280
Eur. J. Immunol. 1997.27: 280-288
C. C. Norbury et a1
Christopher C. Norbury', Benedict J. Chambers', Alan R. Prescott', Hans-Gustaf Ljunggren' and Colin Watts'
' Department of Biochemistry, Medical Sciences Institute, University of Dundee, Dundee, Scotland Microbiology and Tumor Biology Center, Karolinska Institute, Stockholm, Sweden
Constitutive macropinocytosis allows TAPdependent major histocompatibility compex class I presentation of exogenous soluble antigen by bone marrow-derived dendritic cells Dendritic cells expanded from mouse bone marrow (BMDC) with granulocyte/ macrophage-colony-stimulating factor have potent T cell-stimulatory properties both in vitro and in vivo. This has been well documented for major histocompatibility complex (MHC) class XI-restricted responses, and more recently using peptide-loaded and protein-pulsed DC for CD8 responses following adoptive transfer in mice. An unresolved question concerns the capacity of BMDC to present exogenous antigen on MHC class I molecules, an unconventional mode of MHC class I loading for which there is now considerable evidence, particularly in macrophages. Here, we show that BMDC exhibit high levels of macropinocytosis driven by constitutive membrane ruffling activity. Up to one-third of actively ruffling and macropinocytosing BMDC transferred pinocytosed horseradish peroxidase into the cytosol following a 15-min pulse, suggesting that they might be capable of presenting exogenous soluble antigen on MHC class I molecules. We show that BMDC presented exogenous ovalbumin to a T cell hybridoma more effectively, more rapidly, and at lower exogenous antigen concentrations than BM macrophages on a cell-for-cell basis. Presentation was TAP dependent, brefeldin A sensitive, and blocked by inhibitors of proteasomal processing, demonstrating use of the classical MHC class I pathway. Although effective presentation of exogenous antigen by BMDC occurred in the absence of agents which stimulate macropinocytosis, treatment with phorbol myristate acetate (PMA) enhanced both pinocytosis and MHC class I presentation by BMDC. Finally, PMA-stimulated BMDC exposed to exogenous ovalbumin in vitro were able to prime an antigen-specific cytotoxic T lymphocyte response following adoptive transfer in vivo.
1 Introduction Accumulating evidence indicates that dendritic cells (DC) are crucial for triggering primary T cell responses. Progenitor DC arise in the bone marrow and migrate to peripheral tissues where they fulfill a sentinel function, sampling and processing environmental antigens, and expressing these on MHC molecules. In a second and distinct phase promoted by inflammatory mediators, they migrate to lymphoid organs and present previously captured antigens to T cells [l-31. Studies on DC biology have been greatly facilitated by the development of methods to propagate these cells and maintain them in vitro at a stage at least [I 163491 Received October 16, 1996, in revised form November 15, 1996; accepted November 18, 1996.
Correspondence: Colin Watts, Department of Biochemistry, Medical Sciences Institute, University of Dundee, Dundee DDI 4HN, Scotland Fax: +44-1382-201063; e-mail:
[email protected] Abbreviations: BMDC: Bone marrow-derived dendritic cells BMMQ: Bone marrow-derived macrophages TAP: Transporter associated with antigen prozessing HRP: Horseradish peroxidase ER: Endoplasmic reticulum BFA: Brefeldin A E G F Epidermal growth factor Key words: Major histocompatibility complex class / Exogenous antigen /TAP / Macropinocytosis
0014-2980/97/0101-280$10.00 + .25/0
approximating their immature state in vivo [4, 51. These cells have potent antigen-presenting capacity for CD4 T cells, attributable in part to high-capacity pinocytosis [6] as well as expression of mannose receptor family members, which may allow enhanced uptake of bacterial products [6, 71. Bone marrow-derived DC (BMDC) have recently been shown to be potent inducers of CD8+Tcell responses, e.g. following priming with BMDC pulsed in vitro with peptide [8-101 or protein [ll] antigens. Earlier studies using splenic DC had also demonstrated the capacity of these cells to stimulate CD8 responses [12-141. Since MHC class I molecules are classically found to present cytosolic antigens, de novo synthesis of pathogenencoded proteins within DC (or other professional APC) appears to be necessary to trigger effective CD8-restricted Tcell responses. However, pathogens might easily adapt to avoid infecting or replicating within these cell types, perhaps making it desirable, in the case of professional APC, to lift the normal embargo on presentation of exogenous antigens on MHC class I molecules. In fact, several studies have demonstrated in vivo cross-priming of host MHC class I with antigens not made biosynthetically within host APC [15-171. However, a clear consensus for the mechanistic basis of this unusual mode of presentation on MHC class I molecules has not emerged, and it is not clear which cell types might mediate this pathway. In macrophages, two fundamentally different pathways for the presentation of exogenous antigens in vitro have been described (reviewed in [18]), one involving unconventional postGolgi loading of MHC class I [19-211, and another involv0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim, 1997
Eur. J. Immunol. 1997.27: 280-288
TAP-dependent presentation of exogenous antigen by DC
ing unusual access of exogenous antigen to the classical loading pathway [22-251. Generally, these two pathways are, respectively, insensitive or sensitive to disruption of the secretory pathway with brefeldin A (BFA). Exogenous antigen presented by the BFA-sensitive pathway also utilized the proteasomal processing system and the transporters associated with antigen processing (TAP) [24]. The BFA-insensitive pathway presumably uses a post-Golgi population of class I molecules, endosomal, or phagosoma1 processing followed by endosomal loading [19, 201 or regurgitation [21], followed by capture on cell surface class I molecules.
2.2 Cell culture
Given their potency in priming CTL responses, it is tempting to speculate that D C might also engage in presentation of exogenous antigen on MHC class I and may be important in the cross-priming phenomenon. Paglia et al. [ l l ] recently showed that BMDC exposed to intact antigen in vitro were able to prime a CD8 T cell response in vivo. However, direct evidence that D C can present exogenous antigen on class I MHC molecules is lacking and it is not clear which of the two pathways outlined above might occur within DC. Recently, we showed that bone marrowderived macrophages (BMMQ) were able to take up and present a soluble antigen on MHC class I molecules in vitro [22]. Presentation correlated strictly with the level of ruffling and macropinocytosis during exposure to exogenous antigen and morphological evidence demonstrated transfer of exogenous material into the cytosol where it was processed and presented via the classical class I pathway [22].
