MHC-II presentation by dendritic cells (DC) is necessary both for initial priming of CD4 T cells and for induction of peripheral effector function. Although CD4 T cells can be critical for competent immunization-mediated cancer immunosurveillance, unmanipulated CD4 T cell responses to poorly immunogenic tumors result in either complete ignorance or tolerance induction, suggesting inadequate DC function. In this study, we investigated the phenotype, Ag uptake, and MHC-II presentation capacity of normal dermal DC and tumor-infiltrating DC (TIDC) in both lymphoid and peripheral sites. We found that murine tumors were extensively infiltrated by partially activated TIDC that closely resembled dermal DC by surface marker expression. However, in contrast to dermal DC, TIDC were inefficient at MHC-II presentation due to poor intrinsic protein uptake capability. This resulted in both inferior initiation of T cell responses in the draining lymph node and poor peripheral effector cell accumulation. In addition, TLR stimulation selectively enhanced MHC-II presentation of Ag by dermal DC, but not TIDC in the draining lymph node, and did not affect overall peripheral Ag uptake of either. These results show that TIDC are functionally distinct from normal interstitial DC, thus indicating that neoplastic tissues can evade effector CD4 T cells through modification of DC competence.
Cancer immunosurveillance can be mediated through multiple mechanisms that involve cells of the adaptive and innate immune system (1). Although CD8 T cells and NK cells have been the primary focus of cancer immunology due to direct cytolytic capabilities, CD4 T cells have recently been implicated as a critical cell type for successful immunosurveillance in vaccination and lymphopenia-induced cancer therapy models (2, 3, 4, 5, 6). CD4 T cells have been shown to mediate indirect and direct antitumor effects through a multitude of cell contact and cytokine-dependent mechanisms. Initial CD4 T cell activation in the draining lymph node (DLN)3 leads to up-regulation of CD40L surface expression, which through binding of CD40 on dendritic cells (DC) causes DC maturation, increased DC longevity, production of guidance chemokines, and initiation of third signal cytokine production to support CD8 T cell activation (7, 8, 9, 10, 11, 12, 13). In addition, CD4 T cells are considered to be an important source of IL-2, necessary for continued effector CD8 T cell expansion, as well as for formation of functional CD8 T cell memory (8, 14, 15). It has also been shown that effector CD4 T cells can greatly enhance CD8 T cell tumor infiltration after immunization (6). Together these helper effects have been proven necessary for optimal production and maintenance of effector and memory CD8 T cells, and therefore, for direct CD8 T cell antitumor activity (2, 3, 6, 13, 16, 17, 18, 19, 20, 21). In addition, effector CD4 T cells are capable of directly producing IFN-γ, an inflammatory cytokine involved in enhancement of intratumoral (i.t.) chemokine expression and immune cell recruitment, activation of DC and macrophages, stabilization of surface MHC-I/II expression, induction of apoptosis, inhibition of tumor angiogenesis, as well as augmentation of CD8 T cell expansion (22, 23, 24). Through these indirect and direct antitumor effects, CD4 T cells can orchestrate a comprehensive immunosurveillance program that can protect and treat diseased individuals.
In order for CD4 T cells to elicit effector function in 2° lymphoid and nonlymphoid peripheral tissues, such as cancerous tissues or infected sites, they need to be first activated by DLN DC-presenting specific MHC-II-peptide complexes. The DLN DC are a heterogeneous mix of lymph node (LN)-resident as well as recent tissue-migrant DC populations that have differing abilities to present proteins to CD4 and CD8 T cells (25). It has been shown that whereas both LN-resident and migratory interstitial DC can activate CD4 T cells, migratory DC carry greater levels of Ag and display a mature phenotype that allows full effector CD4 T cell activation and prolonged CD40L expression (12, 25, 26, 27, 28). It is thought that subsequent CD4 T cell DC interactions, which are necessary for continued CD4 T cell expansion and differentiation, allow licensing of immature LN-resident DC for complete maturation (29). Thus, CD4 T cells have the potential to license CD8α+ DC, the only DC subset shown to be efficient at processing peripheral proteins for cross-presentation on MHC-I, direct presentation on MHC-II, and production of high levels of cytokines necessary for CD8 T cell effector programming (8, 19, 30, 31, 32). After migration out of 2° lymphoid organs into peripheral target tissues, effector CD4 T cells again need to see the specific MHC-II-peptide complexes to be stimulated to produce effector cytokines and to accumulate to large numbers (33) (J. McLachlan and M. Jenkins, unpublished observations). Considering that most parenchymal and cancerous tissues do not express MHC-II, stimulation of CD4 T cell at these sites critically depends on infiltrating interstitial DC, and not on macrophages, that can acquire proteins through various phagocytic mechanisms and present epitopes on MHC-II following lysosomal degradation (34) (J. McLachlan and M. Jenkins, unpublished observations). Therefore, maximal antitumor CD4 T cell activity requires presentation of tumor-specific proteins by APCs in both DLN and inside the tumor. However, the capacity and kinetics of in vivo peripheral Ag uptake, processing, MHC-II presentation, and DC migration for both normal interstitial DC as well as tumor-infiltrating DC (TIDC) are poorly understood.
