Spatiotemporal photolabeling of neutrophil trafficking during inflammation in live zebrafish

A genetically encoded photolabeling system was established by generating transgenic zebrafish that express a photoconvertible fluorescent reporter Dendra2 in neutrophils.

Abstract

How neutrophils traffic during inflammation in vivo remains elusive. To visualize the origin and fate of neutrophils during induction and resolution of inflammation, we established a genetically encoded photolabeling system by generating transgenic zebrafish that express a photoconvertible fluorescent reporter Dendra2 in neutrophils. Spatiotemporal photolabeling of neutrophils in vivo demonstrates that they emerge from the hematopoietic tissue in close proximity to injured tissue and repeat forward and reverse migration between the wound and the vasculature. Subsequently, neutrophils disperse throughout the body as wound-healing proceeds, contributing to local resolution at injured tissue and systemic dissemination of wound-sensitized neutrophils. Tissue damage also alters the fate of neutrophils in the caudal hematopoietic tissue and promotes caudorostral mobilization of neutrophils via the circulation to the cephalic mesenchyme. This work provides new insight into neutrophil behaviors during inflammation and resolution within a multicellular organism.

Introduction

Neutrophil chemotaxis to injured sites is a critical first step during inflammation and wound healing. Neutrophils arrive at damaged tissues as a first line of defense and sterilize the wound by secreting antimicrobial peptides, proteolytic enzymes, and ROS, contributing to wound resolution [1]. Neutrophil infiltration can also lead to destruction of host tissues and persistent inflammation. A fundamental, unanswered question is how inflammation is resolved. Persistent inflammation contributes to the pathogenesis of many diseases, including cardiovascular disease, cancer, rheumatologic disorders, and multiple organ failure [2]. Although the mechanisms and kinetics of how inflammation starts have been the focus of numerous studies, our understanding of how inflammation persists or resolves remains limited. In addition, the mechanisms that regulate inflammation in physiological environments in vivo are still elusive, as few systems are amenable to direct live imaging of leukocyte trafficking in vivo. Our current understanding of neutrophil chemotaxis and trafficking to inflamed tissues is derived primarily from in vitro analysis of purified mammalian neutrophils or fixed tissues in vivo. However, direct in vivo analyses of neutrophil behaviors have recently provided novel insight compared with in vitro studies. For example, neutrophil lifespan has generally been considered to be less than 1 day, but a recent in vivo labeling study has shown that neutrophil lifespan is 5–6 days in the vasculature in humans, much longer than predicted [3]. Thus, development of a system to study leukocyte homeostasis in physiological settings is critical for increasing our understanding of what is occurring in vivo.

The zebrafish, Danio rerio, has emerged as a powerful vertebrate model to study hematopoiesis and immunity, as it is optically transparent and highly amenable to genetic manipulation and chemical screens [4–6]. The immune systems are highly conserved between zebrafish and mammals: zebrafish develop immune cells, including lymphocytes, mast cells, DCs, eosinophils, macrophages, and neutrophils [7–18]. Recent studies using zebrafish have provided unprecedented insight into how neutrophils respond to wounded tissue [19–21]. A study has shown that H₂O₂, produced at a wound is responsible for neutrophil recruitment to damaged tissue [20], exemplifying the usefulness of live imaging in zebrafish to study neutrophilic inflammation in vertebrates.

Previously, we have shown, using a transgenic zebrafish Tg(mpx:GFP) expressing GFP in neutrophils, that neutrophils perform reverse migration to the vasculature after trafficking to the wound and suggested that reverse migration might be involved in resolution of inflammation [19, 22]. This is in clear contrast to the current, dominant idea that resolution is induced by neutrophil-programmed cell death at wounds [23]. However, neutrophil reverse migration, away from the site of inflammation, has been observed in mammalian systems as well [24, 25]. At present, the exact behaviors and trafficking of neutrophils after leaving wounded sites are not clear, as neutrophil tracking by Tg(mpx:GFP) is spatiotemporally limited as a result of difficulty in distinguishing individual cells in multiple areas. To demonstrate how neutrophil trafficking is regulated during induction and resolution of inflammation, we established a genetically encoded photolabeling system by generating a transgenic zebrafish expressing a photoconvertible fluorescent protein Dendra2 in neutrophils. Tracking of photolabeled neutrophils during inflammation demonstrates the origin and fate of neutrophils during wound responses.