2.3 Generation of BMDC
Here, we extend and strengthen these conclusions by showing that BMDC can also process and present exogenous soluble protein antigen. However, whereas presentation by BMMQ required stimulation with phorbol esters or growth factors, BMDC present exogenous antigen constitutively on their MHC class I molecules following antigen capture via constitutive macropinocytosis. Delivery of an exogenous tracer to the cytosol could be demonstrated in approximately 30% of cells. Using D C expanded from TAP -/- bone marrow, we show that presentation of exogenous antigen via this pathway is TAP dependent. BMDC presented exogenous antigen on MHC class I considerably more efficiently than BMMQ from parallel cultures, and following brief exposure to exogenous antigen in vitro, could prime CTL responses in vivo.
281
B3Z cells [26, 271 were maintained in RPMI, 10% FCS, 100 U/ml kanamycin, 2 mM glutamine, sodium pyruvate, 1 mM nonessential amino acids (MEM, 1 mM), 2mercaptoethanol, 400 U/ml hygromycin B and 1 mg/ml geneticin (G418). X63 cells transfected with the mouse granulocyte/macrophage-colony-stimulatingfactor (GMCSF) gene (X63 Ag8 no. 12; a generous gift from D. Gray, RPMS, London) were cultured and supernatant harvested for use in generation of BMDC.
D C preparations were made by an adaptation of the protocol described by Inaba et al. [4]. In brief, femurs from C57BL/6 mice (5-15 weeks old) were flushed through with 10 ml Dulbecco's Modified Eagle is Medium (DMEM), 10 % FCS, 100 U/ml kanamycin and 2 mM glutamine. The exudate was centrifuged briefly at a low speed (300 rpm) to remove aggregates, transferred to a fresh tube and washed twice. The cells were then harvested and resuspended in DC medium (DCM: DMEM, 20% FCS, 2% X63-GM-CSF supernatant, glutamine and kanamycin. The cells were plated out at a density of 5 x lo5 ml in 24-well plates and cultured for 3 days at 37 "C in 10 YO Cod90 YO air. Following the initial incubation, the plates were swirled vigorously to dislodge nonadherent cells, the supernatant was aspirated off and replaced with fresh DCM. After 2 days the nonadherent cells were removed by swirling, and after a total of 7 days ex vivo, the DC, visible as clumps of loosely adherent cells, were harvested by vigorous pipetting. Following overnight culture in tissue culture-treated dishes, nonadherent cells were harvested for use in fluid-phase uptake and antigen-presentation assays. A large proportion of the cells exhibited typical dendritic morphology and, by FACS analysis, 65-90 YO of the cells were N418 (CDllc)+, MY114 (MHC class II)+, and NLDC-145 (DEC-205)'. DC were also generated by the same method from TAP1 knockout mice bred onto a pure C57BL/6 background [28]. BMMQ cultures were generated as described [22, 291. 2.4 Microscopy
Uptake of the fluid-phase marker horseradish peroxidase (HRP) was detected by microscopy. Exposure to these solute markers was performed as described [30, 311. Prior 2 Materials and methods to use, the HRP solutions used were passed through a 0.22-pm filter. For phase-contrast microscopy DC were 2.1 Media and chemicals plated out in DMEM in poly-L-lysine-coated Nunc slide Cell culture media, fetal calf serum and G418 were from flasks 24 h prior to the experiment. DC were incubated GIBCO BRL (Paisley, Scotland). Lab-Tek tissue culture with HRP for the times indicated, washed five times at dishes (100 mm) and slide flasks were from Nunclon 4"C, then fixed in 0.5 % glutaraldehyde. HRP activity was (GIBCO BRL) and all other tissue culture plastics were developed with 0.5 mg/ml diaminobenzidine (DAB) and from Costar (Cambridge, MA). BFA and hygromycin B 0.05% hydrogen peroxide in PBS. A cell was scored as were purchased from Boehringer Mannheim (Mannheim, showing cytoplasmic HRP staining if it excluded trypan Germany). blue and if the DAB reaction product was excluded from 5-Bromo-4-chloro-3-indolyl-~-~-galactoside (X-Gal) and chlorophenol red P-galactoside (CPRG) were the nuclear region. from Calbiochem (Beeston, GB). Glass coverslips and slides were from BDH Chemicals Ltd. (Poole, GB). All For phase-contrast time-lapse video microscopy, cells were other reagents were from Sigma Immunochemicals (Poole, grown in Nunclon slide flasks in DMEM with or without serum. The flasks were mounted in a heated chamber GB) .
282
Eur. J. Immunol. 1997.27: 280-288
C. C. Norbury et al.
maintained at 37"C, in 5 % C 0 2 in air, fitted to a Zeiss Axiovert microscope, and a phase-contrast image was collected from these cells at regular intervals between 20 s and 1 min. These images were collected on a Coolview CCD camera (Photonic Science) and using a Macintosh 950 Quadra computer. Images were recalled sequentially and replayed as a movie sequence using Ionvision software (Improvision). Individual images from representative parts of the movie sequence were then collected together as a montage using Adobe Photoshop software and printed on a Bromide printer.
