Adult mesenchymal stem cells (MSCs) isolated from bone marrow and mitotically expanded in culture are able to differentiate into a number of mesenchymal phenotypes, including those that form bone, cartilage, muscle, fat and other connective tissues (Figure 1) 1–3. These observed events led me to suggest that MSCs were responsible for the normal turnover and maintenance of adult mesenchymal tissues 3. This suggestion was based on the fact that all cells have half-lives and their natural expiration must be matched by their replacement, with the MSCs as the suggested source of these new replacement cells. This replacement logic was an extension from the known sequence of events involved in the turnover and maintenance of blood cells that are formed from haematopoietic stem cells (HSCs) 4, 5. The original mesengenic process pathway (Figure 1) was fashioned after the haematopoiesis lineage diagrams of the 1980s. Indeed, we knew that osteoblasts and chondrocytes were derived in lineage progression pathways, as shown in studies from a variety of laboratories, including our own 6, 7.
The replacement and turnover role for MSCs was also supported by the very crude data in Figure 28. We would seed 10–20 × 106 cells from the light-cell fraction of a bone marrow aspirate placed on a Percoll gradient into a 100 mm Petri dish, discard the unattached cells on day 3 and by days 10–12 colonies of fibroblastic cells (CFU-F) could be counted 9–11. When grouped by decade, a dramatic decrease in MSCs per nucleated marrow cell could be observed, with a 10-fold decrease from birth to teens and another 10-fold decrease from teens to the elderly. These decreases paralleled the observed fracture healing rates: very rapid in the young and very slow in older patients. In comparison, the titres of HSCs in marrow remain constant throughout life at about one per 104 nucleated marrow cells.
Bone marrow and the ceramic cube
At the very minimum, it has been shown that marrow MSCs contribute to the repair blastema (ie callus) formed at sites of bone breaks, and that both bone and cartilage forms at these sites, depending on its mechanical status—mechanically stable breaks form bone, unstable sites form cartilage that is endochondrally replaced 12. Furthermore, culture-expanded human or animal MSCs from marrow that are seeded into the interstices of a porous calcium phosphate ceramic cube that is then implanted into a subcutaneous site of immuno-incompetent hosts form bone and/or cartilage, depending on the proximity of vasculature: the presence of vasculature orients osteoblasts to secrete osteoid onto the walls of the pores of the ceramic, while the absence of vasculature promotes cell replication, aggregation and subsequent chondrogenic differentiation 13–16. Of direct importance are our early observations that marked MSCs in ceramic cubes that formed bone exhibited monolayers of marked secretory osteoblasts on newly formed bone that had labelled osteocytes within the calcified matrix 14, 15. Since human osteoblasts have half-lives of 8–10 days, the labelled cells were replaced after a few weeks by host-derived osteoblasts (and subsequently labelled osteocytes) that continued the bone formation process, eventually leading to vascularized marrow cavities of purely host origin 14, 15. These observations strengthen the supposition that MSCs naturally function to provide replacement cells for those that expire in highly differentiated tissues such as bone. Likewise, these observations document that the host vasculature brings in host MSCs that give rise to secretory osteoblasts that fabricate the segments of host-derived bone in the pores of the ceramic.
The other facts
Inconsistent with the idea that the key function of MSCs was to supply replacement parts for mesenchymal tissue were the observations that human marrow-derived MSCs could be activated to secrete cytokines/growth factors that inhibited an in vitro mixed lymphocyte assay 17, 18. These observations were used to suggest that MSCs could be used therapeutically as allogeneic cells, a so-called ‘universal cell’, a cell that could function in any host. To support this suggestion, it was further documented that culture-expanded MSCs did not have MHC class II cell surface markers, but rather only MHC class I and no co-stimulator molecules 19. Thus, human MSCs could not be antigen-presenting cells and would be invisible to the host's immune system 20, 21.
Importantly, we first used human marrow-derived MSCs to supplement bone marrow transplantations 22–25 because we assumed that MSCs homed back to the bone marrow 26 and that the reconstituted stroma would more efficiently re-establish the microenvironment for HSC engraftment 27 and subsequent lineage translation of their progeny 25, 26. Our early successful clinical efforts for both safety and efficiency supported the idea that culture-expanded MSCs were indeed capable of promoting successful engraftment of haematopoietic progenitors and their efficient production of circulating mature blood cells 22–24. In retrospect, it may be that haematopoietic stem cell engraftment is enhanced by the constitutive secretion of various bioactive agents by the MSCs 28, not the engraftment and stromal differentiation of the transplanted MSCs.
