Lymph nodal prion replication and neuroinvasion in mice devoid of follicular dendritic cells

Abstract

Variant Creutzfeldt–Jakob disease and scrapie are typically initiated by extracerebral exposure to prions, and exhibit early prion accumulation in germinal centers. Follicular dendritic cells (FDCs), whose development and maintenance in germinal centers depends on tumor necrosis factor (TNF) and lymphotoxin (LT) signaling, are thought to be indispensable for extraneural prion pathogenesis. Here, we administered prions intraperitoneally to mice deficient for TNF and LT signaling components. LTα^−/−, LTβ^−/−, LTβR^−/−, and LTα^−/− × TNFα^−/− mice resisted infection and contained no infectivity in spleens and lymph nodes (when present). However, TNFR1^−/−, TNFR2^−/−, and some TNFα^−/− mice developed scrapie similarly to wild-type mice. High prion titers were detected in lymph nodes, but not spleens, of TNFR1^−/− and TNFα^−/− mice despite absence of FDCs and germinal centers. Transfer of TNFR1^−/− fetal liver cells into lethally irradiated Prnp^0/0 mice restored infectivity mainly in lymph nodes. Prion protein (PrP) colocalized with a minority of macrophages in tumor necrosis factor receptor (TNFR) 1^−/− lymph nodes. Therefore, prion pathogenesis can be restricted to lymphoreticular subcompartments, and mature follicular dendritic cells are dispensable for this process. Macrophage subsets are plausible candidates for lymphoreticular prion pathogenesis and neuroinvasion in the absence of FDCs, and may represent a novel target for postexposure prophylaxis.

Prion diseases are lethal, transmissible neurodegenerative conditions. The causative agent (1) was proposed to be identical with prion protein Sc (PrP^Sc), a pathological conformer of the cellular protein PrP^C encoded by the cellular gene Prnp. PrP^C is expressed in many sites, notably including secondary lymphoid organs. Peripheral inoculation routes are likely to initiate most forms of spongiform encephalopathies such as sheep scrapie, bovine spongiform encephalopathy (BSE), iatrogenic Creutzfeldt–Jakob disease (CJD) and variant CJD (vCJD). Also, intracerebral (i.c.) or peripheral administration of prions to mice induces a rise of infectivity in spleen and in other lymphoid organs long before the development of neurological symptoms and neuropathological changes. Intraperitoneal inoculation has been extensively used to study the pathogenesis of transmissible spongiform encephalopathies because it causes rapid accumulation of infectivity in secondary lymphoid organs (2–4). The question of which compartments within lymphoreticular tissues support prion replication is of relevance to public health: contamination with vCJD prions of germinal centers in lymph nodal and tonsillar follicles, for example, might call for precautionary measures in handling and sterilization of surgical instruments.

Tumor necrosis factor (TNF) and lymphotoxin (LT) α signal through TNF receptor (TNFR) 1, whereas membrane-bound LTα/β heterotrimers signal through LTβ receptor (LTβR) (5). TNFR1 and LTβR signaling is necessary for development and maintenance of secondary lymphoid organs (6–10). LTβR signaling is also required for maturation and maintenance of follicular dendritic cells (FDCs), which are thought to be essential for prion replication and for accumulation of disease-associated PrP^Sc within secondary lymphoid organs. Inhibition of the LTβ signaling pathway with a soluble receptor, which depletes FDCs (11), abolishes prion replication in spleens and prolongs the latency of scrapie after i.p. challenge (12). B cell-deficient μMT mice (13) are resistant to prions i.p. (14), perhaps because of impaired FDC maturation (12, 15). In addition to FDCs, PrP^C-expressing hematopoietic cells are required for efficient lymphoreticular prion propagation (16, 17).

Here, we studied peripheral prion pathogenesis in mice lacking TNFα, LTα/β, or their receptors. We report that ablation of LTβR signaling prevents all peripheral pathogenesis, whereas ablation of TNFR1 signaling prevents prion pathogenesis in spleen but not in lymph nodes, despite the absence of FDCs.

