Lymphotoxin, but Not TNF, Is Required for Prion Invasion of Lymph Nodes

Neuroinvasion and subsequent destruction of the central nervous system by prions are typically preceded by a colonization phase in lymphoid organs. An important compartment harboring prions in lymphoid tissue is the follicular dendritic cell (FDC), which requires both tumor necrosis factor receptor 1 (TNFR1) and lymphotoxin β receptor (LTβR) signaling for maintenance. However, prions are still detected in TNFR1−/− lymph nodes despite the absence of mature FDCs. Here we show that TNFR1-independent prion accumulation in lymph nodes depends on LTβR signaling. Loss of LTβR signaling, but not of TNFR1, was concurrent with the dedifferentiation of high endothelial venules (HEVs) required for lymphocyte entry into lymph nodes. Using luminescent conjugated polymers for histochemical PrPSc detection, we identified PrPSc deposits associated with HEVs in TNFR1−/− lymph nodes. Hence, prions may enter lymph nodes by HEVs and accumulate or replicate in the absence of mature FDCs.


Introduction
Prions are unusual infectious agents thought to be comprised solely of an abnormally folded, aggregated isoform (PrP Sc ) of the endogenous cellular prion protein (PrP C ) [1]. Deposition of PrP Sc aggregates, vacuolation, and neuronal loss in brain tissue are histopathological hallmarks of a group of neurological disorders collectively known as prion diseases or transmissible spongiform encephalopathies (TSEs), including scrapie in sheep, bovine spongiform encephalopathy (BSE) in bovids, chronic wasting disease (CWD) in cervids, and Creutzfeldt-Jakob disease (CJD) in humans [2].
However, extraneural prion accumulation in SLOs is not strictly dependent on the presence of mature FDCs. Although prion titers remain below detection in spleens of i.p.-inoculated TNFR1 2/2 and TNFa 2/2 mice, PrP Sc levels and prion infectivity in TNFR1 2/2 and TNFa 2/2 lymph nodes are only marginally reduced compared to TNFR1 2/2 , LTbR 2/2 , LTa 2/2 , or LTb 2/2 spleens [30,31]. Furthermore, TNFR1 2/2 and TNFa 2/2 mice succumb to terminal disease upon i.p. prion inoculation at a noticeably higher rate than lymphotoxin signaling-deficient mice, indicating that prions are still effectively transmitted to the CNS in the absence of TNFR1 signaling. Since TNFR1 2/2 lymph nodes are devoid of detectable mature FDCs, this implies that other undefined cell types may also be required for prions to colonize SLOs. However, lymph nodes are either absent or profoundly disrupted in LTbR 2/2 , LTa 2/2 , and LTb 2/2 mice compared to TNFR1 2/2 and TNFa 2/2 , making it difficult to formally conclude that LTbR signaling is specifically required for prion accumulation in lymph nodes while TNFR1 is not. To determine whether continuous LTbR signaling is required for prions to accumulate in TNFR1 2/2 lymph nodes, we investigated the ability of prions to colonize SLOs of prion-infected TNFR1 2/2 mice treated with an LTbR-Ig blocking antibody.

