Scrapie Pathogenesis: The Role of Complement C1q in Scrapie Agent Uptake by Conventional Dendritic Cells¹

Transmissible spongiform encephalopathies (TSEs)³ or prion diseases are a group of infectious neurodegenerative diseases that affect humans as well as domestic and wild animals. TSE pathology is characterized by neuronal loss, glial activation, and amyloid accumulation of the (abnormal) prion protein (PrP), scrapie PrP (PrP^Sc), in the brain. The host (normal) cellular PrP (PrP^c) is a broadly expressed, proteinase K-sensitive, detergent soluble, membrane glycoprotein with which presence is crucial for prion disease susceptibility. Intriguingly the difference between PrP^Sc and PrP^c is mainly conformational because PrP^Sc has high β-sheet content in contrast to PrP^c, which has a high α-helix content. This conformational change modifies the biochemical properties of the molecule conferring characteristics such as amyloid formation, detergent insolubility, and proteinase K resistance (1). The “prion hypothesis” postulates that PrP^Sc constitutes the infectious component, which acts as a “seed” to facilitate the conversion of PrP^c into PrP^Sc (2).

Mucosal or epithelial exposure (intestine or skin) to TSE agents has been implicated in the transmission of disease, after which accumulation of PrP^Sc is detected in lymphoid organs, in particular associated with follicular dendritic cells (FDC) in B cell follicles (3, 4). In humans, Kuru represents the classical example of this transmission (5); a similar mechanism occurs in scrapie and bovine spongiform encephalopathy. Neuroinvasion occurs later in the infection and involves the translocation of PrP^Sc via peripheral nerves and its accumulation in the CNS, after which the characteristic pathology of the disease develops.

A role for complement in prion diseases was first recognized in animal models that lacked (genetically or temporally) complement components, in which delayed onset of CNS disease was observed after i.p. inoculation of TSE agents (6, 7). Interestingly this delay was observed only after peripheral inoculation of the TSE agent, suggesting a crucial role for complement at early time points of infection. The mechanisms responsible for this delay are currently undefined. One possibility is reduced PrP^Sc retention by FDC. FDC are stromally derived cells, located within B cell follicles, which express high levels of both complement and Fc receptors, permitting them to retain (complement-containing) immune complexes on their surfaces (8). After oral inoculation of TSE agents, PrP^Sc accumulates on FDCs in Peyer’s patches (PP) and mesenteric lymph nodes (MLN), this accumulation is essential for the ensuing neuroinvasion (9). Recently the role of CD21/35 in Ag retention by FDCs has also been addressed, showing that these complement receptors mediate the early retention and accumulation of PrP^Sc on FDCs (10).

An alternative or additional mechanism underlying the delayed development of prion disease in the absence of complement components may involve conventional dendritic cells (cDCs). The cDCs are the most efficient APC and constitute a link between innate and adaptive immune responses. Widely and strategically distributed in nonlymphoid tissues, immature cDCs efficiently acquire Ags, migrate constitutively to regional lymph nodes, and present Ags to lymphocytes initiating adaptive immune responses (11). After oral inoculation of TSE agents, PrP^Sc must not only cross the intestinal epithelium but also be transported to the B cell follicles where FDCs reside. It is likely that cDCs participate in these events because they would rapidly interact with PrP^Sc after oral inoculation. The cDCs may directly capture PrP^Sc from the gut lumen or after translocation by M cells, and transport it to lymphoid organs. We have shown that cDCs can acquire and transport TSE agents from the intestinal lumen (12). In addition, in the absence of cDCs in the intestinal mucosa, the early accumulation of PrP^Sc in PP is blocked (13), highlighting the crucial role of cDCs in the early stages of the infection.

The complement component C1q is the recognition protein of the classical pathway. When C1q binds to a target surface, the serine proteases C1r and C1s, which are bound to C1q, are activated. C1s then cleaves C4 and C2, forming the C3-convertase complex, C4b2a. When C3 is activated, it binds covalently to the target surface, and is subsequently degraded to iC3b and C3dg. C3b and iC3b are major opsonins (14). C1q binds via its globular heads to a wide range of foreign or altered self-targets. These include Gram-negative bacteria, several viruses, apoptotic cells and products of tissue damage, and certain aggregates such as IgG- and IgM-containing immune complexes, and also amyloids (15, 16). Additionally C1q can act as an opsonin (17). In relation to TSE pathogenesis, the interaction between C1q and recombinant PrP is dependent on conformational modifications of PrP (18), and density of packing of PrP (19), suggesting that this interaction might be enhanced after PrP^Sc conversion. PrP^Sc can also activate the complement classical pathway (19, 20), suggesting again that the PrP^Sc and C1q interact directly, but whether this phenomenon occurs in vivo is unclear.

