Lentiviral-based reporter constructs for profiling chondrogenic activity in primary equine cell populations

Abstract

Successful clinical translation of mesenchymal stem cell (MSC) based therapies for cartilage repair will likely require the implementation of standardized protocols and broadly applicable tools that facilitate comparisons among cell types and chondroinduction methods. The present study investigated the utility of recombinant lentiviral reporter vectors as reliable tools for comparing chondrogenic potential among primary cell populations and distinguishing cellular-level variations of chondrogenic activity in widely used three-dimensional (3D) culture systems. Primary equine MSCs and chondrocytes were transduced with vectors containing combinations of fluorescent and luciferase reporter genes under constitutive cytomeglavirus (CMV) or chondrocyte-lineage (Col2) promoters. Reporter activity was measured by fluorescence imaging and luciferase assay. In 3D cultures of MSC aggregates and polyethylene glycol-hyaluronic acid (PEG-HA) hydrogels, transforming growth factor beta 3 (TGF-β3)-mediated chondroinduction increased Col2 reporter activity, demonstrating close correlation with histology and mRNA expression levels of COL2A1 and SOX9. Comparison of chondrogenic activities among MSC populations using a secretable luciferase reporter revealed enhanced chondrogenesis of bone marrow-derived MSCs relative to MSC populations from synovium and adipose tissues. A dual fluorescence reporter – enabling discrimination of highly chondrogenic (Col2-GFP) cells within an MSC population (CMV-tdtomato) – revealed marked heterogeneity in differentiating aggregate cultures, and identified chondrogenic cells in chondrocyte-seeded PEG-HA hydrogels after 6 weeks in a subcutaneous implant model – indicating stable, long-term reporter expression in vivo. These results suggested that lentiviral reporter vectors may be used to address fundamental questions regarding chondrogenic activity in chondroprogenitor cell populations and accelerate clinical translation of cell-based cartilage repair strategies.

Graphical Abstract

Introduction

For over two decades, adult Mesenchymal Stem/Progenitor Cells (MSCs) have been regarded as highly preferable cell type for autologous cell-based cartilage repair (Matsumoto et al. 2010; Roelofs et al. 2013). Their prevalence in many tissues, high expansion potential and accessibility overcome many of the problems associated with the use of articular chondrocytes, including limited availability, donor site morbidity and phenotypic instability (Jayasuriya et al., 2015; Matricali et al., 2010). However, the clinical advancement of MSC and other cell-based therapies for cartilage repair has been hindered by the wide variation in pre-clinical models used among laboratories and associated differences with cell culture, differentiation protocols and implantation strategies. As a result, despite numerous reports of improved healing in animal studies (Goldberg et al., 2017; Lo Monaco et al., 2018; Huang et al., 2017; Bornes et al., 2014), there is still no consensus regarding efficacy among strategies used, and a clear pathway to clinical translation has yet to emerge. Within the MSC field, this disparity is exacerbated by the poor characterization of primary cell populations and the lack of standardized tools for quantifying chondrogenic potential and cartilage repair. In particular, using current techniques, it is difficult to determine if MSCs uniformly undergo chondrogenesis or directly participate in neocartilage formation in vivo.

A consensus regarding which autologous cell sources, chondrocytes or various MSC subtypes, are most useful for cartilage repair has also yet to be established. MSCs from bone marrow, (Johnstone et al., 1998; Mackay et al., 1998) fat (Nathan et al., 2003; Wickham et al., 2003; Khan et al., 2012; Veronesi et al., 2013) and synovium (Sakaguchi et al., 2005; Shirasawa et al., 2006; Pei et al., 2008) have been widely investigated, and purportedly enhanced chondrogenic activity has been reported for synovial-derived MSCs (Sakaguchi et al., 2005; Yoshimura et al., 2007) and more recently, cartilage-derived stem cells (Williams et al., 2010; McCarthy et al., 2012). However the lack of universally adopted methods for assessing chondrogenic potential prevents meaningful study-to-study comparisons. Histological-based assessments of MSC chondrogenesis in vitro and in vivo are essential for visualizing cartilage ECM components and the nature of chondrocyte differentiation, but provide limited sensitivity for meaningful quantitative comparisons of chondrogenesis among different cell preparations (Orth et al., 2015). On the other hand, biochemical and molecular assays of chondrocyte markers or ECM synthesis offer quantitative measures, but often require pooling of samples, and do not distinguish cellular-level responses (McLeod et al., 2017).

Transcriptional reporter systems comprised of fluorescent or bioluminescent cDNAs under regulatory control of promoters or response elements of chondrocyte marker genes including ACAN (Hung et al., 2000; Palmer et al., 2001) COL11A2 (Outani et al., 2013) and COL2A1 (Stokes et al., 2001; Kan et al., 2009; Padmashali et al., 2014; Porter et al., 2012), are used as quantitative indices of chondrocyte activity in cell culture and transgenic models. However, their application as tools to monitor cartilage differentiation and repair in cell- and tissue engineering protocols has not been explored. In the present study, the utility of lentiviral vector-based, transcriptional reporter constructs that incorporate a modified collagen type 2 (Col2) promoter was explored for evaluation of chondrogenesis of post-natal, primary equine MSCs and chondrocytes. The vectors, containing bicistronic or independent expression cassettes encoding combinations of fluorescence and luciferase-based reporters, were used to demonstrate chondrocyte-specific activity and evaluate chondrogenic potential within and among primary cell populations. Reporter activity was measured by fluorescence imaging, flow cytometry or luciferase assay in widely used cell aggregate and hydrogel culture models of chondrogenesis and validated by conventional assessments of chondrogenic differentiation. The utility of these reporters for assessment of cell fate after in vivo implantation was also evaluated.

