Viral and murine interleukin-10 are correctly processed and retain their biological activity when produced in tobacco

Background Interleukin-10 (IL-10) is a potent anti-inflammatory cytokine, with therapeutic applications in several autoimmune and inflammatory diseases. Oral administration of this cytokine alone, or in combination with disease-associated autoantigens could confer protection form the onset of a specific autoimmune disease through the induction of oral tolerance. Transgenic plants are attractive systems for production of therapeutic proteins because of the ability to do large scale-up at low cost, and the low maintenance requirements. They are highly amenable to oral administration and could become effective delivery systems without extensive protein purification. We investigated the ability of tobacco plants to produce high levels of biologically-active viral and murine IL-10. Results Three different subcellular targeting strategies were assessed in transient expression experiments, and stable transgenic tobacco plants were generated with the constructs that yielded the highest accumulation levels by targeting the recombinant proteins to the endoplasmic reticulum. The best yields using this strategy in T1 plants were 10.8 and 37.0 μg/g fresh leaf weight for viral and murine IL-10, respectively. The recombinant proteins were purified from transgenic leaf material and characterized in terms of their N-glycan composition, dimerization and biological activity in in vitro assays. Both molecules formed stable dimers, were able to activate the IL-10 signaling pathway and to induce specific anti-inflammatory responses in mouse J774 macrophage cells. Conclusion Tobacco plants are able to correctly process viral and murine IL-10 into biologically active dimers, therefore representing a suitable platform for the production for these cytokines. The accumulation levels obtained are high enough to allow delivery of an immunologically relevant dose of IL-10 in a reasonable amount of leaf material, without extensive purification. This study paves the way to performing feeding studies in mouse models of autoimmune diseases, that will allow the evaluation the immunomodulatory properties and effectiveness of the viral IL-10 in inducing oral tolerance compared to the murine protein.


Background
The production of biopharmaceuticals in transgenic plants has become more and more attractive over the last few years, primarily reflecting the low start-up and maintenance costs compared to fermenter-based production platforms and the ability to scale up production rapidly according to demand [1]. Plants can also be used as delivery systems for certain products, making extensive purification unnecessary. This applies in particular to products used for mucosal immunomodulation (e.g. oral vaccination or oral tolerance). Indeed, plant tissues are highly suitable for oral administration, allowing the direct delivery of recombinant protein drugs to the gut-associated lymphoid tissue (GALT) while encapsulating the proteins and protecting them from digestion [2,3].
With these advantages in mind, we investigated the possibility of using plants to produce the anti-inflammatory cytokine interleukin-10 (IL-10), a multifunctional cytokine from the alpha-helical bundle superfamily with diverse effects on most hemopoietic cell types [4]. While it plays a complex role in the immune system, the major activities of IL-10 are to inhibit cytokine production by macrophages and to suppress their accessory functions during T-cell activation [5,6]. Since this causes the termination of inflammatory responses, IL-10 is widely considered as an immunosuppressive and anti-inflammatory cytokine, and many investigations of IL-10 expression in vitro, in animal models and in human patients have indicated a significant role in inflammatory, malignant and autoimmune diseases, highlighting the potential clinical value of this cytokine [7]. Human IL-10 has been produced in stably transformed tobacco plants [8] and the ability of plant-produced human IL-10 to induce antiinflammatory responses has also been demonstrated [9]. However, mammalian IL-10, including the human one, also presents several immunostimulatory properties (e.g., activation of dendritic cells, NK cells, and some T cells) which appear not to be negligible when IL-10 is used in vivo to induce tolerance [10].
Interestingly, IL-10 has orthologs in several virus genomes, and the IL-10 produced by Epstein-Barr virus (vIL-10) is particularly closely related to its human counterpart (71% and 84% identity at the nucleotide and amino acid sequence levels, respectively) and binds to both human and murine receptors [11]. Despite the sequence similarity, vIL-10 exhibits primarily the immune-inhibitory properties associated with the cellular cytokine (e.g., suppression of Th1-polarized responses and monocyte inhibition) but lacks many of the immunostimulatory properties associated with the human IL-10 (hIL-10) and murine IL-10 (mIL-10) [12]. This exacerbated inhibitory effect of vIL-10 makes it an even more attractive therapeutic candidate for tolerance induction, as the viral cytokine could have a profoundly different effect on the outcome of an in vivo immune response compared to hIL-10/mIL-10. [11].
One promising immunomodulatory application of IL-10 is to enhance the induction of oral tolerance to co-administered auto-antigens, leading to the prevention or treatment of autoimmune diseases with lower doses of tolerizing protein. We are particularly interested in the induction of oral tolerance for the prevention of type 1 diabetes mellitus (T1DM) which could be achieved by repeated oral administration of small doses of one of the major auto-antigens associated with the disease, the 65-kDa isoform of the enzyme glutamic acid decarboxylase (GAD65). We have already reported the characterization of transgenic tobacco plants producing immunoreactive GAD65 [13], although feeding studies in a mouse model have not yet been possible due to the low protein expression levels, requiring the consumption of unrealistically large amounts of plant material in order to achieve an oral tolerizing effect. However, simultaneous feeding with plant material containing IL-10 could reduce the amount of GAD65 necessary to induce oral tolerance. Additionally, IL-10 induces the formation of T regulatory-1 cells [10], and this should also drive the immune response towards active rather than a passive tolerance, which should in turn provide a more robust and long-lasting protection from the onset of the autoimmune disease. A comparative study, feeding either recombinant vIL-10 or the recombinant endogenous mIL-10 together with the GAD65 auto-antigen to NOD mice, would therefore allow to evaluate and assess the outcome of the different immunomodulatory properties of the viral cytokine in inducing oral tolerance in the animal model, and to evaluate the possibility of exploiting vIL-10 for the prevention of T1DM.
Here we describe for the first time the production of vIL-10 and mIL-10 in plants. Three different targeting strategies were used to identify the best subcellular compartment for IL-10 accumulation in tobacco leaves. Characterization of the plant-derived cytokines confirmed that tobacco plants can accumulate correctly-processed, dimeric IL-10 and that the recombinant protein retains the activity of its native counterpart. The impact of this achievement on the oral delivery of auto-antigens produced in plant tissues is discussed.