2.5 Measurement of fluid-phase pinocytosis Fluid-phase pinocytosis was quantitated by measuring the uptake of HRP as described [6,30]. DC were resuspended at 1 x 106/mlin HBSS/2% FCS and an equal volume of prewarmed HRP (2 mg/ml in HBSSFCS) added. Following incubation at 37°C for the times indicated, a tenfold excess volume of cold HBSSFCS was added, and the cells washed five times at 4°C. After washing, the cells were lysed in 100 ml 0.2 YOTriton-X 100/PBS and HRP activity assayed using o-dianisidine and H 2 0 2as a substrate [30]. Absolute amounts of HRP taken up were calculated using an HRP standard curve and subsequently converted to volumes. 2.6 Antigen uptake and presentation Since some ovalbumin (OVA) preparations bound to cell surface H-2Khmolecules in the absence of cellular processing and mediated T cell activation, each preparation of OVA (Lorne Laboratories, Twyford, GB) was tested for background binding before use in antigen presentation assays as outlined [22]. Any OVA preparation that triggered the activation of more than 0.1 YOof the B3Z cells was discarded. For presentation assays, cells were pre-incubated in suspension at 37 "C before exposure to OVA in HBSSFCS at the concentration and for the time periods specified. After exposure to OVA, the cells were washed thoroughly three times before plating out in a 24-well plate. After a 2-h chase, the cells were overlaid with an equal number of B3Z and incubated overnight. Following this period, B3Z were assayed for 6-Gal activity with CPRG or X-Gal as described [32]. Activation of single B3Z cells was quantitated by counting the number of blue (LacZ' cells) in a Neubauer counting chamber; the results are expressed as the percentage of the total number of cells present. Alternatively, a colorimetric assayed using CPRG as a substrate was used to detect LacZ activity in B3Z lysates. For analysis of the time course of the presentation of SIINFEKL-Kh complexes, DC were fixed at various time points with 1YO paraformaldehyde as described [22]. Gelonin (purified from Gelonium multiflorum; Sigma, cat. no. G-0528) was used at a final concentration of 1 mM after dilution from a stock solution of 20 mM in DMSO. BMMQ were incubated with gelonin for 20 min prior to OVA pulsing and during the OVA pulse. BFA was prepared at 5 mg/ml in DMSO and was used at 5 pg/ml. Since BFA has a short lifetime in culture, the medium containing
it was replaced hourly. In all cases, control cultures were incubated with carrier at a concentration equivalent to that in the inhibitor preparation.
3 Results 3.1 Mouse BMDC ruffle and macropinocytose constitutively DC were expanded from the bone marrow of C57BU6 mice in the presence of GN- C SF essentially as described by Inaba et al. [4] and outlined in Sect. 2.3. The cells showed striking spiked projections and veils as described [4] and were highly motile in suspension culture or when semi-adherent. Time-lapse phase-contrast video microscopy revealed that continual remodelling of the cell surface occurred even in the absence of exogenous growth factors, phorbol esters, or serum; and that this led to the constitutive formation of phase-bright vacuoles very similar to macropinosomes described in other cell types. An 8-min sequence is shown in Fig. 1 which illustrates typical behavior of these cells. Three episodes of ruffling activity and macropinosome formation can be observed in the central cell. In the first (0-160 s), characteristic circular ruffles coalesce to form two macropinosomes, and this is followed by ruffling (arrowed) and macropinosome formation at a second site. Intriguingly, the same cell is also moving, indicating that motility and ruffling activity can occur simultaneously. Note that the cell to the right also ruffles and generates a macropinosome during this time period (240-300 s). Having already generated two or more phasebright macropinosomes at different sites, the central advancing cell appears to contact a second cell, which then stimulates renewed ruffling activity and the formation of several macropinosomes (300-480 s, Fig. 1). Multiple macropinosomes were often observed to form at a single site of local ruffling (not shown). Incubation with exogenous tracers such as HRP provided unequivocal evidence of extensive pinocytosis, as judged by the appearance of large spherical HRP-labeled structures in the cytosol (Fig. 2a). Within 2 min, 90 YOof pulsed cells contained some HRP-labeled pinosomes (Fig. 2b). At this time point, few cells (less than YO) showed evidence of cytosolic HRP. This changed dramatically, however, following a 15-min pulse, when a significant proportion of cells (up to 30%) showed evidence of cytosolic HRP (Fig. 2a and b), an event shown to take considerably longer (1-2 h) in A431 cells [22]. We compared the pinocytic capacity of BMDC with other mouse lymphoid cell lines, with BMMQ, and with the human epidermal carcinoma cell line A431. Macropinocytosis has been well characterized in A431 cells, but is only observed following stimulation with epidermal growth factor (EGF) [30]. BMDC showed rates of accumulation of HRP some 10-20-fold greater than RMA and A20, and 3-fold greater than the mastocytoma line P815 (data not shown). Strikingly, the constitutive rate of uptake (- 2.0 n1/106 cells per min, Fig. 3a, b) was very similar to the rate of pinocytosis in A431 cells stimulated with EGF [30, 311 or BMMQ stimulated with macrophage (M)-CSF [29, 331. In human DC, the mannose receptor has been shown to play a significant role in accumulation of some
Eur. J. Immunol. 1997.27: 280-288
TAP-dependent presentation of exogenous antigen by DC
283
Figure 1. Constitutive membrane ruffling and macropinocytosis in BMDC. Cells in slide flasks were viewed at 37°C as described in Sect. 2.4. In (a), the cell in the center of each panel shows three different episodes of ruffling and macropinosome formation over an 8-min period. In addition, it is moving from top right to bottom left over the same period. Ruffling in a characteristic circular morphology and the formation of two macropinosomes (0-160 s) is then followed by ruffling and macropinosome formation at a second site (arrowed; 240-300 s). The cell appears to contact a second cell and engages in a third episode of ruffling and formation of several large macropinosomes (300-480 s). Once formed, macropinosomes move in towards the nucleus.
