|Home | About | Journals | Submit | Contact Us | Français|
Following entry into lymph nodes (LNs), B cells migrate to follicles, whereas T cells remain in the paracortex, with each lymphocyte type showing apparently random migration within these distinct areas. Other than chemokines, the factors contributing to this spatial segregation and to the observed patterns of lymphocyte movement are poorly characterized. By combining confocal, electron, and intravital microscopy, we show here that the fibroblastic reticular cell (FRC) network regulates naïve T cell access to the paracortex and also supports and defines the limits of T cell movement within this domain, whereas a distinct follicular dendritic cell (FDC) network similarly serves as the substratum for movement of follicular B cells. These results highlight the central role of stromal microanatomy in orchestrating cell migration within the LN.
LNs are situated at the interface of the blood and lymphatic systems, where they contribute to efficient initiation of adaptive immune responses by bringing cell-associated and soluble antigens draining from peripheral tissues together with circulating lymphocytes entering from the blood (Gretz et al., 1997; Gretz et al., 1996). Two key questions about cell behavior within LNs are how lymphocyte migration is controlled within this densely packed environment and what chemical and/or structural cues guide the compartmentalization of lymphocytes into the follicles (B cell areas) and paracortex (T cell area). Recent explant and intravital imaging studies of LNs have documented rapid movement of both T and B lymphocytes within their respective zones of enrichment (Miller et al., 2002; Wei et al., 2003). For T cells in particular, it has been proposed that such movement is random, allowing these cells to scan without prejudice the many dendritic cells in the paracortex for those bearing relevant antigen (Bousso and Robey, 2003; Miller et al., 2004; Miller et al., 2003; Wei et al., 2003). However, naive T cells show acute turns at the border of B cell follicles back into the paracortex, as well as frequent shifts in their direction within the latter region, behaviors that could indicate an influence of unseen physical and/or chemical guidance cues on their migration.
LNs have a highly organized and complex architecture composed of distinct cellular compartments and structures, at the heart of which is a non-haematopoietic cell backbone (Gretz et al., 1997; Kaldjian et al., 2001). Among the cells that are critical for generating this backbone are the fibroblastic reticular cells (FRCs) present in the T cell area (Katakai et al., 2004a) as well as the follicular dendritic cells (FDCs) present in the B cell follicles (Cyster et al., 2000). FRCs ensheath and produce the basement membrane molecules of the conduit system that allows the transport of lymph-borne molecules from the LN capsule through the cortex. FRCs also provide an adhesive substratum for LN-resident DCs that sample the conduit’s lymph content (Katakai et al., 2004b; Sixt et al., 2005). Other possible functions of these FRCs remain largely unexplored, in particular their potential contribution to guiding and constraining T cell movement.
Based on its dimensions and location, as well as its expression of adhesive and chemotactic molecules recognized by naïve T cells, several investigators have proposed that the FRC network might support T cell motility in the paracortical area of the LN (Gretz et al., 1997; Gretz et al., 1996; Katakai et al., 2004b). However, this hypothesis derives from observations made by static imaging as well as from in vitro chemotactic experiments that do not mimic the complex nature of the LN microenvironment. Given the dense nature of both the FRC network and T cell population, static images alone cannot determine whether contacts between lymphocytes and network fibers produce adhesive forces utilized for T cell motility or are merely the result of close packing of the two within the LN environment. In vitro, T cells can move on various substrates such as fibronectin coated surfaces or 3D collagen matrices (Crisa et al., 1996; Friedl et al., 1998; Stachowiak et al., 2006). However, collagen within the paracortical regions of the LN is located almost exclusively within the conduit system, enwrapped by FRCs, with no more than 10% of such molecules accessible to lymphocytes (Hayakawa et al., 1988). Thus, the exact contribution of the LN FRC network to T cell migration has not been established, in particular, whether FRC fibers actively guide T cell movement or simply pose physical barriers to free migration.
While the presence of the FRC network within the LN paracortex makes it a candidate for regulating T cell migration, its absence from the follicles of secondary lymphoid organs suggests that B cell motility may be influenced by other structures. Located in the B follicles, FDCs form a distinct network of non-hematopoietic stromal cells and play a critical role in presenting antigen to B cells (Tew et al., 1990). In addition, these cells are the main producers of CXCL13 (Gunn et al., 1998; Munoz-Fernandez et al., 2006) and BAFF (B-cell-Activating-Factor) (Hase et al., 2004), two soluble factors implicated, respectively, in the homing of naïve B cells to follicles and in B cell survival (Gunn et al., 1998; Schiemann et al., 2001). Beyond these functions, the potential contribution of FDCs to guiding and constraining B cell movement within the follicle has not been assessed.
Cell-cell interactions and cellular compartmentalization are critical for processes such as efficient development of adaptive immune responses, induction of peripheral tolerance, and homeostatic maintenance of lymphocyte populations. Understanding the relative roles of free migration vs. mechanically or chemically (Castellino et al., 2006) guided cell trafficking is thus essential to developing a better picture of how these events are regulated. Here we examine the possibility that the extensive networks of FRCs within the T cell area and FDCs within the B cell follicles play central roles in controlling the nature and scope of T and B cell movement within the LN environment.
