human SHIP1, both zebrafish SHIP1 and SHIP2a transiently localize to the cell front and at times can be seen in the tail of motile neutrophils in vivo (supplementary material Fig. S2; Movie 5). Neutrophil-specific protein expression using the lyz promoter was confirmed by crossing the transgenic line Tg(lyz:TagRFP), which we generated based on the published lyz promoter sequence (Kitaguchi et al., 2009), with Tg(mpx:dendra2) (Yoo and Huttenlocher, 2011) (supplementary material Fig. S3).
SHIP phosphatases impair recruitment of neutrophils to wounds
To determine if SHIP phosphatases regulate neutrophil motility in vivo, we used SHIP-targeted splice donor blocking morpholinos (MOs) to deplete SHIP1 and SHIP2a. RT-PCR confirmed SHIP knockdown (Fig. 3A). We found that depletion of both SHIP1 and SHIP2a (Fig. 3B) but not single SHIP1/2a (data not shown) increased neutrophil wound recruitment at 0.5 and 1 hour post wound (hpw) in 2.5 dpf zebrafish. The effect of SHIP depletion was confirmed using a second set of morpholinos, which targeted SHIP1 (SHIP1 MO2) and SHIP2a (SHIP2a MO1) (Fig. 3C). It has been reported that there are increased total neutrophils in SHIP-deficient mice (Rauh et al., 2004). By contrast, total neutrophil numbers were decreased in SHIP morphants (supplementary material Fig. S4), despite having more neutrophils infiltrate into wounds. Moreover, we also found that macrophage recruitment to wounds was increased in the SHIP morphants (supplementary material Fig. S5), suggesting that SHIP phosphatases limit both neutrophil and macrophage wound recruitment. After initial recruitment, neutrophils migrate repeatedly towards and away from the wound in a process called reverse migration (Mathias et al., 2006; Yoo and Huttenlocher, 2011). SHIP morphants did not show a significant difference in reverse migration or resolution of inflammation as compared to control (data not shown).
Previous studies have reported that depletion of SHIP1 increases PI(3,4,5)P3 accumulation in leukocytes (Harris et al., 2011; Nishio et al., 2007; Vedham et al., 2005). To test the possibility that SHIP phosphatases modulate neutrophil wound recruitment by affecting PI(3,4,5)P3 levels, we inhibited PI(3,4,5)P3 generation in SHIP morphants using low dose PI3Kγ inhibitor. It is known that PI3Kγ inhibition impairs PI(3,4,5)P3/PI(3,4)P2 generation and neutrophil motility in vivo (Yoo et al., 2010). We found that treatment with low dose PI3Kγ inhibitor did not affect neutrophil wound recruitment in control zebrafish. However, low dose PI3Kγ inhibitor was sufficient to rescue neutrophil wound recruitment in SHIP morphants to control levels (Fig. 3D). These findings suggest that depletion of SHIP increases neutrophil recruitment through a PI3K-dependent pathway. However, we cannot rule out the possibility that the generation of PI(3,4)P2 may also be important for SHIP effects on neutrophil wound attraction. Since PI(3)K activity and leading edge PI(3,4,5)P3 is indispensable for neutrophil motility in vivo (Yoo et al., 2010), it is interesting to speculate that SHIP hydrolyzes PI(3,4,5)P3 into PI(3,4)P2, particularly in the front of neutrophils, and thereby limits neutrophil motility.
SHIP phosphatases limit neutrophil motility in vivo
It is possible that neutrophil wound recruitment is enhanced in SHIP morphants due to increased neutrophil motility in SHIP morphants. To test this possibility, we performed real time imaging of neutrophil random motility in the head region in control and SHIP morphants at 2.5 dpf. We found that there was increased neutrophil random motility in SHIP double morphants compared to control (Fig. 4A–C; supplementary material Movie 6).
