ER intrabody-mediated inhibition of interferon α secretion by mouse macrophages and dendritic cells

Interferon α (IFNα) counteracts viral infections by activating various IFNα-stimulated genes (ISGs). These genes encode proteins that block viral transport into the host cell and inhibit viral replication, gene transcription and translation. Due to the existence of 14 different, highly homologous isoforms of mouse IFNα, an IFNα knockout mouse has not yet been established by genetic knockout strategies. An scFv intrabody for holding back IFNα isoforms in the endoplasmic reticulum (ER) and thus counteracting IFNα secretion is reported. The intrabody was constructed from the variable domains of the anti-mouse IFNα rat monoclonal antibody 4EA1 recognizing the 5 isoforms IFNα1, IFNα2, IFNα4, IFNα5, IFNα6. A soluble form of the intrabody had a KD of 39 nM to IFNα4. It could be demonstrated that the anti-IFNα intrabody inhibits clearly recombinant IFNα4 secretion by HEK293T cells. In addition, the secretion of IFNα4 was effectively inhibited in stably transfected intrabody expressing RAW 264.7 macrophages and dendritic D1 cells. Colocalization of the intrabody with IFNα4 and the ER marker calnexin in HEK293T cells indicated complex formation of intrabody and IFNα4 inside the ER. Intracellular binding of intrabody and antigen was confirmed by co-immunoprecipitation. Complexes of endogenous IFNα and intrabody could be visualized in the ER of Poly (I:C) stimulated RAW 264.7 macrophages and D1 dendritic cells. Infection of macrophages and dendritic cells with the vesicular stomatitis virus VSV-AV2 is attenuated by IFNα and IFNβ. The intrabody increased virus proliferation in RAW 264.7 macrophages and D1 dendritic cells under IFNβ-neutralizing conditions. To analyze if all IFNα isoforms are recognized by the intrabody was not in the focus of this study. Provided that binding of the intrabody to all isoforms was confirmed, the establishment of transgenic intrabody mice would be promising for studying the function of IFNα during viral infection and autoimmune diseases.


Introduction
Interferons (IFNs) are divided into three multigene families (type I, II and III). The type I interferon family comprises the highest number of members: IFNαs, IFNβ, IFNε, IFNƬ, IFNκ, IFNω, IFNδ and IFNξ respectively [1]. Type I IFNs play a major role in the immune response during acute viral and bacterial infections but also take part in induction of tumor cell death and inhibition of angiogenesis [2][3][4]. They play a pathogenic role in autoimmune diseases and in chronic infections [5,6].
The type I family members IFNα and IFNβ are produced by almost all cells after contact with microbial products. Their synthesis is induced after binding of danger signals (PAMPs or DAMPs) to some PRRs, especially TLR 7, 9 and RIG-I-like receptors [4]. Activation of PRRs leads to type I IFN synthesis. Binding of type I IFNs to their receptor (IFNAR) induces multiple downstream signalling pathways leading to activation of a large number of IFN-stimulated genes (ISGs) in infected and neighbouring cells [2,4]. ISG-encoded proteins inhibit the spread of viruses by inhibition of replication, viral transcription and translation, viral assembly and viral egress [7]. Additionally genes are induced that encode cytokines and chemokines, antibacterial effectors and pro-apoptotic and anti-apoptotic molecules [8].
Type I IFNs are produced by virus infected innate immune cells. They can act on innate immune cells including dendritic cells and macrophages enhancing the antigen-presenting function of these cells. In addition virus-infected macrophages and dendritic cells (DCs), the main producers of type I IFNs, secrete IFNα and IFNβ, which can lead to chemokine production in innate immune cells [9]. IFNα and IFNβ activate immature committed DCs to enhance MHC presentation. In addition type I IFNs enhance the antiviral function of adaptive immune cells by promoting CD4 + T cells to activate B cells and positively influencing the cytotoxicity of CD8 + T cells and NK cells [4].
Dendritic cells can be divided in two main cell types: conventional DCs (cDCs) and plasmacytoid DCs (pDCs). cDCs are specialized in antigen presentation for T-cell activation. The primary function of pDCs is the secretion of high amounts of type I interferons (IFNα and IFNβ) in response to viruses and/or virus-derived nucleic acids. High levels of type I IFN are observed in the beginning of systemic infections such as early murine cytomegalovirus, vesicular stomatitis virus (VSV), lymphocytic choriomeningitis virus and herpes simplex virus type 1. Furthermore, pDCs play a role in human autoimmune diseases such as systemic lupus erythematosus, psoriasis and type I diabetes [10,11].
The human genome comprises 14 IFNα genes (including one pseudogene) and mouse has 17 IFNα isoforms (including 3 pseudogenes), in contrast to IFNβ that is encoded by a single gene [1,[12][13][14]. Distinct biological activities of the mouse IFNα isoforms such as induction of IFN-stimulated genes, their expression after viral infection and their anti-viral activity have been studied in vitro [15]. The role of IFNα has been studied in infected mice. For example, the reduction of viral titres and of the proportion of infected cells upon treatment of infected mice with different IFNα isoforms have been demonstrated.. In addition, the activation of virus-specific CD4 + /CD8 + T cells and NK cells by IFNα was shown [16][17][18]. Immunomodulatory effects of IFNα subtypes on NK cells in Hepatitis C infected patients have also been described [19,20]. However, the in vivo functions of IFNα and how they are distinguished from other type I IFN members have remained poorly described. There are no IFNα knockout mice available for analysing the role of IFNα during active and chronic viral infection and in autoimmune diseases in vivo. When IFNAR-knockout mice are used, one cannot distinguish between the effects of IFNα and IFNβ.
The function of proteins can be evaluated by knockout techniques such as gene-targeted knockout animals, targeted gene disruption in mammalian cells and knockdown techniques such as siRNA, shRNA, miRNA, CRISPR and TALEN. Complementing these approaches, knockdown of proteins is achieved by dominant negative mutants or inhibitory molecules. In the last years, intracellular antibodies (intrabodies) became increasingly promising due to their high target specificity and the possibility to specifically knockdown proteins with posttranslational modifications. Now it is possible to inhibit virtually any protein of a cell passing the ER or localizing in the nucleus or cytosol using intrabody molecules [21][22][23][24]. ER intrabodies are applied as scFv, which are stable in the oxidizing environment of the ER [25]. Cytosolic and nuclear intrabodies are mainly single domain antibodies derived from camels or sharks, which are correctly folded in the reducing milieu of the cytosol or nucleus [23]. scFvs can be constructed from hybridoma clones or by selection from human antibody repertoires by phage display or bacterial, yeast, mammalian cell surface display or ribosome display [26].
ER intrabodies are expressed and retained inside the ER and inhibit protein function by blocking the translocation of the antigen from the ER to the compartment where it normally acts. Intrabodies expressed in the cytosol or nucleus have neutralizing capacity or induce an inactive conformation of the target [21][22][23]. Even transitory proteins can be inhibited in transgenic ER intrabody mice [27,28]. Besides intrabodies, small alternative protein scaffolds for targeting intracellular proteins have been developed, such as DARPins, ubiquitin variants, monobodies and affibodies [29]. Some non-antibody binders are already being investigated in clinical studies [30].
Here the intrabody technique was used for knockdown of IFNα secretion in RAW 264.7 macrophages and dendritic D1 cells. To realize this strategy, an ER intrabody was constructed from an anti-mouse IFNα rat hybridoma antibody (4EA1) recognizing five mouse IFNα isoforms. This antibody likely also recognizes other IFNα isoforms, which are highly homologous, but this has not been experimentally confirmed until now. 4EA1 has been used frequenly as a general IFNα antibody, e.g. for measuring total IFNα production by virus infected mouse cells including plasmacytoid dendritic cells [31,32]. The newly developed intrabody was able to knockdown IFNα in stimulated RAW 264.7 macrophages and dendritic D1 cells. This suggests that specific abrogation of IFNα signalling might also be achieved in intrabody-expressing transgenic mice [27,28]. This could lead to mouse models for studying the role of IFNα during infections and autoimmune diseases such as systemic lupus erythematosus (SLE) [33].