In the paracortex, FRCs that express the intermediate filament desmin ensheath the 3-dimensional (3D) conduit system, whose outline can be delineated by its reactivity with the monoclonal antibody ERTR-7 (Figures 1A and 1B; Movie S1) (Katakai et al., 2004a; Sixt et al., 2005). As a first approach to determining the possible influence of FRCs on T cell migratory behavior, we transferred CFSE-labeled T cells intravenously (i.v.) into recipient mice that were subsequently anesthetized and perfused with fixative 12 hours later. We found that this protocol results in preservation of cell morphology that is lost with conventional methods that fix tissues after their isolation. LNs were excised from the perfused animals, sectioned, stained for desmin and ERTR-7 expression, and subsequently analyzed using confocal microscopy (Figure 1C, Movie S2). Consistent with published explant (Hugues et al., 2004; Miller et al., 2002) and intravital (Mempel et al., 2004; Miller et al., 2003) imaging data showing that most lymphocytes are actively migrating within LNs under physiological conditions, the large majority of the T cells in the in situ fixed samples exhibited a polarized, elongated shape with a higher intensity of CFSE in the region of the cell body corresponding to the uropod (Figures S1A and S1B; Movie S3). The presence of a uropod is characteristic of motile cells (Ratner et al., 2003). Among such polarized cells, 298 out of 300 T cells examined were immediately adjacent to or touching an ERTR-7+ fiber (Figure 1C).
The FRC network is quite dense, with the distances between intersections of FRC strands ranging from 5 to 37 μm (17.26 +/− 6.93 μm, n=6 LNs), a space that could accommodate between 1 to 5 naïve lymphocytes (Movie S1) (Gretz et al., 1997; Ushiki et al., 1995). Therefore, the observed association between T cells and FRCs might not reflect specific interactions but merely a close proximity of the two, based on a random distribution of lymphocytes packed into the FRC network. As a first approach to discriminating between these possibilities, we used electron microscopy to analyze LN sections obtained from fixative-perfused animals, a procedure that preserves the microvilli usually covering naïve lymphocytes within the LN (Nijhara et al., 2004; Ushiki et al., 1995). Scanning electron microscopy (SEM) of LNs clearly identified lymphocytes with numerous microvilli. In addition, fine structural elements were visualized, including conduits comprised of collagen bundles enveloped by FRCs (Figure 1D, Figure S2). Because of the dense cell packing in the LN, only rare contact points between lymphocytes and FRCs were unambiguously observed at the imaging angle necessary to determine if individual lymphocytes truly contact adjacent stromal cells. Nonetheless, among the 24 T cells that were appropriate for addressing this issue, all showed microvillus protrusions towards the FRCs. These results are consistent with the hypothesis that T cells specifically interact with FRCs while trafficking in LNs. However, due to the dimensions of the FRC network and the large quantity of enmeshed lymphocytes, these static images alone do not adequately distinguish between a model in which a T cell transiently binds to an obstructing fiber found along a path of free migration and one in which the T cell uses the fiber to guide its directionality in a positive manner. Images of live T cells actively migrating within the FRC network were needed to better distinguish between these alternatives.
To investigate the dynamics of naïve T cell interactions with the FRC network, T cells were visualized in the LNs of anesthetized mice using two-photon (2P) laser scanning microscopy. To image non-hematopoietic stromal cell populations within the LN, we generated chimeric mice by using wild-type bone marrow cells to reconstitute irradiated ubiquitin promoter-GFP transgenic animals in which the fluorescent protein is expressed by all nucleated cells (Figure 2A) (Schaefer et al., 2001). Animals were allowed to reconstitute for a minimum of eight weeks to ensure nearly complete replacement of hematopoietic-derived cells with non-fluorescent populations. To confirm that GFP-expressing cells within the LN T cell area of chimeric animals represented the FRC population, LN sections from such animals were stained for desmin and ERTR-7 expression and analyzed using confocal microscopy (Figure 2B, Movie S4). In the T cell paracortex, GFP+ cells created a 3D network that surrounded the ERTR-7+ conduit system and overlapped with desmin staining, indicating that this network was indeed formed by FRCs.
To analyze the dynamic behavior of T cells in relation to this green fluorescent FRC network, T cells were labeled with the red fluorescent dye SNARF-1 and injected i.v. into chimeric animals. Four hours later, when many fluorescent T cells were located in the outer paracortex at a depth accessible to imaging by intravital 2P microscopy, recipient mice were anesthetized, and a popliteal LN surgically exposed for intravital imaging (Figure 2C, Movies S5 and S6). For most data sets, the imaged volume was small (100 x 100 x 10μm) and the images were acquired at low scan speed (166 or 300 lps) with a small step size in the z dimension (1.5 to 3μm). These specific settings were used to optimize detection of the thin (1–3μm) and densely packed FRC processes. Analysis of 4D (x,y,z and time) datasets suggested that migrating SNARF-1 labeled T cells actively crawled on GFP+ FRCs, following and morphologically adapting to the paths established by the cell bodies and extended processes of these non-hematopoietic cells. Occasionally, some T cells in the processed image stacks seemed to be moving in an “empty” space in which no FRC processes were visible. When such behavior was observed in the intermediate z planes in the imaged volume, detailed analysis revealed that the T cells were still in contact with faint GFP+ extensions of FRCs extending under or above the field of view (Figure S3A, Movie S7). Migrating T cells could be observed to extend filapodia from their leading edge and to probe intersecting fibers diverging from a point in space, followed by T cell movement along one of the fibers that had been examined by the cell extensions (Figure S3B, Movie S8). These data suggest that the T cells actively follow the fibers, rather than merely finding pathways around potential obstructions.