Investigation of different targeting strategies in transient expression assays
Three constructs each were designed for the mIL-10 and vIL-10 transgenes, targeting the recombinant proteins to three different sub-cellular localizations: (i) the endoplasmic reticulum (ER), achieved by retaining the native sig-nal peptide and adding a C-terminal SEKDEL motif for ER retention (Fig. 1, ER-m/vIL-10) [14]; (ii) the plasma membrane (IL-10 facing the apoplast), achieved using the native signal peptide and adding the C-terminal hinge and transmembrane domain from the human T-cell receptor (Fig. 1, Apo-m/vIL-10) [15]; and (iii) the ER membrane (IL-10 facing the cytosol), achieved by deleting the native signal sequence and adding the C-terminal cytochrome b5 (cytb5) membrane anchor domain (Figure 1, Cyt-m/vIL-10) [14,16,17]. These constructs, all bearing a His 6 -tag at the C-terminus to facilitate protein purification, were generated by PCR and inserted into the plant expression vector pTRAk [18] placing the transgene under the control of the CaMV 35S promoter with duplicated enhancer ( Figure 1). All plasmids were transferred into Agrobacterium tumefaciens, and transient expression in Nicotiana tabacum cv. Petit Havana SR1 leaves was carried out at least three times for each construct by agroinfiltration. Analysis of IL-10 accumulation in leaves using a sandwich enzyme-linked immunosorbent assay (ELISA; Figure 2), showed that both mIL-10 and vIL-10 accumulated to the highest levels in the ER (up to 10-16 μg/g fresh leaf weight (FLW)). Furthermore, immunoblot analysis of leaf extracts indicated that the recombinant proteins were correctly processed, as the molecular weights were consistent with those of the mature native polypeptides (taking into account the C-terminal His 6 and SEK-DEL tags; data not shown).

