The ability of cells to temporally acquire different characteristics, also known as cell plasticity, plays essential roles in physiological and pathophysiological processes [1
]. Epithelial cells might transform to a mesenchymal phenotype and return to epithelial phenotype by epithelial–mesenchymal transition (EMT) and mesenchymal–epithelial transition (MET), respectively. EMT has been correlated to the metastasis and invasion potential of many types of malignancies [2
]. The fact that the poor outcome of many types of neoplasias correlates with EMT/MET, makes these molecular phenomena an important focus for research and drug targeting [3
]. Despite the clinical association, the role of EMT/MET in metastasis is inconclusive; for example, mesenchymal-like prostate cancer cells survive in circulation, but, unlike epithelial or cells undergoing EMT, they are unable to form macrometastases [7
]. In fact, breast cancer cell metastasis to lung tissue in mice was not affected by decreasing EMT—by targeting EMT—triggering transcription factors such as SNAI1
) and SNAI2
) by overexpressing miR-200 [8
]. An additional role of EMT in tumorigenesis might be the development of apoptotic tolerance and increased resistance to chemotherapy as has been found in animal models and patients [8
]. Recently, a hybrid stage between the epithelial and mesenchymal phenotypes (hybE/M) has been recognized, and such hybE/M cells migrate outside the primary tumors, displaying some mesenchymal features such as spindle-like morphology, increased nuclear levels of ZEB1 transcription factor, and epithelial properties such as cell–cell adhesion potential [10
In order to study the transitory stages of EMT, MET, and hybE/M in vivo, there is need to generate reporter cells to visualize phenotypical changes. There have been different approaches to this, for example, the use of heterologous fluorescent proteins driven under stage-specific promoters such as mesenchymal (ZEB1
) and epithelial (CDH1
], mesenchymal snai1
and epithelial sox10
in zebrafish [13
], or mesenchymal Vimentin (Vim)
in mice [14
]. Nevertheless, heterologous expression of exogenous reporter genes is hampered by the existence of alternative promoters, cis
- and trans
-regulatory elements, and epigenetic events that modulate promoter activation, making some data unreliable.
To overcome these limitations, we genetically engineered and characterized human lung carcinoma H2170 Vimentin Reporter Cells (VRCs) where the fluorescent protein coding gene (mCardinal) has been knocked-in as a genetic fusion, although as separate proteins due to a T2A self-cleaving peptide, at both alleles of the mesenchymal marker VIM.
2. Materials and Methods
2.1. Cell Culture, Nucleofection, and Transfection
Human squamous lung carcinoma (H2170), human colorectal adenocarcinoma (HT29), and human embryonic kidney (HEK293) cells were cultured according to standard mammalian tissue culture protocols and sterile techniques. The cell lines were cultured in DMEM supplemented with 10% fetal calf serum (FCS), 100 units/mL and penicillin/100 µg/mL streptomycin. All tissue culture media and supplements were obtained from Gibco. H2170, HT29, and HEK293 cells were obtained from ATCC. H2170 cells were nucleofected for genome editing with the use of Nucleofector I Device and Cell Line Nucleofector Kit T (Amaxa). The optimized protocol is as follows: nulceofection of 1 million suspended cells with 2 µg of the plasmid DNA using program X-001, generating a transfection efficiency of 99%. For performing functional experiments on a smaller scale, the H2170 cells were transfected with the use of Lipofectamine 3000 (Invitrogen, Waltham, MA, USA) with transfection efficiency 90–95% after 24 h. A total of 50000 cells were seeded a day before transfection in 24-well plates. Transfection was performed using 500 ng of the plasmid DNA, 1.5 μL Lipofectamine®3000 Reagent, and 1 μL P3000™ Reagent (both from Invitrogen) per well, following the manufacturer’s protocol. HEK293 were transfected with the use of TurboFect reagent (Thermo, Waltham, MA, USA) according to the protocol supplied by the company. Next day, the transfection efficiency was in the range of 80–90%. VRCs were also cultured with 10 ng/mL TGFβ (human TGFβ1, Biorbyt, Cambridge, United Kingdom) for 72 h.
