Monocyte-derived DC (MoDC) can sense minor amounts of danger-associated molecules and are therefore easily primed. We provide a detailed protocol for the isolation of MoDC from blood and tumors and their activation with immune complexes while highlighting key precautions that should be considered in order to avoid their premature activation.
Dendritic cells (DC) are heterogeneous cell populations that differ in their cell membrane markers, migration patterns and distribution, and in their antigen presentation and T cell activation capacities. Since most vaccinations of experimental tumor models require millions of DC, they are widely isolated from the bone marrow or spleen. However, these DC significantly differ from blood and tumor DC in their responses to immune complexes (IC), and presumably to other Syk-coupled lectin receptors. Importantly, given the sensitivity of DC to danger-associated molecules, the presence of endotoxins or antibodies that crosslink activation receptors in one of the isolating steps could result in the priming of DC and thus affect the parameters, or at least the dosage, required to activate them. Therefore, here we describe a detailed protocol for isolating MoDC from blood and tumors while avoiding their premature activation. In addition, a protocol is provided for MoDC activation with tumor IC, and their subsequent analyses.
Since their discovery, dendritic cells (DC) have been a focus of extensive research due to their unique ability to skew T cell differentiation1. Over the past several decades, an extensive research effort has sought to define the various DC subsets and their function during tumor progression and immunity 2. DCs are composed of heterogeneous cell populations that differ from each other in their pattern-recognition receptors, tissue distribution, and migratory and antigen presentation capabilities3,4,5. Compared to other DC subsets, monocyte-derived DC (MoDC) are far more abundant in tumors and can be easily generated from circulating or tumor-infiltrating monocytes6,7. Therefore, many clinical trials seeking to take advantage of their relative prevalence are based on in vivo and ex vivo manipulation of autologous MoDC in order to elicit T cell immunity 8,9.
Similarly, DC-based vaccination of experimental tumor models requires 2-3 serial injections, 5-7 days apart, of 1-2 x 106 activated DC pulsed with tumor antigens. Therefore, to achieve this large number of DC, most mouse studies have primarily used MoDC cultured from bone marrow (BM) precursors in GM-CSF for 7-9 days (IL-4 is not needed in the mouse setting)10,11. Nonetheless, given that GM-CSF knockout mice have overall normal DC compartment 12,13, and given the mixed populations obtained from that culture,14 the physiological relevance of these DC has been called into question.
Alternatively, DC may be routinely isolated from spleen cells. However, DC comprise only about 0.3-0.8% of total spleen cells (resulting in approximately 7 x 105 DC/spleen), and of these cells, only CD103+ DC and MoDC can migrate back to lymphoid organs. Since MoDCs comprise approximately 10-15% of splenic DC populations15,16, most isolation protocols yield approximately 1 x 105 MoDC per spleen. Expansion of MoDC can be achieved by injecting transfected B16 cells that secrete GM-CSF, resulting in a 100-fold increase in splenic MoDC17. However, the use of MoDC for developing DC vaccines is limited since this procedure cannot be done in humans and the obtained MoDC are already highly activated.
In addition to obtaining adequate numbers of DC, another challenge for developing effective DC vaccines against autologous cancer cells involves the lack of sufficient danger signals in the tumor setting to fully activate DC. Induction of co-stimulatory signals is usually achieved through activation of pattern-recognition receptors (PRR), or c-type lectin signaling pathways18,19,20,21. A further approach for activating DC exploits their ability to take up antigens through interactions with surface Fcγ receptors (FcγR). Indeed, a number of important manuscripts have shown that injection of MoDC from BM precursors activated with tumor-IgG IC can prevent tumor growth in prophylactic settings, and can lead to the eradication of established tumors22,23.
In two recent papers, Carmi et al. discovered that in contrast to BMDC and spleen DC, MoDC from the blood and tumors cannot respond to IgG IC without additional stimuli. This was found to be due to the presence of high intracellular levels of tyrosine phosphatases regulating FcγR signaling24,25. By defining a critical checkpoint in DC, this work provided an important insight into the requirements for successful DC-based vaccination. The requirement for additional stimuli to enable FcγR signaling, and presumably signaling from other lectin receptors utilizing a similar phosphorylation cascade, thus underscores the need for avoiding the priming of DC during their isolation.
