Rapid and Sensitive Lentivirus Vector-Based Conditional Gene Expression Assay to Monitor and Quantify Cell Fusion Activity

Cell-to-cell fusion is involved in multiple fundamental biological processes. Prominent examples include osteoclast and giant cell formation, fertilization and skeletal myogenesis which involve macrophage, sperm-egg and myoblast fusion, respectively. Indeed, the importance of cell fusion is underscored by the wide range of homeostatic as well as pathologic processes in which it plays a key role. Therefore, rapid and sensitive systems to trace and measure cell fusion events in various experimental systems are in demand. Here, we introduce a bipartite cell fusion monitoring system based on a genetic switch responsive to the site-specific recombinase FLP. To allow flexible deployment in both dividing as well as non-dividing cell populations, inducer and reporter modules were incorporated in lentivirus vector particles. Moreover, the recombinase-inducible transcription units were designed in such a way as to minimize basal activity and chromosomal position effects in the “off” and “on” states, respectively. The lentivirus vector-based conditional gene expression assay was validated in primary human mesenchymal stem cells and in a differentiation model based on muscle progenitor cells from a Duchenne muscular dystrophy patient using reporter genes compatible with live- and single-cell imaging and with whole population measurements. Using the skeletal muscle cell differentiation model, we showed that the new assay displays low background activity, a 2-log dynamic range, high sensitivity and is amenable to the investigation of cell fusion kinetics. The utility of the bipartite cell fusion monitoring system was underscored by a study on the impact of drug- and RNAi-mediated p38 MAPK inhibition on human myocyte differentiation. Finally, building on the capacity of lentivirus vectors to readily generate transgenic animals the present FLP-inducible system should be adaptable, alone or together with Cre/loxP-based assays, to cell lineage tracing and conditional gene manipulation studies in vivo.


