Dynamics of neutrophil migration in lymph nodes during infection

An experimental model to examine neutrophil migration in lymph nodes

In order to examine the behavior of neutrophils in lymph nodes during infection, we infected mice with the intracellular protozoan parasite, Toxoplasma gondii. Following injection of live fluorescent parasites into the earflap, parasites can be found within one hour in the subcapsular sinus of the draining lymph node in association with LYVE1+ lymphatic vessels (Fig 1A). Many parasites are found within the CD169+ macrophages that line the subcapsular sinus (Fig 1Bi), a distribution similar to that seen with other particulate antigens, such as viruses and immune complexes (Carrasco and Batista, 2007; Junt et al., 2007; Phan et al., 2007).

In order to track neutrophils relative to Red Fluorescent Protein (RFP)-labeled parasites we used reporter mice in which GFP is under the control of the lysozyme M promoter (lysGFP) (Faust et al., 2000). While this reporter is also expressed by macrophages and a subset of dendritic cells, the vast majority of the GFP^high cells in the lymph nodes at 1-5 hours after infection are neutrophils, based on their cell surface phenotype (Ly6G^high, CD11b^high, CD11c^low) (Rydstrom and Wick, 2007; Sasmono et al., 2007) (Fig S1A). Neutrophil recruitment to the draining lymph nodes occurred rapidly and was dependent on the TLR/IL1R adaptor protein, MyD88 (Fig. S1A). Interestingly, many neutrophils were associated with LYVE1+ lymphatic vessels (Figure 1C). This is consistent with the possibility that neutrophils entered the lymph node via lymphatics, as has been reported previously (Abadie et al., 2005; Maletto et al., 2006). We also detected neutrophils within blood vessels of infected lymph nodes (data not shown), suggesting that neutrophils may enter lymph nodes via both blood and lymph.

At early times post infection, neutrophils contained proportionally more parasites compared to macrophages and dendritic cells (Fig. S1B). Although phagocytic destruction of pathogens is one protective mechanism used by neutrophils, the parasites inside neutrophils appear intact (Fig 1Bii). This is consistent with evidence that T. gondii can divide in vitro in human neutrophils (Channon et al., 2000; Nakao and Konishi, 1991b), and indications that the protective effects of neutrophils during T. gondii infection are due to immunoregulation rather than direct killing mechanisms (Bennouna et al., 2003; Bliss et al., 2001; Denkers et al., 2004; Sayles and Johnson, 1996).

Neutrophils form dynamic swarms in the subcapsulary sinus of the lymph node

Analysis of tissue sections of lymph nodes from infected mice showed that reporter positive cells formed large clusters along the subcapsular sinus that generally coincided with the location of parasites (Fig. 1 C, D). Virtually all of the reporter bright cells in the subcapsular sinus (SCS) region of the lymph node express Ly6G (Figure 1D), consistent with flow cytometric analysis (Fig. S1A) and confirming their identity as neutrophils.

To examine the dynamics of neutrophil cluster formation and the relationship of cluster formation to infection, we examined intact draining lymph nodes of lysGFP reporter mice infected with RFP parasites using 2-photon scanning laser microscopy (TPSLM). Visual inspection of time-lapse imaging data revealed the striking dynamics of neutrophil swarm formation (Fig 2A and Videos 1-3). Some swarms, which we term transient swarms, grew by rapid, large-scale, coordinated migration of neutrophils into the swarm and then quickly dissolved as neutrophils migrated out to join nearby growing swarms (Fig 2A and Suppl Video 1,2). Transient swarms formed and dispersed over a period of 10-40 minutes (mean duration 20 minutes) and remained relatively small (<4 × 10⁴ microns³, corresponding to approximately 150 neutrophils). We also noted a distinct type of swarm, which we term persistent swarms, that grew throughout the imaging period (up to 38 minutes) both by continued migration of neutrophils into the swarm and by merging with nearby smaller swarms (Figure 2A and Suppl Video 3). Persistent swarms tended to be large and occasionally grew to fill the entire imaging volume (>6 × 10⁵ microns³). The relationship between the size and persistence of the swarms is consistent with the notion that neutrophils themselves generate signals that induce swarming and that, once swarms reach a certain size they produce a large signaling center that can overwhelm competing nearby signals.