Materials and Methods

Generation of recombinant lentiviral reporter vectors

Bicistronic lentiviral vector - Lv.Col2-GFP/fLuc

A promoter sequence was synthesized based on a previous design (Kan et al., 2009): four repeats of a highly conserved enhancer region within the first intron of COL2A1 (+2126/+2174), which contains a SOX9 binding motif (Zhou et al., 1998), were positioned upstream of the core COL2A1 promoter (−164/+37). These regulatory sequences were flanked by restriction sites for SpeI (5’ end) and BamHI (3’ end), and an NheI site was placed between the four enhancer repeats and the core COL2A1 promoter. This promoter was synthesized by GenScript (Piscataway, NJ) and cloned in place of a truncated, “minimal” cytomegalovirus (mCMV) promoter within the third-generation lentiviral expression plasmid pTRH1-mCMV-dscGFP-T2A-Fluc (System Biosciences, Mountain View, CA). In the resulting reporter construct, pGF-4eCOL2A1 (Addgene ID# 97210), the promoter drives co-expression of copepod green fluorescent protein (copGFP) and firefly luciferase using the Thosea Asigna virus 2A (T2A) self-cleaving peptide (Kim et al., 2011). Replication-deficient lentivirus was generated by transfecting 293T cells with pGF-4eCOL2A1 and the third-generation packaging plasmids pMDLg/pRRE (Addgene ID# 12251), pRSV-Rev (Addgene ID# 12253), and pMD2.G (Addgene ID# 12259) (Dull et al., 1998).

Dual promoter reporter vectors - Lv.CMV-mLuc/EF1-GFP, Lv.Col2-mLuc/EF1-GFP and Lv.CMV-tom/Col2-GFP

For generation of dual lentiviral reporter vectors with independent expression cassettes under separate regulation of constitutive (CMV or EF1α) and chondrogenic (Col2) promoters, the pcDH lentiviral expression system (System Biosciences) was used. A modified pcDH lentiviral expression plasmid, pcDH-CMV-mLuc/EF1-GFP, containing independent CMV-mLuc and EF1-GFP expression cassettes was generated by cloning the full-length cDNA of Metridia luciferase (mLuc) from the pMetLuc reporter vector (Takara Bio, Mountain View, CA) into the multiple cloning site of pcDH-CMV-MCS-EF1-copGFP via compatible BamH1/Not1 restriction sites. To generate a lentiviral expression plasmid with mLuc under control of a Col2 promoter, pcDH-Col2-mLuc/EF1-GFP, the CMV promoter sequence was removed from pcDH-CMV-mLuc/EF1-GFP and replaced with the Col2 promoter sequence from pGF-4eCOL2A1 via compatible Spe1/BamH1 sites. For generation of a dual fluorescence reporter containing independent CMV-tom and Col2-GFP expression cassettes, the full length cDNA of tdtomato was obtained by BamH1/Not1 digestion of pCSCMV:tdtomato (Addgene ID# 30530) and cloned into the multiple cloning site of pcDH to generate pcDH-CMV-tom/EF1-GFP. Then the EF1-GFP cassette was then replaced with a Col2-GFP expression cassette flanked by Not1 (5’end; underlined in the primer sequence) and Sal1 (3’end; underlined in the primer sequence) restriction sites introduced by PCR amplification of pGF-4eCOL2A1 using the primers listed in Table 1, to generate a pcDH-CMV-tom/Col2-GFP lentiviral expression plasmid. For production of replication-deficient lentiviral vectors, each pcDH lentiviral expression plasmid was transfected into 293T cells with the second generation packaging plasmids pSPAX-2 (Addgene ID# 12260) and pMD2.G. Viral supernatants were harvested 48 h after transfection and either used directly for cell transduction, or concentrated to 1/50 th of the initial volume using the Lenti-X concentrator reagent and protocol (Takara).

Table 1.

Primers sequences to generate a pcDH-CMV-tom/Col2-GFP lentiviral expression plasmid.

Primer sequence	Orientation
5′-TTATCGATGCGGCCGCGAACTAGTCTGTGAATCG-3′	forward
5′-CGCATGTTAGTCGACTTCCTCTGCCCTCACCG-3′	reverse

Viral-titering

Quantitation of lentiviral vectors from culture supernatants or concentrated virus was performed using the QuickTiter™ Lentivirus Titer Kit (Cell Biolabs, San Diego, CA). Physical titers were determined based on detection of lentivirus associated HIV p24 and ranged from 2.5 ×10⁹ - 2 × 10¹⁰ vp/mL.

Cell and tissue harvest

Cartilage, bone-marrow, synovium and adipose tissues were collected from horses (age range: 2–12 years) euthanized for health issues unrelated to musculoskeletal conditions at the UF School of Veterinary Medicine, Gainesville, FL, USA.

Articular Chondrocytes

Intact, healthy cartilage was obtained from the femoropatellar, metacarpophalangeal and scapulohumeral joints, minced finely and digested with 0.2% collagenase type II (Worthington, Biochemical Corporation) for 12–16 hours at 37°C, 5% CO₂ in DMEM/F12 (Thermo Fisher, Carlsbad, CA) supplemented with 5% foetal bovine serum (FBS) (Hyclone Laboratories, South Logan, UT). Cell suspensions were filtered through a 70 μm strainer and plated at density of 50,000 cells/cm² in high glucose DMEM (Corning, Oneonta, NY) supplemented with 10% FBS. Within 3 days of harvest, primary chondrocytes were replated at high density until required.