In this study we have assessed the role of complement in the endocytosis of PrP^Sc by cDCs. PrP^Sc uptake was evaluated after in vitro coculture of cDCs with whole brain homogenate (WBH) from terminally ill mice infected with the ME7 scrapie strain as a source of PrP^Sc. We show that in the absence of C1q, there is a significant decrease in PrP^Sc uptake by cDCs and that C1q-mediated uptake is very probably receptor-mediated. Finally, we confirmed the expression of calreticulin, a C1q receptor, on cDCs. Our results demonstrate that C1q participates in PrP^Sc uptake by cDCs, suggesting an important role in initial capture by intestinal cDCs, which will then ferry the protein to secondary lymphoid tissues where it accumulates at early time points of infection.

Specific pathogen-free, adult (6–8 wk) C57BL/6 mice were obtained from Harlan Sprague-Dawley. Scrapie ME7-infected mouse brains were provided by Dr. N. Mabbott (Roslin Institute, Edinburgh, U.K.). Scrapie RML-infected C1q^−/− mouse brains were provided by Dr. A. Aguzzi (Institute of Neuropathology, Zürich, Switzerland). The 10% WBH (10% w/v) were prepared by weighing the brains and adding 9 ml/g of sterile (endotoxin-free) PBS. C1q^−/− mouse serum was a gift from Dr. M. Botto (Imperial College, London, U.K.).

The following mAbs were purchased from BD Biosciences: purified anti-CD16/32, biotinylated anti-CD11c, FITC-conjugated CD11c, and PE-conjugated anti-CD11c, anti-CD11b, anti-CD21/CD35. Polyclonal Ab anti-calreticulin was purchased from AbD Serotec. Anti-mouse C1q mAb was purchased from Cedarlane Laboratories. Anti-PrP mAb 8H4 was purchased from Alicon, and anti-PrP mAb POM2 was provided by Dr. A. Aguzzi. The following secondary mAbs were purchased from Molecular Probes: Alexa Fluor 488-conjugated anti-mouse IgG1, Alexa Fluor 555-conjugated anti-mouse IgG1, Alexa Fluor 555-conjugated anti-goat IgG, Alexa Fluor 555-conjugated anti-rabbit IgG, Alexa Fluor 555-conjugated streptavidin, and Alexa Fluor 647-conjugated streptavidin. HRP-conjugated goat anti-mouse IgG was purchased from Jackson ImmunoResearch Laboratories. FITC-dextran was purchased from Sigma-Aldrich.

MLN and spleens were harvested in cold RPMI 1640 and single-cell suspensions were prepared after enzymatic digestion with 400 U/ml collagenase D (Roche), for 25 min at 37°C. The cDCs were enriched by negative selection using the following MACS beads (Miltenyi Biotec): CD19, CD5, and DX5. Cells were kept in MACS buffer (PBS, 0.5% BSA, 2 mM EDTA) during enrichment. Purity after enrichment was ∼75%, most contaminants being granulocytes, which were excluded during the analysis.

The cDCs (1 × 10⁶ in 1 ml of serum-free medium) were incubated with 1 mg/ml 10% (w/v) brain homogenate from terminally ill ME7 scrapie-infected C57BL/Dk mice. Normal C57BL/6 mouse brain homogenate was used as a control. Cultures were supplemented with 10% v/v untreated mouse serum, 10% heat-inactivated (56°C, 30 min) mouse serum, or 10% C1q^−/− mouse serum as stated. Where indicated, FITC-dextran (500 μg/ml) was also added to the cultures. Incubations were maintained for the indicated times at 37°C in 5% CO₂.

