The cholesterol-binding protein NPC2 restrains recruitment of stromal macrophage-lineage cells to early-stage lung tumours

The tumour microenvironment is known to play an integral role in facilitating cancer progression at advanced stages, but its function in some pre-cancerous lesions remains elusive. We have used the V600EBRAF-driven mouse lung model that develop premalignant lesions to understand stroma–tumour interactions during pre-cancerous development. In this model, we have found that immature macrophage-lineage cells (IMCs) producing PDGFA, TGFβ and CC chemokines are recruited to the stroma of premalignant lung adenomas through CC chemokine receptor 1 (CCR1)-dependent mechanisms. Stromal IMCs promote proliferation and transcriptional alterations suggestive of epithelial–mesenchymal transition in isolated premalignant lung tumour cells ex vivo, and are required for the maintenance of early-stage lung tumours in vivo. Furthermore, we have found that IMC recruitment to the microenvironment is restrained by the cholesterol-binding protein, Niemann-Pick type C2 (NPC2). Studies on isolated cells ex vivo confirm that NPC2 is secreted from tumour cells and is taken up by IMCs wherein it suppresses secretion of the CCR1 ligand CC chemokine 6 (CCL6), at least in part by facilitating its lysosomal degradation. Together, these findings show that NPC2 secreted by premalignant lung tumours suppresses IMC recruitment to the microenvironment in a paracrine manner, thus identifying a novel target for the development of chemopreventive strategies in lung cancer.


Isolation of lung tumour and stroma cells
Harvested lungs were minced and incubated in Buffer-A [RPMI1640 (Invitrogen) with 5% FCS (Sera Laboratories International), 1 mg/ml collagenase (Sigma), and 20 mg/ml DNase (Sigma)] for 30 min at 37°C. After gentle pipetting, digested tissues were filtered through a 100 mm nylon mesh, overlaid onto 50% Histopaque-1119 (Sigma) and centrifuged at 600 x g for 20 min. The resultant pellet was treated with red blood cell lysis buffer, re-suspended in DMEM (Invitrogen) with 10% FCS, and incubated for 30 min to obtain adherent stroma cells enriched for CD11c + CD11b low IMCs. For lung epithelial cell isolation, undigested tissues after 30 min incubation in Buffer-A above were washed 3 times in PBS and incubated in Buffer-B [RPMI1640 with 1 mg/ml elastase (Worthington Biochemical) and 20 mg/ml DNase] for 90 min at 37°C. After vigorous pipetting, digested tissues were filtered through a 100 mm nylon mesh, mixed with FCS to inactivate elastase, and incubated for 30min on culture plates.
Non-adherent cells after 30 min incubation were overlaid onto 40% Histopaque-1119 and centrifuged at 600 x g for 20 min. The resultant pellet was treated with red blood cell lysis buffer, re-suspended in DMEM with 10% FCS, and incubated again for 1 hr on culture plates. Non-adherent cells at this stage were enriched for AT2 cells. To quantify IMCs and AT2 cells in the lungs, whole lung tissues were completely digested in Buffer-A for 3 hrs, treated with red blood cell lysis buffer, and counted using a hemocytometer. Total lung cell number was multiplied by %CD11c/SpC+ cells to calculate IMC and AT2 cell numbers.

Modified Papanicolau staining
Freshly isolated AT2 cells were smeared on glass slides, air-dried, and stained as previously described (Dobbs, 1990).

