Heat shock proteins (HSPs) play a regulatory role for maturation of antigen-presenting cells (APCs) such as dendritic cells (DCs) and macrophages. Whereas HSP70 has been shown to enhance the maturation of human DCs via a nuclear factor kappa-B (NF-κB)-dependent pathway, the regulatory role of calreticulin (CRT), which is a HSP with similar functions to HSP70, is not well studied. To investigate the role of CRT as adjuvant in cell activation and co-stimulatory responses we determined the effects of CRT on human APC maturation in comparison to that of HSP70. To facilitate eukaryotic endotoxin-free CRT protein expression, three different methods were compared. We demonstrate that CRT induces the maturation of human DCs and increases the production of proinflammatory cytokines via the NF-κB pathway. CRT-mediated maturation was qualitatively similar to that induced by HSP70. Interestingly, priming of monocytes with HSPs showed an even more prominent effect on maturation than exposure of immature DCs to these compounds. A higher expression of CD86, CD83 and CCR7 on mature DCs were found in response to CRT. Our data provide novel insights into the role of extracellular HSPs as chaperokines in the processes of APC generation and may thus be useful to improve adoptive immunotherapy.
Heat shock proteins (HSPs) are highly conserved molecular chaperones that work either passively, by preventing the aggregation of damaged proteins, or actively via adenosine triphosphate (ATP)-driven conformational folding of target proteins . They are synthesized in response to stress conditions such as heat, hypoxia or viral transformation . HSPs are also involved in the stimulation of adaptive and innate immune responses after their cellular release [3–6]. HSPs such as HSP27, 60, 70, 90 and 110, glucose-regulated proteins (GRP) 78, 94 and 170 and calreticulin (CRT) are released from cells in a variety of circumstances in the context of either apoptosis/necrosis or induced secretion, after which they may enter the bloodstream [3,4,6,7]. Once extracellular they perform various functions, such as interacting with antigen-presenting cells (APCs), stimulating APCs to secrete inflammatory cytokines or chemokines, and inducing the maturation of dendritic cells (DCs) [8,9]. The ability to activate DCs has been proposed as an underlying basis for their role as endogenous danger signals . Furthermore, recent reports have indicated that HSP-dependent activation of the innate immune system may be sufficient to elicit tumour rejection [11,12].
DCs are the major cell population of professional APCs and co-ordinate antigen-specific adaptive immunity and tolerance [13,14]. They direct the quality and extent of immune responses. The ability of DCs to activate T lymphocytes depends on their maturation status. Immature DCs capture and process antigens with high efficiency, whereas fully mature DCs have a high surface expression of major histocompatibility complex (MHC) class II and co-stimulatory molecules and are therefore potent APCs that stimulate proliferation and differentiation of CD4+ and CD8+ T cells . It has been shown that extracellular HSP binding to surface receptors such as CD40 , natural killer group 2D (NKG2D) , CD14 , CD91 , LOX-1 , C-C chemokine receptor 5 (CCR5) and Toll-like receptors 2 and 4 (TLR-2/4) [21,22] as well as HSP-induced signalling events are involved in the maturation of DCs. This binding is characterized by the up-regulation of several co-stimulatory molecules, such as CD86, CD40 and MHC class II molecules, and the secretion of cytokines interleukin (IL)-1, IL-12, tumour necrosis factor (TNF)-α and granulocyte-macrophage colony-stimulating factor (GM-CSF) . Recombinant human HSP60, human inducible HSP70/HSP72 and human gp96 preparations were found to have an impact on DC maturation . HSP70 was shown to activate DCs via the TLR-4 pathway and to induce DC maturation, as evidenced by increased expression of CD40, CD86 and CD83 surface markers [9,11].
The mechanism by which human CRT acts on human DCs is widely unknown. Interestingly, Bak et al. reported the failure of murine CRT to induce DC maturation and activation as determined by their inability to up-regulate MHC class II molecules and CD86 . In this study, we assessed the potential of endotoxin-free human CRT and HSP70 to induce nuclear factor kappa-B (NF-κB)-regulated maturation of monocyte-derived human DCs compared to standard techniques using a proinflammatory cytokine cocktail [26–28] or monocyte-conditioned medium (MCM) (Fig. 1) [29,30].
We hypothesize that CRT can be used to generate mature DCs and that the interaction with the cells is similar to that shown for HSP70. Three different methods to express recombinant endotoxin-free CRT to be secreted in the supernatant of HEK293 cell were established and compared: (i) the complete, unmodified CRT coding sequence was cloned (CRT); (ii) the C-terminal endoplasmic reticulum (ER)-retrieval KDEL amino acid sequence was mutated into KDQL in order to disturb the endoplasmic retention and support the protein secretion (CRT_KDQL)  and (iii) a shRNA was designed to knock down the expression of aminoacyl-tRNA synthetase-interacting multi-functional protein-1 (AIMP-1), which is known to regulate protein retention in the ER (CRT_shAIMP-1) .
In contrast to the findings in mice, we found that CRT can induce the maturation of human DCs. In particular, priming of monocytes with the HSPs enhanced the expression of MHC class II and co-stimulatory molecules (CD83, CD86) on mature DCs. Furthermore, we found a strong increase in surface expression of C-C chemokine receptor 7 (CCR7) on DCs generated under the influence of the HSPs. Our findings highlight the role of extracellular HSPs as cytokine-like proteins (chaperokines) in the processes of antigen presentation and immune stimulation. Because CRT and HSP70 can act as mediators of in vitro DC maturation they have potential usefulness in T-cell stimulation and expansion protocols, as well as in the development of HSP-based vaccination protocols.
