To investigate the factors that regulate incorporation into uninjured or damaged skeletal muscle of donor markers derived from unfractionated bone marrow (BM) cells or from highly purified c-kit+Thy1.1loLin−Sca-1+ hemato-poietic stem cells (HSCs), we evaluated myofiber chimerism of multiple muscle groups in irradiated and transplanted recipient mice and in unirradiated parabiotic animals. Uninjured panniculus carnosus, diaphragm, and abdominal muscles infrequently incorporated donor markers into myofibers in a subset of animals after either BM or HSC transplantation; however, acute muscle injury was essential to elicit contributions to triceps surae (TS) and tibialis anterior muscles. The low level of incorporation of donor marker–expressing myofibers could not be enhanced either by transplantation into newborn recipients or by induced migration of HSCs into the periphery. Analysis of muscle chimerism in unirradiated animals joined surgically by parabiosis revealed that contributions of circulating cells to myofibers in the TS were injury dependent and that at least some circulating cells with the potential to contribute to regenerating muscle derive from BM, suggesting that hematoablative preconditioning is not required for such contributions. In all cases tested, donor-derived myofibers expressed both donor-specific and host-specific markers, suggesting that they arise by low-level fusion into skeletal muscle of cells that can include the progeny of HSCs. It is not yet clear whether such events represent a normal myogenic pathway or a pathological response to muscle damage.
Stem cells are primitive, self-renewing cells that in many adult tissues function to maintain tissue homeostasis and to regenerate damaged tissue following injury [1–6]. In postnatal skeletal muscle, repair of tissue damage is thought to be mediated by tissue-resident satellite cells, located beneath the basal lamina of multinucleated myofibers [6–8]. However, several recent reports [9–14] have suggested that adult skeletal muscle may additionally derive from bone marrow (BM) precursors and have implicated hematopoietic stem cells (HSCs) or their progeny as possible candidates for this activity. Yet existing studies have not fully defined the dynamics of BM contribution to muscle, and it remains unclear whether transplanted cells home to muscle immediately after intravenous transfer or whether they may circulate in the bloodstream or engraft at other locations before being recruited to damaged tissue. Additionally, the biological processes that allow BM or HSC contributions to skeletal muscle are just beginning to be defined and may involve cell fusion [10, 15–19], transdifferentiation, or differentiation from a pluripotent [20–22] or muscle-committed  stem or progenitor cell.
We previously demonstrated that in irradiation-damaged tissues, single c-kit+Thy1.1loLin−Sca-1+ (KTLS) [2, 23] HSCs can fully repopulate the hematolymphoid system but essentially do not contribute to any nonblood tissues . To extend these data to evaluate the possibility that selective pressure resulting from tissue damage may recruit HSCs or their progeny to nonhematopoietic cell fates, we now have tested the capacity of unfractionated BM cells or of prospectively isolated KTLS HSCs to contribute to muscle cell lineages after transplantation into irradiated adult or newborn mouse recipients. In addition, using a parabiotic mouse model, in which genetically distinct animals are surgically joined such that they develop a common, anastomosed vascular system [24, 25], we have evaluated recruitment from circulation of cells capable of contributing to regenerating muscle. Together, these models allow the quantitative assessment of muscle descendants generated from BM, HSCs, or circulating cells during muscle growth, homeostasis, and repair.
