Myoblasts transplanted into muscles of recipient mice mostly die, only a minor stem cell-like subpopulation surviving and participating in muscle regeneration. To investigate this phenomenon further, we used a retrovirus expressing β-galactosidase to provide a unique marker for satellite-cell-derived muscle precursor cells, before transplanting them into myopathic mdx nu/nu mouse muscle. We employed inverse polymerase chain reaction to identify viral integrations, to follow the fate of clones present within the injected cells.
Mass-infected cultures contained many marked clones, some of which contributed disproportionately to muscle regeneration. Although no particular clones showed overall predominance, some were present in more than one injected muscle, an eventuality unlikely to arise by chance. Conversely, in grafts of muscle precursor cells that had either been labelled as sparse satellite-cell derived cultures, or had been cloned, all clones were shown to be able to survive and form muscle in vivo. Moreover, all clones contributed to further generations of new-formed muscle fibres following a series of injuries administered to injected muscles, demonstrating that some cells of each clone had been retained as stem-cell-like muscle precursors. Furthermore, retrovirally marked satellite-cell-derived clones were derived from muscles that had been injected with marked muscle precursor cells. These cells formed muscle following their transplantation into a new host mouse, confirming their stem cell properties.
Transplantation of muscle precursor cells (MPCs) in mice has demonstrated that only a small proportion of the total transplanted cells survive (Beauchamp et al., 1994; Beauchamp et al., 1997; Beauchamp et al., 1999; Fan et al., 1996; Huard et al., 1994a) to give rise to the new muscle that regenerates at the site. This is true of primary myoblasts (Beauchamp et al., 1999), clonal cell lines of immortomouse MPCs (Beauchamp et al., 1999), and the C2C12 myogenic cell line (Zammit and Beauchamp, 2001). Comparison of a genetic marker (the Y chromosome) and a semi-conserved radiolabel in grafted myoblasts revealed that cells that survived and proliferated within irradiated muscle in vivo constituted a distinct subpopulation that did not incorporate radiolabelled thymidine during overnight culture immediately before their transplantation. On this basis, the surviving cells were considered to represent a muscle stem-cell-like population (Beauchamp et al, 1999), but it was not possible to make more than approximate estimates of their proportion in the original population. Nonetheless, MPCs that survive and form muscle in vivo can proliferate in vitro. Muscle precursor cells are readily infected in tissue culture with retroviruses (which are only incorporated into dividing cells) and these retrovirally marked cells form new muscle in vivo, even after extensive tissue culture (Blaveri et al., 1999; Morgan et al., 2002).
These findings do not tell us if the donor cells that survive do so by chance, or because they are innately different. Chance survival may, for example, be due to cells either being at a particular stage of the cell cycle, or having been transplanted into a particularly favourable microenvironment. Innate differences may be due to some of the cells either expressing, or acquiring, a more primitive stem cell phenotype. We therefore decided to examine the clonal relationship between MPCs in vitro and those that survive transplantation to give rise to new muscle and/or MPCs in vivo. To do this, we used a retrovirus that expresses nuclear-localising LacZ as a genetic marker for MPCs in tissue culture, at titres that would generate, at most, one or two proviral integration events per cell. This type of retroviral marking has been used to follow the fate of haematopoietic stem cells (Capel et al., 1990; Nolta et al., 1996) progenitors in the nervous system (Price et al., 1987; Walsh and Cepko, 1992; Szele and Cepko, 1998; Johansson et al., 1999; Kardon et al., 2002), myoblasts during development (Hughes and Blau, 1990; Hughes, 1999) and following their implantation into adult mouse muscle (Hughes and Blau, 1992; Blaveri et al., 1999; Morgan et al., 2002). As each retroviral insertion into the host genome is both unique and heritable (Lemischka and Jordan, 2001), it constitutes a genetic marker of each transduced cell and can be used to trace its fate. Accordingly, we analysed the retroviral integration sites within muscles previously grafted with retrovirally marked myogenic cells derived from adult mouse muscle satellite cells. To detect the variety of integration events we used inverse PCR (IPCR), which amplifies the genomic DNA adjacent to the retroviral insertion site (Silver and Keerikatte, 1989; Triglia et al., 1988).
Our results show a marked reduction in the clones represented within grafted muscles after transplantation of a population containing hundreds of individual clones. We often found different clones present in different host muscles, although some clones contributed significantly to more than one muscle. However, any clone of myogenic cells that was isolated and expanded in tissue culture was able to engraft irradiated mdx nu/nu mouse muscle successfully.