We, as well as others, have recently shown that whereas tumor-specific CD8 T cells were readily primed in the DLN of s.c. B16 melanoma and EL4 thymoma tumors, CD4 T cells remained completely ignorant of the progressing disease (35, 36). The lack of CD4 T cell activation thus prevented licensing of immature DLN-resident CD8α+ DC that were cross-presenting tumor-specific proteins on MHC-I, and therefore contributed to the ensuing CD8 T cell tolerance induction (35). We found that the absence of CD4 T cell responses was not due to general T cell unresponsiveness in tumor-bearing animals, but due to tumor-mediated inhibition of MHC-II presentation of both tumor-secreted and i.t. injected proteins in DLN DC, as compared with normal s.c. protein inoculations (35). Considering that proteins from nonlymphoid peripheral sites are delivered and presented to CD4 T cells to a large extent by migratory DC, we hypothesized that Ag handling and/or migration of TIDC to DLN were defective (26). Even though TIDC have been shown capable of ex vivo protein uptake and presentation to both CD4 and CD8 T cells after s.c. immunization, we previously observed that TIDC were unable to present tumor-specific proteins acquired directly in vivo to CD4 T cells, arguing that steady-state MHC-II presentation is defective in these cells (35, 37). In addition, defects in DC differentiation leading to poor maturation and cytokine secretion in steady-state and adjuvant-stimulated conditions have been described for human and murine neoplastic disease in both tumor-associated DC and TIDC (38, 39, 40).
To better understand the defects in tumor-Ag presentation to CD4 T cells, we have characterized the kinetics of Ag uptake, processing, MHC-II presentation, and DC migration for both normal interstitial dermal DC and TIDC from poorly immunogenic tumors. Elucidation of the unmanipulated as well as adjuvant-stimulated Ag handling and migratory properties of TIDC in vivo provides insight into the first stages of the innate and adaptive immune system cross-talk during cancer progression and after therapeutic immunization. This information is valuable for understanding the basis for the behavior and antitumor activity of responding tumor-specific T cells, and may ultimately be useful for diagnostic as well as therapeutic applications. In addition, dissection of normal interstitial DC function in peripheral tissues provides an improved perspective on peripheral CD4/DC interactions during s.c. immunizations, and has implications for delayed-type hypersensitivity responses, as well as various bacterial and viral infections.
Materials and Methods
Mice, tumor cell lines, and adoptive transfer
B16.F10 melanoma and EL.4 thymoma cell lines were maintained by in vitro culture in complete RPMI 1640 medium (cRPMI) (21). The B16.OVA melanoma line was maintained in cRPMI with 800 mg/ml G418 (Mediatech). The 4T1 breast cancer cell line was maintained in cRPMI supplemented with 2.5 g/L glucose. Male C57BL/6NCr mice (National Cancer Institute) were s.c. injected with 2 × 105 B16.F10 or B16.OVA cells, or with 2 × 106 EL.4 cells in PBS. Female BALB/cAnNCr mice (National Cancer Institute) were injected in the mammary fat pad with 5 × 104 4T1 cells in PBS. OVA257–264/H-2Kb-specific, TCR transgenic OT-I mice, and OVA323–339/I-Ab specific, TCR transgenic OT-II mice were a gift from F. Carbone (University of Melbourne, Melbourne, Australia). OT-I.PL and OT-II.PL mice were generated through crossing OT-I or OT-II mice with Thy1-congenic B6.PL-Thy1a/Cy (Thy1.1) mice (The Jackson Laboratory) and breeding to homozygosity. OT-II.PL Rag 1−/− were generated by crossing OT-II.PL to C57BL/6J-Rag1 tm1Mom (Rag1-deficient, 2216) mice (The Jackson Laboratory). In some experiments, 1 × 106 naive OT-I.PL cells, greater than 95% purity (21), and OT-II.PL Rag 1−/− were i.v. injected into animals 1 day before tumor inoculation. Experiments were conducted under specific pathogen-free conditions at the University of Minnesota, and were performed in compliance with relevant laws and institutional guidelines and with the approval of the Institutional Animal Care and Use Committee at the University of Minnesota.
Protein and adjuvant injections
Escherichia coli expressing EαGFP was a gift from M. Jenkins (University of Minnesota, Minneapolis, MN). The EαGFP protein was purified from bacterial lysates using a nickel resin His-Bind column (Novagen), as previously described (26). Contaminating endotoxin was removed using Triton X-114 (MP Biomedicals) phase separation and SM-2 Bio Beads (Bio-Rad). Protein was injected s.c. into naive animals or directly i.t. (50 μl vol) into day 10 B16.F10-bearing animals. Where indicated, EαGFP was injected into day 8 EL.4 and day 14 4T1 tumors. A total of 35 μg of CpG 1826 (Coley Pharmaceuticals), 50 μg of E. coli LPS (Sigma-Aldrich), or 35 μg of poly(IC) (Amersham Biosciences) was injected with the protein in some experiments. In some experiments, 5 × 108 1 μM unmodified red fluospheres (Invitrogen) were injected. For immunizations, 100 μg of OVA protein with 50 μg of LPS (Sigma-Aldrich) was injected i.v. into animals that were challenged with B16.F10 and B16.OVA on contralateral sides 6 days previously.
