Nuclear transport of cancer extracellular vesicle-derived biomaterials through nuclear envelope invagination-associated late endosomes

Extracellular membrane vesicles (EVs) function as vehicles of intercellular communication, but how the biomaterials they carry reach the target site in recipient cells is an open question. We report that subdomains of Rab7+ late endosomes and nuclear envelope invaginations come together to create a sub-nuclear compartment, where biomaterials associated with CD9+ EVs are delivered. EV-derived biomaterials were also found in the nuclei of host cells. The inhibition of nuclear import and export pathways abrogated the nuclear localization of EV-derived biomaterials or led to their accumulation therein, respectively, suggesting that their translocation is dependent on nuclear pores. Nuclear envelope invagination-associated late endosomes were observed in ex vivo biopsies in both breast carcinoma and associated stromal cells. The transcriptome of stromal cells exposed to cancer cell-derived CD9+ EVs revealed that the regulation of eleven genes, notably those involved in inflammation, relies on the nuclear translocation of EV-derived biomaterials. Our findings uncover a new cellular pathway used by EVs to reach nuclear compartment.

FEMX-I and MDA cells were transfected with 10 g PS100010 PrecisionShuttle mammalian plasmid encoding CD9 with a green fluorescent protein (GFP) tag at its C-terminus under the control of the cytomegalovirus promoter (RG202000; Origene) using FuGene (Promega) in a 1:3 DNA:lipid ratio. Cells were selected by introducing 400 g/ml of G418 (Life Technology) in the culture medium for seven days, resulting in >99% GFP-positive cells.
Antibiotics were removed from the medium at least one week before experiments.
Likewise for the ER-GFP and mitochondria-GFP encoding GFP fused to the ER signal sequence of calreticulin with KDEL retention signal or leader sequence of E1  pyruvate dehydrogenase, respectively.
To express CD9-GFP fusion protein in MSCs, lentiviral particles containing the pCT-CD9-GFP plasmid (CYTO122-VA-1; System Biosciences) were utilized. The virus were preloaded onto retronectin-coated plates and centrifuged at 960 x g for 30 min at 4 o C. The operation was repeated twice. The supernatant was then removed and plates were washed with PBS before addition of cells. After transduction, stable cell lines were selected by introducing 1 µg/ml puromycin in the culture medium for a week. Cells were kept in culture for a maximum of 5 passages (~3 weeks).
To inhibit CD9 expression, CD9 shRNA lentiviral particles (sc-35032-V, Santa Cruz Biotechnology) containing the puromycin resistant gene were employed [1]. They consist of a pool of concentrated, transduction-ready viral particles containing 3 target-specific constructs that encode 19-25 nt (plus hairpin) shRNA designed to knock down gene expression. Lentiviral particles containing shRNA construct encoding a scrambled sequence (sc-108080) were used as negative control. After transduction, stable cell lines were selected by introducing 2 g/ml puromycin in the culture medium for a week and CD9-negative cells were sorted by flow cytometry.

Flow cytometry
FEMX-I and MDA CD9sh cells were incubated with HI9a anti-CD9 Ab for 60 min, washed, and incubated with Cy5-conjugated anti-mouse IgG secondary Ab for 30 min. CD9-negative cells were isolated from transduced shCD9 cells using an SH800 cell sorter (Sony Global, Champaign, IL) and maintained in complete medium with puromycin (2 g/ml). Antibiotic was removed from the medium at least one week before experiments.

Cell cycle analysis
Cell cycle of proliferating and quiescent MSCs was analyzed using the method of propidium iodide staining and analyzed on a SH800 cell sorter. Cell cycle distribution was calculated by FlowJo version 7 software.