Introduction
Diverse biological phenomena involve fusion between either the same or different cell types (i.e. homotypic and heterotypic cell fusion, respectively). Notable examples include osteoclast and giant cell formation (macrophage fusion), fertilization (sperm-egg fusion), syncytiotrophoblast formation (cytotrophoblast fusion) and skeletal muscle development and regeneration (myoblast fusion). In addition, recent studies indicate that in parallel to these canonical cell fusion processes there are also those involving noncanonical or atypical cell fusion events. For example, it has been advanced that cancers may progress through heterotypic cell fusion between tumor cells and cells with tissue-infiltrating capacity such as macrophages [1,2]. Another example is the occurrence of low-level heterotypic cell fusion between stem or progenitor cells from one tissue and parenchymal cells from another tissue especially under conditions disrupting the homeostasis of the recipient organ. The latter phenomenon stirred up the stem cell field by providing an alternative explanation to some of the early observations on stem cell plasticity whereby cells, in some cases, purportedly transdifferentiated, i.e. acquired a phenotype different from the cells of their tissue of origin [3]. In addition, cellto-cell fusion is also a feature evolved by syncytium-forming viruses (e.g. human immunodeficiency virus [HIV] and measles) to facilitate spreading through their hosts [4,5]. Of note, syncytium formation ''as-a-means-to-spread'' has been proposed as a process that could be exploited to increase the potency of oncolytic virusbased vectors [6][7][8]. Given its relevance in biology, flexible and sensitive assays to detect and measure cell-to-cell fusion are expected to have a broad applicability.
Cell fusion activity is often expressed in terms of the ''fusion index''. The ''fusion index'' is obtained by estimating the frequency of nuclei associated with syncytia in relation to the total number of nuclei. Alternatively, putative fusion partners can be marked with different fluorescent proteins or dyes that produce a distinctive combinatorial spectral signature after fusion between the differently labeled cells takes place [9]. Also, species-specific antibodies as well as species-or gender-specific chromosome probes can be applied to detect fusion events between cells of different organisms or genders. These methods, however, are based on microscopic inspection and cell counting by different researchers and, as a consequence, they are fastidious, time-consuming and error-prone. This makes them suffer from a lack of high-throughput capabilities, accuracy and reproducibility. Additionally, ''fusion index'' estimations cannot discriminate bona fide cell fusion events from aborted cytokinesis.
To overcome these limitations more quantitative and reproducible cell fusion assays have been developed (for a review see [10]). Because the most direct method of demonstrating cell fusion is to ascertain mixing of cellular constituents of the interacting partners, these assays have in common the measurement of a new signal output only after such mixing occurs. The majority of quantitative cell fusion assays are based either on biochemical complementation or on transcription activation principles. The former rely on assembly of tetrameric complexes consisting of two Escherichia coli b-galactosidase subunits that per se are nonfunctional due to the deletion of crucial protein domains [11], whereas the latter depend on bacteriophage T7 RNA polymeraseor bacteriophage Cre recombinase-responsive reporter genes [10]. The Cre recombinase-based systems possess a characteristic arrangement of genetic elements (see below) and allow the detection of sporadic cell fusion events in vivo [12]. Of note, normally low Cre concentrations suffice to induce target DNA rearrangement with subsequent signal amplification being relayed via reporter gene transcriptional activity. In contrast, albeit potentially faster in reporting cell fusion events, the b-galactosidase complementation method require the component protein subunits to be present in high and, ideally, equimolar amounts to facilitate the assembly of catalytically active enzyme complexes.
This report describes the development and testing of a novel two-component transcription activation-dependent cell fusion assay based on the introduction into separate test cell populations of recombinase-encoding and recombinase-responsive transcription units. The recombinase-encoding expression unit directs the synthesis of FLPe, an enhanced version of the Saccharomyces cerevisiae site-specific recombinase FLP [13], whereas the recombinaseresponsive transcription module contains a direct repeat of FLP recognition target (FRT) sites embedded within a unique arrangement of transcription elements that allows the formation of functional transgenes solely upon the generation of circular episomes (see below). This FLPe-responsive molecular switch, which differs from those usually applied in recombinase-activatable transgenes, was chosen to minimize reporter gene basal activity in the ''off'' state and to avoid chromosomal position effects, including transcriptional interference, on reporter gene expression in the ''on'' state. Importantly, to allow rapid and flexible deployment in dividing and non-dividing cells as well as in cells displaying a limited replicative life span (e.g. most primary cells), both components of the assay system were incorporated into lentivirus vector particles. This new conditional gene expression system was validated in an ex vivo differentiation model based on myoblasts from a Duchenne muscular dystrophy (DMD) patient [14] using reporter genes amenable to live-and single-cell imaging and to whole-population measurements. Finally, by deploying the bipartite cell fusion monitoring system together with short-hairpin RNAs (shRNAs) and the specific p38a/b inhibitor SB 203580 we showed the involvement of the p38 mitogen-activated protein kinase (MAPK) pathway in human myotube formation.