Materials and Methods

Inoculation of Mice.

Mice were inoculated i.p. with 100 μl of brain homogenate containing 3–6 log LD₅₀ units of the Rocky Mountain laboratory (RML) scrapie strain (passage 4.1) prepared as described (18). Mice were monitored every second day, and scrapie was diagnosed according to standard clinical criteria. Mice were killed on the day of onset of terminal clinical signs of scrapie.

Infectivity Bioassays with tga20 Indicator Mice.

Assays were done on 1% spleen or lymph node homogenates. Spleens of one individual mouse, or inguinal, mesenteric, and superficial cervical lymph nodes pooled from three animals from each experimental group, were collected. Lymph nodes from every individual mouse were pooled for the experiments with fetal liver cell (FLC)-reconstituted chimeric mice. Tissues were homogenized in PBS/5% BSA with a microhomogenizer and passed several times through 18-gauge and 22-gauge needles. When the suspension appeared homogenous, it was spun for 5 min at 500 × g. Supernatants (30 μl) were inoculated i.c. into groups of four tga20 mice (19). Indicator mice were killed after development of terminal scrapie; the relationship y = 11.45 − 0.088x (y, log LD₅₀/ml of homogenate; x, incubation time in days to terminal disease) was used to calculate infectivity titers (20).

Western Blot Analysis.

Tissue homogenates were adjusted to 5 mg/ml (brain) or 8 mg/ml (spleen) protein and treated with proteinase K (20 μg/ml, 30 min, 37°C). Fifty micrograms (brain) or 80 μg (spleen) of total protein of each sample were electrophoresed through an SDS/PAGE gel (12%). Proteins were transferred to nitrocellulose by semidry blotting. Membranes were blocked with TBST/5% nonfat milk, incubated with antibodies 6H4 (brain) or 1B3 (spleen), and developed by Tris-buffered saline + Tween 20 (0.05%)/5% nonfat milk enhanced chemiluminescence (ECL, Amersham Pharmacia) as described (12).

Immunohistochemistry and Immunofluorescence.

Ten-micrometer frozen sections of spleens or lymph nodes were immunostained with the follicular dendritic cell marker FDC-M1 (clone 4C11), CD35 (8C12, PharMingen), peanut agglutinin (PNA), or the pan-B cell marker anti-CD45RO/B220 (RA3-6B2, PharMingen) as previously described (21).

Two-color confocal analysis was performed with antiserum XN to PrP and ER-TR9, MOMA-1, or F4/80 on frozen acetone-fixed tissue sections. For visualization we used secondary FITC-conjugated F(ab′)₂ fragments of goat immunoglobulins against rat (BioSource, Camarillo, CA) and ALEXA 546-conjugated goat anti-rabbit Ig (Molecular Probes, Leiden, The Netherlands). For controls, preimmune sera were used, or, if this option was unavailable, primary antibodies were omitted.

Construction of Fetal Liver Chimeric Mice.

Eight- to ten-week-old mice were reconstituted with lymphohemopoietic stem cells (LSCs) derived from fetal livers. Timed pregnancies of wild-type, TNFR1^−/− and Prnp^0/0 mice served to produce mouse embryos. Fetal livers were collected at embryonic day 14.5–15.5 in DMEM and dissociated by using 18-gauge and 22-gauge needles. After a short spin (15 s, 170 g), the supernatant containing FLCs was collected and diluted in DMEM. FLCs (5–10 × 10⁶ cells) were injected into tail veins of recipients that had been lethally irradiated (1100 rad) 24 h earlier. Six to seven weeks after grafting, successful reconstitution was assessed by PCR analysis of peripheral blood taken from the retroorbital plexus for the presence or absence of the Prnp or TNFR1 gene locus as described earlier (17).