Materials and Methods
Mice and Antibodies
C57BL/Ka, C57BL/Ka-Thy-1.1, and C57BL/Ka-Ly5.2/ Thy-1.1 mouse stains were bred and maintained at the Stanford University Research Animal Facility. Enhanced green fluorescent protein (GFP) transgenic mice were generated as described  and were backcrossed for at least 10 generations to C57BL/Ka-Thy-1.1 mice. GFP transgenic mice used as BM donors in these studies were 6–12 weeks old, and adult, nontransgenic recipients of transplanted cells were 10–14 weeks old. For parabiosis, animals were typically joined at 6–8 weeks of age, except for previously transplanted animals, which were joined at 16–24 weeks of age. HcRed transgenic mice were generated using a modified pCXEGFP vector (pCXHcRed) in which the enhanced GFP (eGFP) cDNA was replaced by cDNA encoding HcRed (Clonetech, Palo Alto, CA). The construction of the pCXEGFP vector and its use in generating eGFP transgenic mice has been described previously [26, 27]. To generate HcRed transgenic mice, pCXH-cRed was linearized, and the fragment containing the cytomegalovirus-immediate early enhancer, chicken β-actin promoter, eGFP cDNA, and β-globin polyadenylation sequences was gel purified using the Qiaquick gel extraction kit (Qiagen, Valencia, CA). Purified, linearized DNA was injected into F1 zygotes from crossed C57BL/Ka-Thy1.1 × C3H mice, and founder mice were screened by polymerase chain reaction and flow cytometry (A. Terskikh et al., unpublished data). Transplanted recipient β-actin/HcRed mice used in these studies were backcrossed for at least two generations to C57BL/Ka-Thy-1.1 mice. One hundred percent of skeletal muscle myofibers are HcRed+ in these animals (data not shown). The antibodies used in these studies included 19XE5 (anti-Thy1.1, phycoerythrin conjugate), 2B8 (anti-c-kit, allophycocyanin conjugate), E13-161.7 (anti-Sca-1, Ly6A/E, Texas Red conjugate), polyclonal anti-GFP (unconjugated or Alexa488 conjugate [Molecular Probes, Eugene, OR]), JL-8 (anti-GFP [Clontech]), MY-32 (anti-skeletal myosin [fast] [Sigma, St. Louis]), NOQ7.5.4D (anti-skeletal myosin [slow] [Sigma]), EA-53 (anti-α-actinin [Sigma]), MANDYS8 (anti-dystrophin [Sigma]), 30-F11 (anti-CD45, unconjugated or biotin conjugate [BD Pharmingen, San Diego]), goat anti-mouse immunoglobulin G (IgG) (Alexa594 conjugate [Molecular Probes]), goat anti-rat IgG (Alexa594 conjugate [Molecular Probes]), and goat anti-rabbit IgG (Alexa594 conjugate [Molecular Probes]; or biotin conjugate [Vector Labs, Burlingame, CA]). The cocktail of lineage marker antibodies included KT31.1 (anti-CD3), GK1.5 (anti-CD4), 53-7.3 (anti-CD5), 53-6.7 (anti-CD8), Ter119 (anti-erythrocyte specific antigen), 6B2 (anti-B220), 8C5 (anti-Gr-1), and M1/70 (anti-Mac-1). Unless otherwise indicated, monoclonal antibodies were produced and purified in this laboratory.
BM was flushed from the femurs and tibiae of donor mice with Hanks' balanced salt solution (Life Technologies, Gaithersburg, MD) supplemented with 2% fetal calf serum and 2 mM EDTA. Red blood cells were lysed during a 3-minute incubation in 0.15-M ammonium chloride and 0.01-M potassium bicarbonate solution on ice. BM cell suspensions were filtered through nylon mesh before transplantation or staining for HSCs.
Fluorescence-Activated Cell Sorting of HSCs
GFP+ HSCs were isolated by double fluorescence-activated cell sorting (FACS) of c-kit–enriched BM from GFP transgenic mice, based on previously defined reactivity for particular cell-surface markers (KTLS) [2, 23, 28]. All antibody incubations were performed on ice for 15–25 minutes. BM cells were first stained with purified lineage antibodies followed by anti-rat Cy5PE (Caltag, Burlingame, CA). Lineage-stained cells were further stained for c-kit with biotinylated 3C11, and c-kit+ cells were enriched by positive selection using MACS (Miltenyi Biotec, Sunnyvale, CA) streptavidin-conjugated magnetic beads and AutoMACS cell-separator columns according to the manufacturer's instructions. c-kit–enriched cells were stained with fluorescently labeled 2B8, 19XE5, and E13-161.7 monoclonal antibodies to identify HSC. 3C11 and 2B8 recognize distinct, nonoverlapping epitopes of c-kit. Before FACS analysis, cells were suspended in 1 μg/ml of propidium iodide (PI) to identify and exclude dead (PI+) cells. KTLS HSC populations were double sorted to ensure purity, using a highly modified Vantage SE (Becton, Dickinson Immunocytometry Systems, Mountain View, CA), provided by the Stanford University Shared FACS Facility. Flow cytometry data were analyzed using FlowJo (Treestar, San Carlos, CA) analysis software.