Methods and materials
Generation of pcDNA3·1/CRT and pcDNA3·1/CRT_KDQL constructs
RNA was isolated from human embryonic kidney cell line HEK293 (RNeasy Mini Kit; Qiagen, Hilden, Germany) and amplified by reverse transcriptase–polymerase chain reaction (RT–PCR) (OneStep RT-PCR Kit; Qiagen) with the following primers: CRT-S (5′-GAG ATG CTG CTA TCC GTG CCG CT-3′) and CRT_WAS (5′-CAG CTC GTC CTT GGC CTG-3′). The resulting PCR product was cloned into pcDNA3·1V5/His (Invitrogen, Karlsruhe, Germany). As the anti-sense primer (CRT-WAS) did not contain a stop codon, the cloned sequence was followed by the vector sequence for V5/6×His tag.
The resulting pcDNA3·1/CRT construct was used as template to mutate the C-terminal tetrapeptide Lys-Asp-Glu-Leu (KDEL) into KDQL by replacing Glu with Gln (QuikChange® XL Site-Directed Mutagenesis Kit; Stratagene, La Jolla, CA, USA). The following primers were used: CRT_KDQL_001 (5′-GGC CAG GCC AAG GAC CAG CTG AAG GGC AAT TCT G-3′) and CRT_KDQL_002 (5′-CAG AAT TGC CCT TCA GCT GGT CCT TGG CCT GGC C-3′). Vector purification (pcDNA3·1/CRT, pcDNA3·1/CRT_KDQL) was performed using the EndoFree Maxi Plasmid Kit (Qiagen).
Establishment of AIMP-1-silenced HEK293 cells (HEK293_shAIMP-1)
The expression of the AIMP-1, which is known to regulate protein retention of CRT in the ER , was down-regulated using RNA interference (RNAi) technology. The short-hairpin RNA (shRNA) sequences were designed using the web-based siRNA Target Designer (https://rnaidesigner.invitrogen.com/rnaiexpress). Three different shRNA expression cassettes (Table 1) were cloned into the pLVTHm/si vector (Addgene, Cambridge, MA, USA). Lentiviral particles targeting the expression of AIMP-1 were produced as described previously . Silencing effect was verified by real-time RT–PCR.
Table 1. shRNA sequences.
Expression of recombinant CRT in HEK293 and HEK293_shAIMP-1 cells
Using Amaxa Cell Line Nucleofector Kit V (Amaxa, Cologne, Germany), the pcDNA3·1/CRT and pcDNA3·1/CRT_KDQL constructs were transfected into HEK293 cells and, additionally, the pcDNA3·1/CRT construct was transfected into those AIMP-1-silenced HEK293 cells found to be most effectively silenced (shAIMP-1_3, Fig. 2a).
Selection of clones resistant to geneticin (G418, 1000 µg/ml; Invitrogen) was performed 48 h post-transfection and geneticin-resistant clones were subcloned by further limiting dilution. CRT expression in the supernatant of transfected HEK293 (CRT, CRT_KDQL) and HEK293_shAIMP-1 (CRT_shAIMP-1) cells were quantified using a sandwich V5/HIS enzyme-linked immunosorbent assay (ELISA) as described previously . CRT RNA expression levels were measured by real-time RT–PCR using a specific primer/probe combination to detect the V5/6× His sequence of the transfected genes.
In order to identify the best CRT expression strategy 5 × 105/ml HEK293_CRT, HEK293_CRT_KDQL and HEK293_CRT_shAIMP-1 cells, respectively, were seeded into each well of a 24-well plate in the appropriate medium. Untransfected HEK293 cells were used as control. After 3 days, cells and supernatants were harvested and analysed by real-time RT–PCR for mRNA expression and by V5/HIS ELISA for soluble CRT protein secretion.
Real-time RT–PCR to evaluate AIMP-1 silencing and to detect CRT mRNA levels
Silencing of AIMP-1 in HEK293 cells was verified using the primers RT_AIMP1_S 5′-TCC TGC TGT GGC TGT CTC G-3′ and RT_AIMP1_AS 5′-GCT TCA TGA TTT TCT GCC GT-3′ and the MGB-TaqMan probe for AIMP-1 (5′-ACC CGT GGT CCT C-3′). A specific primer/probe combination was used to detect the V5/6 × His sequence of the CRT transcripts (CRT, CRT_KDQL, CRT_shAIMP-1): RT_V5_001 (5′-AGA GGG CCC GCG GTT-3′) and RT_V5_002 (5′-TCG AGA CCG AGG AGA GGG TT-3′). The sequence of the MGB-TaqMan probe specific for the V5/6 × His tag was 5′-AAG GTA AGC CTA TCC C-3′. Untransfected HEK293 cells were used as negative controls. β-actin mRNA was amplified as the reference standard of mRNA levels .
Immobilized metal affinity chromatography (IMAC) purification of CRT from cell culture supernatant
The HEK293_CRT cells selected according to the protein expression results (Fig. 2) were used for large-scale production using the CELLSPIN cultivation system (Integra Biosciences, Chur, Switzerland). Recombinant V5/His-tagged CRT was purified as described for recombinant HSP70  and quality-controlled by Western blot using detection antibodies against CRT (Stressgen, Ann Arbor, MI, USA), V5 and 6× His tag (both from Invitrogen). The V5/His ELISA was used for quantification of the eluted protein.
The limulus amebocyte lysate assay (LAL) was used to exclude endotoxin contamination of the purified protein (Rapid Endo-Test; Lonza, Verviers, Belgium, sensitivity 0·005 EU/ml). In order to confirm cleavage of the CRT signal peptide from mature CRT Edman sequencing was performed to determine the first 6 N-terminal expressed amino acids of the protein (Toplab, Martinsried, Germany).