Histology, tissue dissociation, and flow cytometry
Tissue sections (8 μm) were made of day 12–14 B16.F10 frozen tumor tissues using a CM1800 microtome (Leica Microsystems). Sections were fixed in acetone, air dried, and hydrated for 10 min in PBS. Sections were blocked with CD16/CD32 Abs and 1% normal mouse serum PBS, and stained with Abs for CD11c, I-Ab, and CD11b for 40 min. Sections were then washed and stained with 4′,6-diamidine-2′-phenylindole dihydrochloride (Roche Applied Science). Aqua Poly/Mount medium (Polysciences) was then added to slides, which were then covered with Esco coverslips (Erie Scientific). Tissue sections were visualized on a Leica AF6000 microscope. For flow cytometric analysis of DC, LN or tumors of B16.F10-bearing mice were harvested and treated with 400 U/ml collagenase D (Roche Applied Science) solution, and incubated for 25 min at 37°C. EDTA (10 mM) was added for 1 min, followed by a PBS wash. A total of 1 × 2-cm shaved skin samples surrounding the injection site was incubated for 2 h in cRPMI containing the following: 0.25 mg/ml bovine testes hyalurnidase, 0.1 mg/ml bovine pancreas deoxyribonuclease I, and 2.7 mg/ml collagenase from Clostridium histolyticum (Sigma-Aldrich). Dermis was disassociated by pipetting and washed with 2% FCS/PBS. DC in all tissues were identified by gating on forward light scatter (FSC)high, CD3−, CD19−, CD11c+, MHC-II+ cells with further subgating on CD11b+ populations where indicated. Macrophage populations were identified by gating on FSChigh, CD3−, CD19−, CD11c−, CD11b+ cells, with verification for expression of F4/80 and not Ly6G surface markers. For immunization experiments, tissues were disrupted into single-cell suspensions through homogenization. Transferred OT-I.PL and OT-II.PL T cells were identified as Thy1.1+Vα2+ CD8+ or CD4+, respectively. All Abs, including YAe-bio, were obtained from eBiosciences, BD Biosciences, Invitrogen, or Biolegend. Cell counts were conducted on the entire collected tissues, including DLN and nondraining inguinal LN, spleen, the day 12 B16.F10 or B16.OVA tumor, and the 1 × 2-cm patch of skin by using PKH26 reference microbeads (Sigma-Aldrich). Cell density measurements were approximated by dividing the total cell number by the average collected tissue volume. Day 12 B16.F10 average tumor volume was 10 × 10 × 10 = 1000 mm3. Dermal tissue volume was calculated by multiplying the area of the collected skin by the previously published dermal depth for male C57BL/6NCr mice, as follows: 10 × 20 × 0.5 = 100 mm3 (41). All stained cells were collected on a LSR-II flow cytometer (BD Biosciences), and analyzed using FlowJo software (Tree Star). Statistical significance was calculated by using unpaired two-tailed Student’s t tests, unless indicated. Value of p representation is as follows: less than 0.05 are represented by one asterisk; less than 0.01 by two asterisks; and less than 0.001 by three asterisks.
Multiple transplantable tumor models inhibit MHC-II presentation in DLN DC
To examine Ag uptake and presentation by DC, we used the EαGFP fluorescent protein, a fusion of the MHC I-Eα-chain and GFP. Uptake of the Ag can be directly examined by direct fluorescence detection, and MHC-II presentation of the Eα52–68 epitope can be measured by staining with the YAe Ab that is specific for I-Ab in complex with the Eα52–68 epitope (26). We previously found that i.t. injection of EαGFP protein into B16.F10 melanoma resulted in lower numbers of both GFP+ and MHC-II-peptide-presenting DC in tumor DLN 24 h postinjection, in comparison with s.c. injection in normal mice (35). To test whether decreased Ag presentation in tumor DLN extended to other s.c. transplantable tumors, we injected EαGFP protein either s.c. into nontumor-bearing C57BL/6 or BALB/c animals, or i.t. into B16.F10 melanoma, EL4 thymoma, or 4T1 mammary carcinoma. We found that skin DLN contained a relatively large population of GFP+ DC that were almost exclusively of the CD11b+ mature migratory phenotype (Fig. 1,A, and data not shown). As expected, cell surface I-Ab-Eα52–68 complex was only detected with the YAe Ab on the GFP+ DC in the C57BL/6 skin DLN, and not in I-AE+ BALB/c skin and 4T1 tumor DLN. In sharp contrast to DC from s.c. injections, DLN from all of the tumors contained a smaller number of GFP+ cells, as well as lower levels of YAe staining (B16.F10 and EL4 tumors) (Fig. 1,A). Additionally, the cells that were GFP+ in tumor DLN displayed a significantly lower level of GFP compared with GFP+ DC in skin DLN (Fig. 1 B). We did observe some experimental variability in the differences between B16.F10 and skin DLN in the numerous experiments that were done, which was most likely attributable to variations in drainage from the injected tumor site and the time required for dermal DC migration. Nonetheless, clear decreases in GFP mean fluorescence intensity (MFI) and MHC-II-peptide complex formation in the GFP+ cells were observed in all experiments. These results showed that reduced DC-mediated Ag presentation in the tumor vs dermal DLN is a common feature of at least the three tumors examined in this study in two different mouse strains. This most likely accounts for the complete CD4 T cell ignorance to either tumor-specific Ag or i.t. injected proteins that we previously observed (35).