To determine if the SHIP phosphatase activity is sufficient to alter neutrophil motility, we ectopically expressed a membrane-bound form of the human SHIP1 phosphatase domain (aa364–826) in zebrafish neutrophils (Freeburn et al., 2002). Transient expression was achieved using the lyz promoter driving both the constitutively active SHIP1 phosphatase and EGFP expression with the viral 2A peptide system which allows multiple protein products to be expressed from a single transgene (Provost et al., 2007). Transient expression of this construct in Tg(mpx:mCherry) embryos allowed for mosaic expression of the phosphatase domain labeled with EGFP and mCherry to be compared to control neutrophils that expressed mCherry alone. Live imaging of neutrophil random motility in the head of 3 dpf embryos showed that ectopic expression of the SHIP1 phosphatase domain impaired neutrophil random motility as compared to control neutrophils (Fig. 4D–E; supplementary material Movie 7). By contrast, ectopic expression of EGFP control did not impair neutrophil motility (Fig. 4F). Further analysis demonstrated that there was no significant difference in the meandering index (displacement/track length) measured in the SHIP knockdown embryos and the embryos overexpressing the SHIP1 phosphatase domain (data not shown), suggesting that SHIP functions to limit neutrophil speed but not directional migration or persistence. Taken together, our findings suggest that SHIP limits neutrophil motility through the modulation of PI3K signaling. Moreover, our findings demonstrate that the effects of ectopic expression of SHIP1 are cell autonomous, since SHIP1 expression in neutrophils alone was sufficient to impair migration.
It is well established that localized generation of PI(3,4,5)P3 is critical for neutrophil motility. PI(3,4,5)P3 can activate Rac through the recruitment of Rac guanine nucleotide exchange factors (Nishikimi et al., 2009) and it is known that Rac2 is required for neutrophil motility in vivo (Deng et al., 2011). SHIP prevents PI(3,4,5)P3 accumulation and it is possible that SHIP phosphatases limit neutrophil motility by affecting Rac2 activity at the leading edge. In addition to the role that SHIP plays in dampening PI(3,4,5)P3 signaling, it is also possible that SHIP alters neutrophil motility through the generation of PI(3,4)P2 which can act as a second messenger by binding pleckstrin homology(PH)-containing proteins such as TAPPs (Marshall et al., 2002), Akt, PDK-1 and PKC family members (reviewed by Rameh and Cantley, 1999).
Our findings that SHIP impairs leukocyte motility are consistent with previous reports that bone marrow-derived SHIP-deficient macrophages exhibit spontaneous and enhanced migration (Vedham et al., 2005) and the increase in myeloid infiltration into vital organs in SHIP−/− mice (Helgason et al., 1998). A previous report showed that SHIP-deficient neutrophils have impaired cell polarity and migration in vitro, suggesting a role for SHIP in regulating neutrophil motility (Nishio et al., 2007). However, this did not explain the SHIP-deficient inflammation phenotype observed in vivo. This discrepancy may be due to the differences between in vitro 2D and in vivo 3D motility, which sometimes yield contradictory results. For example, integrins are required for leukocyte migration on 2D surfaces but not in 3D interstitial tissues (Lämmermann et al., 2008). Moreover, MEK-cofilin signaling controls T-cell migration in 3D but not 2D environments (Klemke et al., 2010). Therefore, the use of in vivo model systems is crucial to increase our understanding of how specific pathways regulate leukocyte motility and inflammation in live animals.
Our findings suggest that SHIP is a key brake that limits neutrophil motility in vivo. An important unanswered question is how and when this “brake” is released to allow for motility. Hydrogen peroxide is generated at wounds and mediates rapid wound detection in zebrafish (Niethammer et al., 2009). Moreover, we have recently shown that the Src family kinase Lyn acts as a redox sensor that can mediate leukocyte wound attraction (Yoo et al., 2011). In addition to activating Lyn kinase, hydrogen peroxide can also inactivate phosphatases, such as SHIP, through the oxidation of specific cysteine residues (Paulsen and Carroll, 2010; Poole and Nelson, 2008). Although Lyn can activate SHIP under some conditions (Baran et al., 2003), it is intriguing to speculate that in the presence of hydrogen peroxide, Lyn activation of SHIP phosphatases (the “brake”) is prevented. A future challenge is to understand how these positive and negative regulatory pathways are coordinated and balanced to mediate proper leukocyte wound attraction and prevent untoward inflammation in vivo.
