BioSOX2 and birA mice were previously generated and maintained on C57/Bl6 background under standard conditions and experiments were performed following guidelines of the ethics committee of the Erasmus Medical Center (26, 27).

Lungs from wild type mouse embryos from E11 till E18 were dissected and fixed in 4% PFA overnight at 4°C before processing for paraffin embedding. Sections of 5 μm were dewaxed and rehydrated, followed by antigen retrieval with microwave treatment in 10 mM Tris-HCl, 1 mM EDTA pH9.0 (TE). After quenching endogenous peroxidase with 1.5% H₂O₂, samples were blocked with 1% BSA, 0.05% Tween20 in PBS for 10 min at room temperature and then incubated with antibody against CHD4 (Abcam; ab72418), FOXP2 (Abcam; ab16046), FOXP4 (Santa Cruz; sc-292474), or CUX1 (Abcam, Ab54583) diluted in blocking buffer at 4°C overnight. After washing in PBS with 0.05% Tween20, slides were incubated in blocking buffer with biotin conjugated secondary antibody for 30 min at room temperature. Then samples were incubated with VECTASTAIN^® ABC Reagent (Vector) for 30 min, followed by incubation with diaminobenzidine (Fluka). Slides were counterstained with hematoxylin and mounted with pertex. Sections were analyzed using Olympus BX41 microscope.

Lungs from wild type mouse embryos from E11 till E18 were dissected and fixed in 4% PFA overnight at 4°C before processing for paraffin embedding. Sections of 5 μm were dewaxed and rehydrated, followed by antigen retrieval with microwave treatment in TE. Samples were blocked and incubated with antibodies against CHD4 (ab72418) and SOX2 (GTI5098), FOXP2 (ab16046) and SOX2, and CUX1 (Abcam, Ab54583) and SOX2 diluted in blocking buffer at 4°C overnight. After washing in PBS with 0.05% Triton X-100, slides were incubated in blocking buffer with fluorophore conjugated secondary antibody (Alexa Fluor-488, Alexa Fluor-594; Jackson Immuno Research) for 1 h at room temperature. Slides were mounted in Vectashield Mounting Medium with Dapi (Vector laboratories, Burlingame, CA, United States). Digital images were captured using a ZEISS imager Z1 AX10 microscope.

HEK293 (HEK) cells were cultured in DMEM (Lonza, Verviers, Belgium) with 5% fetal calf serum and 1% penicillin-streptomycin under standard culture conditions. NCCIT cells were cultured in DMEM (Lonza, Verviers, Belgium) with 10% fetal calf serum and 1% penicillin-streptomycin under standard culture conditions.

HEK cells were transfected with different combinations of pcDNA3 based plasmids expressing FLAG-tagged CHD4, FLAG-HA-tagged WDR5, myc-tagged SOX2, myc-tagged FOXP2, myc-tagged FOXP4, myc-tagged TCF3, and FLAG-tagged SOX2, using X-tremeGENE HP DNA Transfection Reagent (Roche, Basel, Switzerland). Similar co-transfections and co-immunoprecipitations were performed using pcDNA3-myc-HA-tagged CUX1 and different SOX2 constructs that were described previously (24, 28–34). X-tremeGENE HP DNA Transfection Reagent (Roche, Basel, Switzerland) was used for the transfection according to the manufacturer’s manual. Cells were harvested 24 h after transfection. Total cell extracts were prepared in 300 μl cell lysis buffer (20 mM Tris pH8, 137 mM NaCl, 10 mM EDTA, 1% NP40, 10% glycerol) with Complete protease inhibitor (Roche, Basel, Switzerland). 50 μl extract was incubated for 2 h at 4°C in 250 μl cell lysis buffer with antibodies against myc (Roche; 1668149) and FLAG (Sigma-Aldrich; F1804), followed by 1 h incubation with protein G beads (Sigma-Aldrich, St. Louis, MO, United States). After washing with cell lysis buffer, the beads were resuspended in 20 μl sample buffer and heated for 10 min at 95°C.