Materials and Methods
Cell culture and retroviral labelling
Satellite cells were prepared by isolated fibre explants (Rosenblatt et al., 1995) from extensor digitorum longus (EDL) muscles of 2-3-week-old heterozygous H-2Kb–tsA58 mice (Jat et al., 1991; Morgan et al., 1994). Fibres were plated in dishes coated with 0.1 mg/ml Matrigel in DMEM (high glucose, with sodium pyruvate) supplemented with 10% horse serum, 0.5% chick embryo extract, 4 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin and grown at 37°C in 5% CO2. The following day, when approximately one cell had migrated from each fibre, cells were incubated for 2 hours at 37°C in the presence of pMFG nls lacZ (Ferry et al., 1991) supernatant, at a titer of 5×103 cfu/ml, with 8 μg/ml polybrene. The retroviral supernatant was then discarded and replaced with fresh medium. After two days, the medium was switched to DMEM containing 100 U/ml penicillin and 100 μg/ml streptomycin, 4 mM L-glutamine, 20% heat-inactivated fetal calf serum and 2% chick embryo extract (growth medium; GM) with γ-IFN (20 U/ml) at 33°C in 10% CO2. After 3-4 days, cells were detached using Hanks' balanced salt solution (HBSS) containing 0.05% trypsin and 0.02% EDTA, seeded in flasks coated in 0.01% gelatin and expanded before their implantation. Cells were passaged every 2-3 days, when they were approximately 40% confluent and plated at 5×104 cells per 175 cm2 flask. No fusion into myotubes occurred under these conditions. Cell death, if it did occur, was not conspicuous.
Population A was derived from 188 isolated fibres, population B from 17 fibres and population C from 31 fibres. As there is a mean of seven satellite cells per EDL muscle fibre (Zammit et al., 2002) this would have given a starting population at the time of infection of, at most, 1316 cells (population A), 119 cells (population B) and 217 cells (population C).
An aliquot of each preparation was fixed in methanol/acetone and stained for β-galactosidase (β-gal) (Dannenberg and Suga, 1981) to determine the proportion of cells that had incorporated the retrovirus and were expressing the marker protein. We also stained for desmin (Allen et al., 1991) to determine myogenicity. DNA was prepared from an aliquot of each cell preparation and analysed by IPCR.
Cells were cloned by dilution-cloning in 96-well plates. From poplulation A, 15 clones were expanded in vitro; five of these clones were transplanted individually into host muscles. From 25 clones obtained from population C, four that were β-gal positive (and had one or two retroviral integration sites) were selected. Population B was also cloned and one β-gal-negative clone was selected. These five clones were expanded individually in culture and mixed together immediately before transplantation.
Detection of integrated provirus
IPCR (Silver and Keerikatte, 1989; Triglia et al., 1988) has been employed to identify clones of retrovirally infected primitive human haematopoietic progenitors following their engraftment into immunedeficient mice (Dao et al., 1997) and autologous transplantation in rhesus macaques (Kim et al., 2000). Here, we have used IPCR to follow the fate of clones of muscle-derived cells implanted into myopathic mouse muscle. Two micrograms of DNA was subjected to restriction enzyme digestion with SacI, chosen for the number and position of the cutting sites within the proviral sequence (one site within the long terminal repeat sequence and one within the internal viral sequence, to generate a fragment that is of constant size and amplifiable by PCR). Fragments were then ligated at 15°C for 15 hours at a concentration of 1 ng/μl in a 500 μl reaction with 3U T4 DNA ligase. The low DNA concentration favours self-ligation of individual fragments and allows PCR amplification of the unknown flanking genomic DNA using two primers in opposite directions on the integrated provirus. The ligase was then inactivated at 68°C for 10 minutes before XbaI digestion to produce a linear template for the subsequent PCR amplification. In the IPCR scheme adopted (Fig. 1), restriction enzyme sites were chosen and primers designed such that each retroviral integration event should yield two products following PCR amplification: a constant-sized fragment from the internal viral sequence, and a variable-sized product, unique for each retroviral integration. IPCR was carried out in 200 μl thin-walled eppendorf tubes with 500 ng linearised DNA in a total volume of 50 μl, with 200 μM dNTPs, 45 ng of each primer (forward: GTTCCATCTGTTCCTGACCT, reverse: GACCTGAAATGACCCTGTGC), enzyme mix and buffer system 3 as supplied with the Roche Expand™ Long Template PCR system. Thermocycling parameters were as follows, denaturation at 94°C for 10 seconds, annealing at 58°C for 30 seconds and elongation at 68°C for 4 minutes. Amplification was carried out for 10 cycles, and a further 20 cycles followed with a 20 second extension to the elongation cycle. A further PCR amplification was carried out with 0.5-1 μl of the original reaction in a 50 μl total volume using standard protocols with Taq polymerase (Advanced Biotechnologies) and 30 cycles under the following conditions: denaturation at 94°C for 30 seconds, annealing at 58°C for 30 seconds and elongation at 72°C for 2 minutes. Internal primer sequences for this nested reaction were as follows F1: GCTTGCCAAACCTACAGGTG, R1: ACCAATCAGTTCGCTTCTCG. PCR products were visualised by agarose gel electrophoresis and their identity confirmed by Big dye terminator cycle sequencing (Applied Biosystems) with both internal nested primers. Sequence analysis was performed on a AB1 3700 genetic analyser.