Out of a total of 40 runs representing ∼20 hours of cumulative imaging time, we observed 30 swarms, of which 16 were transient. We also observed neutrophil recruitment and swarming behavior in lymph nodes following earflap infection with Listeria monocytogenes (data not shown). This indicates that swarming is not unique to T. gondii infection, but may be a more general response to infection. While we occasionally observed neutrophil swarms in uninfected lymph nodes (2 persistent and 0 transient swarms seen in 14 runs), these were infrequent compared to infected lymph nodes. This, together with the clear correlation between neutrophil recruitment and infection seen from analysis of tissue sections and flow cytometry (Fig. 1 and S1), indicates that most neutrophil swarms form as a consequence of infection.

Quantitation of directed, coordinated migration by neutrophils

To relate neutrophil migration patterns to swarm formation, we tracked individual neutrophils before they entered swarms, or during migration without swarming (Fig. 2BE and Suppl. Video 4, 5). Neutrophils that were not in swarms migrated with average speeds of 11.9 micron/min, substantially faster than a recent report of neutrophil migration in the footpad (Zinselmeyer et al., 2008), but similar to that of naive T cells in intact lymph nodes ((Miller et al., 2002) and data not shown). Neutrophils containing parasites were only slightly slower that those that did not contain parasites (Fig. 2B).

In principle, swarms could form either by directed migration, or by random migration and retention of cells at sites where swarms were growing, two possibilities that can be distinguished by dynamic imaging. Visual inspection of time-lapse runs in which neutrophil swarms were forming showed large-scale directed migration of neutrophils into swarm center (Suppl Videos 1-3). To quantitate this directional migration during swarm formation we measured the distance of neutrophils to the swarm center, and plotted the changes in this distance for individual tracks over time. These analyses confirmed that individual neutrophils moved persistently toward the swarm over time, indicative of directed migration (Fig. 2C). Persistent movement toward the swarm centers could be seen even when neutrophils were >70 microns away from the swarm center. As an alternative method for quantitating directed migration, we also calculated the angle (psi) defined by the migration trajectory vector and the vector between initial neutrophil position and swarm center, and plotted the average psi as function of distance to swarm center (Fig. 2D). The mean value for psi should be 90 degrees for random movement, and less than 90 degrees for directed migration into swarms. We find that mean psi is significantly less than 90 degrees even for cells at distances >75 microns away from the swarm center. Together these analyses indicate that the growth of swarms occurs by large-scale directed migration of neutrophils toward swarm centers.

We noted that neutrophils tended to migrate in “streams” with multiple neutrophils following parallel paths while entering or leaving swarms (Suppl Videos 1 and 6, 7). To quantitate this phenomenon, we calculated the angle between the migration trajectories of pairs of cells (alpha) and plotted mean alpha as a function of spatial distance between cells in the pair (Fig. 2E). For runs in which swarming was not observed, the mean alpha was close to 90 degrees, consistent with lack of coordinated movement. In contrast, for runs in which swarms formed, alpha was significantly less than 90 degrees, a trend that could be seen even for cell pairs that were >100 microns apart. This streaming behavior could reflect communication between migrating cells, as been described for swarm formation in Dictyostelium (Kriebel et al., 2003) and/or may reflect a common response to competing attractive signals from fluctuating swarm centers.

Neutrophils swarms are initiated by pioneer neutrophils and parasite egress

To obtain clues about the signals that initiate swarm formation, we carefully examined parasite and neutrophil behavior during each recorded example of swarm formation. In some cases swarms formed in regions in which no parasites were visible and in absence of any obvious initiating event. However, in 40% (9/22) of swarm initiation events examined, swarm formation occurred in 2 distinct temporal stages (Fig. 3, Fig S2, and Suppl Videos 6-9). In these examples, a few “pioneer” neutrophils formed a small cluster, followed a few minutes later by large-scale migration of cells into the cluster. Importantly, during the initial phase of swarm formation, some neutrophils can be seen migrating randomly past the swarm center and do not begin their direction migration toward the swarm until several minutes after the arrest of the pioneer neutrophils (Fig. 3, Fig S2, and Suppl Videos 6 and 8). This suggests that only the pioneer neutrophils responded to the initial signal, and that the late-arriving neutrophils were responding to an amplified signal generated by the pioneers.