Cloning of 4EA1 V H and V L
For all PCRs, the Expand High Fidelity Plus PCR System (Roche) was used with 2.5 units Expand polymerase per 50 μl reaction, 0.2 mM of each dNTP and 0.4 μM of each primer.
Construction of the anti-IFNα intrabody was performed following the methodology described in [34]. 60 μg RNA (A 260 /A 280 = 2.1) was isolated from 10 7 cells of the rat antimouse IFNα hybridoma cell line 4EA1 [12] using the Genematrix Universal RNA Purification Kit (Roboklon, Berlin, Germany). First strand cDNA was synthesized with AMV Reverse Transcriptase native, 5x RT buffer (Roboklon, Berlin, Germany) and an oligo(dT) 20 primer. For cDNA synthesis, 1.6 μg mRNA was combined in a total volume of 34 μl with 2.5 μl of 10 μM oligo(dT) 20 primer and 10 μl of dNTPs, 5 mM each. For annealing, this mixture A was heated to 65˚C for 5 min and then placed on ice. For mixture B, 10 μl 5x RT buffer, 2.5 μl RNase inhibitor (12.5 U/μl, Roboklon, Berlin, Germany), 2.5 μl DTT, 100 mM were mixed, placed on ice, followed by addition of 1.3 μl AMV reverse transcriptase (10 U/μl). The mixtures A and B were combined, and cDNA was synthesized by incubating for 15 min at 42˚C and 45 min at 50˚C, followed by storage at -20˚C.
The synthesized cDNA was used as template for PCR amplification of immunoglobulin variable domains of the heavy (V H ) and the light chain (V L ). 7 combinations of a set of 9 degenerate primers were used in PCR reactions [35,36] (Table 1). Antisense primer Bi4 was used with sense primers Bi3, Bi3b and Bi3d for amplification of the V H domain, and antisense primer Bi5 was used with sense primers Bi6, bi7, Bi8 and NS21 for the V L domain. For each primer pair, four PCRs with annealing temperatures from 52 to 65˚C were performed. The reaction volume was 40 μl with 0.8 μl cDNA template. In the PCR reaction, 5 min at 94˚C were followed by 30 cycles of 30 s at 94˚C, 30 s at the annealing temperature (53-64˚C), and 40 s at 72 C, followed by 10 min at 72˚C. PCR products were only obtained with primers Bi7 and Bi5 for V L with annealing temperatures of 52-60˚C and with primers Bi3d and Bi4 for V H and an annealing temperature of 53˚C. PCR products were cloned with the TOPO TA cloning kit for sequencing (ThermoFisher Scientific, Darmstadt, Germany), and the inserts of three V H cDNA clones and two V L cDNA clones were sequenced. The three V H clone sequences and the two V L clone sequences were identical.

Edman sequencing
5 μg purified 4EA1 rat anti-mouse IFNα antibody was prepared in reducing sample buffer and separated by SDS-PAGE. The gel was blotted onto a PVDF membrane. The bands of the antibody's heavy chain and light chain were visualized by Coomassie staining and cut out for sequencing via N-terminal Edman degradation performed on an Applied Biosystems Procise Protein Sequencer 494C with reagents supplied by the manufacturer (ThermoFisher).