Tumors are infiltrated by a large number of partially activated CD11b+ DC
Because migratory DC bring in the highest levels of peripheral proteins into the DLN, we hypothesized poor initial tumor infiltration by DC might account for diminshed numbers of DC migrants, and hence, the low level of Ag presentation found in tumor DLN (26). We therefore compared the numbers and phenotypes of DC present in the tumor and compared these with a 1 × 2-cm patch of skin surrounding the s.c. injection site. By gating on FSChighCD3−CD19−CD11c+I-Ab+ cells, we found that tumor tissues were infiltrated by large numbers of DC that stained positive for the myeloid and interstitial DC marker CD11b+, and negative for the Langerhans cell marker EpCAM, plasmacytoid DC marker PDCA-1, resident LN/spleen DC marker CD8α, and myeloid DC marker CD4 (Fig. 2,A) (25). Additionally, the TIDC did not express high levels of GR-1 (data not shown), a marker commonly associated with myeloid suppressor cells (40). In addition, the CD11b+ DC in all examined tissues, including the tumor site, appeared to have relatively high levels of surface F4/80 expression (data not shown). Although commonly used to isolate macrophage populations, this marker is also expressed on myeloid DC in lymphoid tissues, dermal DC, Langerhans cells, and other interstitial DC (42, 43, 44, 45, 46, 47). Overall, the TIDC phenotype closely resembled that of normal interstitial dermal DC in the skin, as well as some of the DC found in the skin DLN that are thought to be the recent dermal DC migrants (Fig. 2,A). Interestingly, the number of TIDC was ∼100-fold higher than the number of dermal DC in the skin site. When normalized based on the volume of tissue sampled, TIDC density surpassed that of the dermal DC in the skin by ∼10-fold (Fig. 2,C). There were no differences in total number and density of CD11b+ DC in skin vs tumor DLN (Fig. 2, B and C), even though tumor DLN were often enlarged. Fluorescence microscopic examination of frozen tumor samples revealed that tumors were infiltrated by MHC-II+CD11c+ cells (Fig. 2,D, zoomed-out image) that also stained positive for the CD11b marker (Fig. 2,D, zoomed-in inset). These TIDC appeared to infiltrate throughout the entirety of the tissue, and in some parts, including peripheral edges of the tumor, exhibited areas of high tissue infiltration (Fig. 2 D).
To further characterize TIDC phenotype, we examined overall DC activation status by looking at MHC-II (I-Ab) and costimulatory molecule expression. TIDC expressed intermediate levels of MHC-II as compared with bulk skin LN CD11b+ DC, suggesting that the TIDC population was somewhat heterogeneous and comprised of more and less mature subpopulations with dimmer or brighter levels of MHC-II, respectively (Fig. 2,E). Interestingly, both MHC-II+bright and the majority of MHC-II+dim TIDC displayed high levels of CD86 and CD80 surface expression. However, CD40 expression was overall lower in comparison with MHC-IIhigh LN DC (Fig. 2 E). In addition, TIDC did not produce significant levels of IL-12, further providing support that as a population the TIDC were only partially mature (data not shown). In contrast to the partially activated phenotype of the TIDC, normal dermal DC displayed a fully immature phenotype (data not shown), consistent with the notion that DC migrate out of normal peripheral tissues upon maturation. Taken as a whole, these results indicated that tumors were well infiltrated by large numbers of TIDC that displayed normal interstitial DC markers and exhibited a partially activated phenotype.
TIDC exhibit marked differences in migration to DLN compared with dermal DC
The inability to find many GFP+ DC in tumor DLN (Fig. 1,A), despite large numbers of DC in the tumor, could have been due to poor ability of TIDC to migrate to DLN over the times we examined. Because GFP fluorescence will decrease over time due to proteolysis, we used indigestible, 1-μm-sized red fluospheres to compare the ability of dermal DC and TIDC (FSChighCD3−CD19−CD11c+MHC-II+CD11b+ cells) to migrate from the peripheral injection site to the DLN. Injection of 5 × 108 beads either s.c. or i.t. resulted in phagocytosis of the beads at the respective sites by both dermal DC and TIDC within 4 h (Fig. 3,A), and there was at least a 10-fold greater number of bead-positive DC in the tumor compared with the dermis (Fig. 3,B). This is not surprising considering that the total number of TIDC is much greater than the number of dermal DC in the skin surrounding the s.c. injection site (Fig. 2,B). Although the total number of bead+ dermal DC in the skin declined over time, the overall number of bead+ TIDC remained relatively unchanged, resulting in even greater differences in total numbers of bead+ DC (Fig. 3,B). To examine the kinetics of DC migration to the DLN, we isolated DLN at various time points after bead injection and gated on FSChighCD3−CD19−CD11c+CD11b+I-Ab+bright cells. Considering that all dermal and TIDC are of a mature phenotype upon migration to DLN, this effectively allowed us to reduce the background due to bead uptake by immature LN-resident DC (26, 35). Even though the numbers of DC were drastically different at the injection sites (Fig. 3,B), the total numbers found in the tumor and skin DLN were virtually identical, as were the rates of accumulation (Fig. 3,C). These data suggested that on a population level, TIDC were defective in migrating to DLN, consistent with the fact that the total numbers of CD11b+ DC in skin and tumor DLN were equivalent, even though tumor DLN were draining a site containing large numbers of DC (Fig. 2 B). Even though the TIDC were poor at migrating to DLN as compared with dermal DC, the total number of migrants found in DLN was comparable, suggesting that DC migration capacity could not fully explain the low Ag levels and MHC-II presentation in tumor DLN DC.
TIDC are poor at Ag uptake compared with dermal DC
Following injection of EαGFP, the overall intracellular GFP MFI was substantially lower in tumor DLN DC compared with the skin DLN DC (Fig. 1,B), suggesting that TIDC may take up less protein than dermal DC. To examine this, we performed a kinetic comparison of EαGFP uptake by dermal DC in the skin and TIDC inside the tumor. Although both dermal DC and TIDC quickly took up EαGFP protein, dermal DC were vastly superior to TIDC in overall level of uptake at all time points (Fig. 4, A–C). Kinetic comparison revealed that even though the GFP MFI was substantially lower for TIDC, the overall percentage of GFP+ TIDC at the peak of uptake (∼4 h) was not that dissimilar from dermal DC at the peak of uptake (∼3 h) (Fig. 4,B). Because the intracellular GFP MFI levels were frequently ∼10-fold lower in TIDC than in dermal DC, we tested whether increasing the EαGFP inoculation dose by 10-fold (500 μg) would rescue Ag uptake in TIDC. Interestingly, whereas the overall TIDC GFP MFI levels rose substantially to the levels seen in dermal DC after a 50 μg injection, TIDC Ag uptake remained vastly inferior to dermal DC for the 500 μg dose (Fig. 4, A and D).