Isolation and characterisation of lung fibroblasts from BVE mice
Total lung cells obtained from BVE lung by 3hr collagenase/DNase treatment were resuspended at 5x10 6 /ml in DMEM/10%FCS, plated onto 12-well plates at 1ml/well and incubated for 5-7 days until the culture became confluent. The culture was passaged onto 6well plates by trypsinization. When the culture became confluent again at 3-5 days after re-plating, the cells were counted using a hemocytometer, and subjected to serial passage culture according to the 3T3 protocol. Since IMCs strongly adhere to the culture plate and are not detached by trypsinization, and AT2 cells do not survive after trypsinization, only fibroblasts grew under this culture condition, which allowed us to obtain a highly enriched fibroblast population until passage 5. For transwell co-culture, lung fibroblasts were plated into the bottom wells at 6x10 5 /well before inserts containing isolated AT2 cells were placed on top. Under this condition, lung fibroblasts in the bottom became confluent after 48h of coculture without detectable contamination of other cell types, as confirmed by phase-contrast microscopic observation after removal of the inserts. For immunoblot analysis of proteins secreted by lung fibroblasts, confluent lung fibroblasts in 10cm plates were cultured for 72h in 5ml serum-free DMEM to collect CM. The CM was concentrated using Amicon® Ultra-4 Centrifugal Filter Units (Millipore), and final volumes of the concentrated CM was adjusted by protein quantification of cell lysates obtained from the cells used to collect the CM. 20-30µl concentrated CM was analysed by immunoblotting along with IMC-CM adjusted in an identical manner.

Tumour area quantification
Sequential images covering the whole lung (right lobe) of histological sections stained with H&E were obtained using a Nikon Eclipse Ti microscope, equipped with Nikon CFI Plan Fluor Ph1 10x/0.3NA objective and Andor iXon EM + EMCCD (DU-885) camera, with automatic focus correction (Perfect Focus) system. The images acquired through RGB filter sets were merged and processed using Nikon NIS-Elements software to create whole lung images. Total lung area and tumour area including stroma components were quantified using Image J, and %tumour area (tumour area/total lung area x 100) was calculated.

Preparation of conditioned media
To collect whole lung conditioned media (WL-CM), chopped lung tissue was passed through a 70mm nylon mesh, and cultured in serum-free DMEM for 24 hrs. To collect conditioned media from primary culture of IMCs (IMC-CM), purified IMCs were plated at up to 2 x 10 6 /ml in DMEM and maintained for 72 hrs. Following removal of debris, culture supernatants were concentrated using Amicon® Ultra-4 Centrifugal Filter Units (Millipore). The volume of the concentrated WL-CM was adjusted in proportion to the weight of the initial lung tissue.
Secreted proteins within the WL-CM and IMC-CM were identified by mass spectrometry as described below.

Preparation of NPC2
Bovine NPC2 (bNPC2) was purified from the whey fraction of whole cow milk by the twostep ion exchange chromatography previously described (Larsen et al, 1997) with the following modifications: (1) the pH of CH 3 COONH 4 solution was adjusted to 7.0 for the first DEAE-sepharose chromatography; (2) flow-through from the first chromatography was collected, adjusted to pH 5.0, and applied to the second CM-sepharose chromatography; (3) proteins bound to the CM-sepharose were eluted with a two-step gradient (10-50mM for an initial 1 column volume and 50-200mM for subsequent 25 column volumes) of CH 3 COONH 4 (pH 5.0). bNPC2 was eluted as a first peak in the second gradient. Fractions containing bNPC2 confirmed by immunoblotting were pooled, dialysed against PBS, and concentrated using Amicon® Ultra-4 Centrifugal Filter Units (Millipore). The purity of bNPC2 was confirmed by Coomassie blue-staining on SDS-PAGE gels and mass spectrometry. The concentration of bNPC2 in the final solution was quantified by UV absorbance at 280nm.

Mass spectrometry
Concentrated CM samples were resolved on SDS-PAGE gels and stained with Coomassie Blue. Bands of interest were excised and destained in 200 mM ammonium bicarbonate in 20% acetonitrile, followed by reduction in 10 mM dithiothreitol, alkylation in 100 mM iodoacetamide and trypsin digestion using an automated robot (Multiprobe II Plus EX, Perkin Elmer). In-gel digested peptides were analyzed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) using an RSLCnano HPLC system (Dionex) and an LTQ-Orbitrap-Velos mass spectrometer (Thermo Scientific). The raw data file obtained from each LC-MS/MS acquisition was searched using Mascot (version 2.2.04, Matrix Science Ltd.) against the UniProtKB/Swissprot4 database, and further processed using Scaffold5 (version 3.00.08, Proteome Software). The threshold for protein identification probability was set as 75%, according to the criteria based on the manufacturer's definitions.