Experimental rationale and design
Chromosomal position effects imparted on transgenic DNA by flanking genomic sequences potentially complicate reliable quantification of recombinase-dependent transgene expression. This pitfall may be overcome by using clonal derivatives of the test cells or by targeted insertion of the recombinase-activatable transcription unit into specific chromosomal loci. However, both these approaches rely on the establishment and screening of single-cell clones which makes them laborious and time-consuming. Moreover, recombinant DNA transfer as well as selection and expansion of clones are not easily applicable to cells with a limited replicative life span like most primary mammalian cells. Furthermore, working with single-cell clones may yield biased results due to the possible existence of phenotypic heterogeneity in the test cell population.
Besides their prominent role in gene therapy research [15], lentivirus vectors have become widely used gene transfer tools due to their straightforward production and capacity to stably transduce both dividing and non-dividing cells with high efficiency. The versatility of lentivirus vector technology is further underscored by its successful deployment in establishing transgenic rodent lines [16,17]. Lentivirus vectors achieve permanent genetic modification of transduced cells by integrating DNA copies of their genomes into the host cell's chromosomal DNA. Integration events, however, are not strictly random since the lentiviral pre-integration complexes preferentially target actively transcribed open reading frames (ORFs) [18,19]. As a result, individual DNA integrants become subjected to different chromatin environments and associated regulatory cis-acting elements leading to positional effects on transgene expression (Fig. 1A). In the context of a recombinasedependent conditional gene expression system with a conventional organization of genetic elements these chromatin positional effects may include in cis (depicted) and in trans transcriptional interference on reporter gene expression (Fig. 1A). To minimize these effects we designed a lentivirus vector encoding a recombinase-dependent gene switch module with an arrangement of its genetic elements as illustrated in Fig. 1B. Of note, this lentivirus vector-based conditional gene expression system contains the unidirectional murine metallothionein gene polyadenylation signal (pA) previously shown to block cryptic transcription of a promoterless ORF located downstream of it [20]. This element, as well as the reporter ORF and the rabbit b-globin pA associated with it, are oriented in the opposite direction to that of the lentivirus vector 59 long terminal repeat (LTR)-driven transcription to prevent interference with the generation of full-length vector genomes during particle production. FLPe-mediated site-specific DNA recombination and ensuing reporter gene activation is concomitant with a transition of the expression unit from a chromosomal to an episomal context. This topological change should eliminate possible negative or positive effects of chromosomal cis-acting regulatory elements on reporter gene expression (Fig. 1B). Putative DNA integrant-specific reporter activity in the ''on'' state due to epigenetic modifications acquired by the transgenic sequences following their host cell genome insertion is expected to be avoided neither by the traditional nor by the new gene switch design.
Generation and testing of lentivirus vectors bearing recombinase-responsive reporter genes whose activation is dependent on episome formation We started by testing whether lentivirus vector particles encoding a FLP-dependent gene switch with the structural organization depicted in Fig. 1B could be generated and, following indicator cell transduction, express the reporter gene in a recombinase-dependent manner. To this end, LV.GS.DsRed particles produced by using the lentivirus vector shuttle plasmid pLV.pA+.GS.DsRed (Fig. 2) were added onto myoblasts derived from a satellite cell of a DMD patient [14] at 0, 3, 9 and 15 transducing units (TU)/cell. The pLV.pA+.GS.DsRed construct encodes an enhanced version of the red fluorescent protein (RFP) of Discosoma sp. designated DsRed.T4-N1 [21].
Next, these cells were either mock-infected or incubated with a single dose of the FLPe-encoding lentivirus vector LV.FLPe or of the FLPe-encoding high-capacity adenovirus vector HD.FLPe.F50 [22] (hereinafter referred to as HD.FLPe). Three days post-infection direct fluorescence microscopy revealed that synthesis of RFP was dependent on the presence of FLPe and that more red fluorescent myoblasts were observed in cultures initially exposed to higher doses of LV.GS.DsRed (Fig. 3A). In another experiment, parallel cultures of DMD myoblasts transduced with LV.GS.DsRed at 0, 3, 9 and 15 TU/cell were exposed to increasing amounts of HD.FLPe (i.e. 0, 10, 20, 30 and 40 gene switch-activating units [GSAU]/cell) or LV.FLPe (i.e. 0, 10, 20, 30 TU/cell). Flow cytometric analyses performed at 3 days after infection confirmed FLPe-mediated transgene activation and an LV.GS.DsRed dose-dependent increase in the frequency of RFP-positive cells (Fig. 3B). Dot plots corresponding to the flow cytometric analyses of DMD myoblasts exposed to or not exposed to HD.FLPe, are depicted in Fig. 3C (right and left panel, respectively). Importantly, in the absence of FLPe, the frequency of RFP-positive cells in myoblast cultures transduced with increasing amounts of LV.GS.DsRed remained below 1% (range: 0.28-0.81%).
On the basis of the results described above, we constructed pLV.pA+.GS.Luc to allow quantification of cell fusion activity in whole cell populations. This lentivirus vector shuttle plasmid contains the luciferase gene from Photinus pyralis (Fig. 2) whose product permits the detection of gene expression within a very wide linear range by using simple bioluminescence assays. These assays measure the visible light emitted during the enzymatic conversion of luciferin to oxyluciferin. Lentivirus vector particles were generated by using the shuttle plasmid pLV.pA+.GS.Luc. The functionality of the LV.GS.Luc particles generated with the aid of pLV.pA+.GS.Luc was evaluated by transducing primary human mesenchymal stem cells (hMSCs) [9] and human myoblasts [14] at 30 TU/cell. Mock-transduced hMSCs and mock-transduced human myoblasts served as negative controls. Subsequently, cultures of these two different indicator cell types were exposed to the FLPe-encoding adenovirus vector HD.FLPe at 3, 15 and 75 GSAU/cell [22]. Data shown in Figs. 3D and   Vector-Based Cell Fusion Assay PLoS ONE | www.plosone.org 3E clearly demonstrate that robust luciferase gene expression in both hMSCs (Fig. 3D) and human myoblasts (Fig. 3E) is dependent on the presence of FLPe. For each of the HD.FLPe doses applied, the fold increase in luciferase activity in comparison to cells transduced exclusively with LV.GS.Luc, was similar for both cell types (i.e. 50, 80 and 102 for hMSCs versus 68, 84 and 91 for myoblasts at HD.FLPe doses of 3, 15 and 75 GSAU/cell, respectively).
Taken together these data indicate that (i) the specific arrangement of genetic elements in the FLP-responsive reporter gene-containing shuttle plasmids is compatible with the synthesis and packaging of lentivirus vector genomic RNA and (ii) host cell promoter-driven reporter gene expression is contained as evident from the low basal expression levels measured following proviral vector DNA integration into the genome of the indicator cells.