Results

After i.c. inoculation, all treated mice developed clinical symptoms of scrapie with incubation times similar to those of control mice (Table 1). In all genotypes, topography and intensity of spongiosis and gliosis were similar to wild-type mice (not shown), indicating that TNF/LT signaling is irrelevant to cerebral prion pathogenesis. Upon i.p. prion challenge, wild-type mice showed an inverse logarithmic correlation between inoculum size (3–6 log LD₅₀) and incubation time (ref. 20; Table 1). Instead, mice defective in LT signaling proved virtually noninfectible with ≤5 log LD₅₀ (Table 1). Although these mice occasionally developed clinical disease when challenged with an extremely large inoculum (6 log LD₅₀), latency was frequently longer than in wild-type mice (Table 1). No PrP^Sc was detected in brains (Fig. 1A) and spleens (Fig. 1C) of clinically healthy LTα^−/−, LTβ^−/−, LTβR^−/−, or LTα/TNFα^−/− mice 450 days postinfection (dpi), indicating that defects in the LTα/LTβ pathway prevent establishment of subclinical disease (22).

Table 1.

Susceptibility to scrapie of mice lacking LT/TNF signaling components

LT/TNF genotype	Genetic background	i.p. prion inoculation								i.c. inoculation
		6 log LD50		5 log LD50		4 log LD50		3 log LD50		3 × 105 log LD50 i.c.
		Attack rate	Disease latency (avg ± SD)	Attack rate	Disease latency	Attack rate	Disease latency	Attack rate	Disease latency	Attack rate	Disease latency
TNFR1−/−	129Sv × BL/6	11/11	223  ± 20	6/6	240  ± 10	11/12	244  ± 16, 1× > 350	6/8	251  ± 20, 2× > 350	4/4	147  ± 7
TNFR2−/−	C57BL/6	7/7	217  ± 12	5/5	239  ± 23	6/6	254  ± 15	6/6	265  ± 29	ND	—
TNFα−/−	C57BL/6	6/7	209  ± 23, > 370	5/7	226  ± 10, 2× > 370	4/7	321  ± 56, 3× > 370	2/7	313, 368, 5× > 370	4/4	132  ± 12
LTα−/−	C57BL/6	3/9	277  ± 66, 6× > 570	0/8	>570	1/8	503, 7× > 570	1/8	446, 7× > 570	ND	—
LTβ−/−	129Sv × BL/6	4/9	264  ± 32, 5× > 570	0/5	>570	1/8	363, 7× > 570	0/8	>570	5/5	138  ± 9
LTβR−/−	C57BL/6	4/9	272  ± 62, 5× > 540	1/6	240, 5× > 540	0/6	>540	0/6	>540	5/5	148  ± 7
TNFα−/− × LTα−/−	129Sv × BL/6	4/8	230  ± 18, 4× > 570	1/8	308, 7× > 570	0/5	>570	0/5	>570	4/4	140  ± 5
Wild-type	C57BL/6	10/10	199  ± 12	32/32	209  ± 6	6/6	225  ± 9	10/10	243  ± 15	4/4	142  ± 6
Wild-type	129Sv × BL/6	12/12	203  ± 11	6/6	212  ± 7	6/6	223  ± 6	5/5	241  ± 9	ND	—
Prnp0/0	129Sv × BL/6	0/6	>450	ND	—	ND	—	ND	—	0/4	>450

Mice with intact secondary lymphoid organs (TNFR2^−/− and wild-type) developed scrapie upon i.p. challenge with inocula of all sizes, while all other strains were protected to varying degrees. Mean incubation time and standard errors were calculated for mice that developed scrapie. The total observation time is indicated for mice that remained disease free. ND, not determined. 

Figure 1.