Therefore, the present protocol describes the isolation of MoDC from blood and tumors, which differ markedly from BM and spleen DC, and highlights precautions worth considering during the process.
The protocols below refer to the isolation of mouse MoDC, yet the overall principles may apply to other DC subsets cells, as well. 12 – 16-week-old C57Bl/6j mice were maintained in an American Association for the Accreditation of Laboratory Animal Care–accredited animal facility. All protocols were approved by Stanford University and Tel-Aviv University Institutional Animal Care and Use Committee.
1. Isolation of Tumor Associated Monocyte-Derived DC
2. Isolation of Monocyte-derived DC from Peripheral Blood
NOTE: Since mature MoDCs are relatively rare in mouse blood, the protocol below refers to their derivation in vitro from sorted monocytes.
3. Preparation of Tumor-IgG Immune Complexes
4. Activation of MoDC with Tumor-IgG IC
We initially compared the capacity of antibodies from naïve syngeneic and allogeneic mice to bind to tumor cells. To this end, B16F10 and LMP tumor cell lines were fixed in paraformaldehyde and washed extensively. B16F10 is a melanoma cell line, which was originally isolated from lung metastases in C57Bl/6 mice. LMP is a pancreatic tumor cell that was isolated from KrasG12D/+, LSL-Trp53R172H/+, and Pdx-1-Cre mice, and grows steadily in 129F1 mice. To obtain IC, tumor cells were incubated for 20 min on ice with 2 μg of syngeneic or allogeneic IgG per 1 x 105 tumor cells. IgG and IgM antibodies were isolated from the circulation of naïve 20-24-week-old mice on protein A, followed by size-exclusion chromatography as described24. Cells were then washed and stained with PE-conjugated rat anti-mouse IgG secondary antibody, and the mean fluorescent intensity was analyzed in a flow cytometer. As indicated in Figure 1A, IgG antibodies from C57Bl/6 allogeneic mice bound LMP tumor cells ex vivo far more effectively than antibodies from 129S1 syngeneic mice. Similarly, staining of fixed B16F10 with 129S1 allogeneic IgG was more than tenfold higher, compared to staining with syngeneic IgG from naïve C57Bl/6 mice. Interestingly, mice bearing B16 tumors failed to produce antibodies with the similar binding capacity to that of allogeneic mice, even during the tumor progression (Figure 1B).
We next sought to compare the IC response of MoDC from blood and tumors to that of DC from the spleen and BM. To isolate tumor-associated MoDC, B16F10 tumors were enzymatically dissociated to obtain single cell suspensions. Immune cells were enriched using CD45-magnetic beads, and MoDC were further sorted as SSClo/FSClo/CD11c+/MHCII+/Ly6Clo by FACS (Figure 2A). It is noteworthy that different tumor DC may have different markers that define them. To isolate MoDC from the blood, circulating monocytes were enriched by CD11b-conjugated magnetic beads and further sorted as SSClo/FSClo/ CD115+/MHCIIneg/lo. Cells were then cultured for 1 day in GM-CSF, and the non-adherent and loosely adherent cells were transferred into a new plate and cultured for an additional 4-5 days to obtain MoDC (Figure 2B). As a reference point we used BMDC, serving as the "gold standard" DC for many functional assays, as well as splenic MoDC, which reflect a more physiological DC subset. BMDC were obtained by sorting the BM pro-monocytes (CD11b+/Ly6Chi/CD115hi/MHCIIneg), followed by their culturing for 7 days with GM-CSF, as described24. Splenic MoDC were isolated from the single cell suspension, obtained by mashing the spleen through a cell strainer (pre-incubation with collagenase is not required for splenic MoDC), followed by enrichment with CD11b-conjugated magnetic beads. Cells were then sorted as SSClo/B220neg/NKp46neg/CD3neg/Gr1neg/F4/80neg/MHCIIhi/CD11chi and cultured for 1 hour in complete RPMI at 37oC to restore baseline activity.