Prion accumulation in TNFR1 2/2 lymph nodes is LTbR signaling-dependent
We previously showed that mice devoid of TNF signaling accumulate prion infectivity and PrP Sc in lymph nodes but not in spleen, in contrast to LT signaling-deficient mice which do not accumulate prions in either spleen or lymph nodes [31]. To determine whether prion accumulation in TNFR1 2/2 lymph nodes was dependent on continuous LTbR signaling, we administered weekly 100 mg intraperitoneal (i.p.) injections of an LTbR immunoglobulin fusion protein (LTbR-Ig) or control pooled human immunogloblulin (Ig) to wild-type (WT) and TNFR1 2/2 mice to achieve sustained inhibition of LTbR signaling [32]. One week following the initial LTbR-Ig or control Ig injection, mice were inoculated intraperitoneally (i.p.) with 6 log LD 50 RML6 prions. At 60 days postinjection (d.p.i.), spleens and mesenteric lymph nodes (mLNs) from these mice were assessed for accumulation of prion infectivity and PrP Sc . To confirm that LTbR-Ig treatment effectively inhibited LTbR signaling, we analyzed follicular dendritic cell marker 1 (FDCM1) immunoreactivity in spleens from WT or TNFR1 2/2 mice treated with either control Ig or LTbR-Ig. FDCM1 immunoreactivity was absent in spleens and mLNs from TNFR1 2/2 -Ig, WT-LTbR-Ig, and TNFR1 2/2 -LTbR-Ig spleens in contrast to WT-Ig, indicating that LTbR-dependent FDCs had de-differentiated in response to LTbR-Ig treatment ( Fig. 1A-J). In addition, we analyzed transcriptional targets of the LTbR pathway in spleens from WT or TNFR1 2/2 mice treated with either control Ig or LTbR-Ig. As expected, levels of both NFkB2 (p100) and CXCL13 mRNA were reduced in TNFR1 2/2 -Ig, WT-LTbR-Ig, and TNFR1 2/2 -LTbR-Ig spleens compared to WT-Ig ( Fig. 1K & L).
Next, to determine the effect of inhibited LTbR signaling on prion accumulation in SLOs, we compared the pattern of PrP Sc deposition in spleens and mLNs from prion-infected TNFR1 2/2 -Ig, WT-LTbR-Ig, and TNFR1 2/2 -LTbR-Ig mice to WT-Ig mice by histoblotting. As expected, TNFR1 2/2 -Ig, WT-LTbR-Ig, and TNFR1 2/2 -LTbR-Ig spleens accumulated less PrP Sc than WT-Ig spleens ( Fig. 2A-D), though chronic LTbR-Ig administration seemed less effective at preventing PrP Sc deposition in WT spleens than genetic ablation of TNFR1 (compare Fig. 2B with 2C). Since the 60 day treatment period approaches the limit of effective LTbR-Ig inhibition, this most likely reflects partial FDC re-maturation in WT SLOs near the end of the experiment (J. Browning; personal communication). Regardless, the ability of LTbR-Ig to block prion replication in WT SLOs is well-established [26,27]. Consistent with our previous studies, mLNs from WT and TNFR1 2/2 -Ig-treated mice contained similar numbers of PrP Sc deposits ( Fig. 2E & F), whereas PK-resistant PrP deposits in WT-LTbR-Ig mLNs were less numerous than in WT-Ig or TNFR1 2/2 -Ig mLNs (Compare Fig. 2G with Fig. 2E & F). Background PrP immunoreactivity in PK-digested histoblots from non-infected wild-type spleens was negligible (Supp. Fig. S1).
Of note, TNFR1 2/2 -LTbR-Ig mLNs were devoid of PrP Sc immunoreactivity (Fig. 2H), demonstrating that prion accumulation in lymph nodes in the absence of TNFR1 is dependent on LTbR signaling. To confirm this result quantitatively, we analyzed prion infectivity in mLNs from Ig-treated versus LTbR-Ig-treated WT and TNFR1 2/2 mice using the scrapie cell assay (SCA; [33,34,35]). Consistent with the corresponding histoblots from these mice, LTbR-Ig treatment decreased prion infectivity in WT and TNFR1 2/2 mLNs by 2-3 log tissue culture infectious (TCI) units per gram of tissue compared to Ig treatment (Fig. 2I).