Quantification of cell fusion in ex vivo cultures of human muscle progenitor cells
To investigate the suitability of the present lentivirus vectorbased inducible reporter gene system to detect and quantify cell fusion events we exploited an ex vivo cellular differentiation model based on human myoblasts. In this system, activation of the terminal myogenic differentiation program is simply accomplished by withdrawing mitogens from the culture medium. This triggers the homotypic fusion of the mononucleated muscle progenitor cells resulting in a time-dependent accumulation of nuclei in syncytial structures called myotubes.
FLPe-encoding human myoblasts were generated through transduction with the lentivirus vector LV.FLPe.PurR followed by selection of genetically modified cells using puromycin. Five days after the initiation of puromycin-mediated selection, an LV.FLPe. PurR dose-dependent increase in cell survival could be discerned (Fig. 4A). Repeated passaging of LV.FLPe.PurR-transduced human myoblasts in the presence of puromycin revealed no apparent differences in the growth rate between cell cultures exposed to the different amounts of FLPe-encoding vector tested (not shown). Thus, we selected myoblasts transduced with the highest LV.FLPe.PurR dose (hereinafter referred to as myoblasts FLPe ) to carry out the subsequent experiments. In parallel, reporter gene-modified test cells (hereinafter referred to as myoblasts GS.Luc ) were generated by transducing human myoblasts with LV.GS.Luc at a multiplicity of infection of 30 TU/cell.
To determine whether the lentivirus vector-based inducible reporter gene system displays (i) low basal activity in the presence of FLPe-positive cells and (ii) high fusion-dependent induction factors, we mixed myoblasts FLPe with myoblasts GS.Luc at different ratios (i.e. 90:10, 75:25, 50:50, 25:75 and 10:90) (Fig. 4B). Cultures consisting exclusively of myoblasts FLPe (i.e. 100:0) or of myoblasts GS.Luc (i.e. 0:100) served as negative controls (Fig. 4B). The extent of myogenic differentiation observed in each of the myoblast cultures after mitogen withdrawal was similar (not shown). Bioluminescence measurements immediately prior to as well as after the completion of the myoblast-myotube differentiation schedule clearly revealed a differentiation/fusion-dependent increase in luciferase activity. Indeed, reporter gene expression levels more than 2 orders of magnitude above those in nondifferentiated cells were consistently measured exclusively in cocultures of myoblasts FLPe and myoblasts GS.Luc (Fig. 4B).

Analysis of cell fusion kinetics in ex vivo cultures of human muscle progenitor cells
As previously mentioned, upon induction of myogenic differentiation a gradual accumulation of myotubes of growing size and with an increasing number of nuclei can be observed in cultures of myoblasts as a consequence of ongoing cell fusion. Conversely, parallel cultures kept under normal growth medium are mostly devoid of myotubes. We asked whether this myogenic differentiation phenomenon could be followed as a function of time by deploying the lentivirus vector-based conditional gene expression system presented above (Fig. 5). To this end, myoblasts FLPe were mixed with myoblasts GS.Luc at a 1:1 ratio using two different total amounts of cells (i.e. 10 5 and 2610 5 ). The resulting co-cultures were subsequently either kept in growth medium or were exposed to differentiation medium for different periods of time (Fig. 5A, upper and lower panels, respectively). Luminometric analyses of lysates from cells maintained in differentiation medium revealed a time-dependent increase in luciferase activity, which correlated with the time-dependent rise in the frequency and size of myotubes. In addition, consistent with the previous findings (Fig. 4B), parallel co-cultures of myoblasts FLPe and myoblasts GS.Luc maintained in regular growth medium did not show a time-dependent rise in luciferase expression despite the increase in cell numbers due to mitosis (Fig. 5B, solid bars in right panels). These data suggest that the newly developed lentivirus vector-based conditional gene expression system can be deployed not only to quantify muscle cell differentiation activity but should also be useful to investigate factors that positively or negatively affect the kinetics of this process.