Construction of intrabodies
Intrabodies were constructed from the V H and V L sequences of the complete IgG 4EA1 ( Fig  1A). Using assembly PCR [37], V H and V L were linked by the protein sequence (GGGGS) 3 , and a signal peptide was added. The PCR product, comprising signal peptide, V H , (GGGGS) 3 linker and V L , was cloned between the NcoI and NotI sites of the vector pCMV/myc/ER, thereby adding a myc tag and SEKDEL endoplasmic reticulum (ER) retention signal. First, three PCR reactions (i-iii) were set up (25 cycles, 60˚C annealing temperature). (i) The signal peptide was amplified from pCMV/myc/ER with primers CMV-For and scFv-AKG-SP (Table 2). (ii) V H was amplified with primers scFv-AKG-VH-S-G and scFv-AKG-VH-AS. (iii) V L was amplified with primers scFv-AKG-Linker and scFv-AKG-VL-AS. Then, 0.5 μl of PCR products i, ii, iii were added to a 100 μl-PCR reaction with primers CMV-For and scFv-AKG-VL-AS (25 cycles). The purified PCR product was digested with NcoI and NotI and   ligated to pCMV/myc/ER linearized with the same enzymes, creating the plasmid pCMV-αIFNα-ib. The plasmid is available from Addgene.

Cloning of lentivirus vectors
The intrabody gene was cloned into third generation lentiviral transfer vectors for co-expression of the intrabody with eGFP. The vectors pJSemcveGFP and pHR'SIN-SEW [38] were used for expression with a CMV promoter or an SFFV promoter, respectively. A cassette containing the CMV promoter and αIFNα-ib was PCR amplified from pCMV-αIFNα-ib with primers IB-LVfor/rev ( Table 2) and was cloned between the EcoRI and SalI restriction sites of pJSemcveGFP with, upstream of the EMCV IRES and eGFP elements, creating plasmid pLVCMV-αIFNα-ib (S1 Fig). This plasmid was used for expression of αIFNα-ib in RAW264.7 macrophages. The vector pHR'SIN-SEW contains an eGFP gene activated by an SFFV promoter. The eGFP coding sequence was replaced by a cassette of αIFNα-ib, EMCV IRES and eGFP, which was PCR amplified from pLVCMV-αIFNα-ib with primers SEW-up/down ( Table 2). The vector was digested with AscI and SbfI, removing the eGFP ORF, and the cassette was inserted with a QuickFusion kit (Absource, Munich, Germany), creating plasmid pLVSFFV-αIFNα-ib used for expression of αIFNα-ib in dendritic D1 cells.

IFNα expression plasmid
The coding sequence of mouse IFNα4 (GenBank NM_010504.2) was synthesized (GeneArt, ThermoFisher, Regensburg, Germany). PCR amplification with primers IFNA4-S-Esp3I and IFNA4-AS-Tag was followed by digestion with Esp3I and cloning between the BspI sites of the mammalian expression vector pFlpBtM-II, GenBank KC991095 [39], resulting in the plasmid pFlpBtM-IFNα. The plasmid comprises the IFNα coding sequence without the signal peptide with an N-terminal mouse IgG signal peptide (HVM06_MOUSE) and C-terminal TEV-site, Twin-Strep-tag and His 8 -tag.

Cells and viruses
HEK293T cells, murine RAW 264.7 macrophages and NIH/3T3 cells were provided by the German strain collection (DSMZ, Braunschweig, Germany).
For the luciferase assay, 1 ml of 2 x 10 4 IEC-Mx2Luc cells were grown in a 24-well microtiter plate in muINTEPI medium for 48 h at 37˚C with 5% CO 2 . The IFNα containing supernatants of Poly (I:C) stimulated macrophages and dendritic cells were incubated with the IEC-Mx2Luc cells for 24 h to estimate the IFNα content. Vero 4B cells were cultivated in RPMI 1640, 10% FBS, 1% L-glutamine, 100 μg/ml penicillin and 100 μg/ml streptomycin.

Production of recombinant lentiviruses and transduction into RAW267.4 cells and dendritic D1 cells
Production of recombinant retroviral viruses was performed in HEK293T cells using Lipofectamine 2000 (ThermoFisher). HEK293T cells were grown on a culture plate of 10 cm in diameter with a cell density of 5 x 10 6 cells/ml in culture medium without antibiotics. After 24 h, the medium was exchanged with 5 ml fresh medium. Then 2 μg of the packaging plasmid pLP1 (comprising the cDNA of HIV-1 gag and pol, Invitrogen), 4.5 μg of the packaging plasmid pLP2 (comprising the cDNA of HIV-1 rev, Invitrogen) and 2.5 μg of the packing plasmid VSV-G (comprising glycoprotein G of vesicular stomatitis virus, Invitrogen) and 3.0 μg of the lentiviral transfer vector (pLVCMV-αIFNα-ib (S1 Fig), pLVSFFV-αIFNα-ib or the eGFP control vector pHR'SIN-SEW) were diluted in 1.5 ml DMEM medium without FCS and antibiotics. In parallel, 36 μl Lipofectamine 2000 were diluted in 1.5 ml DMEM medium without antibiotics and FCS. After incubation at RT for 5 min, both solutions were combined, shortly shaken and incubated for additional 20 min at room temperature. The DNA-lipofectamin complex containing solution was pipetted dropwise to the cells. After 18 h, the medium was exchanged and cells cultivated for further 72h. Virus-containing supernatants were obtained by centrifugation for 15 min at 3000 rpm and filtered. 1 ml-aliquots were frozen at -80˚C.
For transduction, 264.7 RAW macrophages or dendritic D1 cells were cultivated at a cell density of 10 6 cells/ml culture medium in a 6-well microtiter plate. After 24 h, the culture medium was exchanged with 800 μl of virus-containing medium supplemented with 8 μg/ml polybren (Sigma-Aldrich, USA) (virus-containing supernatant was diluted 1:2 or 1:4 in medium). 6 h after transduction, 1.2 ml fresh medium was added. The next day, the medium was exchanged and after 24 h, 48h and 96h, the transfection rate determined using flow cytometry by analysing eGFP expression.

