Effector CD4+ T cells recognize intravascular antigen presented by patrolling monocytes

Although effector CD4+ T cells readily respond to antigen outside the vasculature, how they respond to intravascular antigens is unknown. Here we show the process of intravascular antigen recognition using intravital multiphoton microscopy of glomeruli. CD4+ T cells undergo intravascular migration within uninflamed glomeruli. Similarly, while MHCII is not expressed by intrinsic glomerular cells, intravascular MHCII-expressing immune cells patrol glomerular capillaries, interacting with CD4+ T cells. Following intravascular deposition of antigen in glomeruli, effector CD4+ T-cell responses, including NFAT1 nuclear translocation and decreased migration, are consistent with antigen recognition. Of the MHCII+ immune cells adherent in glomerular capillaries, only monocytes are retained for prolonged durations. These cells can also induce T-cell proliferation in vitro. Moreover, monocyte depletion reduces CD4+ T-cell-dependent glomerular inflammation. These findings indicate that MHCII+ monocytes patrolling the glomerular microvasculature can present intravascular antigen to CD4+ T cells within glomerular capillaries, leading to antigen-dependent inflammation.

A growing body of evidence indicates that immune cells can make critical contributions to inflammatory responses while remaining within the vasculature 1 . This concept of "intravascular immunity" is exemplified by the intravascular migration of non-classical monocytes in tissues such as skin, mesentery, muscle, and brain [2][3][4][5][6] . In vivo imaging studies show that this patrolling function involves prolonged crawling on the endothelium independent of the direction of blood flow 2,4 . Patrolling Ly6C − monocytes perform important immune surveillance within the vasculature, internalizing microparticles and soluble material from the bloodstream and responding to microbial infection or tissue injury 3,7 . Upon detection of these signals, intravascular monocytes are positioned to respond rapidly by inducing recruitment of other immune cells or migrating out of the vasculature [3][4][5] . These intravascular activities are not restricted to myeloid leukocytes as in the liver microvasculature, invariant natural killer T (iNKT) cells also constitutively migrate. In this location iNKT cells respond to innate and adaptive signals by modulating their migration and releasing proinflammatory mediators 8,9 . Whether these intravascular functions also contribute to adaptive immune responses involving conventional T cells is less clear.
Intravital imaging studies have revealed that CD4 + and CD8 + T cells in microvessels of the central nervous system and the renal interstitium can undergo intraluminal crawling and subsequently exit the vasculature [10][11][12] . While in contact with the endothelium, effector CD8 + cells can recognize peptide/MHC class I (MHCI) complexes expressed by endothelial cells, leading to T-cell activation and promotion of recruitment and/or tissue retention [12][13][14] . Although some endothelial cells can express MHC class II (MHCII) in inflammatory states, mechanisms of antigen recognition by antigen-specific CD4 + T cells within the vasculature are unclear [15][16][17] One site where intravascular immunity is crucial is the specialized microvasculature of the glomerulus. Monocytes and neutrophils have been shown to undergo constitutive intravascular adhesion and crawling within glomerular capillaries 17,18 . During antibody-mediated glomerular inflammation, the duration of the retention of these cells is increased and intravascular neutrophils generate reactive oxygen species (ROS) responsible for glomerular injury 17,18 . In addition to humoral mediators, CD4 + T cells also have an important function in the development of glomerular injury and dysfunction in severe, rapidly progressive forms of glomerulonephritis. Intraglomerular T cells are detectable in humans with rapidly progressive glomerulonephritis [19][20][21] , and functional studies in animal models demonstrate that disease-inducing effector responses can be directed by CD4 + T cells responding to antigen located intravascularly within the glomerulus 15,16,[22][23][24] .
Evidence indicates that glomerular injury mediated by effector CD4 + T cells in glomerulonephritis involves multiple steps, beginning with loss of tolerance to nephritogenic autoantigens in secondary lymphoid organs 16,25 . This evidence includes the discovery of circulating CD4 + T cells specific for these antigens in patients with autoimmune glomerulonephritis [26][27][28] . In patients with autoimmune disease, circulating autoreactive T cells have a memory phenotype, indicating that they have been exposed to cognate antigen and undergone differentiation into effector or memory T cells 29,30 . However, the existence of circulating, antigen-experienced T cells is insufficient to result in disease. The final steps in the process involve T-cell recognition of antigen in the target tissue, leading to effector T-cell-mediated injury at the site of antigen recognition, the glomerulus. Indeed, analysis of antigen-experienced T-cell responses to antigen in the periphery has shown that these cells can respond within minutes upon recognition of cognate antigen 31,32 . However, the mechanism whereby CD4 + T cells recognize antigens in the unique microvasculature of the glomerulus is not known. Therefore, the aim of this study is to investigate the mechanisms of intravascular antigen presentation to disease-initiating, antigen-experienced effector CD4 + T cells in the glomerulus, using a validated model of T-cell-mediated glomerulonephritis. The findings indicate that in the absence of inflammation, MHCII expression in the glomerulus is restricted to subsets of circulating leukocytes. Of these cells, monocytes undergo the most prolonged retention and migration in the glomerular capillaries and are required for CD4 + T-cell-mediated induction of neutrophil-dependent glomerular inflammation.