To determine the effect of LTbR-Ig treatment on PrP C expression in SLOs, which might explain the differential ability of TNFR1 2/2 -Ig versus TNFR1 2/2 -LTbR-Ig mLNs to accumulate prions, we measured Prnp mRNA levels in spleens and lymph nodes from all treatment groups by quantitative PCR. Prnp mRNA levels were reduced in WT-LTbR-Ig, TNFR1 2/2 -Ig, and TNFR1 2/2 -LTbR-Ig spleens compared to WT-Ig (Fig. 2J), most likely reflecting de-differentiation of FDCs, the primary PrP Cexpressing cell type in spleen. In contrast, no differences in Prnp mRNA expression were found in mLNs from mice of different treatment groups. Of particular note, no difference in Prnp mRNA expression was found between TNFR1 2/2 -Ig and TNFR1 2/2 -LTbR-Ig mLNs (Fig. 2K). Taken together, our data indicate that TNFR1 2/2 lymph nodes accumulate prions in the absence of mature FDCs yet in an LTbR-dependent manner. Moreover, inhibited prion deposition in mLNs upon loss of LTbR signaling seems to be unrelated to local Prnp levels.

MadCam1 expression correlates with prion infectivity in mesenteric lymph nodes
We reasoned that prion accumulation in TNFR1 2/2 lymph nodes might rely on a putative LTbR signaling-dependent, TNFR1 signaling-independent cell present in lymph nodes but not in spleens. In order to identify such a cell, we screened spleens and mLNs from WT-Ig, WT-LTbR-Ig, TNFR1 2/2 -Ig, and TNFR1 2/2 -LTbR-Ig mice for a variety of hematopoietic and stromal cell markers by immunohistochemistry (IHC) whose expression correlated with prion replication ability. As previously noted, the pattern of FDCM1 immunoreactivity was consistent with PrP Sc deposition in spleen, but not in mLNs (

Author Summary
Prions are unique infectious agents thought to be composed entirely of an abnormal conformer of the endogenous prion protein. Prions cause a severe neurological disorder in humans and other animals known as prion disease. Though prion disease can arise spontaneously or from genetic mutations in the gene encoding the prion protein, many cases of prion disease arise due to peripheral exposure to the infectious agent. In these cases, prions must journey from the gastrointestinal tract and/or the bloodstream to the brain. Prions often colonize secondary lymphoid organs prior to invading the nervous system via neighboring peripheral nerves. Prion accumulation in follicular dendritic cells found in germinal centers of lymphoid organs is thought to be a crucial step in this process. However, prion colonization of lymph nodes, in contrast to spleen, does not depend on follicular dendritic cells, indicating that other mechanisms must exist. Here, we identify the signaling pathway required for follicular dendritic cell-independent prion colonization of lymph nodes, which also controls the differentiation of high endothelial venules, the primary entry point for lymphocytes into lymph nodes. Importantly, prions could be found within these structures, indicating that high endothelial venules are required for prion entry and accumulation in lymph nodes.  . PrP Sc accumulation in TNFR1 2/2 lymph nodes requires LTbR signaling independent of Prnp expression. C57BL/6 (WT) or TNFR1 2/2 mice inoculated i.p. with 6 log LD 50 RML6 and treated weekly with control Ig or LTbR-Ig were sacrificed at 60 d.p.i. Histoblots were performed on frozen sections from spleens (SPL; A-D) or mesenteric lymph nodes (mLN; E-H) from mice in each treatment group to visualize PrP Sc deposition. Whole organs are shown in left panels, and corresponding higher resolution images for each treatment group are shown in right panels. Note that lack of TNFR1 signaling can prevent PrP Sc accumulation in spleen (B) but not lymph node (F). However, blocking LTbR signaling can prevent PrP Sc accumulation in TNFR1 2/2 lymph nodes (compare F and H). Prion infectivity titers in mLN homogenates from individual TNFR1 2/2 Ig-treated and LTbR-Ig treated mice were measured using the scrapie cell assay (I). Whereas TNFR1 2/2 -Ig mLNs all harbored $6.1 log TCI units/g tissue, prion infectivity in TNFR1 2/2 -LTbR-Ig mLNs was reduced by at least 2.5 log TCI units/g tissue. Total mRNA was isolated from spleens (J) or mesenteric lymph nodes (mLN; K) of mice from the indicated treatment groups and  PCR in mLNs quantitatively corroborated the MadCam1 IHC findings: MadCam1 mRNA levels were equally reduced in spleens from WT-LTbR-Ig, TNFR1 2/2 -Ig, and TNFR1 2/2 -LTbR-Ig mice, compared to WT-Ig (Fig. 4E). In contrast, MadCam1 mRNA levels were intermediately reduced in TNFR1 2/2 -Ig mLNs compared to WT-Ig, and a further reduction in MadCam1 mRNA levels was observed in TNFR1 2/2 -LTbR-Ig mLNs compared to TNFR1 2/2 -Ig (Fig. 4F). Thus far, our data suggested that the presence of a MadCam1-expressing cell was associated with accumulation of prions in lymph nodes but not in spleen.