Materials and Methods
Tail fin wounding, whole-mount immunolabeling and Sudan Black staining
All animal experiments were performed according to approved guidelines. Embryos pretreated with 0.5% DMSO, AS-605240 (1 µM) for 1 hour at 2.5 dpf were anesthetized by 0.2 mg/mL tricaine and wounded at the tail fin with a 33 gauge needle. Tg(mpx:dendra2) embryos were fixed in 1.5% formaldehyde in 0.1 M PIPES, 1 mM MgSO4, and 2 mM EGTA overnight at 4°C and immunolabeled for L-plastin as previously described (Mathias et al., 2009; Yoo and Huttenlocher, 2011). Wild-type embryos were fixed in 4% formaldehyde overnight at room temperature and stained with Sudan Black as described previously (Le Guyader et al., 2008).
RT-PCR was performed using mRNA from FACS-sorted myeloid cells using Tg(mpx:dendra2) zebrafish embryos as previously described (Mathias et al., 2009; Yoo and Huttenlocher, 2011). Primers used for checking expression of different SHIP isoforms are as follow: SHIP1-RT-F: 5′-CACAGCCAATCAAATCCCACATGAC-3′; SHIP1-RT-R: 5′-GCTTCGTGTGATGGAAGGTCAAC-3′; SHIP2a-RT-F: 5′-CTGTGCAAACCTTCGAGGTGAAAC-3′; SHIP2a-RT-R: 5′-CCACCCATTCTTTATCGCACACAG-3′; SHIP2b-RT-F: 5′- GAGAATTTGAAAGGAGGCGCGTC-3′; SHIP2b-RT-R: 5′-CTCTGAAACTTCACCAACTGCACG-3′.
Primers used for checking MO knockdown efficiency are as follow: SHIP1 MO1-F: 5′-GGCACCATGGCAATATCACTCGG-3′; SHIP1 MO1-R: 5′-CCCATTCTGGACCTGTTCTGCATC-3′; SHIP1 MO2-F: 5′-GTCTCCTGGAGCTGGAAGACTAAG-3′; SHIP1 MO2-R: 5′-GTCATGTGGGATTTGATTGGCTGTG-3′; SHIP2a MO1/MO2-F: 5′-CTTCGCCAGTGCAAAGAAAAGGG-3′; SHIP2a MO1/MO2-R: 5′-CTCTCTCCAGGTTCAGCTGGTC-3′.
DNA, RNA and morpholino injection
All DNA expression vectors contain either the zebrafish myeloperoxidase (mpx) promoter or lysozyme C (lyz) promoter for neutrophil expression (Kitaguchi et al., 2009; Mathias et al., 2006), minimal Tol2 elements for efficient integration (Urasaki et al., 2006), and an SV40 polyadenylation sequence (Clontech Laboratories, Inc.). Constructs with each of the following in the backbone were constructed: TagRFP-GRP1-PH, AcGFP-TAPP1-PH (Furutani et al., 2006), human EGFP-SHIP1 (accession BC113582), zebrafish EGFP-SHIP1 (accession XM_001922972), zebrafish EGFP-SHIP2a (accession DQ272661) and TM-SHIP1(aa364–826)-2A-EGFP [TM, plasma membrane targeting signal (Lyn kinase targeting sequence); human SHIP1b accession BC113582], The primers SHIP1F, 5′-AGCAGATCTATGTTTTCCCCCCAGCCGTG-3′ and SHIP1R, 5′-GCTTCTAGATCACTTTAGCTGTCGACCCATCTTCAG-3′ were used to amplify the full-length coding sequence of zebrafish SHIP1 from WT AB strain cDNA at 3 dpf, which was then tagged with EGFP. The primers SHIP2aF, 5′-AGCGAATTCGTGTGAGTCTCCGCGATGGC-3′ and SHIP2aR, 5′-GCTACTAGTGGAATTCCTCAGGTCAGATTCACCAGTGC-3′ were used to subclone SHIP2a from SHIP2a cDNA and tagged with EGFP.