Determining the sensitivity of the lentivirus vector-based inducible reporter gene assay to detect cell fusion in ex vivo cultures of differentiating human myoblasts
Earlier data showed that transgene expression in co-cultures of myoblasts GS.Luc and myoblasts FLPe increased with increasing fractions of myoblasts GS.Luc (Fig. 4B) suggesting that the number of gene switch modules available for FLP-mediated recombination was limiting. This is consistent with the idea that relatively few molecules of recombinases such as Cre or FLP suffice to bring about site-specific recombination and that, necessarily, the more luciferase copies are present in the bipartite gene expression system the higher the likelihood for reporter protein synthesis to occur.
Thus, to determine the minimum percentage of myoblasts FLPe that still leads to a measurable fusion-dependent signal output, we set up an experiment in which the frequency of myoblasts FLPe was titrated on a constant total amount of 2610 5 myoblasts. The fraction of myoblasts FLPe present in co-cultures of myoblasts FLPe and myoblasts GS.Luc was varied in a series by a factor of 3. Of note, no obvious differences in the frequency and size of myotubes were observed between the co-cultures containing different ratios of myoblasts FLPe and myoblasts GS.Luc (Fig. 6A). Chemiluminescence measurements performed after the regular differentiation period showed that a frequency of myoblasts FLPe as low as 0.04% could still give rise to a luciferase signal above that measured in cultures containing exclusively myoblasts GS.Luc (i.e. background [grey bar]) (Fig. 6B). These results bode well for the deployment of the current lentivirus vector-based conditional gene expression system in instances in which fusion events are rare. Even so, ways by which to optimize the current set up can be envisioned. These include (i) flanking of the gene switch module by insulator sequences to block host cell enhancer activity on reporter gene expression during the proviral vector stage, (ii) placing transgene expression under the control of a promoter stronger than that of the human glyceraldehyde-3-phosphate dehydrogenase gene and (iii) replacing the current RFP-and Photinus pyralis luciferase-coding sequences by those of reporter proteins with higher specific activity like DsRed-Express2 [23] and Gaussia princeps luciferase [24]. Of note, the latter reporter protein in its secreted [24] or membraneanchored [25] form should permit adjustment to novel experimental read-outs with the added advantage of displaying a higher sensitivity than other commonly used luciferases. Finally, analysis of cell fusion kinetics may profit from the deployment of destabilized versions of luciferase or other reporter proteins.