Stable transformation of tobacco with constructs for ER retention
Based on the transient expression data, stable nuclear transformation of N. tabacum cv. Petit Havana SR1 was carried out using the ER-m/vIL-10 constructs. For each construct, 40 transgenic plants were regenerated and the amount of IL-10 in the leaves of the hemizygous T 0 plants was estimated by ELISA. For mIL-10, the best yield was 21.3 μg/g FLW, whereas the best yield of vIL-10 was 8.9 μg/g FLW, mirroring the results of the transient expression assays. Transformants were fertile and the measurements were repeated in T 1 plants, resulting in accumulation levels up to 37.0 μg/g FLW for mIL-10 and 10.8 μg/g FLW for vIL-10. Transgenic plants expressing vIL-10 grew more slowly than wild type plants and were stunted to an extent that correlated with the amount of recombinant protein accumulating in the leaves (see Additional file 1). The plants expressing mIL-10 did not display any mutant phenotype.

Plant-derived mIL-10 and vIL-10 are dimeric
The biological activity of IL-10 depends strictly on its dimeric state [19]. In order to determine whether or not mIL-10 and vIL-10 were properly assembled into dimers, T-DNA region of the pTRAk plant expression vector and gene cassettes targeting murine and viral IL-10 to different subcellular locations Comparison of IL-10 levels in leaves agroinfiltrated with the different targeting constructs Figure 2 Comparison of IL-10 levels in leaves agroinfiltrated with the different targeting constructs. N. tabacum cv. Petit Havana SR1 leaves were vacuum-infiltrated with A. tumefaciens carrying the different plant expression constructs. For each construct, three to four leaves from different wild-type plants were infiltrated; they were incubated on wet filter paper for 3 d, with a 16-h photoperiod at ~20°C. Quantification of IL-10 accumulation levels in the leaf extracts was performed by sandwich ELISA. The gray and solid bars represent the data for the viral and murine IL-10, respectively; the localizations reported on the x axis refer to the constructs described in Figure 1. Values are the means ± SD of three independent experiments.
the His 6 -tagged proteins were purified from stable transgenic tobacco leaves via immobilized-metal affinity chromatography (IMAC) on a nickel-nitrilotriacetic acid (Ni-NTA) column, concentrated, and fractionated by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) under either reducing or non-reducing conditions. Immunoblot analysis confirmed the correct molecular weights of both viral and murine IL-10 (19 and 21 kDa, respectively; Figures 3a and 3b, right panels). When separated under non-reducing conditions, the plant-derived vIL-10 predominantly formed stable dimers (Figure 3a, left panel), whereas mIL-10 was partly dimeric and partly monomeric (Figure 3b, left panel). Since commercial mIL-10 also gives rise to a monomer band under these conditions, this is most likely an artifact of the SDS-PAGE method. This was confirmed by gel filtration analysis, which showed that the plant-derived mIL-10 eluted as a single peak with a profile corresponding to the size of the dimeric form ( Figure 3c).

Investigation of N-glycosylation and signal peptide cleavage
The mature mIL-10 polypeptide has two potential N-glycosylation acceptor sites, whereas vIL-10 has a single site.
To determine the glycan composition, both plant-derived proteins were digested with trypsin and analyzed by mass spectrometry (MS). Despite the presence of a glycan acceptor site at Asn 127 , the plant-derived vIL-10 was not glycosylated. In contrast, mIL-10 was glycosylated, but only one of the two potential acceptor sites was used, as is the case for the native protein [20,21]. Unexpectedly for a protein bearing a SEKDEL tag, oligo-mannose-type (OMT) N-glycans accounted for only 46% of the total Nglycan population, with Man 7 being the most abundant glycoform ( Figure 4, N-glycan abbreviations according to Proglycan [22]). Complex type N-glycans accounted for 44% of the total, with GnGnXF and GnGnX being the most prominent. GnMX was present at lower levels, while the other complex type N-glycans were found in only trace amounts.
For both proteins, the identity of the N-terminal peptide was established by tandem MS analysis. The sequence of the mature mIL-10 polypeptide started with QYSRG while vIL-10 started with TDQCD, confirming co-translational ER translocation and correct cleavage of both signal peptides, as predicted by the SignalP 3.0 server [23].