PrP Sc localizes to HEVs in TNFR1 2/2 lymph nodes To determine whether prions were localized to HEVs in mLNs, we performed co-IF with PrP C and MadCam1 antibodies in prioninfected TNFR1 2/2 -Ig mLNs. Prion-infected TNFR1 2/2 -Ig mLNs contained areas of intense PrP-positive deposits which were not detectable in non-infected TNFR1 2/2 -Ig mLNs (data not shown; [31]). Of note, many of these PrP C -positive areas localized to MadCam1-postive vessels in prion-infected TNFR1 2/2 -Ig mLNs, indicating that HEVs are probable sites of PrP Sc localization in infected lymph nodes (Fig. 6A-C). However, PrP immunostaining of prion-infected tissue cannot reliably distinguish between PrP Sc deposits and sites of high PrP C expression, and HEVs expressed relatively high levels of PrP C in uninfected mLNs ( Fig. 5G-I).
To develop an independent method of distinguishing PrP Sc from PrP C in lymphoid organs, we tested the ability of a series of fluorescent amyloid-binding dyes known as luminescent conjugated polymers (LCPs [39,40,41,42]) to stain PrP Sc deposits in spleen and lymph node. LCPs were previously shown to recognize PrP Sc in brain [43]. One LCP, pentameric formic thiophene acetic acid (p-FTAA; [44]), stained PrP-positive FDC networks of prioninfected spleens from WT mice (Supp. Fig. S5). In contrast, no immunofluorescence could be detected with p-FTAA in spleens and lymph nodes from uninfected mice (Supp. Fig. S6). Using this method, points of PrP Sc /MadCam1 co-localization could be reliably identified in prion-infected TNFR1 2/2 lymph nodes, which again indicated that a proportion of PrP Sc localizes to HEVs in TNFR1 2/2 mLNs (Fig. 6D-F). However, at lower magnification we also noted that a number of PrP Sc deposits were located outside of HEVs (Fig. 6G-I). To confirm this observation and to analyze the tissue-wide distribution of PrP Sc deposits relative to HEVs in prion-infected TNFR1 2/2 -Ig mLNs, we performed histoblot co-stains with PNAd antibody, which is highly immunoreactive to HEVs and can be visualized even after proteinase K digestion (Supp. Fig. S7). This analysis confirmed that some PrP Sc deposits were localized to HEVs in TNFR1 2/2 -Ig mLNs. However, PrP Sc deposits were also present outside of HEVs (Fig. 6J-K), which is consistent with our previous study showing that strong PrP immunoreactivity in prion-infected TNFR1 2/2 mLNs can also be found in other cell types (i.e. macrophages [31]). Since HEVs could potentially serve as entry portals for prionharboring lymphocytes, and it was recently reported that neighboring dendritic cells (DCs) are responsible for HEV differentiation [45], we also performed PrP co-immunofluorescence on prion-infected TNFR1 2/2 -Ig mLNs using the DC marker, CD11c, to determine whether DCs in the vicinity of HEVs might also contain PrP Sc . However, no overlap of PrP and CD11c immunoreactivity in prion-infected TNFR1 2/2 -Ig mLNs was identified (Supp Fig. S8).