To assess quantitatively whether T cells actively followed the paths laid out by the FRC network, we assumed that if the fibers provided guidance for cell movement, then any directional turns made by a T cell should always be associated with a corresponding turn or branch of a supporting fiber. Conversely, a lack of correspondence between T cell directionality and fiber pathways would indicate that spontaneous turns or physical impediments posed by the many other cells in the densely packed LN environment (Figures S4A and S4B) accounted for T cell directional changes. Analysis of T cell turns using the described method (Figure S4C) revealed a 92.7% correlation (203 out of 219 cells) between changes in T cell direction and the presence of T cell-associated GFP+ FRC fibers running at the corresponding angle (Figures 2C, 2D and S5A). In some rare cases, highly mobile T cells with elongated shapes were seen “jumping” between adjacent but non-intersecting fibers (Figure S5B), probably because of the presence of non-fluorescent cellular obstacles such as DCs in their path. During these “jumps”, T cells were always found to contact a fiber either via their uropods or lamellipodia and to continue their movement in a more conventional way once the ”jump” had been negotiated. Because these events were rare despite the extended length of motile T cells, they suggest that the lymphocytes preferentially follow the FRCs rather than jumping from one fiber to another.
These new static and real-time imaging data provide both a cellular and structural explanation for previous reports suggesting that T cell motility is random in the region of the paracortex typically imaged using 2-photon microscopy and that the crawling T cells frequently make sudden turns in the LN paracortex (Miller et al., 2002; Wei et al., 2003). The FRC network, by virtue of its dimensions and the presence of frequent crossing points, supports this apparent random migration, constituting an intersecting roadway for T cell migration. Indeed, the dimensions of the network and the location of crossing points of FRC fibers correspond to the distance a T cell moves on average before showing turning behavior (Miller et al., 2002; Wei et al., 2003). Furthermore, intravital 2P imaging studies have revealed a dense network of dendritic cells (DCs) in the T cell zone of the LN (Lindquist et al., 2004). Other analyses have shown a close physical relationship between DCs and at least the FRC sheath of conduits, which we confirm here (Figure S6) (Katakai et al., 2004a; Sixt et al., 2005). These data indicate that as T cells move along the FRC fibers, the presence of DCs on these same fibers could enhance the interaction frequency of these two cell types, allowing an efficient presentation of peripheral or lymph-borne antigen content to motile lymphocytes.
Within the LNs of chimeric mice, GFP was also expressed by the radio-resistant vascular endothelial cells of blood vessels including HEVs (Figure 3A). Analysis of stained sections from the LNs of the chimeric mice revealed a layer of desmin+ GFP+ FRCs that surrounded the ERTR-7+ conduit system, which in turn enveloped the PNAd+ (Peripheral Lymph Node Addressin) HEVs with their characteristic thick, “cobble-stone” endothelial cell lining (Figure 3B). Because the FRCs appeared to form a cellular sheath around HEVs, we hypothesized a role for these cells in regulating T cell entry into the LN parenchyma. To study this issue, SNARF-1 labeled T cells were injected i.v into a chimeric mouse whose popliteal LN had previously been surgically prepared for intravital imaging. Using this protocol, the LN could be imaged immediately after T cell transfer, permitting visualization of T cell behavior during diapedesis across the HEV and subsequent migration into the surrounding paracortical parenchyma (Figure 3C and Movie S9). Approximately fifteen minutes after transfer, T cells were visualized exiting the HEVs, a behavior that continued throughout the imaging period (~1 hour). Strikingly, T cells did not gain access to the parenchyma using random points of egress along the entire length of the vessel. Rather, many T cells followed each other out through discrete regions (“exit ramps”). As soon as they exited the HEV via these gaps in the investing FRC sheath, the T cells attached to GFP+ FRC processes and began migrating within the paracortical parenchyma along these paths (Movie S10).
The limited set of sites through which T cells emerged into the LN from an HEV could represent, stably differentiated structural features of the FRC network surrounding the HEVs or simply the junction between two FRCs. While data obtained from intravital imaging (Movie S9) showed what appeared to be ‘holes’ in the FRC sheath even in the absence of an extravasating fluorescently-labeled T cell, these lucent structures might also represent the location of a high flux of non-fluorescent endogenous T cells exiting the HEV. Using confocal analysis of LN tissue sections, we were unable to detect well-defined, specific ‘holes’ in the ring of FRCs that enveloped the HEVs (Figure 3B). Electron microscopy was then utilized, allowing high resolution analysis of the various elements present at the exit sites (Figure 3D). HEVs were easily recognized by their shape, as were the lymphocytes moving in the perivascular channel (PVC), which is a narrow and constrained region through which T cells squeeze for 10–100 min prior to exiting into the parenchyma (Gretz et al., 1997; Jansen et al., 1962; Matous-Malbohan and Arnason, 1974). In agreement with our static confocal imaging analysis, the exits did not appear to correspond to constitutively present exit ports in the ring of FRCs, but rather seemed to represent unstructured openings in the junctions between individual FRCs. This is most clearly demonstrated by images showing T cells that had almost completed exit out of the PVC squeezing between two adjacent FRCs. Together, these findings indicate that the FRC network surrounding the HEVs restricts T cell exit from HEVs. As they complete their traverse of this cellular barrier, T cells move along the fibers for migration within the paracortical zone.