MATERIALS AND METHODS

Zebrafish maintenance

Adult fish were maintained in accordance with the University of Wisconsin-Madison Research Animal Resources Center (Madison, WI, USA). Tg(fli1-EGFP) was purchased from the Zebrafish International Resource Center fish center (Eugene, OR, USA). For live imaging or wounding assays, embryos were anesthetized in E3 containing 0.2 mg/ml Tricaine (ethyl 3-aminobenzoate; Sigma-Aldrich, St. Louis, MO, USA). To prevent pigment formation, some embryos were maintained in E3 containing 0.2 mM N-phenylthiourea (Sigma-Aldrich). For all experiments, the temperature was maintained at 29°C–30°C.

Generation of Tg(mpx:Dendra2)

Tg(mpx:Dendra2) was generated as described previously [21]. Dendra2 (Evrogen, Moscow, Russia) [26], with an SV40 polyadenylation sequence, was PCR-amplified and inserted into a backbone vector containing minimal Tol2 elements [27]. The mpx promoter [19], cut with Xho1/BamH1, was inserted into this plasmid. Solution (0.5 nl) containing 25 ng/μl DNA plasmid and 35 ng/μl transposase mRNA was injected into the cytoplasm of one-cell stage embryos. Multiple Tg lines from four founders were generated.

FACS and RT-PCR

Dissociated cells of Tg(mpx:Dendra2) at 3 dpf were sorted by FACS as described previously [28]. RNA was isolated using the RNeasy mini kit (Qiagen, Valencia, CA, USA), and RT-PCR was performed. Primers for c-fms, mpx, and ef1α have been described previously [28].

Live imaging and photoconversion of Dendra2

Embryos at 3 dpf were embedded in 1% low melting-point agarose (Fisher, Waltham, MA, USA) or settled on a custom-made, glass-bottom dish. Time-lapse fluorescence images were acquired with a confocal microscope (FluoView FV1000, Olympus, Center Valley, PA, USA) using a numeric aperture 0.75/20× objective or Nikon SMZ-1500 zoom microscope (Nikon, Melville, NY, USA). For confocal imaging, each fluorescence channel (488 nm and 543 nm) and DIC images were acquired by sequential line-scanning. Z-series was acquired using a 300- to 600-μm pinhole and 2–10 μm step sizes. Z-stacked fluorescence images were overlayed onto a single DIC plane. For photoconversion of one or more cells, a 405-nm laser was focused into a circular area (diameter 10–30 μm), including the cells for 1 s with 40% power 10.0 μ/pixel (tornado function). For photoconversion of neutrophils in the CHT, a 405-nm laser was focused into a rectangular region that covers the CHT for 55 s with 60–70% power, 10.0 μ/pixel. For neutrophil tracking over multiple days, embryos were extracted from agarose to prevent fin damage. Tracking (Manual Tracking plugin) of neutrophils was performed by using ImageJ (National Institutes of Health, Bethesda, MD, USA).

Immunolabeling and Sudan Black staining

One hour after tail transection, Tg(mpx:Dendra2) larvae at 3 dpf were fixed with 1.5% formaldehyde in 0.1 M PIPES, 1.0 mM MgSO₄, and 2 mM EGTA overnight at 4°C and immunolabeled for MPX and L-plastin as described previously [28]. The ventral fins of larvae at 3 dpf were wounded, and neutrophils were stained with Sudan Black 6 h and 12 h postwounding as described previously [21].