Western blot analysis of brains and spleens, as well as determination of prion infectivity titers in spleens and lymph nodes of scrapie-challenged TNF- and LT-deficient mice. (A and B) Western blots of brain material electrophoresed natively (−), or after digestion with proteinase K (PK; +). Large amounts of PK-resistant prion protein (PrP^Sc) were detected in the brain of all mice that had developed scrapie (terminal sick), independently of the genotype. Clinical healthy mice deficient in LT signaling showed no PrP^Sc accumulation, excluding subclinical scrapie (A). (C) Western blots of spleen homogenates electrophoresed natively (−) or after digestion with proteinase K (PK) (+). TNFR1^−/− and TNFα^−/− mice accumulated lower amounts of PrP^Sc as compared with TNFR2^−/− and wild-type mice. No PrP^Sc was detected in the spleens of clinical healthy LT-deficient mice. (D) Prion infectivity in lymphoid organs. Titers were determined in spleens (blue circles), inguinal (red circles), mesenteric (red crosses, x), and cervical (red inverted open triangles) lymph nodes at the time points indicated below. Mice were inoculated i.p. with 6 log LD₅₀ or 4 log LD₅₀ of scrapie prions as indicated. Standard deviations within groups are drawn only when exceeding ±0.75 log LD₅₀. (a and b) In each of the two separate transmissions, one of the four tga20 mice died 24 h after inoculation, most probably because of i.c. bleeding after injection. (j and k) Intercurrent death during incubation time. Symbols on the abscissa indicate prion titers below detection limit (none of the four indicator mice developed scrapie). If one or more indicator mice survived >180 dpi, or the mean incubation time was over 120 days, titer was assumed to be close to the detection threshold of the bioassay. For these samples (labeled with small letters), the numbers of animals succumbing to scrapie of four inoculated tga20 mice and incubation time (in days in parentheses) to terminal scrapie were as follows: c, 1/4 (111); d, 3/4 (93, 117, 132); e, 2/4 (90, 95); f, 1/4 (126); g, 3/4 (112, 123, 128); h, 4/4 (106, 115, 124, 136); i, 3/4 (89, 102, 119); l, 4/4 (115, 115, 126, 135); m, 1/4 (98); n, 2/4 (74, 78); o, 2/4 (125, 151); p, 2/4 (131, 193); q, 3/4 (119, 125, 152); r, 2/4 (109, 117); s, 1/4 (108); t, 1/4 (71); u, 3/4 (80, 96, 98); and v, 2/4 (98, 98).

In contrast, TNFR1^−/− mice were almost fully susceptible to all inoculum sizes, albeit with somewhat longer incubation times than wild-type mice (Table 1). TNFα^−/− mice showed dose-dependent susceptibility: 6/7 mice developed scrapie upon inoculation with 6 log LD₅₀, whereas 2/7 responded to 3 log LD₅₀ (Table 1). Disease latency in those TNFα^−/− mice that developed scrapie after 3–4 log LD₅₀ was vastly prolonged (Table 1). Immunoblot analysis confirmed PrP^Sc accumulation in brains of terminally sick TNFR1^−/− and TNFα^−/− mice (Fig. 1B).

All TNFR2^−/− mice had intact FDCs and germinal centers (not shown) and developed scrapie (Table 1), PrP^Sc accumulation, and prion infectivity in spleens and lymph nodes (Fig. 1 B and D) after i.p. challenge with inocula of any size.

Furthermore, all examined TNFR1^−/−, TNFR2^−/−, and TNFα^−/− mice had all lymph nodes, whereas all LTα^−/−, LTβR^−/−, and LTα^−/− × TNFα^−/− mice exhibited no lymph nodes. Twenty-three of 33 LTβ^−/− mice had some mesenteric, but never cervical, lymph nodes. We then determined prion infectivity over time in spleen and lymph nodes of mice inoculated i.p. (6 log LD₅₀). Infectivity was undetectable in TNF/LT-deficient spleens, with the exception of trace amounts in some TNFR1^−/− and TNFα^−/− spleens, and wild-type prion levels in TNFR2^−/− spleens (Fig. 1D, blue circles). Accordingly, PrP^Sc was undetectable in LT signaling-deficient spleens, and even in spleens of terminally sick TNFR1^−/− and TNFα^−/− mice (Fig. 1C). Mesenteric lymph nodes pooled from three LTβ^−/− mice at 35, 60, and 90 dpi were devoid of infectivity (Fig. 1D, red crosses). We found no infectivity in spleens and lymph nodes of RAG1^−/− (Fig. 1D) and μMT (not shown) mice at 35 dpi, except for trace amounts of prions in RAG1^−/− inguinal lymph nodes (Fig. 1D). Therefore, ablation of mature B cells, the main suppliers of LTα and LTβ, may result in a phenocopy of genetic inactivation of LTβ signaling.