Results
CD4 + T cells migrate in uninflamed glomerular capillaries. We first examined whether CD4 + T cells could spontaneously adhere within uninflamed glomerular capillaries. Endogenous CD4 + T cells were visualized using anti-CD4 mAb and intravital multiphoton imaging of the kidney. In uninflamed glomeruli, CD4 + T cells regularly underwent periods of adhesion (defined as retention for >30 s) on the endothelial surface, at a rate of~5 cells per glomerulus per hour (Fig. 1a, b and Supplementary Movie 1). The majority remained stationary during adhesion ("static" cells), while~20% underwent crawling (Fig. 1b). Typically, the duration of retention, or dwell time, of CD4 + T cells was~4 min (Fig. 1c). Crawling CD4 + T cells were retained in glomeruli for twice as long as stationary cells and migrated at~9 μm min −1 (Fig. 1c, d).
Detection of these blood-borne CD4 + T cells in this fashion does not allow the differentiation between naïve and effector T cells. This is important as for CD4 + T cells to induce disease in the glomerulus in response to local antigen recognition, they require prior activation and differentiation into effector cells in secondary lymphoid organs. Therefore, we next asked whether activated effector T cells also undergo retention in uninflamed glomerular capillaries. To examine this issue, we used OVAspecific T cells from TCR transgenic OT-II mice. OT-II T cells were activated in vitro to a Th1 phenotype, fluorescently labeled and transferred intravenously into recipient mice and subsequently glomeruli were examined via intravital multiphoton imaging. As for endogenous CD4 + T cells, activated OT-II cells underwent retention in glomerular capillaries (Fig. 1e, Supplementary Movie 2). On average,~1.5 OT-II cells underwent adhesion per glomerulus per hour, of which~70% remained stationary (Fig. 1f). OT-II cells arrested in glomeruli at a relatively consistent rate during the 90 min imaging period after transfer (Fig. 1g). The dwell time of OT-II cells in glomeruli was~13 min -this was similar in stationary and crawling OT-II T cells (Fig. 1h) while more than twice that of endogenous CD4 + T cells. The mean migration velocity of crawling OT-II cells was~9 μm min −1 (Fig. 1i). Together these findings indicate that CD4 + T cells, and specifically effector CD4 + T cells, undergo adhesion and migration in the glomerular microvasculature in the absence of inflammation.
T-cell adhesion in antigen-bearing glomerular capillaries. We have previously shown that OT-II cells can trigger antigendependent neutrophil recruitment within 4 h in a planted antigen model of CD4 + T-cell-dependent glomerulonephritis [15][16][17] . In this model, the peptide antigen OVA 323-339 (pOVA) is delivered to the glomerular microvasculature via conjugation to 8D1, an mAb that binds to the NC1 domain of the α3 chain of type IV collagen in the glomerular basement membrane without inducing glomerular injury 33 . The 8D1/pOVA construct was transferred into mice intravenously together with fluorescently labeled OT-II cells. In order to focus on initiation of the T-cell-dependent response, glomeruli were imaged for 2 h after OT-II transfer. Compared to control mice receiving unconjugated 8D1 mAb, more OT-II cells adhered in glomerular capillaries of mice receiving 8D1/pOVA (Fig. 2a), due solely to an increase in the number of crawling cells (Fig. 2b). OT-II cells also crawled more slowly in mice that received 8D1/pOVA (Fig. 2c), a finding consistent with the reduced velocity of effector T cells during antigen recognition described in other organs 31,32,34 . To identify when the increased retention of crawling OT-II cells occurred, the rate of OT-II cell recruitment to the glomerular capillaries was examined in 10 min intervals during the 2 h after transfer. In mice given 8D1/pOVA, a significant increase in crawling OT-II cells was detectable 71-80 min after transfer (Fig. 2d). In contrast, the rate of arrest of static OT-II cells was constant after the first 20 min of the observation  period and did not differ between the groups (Fig. 2e). OT-II cells were also retained in glomerular capillaries for longer periods in mice that received 8D1/pOVA (Fig. 2f), due to a significant increase in the dwell time of crawling OT-II cells (Fig. 2g). To control for possible effects of inflammation-associated alterations in glomerular hemodynamics on the retention of OT-II cells 35 , we also compared retention of OT-II T cells with that of T cells from SMARTA mice, which recognize an unrelated peptide. In co-transfer experiments with differentially labeled activated OT-II and SMARTA cells, in mice treated with 8D1-pOVA, OT-II cells showed prolonged glomerular dwell time relative to cotransferred SMARTA T cells (Fig. 2h). This response was not seen in the absence of antigen (Fig. 2i), supporting the hypothesis that the increased retention of OT-II cells was due to their capacity to recognize pOVA, rather than physical factors at play in the inflamed glomerulus. We also examined intravascular migration by OT-II cells in glomerular capillaries 2-3 h after cell transfer, when the number of adherent OT-II cells was comparable with or without exposure to antigen and similar for crawling and static cells (Supplementary Fig. 1a, b). However, consistent with the earlier time point (Fig. 2g), in mice receiving antigen, crawling effector OT-II cells remained in glomeruli for longer, and crawled more slowly ( Supplementary Fig. 1c, d) during the 2-3 h period post-transfer. These results indicate that effector OT-II cells undergo increased retention in glomerular capillaries in response to a planted glomerular antigen within just over an hour of administration.
