Optimization of lentiviral transduction parameters and its application for CRISPR-based secretome modification of human endometrial mesenchymal stem cells

ABSTRACT

Mesenchymal stem cells (MSCs) hold a great promise for successful development of regenerative medicine. Among the plenty of uncovered MSCs sources, desquamated endometrium collected from the menstrual blood probably remains the most accessible. Though numerous studies have been published on human endometrium-derived mesenchymal stem cells (hMESCs) properties in the past years, there are only a few data regarding their genetic modulation. Moreover, there is a lack of information about the fate of the transduced hMESCs. The present study aimed to optimize hMESCs transduction parameters and apply Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas9 technology for genome and secretome modification. The fate of hMESCs transduced either in presence of polybrene (Pb) or protamine sulfate (Ps) was assessed by alterations in CD expression profile, growth rate, cell size, migration capability, osteogenic, adipogenic, and decidual differential potentials. Here, we postulated that the use of Ps for hMESCs genetic manipulations is preferable, as it has no impact on the stem-cell properties, whereas Pb application is undesirable, as it induces cellular senescence. Plasminogen activator inhibitor-1 was selected for further targeted hMESCs genome and secretome modification using CRISPR/Cas9 systems. The obtained data provide optimized transduction scheme for hMESCs and verification of its effectiveness by successful hMESCs genome editing via CRISPR/Cas9 technology.

Introduction

Human endometrium is a unique tissue that undergoes about 400 cycles of growth, renewal, differentiation, and shedding during a woman’s reproductive life [1,2]. During the secretory phase, two endometrial layers can be defined – functionalis that undergoes a striking progression of histological change followed by exfoliation during menstrual cycle, and basalis that shows only modest alterations [3,4]. Suggestion about the existence of the resident stem cells within endometrium responsible for the cyclic restoration of the functional layer was proposed more than 30 y ago by V. Pryanishnikov [5]. Although the experimental evidence was obtained much later, for now, the existence of human endometrium-derived mesenchymal stem cells (hMESCs) casts no doubt [6,7].

Until recently, the most common way to isolate hMESCs was to take biopsy of endometrial tissue through a cervix [7,8]. In 2007–2008, two independent research groups isolated stem cells from desquamated endometrium collected from menstrual blood [4,9]. Importantly, cells obtained from menstrual blood are morphologically and functionally similar to cells directly extracted from the endometrium [8]. hMESCs display all the necessary characteristics inherent to the stem cells of mesenchymal origin [4,9,10]. Namely, these cells are characterized by high proliferation activity during long-term cultivation, genetic stability, lack of tumorigenicity, low immunogenicity, expression of specific cell surface markers (CD 90, CD 105, CD 73) as well as multilineage differentiation capacity [4,9,11]. A fundamental distinctive feature of hMESCs from menstrual blood in comparison with the other widely used types of mesenchymal stem cells (MSCs) obtained from bone marrow or adipose tissue seems to be noninvasive and nontraumatic isolation procedures. Largely due to these advantages, promising results concerning hMESCs experimental and clinical application for treatment of various diseases, including stroke, colitis, limb ischemia, coronary disease, Duchenne’s muscular atrophy, and streptozotocin-induced type 1 diabetes, are obtained (reviewed in [8]). Therefore, hMESCs seem to be attractive tools for both investigations in stem-cell biology and applications in regenerative medicine.

Currently, there are various strategies actively exploring in the context of regenerative medicine. The most conventional stem-cell-based applications are recapitulating organs and tissues via scaffold fabrication, 3D bioprinting or self-assembly, and direct administration of undifferentiated stem cells [12–17]. The latter among these approaches uncovers at least two possibilities. On the one hand, the unique capacity of MSCs to migrate to the sites of damage following intravenous transplantation along with their proliferation ability makes them promising candidates for MSC-based gene therapy, as they can be used for delivery of therapeutic genes to sites of injury. On the other hand, accumulating evidence suggests that MSCs participate in tissue repair and regeneration through secreted trophic factors; thus, beneficial effects of stem-cell transplantation are primarily mediated via the paracrine action of mediators rather than the direct differentiation of MSCs and substitution of damaged cells [18–21]. In this context, targeted genome editing allows to augment beneficial and diminish deleterious effects of stem cells paracrine activity. Thus, genetically modified MSCs can be used as a so-called drug manufactures that produce a controlled set of mediators.

The bottleneck of the above approaches implying genetic modifications of MSCs is the delivery of the components of genetic constructs. In order to achieve prolonged, stable, and strong transgene expression, lentiviral vectors are typically used [22]. However, today there are some data regarding the adverse impact of lentiviral transduction on the stem-cell properties [23]. Given that hMESCs is a relatively new type of stem cells, there are only a few data regarding their genetic modulation. Moreover, there is a lack of information about possible technical difficulties and consequences of hMESCs transduction. Therefore, the present study aimed to optimize hMESCs transduction parameters and to apply CRISPR/Cas9 technology for their genome and secretome modification.

While choosing target gene for validation of optimized protocol for hMESCs transduction, we focused on SERPINE-1 gene coding plasminogen activator inhibitor-1 (PAI-1) protein. In vivo hMESCs undergo cyclic activation and subsequent differentiation into mature stromal cells, which further differentiate into decidual cells in response to the postovulatory rise in progesterone and increasing endometrial cAMP levels [24–26]. Decidualization of endometrium is known to be an essential process for embryo implantation, placenta forming, and maintenance of pregnancy [24–26]. Therefore, with regard to regenerative medicine, hMESCs may primarily be applied for cell therapy of infertility associated with decidualization insufficiency. Today, it is clearly shown that tightly controlled PAI-1 level is crucial for normal pregnancy progression from implantation till term [27–29]. PAI-1 expression by decidual cells has been reported to play a decisive role in regulating proteolysis, migration of endothelial cells, and remodeling of maternal tissue during human implantation [27,28,30,31]. Any disturbance in PAI-1 levels may lead to various pregnancy complications, including recurrent pregnancy losses, preeclampsia, intrauterine growth restriction, endometriosis and polycystic ovary syndrome, and unrestricted trophoblastic invasion leading to placenta accrete [30–33]. Since rhythmicity in PAI-1 expression and secretion levels must exist at varied intervals to maintain pregnancy till term, in the context of possible hMESCs secretome application, both overexpression and knockout of PAI-1 may have sense, depending on nature and stage of the disease supposed to be cured. Thus, in the present study, we were able to obtain both PAI-1 knockout and PAI-1 overexpressing hMESCs as well as their modified secretome that might further be used for functional testing.