Synoviocytes

The non-collagenous soft tissues, including the synovial lining and subsynovium, were scraped from the capsular tissues of the carpal joints, minced and digested in saline solution with 0.2% collagenase for 2 h at 37°C, 5% CO₂ as previously described (Watson et al., 2013). Non-adherent cells were removed after 24 h and, after 1 week, cells were replated at high density until experiments.

Bone Marrow-Derived MSCs

Bone marrow aspirate (~15 mL) was obtained via needle puncture of the iliac crest, mixed 1:1 with MSC isolation medium [low glucose DMEM with 20% FBS and 10 ng/ml FGF-2 (Peprotech, Rocky Hill, NJ)], and plated at 500,000 nucleated cells/cm². After 3 days, the media was replaced with MSC expansion medium (low glucose DMEM supplemented with 10% FBS and 10 ng/ml FGF-2) and the cells were cultured until reaching confluency (~10 d).

Synovial-derived MSCs

~2–3g of synovial tissue, were harvested from the intercarpal, and femoropatellar joints and digested with 0.1% trypsin in phosphate buffered saline (PBS) for 30 minutes followed by treatment for 2 h at 37°C, 5% CO₂ with 0.1% collagenase in serum-free high glucose DMEM, as previously described (He et al., 2009). After digestion, cells were filtered through a 45 μm strainer and plated until confluency (~7 d).

Adipose-derived MSCs

~8–10g of adipose tissue was harvested from the subcutaneous fat depot lateral to the tail head. The tissue was diced into small pieces and digested for 3 h at 37°C, 5% CO₂ with 0.2 % collagenase in serum-free DMEM. After digestion, cells were filtered through a 45 μm strainer and plated until confluency (~7 days). After initial plating in T-75 flasks (P0), all MSC populations were expanded at 5000 cells/cm² in MSC expansion medium until required.

To verify colony forming potential (CFU-F) of MSC populations, primary cell isolates were plated into 6-well plates and cultured in MSC expansion medium. After 10 d, colonies were fixed with 10% neutral buffered formalin (NBF) for 15 minutes and stained with 1% crystal violet for 1 h at room temperature. Colony numbers per 10³ nucleated cells were as follows: 0.02 ± 0.002 BM-MSCs; 12 ± 3.2 S-MSCs; and 2.9 ± 1.8; A-MSCs.

Lentiviral transduction

Primary cells (passage 2–4) were seeded into 6-well plates and the culture media replaced with 1 mL/well serum-free DMEM containing 8 μg/ml polybrene (Sigma-Aldrich) and 293T viral supernatants or concentrated virus. For transduction, cells were infected (dose range: 1.3 × 10³ - 6.4 × 10³ vp/cell) by centrifugation at 1200 × g for 60 min, followed by an additional 2 h in a tissue culture incubator at 37°C, 5% CO₂. Virus containing medium was then removed and replaced with growth medium, and cells were replated into T-150 flasks until required. Transduction efficiencies, determined by tdtomato or GFP flow cytometry were ~70% or higher for all viral vectors and cell types used in the study. At the vector doses used, impaired chondrogenic activity of equine MSCs was not observed (data not shown).

In vitro chondrogenesis

Aggregate culture

Lentiviral reporter-labelled MSCs and chondrocytes were trypsinized from monolayer cultures and seeded at 5 × 10⁵ cells/ml in 15 ml polypropylene tubes, centrifuged at 500 × g for 5 min to induce aggregate formation and maintained in 0.5 ml of chondrogenic medium consisting of serum-free, high glucose DMEM containing pyruvate (1 mM), 1% ITS™ Premix (BD Biosciences, San Jose, CA) ascorbate 2-phosphate (37.5 μg/ml), and dexamethasone (10⁻⁷ M) (Johnstone et al., 1998). To induce chondrogenesis, media was supplemented 10 ng/ml human transforming growth factor beta 3 (TGF-β3; Peprotech)

Photopolymerizable PEG-HA hydrogels

Labeled cells (2 × 10⁷ cells/ml) were mixed with a polymer solution of 10% w/v PEODA (Laysan Bio, Arab, AL), 0.05% w/v Irgacure 2959 photoinitiator (Sigma-Aldrich), and 2.5 mg/ml Hyaluronic Acid (1100 kDa; Lifecore Biomedical, Chaska, MN) in sterile PBS (Sharma et al., 2007). Next, the cell-polymer solution was pipetted into plastic moulds and placed under a long wave, 365-nm ultraviolet light at 4 mW/cm2 (Omincure S2000, Excelitas Technologies, Mississauga, Ontario, Canada) for 5 min to induce polymerisation and generate 5 × 3 mm cylindrical constructs. The constructs were then transferred to chondrogenic medium in 24-well plates for in vitro studies or implantation experiments.

Subcutaneous implantation

All procedures involving animals were conducted according to protocols approved by the University of Florida Institutional Animal Care and Use Committee. For surgery, athymic nude mice (Charles River, Wilmington MA) were anesthetized by isofluorane inhalation and prepped for surgery following sedation. Under aseptic conditions, subcutaneous pockets were made by generating two 1 cm dorsal midline incisions and gently spreading the cutaneous tissue lateral to each incision line creating a total of 4 pockets per animal. Sterile cell-seeded PEG-HA scaffolds, approximately 5 mm in diameter, 3mm thick, were then securely implanted into each pocket away from the incision area. Following wound closure using wound clips, the animals were removed from anesthesia and returned to free cage activity for the duration of the experiments.