Pharmacological and genetic interference with the p38 MAPK pathway and its impact on the human myoblast fusion process
Various studies, mostly in murine and rat cells, have shown that the p38 MAPK pathway plays a pivotal role in skeletal myogenesis through, amongst others, phosphorylation-dependent activation of myogenic transcription factors and its ability to induce chromatin remodeling of skeletal muscle-specific gene promoters. A recent study deploying neonatal myoblasts derived from mice in which the p38a, p38b, p38c and p38d genes were individually knocked out pinpointed p38a as the most crucial p38 MAPK in this muscle cell differentiation system [26]. In fact, murine muscle progenitor cells lacking p38b or p38d did not present any apparent defect in their fusion ability whereas those without p38c displayed a delayed fusion phenotype [26].
To determine whether the bipartite conditional gene expression system could be used to establish the involvement of the p38 MAPK pathway during the differentiation of myocytes from human adults, we deployed the p38a/b-specific inhibitor SB 203580 or shRNAs targeting p38a transcripts. Co-cultures consisting of myoblasts FLPe and myoblasts GS.Luc kept in regular differentiation medium or in differentiation medium supplemented with either SB 203580 (0.1, 0.5 and 2.5 mM) or vehicle only, revealed a clear drug-dependent reduction in myoblast fusion activity as determined by chemiluminescence measurements (Fig. 7A). Subsequently, by using lentivirus vector-mediated gene transfer and puromycin selection, we screened on the background of myoblasts and myoblasts GS.Luc cells, polyclonal lines encoding shRNAs designed for post-transcriptional down-regulation of p38a (4 shRNAs), eGFP (1 shRNA) or murine hypoxia inducible factor 1a (hif1a) (1 shRNA) expression. Screening of the relative amounts of p38a protein synthesized in parental myoblasts and myoblasts GS.Luc as well as in the various shRNAencoding polyclonal cell populations derived from them was carried out by Western blot analysis. Results depicted in Fig. 7B show that myoblast and myoblast GS.Luc lines encoding shRNAs sh.p38a.33 and sh.p38a.36 displayed consistent p38a knock-down. To measure the impact of p38a knock-down on human myocyte fusion, the experimental set-up outlined in Fig. 7C was followed. Cultures of parental myoblasts, myoblasts sh.hif1a , myoblasts sh.p38a.33 and myoblasts sh.p38a.36 were divided and transduced either with LV.FLPe or with LV.GS.Luc. After sub-culturing, the resulting FLPe-encoding cell populations were mixed at a 1:1 ratio with their respective GS.Luc-containing counterparts (i.e. myoblasts sh.p38a.36FLPe were  5 myoblasts and exposed for 5 days to differentiation medium. The cultures were composed exclusively of myoblasts GS.Luc (grey bar) or of myoblasts FLPe (white bar) or contained myoblasts GS.Luc (90-99.96%) together with low proportions of myoblasts FLPe (i.e. 0.04, 0.12, 0.37, 1.1, 3.3 and 10%). Cumulative data are expressed as means 6 standard error of the mean (n = 3). P values resulting from the comparison of relevant experimental groups were determined using an unpaired one-tailed Student's t-test. P values ,0.05 were considered significant. RLU, relative light units. doi:10.1371/journal.pone.0010954.g006 mixed with myoblasts sh.p38a.36GS.Luc and so forth). Luminometrybased cell-to-cell fusion measurements following the exposure of the various co-cultures to differentiation medium for 4 days showed that specific down-regulation of p38a expression in human muscle progenitor cells results in impaired cell fusion (Fig. 7D). These results together with those presented in Fig. 7A indicate that the p38 MAPK pathway is involved in terminal differentiation of human myoblasts in vitro.
In summary, we introduce herein a novel cell fusion assay based on the activation of a reporter gene by recombinase-dependent episome formation. We showed that, following lentivirus vectormediated gene transfer, the reporter units display low basal activity and a strict dependency on the FLP recombinase for their activation. These features combined with the ability of lentivirus vectors to swiftly generate transgenic mammals [16,17] might render the present FLP-responsive system useful to investigate cell fusion in vivo and, alone or together with Cre/loxP-based assays, assist in cell-lineage and conditional gene expression studies. Finally, in this post-genomic era, the challenge is to go from primary sequence data to phenotypic associations that are underpinned by functional read-outs. Key to this ongoing endeavor is the continuous development of robust high-throughput cell-based assays to link genotypes to specific phenotypes. Related to this, we expect the current gene switch system combined with high-throughput screening of chemical, small interfering/short hairpin RNA and cDNA libraries [27][28][29][30] to serve as a powerful methodology to identify molecules and pathways involved in the ill-understood process of mammalian muscle cell fusion. As an initial proof-of- principle we used the conditional gene expression assay presented herein to demonstrate by two different approaches the involvement of the p38 MAPK pathway during terminal differentiation of human myocytes in vitro.