Antigen-presenting cells patrol glomerular capillaries. Effector CD4 + T cells recognize antigenic peptides presented via MHCII molecules to induce responses in peripheral tissues. However, in non-inflamed glomeruli, it has been reported that resident cells expressing MHCII are rare 36 . We confirmed these findings using MHCII-EGFP mice in which GFP is expressed fused to MHCII, thereby labeling MHCII-expressing cells 37 . Three-dimensional To exclude the possibility that projections from periglomerular mononuclear phagocytes extended into the glomerulus, we performed similar analyses of kidneys from CD11c-YFP mice, in which YFP is highly expressed in renal dendritic cells and cellular projections are readily detectable [38][39][40] .
In agreement with previous studies, these experiments revealed an interdigitated network of dendritic cells in the interstitium, including immediately adjacent to, but not within glomeruli 39,41 . However, no projections from these cells were detected within glomeruli (Supplementary Fig. 2

and Supplementary Movie 4).
These data indicate that mononuclear phagocytes of the renal interstitium are unlikely to contribute to intraglomerular antigen presentation under resting conditions. We next asked whether the MHCII + cells responsible for antigen presentation were blood-borne leukocytes migrating within glomerular capillaries. Several populations of MHCIIexpressing leukocytes circulate in the bloodstream, including B cells and a subset of monocytes 42 . To determine if MHCII + cells also undergo retention and migration in glomerular capillaries, we examined kidneys of MHCII-EGFP mice via intravital multiphoton imaging. Experiments in uninflamed MHCII-EGFP mice revealed that numerous intravascular MHCII + cells migrate within glomerular capillaries (Fig. 3a, Supplementary Movie 5). Approximately 20 MHCII + cells underwent adhesion in the glomerular capillaries in each glomerulus every hour (Fig. 3b). Nearly 80% of these cells remained static (Fig. 3b), with an average dwell time of~5 min (Fig. 3c). However, the minority of MHCII + cells that crawled within glomerular capillaries were retained for more than twice as long (Fig. 3c). We confirmed these findings in glomeruli of intact, non-hydronephrotic kidneys of 3-week-old mice, in which~15 adherent MHCII + cells underwent adhesion per glomerulus per hour, the majority of which were stationary (Fig. 3d, Supplementary Movie 6). These experiments demonstrate that MHCII-expressing leukocytes constitutively undergo periods of retention, and in some cases migration, in glomerular capillaries, even in the absence of inflammatory stimuli. Together with our observations of constitutive intraglomerular migration of CD4 + T cells and altered retention of these cells in the presence of cognate antigen, these findings raise the possibility that circulating MHCII + leukocytes present glomerular antigens to intravascular effector CD4 + T cells.

T cells interact with intravascular antigen-presenting cells.
For intravascular MHCII-expressing leukocytes to present antigens to CD4 + T cells within the glomerulus, these two populations must interact within glomerular capillaries. To determine if this occurs, CMTPX-labeled OT-II cells were imaged 30-90 min after transfer into MHCII-EGFP mice injected with either 8D1/pOVA or unconjugated 8D1. In both groups of mice, OT-II cells and MHCII + leukocytes consistently underwent interactions in the glomerular microvasculature (Fig. 4a, Supplementary Movie 7). On average, each OT-II cell interacted with one MHCII + cell during its time within the glomerulus (Fig. 4b), with the number of interactions being similar in mice receiving either 8D1 or 8D1/ pOVA. The majority of interactions were no longer than 4 min. However, we noted more prolonged interactions (>10 min) in

Antigen-dependent activation of OT-II cells in glomerulus.
To assess whether OT-II cells undergo activation within the glomerular capillaries, we made use of a system enabling visualization of translocation of the transcription factor nuclear factor of activated T cells-1 (NFAT1) to the nucleus 11,43 . In T cells, NFAT1 is a critical transcription factor driving changes in gene expression in response to T-cell activation, and rapidly translocates to the nucleus in response to elevation in cytosolic calcium following TCR engagement 44 . Via use of the fluorescent reporter NFAT1 (1-460) -GFP (NFAT-GFP), NFAT1 translocation to the nucleus, and therefore T-cell activation, can be tracked in vivo using multiphoton microscopy. We generated NFAT-GFPtransduced OT-II (OT-II NFAT-GFP ) cells and demonstrated in vitro that the NFAT-GFP reporter readily translocated to the nucleus following T-cell activation ( Supplementary Fig. 3). We then transferred these cells into mice 30 min after treatment with either 8D1 or 8D1-pOVA and assessed subcellular localization of the reporter (Fig. 4d-f and Supplementary Fig. 3e−g). Following administration of 8D1-pOVA, NFAT-GFP had either partially or completely translocated to the nucleus in 26% of OT-II NFAT-GFP cells, whereas in mice that received 8D1, translocation was seen in only 10% of cells (P < 0.05, Fisher's exact test) (Fig. 4g). Furthermore, typically T cells with cytoplasmic NFAT-GFP were active and highly migratory, while cells displaying nuclear NFAT-GFP remained static (Supplementary Movie 8), consistent with previous descriptions of T-cell arrest during antigen-dependent activation 31 . These experiments provide direct evidence of T-cell activation occurring within the glomerular capillaries.