Biological activity of plant-derived mIL-10 and vIL-10
All the anti-inflammatory effects of IL-10 require the activation of signal transducer and activator of transcription 3 (STAT3), which acts downstream of the IL-10 receptor in both mouse and human cellular models [11]. We therefore tested the ability of plant-derived mIL-10 and vIL-10 to phosphorylate the Tyr 705 residue of STAT3. To exclude possible interference from plant proteins co-purifying with the recombinant IL-10 molecules, equal amounts of wild type leaf material were subjected to the same purification procedure on a Ni-NTA column, and the eluate was used as a control in all biological activity assays. Stimula- tion of the mouse macrophage cell line J774 with different doses of plant-derived mIL-10 and vIL-10 triggered STAT3 tyrosine phosphorylation within 15 min and in a dosedependent manner ( Figure 5). The concentration of plantderived IL-10 required to elicit STAT3 tyrosine phosphorylation to a degree comparable to that observed in response to commercial mIL-10 (20 ng/ml) and vIL-10 (150 ng/ml) was 50 ng/ml for the murine protein ( Figure  5a) and 500 ng/ml for the viral protein ( Figure 5b). STAT3 phosphorylation resulted specifically from the presence of recombinant IL-10, since control eluate from wild type plant extracts had no detectable effect even at the highest dose tested ( Figure 5).

Assessment of IL-10 integrity and dimerization
It has been reported that suppressor of cytokine signaling 3 (SOCS3) is one of the targets of murine IL-10 [24,25]. Indeed, SOCS3 mRNA and SOCS3 protein expression in lipopolysaccharide (LPS)-stimulated J774 cells is enhanced in the presence of commercial mIL- 10 [26]. Similarly, purified plant-derived mIL-10 increased the expression of LPS-induced SOCS3 mRNA ( Figure 6a) and SOCS3 protein (Figure 6b). Finally, we verified that the plant-derived mIL-10 and vIL-10 were fully functional by testing their inhibitory activity on the LPS-induced production of tumor necrosis factor alpha (TNFα). The amount of TNFα released into the culture supernatant of J774 cells stimulated with LPS for 18 h was significantly reduced in the presence of plant-derived mIL-10 to the same extent observed with the commercial mIL-10 ( Figure  7a). Similarly, the ability of plant derived vIL-10 to inhibit LPS-induced TNFα production was equivalent to that of its commercial counterpart (Figure 7b). Addition of control eluate from wild type plants had no effect on the J774 response to LPS, or the suppressive activity of IL-10 (Figure 7). These results confirmed that the inhibition of TNFα production was dependent on the recombinant cytokines, and not on any co-purified plant proteins.

Discussion
We set out to generate transgenic tobacco plants producing biologically-active vIL-10 and mIL-10 at levels sufficient for raw transgenic leaf material to be used in animal feeding studies. Such studies would compare the two versions of IL-10 as potential adjuvants for the induction of oral tolerance to a co-fed auto-antigen in a mouse model of type-1 diabetes or other autoimmune diseases.
The intracellular targeting of a recombinant protein often has a significant effect on its yield and, in some cases, also on its biological properties. Therefore, we targeted both versions of IL-10 to three different subcellular compartments -the ER, cytosol and apoplast -in order to assess the impact on protein accumulation. Preliminary studies of murine IL-10 transient expression driven by a PVXbased system in N. benthamiana performed in our laboratory, had shown that the ER was by far the most suitable compartment for mIL-10 accumulation compared to the cytosol and the apoplast (ER-retained reached around 10 μg/g FLW; cytosolic and apoplastic expression was nearly undetectable as determined by immunoblot analysis; data not shown). As it has been reported that proteins associated with the plasma membrane are protected from degradation in the cytosol and the apoplast [15,27,28], in this work we used membrane anchors that allowed the recombinant proteins to be exposed to the apoplast or cytosol while retaining them in the membrane, in the attempt to increase the overall IL-10 accumulation levels. We evaluated the three targeting strategies by transient expression in tobacco leaves following agroinfiltration, since this provides a reliable indication of performance much more rapidly than the generation of stably transformed plants. Despite the addition of membrane anchors to stabilize the recombinant proteins in the cytosol and the apoplast, both versions of IL-10 accumulated to the highest levels in the ER, confirming our previous observations for mIL-10 and in line with what has been observed for hIL-10 stably expressed in tobacco [8]. Many other recombinant proteins have also accumulated to the highest levels in the ER when different targeting strategies were compared [29-Activation of STAT3 phosphorylation by murine and viral IL-10 produced in tobacco plants