Recombinant DNA
DNA constructions were carried out by using established procedures [31] or following the instructions provided with specific reagents. In brief, construct pLV.CMV.hMyoD.eGFP [32] (GenBank accession number: EU048697) was digested with Bsu15I and Bsp1407I (both from Fermentas). The 6.3-kb DNA segment containing the plasmid backbone was extracted from gel and its 39 recessed ends were filled in with Klenow polymerase (Fermentas). In parallel, pGS.pA+.DsRed [20] was digested with SmaI, HindIII and SspI. The resulting 2.4-kb fragment was extracted from gel and treated with Klenow polymerase. The 6.3kb and 2.4-kb DNA fragments were ligated to each other giving rise to construct pLV.pA+.GS.DsRed. To construct pLV.pA+ .GS.Luc, pLV.pA+.DsRed and pGL3.Basic (Promega) were digested with BcuI plus MluI and HindIII plus XbaI, respectively. Next, the 8.0-kb pLV.pA+.GS.DsRed fragment without the DsRed.T4-N1 ORF and the luciferase-coding DNA segment from pGL3.Basic were extracted from gel, blunt-ended with Klenow polymerase and ligated to each other, yielding pLV.pA+.GS.Luc.
All ligation mixtures were introduced into the GeneHogs strain of Escherichia coli (Invitrogen) and large-scale plasmid purifications were performed using JETSTAR 2.0 Plasmid Maxiprep kits (Genomed) according to the manufacturer's instructions. The nucleotide sequences of pLV.pA+.GS.DsRed and pLV.pA+ .GS.Luc can be retrieved by using GenBank accession numbers GU253312 and GU253313, respectively. In addition, the annotated sequences corresponding to the FLPe-encoding lentivirus vector shuttle plasmids pLV.FLPe and pLV.FLPe.PurR can be accessed via GenBank accession codes GU253315 and GU253314, respectively.

Cells
The culture conditions for the hMSCs [9] as well as the culture and differentiation conditions for the human myoblasts [14] have been described previously [9,14,33].
Physical particle titers were determined using the RETRO-TEK HIV-1 p24 Antigen ELISA kit (ZeptoMetrix) following the instructions provided by the manufacturer. Titers of lentivirus vector stocks expressed in terms of TU/ml were derived by using a conversion factor of 10 TU per pg of p24 protein.

Cell transductions
Human myoblasts and hMSCs were seeded at a density of 2610 5 and 3610 4 cells per well of 24-well plates (Greiner Bio-One), respectively, and were exposed overnight (circa 20 hours) to viral vectors in a humidified-air 10% CO 2 atmosphere. Next, the cell monolayers were rinsed multiple times with large volumes of PBS after which fresh medium was added. Mock infections were carried out in parallel cultures by using exclusively the respective cell type's culture medium.

Co-culture establishment and maintenance
Co-cultures containing 10 5 or 2610 5 cells were established in wells of 24-well plates by mixing myoblasts FLPe with myoblasts GS.Luc in the specified ratios After an incubation period of 48 to 72 hours the growth medium was either substituted by differentiation medium or by fresh growth medium. Unless otherwise indicated, the differentiation was allowed to proceed for 5 days after which the cultures were processed for analysis.