To investigate whether OT-II cells in the kidney showed other changes typically associated with antigen recognition, 4 h after transfer kidneys were digested and the T cells analyzed by flow cytometry, assessing IFNγ production. In mice that received 8D1/ pOVA, renal OT-II cells had significantly increased de novo production of IFNγ, compared with OT-II cells from mice that received 8D1 control mAb (Fig. 5a-c). This response was not seen in mice that received pOVA conjugated to a control antibody (IgG/pOVA) (Fig. 5b, c), demonstrating that pOVA administered in a non-targeted form was insufficient to induce this response.
We previously observed increased glomerular neutrophil retention in mice 4 and 24 h after receiving OT-II cells and 8D1/pOVA 17 . Here we confirmed that this response was also dependent on glomerular targeting of pOVA. Comparison of neutrophil responses 24 h after administration of OT-II cells plus either 8D1, 8D1/pOVA or non-targeted IgG/pOVA revealed that increased neutrophil dwell time was observed in mice that received 8D1/pOVA but not IgG/pOVA (Fig. 5d). As an additional readout of the neutrophil response, we used the oxidant-sensitive fluorochrome, DHE, to assess ROS production by intraglomerular neutrophils 17,18 . These experiments revealed that induction of neutrophil ROS production also required glomerular targeting of antigen (Fig. 5e). Together these data indicate that glomerular localization of pOVA results in T-cell activation in the microvasculature of the glomerulus and induction of neutrophil-dependent glomerular inflammation.
B cells are not required for T-cell-dependent inflammation. We next sought to identify the MHCII-expressing immune cell in the circulation responsible for intravascular antigen presentation to CD4 + T cells in glomerular capillaries. Flow cytometric analysis of blood from MHCII-EGFP mice revealed that the majority (~95%) of MHCII-EGFP + leukocytes were CD19 + B cells (Supplementary Fig. 4a). The remainder consisted primarily of a subset of CD115 + monocytes, while <1% of the MHCII + population were CD11c + dendritic cells. To determine whether B cells are retained in glomerular capillaries, we used anti-B220 to label B cells in vivo ( Supplementary Fig. 4b, Supplementary Movie 9). B cells adhered within glomerular capillaries at a rate of~13 cells per glomerulus per hour ( Supplementary Fig. 4c). The majority of adherent B cells remained stationary ( Supplementary Fig. 4c), with a dwell time of~8 min, while the minor population of crawling B cells had a shorter dwell time of only~3 min (Supplementary Fig. 4d).
To determine whether B cells were required for induction of Tcell-mediated glomerular inflammation, we next assessed neutrophil retention in B-cell-deficient μ MT mice 4 h after transfer of OT-II cells and either 8D1/pOVA or unconjugated 8D1. μ MT recipient mice that received 8D1/pOVA had prolonged retention of neutrophils in glomeruli, compared with mice that received 8D1 ( Supplementary Fig. 4e), similar to the response described previously in wild-type mice 17 . These findings indicate that while B cells comprise the majority of MHCII + cells retained in the uninflamed glomerulus, they are not required for induction of CD4 + T-cell-mediated inflammation. We next examined the contribution of circulating monocytes to the MHCII + leukocyte population undergoing retention in glomerular capillaries. Flow cytometric analysis of blood from MHCII-EGFP mice revealed that 10-20% of circulating monocytes expressed MHCII (Fig. 6a) similar to previous reports 42,45,46 , with these cells being found in both CX3CR1 lo-int (classical) and CX3CR1 hi (non-classical) populations (Supplementary Fig. 5). To determine whether monocytes contributed to the MHCII + leukocyte population patrolling the glomerulus, we treated MHCII-EGFP mice with clodronate liposomes and analyzed migration of EGFP + cells by multiphoton microscopy. Clodronate liposome treatment has recently been shown to almost eliminate patrolling monocytes from the glomerulus 18 . In MHCII-EGFP mice, clodronate liposomes did not significantly affect the number of stationary MHCII + cells in glomerular capillaries (Fig. 6b), consistent with the continued retention of B cells. In contrast, the number of MHCII + leukocytes undergoing crawling was significantly decreased after clodronate liposome treatment, showing that monocytes constitute the majority of the crawling MHCII + leukocyte population within glomerular capillaries.
In the absence of inflammation, monocytes extensively patrol the glomerular microvasculature, being retained in the glomerular capillaries for an average of~20 min 17,18 . To determine whether MHCII + monocytes were retained for a similar extended time, we utilized anti-CD11b to detect monocytes in MHCII-EGFP mice 3 . As B cells may also express CD11b, we first performed flow cytometric analyses of whole blood, demonstrating that the majority (>80%) of CD11b hi MHCII + leukocytes were monocytes, while <10% were B cells (Supplementary Fig. 6). Using intravital multiphoton microscopy of the glomerular capillaries, . Adhesion of these CD11b + MHCII + leukocytes occurred at a rate of~2.5 cells per glomerulus per hour, with the majority undergoing crawling (Fig. 6d). Interestingly, the mean dwell time of CD11b + MHCII + cells in glomeruli was~40 min, compared to <10 min for CD11b − MHCII + cells, with this applying to both crawling and stationary cells (Fig. 6e). Furthermore, cells with dwell times >35 min were almost exclusively CD11b + MHCII + (Fig. 6f). As B cells typically have a short (~6 min) dwell time in glomeruli ( Supplementary  Fig. 4d), these findings support the interpretation that the CD11b + MHCII + leukocytes with prolonged dwell times are not B cells.