RESULTS AND DISCUSSION

To track where neutrophils traffic during wound-provoked inflammation, we generated a transgenic zebrafish Tg(mpx:Dendra2), in which the mpx promoter specifies high expression of Dendra2 in neutrophils [19, 21, 29]. Our previous study indicated that the mpx promoter drives expression in all neutrophils, showing that mpx-driven expression of mutant CXCR4 affects behaviors of all neutrophils stained with Sudan Black [29]. Dendra2 is a monomeric, photoconvertible fluorescent protein derived from octocoral Dendronephthya sp and capable of 1000- to 4500-fold photoconversion from green to red fluorescent states in response to visible blue or UV light [26]. Tg(mpx:Dendra2) begins to express green fluorescence by 2 dpf and shows very bright expression in neutrophils in the head and the caudal hematopoietic tissue (CHT) [30] by 3 dpf (Fig. 1C), which is consistent with neutrophil localization patterns described previously [14, 16, 19, 31]. RT-PCR of FACS-sorted cells showed that cells expressing high and low levels of Dendra2 express mpx and c-fms, respectively, indicating that they correspond to neutrophils and macrophages, respectively (Fig. 1A). This is consistent with previous Tg lines that we have generated and characterized in detail [19, 21, 28, 29]. Macrophages, which have very low expression of Dendra2, could be identified by their distinctively elongated morphology in live larvae at 2–4 dpf, as reported previously [28]. However, it was not possible to detect such low expression of Dendra2 in macrophages after fixation. Accordingly, immunolabeling showed that all MPX+ cells (neutrophils) expressed Dendra2, but only a subpopulation of L-plastin+ cells (neutrophils and macrophages) expressed Dendra2, suggesting that Dendra2 was expressed primarily at a high level in neutrophils (Fig. 1B). In the current study, we focused on neutrophils at 3–4 dpf and used imaging settings that allowed the visualization of neutrophils but not macrophages as described previously [19, 28]. Notably, Tg(mpx:Dendra2) is the brightest among multiple Tg lines that we have generated and has no background expression. Photoconversion of neutrophils using a 405-nm laser changed their fluorescence from green to red, demonstrating that photoconversion works in neutrophils in vivo (Fig. 1C).

Figure 1. Generation of Tg(mpx:Dendra2).

(A) RT-PCR of c-fms (macrophage marker), mpx (neutrophil marker), and ef1α (loading control) from Dendra2-low and -high fractions sorted by FACS. Tg(mpx:Dendra2) larvae at 3 dpf were used for flow cytometry. Dendra2-low and Dendra2-high fractions are boxed (blue). Cells from WT larvae (red) were used to set the lowest level of expression. FSC-A, Forward-scatter pulse area. (B) Tg(mpx:Dendra2) larvae at 3 dpf were fixed 1 h after tail transection and immunolabeled using antibodies to MPX (neutrophil) and L-plastin (neutrophil and macrophage). Dendra2 is highly expressed in MPX+ cells but only in a subpopulation of L-plastin+ cells. Original scale bars = 20 μm. (C) Neutrophils within the three boxes in the CHT were photoconverted by 405 nm laser from green (488 nm) to red (543 nm). Photoconverted cells appear white as a result of merged red fluorescence (543 nm) and the remaining green fluorescence (488 nm). Pigments along the dorsal ridge have autofluorescence detected by 543 nm excitation.

Wounding in the ventral fin recruits neutrophils to the wound site within 1 h [19, 32, 33]. Although some populations of leukocytes have been suggested to emerge from the CHT [32, 33], the origin of neutrophils recruited to wounds in zebrafish larvae has not been demonstrated definitively, as continuous image acquisition to capture rapid neutrophil dynamics is technically difficult. It has been postulated that neutrophils come from the CHT or circulation. To distinguish the two possibilities, neutrophils in the CHT were photolabeled by photoconversion of Dendra2 (Fig. 2A and B and Supplemental Video 1). Wounding in the ventral fin provoked rapid recruitment of photolabeled neutrophils to the wound (Fig. 2C and D and Supplemental Video 1), indicating that neutrophils come directly from the hematopoietic tissue in close proximity to the wound in zebrafish larvae.