Samples were separated on a SDS-PAGE gel and transferred to a PVDF membrane by wet blotting for 2 h at 100 V and 400 mA. Membranes were blocked in TBS with 0.05% Tween-20 and 5% BSA. Membranes were labeled with antibodies against myc (Abcam; ab9106) and FLAG (Sigma-Aldrich; F7425) for 2 h, followed by an 1 h incubation with secondary HRP-labeled antibody. Membranes were developed with ECL incubation (Thermo Fisher Scientific INC., Waltham, MA, United States) on an Alliance Imager (Uvitec, Cambridge, United Kingdom).

Generation of glutathione S-transferase (GST) fusion-tags and bacterial lysates is previously described (24, 35). HEK cells were grown ∼85% confluent and nuclear extracts were prepared. 10 μg of the bacterial lysate was incubated with 25 μl glutathione sepharose UB beads (GE Healthcare)/1% fish skin gelatin for 2 h rotating at 4°C. Beads are washed 4x with bacteria lysis buffer (20 mM Hepes-NaOH pH7.6, 150 mM KCl, 10% glycerol, 0.5 mM EDTA, 10% triton; 0.5 mM DTT + 1x CEF prior to use), 2x with wash buffer (20 mM Hepes pH7.6, 20% glycerol, 100 mM KCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.02% NP-40; 1x CEF prior to use) and are then incubated for 2 h with nuclear cell extracts rotating at 4°C. After washing 4 x with wash buffer, 40 μl 2x sample buffer is added and beads are heated for 10 min at 95°C. Western blotting was used to detect SOX2 binding partners.

∼120 × 10⁶ NCCIT cells were double cross-linked with 2 mM disuccinimidyl glutate (Thermo Fisher Scientific) and 1% formaldehyde, or single cross-linked with 1% formaldehyde (36). Cells were lysed in ChIP cell lysis buffer (10 mM Tris pH 8.0, 10 mM NaCl, 0.2% NP-40, 1x CEF), followed by lysis of the nuclei in ChIP nuclei lysis buffer (50 mM Tris pH 8.0, 10 mM EDTA, 1% SDS, 1x CEF). Samples were sheared with a multiprobe bioruptor (double cross-linked: 75 min, 30 s high, 30 s off; single cross-linked: 30 min, 30 s high, 30 s off). Samples were diluted 10x with ChIP dilution buffer (16.7 mM Tris-HCl pH 8.1, 167 mM NaCl, 1.2 mM EDTA, 1.1% TritonX-100, 0.01% SDS), precleared and incubated O/N with antibodies against SOX2 (#2748 cell signaling) and CHD4 (Abcam, ab72418), and goat IgG (Santa Cruz) and rabbit IgG (Santa Cruz) as negative control. Samples were incubated with Protein A/G agarose beads 1 h rotating at 4°C and then washed with Low Salt Immune Complex Buffer (20 mM Tris-HCl pH8.0, 150 mM NaCl, 2 mM EDTA, 0.1% TritonX-100, 0.1% SDS), High Salt Immune Complex Buffer (20 mM Tris-HCl pH8.0, 500 mM NaCl, 2 mM EDTA, 0.1% TritonX-100, 0.1% SDS), LiCl Immune Complex Buffer (10 mM Tris-HCl pH8.0, 1 mM EDTA, 0.25 M LiCl, 1% NP-40, 1% deoxycholate), and TE (10 mM Tris-HCl pH8.0, 1 mM EDTA pH 8.0). Bound chromatin was eluted with freshly prepared elution buffer (1% SDS, 0.1 M NaHCO₃), de-cross-linked and purified DNA was sequenced. For ChIP-PCR, the purified DNA was analyzed using gene-specific primer sets in a qPCR and the ΔC(t) method was used to determine the enrichment of the SOX2 and CHD4 precipitated DNA relative to the IgG control samples. ChIP-seq analysis was mainly performed as previously described (37). Peaks were called with the following criteria: peak ≥ 20, FDR ≤ 0.001, fold change over control ≥ 2. The iRanges package and the “findOverlaps” function from the GenomicRanges package (38) was used to identify the target genes and the overlap between the binding regions.