Again, based on our bone marrow homing hypothesis and the fact MSCs were able to differentiate into osteoblasts, we and others embarked on clinical efforts to cure gene defects by allogeneic transplantations, using normal marrows that did not exhibit the genetic defect 23, 25, 29. In these cases, culture-expanded MSCs from the allo-donor were used to supplement the bone marrow transplantation, with the assumption that the MSCs homed to marrow and re-established the stroma, to enhance allogeneic engraftment and to ensure that ‘normal’ MSCs in the marrow could differentiate into normal differentiated mesenchymal phenotypes. A clear example of these efforts is documented in the publications of Horwitz and colleagues, in which they describe their attempts to cure osteogenesis imperfecta (OI) by allogeneic bone marrow transplantation, including a ‘booster’ shot of MSCs intravenously 18 months after the original allogeneic bone marrow transplantation 29, 30. The results in six children with OI were mixed, with a few regaining impressive skeletal growth rates while others showed no positive effects. The question was: did the MSCs engraft? If yes, how many and where? And, how long did the MSCs remain active and did they form osteoblasts to cure the bone problems of these afflicted patients? Or did the MSCs supply bioactive factors that directly or indirectly enhanced the growth rates?
It is important to note that during these early stages of MSC commercialization and bone marrow transplantation, clinical trials with MSCs by Mosca and his colleagues at Osiris Therapeutics Inc. documented that MSCs secreted bioactive molecules with immunomodulatory effects 31. Further, in all clinical use by ourselves or others of human adult marrow-derived, culture-expanded MSCs, whether autologous or allogeneic, no adverse events were recorded. This established that isolation and culture expansion is safe and, indeed, clinical benefit from the intravenous delivery of hMSCs could be observed 24 (Osiris Therapeutics Inc. website: http://www.osiris.com).
The new era
Based on its expanded dataset and the detailed measurements of others, Osiris Therapeutics Inc. and a few independent investigators, especially LeBlanc et al32, used culture-expanded human MSCs and their immunomodulatory capacity to combat steroid-resistant graft-versus-host-disease (GvHD). These results were quite striking in that the MSCs very effectively shut down the GvHD processes. The most striking observation comes from the compassionate use of adult marrow-derived culture-expanded allogeneic MSCs by Osiris Therapeutics in 12 children, aged 5 months to 15 years, who were considered completely resistant to any therapies for GvHD and who were expected to die. These children were given 3–21 intravenous infusions of hMSCs. As reported on the Osiris website, 7/12 had complete remission of GvHD at 1 month and 95% were alive at 6 months compared to a maximum survival of 25% for patients given additional immunosuppressive treatments. Also of importance is the fact that 9/12 had complete recovery from their gastrointestinal (GI) GvHD and the remaining three had their severity reduced to Grade 1 GI GvHD. Based on these last observations and others, Osiris has embarked on a clinical trial to treat Crohn's (inflammatory bowel) disease with hMSC infusions and have documented an impressive, dose-dependent improvement in the clinical Crohn's score.
Quite striking is the fact that, to date, in all of the infusions of allogeneic MSCs, even the multiple infusions discussed above, no adverse events have been reported. The question can be raised as to why, if MSCs can differentiate into osteoblasts, have there not been reports of ectopic osteogenesis in the infusion sites or in lung and liver, where we know infused cells first dock 33? Perhaps the answer is that the natural function of MSCs is not to provide replacement units for the turnover of mesenchymal tissues such as bone, and their primary and most important function is to inhibit immunosurveillance and to establish a regenerative microenvironment!
The trophic effects of MSCs
We long ago understood that MSCs in culture secrete a variety of molecules, both bioactive and extracellular matrix 27, 28. In 1996 we published our study of the hMSC secretion of molecules into the medium in log-growth, as they entered osteogenesis and as they entered stromagenesis 28. In Figure 3, I summarize these published observations; the interpretation is that each vertical column documents a secretory signature of the activity state of the cells of each pathway in a 24-h period. We noted that the constitutive secretory quantities varied considerably between individual donors, but that percentage change as hMSCs went from growth to osteogenesis or stromagenesis was tightly clustered. Obvious from our ELISA quantitation of these bioactive molecules was that substantial levels of these molecules were fabricated every 24 h. This turns out to be the key observation and we now understand that one of the primary and substantial functions of MSCs is to secrete large quantities of bioactive molecules in response to their local environment and their activity status.
The assumption that MSCs could differentiate into any number of mesenchymal phenotypes led others to the assumption that MSCs could differentiate into cardiac myocytes 34, 35. Such cardiac myocytes could be therapeutically useful in sites of cardiac ischaemia (infarct) where a large number of cardiac myocytes expire (apoptosis) due to the ischaemia; the replacement of these expired cells would be expected to provide clinical benefit. This would be especially useful if the replacement cells were allogeneic and an ‘off-the-shelf’ product could be available for the treatment of acute myocardial infarcts. Studies using skeletal myoblasts in experimental animals seem to support this supposition 35, 36. Moreover, we had long before shown that normal marrow MSCs could differentiate into myoblasts and integrate into functional myotubes of MDX mice and ‘cure’ those genetically defective myotubes by producing functional dystrophin that was dispersed throughout the length of the myotubes 37, 38. In experiments in mice, rats and pigs, Osiris Therapeutics and others showed that intravenous infusion of culture-expanded allogeneic MSCs could have significant and positive benefits in myocardial infarct models 39. The signalling of the ischaemic tissue to the marrow MSCs, as recently reported, involves a number of secreted factors, including SDF-1 40, MCP-3A 41 and others. Based on these positive preclinical results, Osiris Therapeutics has conducted an FDA-approved Phase I clinical trial whose results include a four-fold reduction in arrhythmias, fewer premature ventricular contractions at all time points studied and prompt return of heart rate after a 6 min walk, and the unexpected result was a substantial improvement in lung function. Presumably, infarct patients have other organ ailments and the infused MSCs seem to go to these injured or affected tissues, such as lung, and provide therapeutic improvement. It seems clear that the infused MSCs do not exert their therapeutic effects by differentiating into cardiac myocytes.
