Mol Microbiol. 2003 Nov;50(3):835-43.

Internalization of OspA in rsCD14 complex and aggregated forms.

Welty DM, Snyder DS.

Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, 903 S 4th St. Hamilton, MT 59840, USA.

Abstract

Although the spirochetal protein OspA is capable of stimulating immune cells in a CD14- and TLR2-dependent manner, little is known about how TLR2 receptor complex ligands, such as OspA, are handled by the cell once delivered. We examine here the internalization of the fluorescently derivatized forms of both the full length OspA lipoprotein delivered as a recombinant soluble CD14 (rsCD14) complex and the corresponding lipohexapeptide given to the cells as an aggregate. Both forms of OspA are internalized in a similar manner to acetylated low density lipoprotein (AcLDL), a scavenger receptor ligand. Acetylated low density lipoprotein is capable of competing for internalization with OspA even when OspA is delivered as a rsCD14 complex. We observe co-localization of OspA with lysosomes but not with the Golgi complex. These phenomena are similar between RAW264.7 macrophages and endothelial cells but change drastically when the cells are deprived of serum. Upon serum starvation, OspA shows some localization to the Golgi apparatus whereas the lipohexapeptide remains on the cell surface. Inhibition of internalization of OspA via treatment with cytochalasin D or of the lipohexapeptide via serum starvation does not interfere with TNF induction activity, consistent with signalling from the cell surface.

PMID: 14617145 [PubMed - indexed for MEDLINE]

Introduction

In both its membrane bound and soluble forms, CD14 interacts with a variety of lipids and lipoconjugates and to facilitate the recognition of pathogen associated molecular pattern (PAMP) ligands by the innate immune system (Haliman et al., 1996). In doing so, it interacts with Toll-like receptor (TLR) complexes (reviewed inBeutler, 2002), which are critical in sensing such ligands and initiating the inflammatory response. Unable to signal on its own, CD14 is thought to act as a delivery molecule, handing off ligands to their appropriate receptors (Vasselon et al., 1997; 1999). With many ligands, the exact point at which the initiation of an inflammatory stimulus begins is unknown. This is significant because inhibition of steps before this event is likely to diminish inflammation in response to ligand whereas inhibition of later steps is unlikely to have such an effect and may even increase the level of inflammation by holding ligands in contact with their receptors. Likewise, any receptor antagonist acting at an internal site must be internalized in a similar manner to the ligand it blocks or it may never see the appropriate receptor.

The best characterized TLR system is the TLR4 complex which responds to bacterial lipopolysaccharide (LPS, Beutler, 2002). When LPS is delivered by soluble CD14 (sCD14), it is sent to the Golgi apparatus and internalization of ligand is coupled with signalling (Thieblemont and Wright, 1999). In contrast, aggregated LPS is known to be taken up primarily by scavenger receptor and sent to lysosomes (Hampton et al., 1991). Inhibition of LPS internalization inhibits signalling (Detmers et al., 1996, Thieblemont, 1999; Cowan et al., 2001,) even though the signalling molecule MyD88 is recruited to the cell surface by LPS stimulus (Ahmad-Nejad et al., 2002). This implies that although internalization and signalling may be coupled, such signalling events originate from the cell surface in the TLR4 complex.

Outside of the LPS arena, it is known that the TLR9 complex can co-localize with its unmethylated DNA ligand inside endocytic vesicles (Takeshita et al., 2001) and that it actually recruits MyD88 to such vesicles (Ahmad-Nejad, 2002). In a similar manner, TLR2 localizes to phagosomes in response to stimulus from whole yeast cells (Underhill et al., 1999). Although inhibition of TLR signalling does not prevent internalization (Underhill, 1999), the issue of whether prevention of internalization abrogates signalling has never been addressed. In fact, whether the TLR2 inflammatory signal originates from the cell surface or from interior compartments remains an open question. Knowledge of the point of origin of the inflammatory signal is important as it can be used to help localize the active agents in that signalling, ruling out proteins found elsewhere in the cell.