Considering that the total number of DC was much greater inside the tumor as compared with the dermal site (Fig. 2,B), it was possible that TIDC were simply competing more for the low levels of available soluble protein. In addition, s.c. and i.t. injections deliver proteins into distinctive sites with different microenvironments and tissue volumes, thus making it possible that DC extrinsic factors account for the differences in uptake. To test whether the protein uptake differences seen were due to cell intrinsic or extrinsic mechanisms, we purified dermal DC and TIDC from the tissues and plated the cells ex vivo with different concentrations of EαGFP for 45 min at 37°C to allow uptake of the soluble protein with cells under similar conditions. However, even under these conditions, the TIDC were inferior to dermal DC in their ability to take up protein to a similar extent as observed in vivo (Fig. 4,E). These profound Ag uptake differences by TIDC were observed in all of the tested tumor systems (Fig. 4 F). These results showed that whereas both normal dermal DC and TIDC mediated maximal protein uptake within several hours of protein delivery, dermal DC were intrinsically superior to TIDC in the overall uptake capacity.
Dermal DC and TIDC exhibit similar protein degradation kinetics
It has been shown recently that enhanced lysosomal protein degradation by DC can lead to reduced levels of intracellular protein both in the periphery and in DLN, and cause a reduction of MHC-II presentation to CD4 T cells (34). Thus, in addition to reduced uptake, enhanced degradative capacity in TIDC could potentially contribute to the lower levels of EαGFP protein observed in the DLN. To examine protein degradation, we determined the amount of intracellular protein remaining over time as a percentage of the level at the peak of uptake (3 h for dermal DC; 4 h for TIDC). The rates of loss of fluorescence for dermal DC and TIDC were essentially identical, suggesting that the speed of protein processing was similar in both cell types (Fig. 5,A). Interestingly, we found that the overall levels of intracellular EαGFP quickly stabilized following an initial phase of degradation in both dermal DC and TIDC, suggesting that some of the protein may be entering into the nondegradative intracellular vesicular compartment (48). Considering that the levels of intracellular protein were much higher in dermal DC than in TIDC (Fig. 4,C), we compared the decrease in GFP MFI over 24 h in dermal DC and TIDC in a situation where the levels of intracellular EαGFP were equivalent by injecting 50 μg s.c. for dermal DC and 500 μg i.t. for TIDC. Here too, the rate of decrease in fluorescence was identical for dermal DC and TIDC (Fig. 5 B), again indicating that the overall rates of protein degradation were similar for these cells and independent of intracellular protein concentration.
To confirm these results in an EαGFP-independent assay, we used a fluorogenic DQ OVA protein (DQ-OVA). DQ-OVA is heavily conjugated to BODIPY fluorescent dyes that reach autoquenching levels, thus only becoming fluorescent after intracellular proteolytic degradation and physical separation of the OVA peptides. We hypothesized that if the decreased intracellular protein content and MHC-II presentation in tumor DLN DC (Fig. 1,A) are solely due to enhanced rates of protein degradation by TIDC, then we would observe greater intracellular FL-1 fluorescence levels in TIDC DLN migrants compared with s.c. dermal DC migrants after DQ-OVA inoculation. In contrast, we observed a reduced percentage of FL-1+ DC, and more importantly, much lower FL-1 MFI of FL-1+CD11b+ DC in the tumor DLN compared with skin DLN (Fig. 5 C). These results were consistent with TIDC, picking up lower amounts of DQ-OVA protein at the tumor site before migrating to the DLN, thus resulting in lower fluorescence levels following DQ-OVA degradation. Together these results strongly argued that dermal DC and TIDC have comparable protein degradation kinetics, and that the primary defect in TIDC is the reduced ability to take up proteins.
TIDC and dermal DC exhibit similar intrinsic MHC-II presentation capacity
The above results demonstrate that TIDC are less efficient than dermal DC in taking up protein Ag. It was also possible that a reduced ability to process the protein and form I-Ab-Eα52–68 complexes might contribute to the reduced levels of MHC-II-peptide complexes on migratory TIDC compared with dermal DC in the DLN (Fig. 1,A). To examine this, we compared the ability of these subsets to present EαGFP on MHC-II at the peripheral site before migration. The overall percentage of both TIDC- and dermal DC-presenting I-Ab-Eα52–68 complexes was relatively low (Fig. 6,A), even though a majority of cells had taken up EαGFP protein (Fig. 4,A). GFP+ TIDC exhibited a lower percentage and MFI of YAe+ cells than dermal DC at the 50 μg dose of injected protein (Fig. 6, A and B), but this difference was not found when a 500 μg dose of protein was injected. Considering that dermal DC took up more soluble protein than TIDC, and that Ag presentation of I-Ab-Eα52–68 complexes appeared to positively correlate with the amount of intracellular protein (Fig. 4,A), we normalized the protein uptake of the DC populations by creating a series of gates of increasing GFP MFI (Fig. 6,C). This allowed us to compare presentation of I-Ab-Eα52–68 complexes by dermal DC and TIDC that had taken up similar amounts of EαGFP. This analysis revealed that both dermal DC and TIDC were indistinguishable in their presentation capacity, because both the percentage of YAe+ cells and the YAe MFI increased comparably with increasing levels of intracellular EαGFP for both cell types (Fig. 6, D and E). These results argued that the observed deficiency in TIDC MHC-II presentation on the population level was not due to an intrinsic inability of the cells to process Ag, but due to the defective uptake that resulted in lower intracellular levels of EαGFP available for processing. These results also demonstrated that peripheral MHC-II presentation in both normal skin and tumor peripheral sites requires fairly robust Ag uptake, and that the level of MHC-II-peptide complex formation appears to directly correlate with amount of ingested protein.