Figure 2. Photolabeling of neutrophils in the CHT.

Neutrophils within the rectangle in the CHT (A) were photoconverted (B), and the ventral fin was wounded (C). Photolabeled neutrophils migrated to the wound (D). Autofluorescent pigment is indicated by arrowheads. Original scale bar = 100 μm.

To track where neutrophils go after leaving the wound, photolabeling of single neutrophils was performed at the wound in the ventral fin. This approach revealed that the same neutrophils repeat forward and reverse migration between the wound and the blood vessel multiple times within the first 4–6 h postwounding (Fig. 3A and H and Supplemental Video 2), suggesting an apparent competition of signaling between the two locations. Some macrophages, which exhibited an elongated morphology and expressed low levels of Dendra2, were also observed in live larvae (Supplemental Video 3), although this dim population was not detected in fixed larvae by immunostaining (Fig. 1B). To investigate whether neutrophils intravasate into the vessel lumen after reverse migration, Tg(mpx:Dendra2) was crossed with Tg(fli1:EGFP) [34] to visualize vascular endothelial cells. This approach, in combination with three-dimensional reconstruction, revealed that some but not all neutrophils intravasate into the lumen of the posterior cardinal vein after leaving the wound (Fig. 3 B–G and Supplemental Video 4). Intravasation of neutrophils could also be identified based on weakening of fluorescence signals within the blood vessel, presumably as a result of circulating blood (Supplemental Video 4). As wound healing proceeds 4–6 h after wounding, the number of neutrophils returning to the wound gradually decreased, and inflammation resolved (Supplemental Video 5). In the time-frame of 4–6 h after wounding, we did not detect any apoptosis of neutrophils around wounds in more than 10 movies, which is consistent with a previous report [35]. Wounding in the tail fin also induced repetition of forward and reverse migration, which was resolved gradually (Supplemental Video 6), suggesting that bidirectional neutrophil trafficking is a general feature of wound-induced migration regardless of the location of the wound.

Figure 3. Neutrophil bidirectional trafficking between the wound and vasculature.

(A) Two neutrophils were photoconverted from green to red at a wound (14 min). A neutrophil indicated with an arrow migrated toward the vasculature (34 min), exited the fin (40 min), and came back to the wound (50 min). Autofluorescent pigment is indicated by arrowheads. (B) Z-projected image of blood vessels and two neutrophils, which were photolabeled at a wound in Tg(mpx:Dendra2/fli1:EGFP). White lines (labeled C–E) indicate sections made though this Z-stack and correspond to panels C–E. Autofluorescent pigment is indicated by the arrowhead. (C) The picture indicates that the two photolabeled neutrophils are in different z-planes. (D) A photolabeled neutrophil is adjacent to, but not in the posterior cardinal vein. (E) A photolabeled neutrophil is in the posterior cardinal vein. (F and G) Images of single planes indicated by the lines in (C). (F) A section between the CHT and the trunk muscles. (G) A section encompassing the posterior cardinal vein and the photolabeled neutrophil, which reverse-migrated into it. (H) Tracking of two neutrophils (green and blue) for 6.5 h after wounding. Note that neutrophils repeat forward and reverse migration between the wound and blood vessel. (I–O) Neutrophil tracking over multiple days. (I) Five neutrophils were photolabeled at a wound at 3 dpf. Arrows indicate photolabeled neutrophils. One neutrophil was identified in the head (J), two in the superficial region of the trunk (K), and one in the tail fin (L) 24 h after photolabeling at 4 dpf. Three neutrophils were observed on the surface of the trunk (M–O) 48 h after photolabeling at 5 dpf. Original scale bars = 50 μm.