In this work, we examine binding and internalization of the TLR2 ligand, OspA. OspA is located on the outer surface of the Lyme disease pathogen, Borrelia burgdorferi (Barbour, 1985). A lipoprotein, OspA interacts with CD14 and provokes an inflammatory response in host immune cells (Sellati et al., 1999). In a similar manner to LPS, the lipidated portion of the molecule is responsible for inflammatory activity as synthetic lipohexapeptide can provoke a similar response to the complete molecule, despite its increased aggregation (Sellati et al., 1996). We determine the internalization patterns of both the CD14 delivered monomeric OspA and aggregated OspA lipohexapeptide. We further determine the effects of inhibiting internalization of OspA on its immunostimulatory activity.

Results

The first goal of this study is to determine the internalization pattern of the TLR2 ligand OspA. To accomplish this, we incubate fluorescently labelled versions of both the full length lipoprotein and bioactive lipohexapeptide with macrophages treated with probes which are known to go to specific cellular locations. Nearly complete co-localization (98% as analysed by Metamorph) is observed between vesicles containing the scavenger receptor ligand AcLDL and CD14 delivered OspA when both probes are incubated with RAW 264.7 macrophages (Fig. 1A). To ensure that OspA:CD14 was not binding directly to AcLDL, cells were also incubated first with AcLDL then washed and incubated with OspA-CD14. Similarly high levels of co-localization (90%) are detected in this case (Fig. 1B). Co-incubation of CD14-OspA with bodipy derivatized LDL showed little to no co-localization (unpubl. data).

Investigating this pathway further, we can see co-localization of OspA with the lysosomal marker lysotracker green (72% co-localization) but not with the Golgi marker 6-{ [N-(7-nitrobenz-2-oxa-1, 3-diazol-4-yl) amino]hexanoyl} sphingosine (NBD ceramide, 3.7% co-localization) (Fig. 1C and D respectively). It should be noted that co-localization with the lysotracker is much more variable both between cells and between experiments than either co-localization with either AcLDL or NBD ceramide and that while all these results are representative of at least three experiments, co-localization with lysosomes is highly variable at this time-point. This may be because of variability in the rate at which vesicle traffic proceeds within the cells. Also note that the background staining with NBD ceramide in 1D has been deliberately enhanced to demonstrate that the OspA is predominantly internalized at this point and that the brighter areas representing the golgi apparatus are perinuclear as expected.

To address the issue of whether aggregation alters the destination of internalized OspA as it does with LPS (Thieblemont, 1999), we compare the internalization of the aggregated OspA lipohexapeptide with the internalization of the monomeric CD14 delivered protein seen above. We note that the OspA lipohexapeptide has been noted to have similar immunostimulatory properties to the full length protein (Sellati et al., 1999) as is consistent with the hydrophobic portion of the molecule having most of the immunostimulatory activity. After 60 min of incubation, 89% co-localization is witnessed between lipohexapeptide and AcLDL, much lower levels are observed with lysotracker green (35%) and virtually no co-localization is noted with the golgi apparatus (0.14%, Fig. 2).

If OspA is being internalized via a scavenger receptor dependent pathway, an excess of AcLDL, should be able to compete for internalization, effectively reducing the level of OspA taken up by the cells. Figure 3 shows the results of adding varying levels of AcLDL or LDL to the culture supernatant with labelled OspA: CD14. We notice a saturable decrease in OspA uptake with increasing levels of AcLDL. Addition of LDL does not inhibit uptake to the same extent. Although there is a slight decrease in OspA uptake in the presence of 300 µg ml⁻¹ LDL, this would be consistent with minor amounts of damaged LDLs in the preparation competing for scavenger receptor. The low levels (approximately 10% of maximal fluorescence) of residual fluorescence we witness in the presence of high doses of AcLDL may be a result of OspA competing with a lower level of AcLDL towards the end of the uptake period or to low levels of OspA binding to AcLDL.

Because serum starvation has been shown to increase uptake of both LPS and HSP60 in tissue culture (Cowan et al., 2001; Vabulas et al., 2001), we have examined the effect of incubating the cells in serum free DMEM for several hours before exposure to OspA. Upon incubation of cells with the OspA: CD14 complex, co-localization to the Golgi apparatus can now be observed in some cells (Fig. 4). It is disrupted by brefeldin, A which disrupts the Golgi apparatus (unpublished data). Surprisingly, when serum starved cells are exposed to OspA lipohexapeptide, rings are formed at the cell periphery. These rings correspond with the outer extent of background staining of NBD C6-ceramide, consistent with a lack of internalization of the lipohexapeptide (Fig. 4B). Serum starvation is necessary for this alteration in internalization as incubation in a serum free, protein free medium capable of supporting cell growth (Vero medium) did not alter localization of OspA (unpublished data). Consistent with results observed for LPS, we also see a qualitative increase in uptake of OspA during serum starvation.