MHCII + monocytes are not recognized as professional antigen-presenting cells. Nevertheless, for them to be important in T-cell-mediated glomerular inflammation, they must be able to induce antigen-dependent responses in T cells. Therefore, we sorted MHCII + monocytes from the blood of MHCII-GFP mice (Fig. 7a) and assessed their capacity to induce OT-II cell proliferation in vitro 39,40 . In the presence of OVA-peptide, MHCII + monocytes were able to induce OT-II T-cell proliferation (Fig. 7b-d). T cells did not proliferate in the absence of monocytes and the level of T-cell proliferation was proportional to the number of monocytes in the assay, supporting the specificity of this response. These observations clearly demonstrate that the MHCII-expressing monocyte subset can present antigen to CD4 + T cells. Finally, using electron microscopy, we observed that mononuclear immune cells patrolling glomerular capillaries extended projections in close proximity to fenestrae of glomerular endothelial cells ( Supplementary Fig. 8), raising the possibility that these leukocytes were probing the location of the antigen in these experiments. Together these findings indicate that a population of MHCII + monocytes undergoes constitutive patrolling within the glomerular microvasculature, characterized by substantially longer intraglomerular dwell times than any other population previously identified, and that MHCII + monocytes have the capacity to induce CD4 + T-cell proliferation.
Monocytes are required for T-cell-induced inflammation. We next assessed whether the presence of MHCII + patrolling monocytes within glomeruli was associated with a role for monocytes in induction of antigen-dependent glomerular inflammation, using clodronate liposomes to deplete monocytes in the OT-II cell-8D1/pOVA model. Firstly we established that OT-II cell trafficking in the glomerulus was unaltered in clodronate-treated mice (Supplementary Fig. 9). Subsequently we used neutrophil retention and ROS production in glomerular capillaries as readouts of T-cell-initiated glomerular inflammation following clodronate treatment. In mice depleted of monocytes, both neutrophil dwell time and ROS production 24 h after transfer of OT-II cells and 8D1/pOVA were reduced relative to mice treated with control liposomes (Fig. 8a-c and Supplementary Movie 11), with the reduction in ROS-producing neutrophils due to reduced production by crawling cells (Fig. 8d). Together these findings support the hypothesis that induction of glomerular inflammation by intravascular OT-II cells in response to planted antigen requires monocytes.

Discussion
CD4 + T cells specific for nephritogenic autoantigens are present in the circulation during autoimmune glomerulonephritis [26][27][28] , and experimental evidence indicates that CD4 + T-cell-dependent glomerular inflammation can be initiated by T cells responding to antigen located within the vasculature 15,16 . However, the cell type that presents antigen to effector CD4 + cells within glomeruli is not known. The present study addresses this by demonstrating that the cellular participants required for a CD4 + T-cell-mediated response-CD4 + T cells and an MHCII-expressing cell-both undergo constitutive periods of intravascular retention and crawling in the glomerular microvasculature. In addition, here we observed that each T cell retained in the glomerulus interacts with on average one MHCII-expressing immune cell. In the presence of antigen, T cells in the glomerular capillaries displayed evidence of antigen recognition including nuclear translocation of NFAT1, prolonged dwell time and reduced migration velocity. The MHCII-expressing cells retained in the glomerulus for the longest duration expressed the monocyte marker CD11b and monocyte Together these findings demonstrate that a CD4 + Tcell-driven immune response in the glomerulus can be mediated by intravascular antigen presentation by an MHCII-expressing subset of circulating monocytes. We hypothesized that a key step in the development of T-cellmediated glomerulonephritis occurs when T cells recognize disease-causing antigens within the glomerular microvasculature. Here we provide direct evidence of T-cell activation in the glomerulus during the initial response to locally planted antigen. Using the NFAT-GFP reporter to reveal T-cell activation in vivo at a single cell level, we visualized increased nuclear translocation of this transcription factor in T cells in glomerular capillaries, in the presence of cognate antigen targeted to the glomerulus. These findings clearly demonstrate that antigen-dependent T-cell activation occurs intravascularly within the glomerular capillaries. Glomerular targeting of pOVA was necessary for both T-cell activation and downstream intraglomerular neutrophil activation, as these responses were not seen when pOVA was delivered systemically in a non-targeted fashion. Together these data provide compelling evidence that T cells respond to antigen within glomerular capillaries and that intraglomerular antigen recognition leads to proinflammatory changes in the glomerular microvasculature. In some studies, immunogenic peptides capable of binding directly to MHCII without intracellular uptake or processing have been used to examine the immediate response to antigen presentation 31 . As our studies use the peptide recognized by OT-II cells, it is conceivable that a similar response could be at work here. However, as the peptide is covalently conjugated to a much larger immunoglobulin molecule, we anticipate that intracellular processing is the more likely route by which the OVA peptide is loaded onto MHCII.