Selected restriction enzymes and primer sequences yield two products for each retroviral integration. A constant (approximately 1.6 kb) fragment is produced by amplification of retroviral sequence between the 3′ LTR SacI site and an upstream SacI site that is constant for every integration. In addition, a variable sized product of retroviral 5′ATP and flanking genomic DNA should be obtained, whose size is dependent on the proximity of the site of retroviral integration to a SacI site in the genomic DNA (Fig. 1). SacI is cited as cutting every 3000 bp in the mouse genome (New England Biolabs catalogue).
Three days before cell implantation, three-week-old mdx nu/nu (Partridge et al., 1989) host mice were anaesthetized with hypnorm and hypnovel (Gross and Morgan, 1999). In most experiments, both hind limbs were exposed to 18 Gy gamma radiation in 25 minutes (Gross et al., 1999), but in one experiment, only the right leg was irradiated. Irradiated mdx nu/nu mouse muscle provides an environment that encourages implanted muscle precursor cells to proliferate (Beauchamp et al., 1999), to form new muscle and to repair damaged host muscle (Morgan et al., 1990; Morgan et al., 1993; Morgan et al., 1996; Morgan et al., 2002) and to give rise to long-lived precursor cells (Morgan et al., 1994; Gross and Morgan, 1999; Blaveri et al., 1999; Heslop et al., 2001).
MPCs were detached from the culture flask by incubation in HBSS. Cells were centrifuged at 350 g and cell pellets containing 5×105 MPCs were prepared. Mice were anaesthetised with hypnorm/hypnovel or isofluorane, and the skin overlaying the tibialis anterior (TA) muscle was opened. We injected 5×105 MPCs into the left and right TA muscles using a Hamilton 7005 syringe. At the time of implantation, an aliquot of cells was taken from each cell population for DNA preparation and IPCR analysis. In some experiments, mice were re-anaesthetised three weeks after cell transplantation and the right TA muscles were induced to degenerate by injecting 10 μl notexin (10 μg/ml in PBS) (Huard et al., 1994b; Gross and Morgan, 1999). Notexin specifically destroys muscle fibres, but spares other cells, such as satellite cells (Harris and Johnson, 1978). Animals were either sacrificed seven days later, or the right TA treated again with notexin three weeks after the first treatment and then sacrificed after a further seven days. Untreated TA muscles were therefore allowed to regenerate for four or seven weeks following initial cell transplantation.
Analysis of injected muscles
TA muscles, and in some experiments, extensor digitorum longus (EDL) muscles were removed for analysis three, four or seven weeks after the initial myoblast transplantation. Injected muscles were either snap frozen in liquid nitrogen and stored at –80°C or mounted in gum tragacanth (6% in water), frozen in liquid nitrogen-chilled isopentane and stored at –80°C.
Isolated muscle fibres were prepared from one TA muscle and one EDL muscle that had been injected three weeks previously with population A MPCs and cultured as described as above. An aliquot of any cells emanating from individual cultured fibres was stained for β-gal, desmin and expression of T antigen, and analysed by IPCR. Cells that emerged from two individual fibres (1 and 5) were expanded in tissue culture and were injected into both TA irradiated muscles of three (fibre 1) and four (fibre 5) new host mice.
Cryosections from muscles implanted with each cell preparation were analyzed to assess the extent of donor cell contribution to repaired muscle fibres.