Even though DC are thought to be the predominant APC population, macrophages can also process proteins for MHC-II presentation, albeit at much lower levels (34). We found that tumors contained a population of tumor-infiltrating CD11b+CD11c− macrophages that expressed high surface levels of F4/80 and low levels of Ly6G, and were relatively similar in total cell numbers as compared with TIDC (Fig. 7,A, and data not shown). Interestingly, macrophages found in the dermis and the tumor took up similar levels of EαGFP as their DC counterparts, indicating that inhibition of Ag uptake by the tumor microenvironment extended to multiple immune cell types (Fig. 7,B). Importantly, macrophages in both the dermis and the tumor had lower levels of surface MHC-II and were significantly worse than DC in processing soluble proteins for MHC-II presentation (Fig. 7 C, and data not shown), suggesting that DC may be the dominant APC in both DLN and in peripheral skin and tumor sites, and that neither TIDC nor tumor-infiltrating macrophages efficiently present soluble proteins (34).
TIDC are poor at presenting proteins to effector CD4 T cells in vivo
We previously showed that DC in tumor DLN were not presenting Ag to naive CD4 T cells, which in turn prevented the potential for indirect helper effects as well as for direct antitumor activity (35). These results suggested that the ignorant status of tumor-specific CD4 T cells might allow for effective therapeutic immunization of the naive CD4 cell pool even after initiation of robust tumor growth. However, the poor ability of TIDC to present soluble proteins on MHC-II (Fig. 6,A) raised the possibility that even if effector CD4 T cells did traffic to the tumor site post immunization, they would not be restimulated to elicit antitumor effects. Consistent with this hypothesis, TIDC purified from B16.OVA melanoma tumors were incapable of activating OVA-specific OT-II CD4 T cells directly ex vivo (35). To test whether TIDC could function as APCs in the tumor in vivo, we adoptively transferred naive OT-I and OT-II T cells (OVA-specific CD8 and CD4 T cells, respectively) into C57BL/6 mice that were then s.c. challenged with B16.F10 and B16.OVA tumor cells injected into contralateral sides (Fig. 8,A). Injection of both B16.F10 and B16.OVA into the same animals allowed for discrimination of T cell tumor infiltration, accumulation, and sensing of presented OVA protein in an Ag-specific manner within the same host. Some of the animals were then immunized with OVA protein and LPS adjuvant i.v. on day 6 of tumor growth to stimulate robust activation and expansion of effector OT-I and OT-II T cells. Without immunization, the total number of OT-I T cells in the spleen and the two tumor sites combined only increased ∼3-fold after 12 days of tumor growth, compared with tumor-free animals, consistent with the previous findings of tolerogenic CD8 T cell activation (35). As previously shown, OT-II T cells remained naive (ignorant), did not undergo expansion, and did not traffic to the tumor site without immunization (Fig. 8,B, and data not shown). OVA/LPS immunization induced robust expansion of both OT-I and OT-II T cells in the examined tissues, with OT-I T cells totally expanding ∼4-fold more than OT-II T cells (Fig. 8,B). Examination of the T cell tissue distribution revealed a large prevalence of OT-I T cells in the B16.OVA tumor, with relatively low numbers of cells in the spleen and the B16.F10 tumor (Fig. 8,C). In contrast, OT-II T cells largely remained in the spleen, did not infiltrate either tumor in large numbers, and did not preferentially accumulate in an Ag-dependent manner in the B16.OVA tumor. This resulted in greater than a 2500-fold difference in OT-I vs OT-II cell numbers in the B16.OVA tumor site (Fig. 8,C). Interestingly, both OT-I and the small number of OT-II T cells in the B16.OVA tumor up-regulated CD69 expression. This suggested that a low level of MHC-II presentation may occur within the tumor, but be insufficient to promote CD4 T cell accumulation and/or division (Fig. 8 D). In agreement with low CD4 T cell accumulation and hence marginal antitumor function, we observed minimal effects on tumor growth after immunization, which was to some degree surprising considering the degree of CD8 T cell infiltration (data not shown). Together these results indicated that whereas activated CD8 T cells could infiltrate the tumor in great numbers and robustly sense MHC-I-peptide complexes, effector CD4 T cells were poor at overall migration and Ag-specific retention at the tumor site, most likely as a result of defective Ag uptake and presentation on MHC-II by TIDC.