Materials and methods

hMESCs culture

Human MSCs were isolated from desquamated endometrium in menstrual blood from healthy donor (hMESCs, line 2804) as described previously [10]. The study was reviewed and approved by the Local Bioethics Committee of the Institute of Cytology RAS, protocol no. 2. The copy of the approval by the Bioethics Committee of the Institute of Cytology is available upon request. hMESCs have a positive expression of CD 73, CD 90, CD 105, CD 13, CD 29, and CD 44 markers and absence of expression of the hematopoietic cell surface antigens CD 19, CD 34, CD 45, CD 117, CD 130, and Human Leukocyte Antigens (HLADR) (class II). Multipotency of isolated hMESCs was confirmed by their ability to differentiate into other mesodermal cell types, such as osteocytes and adipocytes. These cells are characterized by high rate of cell proliferation (doubling time 22–23 h). hMESCs at early passages (between 8 and 12 passages) were used in all experiments to avoid complications of replicative senescence. hMESCs were cultured in complete medium DMEM/F12 (Gibco BRL, USA) supplemented with 10% FBS (HyClone, USA), 1% penicillin–streptomycin (Gibco BRL, USA), and 1% glutamax (Gibco BRL, USA) at 37°C in humidified incubator, containing 5% CO₂. Cells were harvested by trypsinization and seeded at a density of 15 × 10³ cells/cm².

Single guide RNAs design

Single guide RNAs (sgRNAs) for modulation of PAI-1 expression were designed using the CCTop-CRISPR/Cas9 target online predictor and the CRISPR-ERA web applications in accordance with generally accepted rules [34]. Briefly, sgRNA sequences of 20 nucleotides in length were projected according to the common formula 5′ GN18G 3′ (PAM: NGG) to the promoter region of the gene from −200 to 0 bp relative to the transcription start site for transactivation and to the first constitutive exon region for knockout. Selected sgRNAs were further filtered by efficiency and specificity with applications' scores and were additionally checked for specificity in BLAST [35]. sgRNAs complementary to off-targets with more than 16 nucleotides were cut from the design.

Lentivirus vector constructs

In order to determine optimal parameters for hMESCs transduction, the FgH1tUTG plasmid with enhanced green fluorescent protein (GFP) reporter was used (a gift from Marco Herold, Addgene plasmid no. 70183). For CRISPR-mediated PAI-1 expression modulation, lentidCAS-VP64_Blast, lentiMS2-P65-HSF1_Hygro, lentisgRNA(MS2)_zeo backbone, and lentiCRISPR v2 plasmids were used (gifts from Feng Zhang, Addgene plasmid nos. 61425, 61426, 61427, and 52961). For overexpression and knockout, sgRNA coding sequences were cloned into lentisgRNA(MS2)_zeo backbone and lentiCRISPR v2 vectors, respectively, following the protocols described in Shalem et al. and Sanjana et al. [36,37]. Briefly, for construction of each sgRNA expressing vector pair, oligonucleotides that included 5′ 5 bp overhang for the forward (CACCG) oligonucleotide and 5′ 4 bp overhang and 3′ 1 bp overhang for the reverse (AAAC, C) oligonucleotide were annealed and ligated to BsmBI-digested, dephosphorylated, and gel-purified vector plasmids, after that Stbl3 bacteria were transformed with ligation reaction mixtures. All generated constructs were further checked for correct insertions by colony PCR using universal primer to U6 promoter as forward. All oligonucleotides sequences were purchased (Evrogen, Russian Federation); sequences are listed in Table 1.

Table 1.

Oligonucleotides used for sgRNA coding sequences cloning.

Oligonucleotide	Sequence
SAM sgRNA1 – f	5′-CACCGGAATGCTCTTACACACGTAC-3′
SAM sgRNA1 – r	5′-AAACGTACGTGTGTAAGAGCATTCC-3′
SAM sgRNA2 – f	5′-CACCGGGACCCGCTGGCTGTTCAGA-3′
SAM sgRNA2 – r	5′-AAACTCTGAACAGCCAGCGGGTCCC-3′
KO sgRNA1 – f	5′-CACCGCTCCTTGTACAGATGCCGG-3′
KO sgRNA1 – r	5′-AAACCCGGCATCTGTACAAGGAGC-3′
KO sgRNA2 – f	5′-CACCGTACAGATGCCGGAGGGCGG-3′
KO sgRNA2 – r	5′-AAACCCGCCCTCCGGCATCTGTAC-3′
U6 COLONY PCR – f	5′-GCCTATTTCCCATGATTCCTTCATATTTGC-3′

Lentivirus production and titration

Lentiviral particles were produced by transfection of 293T cells cultured in DMEM (Biolot, Russian Federation) containing 10% FBS (HyClone, USA) and 2% gentamicin (Biolot, Russian Federation) at 37°C in humidified incubator, containing 5% CO₂. For virus production, the cells were seeded at 4 × 10⁶ in 100-mm dishes in complete medium without addition of antibiotic. After overnight incubation, the medium was replaced to 9 ml fresh complete medium without antibiotic followed by addition of transfection mixtures prepared as follows: 10 mg of vector plasmid along with 7.5 mg of the packaging construct psPAX2 (a gift from Didier Trono, Addgene plasmid no. 12260) and 2.5 mg of the envelop plasmid pMD2.G (a gift from Didier Trono, Addgene plasmid no. 12259) were diluted and mixed in 1 ml of DMEM, after that 100 mg of linear polyethyenimine MW 25 kDa (Polysciences, USA) was added to the DNA and mixed for 15 s. Mixtures were incubated at room temperature for 15 min and gently added to the cells for 18 h. Then, the medium was changed to antibiotic-free complete medium with 1 mM NaB (Sigma-Aldrich, USA). Virus containing supernatants were collected at 48, 72, and 96 h after the transfection with the fresh medium (supplemented with NaB) replacement, passed through a 0.45-mm PES filters (Biolot, Russian Federation), centrifuged at 47,000g for 2 h at 4°C in polypropylene tubes (Beckman Coulter, USA), resuspended in hMESCs complete medium, aliquoted, and frozen at −80°C until titration and use.