Sections were cut from frozen embedded muscles on a Leica cryostat at 7 μm (hematoxylin and eosin and dystrophin staining) and 15 μm (X-gal staining). Adjacent cryosections were stained for the lacZ reporter gene expression (Dannenberg and Suga, 1981) and immunostained for dystrophin with p6 rabbit polyclonal antibody (Sherratt et al., 1992) as described previously (Morgan et al., 1993). Slides were examined and photographed with a Zeiss Axiophot fluorescence microscope using Metamorph Imaging System (3.5) software.
Detection of integrated provirus
Frozen transplanted muscles were thawed on ice, minced and DNA extracted by digestion in 500 μg/ml proteinase K in 100 mM NaCl, 10 mM Tris.Cl, 25 mM EDTA, 0.5% (w/v) SDS. Proviral sequences within injected muscles were detected by IPCR, gel electrophoresis and sequencing as described above.
Characteristics of the cell populations
Population A, B and C contained 84%, 66% and 73% β-gal-positive cells, respectively. Population A consisted of the 1.6 kb constant band and a smear following IPCR (Fig. 2A, lane C), indicating that it contained many cells each with different viral integration(s). Population B consisted of, at the most, four retrovirally labelled clones (Fig. 3A, lane 1) and population C contained, at most, 12 retrovirally labelled clones (Fig. 4A, lanes p4 and p8). This shows that only a minority of satellite cells placed into culture was able to give rise to clones that were capable of long-term growth.
From population A we obtained 15 clones (Fig. 2C); each contained, at the most, two retroviral integrations. One of these clones was β-gal negative and did not give rise to any IPCR products (Fig. 2C, clone 17); within the other 14 clones, the proportion of β-gal-positive cells ranged from 9 to 100% (Table 1). Although some of the clones contained products that appeared to be of similar size at the resolution of agarose gel electrophoresis (e.g. the 1.5 kb products in clones 8 and 18, and the 600 bp products in clones 6 and 14), sequence analysis showed that these were not identical (data not shown). The myogenicity of the clones from population A was determined (Table 1), and the five clones that were subsequently transplanted were >95% desmin positive.
When cells from preparations B and C were cloned, two of the clones had one IPCR product and two had two IPCR products (Fig. 4B). The products 750 bp in size (Fig. 4B) in clones 3 and 5 were identical by sequence analysis. However, the 450 bp products (Fig. 4B) in clones 1 and 5 were not of identical sequence, indicating either that `clone' 5 comprised a mixture of clone 3 and another clone, or that the retrovirus had integrated into the same, possibly preferential, site in two clones.
Retrovirally marked satellite-cell-derived MPCs give rise to skeletal muscle in vivo
The presence of donor-derived myonuclei within newly regenerated muscle fibres was confirmed with two markers. The progeny of the retrovirally infected cells were located using β-gal. Dystrophin immunostaining identified all muscle fibres derived from donor cells, whether or not they had been retrovirally labelled. Fibres containing β-gal-positive nuclei concurred with dystrophin-positive fibres, but there were generally fewer β-gal-positive than dystrophin-positive fibres (Figs 5, 6). This was to be expected, as not all of the cells were retrovirally infected. All of the marked nuclei were actually within, or tightly associated with, the surface of donor muscle fibres and most were in a peripheral position (Fig. 5b and Fig. 6d), indicating that, as found previously, the injected cells had given rise to muscle and to no other tissue type. Following the injection of population A into the TA muscles of ten mice, the injected TA muscles contained a mean of 418 dystrophin-positive fibres and 281 β-gal-positive fibres (Table 1; Fig. 5A,B).
Inverse PCR as a tool to investigate clonality of muscle regeneration
Although IPCR is more sensitive than Southern blotting for identifying integration sites of retroviruses, it is not quantitative and is limited by the relatively poor efficiency of the ligation step, and the inefficiency of PCR amplification of large DNA fragments. These limitations make it difficult to analyse complex mixtures of clones using IPCR and we may not detect all of the products that are present in the muscle. For example, we did not always detect the constant, retrovirus-derived product in every cell or muscle preparation that contained retroviral integrants. We seldom detected any IPCR products larger than 1.6 kb, although the calculated mean expected frequency of Sac1 sites is 1 in 3000 bp. It is therefore possible that the few clones (e.g. Fig. 2C, clones 3 and 19) that appeared to contain no retroviral insertion apart from the constant band, actually contained an undetectable integrant in a large product of the Sac1 digestion. Despite these drawbacks, IPCR is still a useful tool for analyzing the contribution of MPC clones to regenerated muscle.