NF-κB staining of stimulated monocytes
A flow cytometric assay was used to study translocation of NF-κB from the cytoplasm to the nuclei of the monocytes after CRT stimulation . Therefore, peripheral blood was collected with informed consent from six healthy donors, as approved by the ethical committee of the Hannover Medical School. Peripheral blood mononuclear cells (PBMCs) from blood samples of those donors were obtained by discontinuous gradient centrifugation followed by isolation of monocytes using a negative selection kit (Miltenyi, Bergisch Gladbach, Germany). After stimulating the monocytes with lipopolysaccharide (LPS) (2 µg/ml), CRT (10 µg/ml) and RPMI-1640 (Lonza) for 10, 20, 30, 60 and 90 min, intact nuclei were isolated using the CycleTest PLUS DNA Reagent Kit (BD Biosciences, Heidelberg, Germany). 1 µg per 1 × 106 cells of fluorescein isothiocyanate (FITC)-conjugated mouse anti-human NF-κB (p65) monoclonal antibody (mAB, clone sc 8008, Santa Cruz Biotechnology, Heidelberg, Germany) was added and allowed to react for 20 min at room temperature. Cold propidium iodide solution was added and the preparations were incubated for another 10 min. Acquisition of stained nuclei was carried out on a FACSCanto A (BD Biosciences). The singlet population was analysed for FITC staining using FACSDiva software version 6·1·2. At least 20 000 events were acquired for each analysis.
Immunofluorescence staining of monocytes with CRT-FITC
Three milligrams of purified CRT were conjugated with FITC (Fluoro tag FITC conjugation kit; Sigma-Aldrich, Hamburg, Germany). Monocytes (1 × 106) obtained from three different donors were resuspended separately in RPMI/2% AB serum (C.C.pro, Neustadt, Germany) and incubated for 4 h with 10 µg/ml CRT-FITC in 500 µl medium (37°C, 5% CO2) using four-well chamber slides (Sigma-Aldrich). After incubation cells were washed carefully with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde (PFA). Monocytes were permeabilized by 0·1 % Triton X-100 (Sigma) and stained for 5 min with 10 µg/ml diamidino-2-phenylindole (DAPI; Invitrogen). After washing with PBS the slides were mounted in Gel/Mount mounting medium (GeneTex, Irvine, CA, USA). The cells were placed under constant illumination on the Olympus-IX81 microscope (Olympus, Center Valley, PA, USA) using a 40× objective. Images were acquired using a CCD camera (Olympus) and analysed using Olympus cellIM and cellIR image version 3·0 software (Olympus).
Maturation of DCs under the influence of CRT and HSP70
Procedure A. To investigate the effects of human CRT and HSP70 on DC maturation we adapted the method for generating mature DCs (mDCs) described by Thurner et al. (Fig. 1, Supplementary Table S1) . On day 0, 5 × 107 PBMCs suspended in RPMI/2%AB serum were incubated overnight in 10 ml medium on an 85-mm Petri dish pretreated with 30 µg/ml Sandoglobulin (Novartis, Nuernberg, Germany) (Fig. 1a). After overnight incubation, the supernatant – now referred to as monocyte-conditioned medium (MCM) – was removed and stored at −20°C for further use. The adherent cells were treated with 10 ml RPMI/2%AB serum supplemented with 500 U/ml IL-4 and 800 U/ml GM-CSF (PeproTech, Hamburg, Germany). Five ml fresh GM-CSF/IL-4 solution was added on day 4. On day 5, supernatant samples containing immature DCs (immDC) were removed, centrifuged and the separated immDCs were resuspended in 1 × 106/ml media (RPMI/2% AB serum, IL-4, GM-CSF). The phenotype of immDCs was analysed by flow cytometry using the following mAbs: phycoerythrin (PE)-conjugated CD83, peridinin chlorophyll protein (PerCP)-conjugated CD14, PE-cyanin5 (PE-Cy5)-conjugated CD54, allophycocyanin (APC)-conjugated CD86 and CD40, APC-cyanin7 (APC-Cy7)-conjugated human leucocyte antigen D-related (HLA-DR) and PE-cyanin7 (PE-Cy7)-conjugated CCR5 and CCR7 (all from BD Bioscience). All samples were analysed on a flow cytometer (FACSCanto A; BD Biosciences). At least 25 000 events within the live gate were acquired for each analysis. Gates were set based upon the scatter properties of DCs. For analysis, cells were gated on either CD4+ or CD8+ T cells. The supernatant was harvested and stored at −20°C for further cytokine expression analysis.
On day 6, maturation was initiated by the following stimuli: (a) MCM; (b) cytokine maturation mix (CytoMix: 10 ng/ml TNF-α, 1000 U/ml IL-6, 10 ng/ml IL-1β (all from PeproTech), 1 U/ml prostaglandin E2 (Sigma-Aldrich, Munich, Germany); (c) 10 µg/ml CRT (0·19 µmol/l); and (d) 10 µg/ml HSP70 (0·13 µmol/l). The mature DCs resulting therefrom were classified as mDC_MCM, mDC_CytoMix, mDC_CRT and mDC_HSP70, respectively (Fig. 1a, Table S1). On days 7 and 8, the maturation status of DCs was analysed using the above-mentioned antibodies, and the supernatants were stored at −20°C for further analysis.
Procedure B. The effect of HSP-conditioned media (HSP-CM) generated by incubating monocytes in the presence of (e) CRT or (f) HSP70, respectively, was evaluated (Fig. 1b, Supplementary Table S1). Therefore, on day 0 monocytes were primed with either 10 µg/ml CRT or 10 µg/ml HSP70, and after overnight incubation CRT-conditioned medium (CRT-CM) or HSP70-conditioned medium (HSP70-CM) was removed and stored at −20°C for further use and cytokine analysis. The HSP-primed monocytes were cultured and matured as described above. On day 6, HSP-CM-primed immature DCs (pimmDC_CRT, pimmDC_HSP70) were stimulated with CRT-CM or HSP70-CM to produce the corresponding primed mature DCs (pmDC_CRT-CM, pmDC_HSP70-CM).