To explore where neutrophils go after wound resolution, long-term tracking was performed over days. Fig. 3I–O is a typical example of the long-term tracking. In this case, five neutrophils were photolabeled at the wound (Fig. 3I) at 3 dpf. After 1 day, five neutrophils could still be identified: one was in the head, three were in the superficial regions of the trunk, and one was in the tail fin (Fig. 3J–L). Two days after photolabeling, three neutrophils were observed in the superficial regions of the trunk (Fig. 3M–O). Photolabeled neutrophils were also observed in the circulation within the blood vessel (data not shown). We tried to identify a specific location where neutrophils prefer to go after leaving the wound by performing photolabeling of neutrophils in more than 10 larvae but found that photolabeled neutrophils disperse to diverse tissues without specific preferences. Neutrophils migrated and kept changing their positions throughout the whole body. It is interesting to note that we did not observe stationary photolabeled neutrophils in the CHT after resolution, supporting the idea that the CHT is a hematopoietic tissue where immature cells exist.

To determine whether wounding affects only neutrophils that have traveled to wounds or also affects other neutrophils in the CHT, we focused on the overall dynamics of neutrophils in the CHT. Although neutrophils migrate randomly in the head and are stationary in the CHT at 2–4 dpf [21], the developmental behaviors and fates of neutrophils in the CHT are not known. Thus, we compared neutrophil behaviors during normal developmental processes and wound-provoked inflammation. Neutrophils in the CHT were photolabeled at 3 dpf to track their dynamics (Fig. 4A). Most of the photolabeled neutrophils remained in the CHT in conditions with and without wounding 12 h after photoconversion. However, neutrophils were consistently observed in the heads of the wounded embryos but not controls (Fig. 4B). Quantification indicated that wounding in the ventral fin promotes neutrophil mobilization from the CHT to the head (Fig. 4C). Wounding did not have significant effects on the total number of neutrophils in the CHT or the heads of larvae 6 h and 12 h postwounding (Fig. 4D). We analyzed carefully whether neutrophils localized to specific regions or organs in the head, such as the thymus or kidney marrow, but found that neutrophils were dispersed in the mesenchyme of the head (Supplemental Video 7). Photoconverted neutrophils were still dispersed in the mesenchyme of the head at 6 dpf (3 days postphotoconversion; Fig. 4E). To dissect whether neutrophils mobilize to the mesenchyme of the head by interstitial migration or via the circulation, neutrophil behaviors were analyzed by live imaging. Live analysis revealed that neutrophils in the CHT mobilized to the head via the circulation a few hours postwounding (Supplemental Video 7). This caudorostral mobilization from the CHT to the cephalic mesenchyme was observed, even when neutrophils did not reach the wound as a result of ineffective wounding (Supplemental Video 8), suggesting that arriving at a wound itself is not necessary for neutrophil mobilization to the head.

Figure 4. Wound-induced caudorostral mobilization of neutrophils from the CHT to the cephalic mesenchyme.

(A) Neutrophils in the CHT were photolabeled, and the ventral fin was wounded at 3 dpf. (B) Photolabeled neutrophils were observed 24 h after photoconversion at 4 dpf. Several photolabeled neutrophils were observed in the head of the wounded embryo. Eyes, pigment, and a part of the yolk sac (outlined by white lines) have autofluorescence excited by the 543-nm laser. (C) Quantification of numbers of photolabeled neutrophils in the head 24 h after photoconversion in the CHT. *P < 0.05; two-tailed Mann-Whitney U-test. (D) Quantification of total numbers of neutrophils in the CHT and the mesenchyme of the head 6 h and 12 h after wounding. The ventral fin of larvae at 3 dpf was wounded, and neutrophils were stained with Sudan Black. P > 0.05; two-tailed Mann-Whitney U-test. (E) Neutrophils in the CHT were photolabeled, and the ventral fin was wounded at 3 dpf. Photolabeled neutrophils were observed in various regions of the head 3 days after photoconversion (6 dpf). Original scale bars = 100 μm.