2.2. Plasmid Vectors for the Knock-In of VIM
We modified Cas9/gRNA-expressing plasmid pSpCas9(BB)-2A-Puro (PX459) V2.0 (Addgene plasmid #62988) by inserting the fragment coding the targeting gRNA using digestion/ligation protocol [15
]. The oligonucleotides used to generate the gRNA were gRNA-F and gRNA-R, as seen in Table S1
. The template plasmid used for inserting DYKDDDDK-tagged (FLAG) mCardinal fluorescent protein gene after VIM
was designed as following (gRNAsite—800 nt of Vim
—P2A—mCardinal—FLAG—800 nt of 3′Vim
UTR—gRNAsite) and was synthetized by Thermo. The Cas9-gRNA and the template plasmids were both nucleofected to cells at the same time, and the cells were selected 2 days after nucleofection and cultured for another 2 days in 5 μg/mL of puromycin. The positive single cell clones obtained by dilution were genotyped, as described in the Genotyping section.
2.3. Plasmid Vectors Used in Functional Assays
For functional experiments in H2170 knocked-in cells, we used (FLAG) Snail 6SA (active Snail) plasmid, which was a gift from Mien-Chie Hung (Addgene plasmid # 16221) [16
], TGFB1-bio-His (proTGFβ), which was a gift from Gavin Wright (Addgene plasmid # 52185) [17
], and HA-OVOL2 (OVOL2)-expressing plasmid, which was a gift from Changwon Park [18
]. EMT/MET in VRCs was studied with the use of expressing vectors harboring genomic fragments of the microRNA-200 family (miR-145
) which were cloned in our laboratory (See the Molecular Cloning section). Control cells were transfected with pUC18 subcloning plasmid.
2.4. Molecular Cloning
All the plasmid fragments used for cloning were amplified using tiHybrid proofreading DNA polymerase (EURx), according to the supplied protocol. PCR products amplified on the plasmid DNA template were incubated overnight at 37 °C with DpnI FastDigest enzyme (Thermo) as per the manufacturer’s instructions.
sequence was cloned from cDNA of knocked-in cells upon PCR amplification and linearized by PCR using a pmR-expressing vector (Clonetech) and recombined using Gibson Assembly (NEB). The resulting vector contained a full VIM-T2A-mCardinal
reading frame under the control of a CMV promoter. The primers for insert amplification were KI-F and KI-R, whereas the pair used for backbone linearization were BCB-F and BCB-R, as seen in Table S1
. Mutagenesis was performed by REPLACR methodology [19
], using the SDM-F and SDM-R primers, as seen in Table S1
The vectors harboring miR-145
, and miR-205
genomic fragments were created by inserting each PCR-amplified microRNA gene into the 3′UTR of mNeon-expressing vector (pmR-mNeon). All genomic fragments listed above were amplified using tiHybrid DNA polymerase (EURx) from DNA, which was purified from the blood of healthy volunteer with the use of GeneAll Exgene Blood SV kit (GeneAll). The sets of primers used for amplification of the miR-145
, and miR-205
fragments were miR145-F
, respectively, as seen in Supplementary Table S1
. The amplified products produced sticky ends upon digestion by BglII and HindIII restrictases (both from Thermo). Digested and purified DNA fragments were ligated using T7 ligase (Thermo) in molar ratio 3:1 with 100 ng of linear pmR-mNeon, which was previously cut by BglII and HindIII enzymes.
The resulting vectors were named miR-145, miR-200b, miR-200c, and miR-205. The sequences of all the vectors were verified by Sanger sequencing (Genomed, Warsaw, Poland).
The targeting sequence for CRISPR/Cas9 was in the last intron (intron 8) of VIM
in HEK293 and H2170 cells with the use of Benchling algorithm. Single-cell clones were cultured on 96-well plates to more than 50% confluence (Nunc, Roskilde, Denmark). Upon washing with phosphate-buffered saline (PBS, Gibco), they were genotyped by PCR using Mouse Direct PCR Kit (Bimake), following the manufacturer’s instructions. The primers used for genotyping were 170 and 249, as seen in Table S1
The genotyping was further confirmed by PCR using the same set of primers (170 and 249), tiHybrid DNA polymerase and high-quality genomic DNA, purified from single-cell clones with the use of QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). The KI was verified by sequencing (Genomed).
2.6. RNA Extraction, Reverse Transcription, and qPCR
Total RNA was isolated from cells with the use of Extractme Total RNA kit (Blirt, Gdansk, Poland) according to the manufacturer’s manual, including DNase treatment. The purity and quantity of isolated RNA was estimated spectrophotometrically with the use of a Tecan M200Pro microplate reader supplied with NanoQuant plates (Tecan, Zürich, Switzerland). Only the samples with 260/280 nm OD ratio higher than 1.8 were used for downstream analysis. For molecular cloning, 3 μg of RNA were reverse transcribed for 30 min at 50 °C using an oligo(dT) primer and the Transcriptor High Fidelity cDNA Synthesis Kit (Roche, Rotkreuz, Switzerland) followed by 5 min enzyme inactivation at 85 °C, according to the manufacturer’s instructions. For QPCR, 2 μg of RNA were reverse transcribed using a High-Capacity RNA-to-cDNA Kit (Applied Biosystems, Waltham, MA, USA) following to the manufacturer’s protocol.