Flow cytometry and cell sorting

For assessment of transfection efficiency of Lv.Col2-GFP/fLuc and Lv.CMV-GFP/fluc vectors, monolayer cultures were trypsinized, washed in PBS and resuspended in PBS/0.05% bovine serum albumin at 5 × 10⁶cells per ml. GFP fluorescence was analyzed using an LSRII flow cytometer (BD BioSciences, San Jose, CA, USA). For sorting of GFP+/tdtomato+ and GFP-/tdtomato+ subpopulations in differentiating BM-MSC aggregates by fluorescent activated cell sorting (FACS), Lv.CMV-tom/Col2-GFP transduced aggregates were pooled (n=7–8) and digested with a mixture of 0.1% collagenase and Tryple Select Enzyme (Thermo Fisher Scientific) for 45 minutes at 37°C. Recovered cells were resuspended in sterile PBS/0.05% BSA at 5 × 10⁶ cells per ml, and stained using a LIVE/DEAD Fixable Blue Cell Dead Stain Kit (Molecular Probes™) prior to sorting, according to the manufacturer’s protocol. Cell sorting was performed on a BDFACSAria II Cell Sorter (Beckton Dickinson) by first selecting for viable cells emitting low fluorescence cell under UV excitation, prior to gating under green and red channels to separate GFP+/tdtomato+ and GFP-/tdtomato+ populations. Collected cells were immediately transferred to TRIzol (Invitrogen) after sorting for RNA extraction.

Luciferase assay

Firefly

Assay of Lv.Col2-GFP/fLuc and Lv.CMV-GFP/fluc transduced monolayer cultures was performed on a Zylux tube luminometer using the Luciferase Assay System (Promega).

Metridia

Secreted luciferease from conditioned media supernatants of Lv.Col2mLuc/EF1-GFP- and Lv.CMVmLuc/EF1-GFP transduced cultures were assayed using the Ready-To-Glow™ Luciferase Reporter Assay (Takara Bio).

RT-PCR

To measure relative mRNA levels, total RNA was extracted from either pooled aggregates (n=5) and hydrogels (n=2) after 21 d of chondrogenic induction, or sorted cell populations, by phenol:chloroform extraction using TRizol reagent and isopropanol precipitation. cDNA was synthesized from 1 μg of total RNA using the High Capacity RNA-to-cDNA kit (Applied Biosystems) and random hexamer primers, according to the manufacturer’s instructions. Real-Time PCR amplification of diluted cDNA products was performed on an Eppendorf Realplex Mastercycler using PerFecta SYBR Green MasterMix (QuantaBio, Beverly, MA) and primers designed from equine gene sequences (Table 2). Relative gene expression was determined from C_t values by normalizing to the housekeeping gene, 18S, using the 2^−ΔΔCt method (Livak and Schmittgen, 2001).

Table 2.

Primers sequences for RT-qPCR

Gene	Sequence	Species
COL2A1	F: 5′-CTGGTCT CAAAGGCGAACAAG-3′	Equine
	R: 5′-GCACCTCTTTTGCCTTCTTCAC-3′
SOX9	F: 5′-CAGGTGCTCAAGGGCTACGA-3′
	R: 5′-GACGTGAGGCTTGTTCTTGCT-3′
RPLPO	F: 5′-CTGATTACACCTTCCCACTTGCT-3′
	R: 5′-AGCCACAAATGCAGATGGATCA-3′
18s	F: 5′-CGGCTTTGGTGACTCTAGATAACC-3′
	R: 5′-CCATGGTAGGCACAGCGACTA-3′
GFP	F: 5′-GTGGACAGCCACATGCACTT-3′	Copepod
	R: 5′-GTGCTGGTACTCCACGATGC-3′

Confocal imaging

Single Photon

Cellular aggregates and PEG hydrogels were fixed in 4% paraformaldehyde for 30 minutes at room temperature, washed 3 times in PBS and mounted in Vectashield (Vector labs, USA). Samples were imaged using a Leica TCS-SP5 (Leica, Wetzlar, Germany) equipped with a motorized inverted DM IRE2 microscope (Leica). Serial optical sections (1024 × 1024 pixels) were taken at 4 μm intervals using a 20× objective. For aggregates, specimens were imaged using a previously established protocol for intact tissues (Martin-Pena et al., 2006), covering a total volume of 150 × 150 × 150 μm. For PEG-HA hydrogels, the same volume was scanned in multiple locations within the 3D construct, and regions of interest were randomly selected within each location. Individual channels were set for imaging GFP, and tdTomato fluorescence signals. For each channel, confocal settings were maintained throughout the samples to allow comparison.

Multi-photon

For, full depth, spatial assessment of chondrogenic cells at high resolution, day 21 aggregates were also imaged using a Nikon A1RMPsi-STORM 4.0 multi-photon, super resolution imaging system (Nikon Instruments, Melville NY). Aggregates were cut in half and scanned using a 20× objective using a mosaic of 3 by 3 images (1024 × 1024 pixels each) with 10% overlap in each z-plane, with each z-plane interspaced by 4 μm and covering a total volume of 1500 × 1500 × 1500 μm.

Analysis

Images were processed using ImageJ software (NIH). Chondrogenic cell distribution in aggregates was determined using maximal z-stack projections and the line profile tool to measure pixel intensity vs. depth in the red and green channels. Values were plotted as normalized Green:Red signal intensity vs. depth. Cell counting in PEG-HA hydrogels was performed using the ImageJ cell count plug-in to determine total cell number and mean fluorescence values in red and green channels.