Another clue to the initiation of swarm formation came from examination of runs with particularly heavily infected lymph nodes. In these samples, we occasionally observed groups of closely apposed non-motile parasites that suddenly became motile and migrated rapidly away from the group (Fig. 4A, Fig S2A, and Suppl Videos 9-11). Such behavior is typical for parasite egress from infected cells leading to cell lysis and invasion of neighboring cells (Black and Boothroyd, 2000). Interestingly, egress coincided closely in space and time with the initiation of a swarm in most (4/5) examples observed. The response of neutrophils to parasite egress was extremely rapid, with directed migration detectable at the same time points or even seconds before increased parasite motility was first detectable (Fig. 4B, Fig. S2B).

Neutrophil swarms lead to removal of subcapsulary sinus macrophages

Because neutrophils are known to degrade tissues by releasing matrix metalloproteinases, the appearance of swarms in the subcapsular sinus of lymph nodes raised the possibility that neutrophil recruitment to lymph nodes during infection could alter lymph node structure. Indeed we find that, while uninfected lymph nodes have a continuous layer of CD169+ macrophages along the lymph node sinus (Fig. 5A left panels), infected lymph nodes show gaps in the CD169 staining that often coincides with the location of neutrophil clusters (Fig 5A, B right panels, arrows). We also see a loss of CD169+ cells by flow cytometric analysis of infected lymph nodes (Supplemental Figure 3), suggesting that SCS macrophage did not migrate elsewhere in the lymph node. In vivo depletion of neutrophils prior to infection largely prevented the appearance of gaps in the layer of CD169+ macrophages, even in regions of the subcapsular sinus that were heavily infected (Figure 5C). Thus although parasite egress also produces some cell lysis, neutrophil swarms, rather than infection per se, were primarily responsible for the gaps in the layer of SCS macrophages reported here. To examine the temporal relationship between swarming and the removal of subcapsular sinus macrophages we visualized neutrophil migration in real time in samples from lysGFP reporter mice that had been injected with fluorescent CD169+ antibodies along with RFP-labeled parasites (Fig. 5D and Suppl Video 12). At the beginning of the run, a continuous region of CD169 staining is visible, and after a few minutes a transient swarm appears and dissolves, leaving behind a gap in the CD169 staining in the same location as the transient swarm. Because spectral overlap between the GFP from neutrophils can obscure the loss of CD169 staining, we also examined lymph nodes from infected mice in which CD169 cells were labeled but neutrophils were not (Fig. 5E and Suppl. Video 13). In these samples we also observed regions of CD169 clearing with a size and rate of appearance that suggested that they coincide with neutrophil swarms. Together these data indicate that neutrophil swarms lead to the removal of subcapsular sinus macrophages.

Neutrophil swarming and subcapsulary sinus macrophage removal also occur in mesenteric lymph nodes following oral infection

To confirm our observations using a more physiological route of infection, we infected LysGFP reporter mice orally with cysts generated from RFP-expressing parasites, and examined neutrophil recruitment, localization, and migration in mesenteric lymph nodes. While few neutrophils were detected in mesenteric lymph nodes of uninfected mice, neutrophils (GFP^high Ly6G⁺) were readily detectable by day 4-5 following oral infection (Fig. 6A, B). Neutrophils form clusters throughout the lymph nodes of infected mice including some near the subcapsular sinus (Fig. 6B, white arrows). TPSLM of intact mesenteric lymph nodes revealed migration dynamics of reporter positive cells similar to that observed in draining lymph nodes following earflap infection (Suppl Video 14). Moreover, we also detected gaps in the layer of CD169+ SCS macrophage in mesenteric lymph nodes of infected mice, and these often co-incided with the location of clusters of reporter positive cells (Fig. 6C, white arrow). It is noteworthy that, while parasites and neutrophils are concentrated at the SCS of draining lymph nodes following earflap infection, parasites and neutrophils are found both in the SCS and in deeper regions of the mesenteric lymph node following oral infection. This is likely to reflect the less synchronous arrival of parasites in mesenteric lymph nodes following oral infection compared to earflap infection. Nevertheless, these results confirm that the recruitment and behavior of neutrophil in the SCS of the lymph node described here are not unique to the earflap infection model, but also occur following oral infection with T. gondii.