After incubation, cells were washed and cytospins prepared (∼1 × 10⁵ cells per cytospin). Cells were fixed in cold acetone for 10 min, and Fc receptors blocked with anti-CD16/32 (1 μg/ml) for 1 h at room temperature. Samples were treated with 3M guanidine isothiocyanate (GdnSCN/10 mM Tris-HCl (pH 7.8)) for 10 min at room temperature. Primary mAbs at optimal concentrations were incubated for 2 h at room temperature. PrP was detected using an Alexa 488-conjugated anti-PrP mouse mAb (POM2). Double staining was performed with biotin-conjugated anti-CD11c mAb, and detected with Alexa Fluor 555-conjugated streptavidin. After incubation, cells were washed with PBS/0.1% BSA and mounted using Vectashield (DakoCytomation). Images were acquired with a Zeiss microscope (Axioplan 2).

After incubation, cells were washed with cold RPMI 1640. To eliminate any free WBH, CD11c⁺ cells were enriched by MACS according to the manufacturer’s instructions (Miltenyi Biotec). After enrichment, cells were washed in FACS buffer (PBS/2% FBS/0.01% sodium azide) and processed for multicolor FACS analysis. Before staining, cells were incubated with anti-CD16/32 for 15 min at 4°C. Surface staining was performed first; primary mAbs were added for 20 min at 4°C. Intracellular staining was performed according to the manufacturer’s instructions (BD Biosciences) using an anti-PrP mAb (8H4) for 15 min at room temperature as primary Ab, and an Alexa Fluor 488- or 555-conjugated anti-mouse IgG1 mAb as the secondary Ab. Appropriate isotype controls and nonpermeabilized samples were included in every set of experiments. Cells were analyzed using a FACSCalibur cytometer and FlowJo software.

After incubation, the cells were washed in 0.1 M Tris (pH 7.4) and lysed with 2% (w/v) Sarkosyl (Sigma-Aldrich) dissolved in PBS (pH 7.4), as positive and negative controls 0.5 mg of ME7-infected WBH and uninfected WBH were used. Samples were divided into two, and one set was treated with proteinase K (100 μg/ml; Roche) for 1 h at 37°C with constant agitation. The reaction was stopped with PMSF (5 mM final concentration; Sigma-Aldrich). PrP^Sc was precipitated from untreated and proteinase K-treated samples using 4% sodium phosphotungstic acid (Sigma-Aldrich). Samples were run in a 4–12% acrylamide Invitrogen SDS-PAGE gel and then transferred to an Immobilon membrane. Blots were blocked for 1 h with 2% w/v BSA in PBS at room temperature. PrP was detected using the 8H4 mAb (0.5 μg/ml diluted in 1% w/v BSA in PBS) for 1 h at room temperature. The blot was then washed three times for 15 min with 0.2% v/v Tween 20, 0.5 mM EDTA in PBS. A HRP-conjugated goat anti-mouse IgG mAb (0.5 μg/ml diluted in 1% w/v BSA in PBS) was used as the secondary reagent. Finally, chemiluminescence was developed using the ECL reagent (Pierce).

Data are expressed as the mean ± the SD. Statistical significance comparing multiple experimental groups was analyzed using the one-way ANOVA nonparametric test. The value p < 0.001 was considered significant. All analyses were performed using Prism 4.0 software.

To evaluate uptake of PrP^Sc by cDCs, purified MLN cDCs were cultured with ME7-infected WBH, uninfected WBH or medium alone, in the presence of normal mouse serum or heat-inactivated (complement-inactivated) mouse serum. After incubation, cells were washed, cytospins prepared and PrP uptake by CD11c⁺ cells evaluated by immunofluorescence. As previously described, cDCs express endogenous PrP (21); however, a clear increase in the PrP (green) signal was observed in cDCs that had been incubated with ME7-infected WBH, in comparison with those cells incubated with uninfected WBH or the medium control. Uptake of PrP from infected-WBH was greater in the presence of normal mouse serum than with heat-inactivated serum (Fig. 1, bottom panels), whereas uptake of PrP from uninfected WBH was not affected by heat treatment of serum. Traces of free WBH remained in our preparations, as shown in the samples cultured with uninfected WBH.