B3Z, which recognizes the SIINFEKL peptide in the context of H-2Kb [27,32]. Tcell activation was assessed either by counting Lac Z + cells or by measuring 0-galactosidase activity in cell lysates. As shown in Fig. 4a, BMDC presented the SIINFEKL epitope to the B3Z T cell hybridoma following a 10-min exposure to soluble OVA. Triggering of B3ZT cells increased in proportion to the exogenous OVA concentration and was detectable at concentrations as low as 125 pglml. Importantly, although presentation at all exogenous antigen concentrations was boosted by the inclusion of 30 nglml PMA to enhance membrane ruffling and pinocytosis, the constitutive pinocytic activity of BMDC was sufficient to capture antigen for MHC class I presentation (Fig. 4a). When macrophages expanded in parallel cultures in M-CSF [29] were compared with BMDC, it was evident that BMDC were more potent on a per-cell basis (Fig. 4b). Indeed, in the typical experiment shown, BMDC in the absence of PMA induced a stronger T cell response than BMMQ exposed to PMA. When the 3.2 BMDC constitutively present exogenous antigen on capacity of each cell type to present exogenous peptide to class I molecules the B3Z hybridoma was compared on a cell-for-cell basis, As shown in Fig. 2, BMDC showed clear evidence that macrophages and BMDC were equally potent (data not exogenous HRP gained access to the cytosol, similar to shown). This indicated that the enhanced uptake and proour previous description of this phenomenon in A431 cells cessing capacity of BMDC not differential expression of and BMMQ. Unlike A431 and BMM, however, this was co-stimulatory molecules, accounted for their increased observed in BMDC without growth factor or phorbol ester efficiency of presentation. Given that pinocytosed tracers stimulation. We, therefore, tested the capacity of BMDC could be readily detected in BMDC after less than 1 min to present an exogenous soluble antigen constitutively on of exposure (Fig. 2), we assessed the time of exposure to MHC class I molecules. BMDC were harvested from clus- exogenous antigen necessary to produce a T cell response. ters expanded in GM-CSF and exposed to soluble OVA for Cells were exposed to 5 mglml OVA for different times, different periods of time. The cells were then washed, washed, and co-cultured with B3Z T cells. As shown in chased at 37 "C for different times, fixed with aldehyde and Fig. 4c, exposure times as short as 1 min were sufficient to then co-cultured with the Lac Z-inducible T cell hybridoma capture sufficient exogenous OVA for subsequent presen-
exogenous tracers [6, 71. Incubation of BMDC with mannan decreased the initial rate of HRP uptake to some extent, but had a more significant effect on accumulation of HRP within the D C (Fig. 3a). This is consistent with more effective retention within D C of solutes taken up by the mouse mannose receptor compared with those taken up by fluid-phase pinocytosis. The rate of pinocytosis was increased four- to-fivefold, up to a maximum initial rate of 10 nI/1O6 cells per min when cells were pre-incubated with 30 nglml PMA (Fig. 3b). Thus, BMDC, like human blood-derived D C [6], show unusually high levels of constitutive pinocytosis as a consequence of extensive ruffling activity. This produces large macropinosomes which are particularly efficient vehicles for endocytosis because of their favorable volumehrface ratio compared with smaller (0.1 pm) clathrin-coated vesicles.
-
Eur. J. Immunol. 1997.27: 280-288
C. C. Norbury et al
284
a
50
a
n
_I
0
b
0
30
20
10
2007
1 0 .
Time of exposure (min)
20
'$3
40
Time of exposure (min)
Figure 3. Quantitation of BMDC pinocytic activity. (a) BMDC were incubated with 1 mg/ml HRP for different time periods in of 1 mg/ml yeast mannan, and in the presence (W) o r absence (0) (b), in the presence (0)or absence ( 0 )of 30 ng/ml PMA. At the end of each incubation period, the cells were washed extensively and HRP activity measured in cell lysates and converted to apparent volumes using an H R P standard curve. Readings shown are means of a minimum of three independent experimental points from representative experiments.
b 100
75
Y 8
50 0.5
a? 25
0.4
0
m
0
5
10
15
0.3
2 0.2
Time of exposure 0.1
Figure 2. Constitutive macropinocytosis in BMDC can load exogenous tracers into the cytosol. (a) Phase-contrast image of BMDC exposed for 15 min to HRP. All cells show peroxidase-loaded macropinosomes and several show evidence of cytosolic HRP revealed by DAB cytochemistry. The central well-spread cell shows cytosolic, but not nuclear HRP staining as well as several intact macropinosomes. In (b), the proportion of cells exhibiting labeled macropinosomes (0)or cytosolic loading (W) after different times of pulsing with HRP is shown. For all points, 500 cells were counted, and cells were only scored as cytosolically labeled if the nucleus remained free of HRP.
tation of the SIINFEKL epitope. To analyze the kinetics of presentation, BMDC were pulsed with antigen, washed, and then fixed at different times prior to co-culture with B3ZTcells. As shown in Fig. 4d, some presentation could be detected 1 h after initial exposure to exogenous antigen. In contrast, presentation by BMMQ was only detected 6 h after an initial antigen pulse [22]. Taken together, these observations indicate that BM DC cultured in vitro in the presence of GM-CSF present exogenous antigen on their MHC class I molecules more efficiently than do BMMQ cultured in parallel.
3.3 BMDC present exogenous antigen following access to the classic MHC class I pathway To distinguish between two possible mechanisms for presentation of exogenous OVA by BMDC, that is, unconven-
0
G 25
75
5
10
Ova (mg/ml)
8
8 6
s
ir, Ova (mg/ml)
0.25
;
0.05 o
;
;
r
1
d
0.05 0
0 0
2.5
5
7.5
10
Time of exposure (min)
0
50
100
150
200
250
Time of chase (min)
Figure 4. BMDC constitutively present exogenous soluble antigen on MHC class I molecules. BMDC or BMMQ, ( 5 x 10') were exposed to OVA in the presence (m; +) or absence ( 0 ; 0) of 30 ng/ml PMA. The cells were washed and recultured for 2 h before overlaying with 5 X lo5B 3 Z T hybridoma cells. After 16 h, the cultures were either lysed and P-Gal activity measured using the CPRG substrate (absorbance max 595 nm) or fixed, incubated with X-gal, and the number of blue cells scored. Additional variables were either the OVA concentration (a), BMDC (W, 0 )versus BMMQ, (+, 0)(b), time of exposure to antigen (c), or chase period before fixation (d). Background absorbance values obtained in the absence of OVA (ranging from 0.05 to 0.2 AU, or 0.2 % Lac Z' cells, were subtracted). Readings shown are means of at least three independent experimental points.
TAP-dependent presentation of exogenous antigen by DC
Eur. J. Immunol. 1997.27: 280-288
0.4 7
b
a
-,
T
-
0.4 -
0.3
0
3
0.2
a
9
1
2
3
4
Ova conc. (mglml)
T
0.2-
-
0
T
-
0
0.5-
0.25
0.1
C
0.5
11
0.75
2 !