Discussion
The means by which prions evade the immune system's numerous defense mechanisms and finally transmigrate to, and selectively damage, neurons of the CNS has been the subject of scientific scrutiny for two decades. A number of studies have implicated FDCs in the germinal centers of SLOs as the primary reservoirs of prions prior to neuroinvasion [4,12,13,14,15,16,17]. Yet the ability of TNFR1 2/2 mLNs to accumulate prions with a minimal loss of infectivity compared to WT mLNs presents an apparent paradox, since FDC maintenance depends on TNFR1 signaling [21,22,23].
Here we have established that lymph nodal prion accumulation in the absence of TNFR1 signaling is LTbR signaling-dependent. Crucially, transient loss of LTbR signaling was sufficient to block TNFR1-independent prion accumulation in lymph nodes, indicating that prion accumulation in lymph nodes specifically requires LTbR signaling and is not simply prevented by general developmental defects or architectural disruptions caused by lack of LTbR signaling in LTb 2/2 lymph nodes.
We previously showed that intense PrP immunoreactivity was localized to macrophages in prion-infected TNFR1 2/2 mLNs [31], and others have reported that PK-resistant PrP is localized to macrophages in spleens with PrP C -deficient FDCs [46], indicating that macrophages serve as alternative sites of prion accumulation in the absence of PrP C -expressing FDCs. How this phenomenon was mechanistically linked to LTbR signaling and the pattern of prion accumulation in lymph nodes was initially unclear, since most macrophage populations were preserved in the absence of both TNFR1 and LTbR signaling [31]. Here, we discovered that loss of LTbR signaling in mLNs was also correlated with the dedifferentiation of HEVs -the primary point of entry for lymphocytes into lymph nodes and a likely determinant in the ability of prions to colonize lymph nodes.
Consistent with the pattern of prion accumulation in spleens and lymph nodes of mice lacking TNF and/or LT signaling components, HEVs exist in lymph nodes and other SLOs but not spleens [37], and the maintenance of HEV architecture is TNFR1 signaling-independent yet LTbR signaling-dependent [47]. Moreover, we identified sites of PrP Sc -HEV overlap in TNFR1 2/2 -Ig mLNs, indicating that HEVs might replicate prions and/or serve as points of entry for prions or prion-harboring lymphocytes. FDC-deficient mice can succumb to scrapie even in the absence of detectable splenic prion titers [24]. In light of our current results, this phenomenon is most likely explained by HEV-dependent entry and accumulation of prions in lymph nodes and other HEVcontaining SLOs, such as Peyer's patches.
HEVs can also form ectopically in certain chronic inflammatory conditions. Prion accumulation at sites of chronic inflammation has previously been observed, but in most cases this could be attributed to local formation of FDC networks [48,49,50]. However, we also previously reported LTbR-dependent prion colonization of granulomas in the absence of FDC markers [51]. Since ectopic HEV  formation without any evidence of FDC networks has previously been reported in certain inflammatory conditions [52], ectopic HEV formation might be a source of prion replication and/or uptake in granulomas, as well.
Of note, LTa/b signaling to HEVs differs from LTa/b signaling to FDCs. Though HEVs express LTbR [53], HEVs persist in the absence of B cells [47], in contrast to FDCs which rely on B cells to provide LTa/b ligand [19,20]. This indicates that the LTa/b signal to HEVs emanates from another cell type. Until recently the cell type providing LTa/b to HEVs was not known; however a recent publication reports that dendritic cells (DCs) may be the source of LTa/b signaling to HEVs [45]. If HEV differentiation is indeed B cell-independent, this may explain why prion neuroinvasion can occur in the absence of B cells in some cases, since LTbR signaling to HEVs and hence the ability of prions to enter and accumulate in lymph nodes would be preserved [24,54].