BM and HSC Transplantation
Recipient mice were transplanted with 1 or 100 double-sorted KTLS HSC or 5 × 105 to 5 × 107 BM cells. Newborn recipient mice were 1–2 days old at time of transplantation and received a sublethal dose of irradiation (400 Rad, delivered in a split dose 3 hours apart) before cell transfer. GFP+ BM cells or BM HSCs were transplanted into newborn recipients by direct injection into the liver. Adult recipient mice received a lethal dose of irradiation (950 Rad, delivered in a split dose 3 hours apart) before transplantation with GFP+ cells by retro-orbital injection. Single HSC transplants were performed as described previously . Transplanted recipients were screened by flow cytometry for GFP+ leukocytes in peripheral blood (PB) greater than 8 weeks after transplant. B cells, T cells, and myeloid cells in the PB were identified by FACS staining with anti-B220, anti-CD3, or anti–Mac-1 plus anti–Gr-1, respectively. In this study, the average hematopoietic chimerism of animals transplanted as newborns with unfractionated BM cells was 18 ± 11% and with KTLS HSCs was 27 ± 25%. The average hematopoietic chimerism of animals transplanted as adults with unfractionated BM cells was 96 ± 4% and with KTLS HSC was 91 ± 9%. HSC- and BM-transplanted mice were subjected to muscle injury and regeneration ∼ 2–11 months after reconstitution (see below).
Muscle Regeneration Assays
Muscle regeneration was monitored after mechanical or chemical injury. Mechanical injury was caused by crush  injury to the triceps surae (TS) muscles of the hind limb. Briefly, a 3-mm incision was made over the TS of an anesthetized mouse, a pair of forceps was inserted midmuscle, and the TS muscles were crushed transversely one to three times, causing an injury 3- to 4-mm wide while maintaining the longitudinal continuity of the muscle . Muscle regeneration was also evaluated in a chemical injury model by injecting an anesthetized mouse with 25 μl of 0.3-mg/ml solution of cardiotoxin (from Naja mossambica mossambica, Sigma) directly into the TS or tibialis anterior (TA) muscles  one time only or three times at 1-week intervals. To assay for BM or HSC-derived myogenesis, previously transplanted animals exhibiting high-level, multilineage hematopoietic engraftment were injured and analyzed for donor-derived muscle regeneration 5–8 weeks after injury.
HSCs and progenitor cells were mobilized by treatment with cyclophosphamide/G-CSF (Cy/G), as described previously [26, 30]. On day–1, mice previously transplanted with GFP+ HSC or BM (see above) were injected intraperitoneally with approximately 200 mg/kg (generally ∼4 mg/mouse) Cy (Bristol-Myers Squibb, New York). Then, on each successive day until day +7, the mice were injected subcutaneously with approximately 250 mg/kg (∼5 mg/mouse) human G-CSF (Amgen Biologicals, Thousand Oaks, CA). On day +3, the animals received a single intramuscular injection of cardiotoxin (25 μl, 0.3 mg/ml) in the TS of one leg. The other leg was left uninjured. Daily G-CSF injections were continued through day +6. On day +7, animals were bled via the tail vein for flow cytometric analysis of stem and progenitor cell frequency. Animals were euthanized, and their tissues were harvested for analysis of donor cell engraftment 8 weeks after cardiotoxin-induced injury.
Parabiosis surgery was performed exactly as described  and in accordance with the guidelines established by the Stanford University Administrative Panel for Lab Animal Care for the humane care and use of animals. The TS muscles of the outside legs (those not adjacent to the site of surgical joining) of both partners of the parabiotic pair were injured by mechanical crushing on the day of parabiosis or by injection of cardiotoxin approximately 3–4 months after parabiosis. Muscle engraftment was evaluated by immunofluorescence, as described below, 8 weeks after injury.
Harvesting of Tissues
Transplanted or parabiotic animals were anesthetized and perfused intracardially with 10–20 ml of 10-mM EDTA/1 × phosphate-buffered saline (PBS), followed by 10–20 ml 2% paraformaldehyde/1 × PBS. For FACS analysis of chimerism in blood and BM of parabionts, blood was collected before perfusion with EDTA, and marrow was harvested subsequently but before perfusion with paraformaldehyde. Perfused tissues were dissected and further fixed for 2–4 hours in 2% paraformaldehyde at room temperature. Fixed tissues were washed with 1 × PBS, cryoprotected by overnight incubation in 30% sucrose, and then quick frozen in optimum cutting temperature (OCT) compound, stored at −80°C, and removed for cryosectioning as needed.