Small-molecule drug inhibition of the p38 MAPK pathway in human myoblast cultures
Myoblasts FLPe and myoblasts GS.Luc were seeded in a 1:1 ratio at a final total density of 2610 5 cells per well of 24-well plates. After 3 days of co-culture, the growth medium was replaced with differentiation medium, differentiation medium containing 0.1, 0.5 or 2.5 mM of 4-[4-(4-fluorophenyl)-2-(4-methylsulfinylphenyl)-1H-imidazol-5-yl]pyridine (SB 203580; Promega) or differentiation medium containing the same final concentration of SB 203580's vehicle dimethyl sulphoxide (Hybri-Max; Sigma-Aldrich) as the co-cultures incubated with 2.5 mM of the p38a/b inhibitor. The different types of media were replenished every 24 hours. Generation of cell lysates and luciferase measurements 3 days after the addition of the various types of differentiation media were carried out as described under ''Luciferase assay''.
Western blot analysis of shRNA-mediated knock-down of p38a gene expression The various shRNA-encoding polyclonal lines were established by exposing myoblasts and myoblasts GS.Luc for 24 hours to clarified supernatants of LV.sh.eGFP, LV.sh.hif1a, LV.sh.p38a.33, LV.sh.p38a.34, LV.sh.p38a.35 and LV.sh.p38a.36 producer cells. Subsequently, these polyclonal lines were sub-cultured in the presence of 0.5 mg/ml of puromycin. Control cells corresponding to parental myoblasts and myoblasts GS.Luc did not survive this drug selection regimen.
Monolayers of myoblasts and myoblasts GS.Luc either mocktransduced or stably transduced with the shRNA-encoding sequences targeting eGFP, hif1a and p38a transcripts (2610 5 cells) were rinsed with ice-cold PBS and dissolved in 100 ml of ice-cold lysis buffer composed of 20 mM Tris-HCl (pH 7.6), 150 mM NaCl, 0.1% sodium dodecyl sulfate (SDS), 0.5% sodium deoxycholate, 1% Nonidet P-40 and 10% glycerol plus a cocktail of protease inhibitors (Complete Mini, Roche Applied Science). Next, the lysates were spun for 8 min at 4uC at 20,8006g. Prior to sample loading, the total protein concentration in the various cell lysates was determined by using the BCA Protein Assay Kit (Thermo Fisher Scientific) and a bovine serum albumin standard curve. Instructions corresponding to the microplate procedure, as provided by the manufacturer, were followed except for the use of 3 instead of 25 ml of cell lysate. Also, the absorption measurements were performed at 545 instead of 562 nm in a Wallace 1420 VICTOR 3 multilabel plate reader (PerkinElmer). Next, equal protein amounts were subjected to electrophoresis through an SDS-12% polyacrylamide gel after which they were transferred onto an Immobilon-P membrane (Millipore) by electroblotting overnight. After incubation with blocking solution (10 mM Tris-HCl [pH 8.0], 150 mM NaCl and 0.05% Tween-20 [TBST]) supplemented with 10% (w/v) non-fat dry milk powder (Elk, Campina), the membranes were incubated overnight at 4uC with the rabbit anti-human p38a affinity-purified IgG antibody AF8691 (R&D systems) diluted 1:3000 in blocking solution. Subsequently, the membranes were washed 4 times with TBST at room temperature and subjected to a 3-hour incubation with horseradish peroxidase-conjugated goat-anti-rabbit IgG secondary antibody (SC-2009; Santa Cruz Biotechnology) diluted 1:5000 in blocking solution. Detection of a/b-tubulin proteins by using a rabbit anti-human a/b-tubulin affinity-purified IgG antibody (2148; Cell Signaling Technology) at a dilution of 1:5000, provided for loading controls. Finally, the membranes were rinsed and processed for protein detection by chemiluminescence as described elsewhere [22].
Effect of shRNA-mediated knock-down of p38a gene expression on human muscle cell fusion Myoblasts, myoblasts sh.hif1a , myoblasts sh.p38a.33 and myoblasts sh.p38a.36 were seeded at a density of 1.5610 5 cells in wells of 24-well plates. After an overnight incubation period, the cells were exposed for circa 16 hours to either LV.FLPe.PurR (15 TU/cell) or LV.GS.Luc (30 TU/cell). After being expanded via sub-culturing, the resulting FLPe-encoding cells were mixed at a 1:1 ratio with their respective GS.Luc-containing counterparts (i.e. 10 5 myoblasts sh.p38a.36FLPe were mixed with 10 5 myoblasts sh.p38a.36GS.Luc and so forth). Following 3 days in growth medium, the four different types of co-cultures were incubated in differentiation medium for 4 additional days, after which, reporter gene expression was determined as specified below.

Luciferase assay
Cell monolayers in wells of 24-well plates were washed with 1 ml of PBS and were lysed by incubation for 20 minutes in 200 ml of icecold lysis buffer consisting of 25 mM Tris-phosphate (pH 7.8), 2 mM dithiothreitol, 2 mM EDTA (pH 8.0), 10% glycerol and 1% Triton X-100. During the incubation period with the lysis buffer, the plates were swirled approximately every 5 minutes to homogenize the constituents present in the wells. Subsequently, the recovered cell lysates were spun at 14,000 rpm for 3 minutes at 4uC in a microcentrifuge (Eppendorf 5417R). The resulting supernatants were stored at 220uC until analysis. The luciferase activity present in one-tenth of each supernatant was measured with the aid of a Wallace 1420 VICTOR 3 multilabel plate reader loaded with luciferase assay substrate (Promega) reconstituted in luciferase assay buffer (Promega) as specified by the supplier.

Fluorescence microscopy
The equipment and methods used for the light microscopy analyses have been described before [22,[32][33][34].

Flow cytometry
The flow cytometry instrument and procedures have been specified elsewhere [22,34].