Here, we established a genetically encoded photolabeling system using Dendra2 in neutrophils to visualize neutrophil trafficking in vivo. Using Tg(mpx:Dendra2), we visualized how inflammation starts, persists, and resolves in live zebrafish. Tissue injury induces neutrophil chemotaxis from the CHT to the wound and repetitive forward and reverse migration between the wound and the vasculature. Subsequently, neutrophilic inflammation at a wound is resolved by trafficking of neutrophils to diverse tissues throughout the body of the zebrafish. This process of resolution is in clear contrast with the current dominant view that inflammation resolves by neutrophil apoptosis at sites of tissue injury or infection [23]. It is noteworthy that zebrafish neutrophils were reported to rarely undergo apoptosis (∼4%) after tail transection of older larvae (7–10 dpf) [35]. In addition, it was difficult to capture apoptosis with wounding the ventral fin of 3–4 dpf larvae.

Whether our observation in zebrafish can be applied to mammals in vivo has not been determined. However, recent studies have shown that mammalian neutrophils can undergo reverse migration [24, 25] and that reverse migration is increased in rheumatoid arthritis [24]. Moreover, the numbers of identifiable apoptotic neutrophils are low compared with the total number of neutrophils in mammals [35], and the lifespan of mammalian neutrophils is much longer than expected [3]. Taken together, we speculate that mammalian neutrophils might demonstrate reverse migration during resolution of inflammation.

At present, mechanisms that regulate cycles of forward and reverse migration between a wound and the blood vessel are not known. As each neutrophil demonstrates spatiotemporally different behavior, we speculate that one of the following mechanisms could be responsible for this phenomenon. A cell-autonomous regulatory system may exist to sensitize and desensitize neutrophils to signals from the wound and vasculature. Alternatively, there may be a regulatory mechanism in the immediate vicinity of individual neutrophils. Fully understanding the process of repeated forward and reverse migration will require elucidation of the chemotactic factors used for each direction. H₂O₂ was indicated as an early signal to recruit neutrophils to damaged tissue [20], but signals involved in later recruitment of neutrophils are not known. It also remains unknown what attracts neutrophils to the blood vessel from a wound. Inflammatory mediators, such as platelet-activating factor and angiopoietin, both of which can be secreted from endothelial cells and activate neutrophils [36, 37], might be involved in this process.

It remains to be determined what function neutrophils perform after dispersal from a wound. If inflammation is defined as neutrophil infiltration into the tissue, dissemination of neutrophils to the whole body may be considered a type of systemic inflammation, providing a potential alarm system to modify subsequent immunity. In this context, one can imagine that neutrophil reverse migration has two functions: resolution at the local site of inflammation and dissemination of wound-sensitized neutrophils that may alter subsequent immune function. Notably, in humans, major localized trauma is known to induce systemic inflammation and multiple organ failure, which is a major cause of post-injury death [38–40]. Moreover, neutrophils are released into the circulation from the bone marrow and sequestered in end organs such as the liver and lung during multiple organ failure after major injury [38]. One can speculate that this neutrophil sequestration in end organs is reminiscent of the wound-provoked caudorostral mobilization of neutrophils that we describe in the current study. Thus, we speculate that the underlying mechanisms of dispersal of neutrophils away from the wound and the caudorostral mobilization might be implicated in the pathogenesis of systemic inflammation in mammals.

In summary, we have developed a platform that provides a powerful tool for understanding inflammation and resolution in vivo. The genetically encoded photolabeling is also applicable to spatiotemporal analysis of other cell lineages in vivo. By using the photolabeling tool, we have shown that tissue insult induces acute repetitive forward and reverse migration of neutrophils between the wound and the vasculature, followed by neutrophil distribution throughout the body. We hypothesize that neutrophil reverse migration has two roles in response to tissue damage: local resolution at injured tissue and systemic dissemination of wound-sensitized neutrophils that may affect systemic immunity.

AUTHORSHIP

S.K.Y. designed and performed experiments, analyzed data, and wrote the paper; A.H. supervised the project.

DISCLOSURE

The authors declare no competing financial interests.