Mice

All mice were bred and housed in pathogen-free conditions at the AALAC-approved animal facility at Life Science Addition, University of California, Berkeley. All animal experiments were approved by the Animal Care and Use Committee of UC Berkeley. LysGFP reporter mice were a gift from Thomas Graf (Albert Einstein College of Medicine, Bronx, NY), and have been previously described (Faust et al., 2000). MyD88 gene KO mice (originally generated in Dr. S. Akira’s laboratory, Osaka University, Osaka, Japan) (Adachi et al., 1998) were backcrossed onto C57BL/6 background.

Parasites

To generate a T.gondii cell line expressing the tandem (td) Tomato variant of red fluorescent protein (RFP) (Shaner et al., 2004) we amplified tdTomato using the primers 5′-AGTCCCTAGGGTGAGCAAGGGCGAGGAG and 5′AGTCCCCGGGCTTGTACAGCTCGTCCATGC. We digested the resultant PCR product with AvrII and XmaI and ligated this into the corresponding restriction enzyme sites of the pCTG vector (GvD and R. Opperman, unpublished) to generate the vector pCTR_2T, where expression of cytosol-localised tdTomato is driven from the constitutive α-tubulin promoter. We transfected this construct into the RHΔhxgprt strain of T. gondii and generated stable lines expressing tdTomato by chloramphenicol selection as previously described (Gubbels et al., 2005). Briefly, 1 × 10⁷ parasites were resuspended in cytomix containing 50 μg of linearized pCTR_2T vector. The parasites mixture was added to a 2 mm electroporation cuvette and electroporated using a BTX ECM 630 electroporator at 1500 V, 25 Ω and 25 μF. Transfected cells were passed onto fresh host cells. 16 hours post-transfection, chloramphenicol was added to a final concentration of 6.8 μg/mL. Cells were passaged in the presence of chloramphenicol until stable lines were generated. Clonal lines of the RH/tdTomato parasite strain were obtained by fluorescence-activated cell sorting. Briefly, parasites were filtered through a 3 μm filter, pelleted and resuspend in phosphate-buffered saline. Parasites were sorted using a MoFlo cytometer (Dako, Ft Collins, CO) with an Enterprise 631 laser tuned to 488 nm for excitation and an emission filter with a band pass of 570/40 nm. Bright parasites were sorted into 96 well plates, and wells containing single plaques after one week of growth were considered to consist of a clonal parasite line. All parasites were maintained in confluent human foreskin fibroblasts.

For mouse infections, parasites were prepared from almost fully lysed fibroblast cultures by first releasing the parasites from fibroblasts by passing them through 21g 1 ½ gauge and 23g1 needles 5-10 times. Parasites were then filtered through a 3 μm filter, pelleted and resuspended in phosphate-buffered saline. 5 ×10⁵-1 × 10⁷ parasites (typically 5 ×10⁶) in 10 μl volume were injected into the earflap. No differences in the anatomical distribution of parasites or neutrophils in draining lymph nodes were observed over this range of infection. Some of the mice used for 2-photon experiments were injected with 10⁶ parasites intraperitoneally at least 4 weeks prior to infection and imaging. No differences in neutrophil behaviour were observed between immunized and naïve animals. For neutrophil depletion experiments, mice were injected i.p. with 1 mg of Ly6G antibody (clone 1A8) two days prior to infection and also i.v. with 0.5 mg of Ly6G antibody one day prior to infection.

For oral infections, Prugnaud-RFP parasites were engineered essentially as described above for RH parasites, and cysts were isolated from brain homogenates of CBA/J mice (Jackson Laboratory, Bar Harbor, Maine) infected with 300-400 parasites i.p. 4-6 months prior. Cysts were counted after staining with Dolichos Biflorus Agglutinin (Vector Laboratories, Burlingame, CA). LysGFP reporter mice were infected with 20-45 cysts by gavage. Mesenteric lymph nodes were used for microscopy or analysed by flow cytometry.