Multiple cytokine detection using Luminex technology and evaluation of IL-12 secretion by ELISA
A multiplex bead-based Luminex® assay (Cytokine Human 10-Plex Panel; Invitrogen) that quantifies multiple cytokines in single-sample supernatants was used to analyse cytokine patterns in the supernatants of cells (monocytes, immature and mature DCs) and in the respective conditioned media (MCM, CRT-CM and HSP70-CM). Samples and cytokine standards supplied by the manufacturer were run on each plate. Cytokine detection data were acquired on a Luminex-200 system (Invitrogen).
IL-12p40 and IL-12p70 levels in the supernatants of the monocytes, immature DCs and mature DCs were measured using an ELISA system according to the manufacturer's instructions (BD Biosciences).
Statistical analyses were performed using the non-parametric Mann–Whitney U-test run on GRAPHPAD PRISM version 5·02 software (GraphPad Software, San Diego, CA, USA). Levels of significance are expressed as P-values (*P < 0·05; **P < 0·01; ***P < 0·001).
Significant differences between CRT, CRT_KDQL and CRT_shAIMP-1 expression
We established and compared three different methods to express recombinant endotoxin-free CRT: (i) the unmodified CRT coding sequence was expressed (CRT); (ii) the endoplasmic reticulum (ER)-retrieval sequence was mutated into KDQL (CRT_KDQL); and (iii) AIMP-1-specific shRNAs (Table 1) were constructed to knock down the expression of AIMP-1 (CRT_shAIMP-1).
One week after transduction, mRNA levels of AIMP-1 expression were 40–88% lower than those in non-transduced controls as determined by real-time RT–PCR (Fig. 2a). The shAIMP-1_3 construct exhibiting the highest level of AIMP-1 reduction (88%) and was used for expression of CRT. The degree of AIMP-1 silencing in the cells remained stable during the entire culture period (data not shown).
Real-time RT–PCR was performed to measure the effects of the different cloning and expression strategies for CRT, CRT_KDQL and CRT_shAIMP-1 (Fig. 2b). CRT mRNA levels in HEK293_CRT transfected cells were 12 678 times higher than in untransfected controls. Significant differences between the expression of CRT versus CRT_shAIMP-1 (mean: 4078-fold increase) and between CRT_KDQL (mean: 10 359-fold increase) versus CRT_shAIMP-1 were observed.
In the respective supernatants CRT protein levels were determined by V5/HIS ELISA (Fig. 2c). As expected, based on mRNA level, CRT_shAIMP-1 yielded the lowest protein secretion level (16·1 µg/ml). Unexpectedly, the protein level of CRT_KDQL (34·51 µg/ml) was significantly higher than for CRT (28·6 µg/ml), showing the importance of the KDEL sequence for exporting the protein out of the ER. Although, theoretically, the KDEL mutation should not interfere with the protein's tasks, we decided to use unmodified CRT in further experiments.
Western blotting from sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) demonstrated the purity of the isolated recombinant CRT. Endotoxin levels in the HSP70 and CRT preparations were determined by LAL assay and were found to be below 0·2 EU/ml (HSP70: 0·15 EU/ml; CRT: 0·09 EU/ml), which has been referred to as endotoxin-free .
CRT signals via NF-κB in human monocytes
As HSP70 is known to activate the NF-κB pathway [16,38], we determined the effect of CRT on nuclear translocation of the p65 subunit of NF-κB in monocytes . The highest percentage (mean 63·4%) of positive NF-κB staining [mean fluorescence intensity (MFI): 5145] compared to RPMI-1640 medium (mean 26·7%, MFI: 3234) was reached after 30 min of stimulation (Fig. 3a).
Immunofluorescence staining of monocytes with CRT-FITC
Figure 3b shows the cellular uptake of CRT-FITC by monocytes and the accumulation of HSP near the nuclei of the cells (Fig. 3b). These results indicate that CRT can interact with monocytes and additionally that the uptake of the protein occurred most probably by an endocytosis process mediated by CRT receptors.
Effects of recombinant CRT and HSP70 on DC maturation (Procedure A)
The DC maturation protocols we used first are an adaptation of the method for generating DCs described by Thurner et al. (Fig. 1a) . Phenotype determinations (CD83, CD14, CD86, HLA-DR, CCR7, CD54, CD40 and CCR5) were performed in immDCs harvested on day 5 (Fig. 4, Table 2) and mDCs (day 7 and 8) generated with CRT (mDC_CRT) or HSP70 (mDC_HSP70) versus those generated using cytokine mixture (mDC_CytoMix)  or monocyte-conditioned medium (mDC_MCM) .
Table 2. Expression of surface markers on immature and mature dendritic cells (DCs).
Classically generated immature DCs were stimulated with different combinations of heat shock proteins (HSPs) and conditioned media and surface markers human leucocyte antigen D-related (HLA-DR), CD54, CD40, C-C chemokine receptor (CCR)5 and CCR7 were measured on days 7 and 8. Additionally, monocytes were primed with calreticulin (CRT) [primed immature DC (pimmDC)_CRT] or HSP70 (pimmDC_HSP70), and the HSP-conditioned media [CRT-conditioned medium (CM), HSP7-CM] were used to generate pmDC_CRT-CM and primed mature DC (pmDC)_HSP70-CM. Results are expressed as mean ± standard deviation of six independent experiments. MFI, mean fluorescence intensity; MCM: monocyte-conditioned medium.