Transient transfection and production of αIFNα-ab and IFNα4
HEK293-6E-MGAT1-k.o. cells were cultivated in FreeStyle F17 medium (ThermoFisher), supplemented with 7.5 mM L-glutamine and 0.1% Pluronic F-68 (AppliChem, Germany), in the presence of 25 μg/mL geneticin. Transient production was performed as described in [45,46]. The supernatants were harvested at time point 168 h by centrifugation for 30 min at 3000 g and prepared for purification.

Purification of αIFNα-ab and mouse IFNα4
Both proteins comprising C-terminal poly-histidine tags were purified by immobilized metal ion affinity chromatography (IMAC) using a Profinia protein purification system (Bio-Rad Laboratories, Munich, Germany). Before purification, supernatants were concentrated and dialyzed against 150 mM NaCl, 30 mM HEPES, 30 mM imidazole, 10% (v/v) glycerol, pH 8.0 using a hollow fibre module (cut-off 10 kDa, Spectrum Laboratories, CA). The affinity tag of IFNα, comprising His 8 and a Twin-Strep-sequence, was cleaved with an excess of TEV-protease (2 mg/ml) overnight at 4˚C to prevent binding of αIFNα to the Ni-NTA tips during biolayer interferometry. The protease and uncleaved IFNα was separated from IFNα without tag. This was performed by IMAC using a 1 ml HisTrap FF column (GE Healthcare) and the Ä KTAprime Plus protein purification system.

Determination of the binding kinetics of αIFNα-ab using biolayer interferometry
The dissociation constant of αIFNα-ab was measured by biolayer interferometry with the Octet Red96 system (fortéBio, Menlo Park, CA, USA) using Dip and Read Biosensors (forté-Bio) [47]. Kinetic measurements were performed in kinetic buffer (0.2% BSA in PBST, PBST was PBS with 0.05% [v/v] Tween 20). Kinetic buffer without αIFNα-ab was applied to Ni-NTA tips (fortéBio) for 300 sec. Then, αIFNα-ab was coupled in two steps to the Ni-NTA tips. Firstly, the tip was activated for 80 s in 0.1 M EDC, 0.0025 M NHS for covalent coupling and secondly placed for 600 s in a solution of 10 μg/mL αIFNα-ab in kinetic buffer. After covalent binding of αIFNα-ab to the tip, free amino-groups were saturated with 1 M ethanolamine, pH 8.5 for 60 s, followed by equilibration with kinetic buffer for 600 s and recording a baseline for 60 s in wells of a black 96 well microtiter plate (Greiner Bio-One, Austria) to allow subtraction of a baseline drift resulting from unspecific binding of BSA. Tips loaded with αIFNα-ab were then dipped in parallel in 200 μl of increasing concentrations of IFNα in kinetic buffer. The biosensor tips were finally transferred into kinetic buffer to measure the dissociation of αIFNα-ab. Orbital shake speed was 1000 rpm. Fit curves were calculated with the Octet Data Acquisition program 8.2.0.9.

Immunofluorescence
For colocalization analyses, HEK293T cells were grown on sterile coverslips and transiently transfected with αIFNα-ib and mouse IFNα4 expression plasmids. After 48 h, cells were fixed for 15 min with 4% paraformaldehyde at room temperature followed by permeabilisation with 0.1% Triton X-100 in PBS for 15 min. After blocking overnight with 3% BSA in PBS, triple staining of αIFNα-ib, mouse IFNα4 and calnexin was performed with goat anti-c-myc-FITC

Fluorescence microscopy
Confocal analysis was performed with a Zeiss LSM 510 META inverted laser scanning microscope using a Plan-Apochromat 100x oil immersion objective (1.3 numeric aperture). Cells stained with FITC-coupled antibody were excited with an argon laser at 488 nm, and emission was collected using a 505-530 nm bandpass filter. Chromeo 546 and Cy3 were excited using a 543 nm He-Ne laser line and detected using a 560-630 emission filter. Excitation of Cy5-labeled antibody was done with a 633 nm laser line, and emitted light was detected using a 650 nm longpass filter. AMCA was excited by the 364 nm laser line, and emission was collected using a 380-430 nm bandpass filter. Nuclei staining with DAPI was performed for cell detection. Epifluorescence images were acquired with a Zeiss Axiovert 100 microscope using appropriate filter sets for FITC/GFP, Alexa Flour 594 and Cy3 and a Olympus CKX41 microscope using filter sets for FITC and Cy3.