For virus titration, 293T cells were used. The cells were plated at 10⁵ in 35 mm dishes in 1 ml of the complete medium and the next day were transduced with serially diluted viral stocks in fresh complete medium supplemented with 20 mg/ml of protamine sulfate (Ps) (Sigma-Aldrich, USA) for 18 h. Then, the medium was replaced to fresh complete medium and cells were cultured overnight. In case of fluorescent reporter gene expression, viral titer was estimated after 48 h since the virus addition by fluorescence-activated cell sorting (FACS) analysis. Virus titer was calculated as follows: TU/ml = (F × N × D)/V, where TU/ml is an amount of functional viral particles (Transducing Units) in 1 ml of stock solution, F is a percentage of GFP-positive cells (no more than 0.20% for reliable titer estimation), N is a number of cells at the time of transduction, D is a fold dilution of vector used for transduction, and V is a volume (ml) of diluted vector samples. In case of antibiotic resistance expression, the selection was started after 48 h since the virus addition, antibiotics working concentrations were 10 μg/ml blasticidin (Santa Crus Biotechnology, USA), 500 μg/ml hygromycin (Sigma-Aldrich, USA), 300 μg/ml zeocin (Sigma-Aldrich, USA), and 1 μg/ml puromycin (Sigma-Aldrich, USA). Selection was performed for 4 d with daily replacement of the medium followed by culturing selected cells in the complete medium for several days for well-detectable colonies formation. The viral titers based on colonies formation were calculated as follows: TU/ml = (N × D)/V, where N is a number of colonies on a dish at the end of titration (from 20 to 100 for reliable titer estimation), D is fold dilution of vector used for transduction, and V is a volume (ml) of diluted vector samples.

Lentiviral hMESCs transduction

For all experiments on transduction, hMESCs were seeded at 10⁵ in a 35-mm dish in 1 ml of the complete medium. Next day, the cells were transduced for 18 h according to experimental parameters with either polybrene (Pb) (Sigma-Aldrich, USA) or protamine sulfate (Ps) (Sigma-Aldrich, USA) with certain value of multiplicity of infection (MOI); MOI was calculated as follows: MOI = TU/N, where TU is an amount of functional viral particles required for infection and N is a number of cells at the moment of transduction. In 18 h after transduction the medium was replaced with the fresh one and transduced cells were maintained up to analyses. For optimization of hMESCs transduction parameters, cells were analyzed at 96 h after infection. In experiments on hMESCs secretome modulation at 96 h after infection, the selection was started, 2 μg/ml puromycin was used for knockout, and 600 μg/ml zeocin was used for transactivation and lentiviral control. Selection was carried out for 72 h, after that the medium was changed to fresh complete medium and cells were expanded for further experiments.

Flow cytometry

Effectiveness of viral transduction, cell viability, proliferation, cell size, and immunophenotyping were estimated by FACS. Flow cytometry was performed using the CytoFLEX (Beckman Coulter, USA) and the obtained data were analyzed using CytExpert software version 1.2. Adherent cells were rinsed twice with PBS and harvested by trypsinization. Four days after transduction, the efficiency was measured by detecting GFP fluorescence intensity (excitation: 488, emission: 510). In order to access cell viability after LV infection, 50 μg/ml propidium iodide (PI) was added to each sample just before analysis and mixed gently. To discriminate live and dead cells, two parameter cytograms were used (Log PI vs. Log FS). At least 3000 events from each sample were collected and analyzed at high sample delivery. The cell size was evaluated by cytometric light scattering of PI-negative cells. hMESCs surface markers were also assayed by flow cytometry. Cells in an amount of 10⁶ cells/ml were resuspended in PBS with 5% fetal bovine serum. Antibodies to CD 90 (no. 555596, BD Pharmingen, USA), CD 73 (no. 550257, BD Pharmingen, USA), CD 105 (no. 560809, BD Pharmingen, USA), CD 44 (no. 560568, BD Pharmingen, USA), and CD 13 (no. IM1427U, Beckman Coulter, USA) conjugated with phycoerythrin were used.

Migration analysis

Cell migration was determined using the xCELLigence RTCA DP system (ACEA Biosciences, USA). Cells were seeded into cell migration plate (CIM-Plate; used with the xCELLigence RTCA DP system) that contains electronically integrated Boyden chambers that provide, in real-time and without the use of labels, quantitative kinetic data for migration. The 15 × 10³ cells were seeded in serum-free media into the upper chamber; the lower chamber was filled with the completed growth media. hMESCs migrated from the upper chamber toward chemoattractant (serum-containing media) in the lower chamber passing through a membrane containing 8 μm pores and then adhered to gold impedance microelectrodes. The resultant change in impedance signal correlates with the number of cells attached to the electrodes.

Differentiation assay

For all experiments, hMESCs were seeded at 10⁵ cells in a 35-mm dish.