In this planted antigen model of T-cell-dependent glomerular inflammation, we reasoned that the cell responsible for antigen presentation must either be intravascular or have access to the glomerular capillary lumen. The renal interstitium plays host to an abundant mononuclear phagocyte population 39,41,47,48 , and these cells may access the lumen of the interstitial cortical renal microvasculature via extension of cellular processes 49,50 . However, results of the present experiments indicate that this is not the case in the glomerulus. Furthermore, no intrinsic glomerular cells, including glomerular endothelial cells, expressed detectable MHCII under resting conditions. Therefore, we next examined circulating immune cells with the potential to present antigen to CD4 + T cells. Circulating B cells expressed MHCII. However, OT-II cells could induce glomerular inflammation in the absence of B cells. Therefore we examined the possibility that a monocyte subpopulation mediated this response. In mice, monocytes exist in two major subpopulations: classical/inflammatory (CCR2 hi CX 3 CR1 lo Ly6C + ) and non-classical/ patrolling (CCR2 − CX 3 CR1 hi Ly6C − ). Mounting evidence indicates that the patrolling subset has a specialized role in mediating responses to injury or infection from within the vasculature, including in the glomerulus 2,3,18,51 , and these cells also undergo prolonged retention and migration in the glomerular microvasculature 17,18 . We observed that a subset of the MHCIIexpressing cells that were positive for the monocyte marker CD11b were retained in the glomerulus, often for over 40 min. This prolonged patrolling of the glomerular microvasculature provides opportunity to encounter and internalize antigens while also increasing the probability of this cell encountering intraglomerular T cells. These properties are consistent with an immunosurveillance function for these cells. We also demonstrate that MHCII + monocytes can induce antigen-specific T-cell proliferation in vitro, providing additional support for the contention that these cells are responsible for antigen-specific T-cell activation in glomerular capillaries. With future technical advances, it may be feasible to isolate or image patrolling MHCII + monocytes in glomeruli to demonstrate antigen uptake from within the glomerular microvasculature.
To examine a role for monocytes in T-cell-induced glomerular inflammation, we used clodronate to remove intravascular monocytes. Clodronate did not affect retention of the MHCIIexpressing cells that underwent brief, static retention in glomeruli, consistent with our observations that these were B cells. In contrast clodronate markedly reduced the number of migratory MHCII-expressing cells present in glomeruli, providing further evidence that these cells were monocytes. This depletion strategy resulted in a reduction in the downstream inflammatory readouts of neutrophil retention and ROS-generating activity, consistent with the hypothesis that MHCII-expressing monocytes are responsible for intravascular antigen presentation in this model. We recently reported that monocytes promote innate responses of intravascular neutrophils in acute glomerulonephritis 18 . The present findings add to this by showing that monocytes can also perform the function of intravascular antigen presentation to T cells in T-cell-mediated glomerular injury. These findings are of particular relevance to the initial stages of disease. However in established autoimmune glomerulonephritis, further mechanisms of antigen recognition, including antigen presentation by glomerular endothelial cells with upregulated MHCII expression could contribute to activation of CD4 + T cells 52 .
Alterations in migration are characteristic of T cells undergoing antigen recognition. In naïve cells, altered migration occurs over several hours 53,54 , but effector T cells respond to antigen presentation rapidly via migratory arrest and induction of cytokine production 31,32 . The T-cell responses in the present study are consistent with previous descriptions of effector T cells. Identification of cells undergoing antigen presentation on the basis of nuclear localization of NFAT-GFP revealed that these cells were uniformly static, while cells with non-translocated NFAT-GFP, and therefore not detecting antigen, continued to migrate. This T-cell response occurred within the first hour after T-cell transfer. Analysis of cytokine production provided further evidence of rapid antigen-specific activation by intrarenal T cells in that many intrarenal effector CD4 + T cells were producing IFNγ 4 h after transfer. Finally, changes in neutrophil retention and activation resulting from this rapid T-cell response were apparent within 4 h. These findings highlight the speed with which effector T cells can respond to antigen in glomeruli and promote downstream inflammation. It should be noted that in these experiments, we used effector T cells generated via a standard ex vivo differentiation protocol. It is conceivable that responses of cells generated in this manner may differ from those of cells differentiated in vivo, a possibility that could be explored in future studies.
The actions of neutrophils observed in glomerular capillaries under resting conditions are similar to those reported for neutrophils in uninflamed pulmonary capillaries [55][56][57] . In the lung, this intravascular patrolling may be a form of immune surveillance facilitating rapid responses to local infection. It is possible that this is also an important function of neutrophils in glomeruli. However, microbial infection is less explicitly relevant in the glomerular microcirculation than in the lung. In the glomerulus neutrophil patrolling is a double-edged sword as it also underpins the induction of injurious responses after immune complex deposition or in response to other inflammatory stimuli 17 .
In conclusion, these studies have identified a novel mechanism of intravascular antigen recognition in the glomerular microvasculature, in which circulating CD4 + T cells routinely interact with patrolling antigen-presenting cells. Under these circumstances, deposition of antigen within the glomerular vasculature can result in rapid induction of inflammation in the glomerular capillaries. This is the first study to show the coordinated series of events underlying initiation of intravascular CD4 + T-cell responses in the glomerulus.