Quantitative real-time expression analysis was performed using a LightCycler®
480 II instrument (Roche) equipped with 384 well plates and PowerUp™ SYBR™ Green Master Mix (Applied Biosystems). The primers were as listed in Supplementary data
, as seen in Table S1
. Amplification was performed in 12.5 μL reaction mixture containing cDNA amount corresponding to 12.5 ng of total RNA, 1 × PowerUp SYBR Green Master Mix, and 3.125 pmol of each primer (forward and reverse).
After 2 min of initial incubation at 50 °C followed by 2 min incubation at 95 °C, cDNA was amplified in 45 cycles consisting of 15 s denaturation at 95 °C, 30 s annealing at 60 °C and 20 s elongation at 72 °C. The obtained fluorescence data was analyzed using a relative quantification (RQ) method 2–ΔΔCT
for estimating expression fold changes normalized to dim-VRCs and 2–ΔCT
method for comparison of the expression of each measured gene. The assessed genes expression (VIM
, and ZEB2
) were normalized to GAPDH
level, which was measured with the use of GAPDH-F and GAPDH-R oligonucleotides. GAPDH
was previously confirmed as stably expressed at the mRNA level in H2170 cells as well as in VRCs (mean Cp = 17.14; median Cp = 17.09; SD = 0.5209; SEM = 0.03638; N = 205). Expression of VIM
was measured using VIM-F and VIM-R primers, measurement of mCardinal
level was conducted using mCard-F and mCard-R primers, whereas estimation of CDH1
expression was performed at the mRNA level with the use of Cdh1-F and Cdh1 -R oligonucleotides. ZEB1
quantifications were performed using ZEB1F/R and ZEB2F/R pairs of oligonucleotides, respectively. TWIST1
quantifications were performed using the TWIST1F/R and TWIST2F/R oligonucleotide pairs, respectively. The primers were designed as intron-spanning to avoid any influence of genomic DNA contamination and are listed in Supplementary Table S1
2.7. Flow Cytometry of Living Cells
VRCs or HT29 cells cultured for one day on a six-well plate (Nunc) were washed with HBSS (Hank′s Balanced Salt Solution, Thermo) and detached by Accutase (Corning, Corning, NY, USA). The cells dissolved in 100 µL HBSS with 5 µL of PE Mouse anti-E-Cadherin (562526, BD Pharmingen, San Jose, CA, USA) were incubated for 1 h, washed with HBSS, and suspended in 0.5 mL HBSS. The cells were counted on the FL-3 fluorescence channel of a FACSCalibur flow cytometer (BD).
2.8. Cell Sorting
VRCs seeded one day before were then sorted using a BD FACSAria™ flow cytometer (BD Biosciences, San Jose, CA, USA). In this case, cells were selected into two populations due to the intensity of their mCardinal fluorescence in the far-red channel, as seen in Figure S1
. After sorting, the purity of the sorted cells was confirmed by flow cytometry and reached more than 97%. The resulting two populations of VRCs were named dim-VRCs and bright-VRCs.
2.9. Confocal Imaging of Living Cells
The cells were seeded onto 24-well glass-bottom plates (MoBiTec, Goettingen Germany) a day before transfection. Transfection was conducted as described above. Transfected cells were then visualized under a Nikon Ti Confocal microscope after 24 and 28 h since transfection, using a 563 nm laser for mCardinal fluorescence. Mean fluorescence intensity was measured using NISelements (ver 3.22.08, Melville, NY, USA) software and is shown on the graphs. Each of at least 10 measured regions of interest (ROIs) included a cluster of more than 20 adherent cells. The brightest slices of the z-stacks were chosen for measurement. The measurements were done in duplicate.
