Introduction
Acute myocardial infarction (AMI) and the resulting complications are a leading cause of morbidity and mortality in the Western world. While conventional treatment strategies for AMI may efficiently alleviate symptoms and hinder disease progression, recovery of lost cells and tissue is rarely achievable. Transplantation of primitive progenitor cells of hematopoietic, mesenchymal, and endothelial lineages have, however, been found to enhance endogenous tissue repair in small animal disease models and to improve overall function of the affected tissues in early phase clinical trials[
1]. The exact mechanism of repair is not known but may involve paracrine signaling by the donor cells or direct replacement of damaged tissue by donor cells[
2].
Stem and progenitor cells derived from hematopoietic tissue have attracted much attention as a source of transplantable cells for cell-based regenerative therapy. Hematopoietic, mesenchymal, and endothelial progenitors have been identified in human bone marrow (BM) and umbilical cord blood (UCB) [
3‐
5]. All three progenitor populations can be simultaneously isolated from human BM based on the expression of the cytosolic enzyme aldehyde dehydrogenase (ALDH)[
6], although the relative contributions of the different sub-populations and consequently their relative therapeutic contribution may vary between the different cell sources. We and others have found that lineage depleted (Lin
-) cells from BM and UCB that express high levels of ALDH (ALDH
hiLin) have superior long term repopulating potential in the hematopoietic tissues of NOD/LtSz-scid/scid (NOD/SCID) mice whereas lineage depleted cells that express low levels of ALDH (ALDH
loLin
-) are virtually devoid of long term repopulating potential in spite of an apparent overlap in expression of the putative human hematopoietic stem cell marker CD34 between the two populations [
7‐
10]. Furthermore, as few as 2 × 10
5 ALDH
hiLin
- cells purified from UCB can engraft multiple tissues in the β-glucuronidase (GUSB) deficient NOD/SCID/MPSVII mouse model, including the pancreas, retina, lung, liver, kidney and heart at 10-12 weeks post transplantation[
11].
Xenotransplantation of human hematopoietic stem cells and progenitor cells to immune deficient mice is extensively used to study human hematopoiesis and diseases involving the hematopoietic system[
12]. The studies of diseases of solid organs using xenotransplantation models is, however, hampered by the lack of simple and sensitive methods for identifying human donor cells, an issue which we addressed in the current studies. We adapted the left anterior descending (LAD) coronary artery occlusion model of AMI recently described by van Laake et al[
13] to highly immune deficient NOD/SCID and NOD/SCID β2-microglobulin null mice (NOD/SCID β2m null). The NOD/SCID β2m null mouse strain is deficient in the expression of the MHC class I associated cell surface protein β2-microglubulin (β2m), which is normally expressed on all nucleated cells[
14]. Engrafting donor cells can thus easily be detected by immune staining for β2m.
Macroscopic evaluation of donor cell distribution to various organs following global or localized delivery is key to understanding the dynamics of stem cell engraftment in target tissues and has been described using labeling with radionuclides, fluorescent dyes, or bioluminescent or fluorescent reporter proteins[
15,
16]. We have recently documented that engrafting human donor cells can be visualized in situ without adversely affecting cell viability and engraftment potential by a combination of nanoparticle labeling and whole organ fluorescent imaging[
17]. Using a similar approach, we have in the present study: 1) evaluated donor cell distribution to multiple organs, including the infarcted heart, at 48-72 hours post transplantation and 2) analyzed long term engraftment in multiple organs and the infarct zone as well as the regenerative effects of cell treatment by molecular and mechanistic approaches at four weeks post transplantation. By the combined nanoparticle labeling and whole organ fluorescent imaging, we found a more pronounced infarct-specific distribution of ALDH
hiLin
- stem cells, as compared to committed progenitor cells at 48-72 hours post transplantation. At four weeks post transplantation, ALDH
hiLin
- cells engrafted multiple organs, including the heart, liver and kidney, at higher frequencies than ALDH
loLin
- cells. Under these highly permissive conditions for human cell engraftment, we found no donor derived cardiomyocytes and only few endothelial cells of donor origin at four weeks. Cell treatment was not associated with a significant improvement in cardiac performance at four weeks. There was, however, a significant increase in the vascular density of large caliber vessels in the central infarct zone of ALDH
hiLin
- cell-treated mice, as compared to PBS and ALDH
loLin
- cell-treated animals.
Discussion
In the current studies we have adapted the LAD occlusion model of AMI to immune deficient NOD/SCID and NOD/SCID β2m null mice. We used this model to evaluate the global engraftment potential of purified human UCB cell populations as well as the distribution, engraftment, and regenerative potential for the infarcted heart.