To quantify PrP uptake by cDCs we used flow cytometry as a readout. We confirmed our previous results using spleen cDCs and performing intracellular staining for PrP in permeabilized and nonpermeabilized samples. Importantly, to eliminate any free WBH, CD11c⁺ cells were enriched by MACS after culture. Cells cultured in medium alone expressed PrP, but a significant increase in PrP signal was revealed by the mean fluorescence intensity (MFI) in cells cultured with either normal or infected WBH (Fig. 2,A, top panels). Again, uptake of PrP from ME7-infected WBH but not uninfected WBH decreased when heat-inactivated serum was used (percentage of CD11c⁺/PrP^high cDCs) (Fig. 2,B). To demonstrate that the increase in PrP signal represented internalized PrP, samples were labeled without permeabilization (Fig. 2,A, bottom panels), in which case no increase in PrP signal was detected. To confirm that cDCs had endocytosed proteinase K-resistant PrP^Sc, cells were lysed after incubation and samples were untreated or treated with proteinase K before immunoblotting. Fig. 2 C shows that proteinase K-resistant PrP was present in cDCs incubated with ME7-infected WBH, confirming that cDCs have endocytosed PrP^Sc (the single band present in the proteinase K-treated medium and WBH samples represents background due to the secondary Ab). Together these data show that cDCs can endocytose PrP^Sc and suggest that uptake of PrP^Sc but not PrP^c is increased in the presence of complement.

We have shown that uptake of PrP^Sc by cDCs is decreased in the presence of heat-inactivated serum as compared with normal serum. The complement component primarily denatured by heat treatment is C1q (22), although Factor B and mannose-binding lectin are also inactivated at 56°C. To directly address the role of C1q in the uptake of PrP we incubated cDCs with uninfected WBH, ME7-infected WBH or medium alone, in the presence of normal serum or serum from C1q^−/− mice and analyzed PrP uptake by flow cytometry. In the absence of C1q, uptake of PrP from ME7-infected WBH was clearly diminished (Fig. 3 A). In contrast, the uptake of PrP from normal brain was not affected by the absence of C1q.

Because the brain samples may contain small amounts of blood, some free C1q could be present in the WBH. In addition C1q mRNA is up-regulated in scrapie-infected mouse brains (23). To exclude a role for endogenous C1q during uptake, we used WBH from C1q^−/− mice infected with RML scrapie. When cDCs were incubated with RML-infected WBH in the presence of normal serum a significant increase in PrP uptake was observed (Fig. 3 B, top). Conversely, when C1q^−/− serum was used, PrP uptake was markedly decreased. When cDCs were incubated with C1q^−/− RML-infected WBH in the presence of C1q^−/− serum, PrP uptake was almost abolished, strongly suggesting that uptake of PrP in this setting is largely C1q-dependent. From these results we conclude that C1q is directly involved in PrP^Sc uptake by cDCs.

To show that complement-mediated PrP uptake is an active process we incubated cDCs with ME7-infected WBH in the presence of normal mouse serum for 4 h at 37°C or at 4°C. Cells were then permeabilized or left intact before labeling for PrP. As shown (Figs. 2 and 3), cDCs incubated with ME7-infected WBH at 37°C endocytosed PrP^Sc, indicated by the increased MFI value (medium, 206; WBH, 220; ME7-WBH, 250; p < 0.001). However, when cultures were maintained at 4°C there was a clear reduction of the PrP uptake due to a reduction of the frequency to cells that internalized it (as shown in the indicated gate, additionally a reduction in the MFI value for the whole population was observed: medium, 198; WBH, 204; ME7-WBH, 215) compared with cultures at 37°C, showing that PrP uptake by cDCs is an active mechanism (Fig. 4 A, top panels). Importantly, we observed a similar frequency of cells within the indicated gate in permeabilized and in nonpermeabilized cells when cultured at 4°C, showing that PrP was surface-bound, also suggesting a receptor-mediated mechanism.

To ascertain whether we could induce uptake of surface-bound PrP, cells were cultured with WBH at 4°C or 37°C, then washed and enriched for CD11c⁺ cells (to eliminate any free WBH) and cultured for a further 2 h at 37°C. Cells were then stained for intracellular PrP. We observed internalization of the surface-bound PrP (cells cultured first at 4°C) when the samples were subsequently cultured at 37°C (Fig. 4,B). As we reported previously, there is a statistical difference between the samples cultured with ME7-infected WBH and the control samples. If PrP^Sc uptake is receptor-mediated, surface binding should be saturable. To test this possibility, we incubated cDCs with increasing concentrations of ME7-infected WBH in the presence of normal mouse serum and assessed PrP internalization by the increase in PrP MFI. Fig. 4 C shows that uptake is saturable and plateaus at a concentration of 1 mg/ml.