Co-immunoprecipitation
For co-immunoprecipitation, 10 6 HEK 293 T cells transiently cotransfected with pCMV-αIFNαib and pFlpBtM-IFNα expression plasmids were cultured in a 6-well microtitre plate for 48 h. Cells were harvested by centrifugation at 1000 rpm, and the pellet was incubated with 250 μl NP-40 lysis buffer pH 7.6 (150 mM NaCl, 10 mM Tris-HCl, 1% Nonidet P-40, 5mM EDTA; 1 mM PMSF) for 20 min on ice. 25 μL anti-c-myc agarose beads (anti-c-myc Agarose Affinity Gel; Sigma Aldrich) were washed three times with 500 μL PBST buffer and centrifuged for 5 min at 3000 rpm. 600 μL PBST and 60 μL cell culture lysate were added to the agarose beads and incubated overnight at 4˚C. After incubation, the agarose beads were washed four times with 500 μL PBST buffer. Precipitated complexes were eluted by adding 35 μL Laemmli's sample buffer and 25 μL PBST and heating for 5 min at 95˚C. The eluted fractions were then analysed by Western blot using a mouse anti-c-myc antibody (9E10, Santa Cruz, 1:2500) for detection the intrabodies or a mouse anti-StrepTag antibody (StrepMAB classic, IBA Lifescience, 1:2000), for detection of IFNα4. As secondary antibody, a goat anti-mouse IgG Fc ab-alkaline phosphatase conjugated (Promega, 1:7500), was applied, and proteins were stained with BCIP/NPT substrate.
The luciferase assay was performed as follows: IEC-Mx2Luc cells [48] comprising the luciferase reporter gene fused to the Mx2 promoter, inducible by interferon type I and type III, were cultivated in 24 well cell culture microtiter plates with a cell density of 2x10 4 cells/well in 1 ml medium. After the cells reached 60-70% confluence, 200 μl of 1:5 diluted supernatant of stimulated macrophages or dendritic cells were pipetted to the IEC-Mx2Luc cells and incubated for 24 h at 37˚C, 5% CO 2 . Then the supernatants were discarded, the wells were washed once with 1 ml PBS, and cells were lysed with 125 μl 1 x reporter lysis buffer (RLB, Promega). Plates were stored for at least 3 h at -80˚C. For analysing the amount of luciferase, 10 μl cell lysate were added to 100 μl Beetle-Juice (Beetle-Juice Small Kit, PJK GmbH, Kleinblittersdorf, Germany). Luminescence was measured with a luminometer (Lumat LB 9507) for 10 s.
To distinguish between secreted IFNα and IFNβ, the neutralizing αIFNα antibody 4EA1 [12] and a neutralizing αIFNβ antibody, rat hybridoma 7FD3, [12] were applied. 100 μl 4EA1 ascites (1:1000) or 100 μl 7FD3 ascites (1:1000) or 50 μl from both antibodies were added to 100 μl of supernatants of stimulated RAW 264.7 cells or dendritic D1 cells. The supernatants were incubated for 2 h at room temperature and then added to the IEC-Mx2Luc cells. The luciferase assays were performed in duplicate.

Analysis of eGFP expression in RAW 264.7 and D1 cells by flow cytometry
Lentiviral transduced RAW 264.7 and dendritic D1 cells were sorted by eGFP fluorescence with a FACSAria IIu cell sorter (Becton Dickenson). RAW 264.7 cells were sorted three times and D1 cells twice. Non-transduced RAW 264.7 or D1 cells were applied as negative control. After sorting, cells were cultured in medium containing 10 μg/ml gentamicin. Gentamicin was omitted during following cell passages.
Sorted cells were analysed using an LSR II (Becton Dickenson) or a BD FACSCalibure (Becton Dickenson) flow cytometer (S2 Fig). In preparation for flow cytometry analysis, the cells were harvested, washed once with FACS buffer (PBS containing 2% FBS) and counted. Subsequently they were resuspended in 300 μl FACS buffer containing 1 μg/ml DAPI or 1 μg/ml Pacific Blue and analysed.

ELISAs
To determine the amount of mouse IFNα in supernatants of transiently transfected HEK293 T cells, 0.1 μg purified hybridoma antibody 4EA1 in 100 μl PBS were coated overnight at 4˚C on 96-well MaxiSorb polystyrene assay plates (Nunc). Blocking was performed with 100 μl 3% skimmed milk powder in PBST for 1 h at 37˚C. Serial dilutions of supernatants in PBST-3% skimmed milk were then applied for 2 h at room temperature. After washing 5 times for 3 min with PBST, a mouse anti-Strep-tag antibody conjugated with peroxidase (MAB classic HRP, IBA Lifescience, Göttingen, Germany) was added at 1:2000 dilution for 1 h at room temperature. After washing 5 times for 3 min with PBST, 100 μl TMB-solution (Kem-En-Tec Diagnostics, Taastrup, Denmark) was added to develop the signals. The colour reaction was stopped with 50 μl 2 M H 2 SO 4 and absorption was read at 450 nm. Assays were performed in duplicate.
For detection of mouse IFNα secreted by stimulated dendritic D1 cells, a mouse IFNα Platinum ELISA detecting mouse IFNα4 and mouse IFNα2 was used (Affymetrix eBioscience, San Diego, CA, USA) according to the supplier's manual. Assays were performed in duplicate. After 1 h, the supernatants were discarded and 500 μl fresh culture medium with FCS but without antibiotics was added and incubated for 24 h. The supernatants were harvested and serial 1:10 dilutions in 100 μl medium with FCS but without antibiotics were added to Vero 4B cells for 1 h. The Vero 4B cells had been seeded at 2 x 10 4 cells/well of a 96 microtiter plate at the beginning of the assay. After discarding the supernatants, the cells were overlaid with DMEM containing 50% (v/v) of a methylcellulose solution (20 g/L) and incubated for 3-5 days. Plaques were then visualized by staining of living cells with crystal violet.

Generation of an anti-IFNα ER intrabody
To inhibit the secretion of IFNα by an intracellularly produced antibody fragment, an intrabody was designed with the ER retention sequence SEKDEL. The intrabody was made of the variable domain sequences of the anti-mouse IFNα antibody 4EA1 [12]. The variable regions of heavy (V H ) and light (V L ) chains of the antibody, encoding the antigen-binding domains, were cloned and sequenced (Genbank MH201402, MH201403). Cloned sequences were compared to N-terminal protein sequences obtained by Edman degradation of the purified 4EA1 antibody chains. The cDNA sequences were matched to the Edman degradation results in the following cloning steps. The antibody's V H and V L sequences were fused with a linker, generating a single-chain variable fragment (scFv, Fig 1). Two expression plasmids were generated. For intracellular expression of the intrabody, the scFv was fused with the ER retention sequence SEKDEL (αIFNα-ib, Fig 1). For production and purification of the scFv in soluble form for binding studies, the scFv was fused with a His 6 affinity tag in the second plasmid (αIFNα-ab, Fig 1).