65 203·8 ± 41 536·2
38 807·4 ± 22 771·8
27 747·8 ± 10 831·1
34 240·4 ± 21 972·1
36 418·4 ± 20 830·0
62 728·4 ± 48 625·3
32 180·2 ± 25 769·1
18 020·8 ± 11 340·3
41 317·3 ± 21 067·2
443 94·7 ± 17 647·1
53 974·4 ± 45 391·8
46 389·2 ± 18 840·0
31 796·5 ± 10 000·6
35 387·6 ± 14 962·0
37 435·7 ± 13 341·2
143 253·9 ± 38 197·3
141 955·8 ± 25 163·1
109 036·8 ± 20 066·4
46 137·1 ± 11 715·3
20 366·6 ± 6 012·7
134 373·3 ± 36 213·2
132 183·9 ± 18 277·0
100 737·9 ± 17 930·7
43 028·4 ± 10 283·9
23 508·0 ± 72 41·9
128 508·4 ± 35 198·1
129 827·4 ± 19 312·4
106 284·6 ± 11 380·5
41 509·0 ± 11 064·3
19 874·7 ± 4 646·8
133 884·3 ± 33 271·5
140 604·3 ± 23 836·9
117 788·3 ± 12 882·2
46 494·8 ± 13 356·5
20 384·7 ± 4 937·0
168 389·7 ± 37 002·5
100 594·6 ± 46 377·0
82 978·8 ± 24 178·7
93 257·2 ± 38 403·1
70 734·6 ± 22 390·5
136 441·2 ± 44 799·7
89 675·5 ± 50 118·1
75 495·2 ± 32 840·3
84 904·6 ± 40 953·9
75 669·9 ± 37 126·8
135 875·1 ± 39 121·5
110 762·5 ± 52 653·0
73 860·7 ± 30 820·6
86 168·9 ± 44 399·7
111 934·4 ± 21 902·3
146 664·3 ± 39 682·6
139 982·2 ± 38 641·0
83 286·9 ± 27 695·7
105 162·1 ± 44 283·2
95 833·7 ± 27 965·3
140 329·5 ± 65 238·5
79 631·0 ± 27 437·7
77 163·1 ± 25 353·4
89 094·5 ± 28 086·4
55 292·4 ± 31 809·3
140 927·8 ± 57 017·8
86 141·2 ± 32 697·1
83 575·5 ± 26 590·9
93 418·2 ± 31 535·7
56 491·6 ± 26 272·5
The increase in CD83 and CD86 expression and decrease in CD14 expression levels relative to those in immDCs showed that both CRT and HSP70 are able to induce DC maturation (Fig. 4). On day 7 the most significant decrease in the population of CD14+ cells (Fig. 4a) from immDCs (2·5%) was observed on mDC_CytoMix (1·0%), followed by mDC_MCM (1·3%), which showed slightly lower levels than mDC_CRT (1·5%) and mDC_HSP70 (1·4%). The co-stimulatory marker CD86 (Fig. 4b) was up-regulated after stimulation with CRT (mDC_CRT, 72·4%) and HSP70 (mDC_HSP70, 70·3%) and reached the levels on mDC_MCM (78·3%) and mDC_CytoMix (78·5%). Compared to immDCs showing 7·8% CD83+ cells (Fig. 4c,d), expression of CD83 on mDC_CRT (23·1%) and mDC_HSP70 (23·6%) increased to the same level observed in mDC_MCM (27·5%). The highest enhancement of CD83 expression was assessed on mDC_CytoMix (56·3%). CRT and HSP70 acted like CytoMix and MCM with regard to HLA-DR, CD54, CD40, CCR5 and CCR7 expression (Table 2).
On day 8, the percentage of CD14+ cells in mDC_CRT (0·8%) and mDC_HSP70 (0·5%) were significantly lower compared to immDCs (2·5%) (Fig. 4a). CD86 expression on DCs after stimulation with CRT (mDC_CRT, 76·5%) and HSP70 (mDC_HSP70, 80·2%) reached the levels on mDC_CytoMix (83·9%) (Fig. 4b). Additionally, a significant increase in CD83 expression was observed in mDC_MCM (40·6%), mDC_CRT (38·9%) and mDC_HSP70 (49·7%) (Fig. 4c,d). Compared to day 7, a slight increase in HLA-DR and a strong increase in CCR5 and CCR7 expression on mDC_CRT and mDC_HSP70 was observed (Table 2). The MFI for CCR5 doubled and that for CCR7 tripled in mDC_CytoMix and mDC_MCM. Interestingly, a four- to sixfold increase was observed in mDC_HSP70 and mDC_CRT. In all samples, MFI levels for CD40 decreased, whereas those for CD54 on mDC_CRT and mDC_HSP70 remained unaffected.
Taken together, the expression levels of maturation markers in DCs indicated that more mature DCs were found on day 8; therefore, we used cells at day 8 in all further experiments. Additionally, higher concentrations of HSP (50 µg/ml) caused an increased CD83 and CD86 expression and decreased amount of CD14+ cells in the cultures. Although this dose-dependent maturation effect was reflected by increased levels of HLA-DR, CD54, CD40, CCR5 and CCR7, it did not reach the level of significance (data not shown).
Use of HSP-conditioned medium showed a more prominent effect on DC maturation (Procedure B)
To evaluate the effects of HSP70 and CRT in the early stages of DC generation, monocytes were primed with 10 µg/ml of the HSPs (Fig. 1b). After overnight stimulation, CRT-CM and HSP70-CM were harvested and compared with MCM in terms of cytokine composition (Fig. 5). The media were used to stimulate the corresponding primed immDCs (pimmDC_CRT, pimmDC_HSP70).