To see whether co-localization is consistent between cell types, we also tested the murine endothelial cell line Bend3. Western blot analysis in our lab shows this cell line to express TLR2 in both the presence and absence of serum (unpubl. data). To demonstrate that OspA has biological activity in this cell line, we examined the localization of a transcription factor known to be critical in mediating inflammation, NFκB. NFκB translocates to the nucleus in response to stimulation by both OspA and LPS (Wooten et al., 1998). After exposure for one hour, cells were fixed, then treated with the nuclear stain DRAQ5 and probed with anti-NFκB antibody then Oregon green labelled secondary. As can be seen in Fig. 5, LPS, OspA protein and OspA peptide were all active in inducing translocation of NFκB to the nucleus.

Examining the localization of OspA in Bend3, we observe a similar pattern to Raw 264.7 (Fig. 6). Specifically, a high level of co-localization is observed with the scavenger receptor ligand AcLDL. Although not all vesicles containing AcLDL have OspA, a majority of the OspA containing vesicles co-localize with AcLDL. On the other hand, negligible co-localization is witnessed in the Golgi. Upon serum starvation of the Bend3 cells, OspA protein is once again seen to migrate to the Golgi apparatus (Fig. 6C).

To examine the effects of alterations in cellular localization on immunostimulatory activity, we tested the supernatants of our RAW 264.7 cultures for TNFα via ELISA assay after induction with the probes used in this study. As both the protein and lipohexapeptide rely on the same, hydrophobic, moiety for immunostimulatory activity, any differences in activity should be a result of changes in access to or ability to bind with receptors. As shown in Fig. 7, we see a dose-dependent increase in induction with both the derivatized forms of the OspA protein and peptide. The protein has a significantly higher immunostimulatory activity than the peptide, even when tested at lower doses, a phenomenon similar to LPS and its hydrophobic derivatives ( Wright, 1999). Derivatization alters the response only minimally in either case (unpublished data). Whereas serum starvation increases TNF induction less than twofold in all tested doses of the OspA protein, the increase in TNF induction ranges from two- to eightfold with the lipohexapeptide. This is consistent with signalling before internalization in both cases.

To determine the effect of an inhibition of internalization on the activity of the OspA protein, we treated RAW264.7 macrophages with 2.5 µg ml⁻¹ cytochalasin D, a concentration noted to be capable of inhibiting both internalization and activity of LPS (Detmers, 1996). As can be shown in Fig. 8A, most OspA remains on the surface of the cell in these circumstances. However, as shown in Fig. 8B, TNF induction in these cells actually goes up, consistent with the increase in activity noted previously for the lipohexapeptide under serum starvation conditions. Addition of cytochalasin D by itself resulted in no stimulation. Addition of polymyxin B to the OspA preparation had no effect on activity.

Discussion

Both LPS and OspA are biologically active amphiphiles owing their activity to the hydrophobic portion of the molecule (Galanos et al., 1985;Sellati, 1996). Although both molecules can interact with CD14, they interact with distinct Toll-like receptor complexes (Hirschfeld et al., 1999; Beutler, 2002). Based on the experiments here, these complexes are distinct in both the way they handle ligand and in their requirement for internalization in regards to biological activity. Where LPS internalization is dependent on its aggregation state (Detmers, 1996; Thieblemont, 1999; Cowan et al., 2001), OspA seems to be independent of aggregation state, at least under normal circumstances. Where LPS activity can be blocked by inhibitors of internalization, OspA is even more immunostimulatory when retained on the cell membrane.

Both the CD14 delivered protein and the aggregated lipohexapeptide co-localize with AcLDL and to a lesser extent with lysosomes. This is consistent with uptake by scavenger receptor mediated endocytosis followed by degradation. We would argue against co-localization with AcLDL being the result of some interaction with serum lipoprotein vesicles on three counts. First, changing the incubation medium to a serum free/protein free version does not alter localization unless actual serum starvation occurs. Second, 90% co-localization is witnessed even when the cells are preincubated with AcLDL then washed before incubation with CD14-OspA. Third, co-localization with LDL is much lower than with AcLDL, indicating that lipid/lipid interactions are not responsible for the co-localization witnessed here.