To investigate the effect of IgG IC on MoDC activity, we incubated the isolated DC subsets overnight with fixed tumor cells or with fixed tumor cells pre-coated with allogeneic IgG. Activation of BMDC or splenic DC with IC resulted in increased CD86 and MHCII expression, unlike MoDC, or tumor-associated MoDC (Figure 3A). The ability to uptake CFSE-stained tumor-derived proteins, with or without allogeneic IgG, was also compared. As observed by confocal microscopy, splenic DC showed superior ability to internalize the tumor-derived proteins (Figure 3B).
Taken together, these results suggest that the tumor DC and MoDC respond differently than splenic DC and BMDC to activation with alloIgG-IC. Therefore, vaccination strategies that wish to activate tumor DC cannot be based solely on the activation patterns of BMDC.
Figure 1: Allogeneic mice have naturally-occurring tumor-binding IgG antibodies in their circulation: A. Mean fluorescence intensity (MFI) of LMP tumor cells stained with IgG antibodies isolated from the blood of syngeneic (129S1) and allogeneic (C57Bl/6) naïve mice. B. Mean fluorescence intensity (MFI) of B16F10 tumor cells incubated with IgG antibodies isolated from the circulation of syngeneic tumor-bearing mice (C57Bl/6), or allogeneic (129S1) naïve mice. This figure has been modified from extended data 2A and 2B Carmi Y et al. Nature 521 7550:99-104, 201524.
Figure 2: Sorting scheme of mouse MoDC from blood and B16 tumors. A. Isolation and sorting scheme of MoDC cultured from mouse monocytes. Confocal microscopy DIC image of inflammatory and patrolling monocytes after 4 days in culture (x400). B. Isolation and sorting scheme of tumor-associated MoDC from B16 tumors. Representative confocal microscopy DIC image of MoDC isolated from B16 tumors after overnight culture (x400). This figure has been modified from Supplementary figure 1 Carmi Y et al. JCI insight 1:18:e89020, 201625. Please click here to view a larger version of this figure.
Figure 3: MoDC from the spleen and BM display different patterns of activation than tumor and blood MoDC following incubation with IC. A. Flow cytometric analysis of MHCII and CD86 expression by DC incubated overnight with tumor-IgG IC. B. Confocal immunohistochemistry of tumor uptake (green) and MHCII expression (red) by DC incubated overnight with CFSE-labeled tumor IC. This figure has been modified from figures 3A and 3DCarmi Y et al. Akt and SHP-1 are DC-intrinsic checkpoints for tumor immunity. JCI insight 1:18:e89020, 201625. Please click here to view a larger version of this figure.
Given the large number of DC required for vaccinating mice (approximately 2-4 x 106 DC per one mouse), most of the vaccination strategies in mice are based on isolation of DC from BM and spleen followed by their ex vivo activation. However, attempts to activate tumor DC in vivo, using the same conditions for activating spleen and BM DC, have often been unsuccessful in producing effective immunity. In two subsequent publications, Carmi et al. have found that blood and tumor MoDC differ significantly from spleen and BM DC, given that they bear naturally high intrinsic levels of tyrosine phosphatase and require priming prior to activation with IC24,25. Moreover, these results further stress the need for taking extra precautions in order to maintain DC in an activation state that better reflects their physiological status. Therefore, the present protocol seeks to provide a detailed description of the isolation process of MoDC from tumors and the circulation, and to highlight the steps that may result in their premature activation.