After accessing the parenchyma, T cells remain in the paracortex while B cells leave this region and localize to peripheral B cell follicles. A previous study has revealed that migrating paracortical T cells and B cells already resident in follicles each turn acutely when they reach the T/B border, showing little or no cellular trespass into the neighboring area (Wei et al., 2003). The mechanisms underlying this behaviour are incompletely defined, although chemokines and chemokine receptor sensitivity have been shown to play important roles in promoting the follicular localization of B cells (Ansel et al., 2000). Because FRCs are the substratum for T cell motility and are absent from the B cell follicles (Figures 1A and 1B), we hypothesized that they might contribute to defining T cell territoriality. In the B cell areas of chimeric LNs, a second GFP+ network distinct from the FRC network was present (Figure S7A). This follicular stromal network did not surround the ERTR-7+ conduit system, but could be stained by FDC-M2, an antibody specific for follicular dendritic cells (FDCs). This made it impossible, using intravital 2P imaging alone, to distinguish the adjacent FDC and FRC networks that are both present at the T/B border and express GFP in chimeric LNs. However, confocal imaging of LN tissue sections revealed a ~50 μm region of T/B cellular interspersion at the paracortex-follicle border over which the density of the FRC network decreased and then terminated (Figure S7B, left panel). Therefore, we sought to determine if the T cells within this region still adhered to the FRC network. LNs of recipient mice that had received CFSE-labeled T cells 12 hours earlier were sectioned and stained for B220 and ERTR-7. Because ERTR-7+ conduits are always surrounded by FRCs, this staining allowed us to determine if T cells that had apparently sojourned into the periphery of B cell follicles (B220+ area) were still associated with the sparse network of FRC fibers in this region. High-resolution imaging revealed that 80.6 % of T cells (258 out of 320) at the B cell-enriched border had clear associations with the rare FRC processes in this region (Figures 4A and 4B). In addition, preventing LN entry of blood-borne lymphocytes with a single injection of an anti-CD62L blocking antibody given 1 day prior to analysis led to the withdrawal of B cells from the terminal region of the ERTR-7+ network, but had no affect on the presence of T cells in this area (Figure S7B, right panel), suggesting that this region is actually part of the paracortex and not the B follicle proper. Therefore, in addition to supporting naïve T cell movement, our results suggest that the FRC network defines the T cell zone in the LN.
After accessing the LN via HEVs located in the paracortex, B cells move through the T zone before they enter the B cell follicles (Qi et al., 2006). We considered the possibility that prior to follicular entry, B cells would also move on the FRC network. As a first approach to testing this hypothesis, we transferred CFSE-labeled B cells intravenously (i.v.) into recipient mice that were subsequently anesthetized and perfused with fixative 24 hours later. As observed with T cells, the majority of B cells (88.2%, 225 out of 255) present in the T cell zone were found to be associated with the non-hematopoietic FRC network (Figure S8A). Because most of the B cells present in the T cell zone are adjacent to the B cell zone, an area within which both GFP+ FRC and GFP+ FDC networks overlap in the chimeric animals, it was difficult to unambiguously monitor B cell motility in the T cell zone using intravital microscopy. As a consequence, we analysed the behavior of B cells in thick living LN sections that were first stained for ERTR-7 and then imaged in warmed, oxygenated medium. Using this technique, we were able to identify the precise location of the FRC network and to follow B cell movement with respect to the ERTR-7+ fibers (Figure S8B, Movie S11). Using the turning angle analysis method, we observed an 88.2% correlation (90 out of 102 cells) between changes in B cell direction and the presence of a B cell-associated ERTR7+ FRC fiber running at the corresponding angle (Figure S8B). These findings are consistent with the idea that like T cells, B cells track along the FRC network while they are in the paracortex.
As B cells leave the T zone, they enter the B cell follicle, a structure that contains a unique stromal 3D network composed of desmin+ FDC-M2+ FDCs (Figures 5A and 5B). These stromal cells express CXCL13, the ligand for the chemokine receptor CXCR5 whose expression by naïve B cells is required for their entry into the follicle (Gunn et al., 1998). In addition, FDCs express BAFF, a B cell survival factor (Schiemann et al., 2001), and they use complement receptors to capture and display antigen to B cells (Carroll, 2000). Given the importance of these cells to B cell physiology in the follicle, we considered the possibility that B cells might preferentially adhere to and move along FDCs within LN primary follicles.
To address this issue we performed experiments similar to those described for investigating interactions between T cells and the FRC network. First, CFSE-labeled B cells were transferred i.v into recipient mice that were perfusion-fixed 24 hours later. LNs were then excised, sectioned, stained for desmin and FDC-M2 expression, and subsequently analyzed using confocal microscopy (Figure 5C). Among CFSE labeled B cells present in the B cell follicle, 97% (194 out of 200) were immediately adjacent to or touching a desmin+ FDC-M2+ fiber. Next, the dynamics of naïve B cell interactions with the radio-resistant FDC network was investigated by visualizing B cells in the LNs of anesthetized chimeric mice using 2P laser scanning microscopy. SNARF-1 (red) labeled B cells were injected i.v. into chimeric animals that were subjected to intravital imaging of the popliteal LN one day later. Surprisingly, unlike the bright immotile FRC network, the somewhat dimmer FDC network had more irregular strands that displayed a marked motility (Movie S12). Analysis of 4D datasets suggested that migrating SNARF-1 labeled B cells actively crawled on the moving EGFP+ FDCs, following and morphologically adapting to the cell bodies and extended processes of these stromal cells (Figure 6A, Movie S13). To assess quantitatively whether B cells followed the paths laid out by the FDC network, we again performed turning angle analysis involving these cell types (Figure 6B). This method revealed a 90.2% correlation (148 out of 164 cells) between changes in B cell direction and the direction and the presence of B cell-associated GFP+ FDC fibers running at the corresponding angle.