Adipogenic differentiation

Upon reaching 70% confluence, the medium was replaced to adipogenic medium containing complete hMESCs growth medium supplemented with 1 µM dexamethasone, 0.5 mM isobutyl methylxanthine, 200 μM indomethacin, and 10 μg/ml insulin (Sigma-Aldrich, USA) and cultured for 4 wk with medium changes every 2–3 d. The adipogenic capacity of hMESCs was evaluated by Oil Red staining. For the latter, cells were fixed with 4% paraformaldehyde for 30 min, washed with water, and then incubated for 30 min with a filtered 0.3% Oil Red (Sigma-Aldrich, USA) solution in 60% isopropanol. Oil Red-positive cells were identified by cytoplasmic fat globules.

Osteogenic differentiation

Upon reaching 90% confluence, the medium was changed to osteogenic medium containing complete hMESCs growth medium with 0.1 μM dexamethasone, 10 mM β-glycerol phosphoric acid, and 50 μM ascorbic acid (Sigma-Aldrich, USA) and cells were cultured for 4 wk with medium changes every 2–3 d. The osteogenic capacity of cells was evaluated by Alizarin red staining. For the latter, cells were fixed with 4% paraformaldehyde for 30 min, washed with water, and then incubated for 30 min with a filtered 1% Alizarin red (Sigma-Aldrich, USA) solution in deionized water. hMESCs osteogenic differentiation was determined by mineral depositions stained with Alizarin red.

Decidual differentiation

After the cells reached 80% density, the medium was changed for serum-free medium for the 24 h. The next day medium was replaced by fresh medium containing 2% of serum and 1 mM 8-bromoadenosine 3′,5′-cyclic monophosphate sodium salt (8-br-cAMP, Sigma-Aldrich, USA); such medium was exchanged on the third day. Undifferentiated cells were cultivated in the same medium but without 8-br-cAMP. In 7 d, media from both undifferentiated and differentiated cells were collected for prolactin and IGFBP1 testing.

ELISA

The amounts of prolactin and IGFBP-1 secreted by control or infected hMESCs were quantified in the cell supernatants (not concentrated conditioned media [CM]) by the Prolactine Human ELISA Kit (ab108679, Abcam, USA) and Human IGFBP-1 ELISA Kit (RAB0232-1KT, Sigma-Aldrich, USA). The data were normalized to the total amount of protein determined by the Bradford method. Positive and negative controls provided by the manufacturer were performed in parallel for comparisons.

SA-β-Gal activity

Cells expressing senescence-associated β-galactosidase (SA-β-Gal) were detected with the use of senescence β-galactosidase staining kit (Cell Signaling Technology, USA) according to manufacturer’s instructions. The kit detects β-galactosidase activity at pH 6.0 in cultured cells, which is present only in senescent cells and is not found in pre-senescent, quiescent, or immortal cells. Quantitative analysis of images reflecting SA-β-Gal staining was produced with the application of MatLab package, according to the algorithm described in the methodological paper [38]. For each experimental point, not less than 100 randomly selected cells were analyzed.

Immunoblotting

Immunoblotting analysis was performed as described previously [35]. SDS–PAGE electrophoresis, transfer to nitrocellulose membrane, and immunoblotting with ECL (Thermo Scientific, CA, USA) detection were performed according to standard manufacturer’s protocols (Bio-Rad Laboratories, USA). Antibodies against the following proteins were used: glyceraldehyde-3-phosphate dehydrogenase (clone 14C10) (1:1000, no. 2118, Cell Signaling, USA), phospho-Rb (Ser807/811) (1:1000, no. 8516, Cell Signaling, USA), phospho-p53 (Ser15) (clone 16G8) (1:700, no. 9286, Cell Signaling, USA), p21^Waf1/Cip1 (clone 12D1) (1:1000, no. 2947, Cell Signaling, USA), IGFBP-3 (1:1000, no. 13216, Cell Signaling, USA), PAI-1 (D9C4) (1:1000, no. 11907, Cell Signaling, USA), BAX (clone D2E11) (1:1000, no. 5023, Cell Signaling, USA) as well as horseradish peroxidase-conjugated goat anti-rabbit IG (GAR-HRP, Cell Signaling, USA) (1:10,000) and antimouse IG (GAM-HRP, Cell Signaling, USA) (1:10,000). Hyperfilm (CEA) was from Amersham (Sweden). Equal protein loading was confirmed by Ponceau S (Sigma-Aldrich, USA) staining.

Preparation and concentration of hMESCs CM

Control- and gene-modified hMESCs were seeded at a density of 15 × 10³ cells/cm² and cultured for 24 h in complete medium DMEM/F12 with 10% FBS. Then, the medium was replaced with fresh DMEM/F12 without FBS for 16 h. Culture supernatants were collected and centrifuged at 2500g for 10 min at +4°C to remove cells and cellular debris. The clarified supernatants were then concentrated with a 3000-Da-cutoff Centriprep spin columns (Millipore, USA). The samples were further concentrated using a 3000-Da-cutoff Microcon spin columns (Amicon, USA). For WB, samples were supplemented with Proteinase inhibitor cocktail (Sigma-Aldrich, USA) and lysed with gold lysis buffer [1% Triton X-100, 30 mM Tris–HCl (pH 8.0), 137 mM sodium chloride, 15% glycerol, 5 mM EDTA].

Bioinformatics

In order to determine biological functions of the selected factor in hMECSs CM, the STRING database was used to create a protein network of its most significant interactions. A set of 10 most significant interactors with a high degree of significance (interaction score >0.700) in accordance with the data of text-mining experiments, databases, and coexpression was generated. Further functional enrichment analysis of the predicted set of proteins was carried out in biological processes gene ontology (GO) terms using the cluster-profiler package in R software.

Statistical analysis

All quantitative data were shown as mean ± SD. To get significance in the difference between two groups, two-sided t-test or Wilcoxon–Mann–Whitney rank sum test was applied. For multiple comparisons between groups, ANOVA with Tukey HSD (honestly significant difference) test were used. All statistical analysis was performed using R software (P < 0.001; P < 0.01; P < 0.05, versus control). To perform functional enrichment analysis, the Benjamini method was used to control the false-discovery rate to correct the P-value.