The cells seeded a day before on glass bottom Labtec chamber slides (Nunc) were washed with PBS and fixed with 2% paraformaldehyde for 20 min at room temperature. Upon washing in PBS, the cells were permeabilized by 0.5% triton in PBS for 20 min. The endogenous peroxidase was blocked by incubation in 1% sodium azide and 1% hydrogen peroxide in PBS for another 20 min. Another washing step in PBS was followed by incubation in Blocking Buffer which was 1× Normal Donkey Serum (reconstituted from 20×, Jackson ImmunoResearch, Ely, United Kingdom) containing 1% bovine serum albumin (SantaCruz Biotechnology, Dallas, TX, USA) and 0.1% of triton for 30 min in room temperature. Supplied by Cell Signaling, primary antibodies against VIM (rabbit, #5714S Danvers, MA, USA) and FLAG-tag (mouse, #8146S) were diluted in Blocking Buffer at ratios of 1:100 and 1:1600, respectively. Incubation with the primary antibodies was conducted overnight at 4 °C and was followed by washing in PBS. Secondary Peroxidase F(ab’)₂ Fragment Donkey Anti-Rabbit IgG (H+L) (711-036-152, Jackson ImmunoResearch) for detection of primary rabbit and Peroxidase F(ab’)₂ Fragment Donkey Anti-Mouse IgG (H+L) (715-036-150, Jackson ImmunoResearch) for detection of primary mouse antibodies were diluted 1:1000 in Blocking Buffer. Incubation with secondary antibodies (fluorochrome or peroxidase conjugated) was performed for 1 h at room temperature and was followed by washing in PBS. Peroxidase-conjugated antibodies were stained with Alexa Fluor 555 Tyramide Reagent (B40955, LifeTechnolgies, Carlsbad, CA, USA) or Alexa Fluor 488 Tyramide Reagent (B40953, LifeTechnolgies) following to the manufacturer’s protocol, resulting in enzyme-linked fluorescence signal amplification. When double Alexa Fluor Tyramide Reagents (488 and 555) staining was done, an additional step of 15 min washing in 3% hydrogen peroxide, 0.1% sodium azide in PBS, to achieve complete inhibition of HRP, was added between incubations with two different secondary HRP-antibodies. The cells were finally washed by PBS and stained with Hoechst (Cayman) diluted in PBS (1:1000 from 10 mg/mL stock solution). The cells were washed again, mounted with ProLong Gold mounting medium (LifeTechnologies), and visualized under a Nikon Ti confocal microscope.
2.11. Migration of VRCs
A XCELLigence real-time cell analysis system (Acea) equipped with 16-well CIM-plates served for the VRCs migration assessment. VRCs were first detached using Accutase, then transfected with OVOL2, miR-200c, miR-205, or pUC18 plasmids using Lipofectamine300 reagent, as described above. Then, 30000 cells were seeded on each well of the CIM-plate, which was kept in the standard mammalian tissue culture conditions and measured over 24 h. The cells’ migration toward an attractant, which was 10% DMEM supplemented with 10% FCS, was normalized to cell migration toward DMEM alone and shown as Cell Index. Each experiment was measured at least in triplicate and repeated twice.
Mounting evidence shows that tumors are far more heterogenous than expected in due to genetic diversity of the tumor cells as well as their phenotypic plasticity [22
]. This phenotypic plasticity is in fact phenotype switching, which is understood as a phenomenon whereby cancer cells transition between different phenotypes in response to environmental cues, without acquiring new mutations [24
]. This cellular plasticity has been reported as having high clinical impact because it is crucial for drug resistance development, e.g., in lung cancer patients treated with EGFR inhibitors [25
], reacquiring pluripotency and become a cancer stem cell (CSC) [26
], or maintaining metastatic ability of many types of cancer cells [27
Among tumor cells, there are some that undergo EMT, MET, or hybE/M, and in order to understand the effects of cellular plasticity in biological and pathological processes, there is a need for reporters that can determine the stage of a cell. Observing changes in mesenchymal and epithelial phenotype as they occur is a significant improvement in studying molecular mechanisms in cancer cells. H2170 cells were chosen because they have been a good model for EMT and MET [30
To date, EMT/MET is routinely studied by the use of exogenous reporter genes that are heterologously expressed, exhibiting interference of cis-
-regulatory elements, alternative promoters, or epigenetic events, which result in obtaining unreliable data [32
]. These limitations are overcome by using genome engineering of endogenous genes. There are a few of available cancer cell lines that harbor a C-terminal red fluorescent protein (RFP) tag on VIM: A549 (lung), HCT116 (colorectal), and MDA-MB-231 (breast adenocarcinoma), and they are commercially available at the ATCC cell repository. These cells, however, have RFP
knocked-in to the beginning of last exon of VIM
(exon 9), which results in deletion of a large fragment of that exon and of the gene product (https://www.lgcstandards-atcc.org/en/Global/Products/CCL-247EMT.aspx#documentation
). While these modified A549, HCT116, and MDA-MB-231 cell lines enable near real-time tracking of the EMT/MET status as cells transition from epithelial to mesenchymal phenotype under defined conditions [34
], any approach where an endogenous protein is truncated may affect functions in comparison to endogenously expressed proteins, particularly in the context of a protein such as VIM, which has many signaling modulatory activities during cell plasticity [35
We used a far-red fluorescent protein (mCardinal), which is simultaneously expressed with endogenous VIM, but separate during translation due to the viral self-cleaving peptide (T2A). T2A was chosen because, together with P2A, it has been reported as the most efficient of all the tested self-cleaving 2A peptides [38
]; moreover, T2A resulted in the least amount of “uncleaved” protein product among the family of 2A peptides [39
]. The stability and half-life of the two resulting proteins can result in small changes in total expression, the two proteins are synthesized at a 1:1 ratio [41
expression at the transcriptional level corresponded to that of mCardinal
, and when stimulated, the changes of VIM
expression were hand-to-hand with those of mCardinal
, evidencing that our strategy is fully functional.