FDCs and FRCs form dense and complex three-dimensional networks within distinct geographical regions of the LN. These structural elements may either represent obstacles to free migration of lymphocytes or the underlying roads that are actively follow as the cells move, two opposing possibilities that previous in vitro and static experiments were unable to discriminate. The data presented here document that these two networks do not act as mere barriers to lymphocyte migration but rather actively support and provide direction to these motile cells.
The influence of FRCs on T cell migratory behavior begins with control of T cell access to the paracortex through the restriction of transit points from the abluminal side of HEVs. As the T cells leave these FRC-delimited exit ramps, they immediately associate with and crawl along the strands of the FRC network, which function as local roads that define the paths of T cell migration within the LN parenchyma and may facilitate T cell interaction with DCs colocalized to the same FRC network. Finally, because FRCs are almost exclusively present in the paracortex and naïve T cell movement is largely restricted to the FRC network, the latter also defines the T cell zone in the LN and is likely to play a key role in preventing naïve T cell entry into B cell follicles.
Recent work has shown that B cells can be stimulated by antigen-bearing DCs in the extrafollicular regions of the LN shortly after the lymphocytes leave the circulation and enter the paracortex (Qi et al., 2006). The present study reveals that B cells move on the FRC network while migrating from HEVs to the primary follicles. Thus, just as for T cells, the probability of encountering an antigen-loaded DC is increased by the fact that the B lymphocytes move along paths leading directly to fiber-associated DCs that have either sampled antigen from the lymph content within a conduit or have entered the LN carrying antigen from a site of peripheral infection or vaccination (Itano et al., 2003).
One possible means by which FRCs and FDCs could influence T and B cell trafficking would be through their expression of a unique molecular signature specifically recognized by these lymphocytes. FRCs express and present both adhesion ligands and chemokines that could support and guide the migration of T cells. FRCs secrete fibronectin and ensheath collagen bundles, two extracellular matrix (ECM) components that support T cell motility in vitro (Crisa et al., 1996; Friedl et al., 1998; Haston et al., 1982). However, these molecules are enclosed by the FRCs and are not accessible for direct T cell contact in the normal LN. FRCs and FDCs also express the adhesion molecules VCAM-1 (Vascular Cell Adhesion Molecule-1) and ICAM-1 (Intercellular Adhesion Molecule-1) (Katakai et al., 2004b; Koopman et al., 1994; Ogata et al., 1996), either or both of which could be used for T cell adhesion to the fibers. Both naïve T and B cells express the integrins VLA-4 (Very Late Antigen-4) and LFA-1 (Lymphocyte function-associated antigen), the respective receptors for VCAM-1 and ICAM-1, but at present we have no evidence that these integrin-ligand pairs are involved in the association of naïve T or B cells with stromal elements in the uninflamed LN.
FRCs secrete the homeostatic chemokines CCL19 (ELC), CCL21 (SLC) and CXCL12 (SDF-1), whereas FDCs produce CXCL13 (Gunn et al., 1998; Katakai et al., 2004b; Luther et al., 2000). Naive T cells express CCR7, the receptor for ELC and SLC, and CXCR4, the receptor for SDF-1, whereas naïve B cells, but not naïve T cells, express the CXCL13 receptor CXCR5 as well as CCR7 and CXCR4 (Forster et al., 1994). Interestingly, FRC and FDC not only secrete SLC and CXCL13, they are also decorated by these chemokines. Chemokines are immobilized on cells or extracellular matrix surfaces by interacting with glycosaminoglycans. Specific chemokines bind different types of glycosaminoglycans, expression of which can vary with cell type, location, and inflammatory status (Kuschert et al., 1999). By localizing particular chemokines to distinct and geographically segregated stromal cell types, haptotactic rather than chemotactic guidance cues may promote proper lymphocyte localization and guidance of trafficking within LN subregions. Further investigations will be required to determine if chemokines closely associated with FRCs and FDCs promote chemokinesis of naïve T and B cells, giving rise to the rapid motility of these cells types in LNs, and if the localization of these chemokines to FRC and FDC fibers leads to the physical guidance of locomoting lymphocytes as suggesting by our dynamic imaging data. For both T and B cells, CCR7-CCL19/21 interactions could dominate with FRCs in the T zone [although a role for CXCL12 on FDCs and CXCR4 on the T and B cells cannot be excluded as participants in this process], whereas in the B cell follicle, CXCL13-CXCR5 signaling might be most relevant. Such considerations suggest that, in addition to the positive FRC-associated factors that may help retain naïve T cells in the paracortex, the absence of follicular recruitment signals such as CXCL13-CXCR5 interactions may also contribute to steady-state T cell localization patterns. Without CXCR5 expression, it is likely that, unlike naïve B cells, naïve T cells are not induced to cross the T/B border. Instead, they may return to the T-zone via the FRC network, which may present the sole motogenic signals to T cells in that region (Ansel et al., 2000; Kim et al., 2001). For T cells, chemotactic responses to a sphingosine-1-phosphate gradient (Mandala et al., 2002; Schwab et al., 2005; Wei et al., 2005) would presumably influence the bulk movement of T cells within the FRC delimited network, which reaches the medullary region of the LN from which T cells exit to the efferent lymph. Such large scale relocalization over many hours would not be apparent in the short (typically < 2hr.) imaging experiments conducted by those laboratories reporting an absence of directional migration among naive T cells in LNs.