31
]. Our data therefore confirm the beneficial environment within the ER, favoring correct protein folding and assembly, and high protein stability.
In T 1 transgenic plants, the highest level of mIL-10 accumulation was 37.0 μg/g FLW (0.6% total soluble protein (TSP)) and the highest level of vIL-10 accumulation was 10.8 μg/g FLW (0.1% TSP). These yields fall into the 0.1-1.0% TSP range typically observed for pharmaceutical proteins produced in nuclear transgenic plants [32] and are much higher than the 0.0055% TSP previously reported for hIL-10, which is very similar in structure to the murine and viral orthologs [8,33]. Surprisingly, the expression of vIL-10 caused a stunted phenotype whose severity correlated with the level of protein accumulation. Further investigation is required to clarify the direct impact of vIL-10 on plant development. The accumulation of both versions of IL-10 increased between the T 0 and T 1 generations, indicating that protein yields are partly dependent on the transgene copy number and/or zygosity. Further strategies could be employed to boost IL-10 accumulation, e.g. enhancing its stability through the use of fusion tags or through the co-expression of interacting molecules [34]. Nevertheless, the yields of both proteins are already sufficient for feeding studies in the mouse model. Indeed, on the basis of comparable experiments [35], 2-10 μg of IL-10 per day should be sufficient to elicit immunomodulatory effects and this would require the administration of a few milligrams of lyophilized plant tissue, which could be ingested by a mouse without substantially interfering with its dietary needs [9].
The expression of a recombinant protein containing a native signal peptide requires the signal peptide to be interpreted correctly by the heterologous host cell, otherwise the protein could be incorrectly cleaved or not cleaved at all. This is especially relevant for pharmaceutical proteins, where imprecise cleavage could affect the biological activity of the protein and might raise safety concerns [36]. The mIL-10 signal peptide was correctly processed in tobacco, as shown by sequencing the N-terminal peptide of the mature protein, which had the same QYSRE N-terminus previously reported for native mIL-10 [37]. There are at least two different N-terminal sequence reported for vIL-10: QCDNF and TDQCD. The former is claimed in several publications [38][39][40] and is most likely based on the prediction first reported by Moore and colleagues [20]. The latter is predicted by the latest version of the widely-used SignalP 3.0 Server [41,42], and is reported in at least one publication [43]. Tandem MS analysis of the N-terminal peptide of the purified vIL-10 produced in tobacco showed the N-terminal sequence to be TDQCD, in line with the prediction of the SignalP 3.0 Server. To our knowledge, this is the first time that the N-terminal amino acidic sequence of mature vIL-10 determined by Inhibition of LPS-induced TNFα production by plant-derived murine and viral IL-10 Figure 7 Inhibition of LPS-induced TNFα production by plantderived murine and viral IL-10. J774 cells were stimulated for 18 h with IL-10 alone, LPS alone, or LPS plus IL-10. The eluate of the purification of wild type leaves (wt) was also included, alone or in combination with commercial IL-10, to exclude the interference of plant proteins with the assay. TNFα secretion in the culture medium (med) was quantified by ELISA and is reported as pg/ml supernatant. Values are the means ± SD of four independent experiments for the murine IL-10 (a) and three independent experiments for the viral IL-10 (b). ** p < 0.001. bioinformatic tools, has been experimentally confirmed by the expression in a eukaryotic cell.