Do prions actually replicate in TNFR1 2/2 lymph nodes, or do they simply accumulate? Accumulation seems likely, as the only cell type in SLOs known to replicate prions are FDCs, and prion replication in macrophage populations was not observed [46]. In any case, experimental evidence suggests that prion accumulation in SLOs is sufficient for prions to invade the CNS [55,56,57]. Moreover, deletion of complement factors dramatically impairs prion accumulation in SLOs and subsequent neuroinvasion without affecting PrP C levels [58], indicating that the ability of SLOs to physically capture prions is indeed critical for the development of downstream pathology.
Curiously, we previously showed that PrP C expression is required in either the stromal or the hematopoietic compartment for lymph nodes to accumulate prion infectivity [31], which implies that prion replication can occur in both compartments in lymph nodes. However, an alternative explanation is that a hematopoietic cell is required for delivery of prions to a ''trapping'' cell within the lymph node, and PrP C expression on the hematopoietic cell mediates efficient uptake of PrP Sc in the bloodstream. Hence, HEVs may facilitate the selective uptake and accumulation of prions or prioncontaining lymphocytes into lymph nodes, rather than serving as sites of bona fide prion replication. Based on both current and past findings, it is likely that the ''trapping'' cells in lymph nodes are endothelial cells in HEVs (HEVECs), and the hematopoietic delivery cells are macrophages. Once prions or prion-harboring cells have successfully invaded mLNs through HEVs, they may be transported via conduits to FDCs [59] under normal conditions. However, in the absence of mature FDCs, prions may remain in or be transferred to macrophages.

RML inoculum
CD1 wild-type mice were inoculated i.c. with 30 mL 1% (w/v) RML-5-infected brain homogenate and sacrificed at terminal stage disease. Brains were flash frozen in liquid nitrogen and stored at 280uC. Brains were then homogenized in sterile 0.32 M sucrose in 1x PBS (20% w/v) using a Ribolyser (Hybaid, Catalys). Subsequent dilutions of inoculum were performed in sterile 5% (w/v) bovine serum albumin (BSA) in 1x PBS. For titer determinations, serial dilutions of RML6 were inoculated i.c. into indicator mice, and LD 50 was defined as the dilution that induced a 50% attack rate.

Real time PCR
Frozen spleens and lymph nodes were homogenized in Qiazol (10% w/v) using a TissueLyser, and total mRNA was isolated using an RNeasy Mini kit, according to the manufacturer's instructions (Qiagen). 1 mg mRNA from each sample was used for first-strand cDNA synthesis using random hexamers from a Superscript II kit (Invitrogen). ,100 ng cDNA, 500 nM primers, and 1x Faststart SYBR Green reaction mixture (Roche) in 25 mL reaction volumes were used for Real Time PCR amplification of target sequences from spleen and lymph nodes normalized to GAPDH using an ABI 7900HT (Applied Biosystems). (See Table 1 for list of primer sequences.) Reactions from each tissue were performed in triplicate and averaged. Samples with a triplicate variability exceeding 10% were eliminated from further analysis. Amplification data was analyzed using the relative quantification method (RQ) with WT-Ig samples serving as the calibrator values. RQ values were calculated and averaged for each sample. Average values depicted in the graphs represent the mean value of single spleens or lymph nodes from each individual mouse from each treatment group. N = 2-4 mice per group, depending on the specific tissue and treatment group.