Using electron microscopy, Gretz et al. have detailed the structure of the FRC network and observed numerous lymphocytes filling what they have termed “corridors.” Based on these observations, the density of lymphocytes in LNs, and the dimensions of the FRC meshwork of interacting fibers, they hypothesized that the channels between FRCs might guide lymphocyte motility, but pointed out that this conclusion was based on inferences and needed to be confirmed experimentally (Anderson and Shaw, 1993; Gretz et al., 1997; Kaldjian et al., 2001). The imaging results we present reveal that FRCs do not generate an enclosed labyrinth of corridors within which lymphocytes traffic due simply to peripheral confinement, as these authors previously suggested. Rather, the FRCs constitute an open 3-dimensional meshwork of cell bodies and extended processes, physical interaction with which appears to provide the guidance cues underlying T and B cell movement within the T zone of LNs. In the follicle, FDCs provide a corresponding substratum for B cell locomotion.
The development of adaptive cell-mediated and humoral immune responses depends on the interactions between rare antigen-specific cells in the naïve lymphocyte repertoire, as well as association of such cells with a limited number of activated antigen presenting cells. How these rare cells find each other in the densely packed LN environment is an issue that has been given a great deal of attention in the recent past. Dynamic imaging studies and mathematical calculations of the possible rates of cell-cell contact based on data from such analyses have suggested that random motility can ensure effective interactions among rare cells in LNs (Bousso and Robey, 2003; Miller et al., 2004; Miller et al., 2003) and several models of immune responses have been based on this notion (Catron et al., 2004). Others have argued that a combination of the micro-anatomy of lymphoid tissue and chemical guidance cues combine to enhance the likelihood of appropriate cell-cell contact over what would occur on a random interaction basis (Castellino et al., 2006; Germain et al., 2005; Huang et al., 2004; Katakai et al., 2004a; Katakai et al., 2004b). Here we demonstrate that naïve T and B cells move along two different stromal networks in the LN, providing new evidence that LN micro-anatomy plays a critical role in underpinning the physiological function of these tissues.
C57BL/6 and C57BL/6 ubiquitin-GFP mice (UBI-GFP/BL6, strain 4353) were purchased from the Jackson Laboratory (Bar Harbor, Maine) and maintained in the National Institutes of Health animal facilities. Balb/c mice were purchased from Charles River Laboratories France (L’Arbresle, France), Hu-CD2 GFP mice were originally a gift from D. Kioussis (Mill Hill, London, U.K.), while CX3CR1 GFP mice (Jung et al., 2000) were originally a gift from S. Jung (Rehovot, Israel). These strains of mice were maintained in the animal facility of the Institut de Pharmacologie Moléculaire et Cellulaire, France. For the generation of chimeras, C57BL/6 ubiquitin-GFP mice were γ-irradiated with a single dose of 950 rads (or twice with 500 rads) from a cesium source and were reconstituted with 2 x 106 C57BL/6 bone marrow cells. At 8 weeks after reconstitution, mice were tested for chimerism. Chimeras were used for subsequent experiments only if analysis of blood leukocytes showed the presence of less than 2% of CD3+ T cells of host origin. All procedures performed on animals in this study have been approved by the Animal Care and Use Committee, NIAID, NIH and/or the Institut National de la Santé et de la Recherche Médicale, Université de Nice-Sophia Antipolis.
T cells were purified from the LNs of wild-type mice using a pan T cell isolation kit (Miltenyi Biotec, Auburn, California) and stained with either CFSE (2μM) or SNARF-1 (5μM) (Invitrogen, Carlsbad, California) at 37°C for 15 min. The indicated numbers of cells were transferred into host mice by intravenous injection.
ERTR-7 antibody specific for an unknown FRC-secreted molecule was purchased from Acris Antibodies (Hiddenhausen, Germany). FDC-M2 antibody specific for FDCs was purchased from Immunokontact (Oxon, UK). RA3-6B2 antibody specific for B220, 17A2 specific for the CD3 complex, MECA-79 specific for PNAd, and 3E2 specific for ICAM-1 were from BD Biosciences Pharmingen (San Diego, California). These antibodies were visualized by direct coupling to allophycocyanin, Alexa fluor-488, -568, -647, or through the use of Alexa fluor-488, -568, or -647 coupled secondary antibodies.
Animals were anesthetized with avertin and given an intracardiac injection of 15 ml of 0.05 M phosphate buffer containing 0.1 M L-lysine, pH 7.4, 2 mg/ml NaIO4, and 10 mg/ml paraformaldehyde (PLP). After excision from perfused animals, LNs were incubated 12 hrs with this same medium, washed in phosphate buffer, and dehydrated in 30% sucrose in phosphate buffer. Tissues were snap frozen in Tissue-Tek® (Sakura Finetek). 10–30 μm frozen sections were cut and then stained with the indicated antibodies as previously described (Bajenoff et al., 2003). Immunofluorescence confocal microscopy was performed using a Leica TCS SP confocal microscope. Separate images were collected for each fluorochrome and overlaid to obtain a multicolor image. Final image processing was performed using ImageJ software (National Institutes of Health) and Adobe Photoshop.