Expression of single construct was obtained by injection of 3 nL solution containing 12.5 ng/µL of DNA plasmid and 17.5 ng/µL in vitro transcribed (Ambion) Tol2 transposase mRNA into the cytoplasm of one-cell stage embryos. Expression of two constructs was obtained by mixing 6.25 ng/µL of each DNA plasmid. Two non-overlapping antisense splice-blocking morpholino oligonucleotides (MOs) of zebrafish SHIP1 and SHIP2a were synthesized by Gene Tools: SHIP1 MO1, 5′- TTGGACTGTTCAGATGTACCTGGTT-3′ which corresponds to the putative exon 3/intron 3 boundary upstream of the phosphatase domain of zebrafish SHIP1; SHIP1 MO2, 5′-ATGACTTAAGACATCTCACCCATGT-3′ which corresponds to the putative exon 12/intron 12 region within the phosphatase domain of zebrafish SHIP1; SHIP2a MO1, 5′-TGTGTTGTTGTGTTCCTGACCGAGT-3′ which corresponds to the exon 13/intron 13 of zebrafish SHIP2a; and SHIP2a MO2, 5′-CCCAGAATGCCGTGTTTCACCTGTA-3′ which corresponds to the exon 15/intron 15 region within the phosphatase domain of zebrafish SHIP2a. Morpholinos were resuspended in Danieau buffer at a stock concentration of 1 mM. Final MO concentrations were injected into the cytoplasm of one-cell stage embryos. Individual concentrations used: 100 µM, 3 nl SHIP1 MO1; 250 µM, 3 nl SHIP1 MO2; 25 µM, 3 nl SHIP2a MO1; 200 µM, 3 nl SHIP2a MO2. Standard control MO from Gene Tools 5′-CCTCTTACCTCAGTTACAATTTATA-3′ was used as controls. Injected embryos were cultured in E3 medium at 28.5°C.
Embryos at 2–3 dpf were anesthetized using 0.2 mg/mL tricaine on a glass-bottom dish for live imaging. Time-lapse fluorescence images were acquired using a confocal microscope (FluoView FV1000, Olympus) using a NA 0.75/20×objective or a NA 1.10/60×water immersion objective. Sequential line scanning was used for each fluorescence channel. Z-stacked and ratiometric images were generated by FluoView FV1000 software (Olympus). After the ratio analysis with background subtraction was performed, a median filter was used to generate an image for visualization. For supplementary material Fig. S2 and Movie 5, a spinning disk confocal system (Yokogawa CSU-X) mounted on a Zeiss Observer Z.1 inverted microscope was used. A Photometrics Evolve EMCCD camera and a NA1.3/60×water immersion objective was used to acquire the Z-series images with a 0.4 µm step size and 300 EM gain. Maximum intensity projection images were made using the Zen 2011 (blue edition) software (Carl Zeiss). Ratio images were generated by using the AxioVison SE64 Rel 4.8 Physiology measurement function with confidence mapping.
Experimental results were analyzed with Prism version 4 (GraphPad Software) statistical software. Statistical significance was determined with the unpaired Student t test or one-way ANOVA with Dunnett post-test. The resulting P values are included in the figure legends for each experiment; n = the number of embryos or neutrophils quantified.
We thank Dr David J. Grunwald and Dr Michael J. Jurynec for the zebrafish SHIP2a cDNA and Dr Toshiki Itoh for the TagRFP-GRP1-PH and AcGFP-TAPP1-PH. We thank Dr Erwin Berthier for his help with the meandering index analysis.
This work was supported by National Institutes of Health [grant number GM074827 to A.H.]; and American Heart Association fellowship [number 11PRE4890041 to S.K.Y.]. Deposited in PMC for release after 12 months.