Histology

Aggregates and hydrogels were fixed in 10% NBF for 30 min at room temperature, and embedded in 0.7% agarose for ease of handling. After dehydration in graded alcohols, the samples were paraffin embedded and sectioned to 5 μm thickness. Sections representing the center of each sample were stained using 1% toluidine blue (Sigma-Aldrich) for the detection of matrix proteoglycan.

Statistical analysis

Quantitative results were shown as representative measurements of a single donor, or mean ± standard deviation of biological replicates or multiple donors (n=3). Statistical significance was determined by two-tailed, unpaired t-test (Fig. 1 and Fig. 5), or two-way ANOVA and subsequent Bonferroni post-hoc tests (Fig. 4) using GraphPad Prism 5.0. For all tests, p < 0.05 was considered significant.

Fig. 1. A modified Col2 promoter exhibits chondrocyte-specific activity in primary equine cells.

(a) Green fluorescence images of equine monolayer cultures infected with Lv.Col2-GFP/fluc (Col2) or Lv.CMV-GFP/fluc (CMV). Numbers indicate percentage of GFP+ cells as measured by flow cytometry. Scale bar: 250 μm. (b) Histogram flow cytometry plot showing the proportion of GFP+ cells after Lv.Col2-GFP/fluc transduction (grey: naïve cell control). (c) Relative luciferase activity among same cell populations. (d) Scatter flow cytometry plots showing cell-to-cell variations in GFP fluorescence in Lv-infected chondrocytes. All analyses were performed 72 h after viral transduction. * = p < 0.05 vs Syn and MSC.

Results

Chondrocyte specific activity of a modified Col2 promoter in primary equine cells

To investigate activity of the Col2 promoter in equine cells, a bicistronic lentiviral vector (Lv.Col2-GFP/fluc) was generated, containing the promoter driving co-expression of copepod GFP and firefly luciferase, that has been previously shown to respond to chondrogenic induction of human MSCs (Porter et al., 2012). Monolayer cultures of equine MSCs, synovial fibroblasts and chondrocytes were infected and Col2 reporter activity was compared to a control lentiviral construct where GFP is driven by a constitutive CMV promoter (Lv.CMV-GFP/fluc) (Fig. 1). As expected, CMV-mediated expression of GFP was high in all cultures; analysis of fluorescence by flow cytometry showed ~68–100% of the cells were positive (Fig. 1a,b). In contrast, Col2-mediated GFP expression was present in 72% of primary chondrocytes and ~18% of synoviocytes, but absent in undifferentiated MSCs (Fig. 1a,b). Assay of firefly luciferase in cell lysates revealed the same trend, with chondrocytes exhibiting significantly higher activity (>10-fold) than synoviocytes and MSCs (*p < 0.001; Fig. 1c). Cell-to-cell variations within the Col2-GFP+ population were also detected by flow cytometry (scatter plot; Fig. 1d). While Lv.CMV-GFP/fluc transduced chondrocytes were highly positive for GFP with relatively small variation in fluorescence intensity (middle panel; Fig. 1d), Lv.Col2-GFP/fluc transduced cultures contained both a negative population and GFP fluorescence that varied over 2 orders of magnitude (right panel; Fig. 1d). Given the relative uniformity of transduction indicated by Lv.CMV-GFP/fluc, it was concluded that heterogeneity within the Col2-GFP+ population was attributable to cells with varying chondrogenic activities. Based on these findings, Col2-driven reporter activity was further explored as a sensitive readout for chondrogenesis in 3D culture of primary equine cell populations.

Dual fluorescence reporting reveals heterogeneity within differentiating MSC aggregates

To assess the degree of uniformity of chondrogenesis and visualize temporal, cell-to-cell variations in chondrogenic activity within an MSC population, a dual-fluorescence reporter, Lv.CMV-tom/Col2-GFP was generated. This reporter vector contained independent expression cassettes under separate regulation of CMV (td-tomato) and modified Col2 (GFP) promoters, enabling the identification of chondrogenic and non-chondrogenic cells within differentiating cell populations. BM-MSCs were transduced with the vector, seeded in aggregate cultures, and analyzed by confocal microscopy for red and green fluorescence following TGF-β3 stimulation at days 7, 14, and 21 (Fig. 2a, upper panels). Z-stack projections of intact aggregates by single photon microscopy revealed abundant tomato (red) fluorescence at all time points indicating high expression of the constitutive CMV-driven reporter. Col2-promoter activation, indicated by either green or a merging of green and red fluorescence (orange), was evident in both unstimulated and day-7-stimulated cultures and identified isolated chondrogenic cells sparsely populated throughout the aggregate. At days 14 and 21, GFP expression markedly increased with culture duration, but distinct regions of red fluorescence were still visible indicating the persistence of non-chondrogenic cells. Supporting these findings, toluidine blue metachromasia increased with culture duration in response to TGF-β3 treatment (Fig. 2a, lower panels), reflecting ECM accumulation. However, even at 3 weeks, staining was negligible in some areas consistent with little or no cellular differentiation. Together these observations reveal striking variations of cellular-level chondrogenic activity within an MSC population in 3D aggregate culture.

Fig. 2. Identification of chondrogenic cells in differentiating aggregate cultures of MSCs.

(a) Upper panels: red/green fluorescence images of intact aggregates of BM-MSCs, from a 9 year old donor, modified with a dual fluorescence lentiviral reporter, Lv.CMV-tom/Col2-GFP. Lower panels: toluidine blue staining of aggregate sections at corresponding time points. (b) Representative flow cytometry contour plot showing the frequency of fluorescent cell populations isolated at day 14. (c) Relative mRNA levels in sorted cell populations (from b) in pooled aggregates of 3 separate equine donors, aged 2 (#1), 9 (#2) and 12 (#3) years old. Scale bar: 500 μm.