Immunofluorescence analysis was performed on frozen sections of skeletal muscle (including TS, TA, panniculus carnosus [PC], diaphragm, and abdominal muscles). Frozen sections (5–12 μm) were cut at −20°C from OCT-embedded tissues using a 5030 series microtome (Bright Instruments, Huntingdon, U.K.). Sections were air dried overnight at room temperature and then stained. Sections were blocked using the MOM blocking kit (Vector Labs), the Avidin/ Biotin blocking kit (Vector Labs), or 2% normal goat serum, as appropriate. For staining of intracellular antigens, 0.5% Triton X-100 was added to the blocking solution. Serial sections were stained for immunofluorescence with anti-GFP and with muscle-specific (anti-dystrophin, anti-α-actinin, or anti-myosin) and pan-hematopoietic (anti-CD45) markers. Primary antibodies were detected using the appropriate Alexa594-conjugated secondary antibody (Molecular Probes). Endogenous GFP signals were amplified by staining with Alexa488-conjugated anti-GFP (Molecular Probes) or with the anti-GFP monoclonal antibody JL-8 (Clontech) followed by biotinylated anti-mouse IgG antibody (Vector Labs) and Alexa594-conjugated streptavidin (Molecular Probes). Nuclei were labeled with Hoechst 33342 (Molecular Probes). GFP was also visualized by immunohistoc hemistry using either the Vectastain Elite ABC kit (Vector Labs) containing the chromagen diaminobenzidine tetrahydrochloride or the Vectastain ABC-AP kit (Vector Labs) containing the chromagen Vector Red. Immunofluorescent labeling was analyzed both by standard fluorescence microscopy using a Nikon Eclipse E800 microscope with epifluorescence powered by a super-high-pressure mercury lamp (Nikon, Tokyo) and by laser-scanning confocal microscopy using the LSM 510 confocal Laser Scanning microscope (Zeiss, Thornwood, NY) with a Coherent Mira 900 tunable Ti, Sapphire laser for 2 photon excitation and analyzed with LSM 510 software (Zeiss), provided by the Stanford University Cell Sciences Imaging Facility. For standard epifluorescence, sequential images were acquired using a SPOT RT CCD camera (Diagnostic Instruments, Sterling Heights, MI) for Hoechst 33342, Alexa594, and GFP using UV-2A, HYQ Texas Red, and HYQ FITC (Nikon) filters, respectively, and electronically merged using SPOT RT software (Diagnostic Instruments). For confocal microcopy, images of serial optical sections were recorded every 1.0 mm per vertical step and analyzed with LSM 510 and Axiovision Viewer software analysis tools (Zeiss). Immunohistochemical staining was analyzed using the Nikon Eclipse E800 microscope, with color images captured with the SPOT RT CCD camera. In all cases, appropriate negative and isotype controls demonstrated antigen-specific labeling by each of these antibodies.
Quantitation and Statistics
Frequencies of in vivo myofiber contributions were determined as the number of GFP+ myofibers detected divided by the total number of myofibers examined. Frequencies of GFP+ myofibers were determined for each animal after examination of at least 10 independent sections, representing at least 20,000 myofibers. To achieve a representative sampling of muscle engraftment, serial sections were separated from one another by at least 100 μm. Data were analyzed by nonparametric Kruskal-Wallis one-way analysis of variance and Mann-Whitney U tests, using the WinSTAT statistics add-in for Microsoft Excel. Differences were considered statistically significant at p < .05.