Results

The perspective to apply genetically modified stem cells or their CM in regenerative medicine imposes certain limitations. The most obvious seems to be the maintenance of the unique MSCs properties after the viral infection, including proliferative activity, CD expression profile, and differentiation potential. Thereby, here we elaborated the efficient gene delivery methods in order to achieve high and stable levels of transgene expression without adversely affecting on the main hMESCs properties.

Optimization of the hMESCs lentiviral transduction parameters

First, we focused on the optimization of hMESCs transduction parameters that, on the one hand, would be effective and, on the other hand, would not affect cell viability. To this end, hMESCs were transduced with lentiviruses (LV) that encode GFP and transduction efficiency was determined by analyzing GFP fluorescence by flow cytometry. It is well established that the infection efficiency of primary cell cultures with LV is rather low; thereby, various additives are used in order to increase it [23]. Here, we applied two different agents – polybrene (Pb), also known as hexadimethrine bromide, and polycationic peptide protamine sulfate (Ps), commonly used to enhance transduction efficiency [23,39,40]. Our results presented in Figure 1(a) are in line with the existing data, at the MOI equal to one, we were able to achieve only 13% of GFP-positive cells when infecting hMESCs with LV alone, whereas in presence of 1 µg/ml Ps or 1 µg/ml Pb, it increased up to 20% or 50%, respectively (Figure 1(a)).

Figure 1.

Optimization of the hMESCs transduction parameters. hMESCs were infected with LV containing a reporter gene that encodes GFP. The transduction efficiency was estimated by FACS analysis of the green fluorescent intensity. (a) The effectiveness of hMESCs transduction by LV alone (MOI = 5) or in presence of either 1 µg/ml Pb or 1 µg/ml Ps determined on d 4 after infection. (b) Pb and Ps concentrations titration. hMESCs were incubated with different concentrations of Pb or Ps at a constant concentration of LV coding for a GFP (MOI = 5). (c) Multiplicity of infection titration. Estimation of transduction efficiency of hMESCs incubated at different MOIs, but at the constant concentrations of agents (8 µg/ml Pb or 50 µg/ml Ps). (d) The impact of various concentrations of Pb or Ps on the viability of transduced hMESCs determined by FACS. All experiments were performed in triplicates. (e) Stability of GFP expression in hMESCs transduced by either scheme: 4 μg/ml Pb with MOI = 5 or 20 μg/ml Ps with MOI = 20 determined either by FACS or immunofluorescence. Values are M ± std. dev. *P < 0.05; ***P < 0.001 by Student’s t-test. ns: Nonsignificant. Representative images are shown.

The initial series of the experiments were performed to choose hMESCs transduction parameters with each compound either by varying concentrations of agents or MOIs. As shown in Figure 1(b,c), approximately equal transduction efficiencies (over 80%) were obtained with MOI = 5 for Pb and MOI = 20 for Ps (Figure 1(c)).

We then estimated cell viability after the LV infection with various concentrations of Pb and Ps. As can be seen in Figure 1(d), Ps in a wide concentration range (from 10 up to 100 µg/ml) had no significant effect on hMESCs viability. However, we observed a dose-dependent decrease in cell number upon Pb application so that only about 60% of cells remained viable in 24 h after transduction with 8 µg/ml Pb (Figure 1(d)).

Based on the results presented in Figure 1(a–d), for the further experiments, we selected the following transduction parameters – 4 μg/ml Pb with MOI = 5 and 20 μg/ml Ps with MOI = 20. To check the stability of transgene expression in hMESCs under the selected infection parameters, cells were transduced at indicated parameters and cultured for 5 d, then cells were reseeded and additionally cultured for 4 d with the following estimation of GFP fluorescence. Notably, the percent of GFP expressing cells remained almost constant and was over 80%, despite the agent applied, indicating stability of transgene expression (Figure 1(e)).

The impact of LV infection on the main stem-cell properties of hMESCs

Having chosen MOIs and concentrations for effective transduction with both compounds, further we checked whether transduction and insertion of the transgene affect hMESCs properties. One of the main defining features of MSCs is expression of a panel of specific cell surface antigens. According to the minimal criteria published by the International Society for Cellular Therapy, over 95% of the MSCs population must express CD 90, CD 73, CD 105, CD 13, and CD 44. By using FACS, we assessed the expression levels of the indicated above surface antigens after the MSCs viral transduction either with Pb or Ps and compared it to those of noninfected control cells. Despite the agent used transduction had no effect on the expression profile of hMESCs, as far as more than 95% of cells remained positive for each CD antigen (Figure 2).

Figure 2.

hMESCs immunophenotyping. Expression of the surface markers in control (Ctr) cells and in hMESCs transduced in presence of Pb (Pb + LV) or Ps (Ps + LV) was determined by FACS with the use of phycoerythrin-conjugated antibodies. Images are representatives of at least three independent experiments. ISO: Isotype control. The percentages displayed in the figure reflect the fraction of antigen positive cells.

Another core characteristic of MSCs is self-renewal that reflects ability to give rise to cells itself without differentiation. The use of genetically modified MSCs or their secretome requires profound cell expansion; thus, maintenance of proliferation of modified cells seems to be crucial. Thereby, further we estimated the impact of both transduction schemes on the proliferation rate of hMESCs. As shown in Figure 3(a), Pb addition during LV infection led to a significant reduction in the number of proliferating cells throughout the whole observation period. Notably, Ps addition had no effect on proliferation as compared to control cells (Figure 3(a)).

Figure 3.

(a) Growth curves of control (Ctr), Pb + LV, and Ps + LV hMESCs. Cell number was determined by FACS at indicated time points. (b) Migration analysis was obtained with the use of XCelligence system (the precise description in Materials and methods section). Control (Ctr), Pb + LV, and Ps + LV hMESCs were seeded into the upper chamber of the migration plate in 4 d after infection. Cell migration was estimated for 3 d and the number of migrated cells was determined by the resultant change in impedance signal. Values are M ± std. dev. **P < 0.01 by Student’s t-test. All experiments were performed in triplicates.