Immunoblotting
Detergent cell lysate extracts obtained upon lysis of cells in a buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl and Set III protease inhibitor cocktail (Calbiochem)) containing 1% Triton X-100 were analyzed by sodium dodecyl sulfate-polyacrylamide-gel electrophoresis under reducing conditions except for CD9 and GAPDH when co-incubated with anti-CD9 Ab. 5,000 (MSC) or 30,000 (all other cell lines) cell equivalents were loaded per lane. After protein transfer, membranes were blocked with 1% bovine serum albumin in PBS for 16 h, and then probed with a given primary Ab (see Supplementary Table 2). After washing, membranes were incubated with an IRdye secondary Ab (LI-COR Biosciences) and visualized using an Odyssey CLx system (LI-COR Biosciences).

NTA
We used the light-scattering characteristics of 488-nm laser light on EV preparations undergoing Brownian motion injected by continuous flow into the sample chamber of a Nanosight LM10 unit (Malvern). Data are presented as the mean ± s.e.m of 5 independent preparations, where each is an average of six 30-sec video recordings.

Labeling of EVs with membrane dyes
CD133 + EVs were labeled for 10-20 min at 37 o C by the addition of 1,1-dioctadecyl-3,3,3,3tetramethylindocarbocyanine perchlorate (DiI, Life Technologies) at a final concentration of 5 M. Excess of dye was removed using LS-columns. DiI-labeled CD133 + EVs were eluted with cold PBS, centrifuged at 200,000 x g for 60 min at 4 o C and resuspended in 200 l PBS.
To exclude the possibility that magnetic beads alone formed clusters that nonspecifically absorbed DiI, control CD133-coupled magnetic beads were treated with DiI prior to their incubation with recipient cells. Under these conditions no fluorescent dye was observed in cells (data not shown).

Drug treatments
Cell treatments with dynasore (80 M) or methyl--cyclodextrin (10 mM) were initiated 30 min prior to addition of EVs for 4.5 h, while treatments with importazole (40 M) or leptomycin B (10 ng/ml) were started after the initial 2.5 h of incubation with EVs and continued for the remaining 2 h. When necessary, the appropriate control vehicle (DMSO, methanol) was used. All experiments were repeated independently three times.

Immunocytochemistry
Cells treated under various condition and/or incubated with EVs were washed with PBS, fixed in 4% paraformaldehyde for 20 min at room temperature (RT), washed twice with PBS and permeabilized with 0.2% Tween 20 in PBS for 15 min at RT. Cells were then incubated with primary antibodies directed against SUN2, CD133, CD9, Alix, Annexin A2, Rab7, importin β1 or nucleoporins (see Supplementary Table 2) for 60 min at RT, washed twice with PBS, and incubated with appropriate fluorescent secondary antibodies for 30 min at RT.
As a negative control, only secondary Ab was employed. The isolated MSC nuclei were stained with Hoechst 33342 (5 μg/ml) during the last 30 min of incubation with DiI-labeled EVs. Cells or isolated MSC nuclei were imaged in PBS by confocal laser-scanning microscopy (CLSM).

Immunohistochemistry
Formalin-fixed, paraffin-embedded patient biopsies (4-µm sections) of infiltrating ductal carcinoma of the breast were derived from anonymized archival sources within the Department of Pathology at UCLA. All Ab labeling, primary and secondary, were performed sequentially, except for pan-cytokeratin or vimentin that were detected simultaneously with CD9 or Rab7, overnight at 4 o C with primary Abs that had been raised in different species. A sequential multiple immunofluorescence protocol which circumvented potential problems arising from using multiple primary Abs raised in the same host species (e.g., rabbit) was also employed [2]. As negative control, only secondary Ab was employed. Slides were then washed with PBS and stained with 1 μg/ml 4',6-diamidino-2-phenylindole (DAPI) for 15 min at RT. Anti-fade mountant (ProLong Diamond, Thermo Fisher) was added to the slides, which were then mounted with 1.5-mm coverslips and viewed by CLSM.