Selection of satellite-cell-derived clones within injected muscles
Population A gave rise to many different products in the eight injected TA muscles (Fig. 2A, lanes 1-6,10-11) and in EDL muscles that were contiguous to injected TA muscles (Fig. 2A, lanes 7-9). Some muscles contained the constant product of amplified viral DNA of approximately 1.6 kb and a smear (Fig. 2A, lanes 1,5,7,10), indicating that many different marked clones were present within the muscle. Some muscles contained only one product (Fig. 2A, lanes 4,6,11) and others contained the constant band and several other products (Fig. 2A, lanes 2,3,8,9). None of the muscles analysed contained exactly the same clonal composition. However, two clones were present in more than one muscle: sequence analysis showed that the products of 1kb in lanes 2 and 4 were identical (Fig. 2A, red asterisk) and the products of 500 bp in muscles 2, 3 and 11 (Fig. 2A, yellow asterisk) were identical. None of the products in the injected muscles were identical to the products in the clones derived from population A (Fig. 2C) or the satellite cells extracted from the injected muscles (Fig. 2B). The constant product was absent in some muscles (Fig. 2A lanes 4,6,10,11), probably due to the inefficiency of the IPCR ligation step.
Satellite-cell-derived MPC clones can survive within injected muscles
Fig. 3A shows IPCR profiles of muscles injected with cell population B at two different passages and demonstrates that almost all clonal products present within the transplanted population are also represented within the transplanted muscles. A constant product of amplified viral DNA of approximately 1.6 kb was present in each lane (Fig. 3A, arrow C), and four variable sized products (Fig. 3A, arrows 1-4) were seen. Product 1 was seen only in the cell preparation and in muscle 5. Product 2 was present in the cell preparation and in all of the injected muscles, except in muscle 2, which had only the constant IPCR product, indicating that the viral insertion was too far distant from a Sac1 site in the host genome for the product to be expanded by IPCR. Product 3 was present in the cell preparation and in muscles 3, 4 and 5. Product 4 was present in the cell preparation and in muscles 4, 5 and 6. As no two products co-localised with each other in every muscle in which they were found, we can infer that each individual product represents a single clone of donor cells. In addition, the cell preparation contains at least one β-gal-negative clone identified by dilution cloning (Fig. 4B, clone 2). Sequence analysis showed that the 400 bp product 4 from cell population B (Fig. 3A, lane 1) was identical to the 400 bp product in muscle 5 that had been injected with cells at passage 9 (Fig. 3A, lane 5; Table 2).
More products were detected in cultured cells from population C (Fig. 4A, lanes p4,p8) than population B (Fig. 3A, lane 1), presumably because these cells were derived from a larger number of isolated fibres. There are 12 different products present in either the cell preparation or the injected muscles (Fig. 4A, arrows 1-12). We know, however, from sub-cloning these cells (Fig. 4B) that some clones had more than one viral integration and some clones had no viral integration. All of the injected muscles contained the 900 bp, 750 bp and 500 bp products represented in clones 3 and 4 (Fig. 4A). Muscle 1l contained, in addition, either clone 1 or 5, with a 450 bp product and other products (Fig. 4A, products 9,10,11). Muscle 2l contained products 7, 10 and 12 and muscle 3l contained products 7, 9, 10, 11 and 12.
Transplantation of the mixed clones (one β-gal-negative clone from preparation B and 4 β-gal-positive clones from preparation C) yielded IPCR products that were similar both to the cells grown in vitro to passage 19 and between analysed muscles (Fig. 4C, lanes 1l,2l,3l; Table 2). However, at least one product was missing from each injected muscle, except clone 4, which was present in all muscles. Because the products in injected muscles in this experiment were not sequenced, there is some uncertainty as to the contribution of clone 5 to some injected muscles. We do know that, with the possible exception of clone 5, each individual clone is present in at least one muscle.
When population C was injected into the irradiated right leg and non-irradiated left leg of three mice, fewer IPCR products were present in the non-irradiated (Fig. 4D, lanes 1l,2l,3l) than in the irradiated muscles (Fig. 4D, lanes 1r,2r,3r). This may have been because implanted cells form less muscle in the non-irradiated leg than in the irradiated leg (Morgan et al., 2002); clones may have been present, but in too small amounts to be detected by IPCR. Nonetheless, with the exception of the constant band and the 900 bp product, all products that were represented in the irradiated right legs were present in at least one non-irradiated host muscle.