Although vIL-10 is not glycosylated [44], both the native and recombinant mIL-10 are heterogeneously N-glycosylated at an acceptor site near the N-terminus [20]. This glycosylation has no known influence on mIL-10 activity [11], but the presence of plant complex type N-glycans on therapeutic glycoproteins renders them potentially immunogenic and thus raises safety concerns, particularly in the case of parenteral administration. Even so the potential immunogenicity of plant glycans may also be important for orally administered proteins, especially in patients with severe food allergies. The glycan modifications that occur in the ER produce high-mannose-type N-glycans that are conserved between mammals and plants, so the use of a KDEL signal for ER retention not only increases protein yields, but also prevents the addition of plant complex type glycans. However, our N-glycan analysis indicated that mIL-10, although carrying a SEKDEL tag, is not efficiently retrieved to the ER, and is modified by enzymes located in the trans-Golgi network [45,46]. In addition, the presence of terminal GlcNAc residues on the complex glycan structures would suggest that mIL-10 escaping the ER is secreted by the plant cell and not, for example, directed to the vacuole [47]. The high efficiency of ER retrieval has been documented for a number of recombinant proteins produced in plants [48][49][50], but there are also several examples of KDEL-tagged recombinant proteins containing complex type glycans, indicating that the retrieval system can be leaky and must be evaluated on a case-by-case basis [51,52]. As immunoblot analysis of the purified protein confirmed that the KDELtag is still present (data not shown) and our protein yields are not high enough to suggest saturation of the KDEL retrieval machinery [53], the most likely explanation for the presence of complex type N-glycans on mIL-10 is that the tag is partially occluded following assembly of the IL-10 dimer, and is not completely accessible to the KDEL receptor. Localization studies will be performed in order to complement the N-glycan analysis, and to clarify where both versions of IL-10 precisely accumulate within the plant cell, particularly since vIL-10 cannot be traced using glycan structures.
Non-reducing SDS-PAGE followed by immunoblot analysis and (for mIL-10) gel filtration analysis confirmed that both vIL-10 and mIL-10 assemble into dimers in tobacco leaves. Moreover, it appears that the dimerization of both plant-derived versions of IL-10 is more efficient than that of their commercially produced counterparts. Since the higher dimer/monomer ratio is consistent regardless of the total amount of protein loaded onto the gel, we can rule out the possibility that this is an artifact caused by high protein concentration. It is also unlikely that the presence of plant glycans is responsible, since vIL-10 is not glycosylated. The most probable explanation is that the presence of the C-terminal SEKDEL and His 6 tags may enhance the stability of the dimers. However, to confirm this assumption, we would need to express non-tagged versions of the proteins and compare the dimer/monomer ratios. Increased stability of plant-derived IL-10 dimers was also observed for hIL-10 by Menassa and colleagues [33]. Interestingly, hIL-10 was not glycosylated and the C-terminal tags had been cleaved from the plantderived hIL-10 by thrombin digestion, which should remove all additional amino acids except a lysine and a valine.
The ability of both vIL-10 and mIL-10 to form dimers indicated that the recombinant proteins were likely to retain their biological activity, a hypothesis we tested by exposing LPS-stimulated J774 mouse macrophage cells to different amounts of recombinant IL-10 and assaying the effect on downstream components of the signaling pathway. We observed a clear dose-dependent effect on STAT3 phosphorylation after treatment with both varieties of IL-10, indicating that the plant-derived proteins properly interacted with IL-10 receptor and initiated the signal transduction cascade. Moreover, mIL-10 was able to upregulate the expression of the intracellular negative regulator of cytokine responses, SOCS3. Most importantly, given that IL-10 is best known for its ability to inhibit proinflammatory cytokine and chemokine expression in LPSstimulated macrophages, our experiments demonstrated that both mIL-10 and vIL-10 are able to inhibit the LPSinduced production of TNFα. Higher doses of the plantderived proteins as compared to the corresponding commercial IL-10 products were required, perhaps because of the presence of the C-terminal SEKDEL and His 6 tags. Nevertheless, this will have no impact on the planning of feeding studies since the dose of IL-10 necessary to enhance oral tolerance needs to be determined empirically by administering different amounts of transgenic plant material to the mice.