Sample size determination
In most experiments, we first performed small-size experiments (n=3-5) which are sufficient for detecting statistically significant differences with large effect sizes by t-tests. When the first experiments showed no significant differences but suggested the possibility to show significant differences with relatively small effect sizes, we performed additional experiments to increasing the samples sizes for statistical analyses.

Protease inhibitors (4 proteins)
Alpha-2-macroglobulin-P A2M NP_783327 Antileukoproteinase SLPI NP_035544 Antithrombin-III ATIII NP_543120 Cystatin-C CST3 NP_034106    B. Morphology and cell densities of IMCs and lung fibroblasts after 48h of co-culture with AT2 cells using a transwell culture plate. The fibroblasts developed more confluent cultures during the co-culture even though they were plated at a lower density (6x10 5 /well) than IMCs (2x10 6 /well). Phase-contrast microphotographs were taken after removal of insert wells containing AT2 cells at 48h of co-culture. Scale bars, 50µm. Of note, fibroblast contamination was rarely observed in the IMC culture.
C. Flow cytometry analysis of BrdU incorporation in AT2 cells co-cultured with IMCs or lung fibroblasts as in B. Co-culture with lung fibroblasts modestly increased BrdU incorporation in AT2 cells (right) compared to the transwell culture without IMCs/fibroblasts (left), but this effect was much weaker than co-culture with IMCs (middle).  B. Tumour burden quantification of CCR1 inhibitor-treated mice by %tumour area calculation (corresponding to Fig 4F). %tumour area was calculated using whole lung (right lobe) images as described in Expanded View Methods. Representative whole lung images for vehicle and CCR1 inhibitor (CCR1i)-treated mice are indicated on the right.

D. Correlation between the sum of CD11c+/SPC+ cell numbers and %tumour area in
AdCre-induced tumour-bearing lungs analyzed in B and C. Pearson's correlation coefficient (R) is indicated on the graph.

Figure S7. Increased T-cells in the lungs of wild-type mice administrated with AdCre
A. Total lung cells harvested from wild-type mice (n=6) at 9-13 weeks after nasal inhalation of AdCre (5x10 7 pfu) were analysed by flow cytometry for CD4, CD8a and B220 surface expression, and compared with wild-type mice without AdCre administration (n=4). A statistically significant increase of T-cells (CD4+CD8) was observed in AdCre-treated mice.

Figure S10. Filipin and organelle marker staining of freshly isolated Npc2 +/hypo IMCs
A. Freshly isolated Npc2 +/hypo IMCs were stained with filipin followed by LAMP1/EEA1 immunostaining with Alexa568-conjugated secondary antibodies, and imaging by confocal laser scanning microscopy (CLSM). Scale bars, 10µm. Free cholesterol-rich structures (strongly stained with filipin) were not co-stained with the lyso-endosome markers. Of note, strong filipin staining was mainly observed at the ventral side of the cells (see Fig S11), where EEA1 signal was relatively weak.
B. Filipin-stained Npc2 +/hypo IMCs were immunostained for Giantin (Golgi cisternae marker) as in A, and imaged by CLSM. Scale bars, 10µm. Giantin staining was not co-localized at free cholesterol-rich coarse structures (strongly stained with filipin) at the ventral side of the cells, though some co-localization at the plasma membrane was observed (top panels). Strong Giantin staining was observed at perinuclear regions (bottom panels), likely representing Giantin localization at the Golgi.
C. CLSM imaging of Npc2 +/hypo IMCs stained with filipin followed by TGN46 (trans-Golgi network marker) immunostaining as in A. Scale bars, 10µm. As with Giantin, TGN46 co-localizes with filipin staining at the plasma membrane but not at coarse structures strongly stained with filipin.

Figure S1
A B