To investigate potential contributions of BM cells or BM HSCs to the repair of damaged skeletal muscle, irradiated newborn or adult recipient mice were transplanted with highly purified KTLS HSCs or unfractionated BM cells isolated from β-actin/GFP-expressing donors, and chimeric mice were subsequently injured by either single or repeated intramuscular injection of cardiotoxin or by mechanical crushing of the TS or TA muscles. Both BM- and HSC-transplanted animals showed roughly equivalent, stable, multilineage hematopoietic engraftment by GFP+ cells before muscle injury (see Materials and Methods). Eight weeks after injury, injured and uninjured muscles were harvested and serial frozen sections were prepared and stained with anti-GFP and with antibodies recognizing the muscle-specific markers dystrophin, α-actinin, or skeletal muscle myosin or the pan-hematopoietic marker CD45 [32, 33]. Donor contributions to muscle were identified as cells that expressed GFP, costained for muscle markers, and did not express CD45. In addition, to exclude autofluorescent cells and confirm GFP expression, GFP was visualized both by immunofluorescence and by immunohistochemistry.
In adult animals transplanted with unfractionated BM cells, GFP+ muscle fibers were observed in the skeletal muscle of a fraction of mice injured either once or repeatedly (Fig. 1). Although GFP+ myofibers were never observed in uninjured TS muscles of BM-transplanted animals (n = 4; data not shown), GFP+ muscle cells were present in both the TA (2 of 6 muscles analyzed) and the TS (15 of 19 muscles analyzed) of animals injured by crushing or cardiotoxin injection, indicating that tissue damage significantly increases the contribution of GFP+ cells to muscle (p < .05). In BM-transplanted animals, repeated injury did not significantly increase the rate of incorporation of GFP+ myofibers over that observed with single injury in either the TA or TS (p > .05), and overall the frequency of incorporation of GFP+ myofibers in BM-transplanted mice was exceedingly rare (generally less than 0.4% of total myofibers). Transplantation of BM cells into newborn animals also did not enhance the ability of these cells to contribute to myofibers (Fig. 2). Although GFP+ myofibers were never observed in the uninjured TS of BM-transplanted mice ( and data not shown), they were detected at a very low frequency in uninjured diaphragm (1 of 20 mice examined), PC (2 of 9 mice examined), and abdominal muscles (1 of 20 mice examined) (Figs. 1, 2). BM-derived contributions to myofibers were not enhanced after injection of BM cells directly into cardiotoxininjured TS muscle, as has been described previously , or by Cy/G-CSF–induced mobilization of HSCs and progenitor cells [30, 35, 36] in animals previously transplanted with unfractionated BM (data not shown). In particular, although Cy/G-CSF–treated animals displayed an approximate 30- to 50-fold increase in blood-borne HSCs at the time of muscle injury, they showed no significant increase in the frequency with which GFP+ myofibers were found in the damaged muscle (0.03% GFP+ myofibers, p > .05).
As in BM-transplanted animals, we also observed incorporation of GFP+ muscle fibers into skeletal muscle of a fraction of animals transplanted with purified KTLS HSCs. Although no donor contributions were detected in the TS of HSC-transplanted animals injured only once (0 of 10 adult recipients and 0 of 7 newborn recipients), repeated injury of the TS did allow the incorporation of donor markers into myofibers in a subset of transplanted mice (one of three recipients; p < .05) (Figs. 1, 2). Furthermore, both singly and repeatedly injured TA muscles of HSC-transplanted mice contained GFP+ myofibers (Fig. 1). As in BM-transplanted animals, the frequency of GFP+ myofibers observed in the injured muscle of HSC-transplanted mice was very low (typically <0.2% of total myofibers). Repeated injury resulted in the incorporation into regenerating TS muscle of a greater frequency of donor marker–expressing muscle cells and increased the frequency of chimeric mice in the TS (p < .05) but not in the TA. We also observed very rare GFP+ myofibers in the uninjured PC of 2 of 10 KTLS HSC-transplanted mice, but not in uninjured diaphragm, TA, TS, or abdominal muscle (Fig. 1).
To demonstrate unequivocally that the GFP+ myofibers found in damaged skeletal muscle of HSC-transplanted animals truly arise by incorporation of donor markers derived from KTLS HSCs, we also analyzed mice whose hematopoietic system had been highly reconstituted (>80% GFP+ blood leukocytes) by a single GFP+ KTLS HSC. In one such animal, although no GFP+ myofibers were detected in singly injured TA (0 of 6,720 fibers examined) or TS (0 of 16,198 fibers examined) muscles, triple injury of either muscle on the contralateral leg did result in low-level incorporation (0.015% [2 of 13,498] of TA fibers examined and 0.033% [4 of 12,052] of TS fibers examined) of cells coexpressing GFP and skeletal muscle markers (Fig. 3).