285
0.1 -
5
Fixation (at 6 h )
+
BrefeldinA
-
-
+
+
0
,
Ova LLnl LLM
-
+
+ 4
Treatment
Treatment
Figure 5. Presentation of exogenous antigen utilizes the cytosolic processing machinery and early secretory pathway. (a) BMDC were incubated with 5 mg/ml OVA in the presence (m) or absence (0)of 1 mM gelonin for 10 min, then washed and incubated for 6 h before overlaying with 8 3 2 cells. (b & c) BMDC were incubated with 5 mg/ml OVA in the presence or absence of 5 @ml BFA (b), 50 mM N-acetyl-L-Leu-L-Leu-L-norleucinal o r N-acetyl-L-Leu-L-Leu-L-methional (c), as shown. BFA, N-acetyl-L-Leu-L-Leu-L-norleucinal, or N-acetyl-L-Leu-L-Leu-L-methional were added immediately after the antigen pulse and were present during the chase period up to 6 h. In (b), DC were fixed where indicated with 1% paraformaldehyde prior to co-culture with B3Z cells. Readings shown are the mean of three independent experimental points, with SEM shown as error bars.
tional access to the conventional pathway versus uncon- conditions, we tested whether or not a functional TAP ventional post-Golgi loading of MHC class I , we per- transporter was required. D C and macrophages were formed studies with inhibitors of protein synthesis, protea- expanded from the bone marrow of mice carrying homosoma1 processing and membrane traffic. Gelonin, a plant zygous deletions for the TAP1 subunit [28]. Fibroblasts toxin which inhibits protein synthesis which is unable to loaded with antigen by osmotic lysis of pinosomes [28], gain access to the cytosol [34] under normal conditions, and BMMQ loaded with antigen by electroporation (CCN substantially inhibited presentation, particularly when the and BJC, unpublished observations) from these mice are exogenous OVA concentration was limiting (Fig. 5a). unable to present antigens on MHC class I molecules. DC However, the overall inhibitory effect of gelonin, and by expanded with GM-CSF from TAP -/- and TAP+/+ mice implication the requirement for de novo class I biosynthe- were pulsed with different concentrations of exogenous sis, appeared less complete than we previously observed OVA and presentation to the B3Z T cell was measured as with BMMQ [22]. However, the effects of inhibitors of the before. In parallel, the level of pinocytosis in DC from proteasome and of pre-Golgi secretory pathway traffic both genetic backgrounds was also measured using HRP were fully consistent with processing of the SIINFEKL [30]. As shown in Fig. 6, DC from the TAP1-/- mice epitope in the cytosol and loading onto a pre-Golgi were equally active in fluid-phase pinocytosis compared population of MHC class I molecules. Presentation of exo- with TAP+/+ cells, but were completely unable to present genous OVA was blocked in the presence of the peptide exogenous OVA at any of the concentrations tested, aldehyde N-acetyl-L-leucinyl-L-leucinyl-L-norleucinal, which blocks the proteasomal activity required for OVA b processing, but not in the presence of the methional anaa 0.5. log N-acetyl-L-leu-L-leu-L-methional (Fig. 5c) which does not affect the proteasome [24]. Presentation to B3Z T cells was eliminated when BFA was included during the chase period following exposure to exogenous OVA (Fig. 5b), but recovered fully following removal of BFA, demonstrating the presence of either a persistent pool of cytosolic OVA or of loaded MHC class I molecules trapped, presumably, in the endoplasmic reticulum (ER). _.. 0.1
3.4 Presentation of exogenous antigen by BMDC is TAP dependent
The inhibitor studies presented above still left open the possibility that proteasomal degradation of OVA might provide a source of antigen for loading onto a population of MHC class I other than that in the ER. Conceivably, export of these molecules to the cell surface might be BFA-sensitive in DC, where to our knowledge the effects of this drug are not well characterized. To establish unequivocally the pathway of MHC class I loading under these
1
Ova conc. (mglml)
10
Time 01 exposure (min)
Figure 6. Presentation of exogenous antigen is TAP dependent in orTAP1.’- mice BMDC. D C expanded from B6TAP”’ mice (0) (W) were incubated with either 5 mg/ml OVA for 10 min (a) or 1 mg/ml HRP for the times indicated (b). Pinocytosis and antigen presentation were measured as described in Sect. 2.5 and 2.6. Cells exposed to H R P at 0°C was considered background loading. Triggering of B3Z Tcells byTAPl-’- D C never increased above the background in the absence of OVA which is not subtracted in this experiment. Readings shown are means of at least three independent experimental points f SEM.
C. C. Norbury et al.
286
Eur. J. Irnmunol. 1907.27: 280-288
demonstrating that a functional TAP transporter is necessary. The TAP - - - DC were equally effective at presenting the SIINFEKL peptide when this was added exogeneously, demonstrating the presence of Kh class I molecules in these cells (not shown). Essentially parallel results were obtained using macrophages expanded from TAP 1 -/bone marrow, i.e. TAP-/- macrophages could not present exogenous OVA even when phorbol ester-stimulated (not shown). Taken together, these studies demonstrate in vitro that BMDC can take up soluble exogenous antigen and deliver it to the cytosol, where it utilizes the proteasome/ TAP pathway of MHC class I loading and presentation.