To investigate the relationship of receptor-mediated PrP^Sc uptake with other internalization pathways, we used FITC-dextran as a marker for fluid phase internalization (24). Cocultures were performed as previously described, but in addition, FITC-dextran was included in the cultures. After 12 h in culture, cells were washed and CD11c^high cells analyzed by FACS for both PrP and FITC-dextran uptake. Fig. 4 D (bottom) shows that in the presence of normal mouse serum or heat-inactivated serum (and independently of the coculture condition), cDCs internalize similar amounts of FITC-dextran (cDCs cultured in medium alone were also used) evaluated by the MFI. When PrP uptake was evaluated on the same cells, the same pattern of PrP uptake was observed (increased PrP uptake in the presence of normal mouse serum, which is statistically significant). Taken together these results suggest that complement-dependent PrP^Sc uptake is receptor-mediated and that this process is distinct from fluid phase internalization.

We have shown that the increased PrP uptake by cDCs in the presence of normal serum is C1q-dependent. To determine whether PrP and C1q are endocytosed together, cytospins were prepared after coculture and cells stained for PrP and C1q. Colocalization was observed in cells that were cultured with ME7-infected WBH (Fig. 5,A), in most cases colocalization was seen in intracellular vesicles. Heat inactivation denatures C1q; accordingly no C1q signal was detected on the samples incubated in presence of heat-inactivated serum. Because WBH may contain traces of serum and as C1q expression may be up-regulated on ME7-infected WBH (23), the cultures were supplemented with mouse serum or FCS to evaluate the participation, of any mouse C1q present in the WBH. Fig. 5 B shows representative pictures of samples cultured with ME7-infected WBH. Colocalization was only observed when cells were cultured with mouse serum, strongly suggesting that the internalized C1q originates from serum added to the culture. These results suggest that PrP and C1q are being cointernalized by cDCs.

We showed that not all cDCs endocytose PrP following in vitro culture with ME7-infected WBH. To show that this result was not concentration-dependent, we incubated splenic cDCs with increasing concentrations of ME7-infected WBH in the presence of normal mouse serum. The proportion cDCs that endocytosed PrP showed a dose-dependent increase that plateaued at 1 mg/ml (Fig. 6 A). To determine whether uptake was associated with a particular cDC subset we analyzed cDCs for coexpression of CD8α and PrP. Similar proportions of CD11c⁺/CD8α⁺ and CD11c⁺/CD8α⁻ cDCs endocytosed PrP (data not shown) showing that uptake was not restricted to one of the well-defined cDCs subsets in the spleen.

Because the previous results suggested that internalization was receptor-mediated, we used FACS to screen expression of complement receptors on different populations of spleen and MLN cDCs. Fig. 6,B shows histograms of complement receptor expression on CD11c⁺ cDCs, which is itself the α-chain of the iC3b receptor, CR4. Although cell surface calreticulin is expressed on cDCs from both spleen and MLN, it is important to note that MLN cDCs have a higher proportion of calreticulin-positive cells and also a higher level of expression. In contrast, CD21/CD35 (receptor for C3b, C3dg, and C4b) was expressed only on splenic cDCs. Finally, CD11b (part of the CD11b/CD18 CR3 complex that recognizes iC3b) is highly expressed on cDCs, especially on splenic CD11c⁺CD8α⁻ cells. To characterize in more detail the phenotype of the calreticulin-positive cDCs coexpression of several markers was evaluated in splenic and MLN cDCs (Fig. 6 C). We conclude that several splenic and MLN cDCs express several candidate receptors for C1q. Nevertheless cDCs populations relevant to PrP^Sc transmission in vivo, such as MLN cDCs, express the C1q receptor calreticulin at high levels and it correlates with the expression of CD103.

Finally, to address the participation of calreticulin as a candidate receptor in the complement-mediated uptake of PrP^Sc by cDCs, cultures supplemented with normal serum were set up as previously described (WBH, ME7-infected WBH, or medium alone). After culture, cDCs were stained to evaluate PrP uptake by FACS, additionally the surface expression of calreticulin was evaluated. Fig. 7 A shows the increase in the PrP signal from cDCs cocultured with ME7-infected WBH. Interestingly ∼70% of cDCs that had taken up PrP^Sc (CD11c⁺/PrP^high, population within the indicated gate) express cell surface calreticulin. Conversely, cells cultured in presence of uninfected WBH that take up PrP do not express calreticulin.