All clones of MPCs give rise to muscle
The above results show that all marked clones were present within injected muscles, but there remains the unlikely possibility that one or more clones had not given rise to muscle. To confirm that individual clones could give rise to muscle, we injected 5 clones derived from population A into mdx nu/nu irradiated TA muscles. All of the injected clones gave rise to muscle (Table 1; Fig. 5). Three clones (15, 18 and 20) gave rise to similar amounts of donor muscle to population A, from which they were derived (Table 1; Fig. 5C,D). Two clones (clones 2 and 6), however, gave significantly less muscle (Table 1; Fig. 5E,F).
Retrovirally marked cells persist as muscle precursor cells and give rise to new muscle following transplantation
To ascertain whether all clones of donor cells gave rise to long-lasting stem-cell-like muscle precursors following their transplantation, we injected notexin into regenerated, cell-implanted muscles. Notexin destroys the majority of mature muscle fibres. The presence of newly regenerated muscle fibres of donor origin after notexin injection is evidence that some of the donor cells had been retained as precursor cells (Gross and Morgan, 1999). IPCR analysis of these newly regenerated muscles will demonstrate whether all, or just some, of the original clones give rise to stem-cell-like muscle precursors. Following one notexin injection, right TA muscles that had been injected with population B contained the same products as the non-notexin-treated left TA muscles (Fig. 3B, lanes 3l,3r,4l,4r). Muscles that had been injected with population C cells showed some differences in products between the right and left TAs (Fig. 4A). Muscle 1l had a 450 bp product 9 that was not present in the contralateral, notexin-treated muscle (1r) (Fig. 4A, lanes 1l,1r). Following cloning of this population, this 450 bp product can be identified as either clone 1 or clone 5 (Fig. 4B; Table 2). Muscle 2r had products 3 and 9 that were not present in the contralateral, non-notexin-treated TA (2l) (Fig. 4A lanes 2l,2r). Muscle 3r had products 2 and 3 that were present neither in the cell preparation at passage 4 or 8, nor in the contralateral muscle (3l) (Fig. 4A, lanes p4,p8,3r,3l). When mixtures of clones were injected, there were again differences in the products found in the notexin-treated muscles and the non-injured, contralateral muscles, (Fig. 4C; Table 2), but as there were also differences between the left leg muscles, one cannot assume that the differences were caused by notexin treatment.
Mice that received two consecutive notexin injections in the right TA muscle following the injection of population B cells contained the same products that were present in contralateral, non-notexin injected muscles (Fig. 3C, lanes 6l,6r,7l,7r,9l,9r). The one exception was muscle 8r, which contained only the constant product, indicating that the IPCR analysis of this muscle had failed.
Donor-derived muscle was, again, confirmed with two markers, β-gal to localize the progeny of the retrovirally infected cells, and dystrophin to identify all donor derived fibres. The two markers concurred; all of the β-gal marked nuclei were actually within, or tightly associated with, the surface of donor muscle fibres and most were in a peripheral position (Fig. 6d), indicating that, as found previously, the injected cells had given rise to muscle and to no other tissue type. Following notexin treatment, the majority of muscle fibres were smaller than those from muscle that had been transplanted with donor cells and had not been subsequently damaged (Fig. 6, compare sections a and c). Only these small diameter, newly regenerated fibres were dystrophin positive (Fig. 6a and e), indicating that they had been derived from the implanted cells.
Retrovirally marked cells migrate to adjacent muscles following transplantation
We also found IPCR products in EDL muscles adjacent to the injected TA muscles, whether or not the TA muscle had been notexin-treated, indicating that all donor clones were capable of entering adjacent muscles (Fig. 3B,C).
Retrovirally marked cells give rise to satellite cells that contribute to new muscle in a second host
Cells emanated from some single fibres prepared from regenerated muscles that had been transplanted with population A MPCs. Four fibres from one EDL muscle, and two fibres prepared from a TA muscle from a second mouse, yielded cultureable cells. These were expanded in tissue culture, as described above.
All of the cells expressed large T antigen, indicating that they were of donor origin (not shown). From the EDL muscle, cells from two of the fibres (Fig. 2B, lanes 3,4) were β-gal negative, and the other two fibres gave rise to both β-gal-positive and β-gal-negative donor cells. One fibre from the TA muscle gave rise to donor cells that were β-gal negative and the other fibre gave rise to cells that were 99% β-gal positive.