CLSM and time-lapse video microscopy
All images were acquired under the same microscope settings for subsequent calculations of mean fluorescence and recorded using NIS Elements software (Nikon). Raw images were processed using Fiji [3]. The number of nuclear envelope invagination-associated late endosomes per cell and the localization of EV-derived biomaterials therein and/or nuclei from scanned samples were manually counted. Each z-section through the nucleoplasm (10- For the live imaging, Rab7-RFP-expressing cells were incubated with CD9-GFP + EVs (1 x 10 9 particles/ml) at 37°C, 5% CO 2 humidified chamber for 4.5 h. Hoechst 33342 (5 μg/ml) was added for the last 60 min. Images were acquired at 20-second interval for 5 min with microscope described above. A short video was then created from the acquired time points using the NIS software.

RNA sequencing
To establish the early effect of FEMX-I-derived EVs on the transcriptome of MSCs in the absence or presence of modulators of nuclear uptake, differentially expressed genes in MSCs were analyzed by RNA-seq. 7 experimental groups in biological duplicates were analyzed, including solvent alone, EVs, EVs plus importazole, EVs plus leptomycin B, EVs from FEMX-I CD9sh, and importazole or leptomycin B alone. Samples with RNA Integrity Numbers of 8 or greater were prepped using the Illumina TruSeq mRNA Sample Prep v2 kit. mRNA prepared from total RNA (0.1-3 g) was then fragmented and copied into first strand cDNA. The 3' ends of the cDNA were then adenylated and adapters ligated. The products were purified and enriched by PCR to create the final cDNA library. Libraries were checked for quality and quantity, clustered on the cBot (Illumina) and sequenced on a HiSeq 2500 (Illumina) in High Output mode.

Bio-informatic analysis
To analyze RNA-seq data, NCBI human reference genome (GRCH38.84) and its GTF file and associated index files were downloaded from the Illumina iGenome website (http://support.illumina.com/sequencing/sequencing_software/igenome.html). TopHat2 software [4] was used to map short read sequences to the whole reference genome. The BAM alignment files were further processed by CuffDiff (version 2.2.1) for estimating transcripts' abundances and testing for differential expression between groups. Up-or down-regulated gene lists were selected using the cutoff criteria of log2 absolute fold change values > 1.1.
RNA-Seq short read sequence files were submitted as DNA Sequencing Data (traces and short reads) to the NCBI Gene Expression Omnibus (GEO) database.

Real-time quantitative reverse-transcription PCR (qRT-PCR)
Total RNA collected from MSCs exposed for 4.5 h to solvent alone, FEMX-I-derived EVs, or EVs plus importazole were analyzed by two-step qRT-PCR. First, genomic DNA was removed in the RNA preparation using the DNase I recombinant, RNase-free kit (10U; Roche). Reaction was carried out at 37°C for 15 min followed by 10 min at 75°C to inactivate the enzyme. Afterward, SuperScript IV VILO master mix (Invitrogen), which includes both oligo (dt)18 and random hexamer primers, was added to reverse transcribe RNA. The samples were incubated 10 min at 25°C (primer annealing), 10 min at 50°C (reverse transcription) and 5 min at 85°C (enzyme inactivation). Second, PCR reaction plate was then setup using cDNA templates and the following reagents from Applied Biosystems: The number of cycles required for the fluorescent signal to cross the threshold (C T ) was determined by setting the baseline and threshold values between the 3 rd and 15 th cycle and the exponential growth phase of the amplification curve, respectively. To calculate ΔC T and foldchanges, all quantitations were normalized to the GAPDH endogenous control.

Data availability
The raw data of RNA sequencing reported in this paper are accessible since October 1, 2016.  Figure 3F.

Supplementary Video 3: Subdomains of Late Endosomes Invading Nuclear Envelope
Invaginations. The video (format: mp4; size: 2.9 MB) depicts a FEMX-I cell expressing Rab7-FRP (red) immunolabeled with SUN2 Ab (purple). Cell was analyzed by CLSM. 3D reconstruction of 3 sections (green slice) is displayed. Still image from this movie is shown in Figure 4A.

Supplemental Tables
Supplementary Table 1. Details of data presented in Figures 4, 5, 6 SD, standard deviation; SEM, standard error of the mean; NS, not significant. >50 cells were evaluated per experiment, n = 3.
Color code refers to the pie chart presented in the figure.