mCardinal was chosen mainly because this far-red monomeric protein is better for in vivo settings due to it is excitation as wavelengths above 600 nm, which penetrate through hemoglobin-rich tissues far better than lower wavelengths [42
]. The second reason is that mCardinal is suitable for precise monitoring of its changing expression due to its short maturation half-time (27 min) [43
], which is about three times shorter compared to other commonly used fluorescent markers, e.g., RFP [44
The subcellular localization of mCardinal in VRCs by confocal imaging show nonhomogenous distribution of the fluorescence, which was observed at moderate levels in the cytoplasm and in brighter glowing foci. Immunofluorescent staining of VIM and FLAG-tag mCardinal showed colocalization at the “foci”, whereas Vimentin was also present in the cytoskeleton of VRCs. Moreover, recloning of the VIM–P2A–mCardinal
and further overexpressing it in HEK293 cells shown similar subcellular distribution of the fluorescence. The presence of foci is probably the result of inefficient T2A peptide cleavage, with the fusion protein being unable to form normal VIM polymers [39
], which is more conspicuous in the case of the fusion of VIM–mCardinal. In fact, C-tags in VIM might result in an impaired protein [45
], and its expression may give a dominant-negative phenotype, similar to the phenotype observed in the cells which express only the N-terminal domain of VIM (NTD-VIM) [46
]. NTD-VIM (VIM that lacks C-terminus) domain-expressing cells show its abnormal localization, as formation of the granules in the cytoplasm together with disrupted endogenous VIM localization examined by immunostaining [45
]. The role of the tail domain in VIM organization is not fully understood, although it is known to be necessary for appropriate network formation [47
]. Thus, it seems to be possible that abnormal VIM localization in VRCs may be produced by the fluorescent protein. Yet, the presence of bright foci is advantageous as these structures are brighter than diffused cytoplasmic mCardinal under low VIM expression.
Dim-VRCs presented a reduced mesenchymal phenotype compared to bright-VRCs. The dim-VRCs could be driven to a more mesenchymal phenotype via expression of proTGFβ [17
] or constitutively active SNAI1 transcription factor [16
]. We also noticed insignificant downregulation of CDH1
in proTGFβ overexpressing dim-VRCs, what suggests that E-cadherin may be only partially regulated via the TGFβ-dependent pathway [48
-overexpressing VRCs respond by E-cadherin expression, which is in accordance with data describing OVOL2 as a strong epithelial regulator that maintains transcriptional programs in epidermal keratinocytes and mammary epithelial cells by repressing ZEB1 and ZEB2 [49
The MiR-200 family was identified as double-negative regulators by silencing CDH1
and transcription repressors ZEB11
]. We confirmed the activatory role of miR-145
, and miR-205
expression in VRCs. We also confirmed that miR-200c
expression decreased ZEB1
levels in VRCs [53
], but we did not observe any effect of miR-200c
]. VRCs that overexpressed miR-205
downregulation, which is in accordance with previous works [55
]. Our results suggest that the molecular mechanisms of the miR-200 family on EMT might be cell-type-specific, and more work needs to be done to clarify this. Our data is in accordance with previous data in that the migratory potential of mesenchymal cells is inhibited by ZEB1 or ZEB2 downregulation produced by miR-200 family overexpression [57
The very similar levels of CDH1, VIM, SNAI1, ZEB1, ZEB2, TWIST1, and TWIST2 transcripts between VRCs and the parental H2170 cell line suggests that VRCs may be useful in broad range of studies.
In conclusion, our data uniquely illustrate a reporter line, VRCs, as a reliable reporter model for studying EMT and MET. VRCs allow one to directly observe cellular plasticity with respect to their mesenchymal/epithelial state in vitro and, in the future, in vivo. These cells could also be used as a robust platform for drug development.