In either BM-transplanted or HSC-transplanted animals, donor-derived GFP+ muscle fibers detected in the TA or TS muscles occurred predominantly as single, isolated cells, were well integrated into the muscle tissue, and were otherwise indistinguishable from surrounding GFP− myofibers (Figs. 1–3Figure 3.). Most of these GFP+ muscle fibers in both the TA and TS muscles were type II fibers, expressing fast, but not slow, skeletal muscle myosin (Fig. 1 and data not shown). Taken together, these data indicate that both BM and HSCs can contribute donor markers to injured and regenerating skeletal muscle. In addition, different muscles seem to differ both in the rate with which they incorporate HSC- or BM-derived markers into myofibers and in their requirement for acute muscle injury to evoke such contributions.
HSCs, as well as some nonhematopoietic, tissue-specific progenitor cells, constitutively circulate in the bloodstream [31, 37]. To evaluate the possibility that HSC- or BM-derived cells, or circulating muscle progenitors, may be recruited to skeletal muscle from normal circulation in the absence of lethal irradiation and iatrogenic infusion of these cells, and thereby contribute to injured muscle, we used a parabiotic model in which two mice, one GFP+ and one GFP−, were surgically joined such that they developed a common, anastomosed, circulatory system . In parabiotic mice, blood chimerism is detectable within 2–3 days of joining, reaches approximately 50% 8–10 days after parabiosis, and is maintained at this level thereafter . TS muscles of nontransgenic (GFP−) partners of parabiotic pairs were injured by crushing at the time of joining or by cardiotoxin injection 3–4 months after parabiosis. Eight weeks after injury, injured and uninjured muscles were harvested and analyzed for the presence of GFP+ (partner marker+) myofibers. All injured parabiotic mice analyzed (five of five) incorporated GFP+ myofibers at the site of injury in the TS, regardless of whether the injury occurred at the time of parabiosis or several months after parabiosis (Fig. 4). A cluster of GFP+ myofibers was observed in one animal; however, most GFP+ myofibers in injured parabiont muscle were single, isolated cells. Consistent with previous studies , no GFP+ partner-derived myofibers were detected in uninjured muscles (diaphragm, abdominal, or TS) of parabiotic partners (Fig. 4). Thus, injury promotes contributions of circulating cells to muscle in parabiotic animals (p < .05). Because parabiotic mice are connected only through their shared vasculature, these data indicate the existence of circulating cells that detectably engraft skeletal muscle only after injury; such cells may be constitutively present in the bloodstream or may be induced to enter the blood in response to muscle injury.
To investigate the relationship between BM-derived and circulating myogenic cells, we also generated parabiotic pairs using two nontransgenic animals, one of which had previously undergone GFP+ BM transplantation. The TS muscles of both animals were injured by crushing at the time of parabiosis and were analyzed 8 weeks later for the presence of GFP+ myofibers. GFP+ myofibers were found at low frequency in the injured TS muscles of both the BM-transplanted mice (three of three mice; data not shown) and their untransplanted partners (two of three mice; Fig. 4). In one pair, rare GFP+ myofibers were also observed in the uninjured abdominal muscle of the BM-transplanted partner but were never found in uninjured muscle in untransplanted parabionts. Thus, these data indicate that the incorporation of genetic markers contained in BM-derived cells into muscle in response to injury can occur long after irradiation and transplantation of BM cells and that at least some circulating cells capable of contributing to skeletal muscle are BM-derived; nonetheless, it remains possible that distinct cell populations with identical or overlapping function may exist also in circulation.