Scrapie cell assay in endpoint format
Scrapie-susceptible neuroblastoma cells (subclone N2aPK1, [33]) were incubated with uninfected brain homogenate, defined titers of RML6-infected brain homogenate, or 10 23 to 10 26 dilutions of mesenteric lymph node homogenate from WT-Ig, WT-LTbR-Ig, TNFR1 2/2 -Ig, or TNFR1 2/2 -LTbR-Ig for 3 days. Infected N2aPK1 cells were passaged 1:3 three times every 2 days, and then 1:10 four times every 3 days. After reaching confluence, 2610 4 cells from each well were filtered onto the membrane of an ELISPOT plate (Millipore; MultiScreenHTS filter plates with Immobilon-P PVDF membrane) and denatured with 0.5 mg/mL proteinase K (PK). Individual prion-infected cells were immunodetected with POM1. Wells were scored positive if the spot number exceeded mean background values, determined as three times the standard deviation of the uninfected control. In this experiment, an ELISPOT membrane with $3 PrP Sc+ colonies was regarded as infected. From the proportion of negative to total wells, the number of tissue culture infectious units per mL was calculated with the Poisson equation. In two independent experiments, a 10 28 dilution of a standard inoculum (brain homogenate from a terminally scrapie-sick mouse) yielded 11/24 or 12/24 positive wells, corresponding to a titer of ,8.3 log tissue culture infectivity (TCI) units/g of brain tissue for the initial inoculum.
The sensitivity threshold was calculated to be 2.8 log TCI units/g of brain tissue. Immunohistochemistry 7 mM frozen sections on glass coverslips (Thermofisher) were dried for several hours at room temp, fixed in 4% formalin for 2 min, 50% acetone for 2 min, 100% acetone for 2 min, and 50% acetone for 2 min. Sections were then washed in 1x PBS, then 1x PBS+ 0.05-0.1% Tween 20, and blocked for 1 hr. in SuperBlock (Pierce). Sections were then incubated overnight at 4uC in primary antibody (see Table 2 for antibodies and dilutions) diluted in 1:10 SuperBlock. For vessel stains, isotype controls (rat IgM for PNAd, Pharmingen # 553941; rat IgG2a for MadCam1, eBioscience # 14-4321; and mouse IgG1 for PrP, Sigma # 15381) were performed using equivalent concentrations to the corresponding primary antibodies (Supp. Fig. S4). Sections were then washed in 1x PBS, followed by 1x PBST, then incubated with 0.2-0.4 mg/mL Alexa Fluor secondary antibody (Invitrogen) diluted in 1x PBST (for immunofluorescence [IF]) or 5.3 mg/mL unconjugated goat anti-rat secondary antibody (Caltag Laboratories) or goat antimouse (Jackson Immunoresearch) diluted in 1:10 SuperBlock for 1 hr. at room temperature (for light microscopy [LM]). Sections were then washed in 1x PBS followed by 1x PBST and incubated in 7.5 mg/mL AP-conjugated donkey anti-goat tertiary antibody (Jackson Immunoresearch) in 1x PBST for 1 hr for LM. Sections were then washed and developed with Fast Red (Sigma) staining kit and counter-stained with hematoxylin & eosin (H&E) for LM. Sections were then mounted (fluorescent mounting medium for IF or aqueous mounting medium for LM; Dako) and coverslipped. Sections were imaged using an Olympus BX61TRF fluorescent microscope, a Zeiss Axiophot light microscope, or a Leica SP5 confocal microscope (where indicated).  Pentameric formic thiophene acetic acid (pFTAA) costaining 7 mM frozen sections on glass coverslips were dried and then fixed in pre-chilled 100% acetone or ethanol at 220uC for 10 min. Sections were dried for 1 min., re-hydrated in 1x PBS for 10 min., blocked in SuperBlock, and then incubated in 20 mg/mL Mad-Cam1 or 6.25 mg/mL FDCM1 in 1x PBS overnight at 4uC. Sections were then washed in 1x PBS and incubated with 0.2-2 mg/mL Alexa Fluor 594-conjugated goat anti-rat secondary antibody for 1 hr. Sections were then washed in 1x PBS and incubated with 30 mM pentameric formic thiophene acetic acid (p-FTAA; [44]) in 1x PBS for 30 min. Sections were then washed in 1x PBS, mounted with fluorescent mounting medium (Dako), coverslipped, and imaged using an Olympus BX61TRF fluorescent microscope. Sections from uninfected WT spleens were used as negative controls for non-specific pFTAA staining (Supp. Fig. S6).