TLR-mediated TIDC maturation does not rescue Ag uptake and DLN MHC-II presentation
TLR stimulation has been shown to induce peripheral DC maturation and to cause enhanced human and murine TIDC function and cytokine secretion ex vivo (38, 39). Additionally, DC maturation has been observed to transiently increase Ag uptake in vitro and induce migration of DC from the peripheral tissues to the DLN in vivo (25, 49). We therefore tested whether TLR-induced TIDC activation could rescue Ag uptake in the tumor and presentation in the DLN by coinjecting EαGFP protein with TLR9 ligand CpG, TLR3 ligand poly(I:C), or TLR4 ligand LPS. As expected, all TLR ligands induced a robust increase in MHC-II presentation in skin DLN after s.c. inoculation into nontumor-bearing animals (Fig. 9,A). In contrast, i.t. administration of TLR ligand induced marginal effects on Ag presentation in tumor DLN, consistently showing significantly lower percentages of GFP+YAe+ DC as compared with skin DLN (Fig. 9,A). In addition, whereas the overall intracellular GFP levels were increased in migratory dermal DC in skin DLN, there was no effect on GFP levels in the GFP+ DC found in tumor DLN (Fig. 9,B). Importantly, peripheral tissue Ag uptake by either dermal DC or TIDC was unaltered in response to TLR stimulation (Fig. 9,C). We consistently observed enhanced maturation of peripheral dermal DC in response to CpG at 3 h and to poly(I:C) and LPS at 24 h postinjection, and clear TIDC maturation in response to CpG and LPS TLR ligands, as evidenced by an increased proportion of I-Ab+brightCD86+ cells in the total DC population, indicating that both types of DC were at least partially responsive to TLR stimulation (Fig. 9 D, and data not shown). Additionally, most DLN DC exhibited a mature phenotype 24 h post-TLR injection, demonstrating the effectiveness of immunization (data not shown). These data suggested that increased GFP MFI levels observed in dermal DC migrants in skin DLN after TLR ligand injection were not primarily due to enhanced peripheral protein uptake, but potentially due to increased DC migration speed and/or decreased degradation rates. In agreement with this, we observed a decrease in the rate of protein degradation in peripheral dermal DC after poly(I:C) administration (data not shown). Additionally, although not being able to measure TLR-induced dermal DC migration, we observed an enhancement of TIDC migration to DLN post-CpG treatment in experiments using fluorescent microspheres (data not shown).
Interestingly, in several experiments, we observed a mild, but significant enhancement of TIDC Ag presentation 24 h after CpG administration, even though no significant increase was seen for dermal DC (Fig. 9 E). However, a CpG immunization trial failed to induce naive CD4 T cell priming in the DLN and i.t. effector CD4 T cell accumulation, suggesting that the major limitation for optimal DC function continued to be poor tumor-Ag uptake and not overall MHC-II processing (data not shown). These results suggested that even though the tested TLR ligands could cause enhanced maturation and migration to DLN in dermal and TIDC, they did not significantly enhance in vivo peripheral tissue Ag uptake, and therefore had minimal effects on peripheral MHC-II presentation.
Adaptive immune responses are initiated by the presentation of peptides from foreign proteins on MHC molecules by DC in the 2° lymphoid sites. We have previously observed that CD8 T cell responses to cytoplasmic and secreted tumor Ag can be robust, but result in overall tolerance induction (35). In contrast, CD4 T cells remain ignorant throughout disease progression, due to defective MHC-II presentation of tumor-derived proteins on DLN DC. We have now examined the basis for this defective MHC-II presentation in several s.c. cancer models and on two murine backgrounds. We found that tumors are infiltrated by large numbers of CD11b+ DC, a large fraction of which have a partially activated phenotype. In comparison with normal dermal DC, these TIDC are very inefficient at taking up soluble proteins, but are comparable in the relative rates of protein degradation and formation of MHC-II-peptide complexes. Although TIDC exhibit a partial block in migration to the DLN, these data show that the inability to present sufficient Ag to activate naive CD4 T cells in the DLN, or stimulate effector CD4 T cells within the tumor appears to result primarily from the limited capacity of TIDC to internalize protein Ag.
Interestingly, the contrasting responses of CD8 and CD4 T cells to the same tumor-secreted protein suggest that MHC-I and MHC-II presentation in the DLN is physically uncoupled and is differentially regulated. In agreement with this notion, we found that whereas the LN-resident CD8α+ DC were efficient at cross-presenting tumor-derived proteins to CD8 T cells, only the activated/migratory CD11b+ DC had any detectable MHC-II-peptide complexes (35). These findings suggest a model in which the DLN receives tumor-derived lymph fluid carrying cell debris along with various tumor proteins and cell membranes, which then gets efficiently taken up by immature CD8α+ DC and selectively directed toward MHC-I processing (50). At the same time, TIDC are poor at protein uptake and to some degree migration, and are thus unable to function as professional APC in the DLN for initiating CD4 T cell priming. The proposed scenario for antitumor responses is similar to one described by Jenkins and colleagues (26) for soluble protein immunization, in which Ag presentation to CD4 T cells in the DLN occurs via both resident and migratory DC, with the exception that in the tumor model the draining proteins are cell associated and hence are targeted to MHC-I presentation, and migratory DC are defective at Ag uptake and therefore MHC-II presentation. However, it is entirely plausible that in addition to TIDC defects, resident DLN DC have defective MHC-II presentation machinery and contribute to CD4 T cell ignorance. Elucidating these nonmutually exclusive mechanisms behind the apparent MHC-I/II presentation paradox will require further study.