Co-localization is consistent between both of the cell types tested, however, the pattern shifts once cells enter a state of serum starvation. In serum starvation, the OspA peptide stays at the cell membrane while the lipoprotein is internalized to the Golgi apparatus. Neither of these effects diminishes TNF induction activity. In fact, TNF induction actually increases in both cases and across all tested doses. If signalling occurs from an internal cellular compartment, one would expect either a shift in localization or a block in internalization to impede immunostimulatory activity. Instead, we see an increase, consistent with signalling from the cell surface.

It seems self evident that the process of signalling/internalization of ligand consists of a series of steps with the toll like receptor complexes. A necessary first step is the binding of ligand to an initial shuttle receptor. In the case of OspA, membrane bound CD14 fills this role. Importantly, the initial signal in the inflammatory cascade does not begin at this stage. Membrane bound CD14 (mCD14) does not have a transmembrane region to start the process (Vasselon, 1997). After initial binding, the receptor complex must be assembled. As TLR4 and CD14 do not start out together on the membrane (Jiang et al., 2000) it seems unlikely that other TLRs start out near CD14 especially given the fact that TLR2 molecules migrate to distinct patches on the cell surface following stimulation with inflammatory ligand (Flo et al., 2002). The inflammatory signal cascade must also be started and the ligand must be prepared for internalization. The events need not occur in this order. At least the preparation for internalization is necessary for signalling to occur in the TLR4 complex with LPS (Thieblemont, 1999). Either such preparation is not necessary for OspA or it can occur without actual internalization occurring. It is possible that inflammatory signalling occurs before preparation for internalization with OspA but after it in the case of LPS. Such a coupling of internalization to inflammatory signalling would also account for the need for TLR4 complexes to internalize ligands such as LPS (Detmers, 1996; Thieblemont, 1999; Cowan et al., 2001) but at the same time recruit signalling molecules such as MyD88 to the cell surface (Ahmad-Nejad, 2002).

It has been shown that TLR2 and TLR4 localize to distinct portions of the cell membrane in response to bacterial stimuli (Flo, 2002). The fact that TLR2 seems to function in conjunction with scavenger receptor, regardless of the aggregation state of the ligand implies that these two proteins may be in the same complex. Given the variety of the functions required for such a complex to operate, the TLR complexes may consist of a greater number of proteins than has previously been appreciated.

Experimental procedures

U251JMPNaoki cells were the generous gift of Dr Sanna Goyert, New York University School of Medicine. OspA protein was the generous gift of John Dunn, Brookhaven National Laboratories. Rhodamine Red OspA peptide was synthesized by Jan Lukszo, National Institute of Allergy and Infectious Diseases. EndX endotoxin removal medium was purchased from Associates of Cape Cod.

Cell culture

Both RAW264.7 and Bend3 cells were routinely cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% by volume fetal bovine serum, 100 U ml⁻¹ penicillin, 2 mM glutamine and 0.1 mg ml⁻¹ streptomycin.

Purification of rsCD14

rsCD14 was purified from U251JMPNaoki neuroblastoma cell supernatants. Cells were grown in Vero serum free medium. After two days in culture, supernatant was harvested, concentrated through a 30 000 molecular weight cutoff filter, equilibrated into 10 mM histidine pH 6.0 and passed through affi-gel blue and CM sepharose columns of approximately 2 ml each. Fall through was collected and loaded onto a DEAE sepharose CL-6b column (Sigma chemicals). After loading column was eluted with 10 mM histidine, 50 mM NaCl to get rid of unwanted proteins then eluted with 120 mM NaCl to elute rsCD14.

Labelling of OspA protein

OspA protein was labelled with bodipy 576/589-succinimidyl ester. Protein and bodipy were reacted for 1 h in 0.1 M NaHCO₃ at 25°C. Unreacted bodipy was then removed by spinning through a sephadex G-15 spin column pre-equilibrated with 29 mM sodium phosphate, 130 mM NaCl pH 7.4 (PBS buffer).