Unexpectedly, all examined lymph nodes (Fig. 1D, red symbols) of TNFR1^−/− and TNFα^−/− mice had consistently high infectivity titers across the entire incubation time (35 dpi to terminal disease). Even inguinal lymph nodes, which are distant from the injection site and do not drain the peritoneum, contained infectivity titers equal to all other lymph nodes. High lymph node titers were detected in wild-type mice and even in TNFR1^−/− mice exposed to 4 log LD₅₀ (Fig. 1D), suggestive of active prion replication. We conclude that TNF deficiency prevents lymphoreticular prion accumulation in spleen but not in lymph nodes.

Differences in germinal center architecture are unlikely to account for the differential prion pathogenesis in TNF- and LT-deficient lymph nodes. Although a thin rim of CD35⁺ marginal zone B cells was present in all analyzed spleens, lymph nodes and spleens of TNFR1^−/− and LTβ^−/− mice were devoid of FDC-M1⁺ or CD35⁺ germinal centers, confirming the absence of mature FDCs. No PNA-positive B cell clusters (indicative of germinal center activation) were detected in lymph nodes and spleens of TNFR1^−/− and LTβ^−/− mice (Fig. 2).

Figure 2.

Absence of mature FDCs and germinal center response in spleen and mesenteric lymph nodes of TNFR1- and LTβ-deficient mice. Frozen sections of spleens (three Upper rows) and mesenteric lymph nodes (three Lower rows) at 35 dpi were immunostained with FDC-M1 and CD35 (for FDCs), as well as with PNA (for activated germinal center B cells). Mature FDCs were detected in secondary lymphoid organs of wild-type mice but not in those of TNFR1- and LTβ-deficient mice, as shown by the lack of FDC-M1 and CD35 staining. Germinal center B cells were seen only in wild-type mice, as indicated by the strong PNA immunostaining.

Next, we generated fetal liver chimeric mice to determine the cellular subsets and hematopoietic and/or stromal components involved in prion propagation in TNFR1^−/− lymph nodes (Table 2). FLCs of the TNFR1^−/−, Prnp^0/0 or Prnp^+/+ genotypes were administered intravenously to lethally irradiated recipient mice (Table 2). Successful reconstitution was assessed by genotyping peripheral blood cells 6–8 weeks after grafting (data not shown). We then determined prion infectivity in spleen and lymph nodes of chimeric mice inoculated i.p. (6 log LD₅₀) 35 days after prion challenge. In Prnp^0/0 mice grafted with TNFR1^−/− FLCs, high infectivity loads (3.4–4.2 log LD₅₀) were detectable in lymph nodes but not in spleens, strongly suggesting that TNFR1^−/− hematopoietic cells support efficient prion propagation within lymph nodes. However, lymph nodes of TNFR1^−/− mice grafted with Prnp^0/0 FLCs also contained substantial amounts of infectivity. Therefore, stromal cells also contribute to the capability of TNFR1-deficient lymph nodes to replicate prions. Reciprocal reconstitution of Prnp^+/+ and Prnp^0/0 mice showed once again that efficient lymphoreticular prion propagation required PrP^C expression in both stromal and hematopoietic compartments, in our hands an extremely consistent finding (16, 17).

Table 2.