The transcription factor SOX2 is expressed early in the developing lung and is one of the first molecular signs of epithelial cells that are committed to the proximal airway fate. To identify molecular switches that hallmark the differentiation of the proximal cell toward the different cell lineages eventually covering the upper airways, such as ciliated cells, club cells and basal cells, we generated a mouse line expressing a biotinylatable tagged SOX2 (26). An additional mouse strain ubiquitously expressing the bacterial biotin birA ligase was used to biotinylate the bioSOX2 protein in vivo (27). The biotin-tagged SOX2 is subsequently purified with high affinity using streptavidin, which has a much higher affinity than regular antibodies have for antigens (39). Using this mouse strain, we performed a large-scale screen to identify in vivo SOX2 interacting proteins using embryonic day 18.5 (E18.5) mouse lungs. Biotinylated SOX2 and associated proteins were precipitated with streptavidin magnetic beads and subsequently identified by mass spectrometry. A list of possible candidate partners of SOX2 was created based on the mascot score and peptide enrichment. From this list, a select number of proteins that could be of importance for lung epithelial differentiation were selected for further analysis (Table 1).

As the screen for potential SOX2 interactors was based on nuclear protein extracts of total embryonic lungs, we first analyzed the spatial and temporal expression pattern of selected proteins during lung development in mice using immunohistochemistry. The Forkhead box protein FOXP2 expression was restricted to the distal lung epithelium during the late pseudoglandular, canalicular, and saccular phases of lung development (Figures 1A–C) and FOXP4 expression was first detected at E14 in the epithelium of the developing airways and the surrounding mesenchyme, but the expression became restricted to the proximal and distal airway epithelium during the canalicular phase (40). A similar expression pattern persisted in the saccular phase and postnatally (Figures 1D–F). FOXP2 and SOX2 were co-expressed in cells localized at the branching region between the proximal and distal region during the embryonic and early pseudoglandular phases of lung development (Figures 2E,F; arrow). During later stages of lung development, FOXP2 is restricted to the distal epithelial cells, while SOX2 is restricted to the proximal epithelial cells (Figures 2G,H). CHD4, one of the catalytic subunits of the nucleosome remodeling and deacetylase (NuRD) complex, was detectable in the embryonic lung until E16, but became significantly expressed in the epithelial cells of the proximal airways from the canalicular stage onward (Figures 1G–I). In the adult lung, CHD4 remained expressed in the epithelial cells of the large airways albeit at lower level (data not shown). CHD4 colocalized at E18 with SOX2 in the epithelial cells lining the proximal airways, thus, at the final stages of embryonic lung development, the spatial and temporal expression patterns of SOX2 and CHD4 overlap (Figures 2A–D; arrows). The Cut-Like Homeobox1 (CUX1) was initially expressed in the mesenchyme at the earliest stages of lung development until E16, but during the canalicular and saccular phases of the lung significant expression was detected in the epithelial cells of the airways (Figures 1J–L). In the adult mouse lung, CUX1 remained exclusively expressed in the epithelial cells of the large airways. CUX1 expression colocalized with SOX2 in epithelial cells of the large conducting airways at the canalicular stage (E16.5) and saccular stage (E18.5, Figures 2I–L). In conclusion, the spatial and temporal expression of these potential SOX2 interacting proteins correspond to the pattern of SOX2 expression.

Since the spatial and temporal expression of SOX2 and its putative partners overlap, the direct interaction between the proteins was validated. Therefore, transient transfections were performed using constructs to express epitope-tagged proteins in different combinations. Immunoprecipitation of a FLAG-tagged SOX2 and either a myc-tagged isoform of FOXP2 or FOXP4 showed that SOX2 interacts with FOXP2 and FOXP4 (Figures 3A,B). Comparable immunoprecipitations were performed to validate the interaction between SOX2 and CHD4 using a FLAG-tagged CHD4 and a myc-tagged SOX2 protein, indicating that SOX2 and CHD4 also interact (Figure 3C). The interactions between SOX2 and TCF3 or WDR5 were used as positive controls, as we previously showed that these proteins interact with SOX2 (26).