IPCR analysis (Fig. 2B) showed that cells from fibre 1 contained 2 products at 1.5 kb and 800 bp, in addition to the constant band. Cells from fibre 2 contained only one product at 1.7 kb, and cells from fibre 3 contained one product at approximately 800 bp, as well as the constant band. Cells from fibre 4 contained only the constant band and cells from fibre 5 contained, in addition to the constant band, products at 2.5 and 1.2 kb. Where two variable IPCR products were present, the cells from that single fibre may be derived from one retrovirally labelled clone with 2 integrants, or from 2 separate clones, each with one integrant. The two products of similar size in cells emanating from fibres 1 and 3 had different sequences (data not shown).
Cells from fibres 1 (from an EDL muscle) and 5 (from a TA muscle) were expanded in tissue culture, and injected into irradiated TA muscles of mdx nu/nu mice. Both cell populations gave rise to new muscle, but significantly less donor muscle than did the same number of the parental cell population A (Table 1). These clones behaved as classic stem cells, as they gave rise to satellite cells following their first injection and these donor-derived satellite cells gave rise to skeletal muscle following their injection into a second series of host mice.
Except in a few favoured systems, the detailed behaviour of cells in vivo is cryptic. True, it is possible to follow the fates of grafted cells carrying a heritable marker that distinguishes them from the host, but this gives only the broad picture, with little information as to the relative roles of subsets of cells within the grafted population. In skeletal muscle, a number of studies have shown that only a small fraction of a grafted myogenic cell population survives and participates in subsequent formation of new muscle (Beauchamp et al., 1994; Beauchamp et al., 1999; Fan et al., 1996; Huard et al., 1994a). Our previous work (Beauchamp et al., 1999) indicates that these surviving cells are a distinct stem-cell-like subset whose behaviour runs counter to that of the bulk population; being the least proliferative cells in vitro, while in vivo they rapidly proliferate in conditions where the major part of the population dies.
The idea that only a predetermined minority of myogenic cells survives transplantation is based on the preferential loss of cells that had been marked with radiolabelled thymidine during overnight culture. Here, we have used positive markers, to examine whether the surviving, muscle-forming cells are a minor subset of the implanted cells. The unique insertion sites of retrovirus coding for nuclear-localising lacZ have been used to follow the in vivo fates of individual marked myogenic clones from either a mass infected culture, or from mixtures of small numbers of marked clones, or from individual clones.
This same basic strategy has been used previously to demonstrate the oligoclonal basis of the repopulation of the haematopoietic system from injected bone marrow stem cells (Dzierjak et al., 1988; Dao et al., 1997; Kim et al., 2000) and the developing neural system (Price et al., 1987; Walsh and Cepko, 1992; Szele and Cepko, 1998; Johansson et al., 1999; Kardon et al., 2002) both characterized by high cell turnover that emphasizes any selective process. This contrasts with the relatively low turnover involved in formation of muscle tissue, in which, accordingly, there is relatively little room for selective processes to emerge. The strategy relies on each individual virus integrating into a unique site in the genome, being transmitted to the progeny of the infected cell (Dick et al., 1985) and continued expression of the marker gene both in vitro and in vivo. The retrovirus that we have used to mark our cells fulfils all three requirements.
In a culture where several thousands of cells were marked by retroviral infection, a large diversity of integration sites was indicated by a smooth smear of IPCR products, none of which had come to predominate during expansion of the culture. This population, when grafted, showed a marked simplification of its diversity, distinct bands appearing above the background in most instances. As many integrants were not resolvable into individual products (Fig. 2) there may have been some covert gains or losses of particular clones within these muscles. At the same time, it is clear that a few clones had expanded conspicuously by comparison with their representation in the myoblast preparation that had been implanted. Moreover, two of these clones had expanded in more than one graft site. This contingency suggests a degree of predetermination within the cells of these clones, because it is unlikely to arise by chance. Certainly, dilution cloning of this preparation in tissue culture yielded no instances of clones that were identical either with one another or with any of the major clonal bands that were present in the muscles into which this population was grafted (Fig. 2).
When muscle fibres were isolated from muscles that had been injected with marked MPCs and tissue cultured, we found that some retrovirally marked myogenic cells had given rise to satellite cells on the fibre surface. Satellite cells from individual fibres were expanded in culture and were found to be of separate clonal origin from one another and from any of the clones that had expanded within the muscle graft (Fig. 2C). MPCs derived from satellite cells on two individual fibres gave rise to new muscle when transplanted into a second mouse host. It remains to be determined whether satellite cells that have precursor/stem cell function can be formed by all clones of MPCs, but our data argues against the idea that satellite cells are formed from the same clones that make major contributions to muscle fibre regeneration.