Methods
Mice. C57BL/6 wild-type mice were obtained from Monash Animal Research Platform at Monash University and housed in specific pathogen-free conditions. MHC Class II-EGFP knock-in (MHCII-EGFP) mice on a C57BL/6 background, generously provided by B. Fazekas de St Groth (University of Sydney) were bred in-house 37 . μ chain-deficient (μMT) mice, CD11c-YFP mice, Cx3cr1 gfp/+ mice, and OT-II mice, all on a C57BL/6 background, were bred in-house. SMARTA-GFP mice (C57BL/6 background) were generously provided by S. Mueller (University of Melbourne). Male mice between 3 and 22 weeks of age were used in all experiments. All experimental procedures were approved by the Monash Medical Centre Animal Ethics Committee 'B'. Sample size calculation was not performed a priori because the effect sizes of our observations and interventions could not be determined before experiments. This study was not randomized and was not blinded. All experiments were included in the analyses.
The following mAbs were used for flow cytometry (see Supplementary Table 1  from C57BL/6 mice were treated with mitomycin C (50 μg mL −1 , 20 min at 37°C), then washed thoroughly and cultured with naïve OT-II cells at a ratio of 10:1 in medium containing 1 μM OVA 323-339 peptide. Cells were maintained in media containing rmIL-12 (2 ng mL −1 ) and anti-IL-4 mAb (11B11, 10 μg mL −1 ). After 48 h, rmIL-2 (5 μg mL −1 ) was added and maintained thereafter. Cells were split once and harvested on day 7. Cell preparations were centrifuged over Histopaque 1083 or Optiprep (1.09 g mL −1 ) for 30 min at 400 g at room temperature to remove dead cells. 1×10 7 OT-II T cells were transferred into recipient mice via jugular or tail vein unless otherwise stated. For analysis of migration, OT-II T cells were labeled with 5 μM CFSE or 1 μM CMTPX prior to transfer. In some experiments, CD4 + T cells from GFP-expressing SMARTA TCR transgenic mice (specific for the LCMV GP-derived P13 peptide) underwent the same activation protocol, using P13 peptide (1 μM) as the stimulating antigen 63 . In these experiments, SMARTA cells showed comparable levels of activation as OT-II cells (Supplementary Fig. 10).
Antibody peptide conjugation. The 8D1 mAb (of IgG1 subclass) was conjugated to OVA 323-339 using a previously published technique 15 . In brief, 8D1 mAb was mixed with tenfold molar excess of EMCS for 2 h at room temperature. Unreacted EMCS was removed by buffer exchange chromatography, and the activated 8D1 was mixed with a tenfold molar excess of modified OVA 323-339 . After incubation for 3 h at room temperature, the reaction was halted by adding 2 mM cysteine, and the modified 8D1/pOVA antibody was dialyzed in PBS to remove excess OVA 323-339 . For some control experiments, pOVA was conjugated to an IgG1 control antibody of irrelevant specificity (MOPC-21) using the identical protocol.
Induction of T-cell-dependent glomerular inflammation. In order to specifically localize OVA 323-339 to the glomerular vasculature, the 8D1/OVA 323-339 conjugate was used 15,17 , taking advantage of the selective binding of 8D1 to the NC1 domain of α3(IV) collagen in the glomerular basement membrane 33 . To induce glomerular inflammation, 150 μg of 8D1/pOVA was transferred intravenously into recipient mice together with OT-II T cells. Unconjugated 8D1 mAb and pOVA conjugated to MOPC-21 ("IgG/pOVA") served as controls.
Renal intravital multiphoton microscopy. To prepare the kidney for glomerular intravital microscopy, 4-5-week-old mice underwent unilateral ureteric ligation 17,18 . Mice were housed for 12 weeks to allow the kidney to undergo hydronephrosis. Intravital imaging experiments performed on intact kidneys were carried out in 3-4-week-old mice.
Multiphoton microscopy was used for intravital imaging studies 17,18 . Mice were anesthetized by an initial intraperitoneal injection of 150 mg kg −1 ketamine hydrochloride and 10 mg kg −1 xylazine. The left jugular vein was catheterized to allow delivery of fluorescent probes and to maintain anesthesia as required. A heating pad was used to maintain the temperature of the mice at 37°C. The hydronephrotic kidney was exteriorized through a lateral incision and drained of urine using a 30G needle. The kidney was extended over a heated platform and secured with 4-0 silk tied to the kidney capsule. The kidney was superfused with saline (0.9% sodium chloride (wt per vol)) and covered with a coverslip held in place with vacuum grease. For imaging experiments using intact kidneys, mice were anesthetized and catheterized as above. The left kidney was exteriorized through a dorsal incision and the kidney was immobilized in a heated well incorporated into a custom-built stage. The exposed kidney was bathed in normal saline and coverslipped.