Coupling of OspA to rsCD14

OspA was coupled to rsCD14 by adding 20 µg ml⁻¹ labelled OspA to 70 µg ml⁻¹ rsCD14 in PBS. Reaction was allowed to proceed overnight at 37°C. Aggregates were removed by filtration through 500 000 molecular weight cutoff membranes and residual free monomer was removed by filtration through 50 000 molecular weight cutoff membranes. Analysis of the resultant material was done via native gel electrophoresis using 4–15% Bio-Rad gradient mini-gels. Visualization of red fluorescence was performed using a Molecular Dynamics 9410 variable mode imager set for 532 nm excitation and using a 580BP30 emission filter. Protein staining was done using the Pierce gelcode blue system according to the instructions of the manufacturer. Upon coupling with rsCD14, labelled OspA shifted migration to a position just above uncomplexed rsCD14 and coincident with the majority of protein staining from rsCD14. Aggregated OspA lipohexapeptide moved with an apparent size of well over 100 kDa with much not entering the gel. In contrast, virtually all rsCD14 complexed OspA both entered the gel and migrated as a single band.

Incubation of cells with probes

Before incubation with probe, cells were first incubated with either DMEM with 10% normal mouse serum, DMEM with no serum added or Vero medium for at least 3 h at 37°C. Incubations were done in Laboratory-Tek chambered coverglass wells. For incubations with AcLDL, cells were incubated with probe and 10 µg ml⁻¹ AcLDL on ice for 30 min then transferred to 37°C for 1 h. Incubations with lysotracker green were done with 50–75 nM LTG for 30 min on ice followed by 30 min at 37°C. Incubations with NBD ceramide were done in the same way as LTG with the exception that 5 µM NBD ceramide precomplexed to BSA (in the manner recommended by Molecular Probes) were used. Typical incubation volumes were either 300 µl or 500 µl, depending on well size. Samples were then washed three times and imaged.

Localization of NFκ-B

Cells were treated with E. coli J5 LPS, PBS buffer or OspA for one hour in the presence of normal mouse serum for one hour at 37°C. OspA peptide was premixed with 10 µM polymyxin B immediately before use to avoid potential contamination problems with LPS whereas OspA protein was pretreated with EndX endotoxin removal medium. Cells were then imaged by the method of Jiang et al. (2000) with the exception that cells were blocked with fetal bovine serum rather than normal goat serum. Briefly, cells were fixed in − 20°C methanol, permeabilized with 2% Nonidet P-40, blocked with 20% fetal bovine serum, incubated with 1 µg ml⁻¹ rabbit polyclonal anti-NFκB for 1 h at room temperature, washed three times and incubated with Oregon green goat anti-rabbit secondary antibody. Cells were then incubated with 5 µM DRAQ5 to stain nuclei, washed three times with PBS and imaged.

Imaging of cells

Imaging was performed on a Bio-Rad MRC 1024 laser confocal scanning imaging system connected to a Zeiss 135 inverted microscope and equipped with a Zeiss 63× Plan Apochromat 1.4NA oil immersion objective. For all images, green pixels represent intensity from 488 nm laser excitation with a 522DF35 emission filter while red pixels represent 568 nm excitation with an HQ598DF40 emission filter and blue pixels represent 647 nm excitation with a 680DF32 emission filter. Both cross-talk and background fluorescence were negligible under the conditions used. Data collection was performed using Lasersharp 2000. Images were prepared for presentation using Confocal Assistant, GIMP and/or Adobe Photoshop. Quantification was performed using the co-localization feature of Metamorph with thresholded images.

TNF assay

RAW 264.7 macrophages were incubated in DMEM either with or without (for serum starvation) 10% fetal bovine serum. Cells were then exposed to activators for 5 h and 50 µl of supernatant was diluted 1:4 with PBS (for OspA samples) or used straight (for lipohexapeptide samples) and assayed for TNFα using the Pierce/Endogen ELISA kit.

Acknowledgements

We would like to thank John Dunn, Barbara Lade, Sanna Goyert and Jan Lukszo for their support in providing us with both materials and advice and Jay Carroll for his insightful commentary regarding the manuscript.

References

Full Article Here: http://onlinelibrary.wiley.com/doi/10.1046/j.1365-2958.2003.03769.x/full