Freshly isolated, fluorescent dye-labeled T cells were injected i.v. into chimeric recipient mice. Three to four hours later (unless otherwise specified), the right popliteal LN of an anesthetized mouse (2.5% isofluorane in an air/O2 mix) was surgically exposed and imaged with a Bio-Rad Radiance 2100 MP system attached to a Nikon 600 FN upright microscope fitted with a 20X water immersion lens (NA = 0.95, Olympus). The 2-P laser was either a Mira900 Sa: Ti femtosecond pulsed laser tuned to 880 nM that was driven by a 10 Watt Verdi pump laser or a Chameleon XR femtosecond pulsed laser tuned to 880 nm (Coherent). Images in movies were collected with typical voxel size = 0.91 x 0.91 x 1–3 μm and a volume dimension = 100 x 100 x 10 μm, unless indicated otherwise. Images were typically collected between 50–140 μM below the capsule. This volume collection was repeated every 20–30 s to create 4D data sets that were then processed with Imaris software (Bitplane) and Adobe AfterEffect (Adobe). Supplementary videos created from these image stacks are maximum intensity projections and play at 150 or 300 x real time.
Freshly isolated LNs were embedded in a 1.5% agarose in PBS. After agarose solidification, 300-μm slices containing LN sections were cut with a Leica VT1000 S Vibratome in a bath of ice-cold PBS. Retrieved sections were kept until use at 37 °C in complete RPMI 1640 medium (Biosource) supplemented with 10% fetal bovine serum, L-glutamine, penicillin, streptomycin, sodium pyruvate, non-essential amino acids, and 10mM of HEPES. For ERTR-7 staining, LN sections were placed in complete RPMI medium containing the primary or secondary antibody for 2 hours at 37 °C. For imaging, the section was placed in a glass bottom microwell dish (MatTek Corporation), held down with a slice anchor, and perfused continuously with pre-warmed phenol red-free RPMI 1640 medium buffered with 10mM of HEPES and bubbled with 95% O2 and 5% CO2. Bath temperature was maintained at 37 °C with a stage heater and objective heater. Time-lapse images were acquired by a Leica TCS SP2-AOBS MP system fitted with a Leica 20X glycerol objective. The 2P laser was a Spectra Physics wide band MaiTai laser tuned to 800 nM. 4D data sets were processed with Imaris software (Bitplane) and Adobe AfterEffect (Adobe).
PLP-fixed LNs were rinsed in 0.1 M phosphate buffer and immersed for 1h in a 20% glycerol solution. LNs were then frozen in liquid nitrogen and freeze-cracked before being dehydrated through graded concentrations of ethanol. LNs were desiccated with hexamethyldisilazane and affixed on aluminum stubs with adhesive, coated with gold-palladium, and examined in a SEM (JEOL 6700F, Japan) microscope with an accelerating voltage of 3 kV.
PLP-fixed LNs were rinsed in 0.1 M phosphate buffer and post-fixed for 1h in OsO4. LNs were then rinsed in water, dehydrated through graded concentrations of acetone (50, 70, 95, 100% pure, 3 times) before incubation for several hours in a 1:1 vol/vol acetone-Epon mixture. LNs were then incubated overnight in pure Epon before final embedding in Epon. 80nm thick LN sections were generated using a LEICA Ultracut S ultramicrotome. Sections were stained with uranyl acetate followed by lead citrate and were imaged using a CM12 Philips microscope.
Figure S1: CFSE accumulates in the uropods of motile T cells.
CFSE-labeled T cells were transferred i.v into a wild-type animal. Twelve hours later, the mouse was anesthetized and perfused with PLP fixative. LNs were sectioned and subsequently stained for ICAM-1, an adhesion molecule known to relocate to the uropod of motile cells. Pictures show two polarized T cells with a high CFSE content at one end (A) See also Movie S3. ICAM-1 staining reveals this end to be the uropod (B).
Figure S2: Additional examples of lymphocytes contacting FRCs.
SEM pictures of lymphocytes associated with FRC fibers in the T cell zone. The arrowheads indicate lymphocytes microvilli extensions from the T cell to the FRC fibers. Scale bar: 1 μm.
Figure S3: T cells can be observed probing the FRC network during their displacement.
(A) Several adjacent optical sections from the same time points of 4D datasets show that thin fibers (green) supporting T cell (red) motility can be missed it they are outside of the z slices represented in an image. (B) One example of a migrating T cell extending its filapodia from its leading edge, probing intersecting fibers and following one of the fibers that had been examined. See also Movie S8.
Figure S4: “Non-fluorescent” LNs spaces are not empty spaces for free lymphocyte motility.
(A) SEM of a wild-type LN showing dense packing of leucocytes near HEV. (B) 3 x 105 CMTMR T cells (red) were transferred i.v into a Hu-CD2 GFP mouse. Twelve hours later, LNs were harvested. 10 μm thick sections were prepared and stained with phalloidin (blue) to reveal endogenous T cells (green), as well as other cells. This image emphasizes the many physical contacts a single T cell makes in addition to those made with FRC fibers. Originally imaged with a 63x objective. These images reveal a dense packing of lymphocytes in the interstices between fibers under normal circumstances, suggesting that the turning behavior is not a result of path obstruction due to the mere presence of another physical object, but rather is regulated by the spatial design of the fiber arrays. (C) Experimental protocol used to quantify, from 4D datasets, the correspondence between the turning angle of a migrating lymphocyte and that of the underlying GFP+ fibers.
Figure S5: Additional examples of T cell motility on FRC fibers.