On day 5, there were no differences in the expression of CD83, CD86 and other surface markers on pimmDC_CRT and pimmDC_HSP70 compared to immDC; CD14 was the only exception (Fig. 4, Table 2). Classically generated immDCs had 2·5% CD14+ cells compared to only 0·7% in pimmDC_CRT and pimmDC_HSP70 (Fig. 4a). On day 8, pmDC_CRT-CM and pmDC_HSP70-CM still had the lowest percentage of CD14+ cells (0·4% pmDC_CRT-CM, 0·8% pmDC_HSP70-CM). In pmDC_CRT-CM (89·5%), and pmDC_HSP70-CM (84·4%), CD86 expression was slightly higher than those in mDC_MCM (79·1%) (Fig. 4b), but strong differences in CD83 expression were observed (Fig. 4c and d). The lowest CD83 expression was found in mDC_MCM (40·6%) and could even be outperformed by pmDC_CRT-CM (63·8%) and pmDC_HSP70-CM (59·4%). Interestingly, HLA-DR levels (MFI) for pimmDC_CRT (62 728·4) and pimmDC_HSP70 (53 974·4) were lower than those in immDCs (65 203·8, Table 2); they increased drastically following cell maturation after treatment with the respective conditioned media (pmDC_CRT-CM: 140 329·5; pmDC_HSP70-CM: 140 927·8). The highest levels of MFIs for CD54 and CD40 were detected on pimmDC_HSP70 (86 141·2 and 83 575·5), and the highest for CCR5 and CCR7 were observed on pimmDC_CRT (89 094·5 and 55 292·4). In conclusion, we found that priming monocytes with HSPs for 24 h and using the HSP-CM for maturation yielded the best results in terms of the expression of DC maturation and co-stimulatory markers. Interestingly the results showed that the CRT-mediated maturation was qualitatively higher to that induced by HSP70.
Cytokine composition of MCM, CRT-CM and HSP70-CM
We used Luminex technology to assess the concentrations of GM-CSF, interferon (IFN)-γ, IL-10, IL-1β, IL-2, IL-5, IL-6, IL-8 and TNF-α in the monocyte- and HSP-conditioned media (Fig. 5). HSP-conditioned media was harvested after stimulation of monocytes with either 10 µg/ml CRT (CRT-CM) or 10 µg/ml HSP70 (HSP70-CM) for 24 h. Interestingly, all media contained high amounts of IL-4 and GM-CSF, which are normally used to generate immDCs. Compared to MCM, HSP-CM contained higher levels of IL-4 and lower levels of GM-CSF. HSP-CM also contained ample quantities of TNF-α, IL-6 and IL-1β, the main components of the cytokine mixture used commonly to maturate DCs. CRT-CM contained far lower concentrations of these cytokines than HSP70-CM, but the lowest concentrations were found in MCM. None of the analysed media contained IFN-γ, IL-2 or IL-5 (data not shown).
Cytokines detected in the supernatants of immature and mature DCs
The cytokine content of immDC, HSP-primed immDC (pimmDC_CRT, pimmDC_HSP70) and mature DC supernatants were assessed by Luminex technology (Table 3). All mature DCs generated with HSP-CM (pmDC_CRT-CM, pmDC_HSP70-CM) expressed higher levels of GM-CSF, TNF-α, IL-1β and IL-8 than mDCs generated with MCM (mDC_MCM) or CytoMix (mDC_CytoMix). Consistent with the observed high levels of IL-4, GM-CSF, IL-6, TNF-α and IL-1β in HSP70-CM (Fig. 5), high expression levels of these cytokines were also observed in pimmDC_HSP70 and pmDC_HSP70-CM supernatants, respectively.
Table 3. Secretion of cytokines by monocytes, immature dendritic cells (DCs) and mature DCs stimulated with CRT, HSP70, CytoMix and different conditioned media.
The capacity of immature and mature DC populations from six healthy donors to secrete different cytokines was assessed by determining the secretion levels of proinflammatory cytokines using a multiplex bead-based Luminex assay (pg/ml). Results are expressed as means of six independent experiments. IL: interleukin; GM-CSF: granulocyte-macrophage colony-stimulating factor; TNF: tumour necrosis factor; HSP: heat shock protein; CM: conditioned medium; pimmDC: primed immature DC; pmDC: primed mature DC; mDC: mature DC; MCM: monocyte-conditioned medium; CRT: calreticulin.
23 461·8 ± 5 380·4
18 859·9 ± 5 604·9
0·0 ± 0·0
11·5 ± 5·6
7·8 ± 7·6
4·2 ± 2·0
155·0 ± 125·9
28 140·9 ± 19 791·5
16 755·1 ± 6 569·1
0·1 ± 0.1
101·9 ± 204·9
10·6 ± 20·3
4·5 ± 1·3
124·7 ± 75·8
34 389·8 ± 22 959·7
21 417·8 ± 6 104·6
2·6 ± 4·6
737·5 ± 1 112·3
37·4 ± 16·9
6·8 ± 3·3
1 776·5 ± 1 555·6
20 844·6 ± 11 398·5
9 049·5 ± 4 032·6
8·3 ± 5·5
11 224·0 ± 5 401·8
327·4 ± 185·3
11·3 ± 6·3
7 097·3 ± 5 467·7
18 763·5 ± 8 903·1
14 791·8 ± 5 473·3
12·3 ± 11·1
115·4 ± 85·8
33·2 ± 12·9
16·7 ± 8·5
17 085·3 ± 4 680·4
23 933·8 ± 5 579·7
11 935·1 ± 6 463·2
8·1 ± 5·3
103·9 ± 55·1
13·0 ± 7·6
10·8 ± 3·5
10 442·4 ± 3 832·7
23 897·2 ± 10 180·5
11 584·3 ± 6 796·8
20·1 ± 10·3
2 813·2 ± 2 642·0
32·0 ± 12·1
30·1 ± 16·9
34 692·6 ± 19 315·2
16 964·7 ± 10 056·0
16 423·8 ± 3 099·6
27·7 ± 15·3
268·9 ± 246·8
45·6 ± 15·5
12·4 ± 4·0
19 381·0 ± 14 748·6
16 935·8 ± 10 392·1
15 805·8 ± 2 875·4
253·6 ± 199·4
5 916·6 ± 6 621·3
443·7 ± 228·2
20·8 ± 10·9
26 179·5 ± 4 676·1
Measurement of IL-12p40 and IL-12p70 secretion by ELISA
IL-12p70 and IL-12p40 levels in supernatants of the various immature and mature DC populations were measured by ELISA (Fig. 6). All mature DC populations produced significantly more IL-12p40 than immature DCs. The difference between mature DC populations was not significant except in the case of CytoMix (Fig. 6a). Immature DCs stimulated with CRT or HSP70 (mDC_CRT, mDC_HSP70) expressed higher levels of IL-12p70 than unstimulated immDCs (Fig. 6b). On day 8, IL-12p70 expression levels in mDC_CRT were higher than those in mDC_CytoMix, mDC-MCM and pmDC_CRT-CM. Among the groups treated with HSP70, pmDC_HSP70-CM had the highest levels of IL-12p70. CytoMix did not yield exceptionally high levels of IL-12p70.