To confirm enhanced chondrogenic activity of GFP expressing cells in differentiating aggregates, cells were isolated by enzymatic digestion after 14 d in culture, separated into tdtomato+/GFP- and tdtomato+/GFP+ cell populations by FACS, and analyzed for expression of chondrogenic mRNAs (contour plot, Fig. 2b). Figure 2c shows expression levels in pooled aggregates of 3 separate MSC donors, indicating upregulation of SOX9 (1.3–2.4-fold) and COL2A1 (3.3–9.2-fold) in the tdTomato+/GFP+ fraction relative to tdTomato+/GFP- fraction. GFP mRNA levels, included as a cell sort control were also elevated (~3-fold), while levels of the non-chondrogenic housekeeping gene, ribosomal protein lateral stalk subunit P0 (RPLP0) did not change between the cell populations.

Spatial variations in chondrogenic activity were further assessed by high resolution (2-photon) imaging of 21 d aggregates of TGF-β3-treated BM-MSCs (Fig. 3). Aggregates were halved and imaged through the center to obtain red (tomato) and green (GFP) image stacks that were analyzed separately to determine fluorescence signal vs depth (Fig. 3a). Line plot analysis showing fluorescence intensity profiles (along the x axis) was performed at varying y-plane depths (plots i-v, Fig. 3b). Fluorescence ratios >2 were marked green denoting high chondrogenic activity, while ratios <2 were marked red, denoting lower/absent activity. Analysis revealed that highly chondrogenic cells were non-uniformly distributed, with the majority of chondrogenic peaks located (~50–400 μm) beneath the aggregate surface (in both x and y planes). This was most evident in plots i and v (y-plane depth ≤ 400 μm), and plots ii and iv (x-plane depth 0–250 μm). Areas of low chondrogenic activity (red) were also concentrated in the deepest regions of the aggregate, represented in plots ii-iv, ~300–900 μm, although some chondrogenic cells were present. These regional variations likely reflected the combined effects of oxygen, nutrient and TGF-β3 gradients on chondrogenic differentiation within the 3D microenvironment.

Fig. 3. Regional heterogeneity of chondrogenic cells in differentiating MSC aggregates.

(a) Merged Z-stack projection showing red/green fluorescence, captured by multi-photon imaging of MSC aggregates labeled with Lv.CMV-tom/Col2-GFP and treated with TGF-β3 until day 21. Aggregates were cut in half and imaged through the center (z-plane depth = 1500 μm). (b) Line plots, i-v, showing fluorescent pixel intensity vs depth analyzed using ImageJ software. Fluorescence ratios >2 are marked green denoting high relative chondrogenic activity, while ratios <2 are marked red denoting low/absent activity. Analyses are representative of 4 BM-MSC aggregates.

A secretable luciferase reporter reveals marked variations in chondrogenic potential among MSC populations.

For luminescence-based determination of chondrocyte differentiation, a lentiviral construct containing the same modified Col2 promoter driving expression of secreted Metridia luciferase (mLUC), and a constitutive EF1α promoter driving expression of GFP for normalisation of transduction efficiency were generated. Through luciferase assay of the conditioned media at various time points, it was assumed that Col2-Luc reporter labeled cells would enable longitudinal, non-invasive assessments of differentiation in 3D culture systems that are predictive of overall chondrogenesis, determined by conventional endpoint assays.

Using cells obtained from the same equine donor chondrogenic activities were compared among MSCs derived from bone marrow (BM), synovium (S) and adipose (A) tissues. MSCs were transduced with Lv.Col2-mLuc/EF1-GFP and grown in aggregate cultures and photopolymerisable PEG-HA hydrogels (Fig. 4a), which are transplantable scaffolds compatible with cell-based cartilage repair (Sharma et al., 2007). To facilitate direct comparisons among cell types, activity was normalized to Lv.CMV-mLuc/EF1-GFP modified aggregates cultured in parallel (Fig. 4a). Articular chondrocytes, also harvested from the same donor, were included as a reference for chondrogenic activity. Among MSC types, the luciferase profiles revealed a common trend in both aggregate and hydrogel cultures, with highest chondrogenic activity in BM-MSCs, followed by S-MSCs and A-MSCs at all time points (Fig. 4b,c). Differences among cell types were significant at day 21 in aggregate culture, with A-MSCs exhibiting lower activity compared to other cells (p < 0.05, Fig. 4b), and at day 14 and 21 in PEG-HA cultures, where BM-MSCs had higher activity than all other cell types (p < 0.05, Fig. 4c). Somewhat surprisingly, activity in BM-MSCs was also higher than chondrocytes in aggregate cultures at day 14 and 21, and at all time points in PEG-HA cultures. The luciferase profiles were mostly consistent with mRNA levels for chondrogenic markers SOX9 and COL2A1 at endpoint (day 21), indicating the same trend among MSC types (Fig. 4d,e). Similarly, toluidine blue staining intensities of cultures harvested at endpoint reflected superior chondrogenesis of BM-MSCs followed by S-MSCs and A-MSCs in both culture systems, in agreement with the luciferase data (Fig. 4f,g). The endpoint analysis also revealed similar chondrogenesis of BM-MSCs relative to chondrocytes in aggregates, also consistent with the luciferase reporter profile (Fig. 4b,d,f). In contrast, chondrocytes in PEG-HA hydrogels had roughly equivalent matrix staining and > 2-fold higher levels of COL2A1 compared to BM-MSCs, despite exhibiting lower luciferase activity. This discordance could potentially reflect differences in innate stabilities between chondrocyte and MSC phenotypes that are not detectable at the level of Col2 reporter activity.