In an attempt to simplify analysis, we examined the behaviour of individual clones within populations containing a small number of marked clones. Two such preparations, marked by retroviral infection of the satellite cells present on a small number of isolated muscle fibres, generated four or 12 distinct IPCR bands in culture, each of which was prominently present in at least one muscle transplanted with the preparation, indicating that all of clones were capable of contributing to muscle regeneration. However, some bands were lost within individual grafted muscles, confirming preferential takeover by a minority of myogenic cells within some regenerates. This same sporadic loss of individual clones was also noted when a mixture of four marked clones was grafted: some clones were absent from one or more grafts, but each individual clone was represented in at least one graft. When individual clones were injected separately into host muscles, all clones gave rise to muscle. However, some clones gave rise to significantly less donor muscle than any of the other clones, or the original cell preparation (Table 1).
Our data confirm a conundrum that was implied by previous work, namely, that any myogenic cell that could be clonally expanded in culture generates a minority subpopulation that exhibits stem-cell-like activity in graft sites (Morgan et al., 1994; Beauchamp et al., 1999; Blaveri et al., 1999). As this population is characterized by its slow cycle time in culture (Beauchamp et al., 1999), it ought to be diluted out of existence in prolonged clonal expansion. The ability of all individually cultured clones to form muscle in vivo implies that a muscle precursor cell clone grown in tissue culture contains a fraction that dies and a minor fraction that survives on grafting. It is possible that the surviving cells happened to be at a particular stage in the cell cycle at the time of implantation, whereas the majority of cells in the clone were at different stages. In the haematopoetic system, engraftment potential of cultured cells is lost at certain stages of the cell cycle (Glimm et al., 2000). The conferral of stem cell properties to a small proportion of grafted cells might arise in consequence of microenvironmental variation in vivo, as seen for instance in the oesophageal epithelium (Seery and Watt, 2000). These results are encouraging for the use of satellite cells to repair skeletal or assist cardiac muscle (Menasche, 2003), as the in vivo myogenic capacity of mouse muscle cells is not lost following extensive tissue culture. In addition, the ability of most retrovirally infected clones to make as much muscle as the parental polyclonal culture means that myogenic cells that are transduced and selected in vitro before their implantation need not necessarily have a reduced myogenic capacity in vivo.
It is difficult to fit our findings into a lineage-based hierarchical system of stem cell determination. If stem cell properties are acquired by lineage, then one would expect a progressive loss of slow-growing stem cell clones in culture. An alternative possibility is that conditions in which muscle precursor cells are grown may maintain or enhance stem cell behaviour. Culturing MPCs under low-density, proliferative conditions may either preserve precursor-like or stem-cell-like properties, or cause a cell to generate a more stem-cell-like phenotype. As virtually all cells that emanate from an isolated fibre of the immortomouse can be cloned, the latter proposition is the more likely. Support for this view is provided by the findings some years ago that the pattern of chromatin arrangement in the nuclei of mouse fibroblasts, presumably reflecting patterns of gene expression, became progressively disparate with increasing number of cell divisions (Abercrombie and Stephenson, 1969). Broad support for this idea also comes from the recent demonstration that haematopoietic stem cells in tissue cultures of skeletal muscle are far more potent (Jackson et al., 1999; Dell'Agnola et al., 2002) than those obtained by directly sorting the stem-cell-containing side population from fresh muscle (Gussoni et al., 1999; Lemischka, 1999). The diversification that occurs under certain cell culture conditions has been shown to include the acquisition of an earlier precursor status by the progeny of highly committed cells, e.g. conversion of oligodendrocyte precursor cells to neural stem cells (Kondo and Raff, 2000). Any such diversification mechanism, if randomly distributed, would have a high probability of being conferred on some members of most clones in oligoclonal cultures but only on a small proportion of clones in polyclonal cultures. This prediction concurs with our findings. On this basis, we may be able to educate cells to become functional stem cells by altering tissue culture conditions, rather than by selecting cells with particular stem cell markers.
This work was funded by the Medical Research Council and the Muscular Dystrophy Campaign. We would like to thank Robin Weiss and Yasuhiro Takeuchi for giving us the pMFG nls LacZ retrovirus and Olivier Danos for the plasmid from which the virus was derived. We would also like to thank Louise Heslop for her help with the isolated fibre cultures and Sarah De Val for her help with the sequencing.
- Accepted February 16, 2004.
- © The Company of Biologists Limited 2004