First, in order to obtain sufficient numbers of DC from blood and tumors, there is a need for a much greater number of mice compared to that of protocols for isolating BMDC. To increase the yield of the DC we use 16-week-old mice, which have larger blood volumes and overall cell counts. We routinely get 5-8 x 104 MoDC per 1 mL of blood without injection of GM-CSF, and 4-5 x 105 MoDC following injection. For tumor DC, we obtain approximately 6-8 x 104 MoDC from a 100 mm3 tumor, though the number may vary between individual mice and between tumor models. Increasing tumor size results in a lower DC yield, as a 1,000 mm3 tumor will have only about 5,000 DC. Instead, we inject mice with tumor cells at multiple sites (typically 4-6), thereby obtaining up to 4 x 105 MoDC per mouse.
Furthermore, we recommend that prior to their experimental use, antibodies, cell lines, tubes and general lab reagents should be tested for endotoxins by LAL assay, and by plating the media on AC bacterial agar. Tumor cell lines are often infected with Mycoplasma and should therefore be tested by PCR prior to their incubation with DC. Use of crude collagenase preparations, such as Types 1 and 4, is a primary source of endotoxins due to isolation of the collagenases from Clostridium histolyticum. Standard preparations of collagenase may contain as much as 10 EU/mg of endotoxins, which is about 1-2 ng of endotoxin per mL of digestion mixture. Jahr et al. have found that collagenase preparations containing 2.7-6.7 ng/mL of endotoxins will induce 1,415-3,967 ng/mL of IL-1β following culture with PBMC29. Indeed, priming of DC is routinely done with 1 ng of LPS per mL media, yet as little as 100 picograms/mL is sufficient to prime them30,31,32. Another potential source of endotoxins is FBS, which may contain 25 EU/mL or more of endotoxins, or less than 10 EU/mL. Assuming that the culture media contains 10% FBS, the endotoxin load could reach 0.5 ng/mL, which is sufficient to prime DC.
In addition, many protocols use anti-CD16/32 antibodies to block Fc receptors as a means of increasing the specificity of their antibody panel prior to sorting. Nonetheless, cross-linking of FcγRIII (CD16) leads to phosphorylation of Syk/ZAP-70 and PLC-γ, release of intracellular Ca2+ 33,and phosphorylation of P38 and ERK1/2in both human34 and mouse monocytes35. In our hands, addition of anti-CD16/32 to MoDC cultures consistently induces a strong phosphorylation of P38 and ERK1/2 kinases in DC within one minute of exposure. We, therefore, strongly suggest avoiding the use of anti-CD16/32 when isolating MoDC for in vitro functional assays.
Another common protocol uses enrichment of DC by CD11c magnetic beads prior to their sorting by FACS. CD11c is a type I transmembrane glycoprotein that recognizes a variety of ligands, including fibrinogen, LPS, type I collagen,and the inactivated C3b subunit. Rezzonico et al. have shown that ligation of CD11c with antibodies induces potent NFκB activation and secretion of chemokines36. In our experience, immobilized anti-CD11c antibodies induce cell activation and apoptosis, while the use of their soluble form in FACS sorting does not induce phosphorylation of either P38 or ERK1/2.
Overall, this protocol is designed to achieve isolation of MoDC with as little activation as possible, so that their subsequent activation in vitro reflects the conditions required for their activation in vivo.
The authors have nothing to disclose.
None
Ficoll-Paque PREMIUM | GE-Healthcare | 17-5442-02 | |
OptiPrep | StemCell Technologies | 07820 | |
CD45 MicroBeads | Miltenyi | 130-052-301 | |
EasySep Monocyte Isolation Kit | StemCell Technologies | 19861 | |
Collagenase IV | Sigma | C9697-50MG | Test each lot for endotoxin |
DNase I | Sigma | DN25-10MG | |
HBSS | ThermoFisher | 14025092 | |
FBS | ThermoFisher | 16140071 | Test each lot for endotoxin |
PE-CD11c | Biolegend | 117307 | |
APC-CD11b | Biolegend | 101211 | |
Brilliant Violet 650 MHCII | Biolegend | 107641 | |
AF48- CD86 | Biolegend | 105017 | |
APC/Cy7-Ly-C6 | Biolegend | 108423 | |
PE/Cy7-CD15 | Biolegend | 135523 |