Because the direct participation of calreticulin during the uptake of PrP^Sc will result in the internalization of both molecules, we evaluated by microscopy the colocalization of PrP, calreticulin, and C1q. Fig. 7,B shows is representative of the triple staining, where colocalization was always observed as intracellular vesicles. As we showed previously in Fig. 5, PrP and C1q clearly colocalize only in cDCs cultured with ME7-infected WBH. Importantly we show that calreticulin also colocalizes in the same vesicles. From these results we conclude that calreticulin is directly involved in the C1q-mediated uptake of PrP^Sc by cDCs.

A role for complement in prion diseases was first revealed by delayed disease development in animal models lacking complement components (6, 7). This development could represent a primary defect in accumulation/retention of PrP^Sc by FDCs, but could also be a consequence of decreased uptake and transport from the periphery by cDCs. It is known that migratory cDCs can acquire and transport intestinal PrP^Sc (12) and temporary depletion of cDCs before feeding infected WBH greatly slows down the accumulation of PrP^Sc on FDCs and the development of disease (13). How cDCs acquire PrP^Sc is not clear and the current experiments were designed to test the hypothesis that complement may be involved in the uptake.

PrP^Sc uptake by cDCs has been previously studied using bone marrow-derived cDCs (25), splenic cDCs (26), and Langerhans-like cells (27). In all cases PrP^Sc internalization has been observed after early time points (between 30 min and 2 h); however, the frequency of the cDCs that take up PrP^Sc was not addressed. An important feature of these reports is that cultures were supplemented with heat-inactivated serum, which would exclude a role for complement. We chose to use cDCs obtained from lymphoid tissues in the steady state instead of bone marrow-derived cDCs because the former are likely to be the cells that would relate most closely to in vivo events. Our results show that both MLN and splenic cDCs internalized PrP^Sc after culture in the presence of normal mouse serum (Figs. 1 and 2). However, whereas only 30–40% of the CD11c^high cells were able to internalize PrP^Sc even after 18 h of culture, all CD11c^high cDCs endocytosed FITC-Dextran after 2 h in culture, suggesting that only a subpopulation of cDCs is capable of complement-mediated PrP^Sc uptake. Additionally, that cDCs internalized similar amounts of FITC-Dextran under all culture conditions suggests that WBH was not inducing cDC activation during the culture periods (Fig. 4 D).

The cDCs cultured with uninfected WBH also showed uptake of PrP^c but importantly no difference was observed when heat-inactivated rather than intact mouse serum was used (Fig. 2). These results indicate that PrP^Sc but not PrP^c is internalized by a complement-dependent mechanism. C1q is known to interact with conformationally modified PrP^c (18), suggesting that it may only interact with PrP^Sc. Moreover, PrP^Sc can activate the classical complement pathway (19, 20), which strongly suggests that PrP^Sc will interact with the globular portion of C1q.

We used FACS analysis to evaluate PrP^Sc uptake and compared staining in cells with and without permeabilization. In the nonpermeabilized samples, only extracellular PrP is detected (i.e., naturally occurring protein and surface-bound protein). The expression of PrP on cDCs is well documented (21, 28, 29) and is up-regulated on mature cDCs (30). In our system we did not observe up-regulation of extracellular PrP over the culture period (Fig. 2 A); confirming first that the difference in PrP signals between permeabilized and nonpermeabilized samples represents the actual uptake, and secondly that the maturation stage of the cDCs is not modified during the culture. Some PrP^c may be intracellular (31, 32), which would explain why the expression of PrP in the medium control differs between permeabilized and nonpermeabilized samples.

The direct role of C1q in PrP^Sc uptake by cDCs was confirmed by the reduced uptake of PrP^Sc in cultures in the presence of C1q^−/− serum (Fig. 3,A). C1q represents the first element of the classical complement cascade, and participates in several functions besides complement activation such as removal of apoptotic cells (17, 33, 34) and immune complexes (35). Importantly to evaluate the PrP^Sc uptake in a C1q free assay we used a different strain of scrapie as a source of prions, RML. In this case a similar uptake pattern with the participation of C1q was observed, confirming that our observation is not strain-dependent (Fig. 3,B). In particular, in this setting we observed a complete abrogation of PrP^Sc uptake by cDCs in the absence of C1q. This could represent a strain-specific phenomenon because we cannot rule out the possibility that some PrP^Sc is being taken up by fluid phase endocytosis (see Figs. 2 and 3).