The precise biological processes that allow BM or HSC contributions to skeletal muscle are clearly of interest in understanding this phenomenon and potentially could involve cell fusion [10, 15–19], transdifferentiation, or differentiation from a pluripotent cell [20–22] or muscle-committed stem or progenitor cell . Therefore, to evaluate the relative importance of de novo myogenesis or cell fusion in the formation of BM- or HSC-derived GFP+ myofibers, we developed an additional transgenic mouse model that ubiquitously expresses a spectrally distinct fluorescent protein (HcRed) from the β-actin promoter (A. Terskikh et al., unpublished data) and transplanted these β-actin/HcRed mice, which express the unique host-specific marker in 100% of skeletal myofibers, with GFP+ BM or KTLS HSCs. We then analyzed skeletal myofibers in the TA and TS muscles of these mice after cardiotoxin-induced injury for possible coexpression of the donor-specific marker (GFP) and the host-specific marker (HcRed). All of the GFP+ myofibers detected in the regenerating muscle of HcRed recipient mice transplanted with either GFP+ unfractionated BM or KTLS HSC coexpressed GFP and HcRed, indicating that these fibers formed through cell fusion rather than de novo myogenesis (Fig. 5) and ruling out direct differentiation or transdifferentiation of BM cells or KTLS HSCs on their own into mature skeletal myofibers in these injury models.
Using transplantation and parabiotic mouse models, we observed low-level incorporation of GFP-marked donor myofibers into uninjured and regenerating skeletal muscles of animals transplanted with unfractionated BM cells or with highly purified HSCs and of parabiotic animals. The capacity of BM or HSCs for muscle engraftment seems to depend both on the particular muscle group analyzed and on the severity of muscle injury. In the TS muscles, injury was essential to reveal muscle contribution. In addition, although donor myofibers were incorporated into the injured TS of BM-transplanted adult mice in response to a single acute injury, repeated injury was necessary to evoke HSC-derived contributions to TS muscle regeneration. Similarly, in animals transplanted at birth, singly injured TS muscles of BM-transplanted animals, but not HSC-transplanted animals, infrequently incorporated GFP+ myofibers. Interestingly, the frequency of animals exhibiting chimeric muscles was substantially lower if the transplant was performed at birth rather than in adulthood (p < .05), which may relate to differences in the conditioning regimen for engraftment (sublethal versus lethal irradiation), differences in overall hematopoietic chimerism (average hematopoietic chimerism of ∼20% versus >90%), developmental regulation of the availability of BM or muscle niches, or other, presently unknown factors.
In contrast to the TS, in injured TA muscles, single injection of cardiotoxin was sufficient to elicit low-level contributions of GFP+ myofibers in a subset of animals after either BM or HSC transplantation. This difference likely relates to differences in the severity of cardiotoxin-induced muscle injury, because the same dose of cardiotoxin may be expected to create a more significant injury in the small TA muscle as opposed to the larger TS muscle. Finally, although incorporation of donor markers from transplanted cells was never seen in uninjured TA or TS muscles of either BM or HSC transplant recipients, in a subset of animals, the PC muscle, a striated, twitching muscle located under the dermis, incorporated BM- or HSC-derived markers into myofibers, albeit at a very low frequency, in the absence of acute muscle injury. In addition, in two individual BM-transplanted animals, we observed GFP+ myofibers in uninjured abdominal muscle and uninjured diaphragm. The underlying cause for the lack of injury requirement in engraftment of the PC, abdominal muscle, and diaphragm is unknown but could relate to the distinct ontological derivation of these muscles compared with the limb muscles or to differences in the normal rate of turnover of myofibers within these different muscle groups.
Taken together, these studies demonstrate rare incorporation of donor-derived genetic markers into myofibers within injured and regenerating skeletal muscles (TS and TA), and, even more infrequently, within the uninjured PC, abdominal muscle, and diaphragm after transplantation with either unfractionated BM cells or highly purified HSCs. Significantly, we and others [10, 11] have confirmed the ability of individual HSCs to generate both GFP+ hematopoietic cells and myofibers by analysis of mice reconstituted by a single, transplanted HSC. Nonetheless, our data may also suggest that both HSC and non-HSC stem/progenitor cells within the BM may contribute to damaged muscle tissue under appropriate conditions. In particular, the observation that unfractionated BM cells but not highly purified HSCs contribute to TS muscle regeneration after a single injection of cardiotoxin may indicate that, in this particular model, non-HSC BM cells are required to facilitate muscle engraftment by HSCs or their progeny or that less-severe injury conditions support contributions from circulating, transplantable, fusogenic, or myogenic BM cells distinct from KTLS HSCs. Although the identity of such cells remains to be determined, they could conceivably be mesenchymal or myogenic cells—perhaps stem or progenitor cells in these lineages—or pluripotent cells [20–22, 38] that include muscle in their progeny and have the unexpected property of circulating to and incorporating into skeletal muscle myofibers.