The next fundamental feature of MSCs is the migration capability. In order to assess possible effects of LV infection with either Ps or Pb on hMESCs mobility, we used the label-free real-time impedance-based xCELLigence system. We revealed that transduction with either compound negatively affected hMESCs migration during the observation period; however, adverse effect of Pb was far more pronounced (Figure 3(b)).

Finally, we analyzed differentiation potential of either control or LV infected hMESCs. Originally, hMESCs localized in human endometrium that lines the inner-surface of uterine cavity undergo differentiation into specialized decidual cells during each menstrual cycle in response to elevated progesterone levels and increased cAMP production [24]. Therefore, we examined possible impact of LV infection in presence of Pb or Ps on the ability of hMESCs to differentiate into decidual cells. By using ELISA, we revealed that neither compound added during LV infection had an impact on the levels of prolactin (PRL) and IGFBP1 secretion, the most reliable markers of decidualzation, indicating the absence of any negative effects on decidual differentiation (Figure 4(a,b)) [41,42].

Figure 4.

The content of prolactin (PRL) (a) and IGFBP1 (b) in control (Ctr) and Pb + LV or Ps + LV transduced cells supernatants quantified by the ELISA and normalized to the total protein determined by the Bradford method. Decidual differentiation analysis was performed in 7 d after induction by 1 mM 8-br-cAMP. Values are M ± std. dev. All experiments were performed in triplicates.

Certainly, the unimpaired ability of hMESCs to generate decidual cells is insufficient to assert preservation of multipotency of hMESCs under LV infection conditions. Thus, further we estimated differentiation ability of hMESCs in adipogenic and osteogenic directions by Oil Red and Alizarin staining procedures revealing fatty drops or calcium deposits, respectively. Interestingly, upon Ps + LV infection, adipogenic and osteogenic differentiations were unaffected as compared to control cells, while Pb application led to remarkable impairment of either staining (Figure 5(a,b)).

Figure 5.

Effect of transduction either with Pb or Ps on hMESCs differentiation capacity. (a) Ctr, Pb + LV, and Ps + LV cells were plated at a density of 15 × 10³/cm² and cultured in complete growth media until 70% confluence, followed by switch to adipogenic induction media. Cells were stained with Oil Red at d 30 to visualize fat droplets. Images were taken at magnifications 10× and 40×. (b) Ctr, Pb + LV, and Ps + LV cells were plated at a density of 15 × 10³/cm² and cultured in complete growth media until 90% confluence, followed by switch to osteogenic induction media. Cells were stained with Alizarin Red at d 30 to visualize mineralized bone matrix. Images were taken at magnifications 4× and 10×. All experiments were performed in triplicates. Representative photomicrographs of the staining are shown.

Summarizing the obtained data, we can make several important conclusions: (1) infection efficiency of hMESCs with LV alone is very low; (2) addition of Pb or Ps can significantly enhance hMESCs infection efficiency without affecting cell viability; (3) hMESCs transduction with Ps has no or little effect on cell properties, while Pb has the adverse effects on the main stem-cell properties, namely on differentiation potential, migration, and proliferation ability. Overall, in terms of MSCs secretome modification, genetic manipulations with the use of Ps are preferable, whereas Pb application is undesirable.

Transduction with Pb induces senescence in a part of hMESCs population

According to the modern data, reduced proliferation and migration rates as well as the decreased differentiation capability of MSCs can testify in favor of senescence progression in the population [43]. Previously, we have convincingly shown that senescence is more preferable reaction of hMESCs in response to damage or stress than apoptosis [44,45]. Based on the data described above, further we analyzed whether senescence induction can be the underlying cause of the observed negative effects of LV infection with Pb. The main features of senescent cells, besides proliferation reduction, are enlarged and flattened cell morphology, emergence of SA-β-Gal activity (SA-β-Gal staining) as well as the activation of the p53/p21Waf1/Cip1/Retinoblastoma protein (Rb) signaling pathway.

First of all, we checked whether Pb + LV treatment led to hMESCs hypertrophy. Indeed, we observed an increase in cell size already in 4 d after transduction (Figure 6(a)).

Figure 6.

Pb + LV transduction resulted in premature senescence induction in the part of hMESCs population. (a) Pb + LV-induced cell size increase. Ps + LV had no effect on cell size. Forward scatter (FS) reflects the average cell size. Values are M ± std. dev. ***P < 0.005 by Student’s t-test. (b) Activation of p53/p21/Rb pathway in hMESCs. Western blot analysis of p53 and Rb phosphorylation levels and p21 protein expression performed at indicated time points. Representative results of the three experiments are shown in the figure. GAPDH was used as loading control. (c) SA‐β‐Gal staining of control, Pb + LV, and Ps + LV infected hMESCs. In 6 d after the infection, cells were reseeded and additionally cultured for 3 d, in order to perform staining in nonconfluent cultures. Scale bar is 500 μm and valid for all images. Quantification of β‐galactosidase activity values. ***P < 0.005 by Mann–Whitney test. Ctr: Control cells.

Furthermore, cell infection in presence of Pb resulted in enhanced SA-β-Gal staining (Figure 6(c)). We then analyzed the activity of the canonical p53/p21/Rb pathway mediating senescence initiation and progression. As shown in Figure 6(b), hMESCs transduction in presence of Pb led to a slight increase in p53 phosphorylation levels, elevation of p21Waf1/Cip1 (p21) expression, and hypophosphorylation of Rb protein throughout the whole observation time. Collectively, these data suggest that hMESCs transduction in presence of Pb results in the premature senescence induction via p53/p21/Rb pathway. In contrast, hMESCs transduction had almost no effect on cell size, SA-β-Gal activity, and activation of the p53/p21/Rb pathway in presence of Ps (Figure 6), what is consistent with the absence of any negative effects of Ps on proliferation and differentiation of hMESCs (Figures 3(a), 5(a,b)). Together, these data provide an additional advantage for Ps application to increase hMESCs transduction efficiency.