Fig. 4. Luciferase-based, real-time monitoring of chondrogenesis in equine cell populations cultured in aggregates and PEG-HA hydrogels.

(a) Strategy for cell labeling, culture and analysis. Cells were harvested from a single equine donor (female, 2 years). (b,c) Secreted Metridia luciferase levels in the conditioned medium of aggregate and hydrogel cultures shown as a ratio of Col2-mLuc/CMV-mLuc. Prestimulated controls (Prestim) represent luciferase levels in cultures without TGF-β3 treatment (day 3–4). (d,e) Relative mRNA levels of pooled aggregates (5) and hydrogels (2) for each cell type at day 21 endpoint. (f,g) Toluidine blue staining of Lv-transduced aggregates and PEG-HA cultures at day 21. * = p < 0.05. Scale bars: 500 μm.

Dual fluorescence reporting enables long-term monitoring of chondrogenic cell fates in PEG-HA implants

To investigate the utility of using fluorescent reporter constructs to determine chondrogenic cell fates in vivo, a subcutaneous implantation model was used to track the phenotype of chondrocytes seeded in PEG-HA scaffolds. Chondrocytes were labeled with Lv.CMV-tom/Col2-GFP, expanded in monolayer and seeded into PEG-HA hydrogels encapsulated with 100 ng TGF-β3. After 24 h, the cell-loaded hydrogels were either harvested for analysis (baseline control group), cultured in vitro with continuous exposure to chondrogenic induction medium (in vitro group), or implanted subcutaneously in athymic mice (implant group) (Fig. 5a). Fluorescence image analysis of recovered hydrogels revealed that over 90% of the total cell population was positive for GFP following in vitro differentiation for 6 weeks, increasing from 10% in the baseline group (p < 0.005; Fig. 5b upper panels and c). This likely reflected re-differentiation of the chondrocyte phenotype after monolayer expansion via prolonged culture in a 3D microenvironment with continuous exposure to TGF-β3. After 6 weeks in vivo, ~45% of the total cell population was GFP+, indicating that initial encapsulation with TGF-β3 can partially restore the chondrocyte phenotype, but to a lesser degree than under controlled in vitro conditions (Fig. 5c, p < 0.05). In support of these findings, toluidine blue metachromasia indicated greater cellular differentiation under in vitro conditions with intense staining in both peri- and extracellular regions (Fig. 5b lower panels). In contrast, recovered implants had less staining, which appeared to be restricted to pericellular regions, while in the baseline group, no significant metachromasia was present. Together these findings indicate that lentiviral-based fluorescent reporters can be used to quantify chondrogenic cell populations in 3D scaffolds and evaluate cellular-level chondrogenesis, providing a platform to compare chondrogenic potencies among chondrocytes and progenitor cell populations in established in vitro and in vivo models.

Fig. 5. Comparison of chondrogenic cell fates in PEG-HA hydrogels in vitro and in vivo using the dual fluorescence reporter, Lv.CMV-tom/Col2-GFP.

(a) Strategy for cell labeling, encapsulation and in vitro culture/implantation. (b) Upper panels: red/green fluorescence images of Lv.CMV-tom/Col2-GFP transduced chondrocytes in PEG-HA hydrogels. Lower panels: toluidine blue staining of hydrogel sections at corresponding time points. (c) Cell counts performed by separate analyses of red and green channels from captured confocal images using ImageJ to determine the relative proportion of chondrogenic (Col2-GFP+) cells within the total cell population (CMV-tdtomato+). Values represent mean of 3 cultures/implants for each group. * = p < 0.05; ** = p < 0.005. Scale bar: 500 μm.

Discussion

The presented results demonstrated the use of lentiviral reporter vectors as tools for quantification and spatial assessment of chondrogenesis in primary adult equine cell populations in 3D culture systems. Lentiviral vectors encoding either GFP or luciferase reporters under conditional activation of a modified Col2 promoter exhibited stable, chondrocyte-specific reporter activity in primary equine chondrocytes, and MSCs undergoing TGF-β3-induced chondrogenesis, correlating with histology and mRNA expression levels of COL2A1 and SOX9. Using dual reporter vectors that provided readouts of chondrogenic (Col2) and constitutive (CMV/EF1α) activity it was possible to (i) discriminate chondrogenic vs. non-differentiated cells in aggregate and PEG hydrogel culture systems, (ii) directly compare chondrogenic activities among primary cell populations and (iii) monitor chondrogenic cell fates in hydrogels following subcutaneous implantation in vivo.

For in vitro monitoring of chondrogenesis, the use of recombinant lentiviral reporter vectors offers several advantages for screening chondrogenic activity of primary cells. They provide stable transgene expression, enabling long-term monitoring of cell fates, and high transduction efficiency (Zhang et al., 2002), reducing the need for additional ex vivo manipulations, such as cell sorting or antibiotic selection, that can potentially affect differentiation, (Cohen et al., 2006; Connell et al., 2016). Transducing titers can be generated with large insert sizes (up to ~18 kb), (Kumar et al., 2001), which allows for the inclusion of multiple reporter genes within single or multiple expression cassettes, enabling multiple cellular readouts from a single infection.