Prion load of spleens and lymph nodes in individual FLC-reconstituted chimeric mice

Chimera	Spleen titrations		Lymph node titrations
	Infectivity (log LD50)	Attack rate (mean ± SD, days)	Infectivity (log LD50)	Attack rate (mean ± SD, days)
Wild-type → Prnp0/0	3.5	4 /4 (91 ± 6.6)	3.2	4 /4 (94 ± 10.9)
	3.4	4 /4 (93 ± 5.8)	3.7	4 /4 (87 ± 4.7)
	3.6	4 /4 (89 ± 5.9)	4.6	4 /4 (78 ± 4.9)
Prnp0/0 → wild-type	5.7	4 /4 (66 ± 8.4)	5.1	4 /4 (72 ± 4.7)
	4.6	4 /4 (78 ± 4.3)	4.3	4 /4 (82 ± 7.0)
	5.7	3 /3 (65 ± 4.0)	4.0	4 /4 (85 ± 12.6)
TNFR1−/− → Prnp0/0	<1.5	1 /4 (105)	3.7	4 /4 (88 ± 12.9)
	<1.5	2 /4 (113, 152)	3.4	4 /4 (91 ± 4.4)
	<1.5	0 /4	3.6	3 /3 (90 ± 6.4)
	<1.5	0 /4	4.2	4 /4 (81 ± 8.4)
Prnp0/0 → TNFR1−/−	<1.5	1 /4 (85)	4.1	4 /4 (83 ± 4.4)
	<1.5	0 /4	<1.5	2 /4 (65, 67)
	<1.5	2 /4 (113, 152)	2.8	4 /4 (98 ± 10.7)
	<1.5	1 /4 (97)	<1.5	3 /4 (87, 87, 89)
Wild-type → wild-type	4.3	4 /4 (81 ± 8.1)	4.1	4 /4 (84 ± 3.5)
	4.5	4 /4 (80 ± 1.9)	4.7	3 /3 (76 ± 3.0)
Prnp0/0 → Prnp0/0	<1.5	0 /4	<1.5	0 /4
	<1.5	0 /4	<1.5	1 /4 (137)
TNFR1−/− → TNFR1−/−	<1.5	2 /4 (87, 89)	4.1	4 /4 (83 ± 4.9)
	<1.5	1 /4 (107)	4.5	4 /4 (79 ± 7.5)

Hematopoietic cells from TNFR1^−/− mice are able to confer prion replication capability to Prnp^0/0 FLCs-reconstituted TNFR1^−/− mice indicate that a stromal component plays an additional role in prion propagation. Attack rate, number of indicator mice succumbing to scrapie/number of mice inoculated with prions. 

Two-color immunofluorescence evidenced strong PrP signals in scrapie-infected TNFR1^−/− (Fig. 3A) and TNFα^−/− (not shown) lymph nodes in the absence of germinal centers and mature FDCs. In wild-type mice, PrP colocalized mainly with FDC networks, but strong PrP immunoreactivity was also detected in areas devoid of FDC-M1⁺ cells, suggesting that cells other than FDCs may accumulate PrP^Sc in lymph nodes (Fig. 3A, arrowheads). Indeed, PrP colocalized with the macrophage marker ER-TR9 and MOMA-1 in wild-type lymph nodes (Fig. 3A). No bright PrP signal was detected in secondary lymphoid tissues of LTβ^−/− and Prnp^0/0 mice, concordantly with the finding that these tissues lacked prion infectivity.

Figure 3.

(A) PrP distribution in mesenteric lymph nodes of scrapie-inoculated mice by two-color immunofluorescence. Sections were stained with antibody FDC-M1 to FDCs, ER-TR9, or MOMA-1 to macrophages (green, Upper) and with antiserum XN to PrP (red, Lower). In wild-type mice, PrP reactivity colocalized mainly with FDC-networks. Some PrP immunoreactivity was also detected in areas devoid of FDC-M1-positive cells, which were positive for the macrophage stains (arrows). In TNFR1^−/− lymph nodes, PrP immunoreactivity was detected in the absence of mature FDCs, as indicated by the lack of FDC-M1 staining. No PrP^Sc accumulation was detected in secondary lymphoid tissues of LTβ^−/− and Prnp^0/0 mice. (B) Two-color immunofluorescence confocal analysis of secondary lymphoid tissues of TNFR1^−/−, LTβ^−/−, and wild-type mice. Frozen sections of spleen and mesenteric lymph nodes (MLN) were stained with macrophage-specific antibodies F4/80, ER-TR9, MOMA-1 (green, Top) and with antiserum XN to PrP (red, Middle). In lymph nodes and, to a lesser extent, in spleens of TNFR1^−/− mice, immunoreactivity for PrP colocalized with F4/80-, MOMA-1-, and ER-TR9-positive cells (yellow-orange color, Bottom). No PrP signal associated to macrophage subsets was detected in LTβ^−/− or in Prnp^0/0 lymphoid tissues.