Effective CRISPR-based hMESCs genetic modification

In order to validate optimized conditions for effective hMESCs transduction, we selected SERPINE-1 gene coding PAI-1 protein for hMESCs genetic modification. According to our unpublished dataset obtained with the use of high-resolution mass spectrometry, we revealed a relatively high content of PAI-1 among a plenty of identified proteins secreted by hMESCs. When analyzing results of text-mining experiments, databases, and coexpression data from the STRING database, we observed that PAI-1 and its close interactors participate in numerous processes, such as response to wounding, regulation of blood coagulation, cell motility and adhesion, critical for proteolysis, and extracellular matrix remodeling that indeed are important during embryo implantation (Figure 7(a,b)).

Figure 7.

(a) Protein network for 10 most significant PAI-1 interactors created with the use of STRING database. (b) Functional enrichment analysis of PAI-1 interaction network in gene ontology terms. (c) Schematic representation of sgRNAs design for SAM and GeCKO systems. (d) Western blot analysis of PAI-1 expression in control (Ctr), knockout (KO), and overexpressing (SAM) hMESCs. Representative results are shown. GAPDH was used as loading control. (e) Levels of extracellular PAI-1 secreted by control (Ctr), knockout (KO), and overexpressing (SAM) hMESCs. Conditioned media was collected and concentrated according to the procedure described in Materials and Methods section. Secreted IGFBP3 and BAX were used as a loading controls.

In order to modify hMESCs secretome content, we applied one of the most powerful tools for genome editing termed CRISPR technology. Today, various approaches for CRISPR-based regulation of gene expression are developed, among which lentiviral CRISPR/Cas9 knockout (GeCKO) and CRISPR/Cas9 synergistic activation mediator (SAM) are some of the most efficient for gene knockout and overexpression, respectively [46]. For hMESCs secretome modulation, we used either one-vector-based GeCKO format for knockout or three-vector SAM format for activation. Based on the known sgRNA targeting rules, we designed two sgRNAs for each modification system so that in case of GeCKO targeted sequences were in the first 5′-exone of PAI-1 gene, whereas in case of SAM, they localized within the 200-bp upstream of the transcriptional start site [34] (Figure 7(c)). We obtained five LV types: two for PAI-1 knockout (with different sgRNA sequences inserted), two for PAI-1 overexpression (with different sgRNA sequences inserted), and the one that was used as transduction control (containing sgRNA designed for SAM system but without Cas9). We then infected hMESCs using the optimal transduction scheme designed for Ps described in the first part of the Results section. As shown in Figure 7(d), we were able to obtain PAI-1 knockout using sgRNA1 and PAI-1 overexpression using sgRNA2. These results highlight the necessity of more than one sgRNA sequences construction. In order to check whether such genetic manipulations of hMESCs would be suitable for modulation of PAI-1 content within the secretome, we then expanded genetically modified and control hMESCs, collected CM, and concentrated it for further immunoblotting. As shown in Figure 7(e), CM from PAI-1 knockout and overexpressing hMESCs contained reduced and elevated levels of PAI-1, respectively, as compared to CM collected from control cells. Importantly, the levels of other proteins secreted by hMESCs, namely, IGFBP3 and BAX, remained almost constant independently of producing cells, testifying to the high accuracy of the CRISPR approach. These data confirm the effectiveness of our optimized protocol of hMESCs LV transduction and demonstrate the possibility of their targeted genome and secretome manipulation with the use of either CRISPR-based systems.

Discussion

Not so long ago, the perspective of MSCs clinical application seemed elusive. Today, a plenty of impressive results regarding the use of stem-cell-based therapies for treatment cardiovascular, immunological diseases, spinal cord injury, diabetes, cancer, etc. are obtained [47]. In keeping with the latter observations, the paracrine activity of MSCs is assumed to underlie the positive effects of stem-cell-based therapy [18–21,47]. These data inspired researches to develop approaches to augment MSCs paracrine activity. In this context, genetic modification allows to manage MSCs secretome precisely by varying the expression levels of the exact genes (reviewed in [21]).

Though the potential of hMESCs in regenerative medicine has began to be explored only in the last decade and most of available clinical studies are still ongoing, there is some evidence confirming therapeutic effects of hMESCs secreted factors (reviewed in [8]). For example, hMESCs injected either by intravenous or intratumoral routes successfully reduced tumor size, at least in part due to the secretion of inhibitory factors as considered by the authors [48]. Furthermore, vascular endothelial growth factor (VEGF), brain-derived neurotrophic factor, and neurotrophin-3 were detected as possible neuroprotective trophic factors released by hMESCs [8,49]. Despite the numerous studies published on hMESCs properties in the past years, only a few data describe their genetic modulation [50,51]. Moreover, there is an informational gap regarding the impact of transduction itself on the fate of hMESCs.

MSCs gene editing has been described using both viral and nonviral (physical and chemical) genetic constructions delivery [22]. Though the nonviral delivery techniques can be easily carried out, low efficiency, transient transgene expression, inability to transfect large cell populations as well as the risk of cell membrane damage limit their use for primary cultures [52,53]. These notions tended us to select viral delivery method; particularly, we have chosen LV, as they integrate their vector genome into the host genome of dividing, nondividing, or slow-dividing cells, without affecting their viability and differentiation potential what is necessary to obtain successful long-term transgene expression for further modifying of the secreted factors.

However, we and others revealed that incubating MSCs with LV alone has rather low transduction efficiency [23]. Thus, here we applied two different positively charged polycations – Pb and Ps – that are known to reduce the electrostatic repulsion forces between a negatively charged cell and an enveloped lentiviral particle [39,54]. Although a lot of research is done on MSCs with the use of LV infection, the relative effects of various types of polycations on the cell fate are not fully understood. Moreover, the data regarding effective lentiviral transduction of hMESCs are almost lacking. Based on the MOIs and agents concentrations titrations, we were able to develop two transduction schemes for each agent that led to a significant increase in infection efficiency of hMESCs as compared to LV alone.