Much of our evidence suggests that C1q-dependent PrP^Sc uptake is receptor-mediated. The reduced internalization observed in cultures maintained at 4°C is indicative of an active process. Nevertheless, we detected an increased PrP signal in nonpermeabilized cells cultured at 4°C, showing the presence of surface-bound PrP (Fig. 4,A). By using a two-step incubation system, we confirmed that the FACS signal observed at 4°C in the nonpermeabilized samples was surface-bound PrP^Sc (Fig. 4,B). Furthermore, saturation of PrP^Sc uptake by culturing cDCs with increasing concentrations of ME7-infected WBH strongly suggests that the uptake mechanism is receptor mediated (Fig. 4 C).

We also showed by immunofluorescence that cDCs that had endocytosed PrP^Sc showed clear colocalization with C1q in the form of intracellular vesicles. Importantly colocalization of PrP and C1q was not observed when the cells were cultured with uninfected WBH, confirming C1q only interacts with PrP^Sc. As our results show that C1q plays an important role in the PrP^Sc uptake by cDCs, an important element to consider is the availability of C1q in the intestinal mucosa. In contrast with most complement components, which are synthesized by the liver, C1q is synthesized by macrophages, cDCs and, importantly, by epithelial cells in the intestine (36). Thus in vivo, free C1q from any of the mentioned sources could be available to interact with PrP^Sc.

Several potential receptors have been described for C1q. These include CR1/CD35 (also a receptor for C3b and C4b), C1qRp/CD93, calreticulin (37, 38), and recently the α₂β₁ integrin (39). CR1/CD35 binds immune complexes and transports them to the liver and spleen mediating their clearance, and can also bind C1q (40). However, although CR1/CD35 is expressed by splenic cDCs (41), it is not expressed on MLN cDCs thus excluding it as a candidate receptor in our system. C1qRp is now known to act as an adhesion receptor and does not interact with C1q (42). Calreticulin, also known as the collectin receptor, is expressed in the cytoplasm and the endoplasmic reticulum by all nucleated cells. However, when expressed on the cell surface it binds the collagenous portion of C1q and mediates its internalization in association with CD91 (34, 43).

Extracellular expression of calreticulin has been described on macrophages and cDCs, particularly immature cDCs (44). We showed surface expression of calreticulin by cDCs from spleen and MLN (Fig. 6,A). In both cases, receptor-expressing cells represent only ∼30–40% of the CD11c⁺ population, as has also been shown for human cDCs (45). Importantly in MLN cDCs calreticulin is expressed by the CD103⁺ cells, which are most likely derived from the intestine (Fig. 6 B). By evaluating the expression of CD8α and CD11b, calreticulin does not appear to be expressed by a discrete population of cDCs. Additionally, calreticulin is expressed by MHC class II^high cells, which suggests that the molecule is not exclusively expressed on immature cDCs, as it has been proposed in studies using human cDCs (44). Importantly we confirm that a high proportion of the cDCs taking up PrP^Sc expressed calreticulin, which suggests the direct participation of this receptor. Alternatively other molecules such as scavenger receptors might be involved because the molecular properties of the C1q-PrP^Sc aggregates could change, affecting the affinity and the avidity to the described receptors.

In conclusion efficient uptake of PrP^Sc by cDCs is C1q-dependent. Our results therefore suggest an alternative or complementary mechanism to explain the delay of prion disease development in the absence of C1q. Altogether our results provide new evidence that highlights the role of cDCs and complement at very early time points in prion disease pathogenesis, particularly after oral inoculation.

We thank Dr. Neil Mabbot for kindly providing scrapie ME7-infected mouse brain, Dr. Adriano Aguzzi and Dr. Mathias Heikenwalder for providing scrapie RML-infected C1q^−/− mouse brain and the POM2 Ab, Dr. Marina Botto for the generous gift of C1q^−/− serum, and Dr. Leopoldo Flores-Romo for critical reading of the manuscript.

The authors have no financial conflict of interest.