What is the nature of the compartment responsible for the high infectivity of TNFR1^−/− and TNFα^−/− lymph nodes? The above results indicate that this compartment does not consist of mature FDCs, but that both hematopoietic and stromal cells are involved. Confocal two-color immunofluorescence analysis revealed intense PrP signals colocalizing with F4/80-, ER-TR9-, and MOMA-1-positive cells (Fig. 3B) in lymph nodes and, less frequently, in spleens of TNFR1^−/− and TNFα^−/− mice. No colocalization was found with follicular dendritic cell markers (IDC-M1, FDC-M2, and CD35), T cell markers (CD3, CD4, and CD8), B cell markers (B220 and CD35), dendritic cell marker (NLDC-145), natural killer cell marker (Pan-NK), and endothelial cell marker (CD54). As expected, lymphoreticular organs devoid of prion infectivity (LTβ^−/− and Prnp^0/0) displayed no bright PrP immunoreactivity (Fig. 3 A and B).

Discussion

We show here that deletion of TNFR1 or TNFα preserves susceptibility to peripheral prion challenge, whereas deletion of LT signaling components confers high resistance to peripheral prion infection. These differential effects on scrapie susceptibility are surprising, because all defects (except TNFR2^−/−) abolish FDCs, which were thought to be crucial for pathogenesis. However, LT/TNF family members play distinct roles in lymph node induction and development (23): lymphoid organ architecture, including organization of T and B cell zones, is more severely impaired in LT- than in TNF-deficient mice (ref. 23 and data not shown). Because LT and TNF signaling defects may potentially affect several immune cell lineages, we have performed additional adoptive fetal liver cell transfer studies, which complement the gene ablation experiments.

The data above imply that prion replication can take place in secondary lymphoreticular organs even in the absence of mature FDCs, and that other cells can maintain replication of prions to titers that are similar to those of wild-type mice. Moreover, prion pathogenesis in the lymphoreticular system can be topographically compartmentalized, and lymph nodes (rather than the spleen) can represent an important reservoir of prion infectivity during disease.

We then investigated the role of hematopoietic and stromal cells for peripheral prion pathogenesis in TNFR1^−/− mice by using fetal liver chimeras. We found that hematopoietic cells from TNFR1^−/− mice restored prion replication in lymph nodes but not in spleens of Prnp^0/0 mice. On the other hand, prion loads were present also in reciprocal (Prnp^0/0→ TNFR1^−/−) chimeras, indicating that immature, non FDC-M1-positive FDC precursors, or other stromal cells might play an additional role in prion propagation in the lymph nodes of these mice. These findings are in line with the previous results that PrP-null hosts, recolonized with PrP-expressing hematopoietic cells, accumulated prions in lymphoid organs for >200 days after inoculation (16, 17), and reinforce once again that a hematopoietic component, in addition to a stromal one, is required for efficient prion replication in secondary lymphoid organs in this model of prion pathogenesis.

In addition, we found now that, in the absence of mature FDCs, the PrP signal colocalized with a subset of single red-pulp, metallophilic, and marginal-zone macrophages in TNFR1^−/− lymph nodes. All of the above imply that the cells involved in prion replication and PrP^Sc accumulation in the lymph nodes of TNFR1^−/− and TNFα^−/− mice are likely to be macrophages, or some subsets thereof. This hypothesis is strengthened by the fact that, even in wild-type lymph nodes, bright PrP signals outside FDC-networks colocalized with a subset of ER-TR9- and MOMA-1-positive cells (Fig. 3A). TNFα signaling through TNFR1 is required for proper homing of macrophages to the splenic marginal zone, and their absence can cause strong aberrations in macrophage subsets. However, MOMA-1⁺ and ER-TR9⁺ macrophages were normally distributed in the subcapsular area of TNFR1^−/− and TNFα^−/− lymph nodes (ref. 24, and data not shown), but were strongly disturbed in mesenteric lymph nodes of LTβ^−/− mice. If the distribution of macrophages is important for peripheral prion pathogenesis, these histoarchitectural differences may account for the differences in splenic vs. lymph nodal prion load of infected TNFR1^−/− and TNFα^−/− mice.