Microscopic examination revealed that TIDC heavily infiltrated the peripheral and inner parts of the tumor tissues with some indication of infiltration-rich regions. Phenotypic examination of TIDC showed that they were largely indistinguishable from normal interstitial dermal DC, which is in agreement with other data that have described a large prevalence of CD11b+ DC infiltrating various murine tumors (36, 51). Interestingly, in contrast to the immature peripheral dermal DC, TIDC expressed intermediate to high MHC-II and costimulatory levels, and low levels of intracellular IL-12. These results suggest that TIDC undergo a partial maturation program, possibly through experiencing inflammatory cytokines, such as TNF-α, at the tumor site (52). It has been shown that DC maturation decreases phagocytosis and prevents uptake and presentation of proteins on MHC (49, 53). It is thus possible that the partial activation state of TIDC is at least in part responsible for causing poor Ag uptake and MHC-II presentation. This is consistent with the fact that the immature dermal DC were efficient at Ag uptake, which allowed them to maintain high levels of available intracellular protein after maturation and migration to the DLN. These high intracellular protein levels appeared to directly correlate with the amount of detectable surface MHC-II-peptide complexes in both the periphery and the DLN, suggesting that efficient Ag acquisition is required for optimal MHC-II presentation. We found that TLR ligand coadministration caused an increase in maturation and migration in both dermal DC and TIDC and greatly enhanced Ag presentation in skin DLN. This treatment, however, failed to alter peripheral Ag uptake in both subsets. Although it has been shown that DC undergo a transient increase in Ag phagocytosis post-TLR-induced maturation in vitro, our results argue that Ag acquisition in vivo occurs relatively rapidly and independently of simultaneous TLR signaling (49). Interestingly, after the initial uptake and degradative phase, dermal DC appeared to store the protein at fairly stable levels, suggesting that some of the soluble protein can enter the nondegradative intracellular vesicular compartment that has been described to play a role in Ag transfer to B cells (48, 54). It is also possible that limiting Ag processing for MHC-II presentation in the periphery could be important for reducing tissue inflammation and destruction during the effector CD4 T cell response. It is important to note that the observed DC Ag-handling characteristics describe responses to soluble proteins, which are likely to be different from bacterial or viral infections.
It has been shown recently that i.t. T cell infiltration and accumulation directly correlate with therapeutic efficacy (55). In our hands, poor Ag uptake and presentation by TIDC not only precluded the initial priming of CD4 T cells in DLN, but also appeared to prevent effector CD4 T cell accumulation at the tumor site and correlated with a lack of tumor regression (35). Even i.t. administration of CpG, which was observed to partially enhance MHC-II presentation, but not Ag uptake, was incapable of rescuing effector CD4 T cell accumulation (data not shown). Interestingly, we also observed differences between Ag-nonspecific CD8 and CD4 T cell tumor infiltration, with CD8 T cells being much better than CD4 T cells at infiltrating the neoplastic tissues. In agreement, we have previously observed that endogenous CD8 T cells vastly outnumbered CD4 T cells inside the tumor (35). Although we initially attributed these results to reduced initial priming of endogenous CD4 T cells in tumor DLN, it is possible that both priming and ability to infiltrate the tumor are defective in the CD4 T cell population. Although defective tumor infiltration by T cells has been previously observed and attributed to a lack of integrin expression on the tumor endothelium, the disparity between CD4 and CD8 T cell infiltration observed in our model suggests that differences in various adhesion molecules exist directly on CD4 and CD8 effector T cells (55, 56). Interestingly, TIDC were efficient at infiltrating and accumulating to great numbers inside the tumor tissue, suggesting that this immune cell subset carries the appropriate adhesion molecules for proper tumor infiltration. In contrast, TIDC were relatively poor at migrating to the DLN, suggesting that TIDC do not possess the necessary surface molecules to mediate exit or to guide them into the DLN. However, it is also possible that disease-associated tissue microenvironment is simply not permissive for TIDC migration after initial infiltration. Further characterization of integrin and adhesion molecule expression and function on tumor endothelium, effector CD4 and CD8 T cells, and TIDC is necessary to understand the requirements for immune cell trafficking into and out of neoplastic tissues.
Although poor TIDC migration most likely contributed to CD4 T cell ignorance, it is clear that defective Ag acquisition by TIDC served as the major limiting step in both initiation and facilitation of antitumor CD4 T cell responses. This defect can be attributed to several different nonmutually exclusive mechanisms. In addition to partial maturation, improper TIDC differentiation due to a lack of appropriate growth factors in the tumor microenvironment can cause poor Ag uptake. It is also possible that inhibitory factors present in the tumor may cause specific deregulation of TIDC function (57, 58). Although TGF-β and IL-10 have been described to negatively modulate antitumor immune responses, we did not observe enhancement of Ag uptake by TIDC or initiation of CD4 T cell priming in DLN after TGF-β Ab blockade or in IL-10-deficient animals, respectively (our unpublished observations). However, it is possible that several suppressive factors act to inhibit TIDC functional competence and/or differentiation, and neutralization of all of them is necessary for restoration of TIDC function (57, 58). In addition, as shown for induction of IL-12 production by TIDC, combining both anti-inhibitory and proinflammatory signals might be required for inducing full TIDC function (51). Additionally, nutrient starvation is likely to occur at the tumor site and has been previously shown to induce autophagy and reduce soluble Ag capture, thus specifically decreasing presentation of peripheral proteins and enhancing presentation of self-proteins on MHC-II (59). Interestingly, we observed that tumor-infiltrating macrophages also took up lower levels of Ag as compared with their normal skin counterparts. Although clearly not being as efficient at Ag processing for MHC-II presentation as DC, these results suggest that inhibition of Ag uptake by the tumor microenvironment is a global phenomenon that extends to multiple immune cell subsets. Understanding the mechanisms behind defective Ag acquisition as well as immune cell migration should allow for better understanding of clinical data from various immunotherapy trials, and more importantly, for enhancing design of future therapeutic regimens.
We thank Dr. Stephen C. Jameson, Dr. Marc K. Jenkins, Dr. Christopher A. Pennell, Dr. Dan Kaplan, and Kerry A. Casey for critical discussion.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants AI34824 (to M.F.M.), CA82596 (to M.F.M.), and NCI 2 T32 CA009138-31 (to M.Y.G.).
Abbreviations used in this paper: DLN, draining lymph node; cRPMI, complete RPMI 1640 medium; DC, dendritic cell; FSC, forward light scatter; i.t., intratumoral; LN, lymph node; MFI, mean fluorescence intensity; TIDC, tumor-infiltrating DC.