3.5 BMDC pulsed with exogenous antigen in vitro can prime CTL responses in vivo Previous studies with DC or macrophages pulsed in vitro with exogenous particulate or soluble antigen showed that CD8' T cell responses were induced following immunization with pulsed cells [ l l , 35, 361. However, this has not been demonstrated for a soluble exogenous antigen known to be presented via the conventional class I pathway. To determine whether antigen pulsing in vitro under the conditions described above was sufficient to trigger de novo T cell responses in vivo, we immunized mice either with BMDC pulsed with soluble OVA in the presence of PMA, BMDC pulsed with SIINFEKL peptide, or BMDC treated with PMA alone. Prior to injection, the pulsed cells were extensively washed and cultured for 2 h. After 9 days, the spleens were removed and after one round of restimulation in vitro, the presence of OVA-specific cytotoxic effector cells was assayed using targets pulsed with the SIINFEKL peptide. As shown in Fig. 7, OVA-specific effector CTL were induced following introduction of PMAstimulated DC pulsed in vitro with soluble OVA, but not when unpulsed DC were introduced. D C stimulated with a 8o 1
1
1OO:l 201
4:l
1:l
Effector:Target Ratio
b
1001 201
41
1:l
Effector:Target Ratio
Figure 7. BMDC pulsed with OVA in vitro can prime CTL responses in vivo. BMDC incubated for 1 h in the presence of 5 mg/ml OVA ( O ) , 10 nM SIINFEKL peptide (0). or neither (A), and 30 ng/ml PMA, and then washed. Under these conditions, presentation to B3Z was still completely TAP-depcndent in v i m . After a 2-h chase, 5 X lo5 cells were used to immunize B6 mice intravenously. After 9 days, spleens were removed and 2.5 x lo7 responder cells restimulated with 1.2 x lo7 syngeneic stimulator cells and 0.5 mM SIINFEKL peptide for 5 days. Cytotoxicity was measured in a conventional 4-h chromium release assay with RMA-S cells in the presence (a) or absence (b) of SIINFEKL peptide. Equivalent activity could be measured using EL4 and OVA-transfcctcd EL4 (EG7.1) as targets.
PMA in the presence of OVA often gave levels of CTL induction at least comparable to that obtained with peptide-pulsed DC. The CTL induced by OVA-pulsed DC were also able to kill EL4 transfected with OVA (EG7.1) when used as targets, but not untransfected EL4 cells. Thus, a soluble antigen captured via the macropinocytic pathway in vitro is able to induce primary CTL responses in vivo.
4 Discussion Many studies suggest that synthesis of antigen de novo within host AP C is not always an obligatory requirement for the induction of CD8' CTL effector cells in vivo [16-18, 371. A variety of cellular, particulate [35, 36, 381, and even soluble antigen [39-411 preparations have proven effective at inducing primary cytotoxic T cell responses. However, neither the mechanism of loading of exogenous antigens onto MHC class I molecules in vzvo nor the antigen-presenting cell type(s) involved in triggering these CD8' responses have been resolved. Given their potency in triggering both CD4 and CD8 responses, it is reasonable to ask whether DC could be involved in presenting exogenous antigen on MHC class I molecules. Previous studies have demonstrated in vitro that although DC isolated from spleen are potent stimulators of male-specific, [42] allospecific [43], and viral-specific [12] MHC class I responses, in each case, antigen was synthesized de novo within splenic DC. Although this requirement could be bypassed by pulsing isolated splenic DC with peptide in vitro [13], these cells were not able to stimulate expansion of CTL precursors following exposure to noninfectious UV-inactivated or brornelain-treated virus [ 141. This indicates that this exogenous antigen, at least, could not be presented on MHC class I by splenic DC. Nonetheless, other evidence indicates the presence of a population of cells in freshly isolated spleen capable of presenting some exogenous antigen on MHC class I molecules [23, 441 and splenic DC were able to prime CTL responses in vitro or in vivo when incubated with antigen encapsulated into pH-sensitive liposomes [45, 461. An important development has been the demonstration that substantial numbers of DC cells can be expanded from bone marrow precursors or differentiated from blood in the presence of GM-CSF [4, 51. These cells are likely to represent an earlier, less-mature stage of DC development and appear to have a considerably enhanced capacity to capture and process antigen compared with their splenic counterparts. Like splenic DC, they are potent inducers of CD4 T cell responses [ 1, 51 and recent reports have demonstrated that, like splenic DC, BMDC pulsed with peptide in vitro can induce potent CTL responses in vivo capable of protecting against tumors expressing those antigens (8, lo]. Importantly, one report has now shown that both BMDC and an immortalized DC line were able to prime CTL responses in vivo following exposure in vitro to soluble intact protein [ l l ] , implying that BMDC can load exogenous antigen onto MHC class I molecules. Here, we demonstrate directly that BMDC have this capacity, at least in vitro. In addition, we have shown that of the two pathways that have been described in macrophages for presentation of exogenous antigens, BMDC presented exogenous OVA exclusively via the classical MHC class I pathway.
Eur. J . Immunol. 1997.27: 280-288
We observed sevcral features of BMDC which may explain enhanced presentation of exogenous antigens on MHC class I when compared to splenic DC. First, DC cultured in vitro are shown to exhibit pronounced surface membrane ruffling activity. These ruffles frequently resolve to form large phase-bright vacuoles o r macropinosomes, which then migrate from the cell periphery towards the nucleus. This is reminiscent of macropinosome biogenesis and movement observed following M-CSF stimulation of BMMQ [29, 331. BMDC exhibited constitutive ruffling and macropinocytosis as previously described for human blood-derived D C [6]. This may therefore be a general property of GM-CSF-expanded or differentiated cells of the DC lineage. Macropinocytosis and presentation of exogenous antigen on MHC class I molecules could be further increased by exposure of bone BMDC to 30 ng/ml PMA and therefore may also be sensitive to up-regulation by more physiological agents in vivo. In culture, the ruffling/macropincytotic response and cell movement occurred simultaneously, suggesting that DC migrating through peripheral tissues may continuously sample the external milieu. Sallusto et al. [6] recently described remarkably high rates of pinocytosis in human blood-derived DC, corresponding in 1 h to approximately 1000 pm3, or 1 pl per cell. Our measurements using HRP in mouse BMDC were between 60-150 nl/lOh cells per h, i.e. approximately 6-16-fold lower than in the human cells. Stimulation with PMA increased pinocytosis in mouse DC to within approximately a factor of 2 of that measured in human DC. Absolute comparisons need to be made with caution, however, since a proportion of HRP accumulation in mouse DC was inhibited with yeast mannan, indicating, as in human DC, that this tracer is taken up by both receptor-mediated and fluid-phase pinocytosis. Interestingly, the effect of mannan on HRP