Our research focused on the role of two heat shock proteins, CRT and HSP70, on the maturation and activation of human DCs. To reduce the risk of bacterial contamination and to obtain a pattern of post-translational modification similar to physiological conditions, we developed suitable strategies to express human CRT and HSP70 sequences in a eukaryotic system. Our data suggest that extracellular CRT and HSP70 can strongly influence the maturation of DCs. Interestingly, preactivation of monocytes with HSPs leads to the generation of an HSP-conditioned medium (HSP-CM), containing high levels of IL-4 and GM-CSF and nearly all cytokines needed for the maturation of DCs. Additionally, mature DCs generated under the influence of extracellular HSP exhibited high levels of CCR7 expression. These findings extend the potential role of HSPs as cytokine-like proteins in the regulation of immune responses. Interestingly, we found that DCs, which were generated by priming monocytes with CRT and matured in the presence of CRT-CM showed the most prominent mature phenotype (pmDC_CRT-CM) needed to induce efficient T-cell responses. The present study confirms the hypothesis that CRT has an enhanced effect to elicit the maturation of human DCs and increases the production of proinflammatory cytokines, most probably via interaction with its cognate TLRs and the activation of the NF-κB pathway.
In recent years, the use of HSPs expressed in Escherichia coli to stimulate immune responses has been discussed controversially. It is now known that many of the proinflammatory responses thought to be induced by HSPs turned out to be caused by contamination of the recombinant HSP preparations with bacterial LPS, endotoxin and other bacterial structures belonging to the group of pathogen-associated molecular patterns (PAMPs) [39,40]. As LPS is the bacterial ligand of the TLR-4/CD14 receptor complex, which is also involved in HSP signalling, it is imperative to use LPS-free reparation in studies assessing the role of HSP in adoptive and innate immune responses [37,41]. In order to circumvent major PAMP contamination, we recently developed a strategy to express eukaryotic, soluble HSP70, which was proved to be highly pure, endotoxin-free and functional . This strategy was also applied for the expression of CRT, with the exception that the signal sequence of the protein was efficient for secretion of the protein outside the HEK293 producer cells. CRT plays an important role in molecular chaperoning and Ca2+ signalling; therefore, it is not surprising that the majority of cellular CRT is located in the ER . It was also found in the extracellular lumen, where it was shown to exert a number of physiological and pathological effects [43–45]. CRT has also been found on the cell surface of cells undergoing apoptosis. A significance of its recognition or APC activation properties has been discussed in diseases with disturbed clearance of apoptotic cells such as lupus erythematosus . None the less, it is still unclear how CRT reaches the extracellular environment. There are several theories attempting to explain its relocation from the ER to the cell surface and extracellular space. First, CRT might proceed through the secretory pathway to the cell surface, given that it has been shown to shuttle between the ER and the Golgi . Secondly, CRT might be expressed in different isoforms which do not contain the four-amino acid C-terminal ER-retrieval sequence Lys-Asp-Glu-Leu (KDEL) responsible for the retention of CRT in the ER lumen . Thirdly, it was found that CRT co-localizes with lytic perforin to the lysosome-like CTL secretory granules to prevent organelle autolysis. The release of granule content might be the mechanism for extracellular targeting of CRT . We established three strategies to express recombinant endotoxin-free CRT to be secreted in the supernatant. Although there was no significant difference between CRT and CRT_KDQL on the mRNA level, we found higher amounts of CRT_KDQL protein in the supernatant, indicating that the disturbance of the interaction between the KDEL sequence and the KDELR-1 receptor by a single-point mutation no longer mediates the retrieval of the protein. Interestingly, it was shown that AIMP-1-deficient cells have reduced interaction between gp96 and KDELR-1, which disturbs the ER retention of gp96 . Although we could reach a significant level of silencing AIMP-1 in HEK293 cells (more than 80%), CRT_shAIMP-1 yielded significantly lower levels in mRNA and protein. We therefore chose the first strategy using the unmodified CRT sequence (CRT) for large-scale protein expression. Our findings indicate that saturation of the ER retention machinery  by over-expression is efficient to produce large amounts of recombinant protein.
Molecular chaperones, including gp96, HSP70 and CRT, are potent inducers of immune responses. They act as chaperones for peptides and inducers of APCs and are thus referred to as ‘immunochaperones’ or ‘chaperokines’. However, the mechanism by which chaperones can activate DCs, the most potent APCs, are not well understood and are indeed controversial. DCs are activated not only by pathogens, but also by endogenous signals received from cells that are stressed, virally infected or killed by necrosis. HSPs such as HSP70 and CRT are such endogenous activating substances, which can function as natural adjuvants to stimulate a primary immune response and may represent natural initiators of spontaneous tumour rejection, transplant rejection and some forms of autoimmunity . The potential of HSP70 [9,38] to function as a molecular adjuvant by binding to different receptors on APCs and the signal transduction pathways responsible for this activity were analysed in various studies [16,18,21,22]. HSP70 has been shown to activate DCs via the TLR-4 pathway  and that was even hypothesized to also be so for CRT. Little is known about the role of extracellular CRT in the maturation and activation of human DCs. Here we used endotoxin-free preparations of CRT and HSP70 to prove and compare their role in the process of DC maturation. DC maturation leads to changes in antigen uptake, processing and presentation and is characterized by the up-regulation of MHC class II molecules and co-stimulatory molecules such as CD83, CD86, CD40 and CCR7 , indicating that extracellular CRT induces activation and maturation of DCs via its cognate TLR receptors.