Using a dual fluorescence reporter vector, Lv.CMVtom/Col2GFP, the number of chondrogenic cells (tdTomato+/GFP+) increased in parallel with the level of toluidine blue staining over time in aggregate cultures of MSCs. Notably, despite continued exposure to TGF-β3, large numbers of non-chondrogenic, tdTomato+/GFP- cells were visible at endpoint (day 21), indicating that a significant population of MSCs exhibited poor or no differentiation within this culture system. This was consistent with flow cytometric analysis after day 14, where approximately 50% of the total cell population were GFP+/tdTomato+. Although the persistence of non-differentiated cells within pellet cultures of MSCs was first reported in 1998 (Yoo et al., 1998), the proportion of chondrogenic cells and degree of uniformity within cultures are rarely considered. A non-uniform distribution of differentiating cells was found within TGF-β3-stimulated cultures, with most of the highly chondrogenic cells located proximal (~400 μm) to, but not along, the aggregate edge. This may reflect an optimal oxygen level just beneath the pellet surface, as hypoxia (i.e. 2–5% oxygen) increases bone marrow MSC chondrogenesis in pellet culture (Boyette et al., 2014; Anderson et al., 2016). Anderson et al. (2016). observe that the MSC response to hypoxia is enhanced in relatively chondrogenic donor preparations. In the present study, fewer cells with high chondrogenic activity were detected within the central regions, which may reflect impaired cellular function resulting from diffusion limitations within the densely packed aggregates. Indeed, tumor spheroids, >500 μm, are frequently characterized by proliferation gradients, hypoxic regions and reduced cell viability leading to the formation of necrotic cores (Hirschhaeuser et al., 2010; Laurent et al., 2013; Zanoni et al., 2016). Thus, size and geometric constraints associated with aggregate (sphere)-based culture may mask the true chondrogenic potential of a cell population or chondrogenic factor and should be taken into consideration when performing a screening assay.

For direct, real-time comparisons of chondrogenic activity among cell populations and 3D culture systems, a Col2-driven secretable luciferase reporter was used to reveal greater TGF-β3-mediated chondrogenesis of BM-MSCs relative to S-MSCs and A-MSCs in both aggregate and PEG-hydrogel culture models. Moreover, relative luciferase activity between the cell types remained constant over the differentiation time course and correlated with histologic and mRNA expression levels at endpoint. Thus, secretable Col2-Luc reporter activity may serve as a potency marker that can predict chondrogenic differentiation of MSCs, providing the opportunity for modification of culturing protocols in real-time, to facilitate the generation of stable, functional cartilage by a particular cell type. This would be of particular benefit to tissue engineering manufacturing processes which are currently constrained by long-term culturing protocols that rely on endpoint assessments of the final product (Roh et al., 2016).

While interpretation of the study’s findings is limited to TGF-β3-driven chondrogenesis from a single equine donor, the superiority of BM-MSCs over A-MSCs has also been reported by others (Vidal et al., 2008; Danisovic et al., 2009; Kohli et al., 2015). However, contrary to our observations in equine cells, comparative studies have also demonstrated enhanced chondrogenesis of S-MSCs relative to BM-MSCs in humans (Shirasawa et al., 2006; Yoshimura et al., 2007) and rats (Sakaguchi et al., 2005) subjects. These differences may be attributable to variations of MSC phenotype among species, or the use of different growth factor combinations for chondro-induction between studies. Intriguingly, among two very different 3D culture formats – high density aggregates or PEG-HA suspension culture – the relative chondrogenic potentials among BM-, S- and A-MSCs did not change, suggesting that cell-cell adhesion and cell-HA interactions were not critical factors in determining the overall chondrogenic response of the cells to TGF-β3 induction, in the absence of mechanical co-stimulation. However, for optimal chondrogenesis of each MSC type in both culture systems, it will necessary to perform a broader investigation that incorporates other TGF-β superfamily members, such as BMP-6 (Estes et al., 2006; Hennig et al., 2007) and BMP-7 (Kurth et al., 2007; Miyamoto et al., 2007; Shintani and Hunziker, 2007), which have been shown to enhance chondogenic differentiation of A-MSCs and S-MSCs, respectively.

With the long-term goal of application in cartilage defects, the feasibility of using fluorescent reporters to monitor cellular-level chondrogenic cell fates in vivo by tracking Lv.CMV-tom/Col2-GFP labeled chondrocytes in subcutaneously implanted PEG-HA hydrogels was tested. After 6 weeks in vivo, almost 50% of the cell population was chondrogenic (GFP+), significantly more than in monolayer expanded cultures, but less as compared to PEG-HA hydrogels cultured in vitro. This was not surprising as in vivo implants received only a single 100 ng bolus of TGF-β3 during encapsulation, while TGF-β3 stimulation in vitro was continuous, with repeated growth factor addition every 4 d. Indeed, a single TGF-β3 dose was also not sufficient for reproducible equine MSC differentiation in the same study (data not shown) reflecting the need for sustained growth factor delivery systems to promote cellular differentiation or long-term maintenance of chondrocyte phenotypes in vivo (Lam et al., 2015; Rey-Rico and Cucchiarini, 2017).

Conclusions

Variations associated with MSC isolation and culture, as well as poor characterization of primary cell populations, has greatly impeded the clinical translation of MSCs in cartilage-repair applications (Tanavde et al. 2015). In large animal models, such as the horse – an essential preclinical model for orthopaedic conditions (Hurtig et al., 2011) – this problem is exacerbated by the lack of equine specific antibodies, as well as the relatively low number of equine specific gene and protein markers. In the present study, using equine cells as a test system, the utility of using transgenic lentiviral-based reporter vectors as tools for objective assessment of chondrogenesis in preclinical cell-based repair studies was demonstrated. The implementation of these reporters across laboratories will advance the development of standardized protocols and help promoting consensus among investigators with regard to cell types, scaffolds and chondro-induction methods for clinical translation of cell-based therapies for cartilage repair.