(A) Intravital snapshots of another single T cell (red) moving over time on the cell body and processes of GFP+ FRC (green). (B) Occasionally, elongated T cells are observed “jumping” between fibers. Arrowheads indicate contact points between T cells and the fibers used during the “jump”.
Figure S6: DC and macrophages lie on the FRC network.
LNs from a CX3CR1 GFP+/+ mouse, in which DC, monocytes, and macrophages express GFP (green), were sectioned and stained for ERTR-7 expression (red). Data show confocal pictures of GFP+ cells ensheathing the conduit system. Originally imaged with a 63x objective.
Figure S7: FRCs delineate the T cell zone in the LN.
(A) Confocal image of a chimeric LN section stained for B220 (left panel, blue) or FDC-M2 (right panel, red) expression. S.C: subcapsular sinus. (B) WT mice were either untreated (left panel) or injected with 200 μg of an anti-CD62L blocking Ab (right panel). One day later, LN sections were stained for CD3 (blue), B220 (green), and ERTR-7 (red) expression to highlight the loss of B cells from the terminal peri-follicular ERTR-7+ network when lymphocyte entry into the LN is blocked. Arrowheads indicate numerous B cells close to HEVs while *s indicate the location of the magnified region (inserts). Originally imaged with a 16x and 63x (inserts) objective.
Figure S8: B cells move on the FRC network in the T zone.
(A) Wild-type mice were injected i.v with 10 x 106 CFSE-labeled B cells. Twenty four hours later, mice were perfused with a fixative solution. Representative confocal images of sections from recipients LN showing transferred CFSE-labeled B cell (green) tightly associated with ERTR-7+ (red) and desmin+ (blue) FRC fibers. Originally imaged with a 63x objective. (B) Wild-type mice were injected i.v with 20–30 x 106 SNARF-1-labeled B cells (red). Twenty four hours later, LN were harvested, vibratome sectioned (300μm thick), stained for ERTR-7 (green), incubated in a warm and oxygenated medium and imaged by 2P microscopy. Data show snapshots of a single B cell (red, arrowhead) moving over time on ERTR-7 fibers (green) in a 12 μm thick volume. See also Movie S11. (C) Quantitation of B cells showing turns in dynamic 4D datasets with respect to their location on or off ERTR-7+ fibers.
An ERTR-7 stained thick section (30 μm) presented in 3D, showing a fibrous and complex network of interconnected strands.
The same T cell (blue) imaged in Figure 1C is shown in 3D, along with the associated FRC fibers stained with ERTR-7 (green) and desmin (red).
CFSE-labeled T cells were transferred i.v into a wild-type mouse and their dynamic migratory behavior in the popliteal LN was captured using intravital 2-P imaging. CFSE intensity is represented with false colours, with the more intense green signal indicating a higher CFSE content.
The popliteal LN of a chimeric animal was imaged using 2-P microscopy. Note the presence of a thick blood vessel in the FRC network.
Dynamic image of T cell (red) migration along the FRC network (green). The trails of three of the T cells are highlighted in the second movie with colored dots to help visualize the path taken along the fibers by a given T cell. (z stack = 12 μM). The playback speed is 300x in the first part of the movie and 150x in the second part when the tracks are highlighted.
Another image of dynamic T cell (red) migration along the FRC fiber network (green). z stack = 12 μM, playback speed is 300x.
Comparative sections from a 4D dataset demonstrating that, depending on the thickness of the visualized z stack, thin FRC fiber strands supporting T cell motility can be absent from the represented volume but still constitute attachment sites for the T cells. The playback speed is 300x.
Dynamic image of T cell (red) migration along the FRC network (green). The trails of two of the T cells are highlighted. Red colored dots highlighted the path of a highly motile cell while the white dots track the path of a slower motile cell protruding its filapodia on fibers during its displacement (z stack = 12 μM). The playback speed is 150x.
A single z slice from an intravital 4D dataset showing numerous T cells exiting HEV via lucent areas that appear to be gaps in the FRC sheath (“exit ramps”). The playback speed is 300x for both the main and zoomed image.
T cells (red) follow the FRC fibers (green) to direct the path of their movement as soon as they exit the HEV (z stack = 9 μM). The playback speed is 300x.
In warmed and oxygenated LN vibratome sections, adoptively transferred B cells (red) follow the ERTR-7+ fibers (green) while moving in the T cell area (Z stack=12 μM). The playback speed is 150x.
FRC and FDC networks present different characteristics in vivo. Note the higher density and less regular shape of the FDC network as well as its capacity to display movement over time. Arrowheads point out a highly motile region of the FRC network.
Dynamic image of B cells (red) migration along the motile FDC network (green). The trails of two of the B cells are highlighted in the second movie with colored dots to help visualize the path taken along the fibers by a given B cell. (z stack = 12 μM). The playback speed is 300x in the first part of the movie and 150x in the second part when the tracks are highlighted.
The authors wish to thank J. Ashwell, A. Anderson, G. Lauvau, and E. Mougneau for their suggestions on the manuscript, Hai Qi for helpful discussions and P. Gounon and S. Pagnotta for advice and assistance with electron microscopy. A. Anderson also provided helpful advice during the course of this study. The CX3CR1 GFP mice were provided by Steffen Jung and Dan Littman, HHMI. This research was supported in part by the Intramural Research Program of NIAID, NIH, DHHS and by the Institut de la Santé et de la Recherche Médicale (INSERM). M. Bajénoff was supported by a postdoctoral fellowship grant from INSERM. The authors declare that they have no competing financial interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.