CUX1 is a large, 200 kDa protein, which is proteolytically processed both at the N- and C–terminus into several isoforms (41). We used a full length CUX1 expression construct containing an N-terminal myc-tag and a C-terminal HA-tag and analyzed which isoforms would be precipitated with the antibodies against the different tags (41). Immunoprecipitation of extracts from transient transfected HEK cells using antibodies against the myc- or HA tag showed that the full length CUX1 protein is proteolytically processed at both termini, as the myc and HA antibody detected different CUX1 isoforms (Figures 4A,B). The myc antibody detected isoforms containing the N-terminus, such as the full-length CUX1 protein (p200) and a p55 isoform, whereas the HA antibody detected different isoforms with the C-terminus intact (Figure 4A, asterisk). Thus, the full length p200 isoform generates multiple isoforms, presumptively corresponding to the p150, p110, p90, p80, and p75 isoforms as previously shown (Figure 4A, bracket; 41). In addition, a commercial CUX1 specific antibody raised against amino acids 521-621, a region between cut repeat 1 and 2 (CR1 and CR2), primarily recognized the p200 isoform (Figure 4A, right panel). To avoid missing specific interactions between CUX1 and SOX2, we used this antibody and another commercial antibody raised against the C-terminus of CUX1 to perform immunoprecipitations. Both antibodies precipitated the transiently transfected CUX1 protein, and efficiently co-precipitated the SOX2 protein, indicating that CUX1 and SOX2 interact (Figure 4B). Given the complex pattern of processed CUX1 isoforms, we decided to perform a streptavidin precipitation of embryonic trachea and brain protein extracts from birA and bioSOX2/birA mice to verify the interaction between SOX2 and CUX1. Comparing the birA with the bioSOX2/birA, it is clear that the biotinylated SOX2 is specifically precipitated, although the trachea showed lower amounts reflecting the fact that the trachea consists of several cell types, of which some express SOX2 (Figures 4C,D). Concomitantly, CUX1 is unambiguously co-precipitated, albeit more abundantly in the brain than in the trachea (Figures 4C,D, arrows). Finally, a bacterial produced GST-SOX2 fusion protein was used to investigate the nature of the interaction between SOX2 and CUX1 (24, 35). Nuclear extracts of HEK cells were incubated with GST-only or the GST-SOX2 fusion protein, resulting in a SOX2 specific precipitation of the p75 CUX1 isoform (Figure 4E). Collectively, these data show that CUX1 directly interacts with SOX2 in the developing airway epithelium.

Previously, we showed that the HMG domain of SOX2 is important for the interaction with another interacting protein, XPO4 (24). Therefore, we used three SOX2 deletion proteins to analyze potential domains of interaction (Figure 5A). As we already showed the interaction of SOX2 with the full length CUX1, we also used two constructs expressing CUX1 truncated proteins based on their natural processing (33; Figure 5B). The full length SOX2 interacts with both CUX1 isoforms, but deletion of the transactivating domain of SOX2 results in reduced association of the short CUX1 isoform. This is even more pronounced when the amino acids 8–40 are removed (F-SOX2-ΔTAD2). Removal of the HMG domain completely abolishes the interaction with the CUX1 isoforms (Figure 5C). Collectively, these data show that the N-terminal and DNA-binding HMG domain of SOX are important for the interaction with CUX1. Moreover, the CUX1 domains that interact with SOX2 are within the short isoform of CUX1 containing the CR2, CR3, and HD domains.

Lastly, we performed a chromatin immunoprecipitation (ChIP) followed by sequencing, to identify if the SOX2 interacting protein CHD4 would also bind to DNA in close vicinity of SOX2. Therefore, the human pluripotent embryonal teratocarcinoma NCCIT cell line, which endogenously expresses SOX2 and CHD4, was used (Figure 6A). In total, over 6000 SOX2 specific binding sites were identified and just over 400 sites for CHD4. Gene ontology analysis showed that some of the identified binding sites that were in close proximity to transcriptional start sites in the SOX2 ChIP are associated with lung development, such as ETV4, JAG1, GLI2, SOX21, TP63 (ΔNTP63 isoform promoter), and NOTCH2 (Figure 6B, top to bottom). Other interesting lung-associated genes are HHIP, PDPN, BMPR1A, and EPAS1. For CHD4 we identified fewer binding sites, of which FOXK2, CDH15, and S1PR4 are shown (Figure 6C, top to bottom). Of the specific binding sites, 145 binding sites were shared by SOX2 and CHD4, of which RIF1, STRA6, and KNG1 are interesting examples (Figure 6D, top to bottom). Overall, we highlight here several examples of genes which may be regulated by SOX2 alone or in combination with CHD4.