Usually, the first step in the DC generation is the cell culture of monocytes in the presence of GM-CSF and IL-4, followed by a maturation step induced by monocyte-conditioned medium (MCM)  or a cocktail of proinflammatory cytokines and prostaglandin E2 (CytoMix) . However, the last approach in particular is highly expensive, and alternative maturation methods are needed. Regarding the expression of maturation markers and the cytokines, we found that HSPs are a promising alternative and more effective than MCM in maturation. Surprisingly, treatment with CRT or HSP70 resulted in the generation of DCs with a maturation status similar to that of those generated using the established cytokine cocktail.
Interestingly, under the influence of HSPs a strongly increased expression of CCR7, an essential receptor for DC mobilization, was observed. CCR7 is involved in homing of various subpopulations of T cells and antigen-presenting DCs to the lymph nodes [50,51]. Within the lymph nodes, T cells establish close physical contacts with DCs allowing their antigen-specific activation. Therefore, it is assumed that increased expression of CCR7 results in an increase of adoptive antigen-dependent immune response. Recent studies also demonstrate that the CCR7-dependent contacts of DCs and T cells are not only necessary for optimal initiation of protective immune responses, but also essential for the induction of peripheral tolerance and the regulation of the immune response by regulatory T cells (Tregs) .
Bak et al. found that murine CRT does not result in DC maturation in vitro or in vivo and does not elicit a measurable proinflammatory cytokine response . Conversely, human CRT acted as a potent adjuvant in the maturation process in all approaches investigated in the present study. The best DC maturation was achieved when monocytes were primed with CRT and matured in CRT-CM (pmDC_CRT-CM). Although the expression of maturation markers was higher in pmDC_CRT-CM, secretion of proinflammatory cytokines such as IL-6 and TNF-α as well as IL-12 was higher in pmDC_HSP70-CM. In light of our findings, the reasons why CRT failed to result in the DC maturation in the murine system are unclear.
There is no doubt that immunotherapeutic T-cell approaches, including DC immunization, can be improved by a better understanding of the control of DC differentiation. Inflammatory (e.g. IFN-α), microbial (e.g. bacterial lipopolysaccharide), ‘alarm’ (e.g. HSPs) and cognate T cell signals (e.g. CD40 ligand; CD40L) can all induce a programme of maturation in DC . It was found previously that ligation of CD40 enhances DC maturation and function, including conditioning for CTL priming [13,52]. Conversely, it is known that CD40 acts as receptor for the HSP70 interaction , and this interaction results in the activation of NF-κB, maturation of the DCs and secretion of cytokines (e.g. IL-15; [54,55]). These effects are also interesting with regard to findings, showing, that CD4+ T cells activated by stressed DCs can induce the maturation and IL-12p40 production of DCs . In order to produce significant amounts of IL-12p70, human APCs need a priming signal such as IFN-γ followed by a second signal (e.g. LPS or CD40L) [57,58]. This is in contrast to murine monocyte/macrophages which can be stimulated with LPS alone for IL-12 secretion. The experimental setup shown in our study was devoid of such a priming signal, hence only low levels of IL-12 are detectable. However, substantial amounts of IL-4 were present and this cytokine is known to counter-regulate IL-12 . Differences measured between the different DC subsets are significant and of importance given the strong immunoregulatory potential of IL-12 even at low doses.
The data from our study provide the basis for further investigations into the effect that HSP70 and CRT have on the phenotype and functionality of specific APCs. Further studies should now include the determination of the immunogenicity, antigen presentation capacities and the induction and stimulation of defined T-cell populations (e.g. purified CD4+CD25+ Treg cells, CD4+ T helper cells, antigen-specific CD8+ CTLs) as well as an extensive monitoring of relevant markers and functional parameters on those cells. In particular, the role of CRT in the inhibition of FasL (CD95L)-mediated apoptosis of T cells and its influence of the restricting DC priming functions through Fas–FasL interactions as a potent mechanism employed by CD4+CD25+ regulatory cells to restrict CD8+ T-cell immune responses are of interest [60,61].
In summary, systems for in vitro culture of DC progenitors using exogenous haematopoietic cytokines to support their growth, differentiation and maturation have made it feasible to generate DCs in large numbers. Because they initiate adaptive immune responses and determine tolerance, DCs are ideally suited to serve as natural adjuvants for vaccination and immunotherapy purposes. For the therapeutic use, it is essential to generate APCs that can take up and present antigens efficiently, resulting in potent and adjusted activation of effector T cells. This study demonstrates for the first time that human extracellular CRT can induce DC maturation and monocyte activation by using the NF-κB pathway. The fact that preactivation of monocytes with HSPs resulted in the generation of nearly all cytokines needed for the maturation process underlines the important role of chaperokines in the adoptive immune response. To finally confirm the functionality of HSP-generated DC populations, further experiments are needed to demonstrate improved antigen uptake and presentation.
The authors would like to thank Sarina Lukis, Doerthe Rokitta and Julia Struss for their excellent technical assistance. This study was supported in part by the German Federal Ministry of Education and Research (reference number: 01EO0802) and Deutsche Forschungsgemeinschaft (DFG Wi1822/5–1).