Binding kinetics of soluble αIFNα-ab by biolayer interferometry
To verify that the anti-IFNα scFv is functional, mouse IFNα4 and the αIFNα-ab scFv were produced and purified and a binding experiment was performed. IFNα4 and the soluble scFv αIFNα-ab were purified from supernatants of transiently transfected HEK293-6E-MGAT1-k. o. cells by nickel affinity chromatography. Subsequent size exclusion chromatography resulted in a major monodisperse elution peak for each protein.
The binding of αIFNα-ab to IFNα4 was analysed by Biolayer interferometry using an Octet RED instrument. Binding of molecules to this instrument's biosensors causes a wavelength shift in the interference pattern, which can be measured in real time. The kinetics of binding and dissociation of IFNα4 to αIFNα-ab immobilized on biosensors was measured at different IFNα4 concentrations. Sensorgrams were recorded consisting of the wavelength shift in the interference pattern over time (Fig 2). The sensorgrams were evaluated assuming a 1:1 interaction and theoretical curves were fitted to the sensorgrams by the instrument software. A reliable curve fit is characterized by the statistical parameters R 2 (>0.95) and Χ 2 (<3) [47]. Curve fitting resulted in K D = 38.6 nM for the binding of αIFNα-ab to IFNα4, with R 2 = 0.998 and Χ 2 = 1.2563 (Table 3), indicative of moderately strong binding between the scFv and the antigen.

Colocalization and binding of intrabody and IFNα4 inside the ER
The intracellular localization of intrabody and IFNα4 was analysed by immunofluorescence of HEK293T cells co-transfected with the IFNα4 expression construct and αIFNα-ib. Cell staining demonstrated colocalization of αIFNα-ib and IFNα4, indicating complex formation of both molecules (Fig 3A). An anti-TLR2 intrabody, αTLR2-ib [49], was used as a negative control and did not have this effect (Fig 3B). Specific intracellular binding of αIFNα-ib and IFNα4 was further demonstrated by co-immunoprecipitation (Fig 3D). Control intrabodies, αTLR2-ib and αNCAM-ib [50], did not precipitate IFNα4, demonstrating specificity of the assay. Another negative control, transfection with only the IFNα4 expression plasmid, did not result in immunoprecipitated IFNα4. The αIFNα-ib:IFNα4 complex colocalized with the ER marker calnexin. Triple staining revealed a lattice structure typical for the ER. These results indicate that the intrabody-antigen complex was localized inside the ER.

Inhibition of IFNα4 secretion by HEK293T cells after co-transfection of αIFNα-ib and IFNα4 expression plasmids
αIFNα-ib-mediated inhibition of IFNα4 secretion by HEK293T cells was measured with an IFNα4 ELISA. HEK293T cells were transiently transfected with the IFNα4 expression construct and αIFNα-ib. αTLR9-ib and αNCAM-ib were used as negative controls. Without αIFNα-ib, high amounts of IFNα were secreted and ELISA signals were only linear up to a dilution of 0.03. Only very low amounts of secreted IFNα4 were detectable in the presence of the αIFNα-ib intrabody, whereas expression of the other intrabodies had no effect (Fig 4). Thus it could be demonstrated that IFNα4 secretion was strongly inhibited by the intrabody.

Characterization of stably transfected intrabody-expressing RAW 264.7 macrophages and dendritic D1 cells
Stably transfected RAW 264.7 macrophages co-expressing αIFNα-ib and eGFP from a CMV promoter were generated by lentiviral transduction and cell sorting (RAW-αIFNα-ib-eGFP). RAW-eGFP control cells, expressing only eGFP, were created in parallel (S2 Fig). Expression of αIFNα-ib was demonstrated by immunoprecipitation. (Fig 5). It is clearly visible that αIFNα-ib is expressed in RAW-αIFNα-ib-eGFP cells.
The D1-αIFNα-ib-eGFP cells were made by transducing D1 dendritic cells with lentiviral particles for co-expression of αIFNα-ib and eGFP, mediated by an SFFV promoter. D1-eGFP control cells expressing only eGFP were also established (S2 Fig). Expression of αIFNα-ib was verified by immunoblot analysis. (Fig 5). The calculated molecular weight of αIFNα-ib, 28 kDa, was confirmed.
To verify the localization of the intrabody-IFNα complex in the ER of stably transfected intrabody expressing immune cells, the stably transfected intrabody-expressing macrophage and dendritic cells were stimulated with Poly (I:C). Intracellular IFNα and intrabody expression were detected by immunofluorescence in both cell populations. Both molecules colocalized with the ER marker calnexin (Fig 6), indicating complex formation of intrabody and IFNα in the ER.