Primary and secondary follicles may be functionally different in spleen vs. mesenteric lymph nodes of TNFR1^−/− mice (25). However, we did not identify morphological differences between splenic and mesenteric germinal centers, and the TNFR1^−/− line used here (26) did not show abnormal germinal center responses after infection with vesicular stomatitis virus (21).

The unexpected finding of high prion titers in inguinal, mesenteric, and cervical lymph nodes of TNF-deficient mice, but not in mesenteric lymph nodes of LTβ-deficient mice, indicates that prion replication within secondary lymphoid organs is LTβR dependent, yet may occur in the absence of mature FDCs and of functional germinal centers, as revealed by the lack of PNA-positive B cell clusters. Therefore, cell types other than FDCs participate in the process of prion replication/accumulation in lymph nodes and, probably, in spleens. Because marginal zone macrophages might entertain close contacts to immature FDCs in the marginal zone, whose presence was postulated for the TNFR1^−/− mice (24), and also interact with marginal zone B cells, this hematopoietic cell type is certainly a candidate for supportive effects in the process of prion uptake and replication. There is a caveat to this interpretation: immunofluorescence detects prion protein rather than infectivity, and does not differentiate unequivocally between PrP^C and PrP^Sc. Therefore, further studies will need to focus on whether macrophage ablation, e.g., by using macrophage-specific suicide transgenes, can suppress the infectibility of TNF-deficient lymph nodes.

These findings are at striking variance with reports that LTβ^−/− mice are fully susceptible to infection with CJD prions (27), and that TNFα^−/− mice peripherally challenged with ME7 prions were largely protected (28). These and other discrepancies have been attributed to the use of different prion strains in these studies. This may well be the case, but the present data indicate that resistance in each mouse strain is dose dependent and can always be overridden (Table 1). Therefore, challenge with one single size of inoculum, as done in other studies, may yield misleading results. Invasion of lymphoid organs by prions occurs very rapidly after peripheral inoculation, and consistently high infectivity titers are detected until terminal disease. Lymphoinvasion most likely plays an important role in the pathogenesis of vCJD, because prion infectivity can be detected in tonsils of virtually every vCJD patient (29, 30). After lymphoinvasion, neuroinvasion occurs via autonomic nerves (31–35), but the nexus between germinal centers and nerves is still elusive. By virtue of their mobility, macrophages may represent a plausible candidate for transport of prion infectivity from germinal centers to sympathetic nerve terminals.

How could a possible prion amplification in macrophages be reconciled with their apparent protective role, at least in the very early phase of prion pathogenesis (36)? Maybe the action of macrophages is dose dependent: small inocula may be destroyed by phagocytosis, whereas larger inocula cannot be digested and will be transported or amplified. Alternatively, the absence of TNFR1 may interfere directly with the interaction of macrophages and prions, because ablation of TNF signaling reduces the phagocytic ability of macrophages in several infectious models (37, 38). Be that as it may, the fact that a cell type other than mature FDCs is involved in prion replication and accumulation within secondary lymphoid organs may help developing postexposure prophylaxis strategies aimed at blocking prion neuroinvasion.

Abbreviations

FDC

follicular dendritic cell

LT

lymphotoxin

PNA

peanut agglutinin

PrP

prion protein

TNF

tumor necrosis factor

vCJD

variant Creutzfeldt–Jakob disease

TNFR

TNF receptor

FLC

fetal liver cell

dpi

days postinfection