In addition, in a goat model of severe osteoarthritis, MSCs injected into knee joints in a hyaluronan delivery vehicle showed the regeneration of the surgically amputated meniscus 42. Last, Michael Chopp and others have reported that both direct injection or intravenous infusion of human marrow-derived and culture-expanded MSCs into rodent stroke models 1 week after interrupting blood flow to the brain results in the recovery of coordinate function by 6–8 weeks 43, 44.
In all of the above cases, it appears that MSCs do not differentiate into cardiac myocytes, meniscal cells or neurons. Rather, the MSCs exert their influence by the secretion of massive amounts of growth factors and cytokines to effect a therapeutic outcome. We call these effects ‘trophic’ activity 45, named after this term that was related to neural effects 46. I interpret all of these models to have a common trophic mode of action of the MSCs: the MSCs secrete bioactive molecules that: (a) inhibit apoptosis and limit the field of damage or injury; (b) inhibit fibrosis or scarring at sites of injury; (c) stimulate angiogenesis and bring in a new blood supply; and (d) stimulate the mitosis of tissue-specific and tissue-intrinsic progenitors, such as cardiac or neural stem cells 45. The infused MSCs also secrete immunomodulatory agents that turn off T cell surveillance and chronic inflammatory processes; thus, allogeneic or, in the case of rodent models, xenogenic hMSCs can be therapeutically effective.
The MSC niche: the pericyte
In the early era of MSC technology, we tried to identify the MSC niche. For example, we showed that the hMSC antibody SH-2 was associated with blood vessels in skin of very young, but not older, patients 47. However, we never followed up on these preliminary observations, although sections of human bone marrow showed rare SH-2-positive cells. The key question was never addressed: where was the MSC normally located and what was its natural function at specific anatomical sites? With recent investigations into aspects of the control of vasculature in the human dermis 48, 49 and the substantial observations of others 50–52, we now understand that MSCs can function as vascular pericytes 53. We have recently proposed that the pericyte is released from its position on a vascular tube in the case of a focal injury and, as such 53, it functions as an immunomodulatory and trophic MSC. This immune modulation turns off T cell surveillance of the injured tissue and thus provides a block to autoimmune reactions, while its trophic activity ensures that the field of damage is limited, that scarring does not occur and that tissue-intrinsic progenitors replace the expired cells. Angiogenesis is brought about by MSC secretion of bioactive factors, such as VEGF, and by the stabilization of newly forming vessels by the repositioning of the MSC into its pericyte role as a perivascular cell 49, 54. Thus, the activated pericyte that is displaced from its natural position on blood vessels functions as a secretory MSC that is both trophic and immunoregulatory.
Further, we noted (Figure 2) that the quantity of MSCs (CFU-F) per nucleated marrow cell decreases with age from one MSC in 104 marrow cells in newborns, to one MSC in 105 cells in teenagers, to one MSC in 106 cells in older individuals. This dramatic decrease also mirrors the decrease in vascular density with age and strongly argues that MSCs are in coordinate contact with blood vessels. The factors that govern vascular density with age are not known.
The new era of cell-mediated therapy
Given the above, we have reconsidered our MSC logic for their therapeutic use. Surely, MSCs can still be of great value for use in tissue engineering therapies by virtue of their ability to differentiate into distinctive and specialized cells. The use of inductive or instructive delivery vehicles, or the jump-starting of MSCs down specific lineage pathways, would seem to be necessary.
However, the use of MSCs as site-regulated multi-drug dispensers, especially allogeneic MSCs, opens avenues of therapy unimagined 5–10 years ago. For example, uses of MSCs in asthma 55, radiation exposure 56, neurological disorders 57, etc. are now being explored. The next technical hurdle is how to deliver a precise dose of effective MSCs to their site of action. Indeed, where is their main site of action? Is it the ischaemic heart or brain, is it in a central or systemic location, or is it the understudied lymph system? We have recently described a cell targeting technology for such site-directed therapeutics 58. The issue is: where do we target these therapeutic cells?
Also clear is that since all blood vessels are not the same, all pericytes are not the same. Thus, the MSCs from fat must be different from the MSCs of marrow with regard to the details that govern their responses and activities; some publications now document these differences 59, 60. Are the MSCs from marrow ‘better’ than the MSCs from fat, muscle or liver for a specific site relative to trophic activity or immunomodulation? Only time and a great amount of careful, quantitative data will help us answer these questions.
Supported by the David and Virginia Baldwin Fund and by grants from NIH.
Power Point slides of the figures from this Review may be found in the supporting information