Intrabody-mediated inhibition of IFNα secretion by RAW 264.7 macrophages and dendritic D1 cells
The efficiency of the intrabody to inhibit IFNα secretion by RAW 264.7 macrophages was analysed using the RAW-αIFNα-ib-eGFP cells and, as controls, the RAW 264.7 wild type macrophages and the RAW-eGFP cells. The activity of secreted type I IFNs was measured in cell supernatants with an Mx2 promoter-dependent luciferase assay. Supernatants were treated with neutralizing antibodies against IFNα or IFNβ for measuring the IFNs specifically. Neutralization of IFNβ allowed measuring mainly IFNα activity and vice versa. Treatment with neutralizing antibodies against IFNα did not reduce IFN activity in the supernatants of wild type and eGFP expressing RAW cells, indicating that IFNβ was the dominant type I IFN (Fig  7). Mx2 promoter activity was lost when both IFN types were neutralized. In contrast, the  αTLR2-ib, B). Expression of IFNα4 and αIFNα-ib was analysed with StrepMAB Classic Chromeo 546 antibody and goat anti-c-myc-FITC antibody, respectively. Expression of calnexin was visualized with rabbit anti-calnexin and goat anti-rabbit Cy5 antibodies. C, As a negative control, cells were co-transfected with the IFNα4 expression plasmid and the empty intrabody vector pCMV/myc/ER. Staining as in A resulted in no signal. D, Co-immunoprecipitation. HEK293T cells were co-transfected with the αIFNα-ib and IFNα4 plasmids. Negative controls were transfected with IFNα4 and the unrelated, myc-tagged intrabodies αTLR2-ib or αNCAM-ib or the empty vector pCMV/myc/ER. 48 h after transfection, co-immunoprecipitation was performed with anti-myc agarose beads, and eluted proteins were visualized by immunoblotting using an anti-c-myc antibody for detection of the intrabodies or an anti-StrepTag antibody for detection of IFNα4. https://doi.org/10.1371/journal.pone.0215062.g003 IFNβ-neutralizing antibody completely abolished reporter gene induction by supernatants of RAW-αIFNα-ib-eGFP cells, indicating that no detectable IFN activity remained. It was concluded that the intrabody mediated pronounced inhibition of IFNα secretion without affecting IFNβ activity. In a second set of experiments, it was analysed whether αIFNα-ib is able to inhibit secretion of IFNα by dendritic D1 cells. Retention of IFNα by αIFNα-ib was clearly demonstrated by immunofluorescence (Fig 8). Intracellular IFNα was only detectable in Poly (I:C)-stimulated D1-αIFNα-ib-eGFP cells expressing the intrabody. No retained IFNα could be seen in the control D1-eGFP cells or non-transduced D1-wt cells. Retention of IFNα by αIFNα-ib was further confirmed by ELISA measurements of secreted IFNα in D1 cell supernatants (Fig 9). The amount of secreted IFNα in the stimulated D1-αIFNα-ib-eGFP cells was greatly reduced in comparison to the control cells and was similar to the amount obtained from unstimulated D1 cells. In summary, the intrabody effectively inhibited IFNα secretion by Poly (I:C)-stimulated RAW264.7 macrophages and dendritic D1 cells.

Effect of the intrabody on VSV-AV2 replication in RAW 264.7 and D1 cells
The intrabody's effect on virus replication upon infection of RAW 264.7 macrophages and D1 cells with VSV-AV2 was tested. Upon infection, the VSV mutant AV2 induces type I IFN production and viral titre increases in IFNAR-deficient cells [51]. An IFNβ-neutralizing antibody accordingly lead to an increase in VSV-AV2 titre in wild type and eGFP expressing cells at low MOIs (Fig 10). In contrast, virus replication in cells expressing the αIFNα intrabody and treated with the neutralizing IFNβ antibody was also enhanced at high MOIs (MOI 1 for RAW

Fig 4. The intrabody inhibits recombinant IFNα4 secretion by co-transfected HEK293T cells.
Serial dilutions of supernatants of HEK293T cells transiently co-transfected with Twin-Strep-tagged IFNα4 expression plasmid and αIFNα-ib or control intrabody expression plasmids or empty vector pCMV/myc/ER were applied to ELISA plates immobilized with purified anti-IFNα hybridoma antibody 4EA1. Secreted IFNα4 was detected with anti-StrepTag peroxidase-labelled antibody. Bars indicate standard deviations (SD) calculated from two independent experiments. IFNα4 concentration differences between the 1:32 diluted supernatants of the αIFNα-ib plasmid-treated cells (asterisk) and supernatants treated with any of the three other plasmids were statistically significant with p < 0.05 (n = 2, one way ANOVA followed by Holm-Sidak test).
https://doi.org/10.1371/journal.pone.0215062.g004 264.7 cells and MOI 10 for DCs) and was even stronger at low MOIs compared to control cells (Fig 10, red frames). Together it was concluded that intrabody-mediated inhibition of IFNα secretion removes the virus-inhibitory effect of IFNα.