Conclusion
Taken together, these results clearly demonstrate that tobacco plants can express the viral and murine IL-10 genes, process and assemble the corresponding proteins into functional, biologically-active dimers. The accumulation levels of both viral and murine IL-10 in tobacco leaves are high enough to provide sufficient material for oral administration in oral tolerance studies using the available mouse models. This will allow to determine whether simultaneous feeding with plant material containing GAD65 and IL-10 could reduce the amount of auto-antigen necessary to prevent the onset of T1DM, and to comparatively evaluate the effectiveness of the viral and murine IL-10 as immunomodulators.

Plant transformation vectors
Six different constructs were generated using the polymerase chain reaction (PCR) to amplify the murine and viral IL-10 cDNA sequences from the pBluescriptSKII+mIL-10 [54] and pAAV-vIL-10 [55] plasmids, respectively, with forward primers incorporating an NcoI restriction site at the 5' end (for amplification of the complete IL-10 coding region including the native signal peptide), or a BspHI site (for amplification of the mature mIL-10 sequence) or an XmnI site (for amplification of the mature vIL-10 sequence). The appropriate targeting and His 6 -tag sequences were added by PCR in-frame to the 3' end of the IL-10 coding region, and an XbaI site was created at the 3' end of all the constructs. The plasmids were electroporated into A. tumefaciens strain GV3101::pMP90RK [56].

Transient and stable expression of IL-10 in tobacco
Agrobacterium-mediated transient expression (agroinfiltration) was carried out essentially as described by Kapila et al. [57]. The bacterial cultures were adjusted to OD 600 = 1 with 2× induction medium (10% (w/v) sucrose, 3.6% (w/ v) glucose, 8.6% (w/v) Murashige-Skoog salts) and water, and induced for an additional 2 h with 200 μM acetosyringone. N. tabacum cv. Petit Havana SR1 leaves were vacuum-infiltrated at 70 mbar for 20 min in the bacterial suspension then placed in plastic trays with moistened Whatman paper. The trays were sealed with Saran wrap and incubated at 25°C with a 16-h photoperiod, 7000 Lux, for 3 d.
Stable transgenic tobacco (N. tabacum cv. Petit Havana SR1) plants were produced using the leaf disc transformation method [58]. T 0 plants were grown on Murashige-Skoog medium containing 100 mg/l kanamycin prior to transfer to soil in the glasshouse, and then selfed to produce the T 1 generation. Transgenic T 1 plants were also selected on kanamycin-containing medium and maintained in soil.

Immunoblot analysis
Soluble proteins were extracted from leaf discs with three volumes of PBS (pH 7.4) containing 0.05% (w/v) Tween-20, 5 mM EDTA and 2% (w/v) polyvinylpolypyrrolidone (PVPP). Equal volumes of the supernatant were either separated by SDS-PAGE on a 17% (w/v) gel prior to blotting, or spotted directly onto a nitrocellulose membrane, blocked for 1 h with 5% (w/v) non-fat milk and incubated overnight at 4°C with either a rat anti-mouse IL-10 antibody (BD Pharmingen, San Diego, CA, USA) diluted 1:2000 or a rat anti-human IL-10 and viral IL-10 (BD Pharmingen) antibody diluted 1:1000. The membrane was then incubated for 1 h at room temperature with an alkaline phosphatase (AP)-conjugated rabbit anti-rat immunoglobulin G (IgG) (Sigma-Aldrich, St. Louis, Mo, USA) diluted 1:5000. The signal was detected by incubation with nitroblue tetrazolium and 5-bromo-4-chloro-3indolyl phosphate (NBT-BCIP). Recombinant mIL-10 produced in insect cells (BD Pharmingen) and vIL-10 produced in Escherichia coli (R&D Systems, Minneapolis, MN, USA) were used as standards.

Quantification of IL-10 by ELISA
Microtiter plates (High-binding; Greiner Bio-One GmbH, Frickenhausen, Germany) were coated overnight at 4°C with either a rat anti-mouse IL-10 antibody (BD Pharmingen) diluted 1:1000 in 0.2 M sodium phosphate, pH 6.5, or a rat anti-human IL-10 and viral IL-10 antibody (BD Pharmingen) diluted 1:2000 in 50 mM sodium carbonate buffer, pH 9.6. Wells were blocked with 1% (w/v) bovine serum albumin (BSA) in PBS containing 0.05% (w/v) Tween-20 (PBST) for at least 1 h. Samples diluted in PBS were applied to the coated plates and incubated overnight at 4°C. The corresponding biotinylated antibodies (554465 and 554499; BD Pharmingen) were added to the plates at 1:1000 dilution in PBST and incubated for 1 h at room temperature. AP-conjugated streptavidin (Jackson ImmunoResearch Laboratories, West Grove, PA, USA) was added to the wells at 1:2500 dilution in PBST, and incubated for 30 min at room temperature before the signal was detected with p-nitrophenyl phosphate. After 15-20 min, the signal was quantified by measuring the absorbance at 405 nm. Ten two-fold serial dilutions of the standards described in the previous section, starting from a concentration of 50 ng/ml for the murine and 1 ng/ml for the viral IL-10, respectively, were used as positive controls, and the linear range of the dose-response curve obtained was used for quantification of IL-10 in the samples. Extract from wild-type tobacco was used as negative controls. The plates were washed with PBST between incubations.