Glomeruli were observed with a Leica SP5 multiphoton microscope (Leica Microsystems), equipped with 20× 1.0 NA WI objective lens and a MaiTai pulsed infrared laser (SpectraPhysics). Experiments in C57BL/6 and μMT mice were performed at 810 nm excitation, while experiments in MHCII-EGFP mice were performed at 900 nm excitation. In most experiments, images were taken every 30 s by collecting z-stacks of approximately 150 μm depth, with 6 μm step size. For 2 h imaging experiments, z-stacks were collected every 60 s. Emitted fluorescence was detected by non-descanned detectors with 432-482 nm, 500-550 nm, 575-605 nm, and 625-675 nm emission filters. Pre-defined settings for laser power and detector gain were used for all experiments. For visualization of the vasculature, either Qtracker ® 655, rhodamine dextran or Pacific Blue-conjugated 8D1 was used. To label endogenous leukocytes, the following mAbs were used as appropriate (2 μg i.v. each unless otherwise stated): anti-CD4 PE (CD4 + T cells); anti-B220-PE (B cells); anti-Gr-1 PE or Alexa488 (neutrophils); anti-CD11b NC650 (monocytes -5 μL). OT-II cells were labeled with either CFSE or CMTPX according to the manufacturer's instructions. In experiments examining neutrophil ROS production, mice received 2 mg kg −1 of pre-warmed DHE intravenously 20 min prior to imaging and were examined using 810 nm excitation 17 .
Image analysis. Images were analyzed using Imaris software (Bitplane) 17 . In brief, leukocytes arrested in glomerular capillaries for at least 30 s (two consecutive frames) were defined as adherent and subsequently categorized as crawling or static. Dwell time was defined as the duration of leukocyte adhesion in the glomerulus. To measure the velocity of crawling cells, images were tracked in three dimensions over time. For neutrophil ROS production, neutrophil DHE positivity was determined after adjusting images using pre-determined thresholds to remove background staining.
Renal leukocyte isolation and analysis. For flow cytometry analysis of OT-II cell cytokine production, OT-II cells were isolated from kidneys using a modification of a previously published method 64 . Kidneys were harvested 4 h after transfer of OT-II T cells, infused with RPMI 1640 containing collagenase (1 mg mL −1 ), DNase I (100 μg mL −1 ) and Brefeldin A (10 μg mL −1 ) and incubated at 37°C for 25 min. Kidneys were gently dissociated and incubated at 37°C for a further 25 min then washed and resuspended in RPMI with 5 μg mL −1 Brefeldin A. The cell suspension was left for 10 min to settle tubular debris. The supernatant was filtered through a 70 μm cell strainer and erythrocytes were lysed. Cells were treated with Fc block, then stained for CD45, CD4, and Vβ 5.1/5.2 TCR. For intracellular cytokine staining, cells were fixed and permeabilized using the BD Cytofix/Cytoperm TM kit, as per the manufacturer's instructions, and then stained for IFN-γ.
OT-II cell proliferation was assessed using a modification of a previously published technique 39,40 . In brief, CD4 + T cells were positively enriched from LNs of OT-II mice using anti-CD4 microbeads (Miltenyl-Biotec) (purity >98%) then labeled with CFSE (5 μM). Cells were subsequently washed twice and resuspended at 1×10 6 mL −1 . Sorted monocytes (ranging from 1.6 to 8.8 ×10 4 per well) were pulsed with 1 µM OVA 323-339 and co-cultured with 10 5 CFSE-labeled OT-II cells. Negative controls included T cells pulsed with peptide alone (no antigenpresenting cells) and unpulsed splenocytes (no peptide). As a positive control, splenocytes pulsed with antigen as above were used as antigen-presenting cells. Cells were harvested after 65 h and stained with anti-Vα2 APC (clone B20.1) and anti-CD4 APC-Cy7 (clone GK1.5). Dead cells were excluded by propidium iodide staining and viable cells were enumerated using BD Calibrate beads (BD Biosciences). Samples were acquired on an FACS CantoII flow cytometer and analyzed with FlowJo software (Tree Star), using the level of CFSE staining as an index of proliferation.
OT-II cells were transduced with 1.5-3.2×10 7 transducing units of virus on days 1 and 2 of the activation protocol, in six-well plates in the presence of polybrene (4 μg mL −1 ), using spinfection at 32°C, 800 g for 1 h. On day 7, transduced NFAT-GFP OT-II cells (OT-II NFAT-GFP ) were isolated via flow cytometric sorting on the basis of GFP expression. Typical transduction efficiencies were between 10 and 20%. Preliminary analyses demonstrated that following in vitro activation, the level of IFNγ expression in OT-II NFAT-GFP cells was comparable to that in activated nontransduced OT-II cells ( Supplementary Fig. 3). Sorted OT-II NFAT-GFP cells were labeled with CMTPX and used in in vivo imaging experiments assessing the subcellular localization of GFP 11,43 .
Statistical analysis. All data presented are shown as mean ± s.e.m. Experimental groups were compared using Student's unpaired t tests, or if variances were unequal, unpaired t-tests with Welch's correction or non-parametric Mann −Whitney tests (all one-tailed). In experiments involving more than two groups, one-way ANOVA or Kruskal−Wallis non-parametric analysis was performed. In experiments involving categorical analysis of T-cell phenotype, Fisher's exact test was used. The numbers of mice used in groups were based on the expected degree of variability observed in standard kidney imaging experiments, in parameters such as number of adherent cells and dwell time. Typical group sizes ranged from 6 to 8 individual mice examined on multiple different days, assigned randomly to control or treatment groups. Significance was set at P < 0.05.
Data availability. The data that support the findings of this study are available from the corresponding author on request.