accumulation was most pronounced after longer incubation times, suggesting that HRP taken up via a putative mannose receptor was better retained within D C than HRP taken up by true fluid-phase pinocytosis. Further analysis of the endocytic capacity of mouse DC is necessary, but it is clear that these cells show remarkably high rates of constitutive pinocytosis some ten-fold higher than most other cell types tested, and equal to those seen in EGF-stimulated human A431 cells [30]. The second feature of mouse BMDC relevant to their capacity to present exogenous antigens on MHC class I was the appearance of pinocytosed HRP in the cytosol of a significant proportion of cells following brief periods of exposure to exogenous tracer. Compared with HRP translocation into the cytosol of A431 cells, where 7.5 % of cells showed evidence of cytosolic loading 1 h after HRP pulsing [22], the number of D C involved was greater (30% and the kinetics of appearance of HRP in the cytosol was faster (15 min compared to 1-2 h) (Fig. 2b). Although it is still unclear how some macropinosomes transfer their content into the cytosol, this has now been observed in A431 cells, in mouse BMMQ [22], and now in DC. The possibility that the appearance of HRP in the cytosol is artifactually induced, for example by aldehyde fixation, is unlikely since cytosolic staining could be visualized in a proportion of living BMDC using FITC-dextran as an exogenous marker ([22] and data not shown). Moreover, recent studies demonstrate that when A431 cells are stimulated with
TAP-dependent presentation of cxogcnous antigen by DC
287
EGF in the presence of HRP coupled to a nuclear localization sequence, a clear nuclear staining pattern is observed following aldehyde fixation and processing for HRP cytochemistry, demonstrating protein transfer into the cytosol and nuclear targeting in living cells (data not shown). Based on studies in vitro, two pathways have been proposed for the presentation of exogenous antigen on MHC class I molecules. These pathways (reviewed by Rock [ 181) essentially differ with respect to the population of MHC class I molecules that become loaded with exogenous antigen. In several studies, unconventional access to the conventional MHC class I pathway has been clearly demonstrated, while in others, unconventional loading of a population of post-Golgi MHC class I molecules takes place [19-21,35,40]. Although both pathways may be utilized in vivo, perhaps for different antigens, we [22] and others [47] have argued that for induction of effector CTL against, for example, virally infected cells, the former pathway is likely to be more important because it uses the same cytosolic machinery for generation and 'loading of MHC class I epitopes to which viral antigens synthesized de novo will be exposed. A similar argument can be made for tumor antigens. Given the likelihood that many antigens of viral and tumor origin will not be expressed de novo within DC, a conceptual problem arises in considering how such foreign antigens can reach D C class I molecules [47]. The appearance of exogenous HRP in the cytosol of mouse BMDC suggests an access route to the classical MHC class I pathway. We do not yet know whether such a pathway exists in vivo. Other workers have shown that phagocytosed antigen coupled to or taken up in association with synthetic beads is presented on MHC class I more efficiently than soluble antigen [24, 251. Whether this is simply due to enhanced antigen uptake or to enhanced access to the cytosol from phagosomes compared with other endosomes is not known. It will be worthwhile to compare the efficiency of presentation of the same amount of exogenous antigen taken up by macropinocytosis (soluble) versus phagocytosis (particulate). Although the concentrations of antigen used are higher in our studies compared with those using particle-driven uptake, the time of exposure to antigen is considerably shorter (< 10 min versus 4-24 h [24,25]). We suggest that both phagocytic and macropinocytic mechanisms may operate in vivo. Compared with macrophage cultures expanded in parallel, DC were more effective at triggering a MHC class I-restricted response and appeared to do so more rapidly following pulsing with exogenous antigen. As described previously [22, 241 in BMMQ, presentation utilized the classical class I pathway following access to the cytosol, as judged by the ablation of the response when inhibitors of either proteasomal processing or the secretory pathway were used. Importantly, DC expanded from TAP1 -/- mice showed equal pinocytic activity, but were unable to present exogenous OVA on MHC class I molecules. Although the cell types involved in cross-priming are still unknown, it was recently shown that a TAP-dependent pathway of loading of MHC class I molecules was responsible for this phenomenon in vivo. Bone marrow chimeric mice reconstituted with H-2h TAP-'- marrow were unable to mount a CTL response to influenza nucleoprotein (NP)
288
C. C. Norbury et al.
366-374 introduced in a H-2dt colorectal tumor cell line [15]. Mice reconstituted with TAP+'+ marrow developed a strong NP-specific H-2h-restricted CTL response as described [37]. TAP-dependent cross-priming suggests that in vivo, exogenous tumor cell-derived material must have access to the cytosol of host APC. Clearly, macropinocytosis, phagocytosis [24,48], or both would maximize the capture of tumor-derived material and, potentially, deliver a portion of it to the cytosol. In summary, BMDC expanded in GM-CSF are able to capture, process, and present an exogenous antigcn on MHC class I. Expression of the OVA epitope we have examined is strictly TAP dependent and is blocked by BFA and proteasome inhibitors. Cytosolic access is achieved via constitutive membrane ruffling and formation of large macropinosomes, some of which deliver their content to the cytosol. Only a proportion of cells achieves visible access of exogenous tracers to the cytosol (Fig. 2 and [22]) in vitro, and expression of exogenous antigen o n MHC class I appears to be substochiometric [25]. In fact, this might be a desirable feature in vivo, since presentation of pathogen-encoded antigens on a limited number of professional APC can occur, even when those APC are not themselves infected. In other words, a few APC displaying peptides derived from exogenous antigens are used disposably to ensure CTL priming. Potentially, this route could be manipulated therapeutically to allow the generation of responses in vivo to tumors or cells infected with intracellular pathogens, where the dominant peptide-specific responses are not known. We thank Nilabh Shastri for the B3Z cell line, David Gray for providing X63 Ag 8 no. 12 cells, Luc Van Kaer for the kind gift of the 8 6 TAP-I- mice, Barbara Spruce for access to video microscopy facilities; and Alexandra Livingstone, Gitta Stockinger, and Paul Crocker for antibodies. We would also like to acknowledge members of the lab for critical reading of the manuscript. C . W is supported by the Wellcome Trust and C . C . N . was the holder of an MRC studentship. H . G .L . is supported by the Swedish Medical Research Council and Swedish Cancer Society, and B.J.C. is the holder of a Karolinska Institute Fellowship.