Antibodies and flow cytometry

Lymph nodes were dissociated by collagenase digestion. Cell suspensions were filtered, stained and analyzed by flow cytometry. The following antibodies were from eBioscience: phycoerythrin-Cy5 conjugated anti-CD11b (clone M1/70), and phycoerythrin-Cy5 conjugated anti-CD11c (clone N418). Fluorescein isothiocyanate-conjugated anti-Ly6G (clone 1A8) was from BD Biosciences. Acquisitions were performed with a Coulter Epics XL-MCL flow cytometer (Beckman-Coulter) and data were analyzed with the FlowJo software (Tree Star, Ashland, Oreg.).

Statistical analysis

Values were expressed as mean ± standard error (SD). Levels of significance were calculated by unpaired t tests using the GraphPad Prism program (San Diego, CA, US). Differences were considered significant at P<0.05.

2-Photon Imaging

One to five hours following ear flap injection, mice were sacrificed, and dorsal cervical (draining) lymph nodes were isolated and imaged by Two-Photon Laser Scanning Microscopy (TPLSM) while being perfused with warmed, oxygenated medium as described previously (Witt et al., 2005). Imaging was performed using either upright Zeiss NLO 510 with a Spectra-Physics MaiTai laser tuned to 900-920nm. In that case GFP and tdTomato emission light was separated using a 560 dichroic mirror and collected using non-descanned detectors. Alternatively, 2-photon imaging was performed using a custom-built microscope with a Spectra-Physics MaiTai laser tuned to 900-920nm, the emission light was separated using 495 and 560 dichroics and collected using 3 detectors. Bandpass filters HQ 450/80M and HQ 645/75M were used to minimize spectral overlap.

Typical imaging volumes (164 × 164 × 40 microns or 172×143×80 microns respectively) corresponded to regions of the lymph nodes extending up to 200 microns below the surface of the capsule and were scanned every 13-37 seconds for 20-40 minutes. Mesenteric lymph nodes were imaged under similar conditions 4-5 days after oral infection.

Data Analysis

The x,y,z co-ordinates of individual cells over time were obtained using Imaris Bitplane Software. Motility parameters were calculated using Matlab (code available upon request). Parameters reported here include Speed (defined as path length over time; microns/minute) and distance to swarm center (microns). To measure the volumes of individual neutrophil swarms, isosurfaces were generated in the GFP channel (threshold of 130, 6 μm Gaussian filter), and the volume of each isosurface was determined over time. The volume of a single neutrophil was estimated using the same method (using 1 μm Gaussian filter), and 15 individual cell volumes were averaged.

Immunofluorescence

Isolated lymph nodes were fixed with 4% formalin in PBS for 1 hr, sequentially submerged in 10, 20, 30% sucrose for 18-24 hrs each, and frozen over dry ice in OCT. 20 micron serial sections were generated by cryosectioning (MICROM H550,Microm GmbH, Walldorf, Germany) and stored at -80°C. Sections were brought to room temperature, fixed with cold acetone for 10 min, air dried and incubated with 10% mouse serum in Fc blocking reagent (2.4G2 culture supernatant) for 1hr. The tissue was stained with unconjugated anti-Ly6G (BD Biosciences), purified rat anti mouse CD169 (AbD Serotecor), or biotin-conjugated anti-LYVE-1 (R&D systems) overnight at 4°C. After primary staining slides were washed 4x in PBS and incubated with either anti-rat Alexa 647 or streptavidin Alexa 633 (Invitrogen) for 2 hrs at room temperature. Sections were then washed 4x in PBS and coverslipped using VectaShield (Vector Laboratories, Burlingame, CA) mounting medium. Stained lymph node sections were visualized on Zeiss 510 Axioplan META NLO upright microscope with a 10x air objective (Plan-Neofluar 10x/0,3) and a 40x oil objective (Plan-Neofluar 40x/1.3 oil WD=0.17mm) using 488nm, 543nm and 633nm laser lines. Images were analysed and assembled using Adobe Photoshop and Imaris Bitplane.