Important results were obtained when we analyzed the fate of hMESCs transduced with Pb. hMESCs infected in presence of Pb exhibited reduced proliferation and migration rates and impaired adipo- and osteogenic differentiation abilities. Similar effects of Pb on the growth rate were described for MSCs derived from bone marrow and for keratinocyte stem cells [23,55]. There are preliminary data regarding the negative effect of Pb on MSCs differentiation potential [23]. Contrarily to our results, Zhao et al. did not observe any side effects of 40 µg/ml Pb on MSCs viability and successfully transduced cells using Pb, postulating that the agent is safe, effective, and might be routinely used for gene transfer in MSCs [56]. However, the last mentioned results were obtained on mouse cells, whereas in all data highlighting the negative Pb effects, human MSCs were used.

Unexpected finding was observed when we induced decidual differentiation of hMESCs transduced with Pb. Even though LV infection with Pb had a pronounced effect on both adipogenic and osteogenic differentiation of hMESCs, it has no impact on their decidual differentiation. This observation is most likely due to the predisposition of hMESCs toward decidualization. Recently, it was shown that hMESCs exhibited higher capacity for differentiation into decidual cells than bone marrow MSCs. Furthermore, in MSCs derived from adipose tissue, the decidualization ability was almost lacking [41].

According to our previous data together with the results described in a latter review, MSCs senescence is accompanied by reduced proliferation rate, decreased differentiation potential as well as impaired migratory ability, the outcomes that we revealed after hMESCs infection supplied by Pb [35,43–45]. Therefore, we speculated that the negative effects of transduction with Pb might be mediated by senescence initiation in a part of hMESCs population. In favor of this assumption, we detected such features typical for senescent cells, as cell hypertrophy, increased SA-β-Gal staining, and p53/p21/Rb activation. To our knowledge, it is the first evidence that MSCs infection in presence of Pb is accompanied by senescence induction. Probably, the underlying cause of the adverse effects of Pb + LV infection on MSCs proliferation rate observed previously by other authors might also be cell senescence [23]. Of note, in the modern studies, Pb is actively used to increase the effectiveness of MSCs transduction for either purpose, what with regard to our results may somehow distort the data obtained by the authors [57–60].

Based on our results, Ps might be used as an excellent alternative to Pb. On the one hand, by using Ps, it is possible to achieve similar enhancement of hMESCs transduction efficiency as with the Pb. And, on the other hand, hMESCs transduction in presence of Ps has no significant impact on the main stem-cell properties, including CD expression profile, differentiation, migration, and proliferation capabilities. The proposal that Ps might be used as an alternative to Pb during MSCs LV transduction was stated by Lin’s group [40]. They observed that high concentrations of Ps (100 µg/ml) can result in a good transduction efficiency without affecting MSCs proliferation and differentiation abilities. In line with their results, we obtained almost equal transduction efficiency (about 36–37%) using the same parameters 50 µg/ml Ps MOI = 5. In order to achieve more pronounced effect of Ps on the infection efficiency, Lin et al. increased Ps concentration up to 100 µg/ml and additionally applied FGF-2. However, according to our data, there was no need to increase Ps concentration higher than 20 µg/ml, as transduction efficiency reached plateau at that point. Thereby, we increased MOI up to 20 transducing units per cell what allowed us to obtain more than 80% of GFP-positive hMESCs.

In keeping with the latter observations, managing MSCs secretome via genetic manipulations is on the cutting edge of stem-cell-based therapy. For example, IL10-overexpressing MSCs turned out to exert a synergistic anti-inflammatory effect to alleviate cardiac injury after myocardial infarction [61]. Moreover, VEGF-engineered MSCs may improve the transplantation strategy for the Parkinson’s disease treatment [62]. Additionally, Flt-1-engineered MSCs caused regression of endometriotic lesions [50]. A plenty of other fascinating results are obtained by applying MSCs with genetically modified paracrine activity [57,58,63,64]. However, in the vast majority of studies, the gain-of-function genetic perturbations are obtained by the use of cDNA overexpression systems. CDNA overexpression has several crucial complications: it is difficult to capture the complexity of transcript isoform variance using these libraries, and large cDNA sequences are often difficult to clone into size-limited viral expression vectors [36]. In order to overcome these limitations, guided nucleases can be used. In the context of MSCs secretome modification, we were able to find the only article where HGF-secreting MSCs were obtained by TALEN-mediated genome editing [64]. In the present article, we applied another type of guided nucleases, CRISPR/Cas9, that is characterized by target design simplicity and possibility of multiplex gene expression regulation compared to TALENs and ZFNs [65]. Notably, by using CRISPR-Cas9 knockout and transcriptional activation systems, we were able to create both PAI-1 knockout and PAI-1 overexpressing hMESCs, respectively. Finally, we managed to obtain either PAI-1 enriched or depleted CM, reflecting the effectiveness and specificity of the chosen approach. Thereby, the present article provides the first evidence of successful and effective MSCs secretome managing via CRISPR/Cas9 genome editing technology.

The results obtained within the present study trigger our future investigations. Taking into account important role of PAI-1 in regulating implantation process, we are planning to check the ability of PAI-1 knockout and PAI-1 overexpressing hMESCs toward decidual differentiation as well as to test the impact of PAI-1 expression levels on in vitro implantation model. Also, we are going to use modified CM to improve implantation using model of impaired decidualization.

Funding Statement

This research was funded by the support of the Russian Foundation for Basic Research: [18-29-09101-mk].

Acknowledgments

The authors are thankful to A.P. Domnina and M.A. Shilina (Institute of Cytology, RAS) for providing reagents for MSCs differentiations and to A.D. Nikotina (Institute of Cytology, RAS) for kind assistance with XCELLigence system.

Disclosure statement

No potential conflict of interest was reported by the authors.