5 References 1 Romani, N., Koide, S., Crowley, M., Witmer-Pack, M., Livingstone, A. M., Fathman, C. G . and Steinman, R. M., J. Exp. Med. 1989. 169: 1169. 2 Steinman, R. M., Annu. Rev. Immunol. 1991. 9: 271. 3 Austyn, J. M., J. Exp. Med. 1996. 183: 1287. 4 Inaba, K., Inaba, M., Romani, N., Aya, H., Deguchi, M., Ikehara, S., Muramatsu, S. and Steinman, R. M., J . Exp. Med. 1992. 176: 1693. 5 Sallusto, F. and Lanzavecchia, A., J . Exp. Med. 1994. 179: 1109. 6 Sallusto, F., Cella, M., Danieli, C. and Lanzavecchia, A,, J . Exp. Med. 1995. 182: 389. 7 Jiang, W., Swiggard, W. J., Heufler, C., Peng. M., Mirza, A., Steinman, R. M. and Nussenzweig, M. C., Nature 1995. 375: 151. 8 Celluzzi, C. M., Mayordomo, J. I., Storkus, W. J., Lotzc, M. T. and Falo, L. D. Jr., J . Exp. Med. 1996. 183: 283. 9 Young, J. W. and Inaba, K., J . Exp. Med. 1996. 183: 7. 10 Porgador. A. and Gilboa, E., J . Exp. Med. 1995. 182: 255. 11 Paglia, P., Chiodoni, C., Rodolfo, M. and Colombo, M. P., J. Exp. Med. 1996. 183: 317.
Eur. J. Immunol. 1997.27: 280-288 12 Macatonia, S. E., Taylor, P. M., Knight, S. D. and Askonas, B. A.. J . Exp. Med. 1989. 169: 1255. 13 Takahashi, H., Nakagawa, Y., Yokomuro, K. and Berzofsky, J. A , , Int. Immunol. 1993. 5: 849. 14 Nonacs, R., Humborg, C., Tam, J. P. and Steinman, R. M., J. Exp. Med. 1992. 176: 519. 15 Huang, A. Y. C., Bruce, A. T., Pardoll, D. M. and Levitsky, H. I., Immunity 1996. 4: 349. 16 Bevan, M. J., J. Immunol. 1976. 117: 2233. 17 Bevan, M. J., J . Exp. Med. 1976. 143: 1283. 18 Rock, K. L., Immunol. Today 1996. 17: 131. 19 Liu, T., Zhou, Z., Orvell, C., Lederer, E., Ljunggren, H.-G. and Jondal, M., J. Immunol. 1995. 154: 3147. 20 Schirmbeck, R., Malber, K. and Reimann, J., Eur. J . Immunol. 1995. 25: 1063. 21 Harding, C. V. and Song, R., J. Immunol. 1994. 153: 4925. 22 Norbury, C. C., Hewlett, L. J., Prescott, A . R., Shastri, N. and Watts, C., Immunity 1995. 3: 783. 23 Rock, K. L., Gamble, S. and Rothstein, L., Science 1990.249; 918. 24 Kovacsovics-Bankowski, M. and Rock, K., Science 1995. 267: 243. 25 Reis de Sousa, C. and Germain, R. N., J. Exp. Med. 1995. 182: 841. 26 Kartunnen, J. and Shastri, N., Proc. Natl. Acad. Sci. USA 1991. 88: 3972. 27 Kartunnen, J., Sanderson, S. and Shastri, N., Proc. Natl. Acud. Sci. USA 1992. 89: 6020. 28 Van Kaer, L., Ashton-Rickardt, P. G., Ploegh, H . L. and Tonegawa, S., Cell 1992. 71: 1205. 29 Racoosin, E. L. and Swanson, J. A,, J. Exp. Med. 1989. 170: 1635. 30 West, M. A., Bretscher, M. S. and Watts, C., J . Cell Biol. 1989. 109: 2731. 31 Hewlctt, L. J., Prescott, A. R. and Watts, C., J . Cell Biol. 1994. 124: 689. 32 Sanderson, S. and Shastri, N., Int. Immunol. 1994. 6: 369. 33 Racoosin, E. L. and Swanson, J. A., J . Cell Sci. 1992. 102: 867. 34 Stirpe, F., Olsnes, S. and Pihl, A.. J. Biol. Chem. 1980. 255: 6947. 35 Schirmbeck, R., Bohm, W., Melber, K. and Reimann, J., J. Immunol. 1995. 155: 4676. 36 Bohm, W., Schirmbeck, R., Elbe, A., Melber, K.. Diminky, D., Kraal, G., van Rooijen, N., Barenholz, Y. and Reimann. J., J. Immunol. 1995. 155: 3313. 37 Huang, A. Y. C., Golumbeck, P., Ahmadzadeh, M., Jaffee, E., Pardoll, D. and Levitsky, H., Science 1994. 264: 961. 38 Falo, L. D. Jr., Kovacsovics-Bankowski, M., Thompson, K. and Rock, K. L.. Nature Med. 1995. I : 649. 39 Wraith, D. C., Vessey, D. C. and Askonas, B. A., J . Gen. Virol. 1987. 68: 433. 40 Bachmann, M. F., Oxenius, A., Pircher, H., Hengartner, H., Ashton-Rickardt, P. A , , Tonegawa, S. and Zinkernagel, R. M., Eur. J. Immunol. 1995.25: 1739. 41 Staerz, U. D., Karasuyama, H. and Gamer, A. M., Nature 1987. 329: 449. 42 Boog, C. J. P., Kast, M., Timmers, H. T. M.. Boes, J., de Waal, L. P. and Melief, C. J. M., Nature 1985. 318: 59. 43 Young, J. W. and Steinman. R. M., J. Exp. Med. 1990. 18: 1315. 44 Carbone, F. R. and Bevan, M. J., J . Exp. Med. 1990.171: 377. 45 Nair, S., Zhou, F., Reddy, R., Huang, L. and Rouse, B. T., J. Exp. Med. 1992. 175: 609. 46 Collins, D. S., Findlay, K. and Harding, C. V., J . Immunol. 1992. 148: 3336. 47 Bevan, M. J., J. Exp. Med. 1995. 182: 639. 48 Reis de Souse, C., Stahl, P. D. and Austyn, J. M., J . Exp. Med. 1993. 178: 509.