Discussion
IFNα activates intracellular anti-viral and anti-bacterial programs and is involved in the development of innate and adaptive immune responses [4]. Furthermore, dysregulation of IFNα responses can contribute to autoimmune diseases and chronic viral infections [5,6]. IFNαβ has well-documented antiviral effects, and indications for IFNα treatment are chronic viral hepatitis, haematological disorders and solid tumours. Independently of the form (IFNα can be given alone or conjugated to PEG) and administration route, side effects including acute toxicity have been reported [52]. The underlying mechanism of some side effects have been illuminated. For example, Davidson et al. showed that IFNα and β mediate increased mortality in influenza infected mice [53]. They demonstrated that excessive amounts of IFNα and β are secreted by plasmacytoid DCs of influenza infected mice, leading to interaction of the deathinducing receptor DR5 on lung epithelia cells with Trail on inflammatory monocytes, which results in lung tissue damage.
The anti-viral properties of IFNα have been studied in vitro and with a few infectious mouse models [15][16][17][18]. Nevertheless only little is known about the mechanism of in vivo  RAW-αIFNα-ib-eGFP macrophages and D1-αIFNα-ib-eGFP dendritic cells were stimulated for 24 hours with Poly (I:C). After fixation and permeabilization of cells, endogenous IFNα, αIFNα-ib and calnexin were stained with rat monoclonal anti-IFNα F18, mouse anti-myc and rabbit anti-calnexin, followed by incubation with goat anti-rat Cy3, goat anti-mouse AMCA and goat anti-rabbit Cy5. Merged images of IFNα and intrabody, and of all three proteins were generated.
https://doi.org/10.1371/journal.pone.0215062.g006 A, Mx2-dependent luciferase assay. RAW-wt, RAW-eGFP and RAW-αIFNα-ib-eGFP cells were stimulated with Poly (I:C) and the concentration of IFNα/β in the supernatants was quantified by an Mx2-dependent luciferase assay. In addition, IFNα or IFNβ or both interferons were neutralized by addition of the neutralizing antibodies 4EA1 and 7FD3 to the supernatants, as indicated. The diagram shows means ± SD, analysed by two way ANOVA followed by Holm-Sidak test, n = 2, � p < 0.05, �� p < 0.01. Stimulations assays were performed in duplicate and IFNα/β was measured with triplicate luciferase assays.
https://doi.org/10.1371/journal.pone.0215062.g007 action of IFNα during viral infections, particularly in macrophages and dendritic cells. Gene knockout mice established by gene targeting or CRISPR/CAS technology are valuable for studying the function of proteins in vivo. Mice have 17 genes of IFNα isoforms, including 3 pseudogenes. Genetic knockout of all genes would require a large effort. Therefore an IFNα knockout mouse has not been reported yet. The intrabody technology was used here as an alternative to a genetic IFNα knockout. It is the first intrabody inhibiting a secretory antigen in dendritic cells.
The K D of αIFNα-ab to mouse IFNα4 was estimated by biolayer interferometry to 38.6 nM. This value is moderate for an scFv. Interestingly the K D values of an antibody Fab fragment broadly neutralizing human IFNα isoforms were estimated to range from 0.018 nM to 35.4 nM for different IFNα isoforms [54]. The binding affinity of the αIFNα-ab scFv lies between the binding affinities which were estimated for 12 human IFNα subtypes to IFNAR1 and IFNAR2, namely 0.5-5 μM for IFNAR1 and 0.4-5 nM for IFNAR2 [55]. It is known that high antigen affinity is not required for an antibody to function efficiently inside the ER as an ER intrabody [56,57]. This might be explained by the relatively high concentrations of intrabody and antigen in the ER compartment. It could be shown by ELISA that secretion of recombinant IFNα4 is inhibited by the αIFNα-ib intrabody in HEK293T cells (Fig 4). Retention of IFNα4 in the ER was based on intracellular binding of both partners, as demonstrated by co-immunoprecipitation (Fig 3). Co-staining of the ER marker calnexin with the intrabody and IFNα4 revealed that the antigen-intrabody complex is localized inside the ER (Fig 6). Similar results had been demonstrated with anti-TLR2, anti-TLR9 and anti-polysialyltransferase intrabodies [34,49,50].
Stably transfected cell pools were generated for demonstrating intrabody-mediated inhibition of IFNα secretion by macrophages and dendritic cells. For generation of stably transfected RAW 264.7 macrophages, a lentiviral transfer vector was used for bicistronic expression of the intrabody and eGFP by a CMV promoter. Transduction of dendritic D1 cells with the same vector lead to only 2% eGFP positive cells. One explanation could be that the CMV promoter was inactivated in dendritic D1 cells due to methylation, as demonstrated in some CHO cell lines [58,59]. For generating the intrabody-producing D1 cells, another vector, based on pHR'SIN-SEW and containing the retroviral spleen focus-forming promoter (SFFV), was constructed. The new vector lead to much better eGFP expression and resulted in 13% eGFP positive D1 cells. Other publications reported 5-20% of GFP expression after retroviral transduction of mouse dendritic cells [60,61]. In addition to the SFFV promoter, the new vector contains the woodchuck hepatitis post-transcriptional regulatory element (WPRE), which favours gene expression and mRNA stability [62]. This could also explain the vector's higher efficiency. Interestingly, transduction of D1 cells with the pHR'SIN-SEW control vector resulted in about 34% eGFP positive cells, and these cells expressed higher levels of eGFP (Fig  8). A possible reason is that the distance of the IRES sequence to the first codon of the eGFP cistron was not optimal in the bicistronic construct [63].
The inhibition of IFNα secretion by intrabody-expressing RAW 264.7 macrophages was clearly demonstrated by an Mx2 reporter-dependent luciferase assay (Fig 7). Analysis showed that less type I IFN was found in supernatants of RAW-eGFP cells compared to RAW 264.7 wild type cells. The overexpression of eGFP could have had an influence on cytokine expression. Whether the insertion of small HIV genome elements and the Hepatitis B virus posttranscriptional regulatory element from the lentiviral vector have also an effect on cytokine expression remains unknown. Inhibition of IFNα secretion by intrabody-expressing dendritic D1 cells was also clearly demonstrated (Figs 8 and 9).
At the moment it is not known which IFNα subtypes are secreted after stimulation of RAW 264.7 macrophages and dendritic D1 cells with Poly (I:C) or upon VSV-AV2 infection. Clearly, IFNα secretion is effectively inhibited by the αIFNα-ib intrabody. Preliminary results with VSV-AV2 infected cells showed a significant influence of the intrabody on virus replication by RAW 264.7 macrophages and dendritic D1 cells under IFNß-neutralizing conditions (Fig 10).
In the paper significant inhibition of IFNα secretion by RAW264.7 macrophages and dendritic D1 cells has been demonstrated. However, before establishing a transgenic intrabody mouse, it should be analysed whether the intrabody is also able to efficiently inhibit IFNα secretion in plasmacytoid dendritic cells after viral infection. A further prerequisite for intrabody functionality in a transgenic mouse would be that it recognizes all IFNα isoforms. The 4EA1 antibody was used in several studies for detection of total IFNα [31,32]. Nevertheless, it would be recommended to clarify this before establishing transgenic intrabody mice.
In summary, the new intrabody can be used in vitro for evaluating the effect of IFNα on virus infection in macrophages and dendritic cells. IFNα knockdown in transgenic intrabodyexpressing mice could enable studying the role of IFNα in virus infections and in autoimmune diseases, such as SLE [33].