Purification of IL-10 by IMAC
After removing the midrib, the leaf tissue was ground in a mortar with liquid nitrogen and extracted with three volumes of buffer (PBS, 10 mM ascorbic acid, 0.1% (w/v) Tween-20, pH 6.0). Insoluble material was removed by centrifugation (15 min, 15,000 × g, 4°C). The supernatant was passed through a paper filter (#MN615, Macherey-Nagel, Düren, Germany), adjusted to 500 mM NaCl, 5 mM imidazol, pH 8.0, and placed for 1 h on ice, gently shaking. After another centrifugation step (30 min, 30,000 × g, 4°C), the clear supernatant was loaded on a disposable column (Bio-Rad, Hercules, CA, USA) packed with 1 ml Ni-NTA Superflow resin (QIAGEN, Hilden, Germany) using a flow rate of approximately 2 ml/min. The column was extensively washed with PBS (pH 8.0) containing 500 mM NaCl and 5 mM imidazol to remove nonspecifically bound material and the IL-10 was eluted with PBS containing 500 mM NaCl and 500 mM imidazol (pH 8.0). IL-10 fractions were pooled, dialyzed against water, and lyophilized. Lyophilized material was then dissolved in PBS (pH 7.4), the preparation was centrifuged to remove residual insoluble material and IL-10 concentration in the supernatant was determined by ELISA.

Gel filtration
Gel filtration was performed using a Superdex 200 (10/ 30) column (GE Healthcare, Uppsala, Sweden), with PBS (pH 7.4) as the eluent and a flow rate of 0.35 ml/min. Purified recombinant mIL-10 (450 ng) was loaded on the column, and 1-ml fractions were collected for the elution volumes 7-12 ml and 17-28 ml. Finally, 0.2-ml fractions were collected from 12 to 17 ml. IL-10 was detected by ELISA.

MS analysis of tryptic peptides and glycopeptides
Murine and viral IL-10 purified from leaf material by IMAC were separated on a 17% (w/v) gel by SDS-PAGE under reducing conditions, and the proteins were visualized by Coomassie brilliant blue staining. The bands were then excised, destained, carbamidomethylated, digested with trypsin and extracted from gel pieces, as previously described [59,60]. Subsequent fractionation of the peptides by capillary reversed-phase chromatography, with detection by a quadrupole time-of-flight (Q-TOF) Ultima Global mass spectrometer (Waters Micromass, Manchester, UK), was performed as described previously [59,61]. The MS data from the tryptic peptides were compared with data sets generated by in silico tryptic digestion of the recombinant protein sequences using the PeptideMass program [59,60,62]. Sequences of the N-terminal peptides were confirmed by tandem MS experiments. Data were analyzed with MassLynx 4.0 SP4 Software (Waters Micromass, Milford, MA, USA).
Determination of TNFα levels in cell-free supernatants was performed using a commercial ELISA kit (DY410; R&D Systems), according to the manufacturer's instructions.

Statistical analysis
Statistical changes in TNFα production and SOCS3 mRNA expression levels were determined using the one-way ANOVA with α set to 0.05, according to the Newman-Keuls test.

Additional file 1
Phenotype of tobacco plants stably transformed with the construct encoding for the ER-retained viral IL-10. Three independent primary transformants (8 weeks after transfer to soil) are shown and viral IL-10 accumulation levels, determined by ELISA, are indicated for each plant. The correlation between the viral IL-10 accumulation levels and the stunted phenotype was observed both in the T 0 and T 1 generation plants.