To further assess skeletal muscle contributions from circulating cells and to evaluate the kinetics of muscle engraftment from transplanted BM cells, we extended our transplantation studies to assay incorporation of donor markers into regenerating muscle in parabiotic mice. These data clearly demonstrate the existence of circulating cells that detectably engraft skeletal muscle only following injury. Such cells may be constitutively present in the bloodstream or may be induced to enter the bloodstream specifically in response to muscle injury. As demonstrated in parabiosis experiments joining previously transplanted animals, at least some circulating cells capable of contributing to skeletal muscle are BM-derived and may be recruited through the circulation to damaged muscle long after transplant; nonetheless, it remains possible that distinct cell populations with identical or overlapping function may exist also in circulation.
Although BM-derived, HSC-derived, and circulating cells all contribute genetic markers to muscle after injury, the mechanisms responsible for each of these activities has not been entirely clear. However, direct injection of BM or HSCs into muscle did not yield donor marker–expressing myofibers ( and data not shown), suggesting that muscle contributions from BM cells may require prior engraftment of host BM or other nonmuscle compartments, passage through the bloodstream, or specific damage induced by irradiation. In addition, pharmacological mobilization to increase the frequency of circulating HSCs and other progenitor cells coincident with the induction of muscle injury did not significantly increase the frequency with which myofibers expressing BM-derived markers incorporate into the injured muscle, suggesting that the availability of these cells in the bloodstream at the time of injury is not a limiting factor.
To test the role of cell fusion in the generation of donor marker–expressing myofibers, we transplanted GFP+ BM or HSCs into HcRedlabeled hosts and assayed injured muscles for expression of donor GFP and host HcRed. In all cases tested, both donor-specific and host-specific markers were coexpressed in the same myofibers, suggesting that injury promotes low-level fusion into skeletal muscle of cells that, in highly injured muscle, can include progeny of HSCs. However, importantly, because cell fusion from myogenic precursors is the process by which multinucleate skeletal muscle cells are typically generated in vivo, these experiments only rule out direct differentiation or transdifferentiation of BM cells or KTLS HSCs on their own into mature skeletal myofibers in these injury models. Thus, our data may indicate that the detection of BM- or HSC-derived markers in muscle results from fusion of circulating hematopoietic cells into existing, possibly damaged, mature skeletal myofibers and not through conventional myogenic pathways, as has been suggested previously . Such a mechanism is supported by data from Goodell and colleagues  that strongly implicate cells of the myeloid lineage in muscle fusion events. It remains unclear whether these fusion events represent a normal pathway of muscle regeneration or result from the activity of cells capable of cell fusion in pathologic circumstances such as occurs in the generation of multinucleate macrophage-derived giant cells in unresolved tuberculosis infections .
Finally, the very low frequency with which donor-derived myofibers are detected in BM-transplanted and parabiotic animals and the severity of injury necessary to evoke these low-level contributions strongly suggest that BM and blood are not primary or physiological sources of cell replacement in normal or regenerating muscle. Thus, although BM transplant for treatment of muscle degenerative disease has been proposed, given the infrequent contributions of these cells to myofibers, even after severe muscle injury, such approaches seem unlikely at present to yield significant improvements in muscle function. Nonetheless, a more detailed understanding of the mechanisms and regulators of muscle recruitment of BM and circulating cells may lead to strategies that enhance this unexpected phenomenon to therapeutically relevant levels.
We thank L. Jerabek for laboratory management; S. Smith for antibody preparation; A. Terskikh for provision of β-actin/HcRed transgenic mice; L. Hidalgo, J. Dollaga, and D. Escoto for animal care; J. Mulholland for assistance with confocal microscopy; and I. Conboy and T. Rando for helpful discussions and critical reading of the manuscript. This work was supported in part through NIH grant CA86065 to I.L.W.; the VPUE Faculty Grant to R.I.S. and I.L.W.; NIH Training Grant in Molecular and Cellular Immunobiology, 5T32A I07290-16, to J.L.C.; American Cancer Society Grant PF-00-017-01-LBC to A.J.W.; and the Frederick Frank/Lehman Brothers, Inc.—Irvington Institute Fellowship to A.J.W.
Richard I. Sherwood and Julie L. Christensen contributed equally to this work.