We first used fluorescent nanoparticle labeling to trace the donor cell distribution to various organs, including the infarcted myocardium, following IV injection. We have recently documented that sorting of the labeled cells is essential to avoid infusing large numbers of unbound nanoparticles[
17]. Non-cell mediated splenic sequestering of fluorescent nanoparticles was indeed pronounced in our previous report when control NOD/SCID β2m null mice received free 750 nm fluorescently conjugated Feridex nanoparticles[
17]. The fluorescent intensities found in the NOD/SCID mice transplanted with QD655 labeled cells in the present study may thus include both cell specific and unspecific non-cell mediated fluorescence. Our present results from animals transplanted with 750 nm Feridex labeled cells sorted prior to infusion, however, confirm a significant distribution of labeled donor cells to the infarcted tissue in the absence of nonspecific signal from free nanoparticles. We have previously found a labeling efficiency between 28% and 40% with fluorescently conjugated Feridex nanoparticles, depending of the purification method[
17]. Specifically, the Feridex labeling efficiency of UCB CD34
+ purified cells was approximately 32% while Lin
- purified UCB cell labeled at approximately 39%. Although we did not measured the QD655 and Feridex nanoparticle labeling efficiency of the Lin
-ALDH
hi and Lin
-ALDH
lo purified cells used in the present study, we expect that differential labeling efficiency is not responsible for the observed difference in signal intensity. Although we were clearly able to visualize a specific trafficking of ALDH
hiLin
- cell to the site of injury, we were unable to image the organs non-invasively thus precluding a longitudinal evaluation of donor cell distribution. Using a similar cell sorting and labeling strategy we, however, recently demonstrated that donor cells could be detected in the ischemic hind limb up to seven days after transplantation[
17]. The difference in sensitivity between our previous study and the present one is likely due to interference from the additional overlying tissue of the thoracic cavity and localized transplantation and/or labeling with fluorescent nanoparticles emitting in the far red range may be needed in order to improve tissue penetration and allow non-invasive visualization of labeled cells in situ[
17]. Also, the electron-dense properties of the fluorescent nanoparticles presently employed potentially allow for multimodal non-invasive visualization of labeled cells using both fluorescent and magnetic resonance imaging[
17]. We have also recently worked with perfluorocarbon nanobeacons, which have a higher emission and penetrance without background and might be better suited for in vivo imaging of deep tissues[
17].
Both the NOD/SCID and the NOD/SCID β2m null strains presently used are known to support multi-lineage engraftment of human hematopoietic cells. Identification of engrafting human cells in solid organs is, however, difficult and requires labeling of donor cells prior to transplantation by ex vivo manipulation of target cells prior to transplantation or by application of complex immunoassay techniques. Extensive ex vivo manipulation of the donor cells is undesirable and may adversely affect the cells and increase the risk of contamination while antibody staining for specific human lineage markers typically requires knowledge of the expected differentiation pattern of the transplanted cells, so unexpected cell phenotypes may go unnoticed. Antibody staining for β2m is, on the other hand, quick and versatile, and requires no ex vivo manipulation of the donor cell. Moreover, no nonspecific staining of endogenous β2m is seen in NOD/SCID β2m null strain and donor derived cells are detected regardless of post transplantation phenotypic fate. A drawback of the β2m staining approach relates to the possible down regulation of β2m expression by some types of cancer cells as a mechanism to avoid normal host cancer surveillance[
25]. Although we are not aware of any literature describing a similar down regulation of β2m expression by non-carcinogenic cells in the setting of xenogeneic transplantation, we cannot exclude the fact that we may underestimate the number of engrafting human cells by this method. To compensate for this shortcoming and to confirm the human specificity of our β2m staining, we employed human specific lineage specific antibodies throughout the study. Alternatively, we have also recently described an alternative murine xenograft model based on the β-glucuronidase (GUSB) deficient NOD/SCID/MPSVII mouse strain[
17,
23]. The lack of GUSB expression by the host tissue similarly allows rapid and precise identification of engrafting human cells by staining for donor GUSB activity. Using the NOD/SCID/MPSVII model, we demonstrated multi-organ engraftment of human UCB-derived ALDH
hiLin
- cells 10-12 weeks post transplantation[
11]. Both the present model and the NOD/SCID/MPSVII model are thus ideally suited for pre-clinical evaluation of prospective cell populations and application strategies in cell-based regenerative therapy.
We and others have previously shown that ALDH
hiLin
- cells have a superior hematopoietic repopulating potential in the BM and spleen of NOD/SCID and NOD/SCID β2m null mice, as compared to CD34
+ or ALDH
loLin
- cells [
7‐
10]. ALDH
loLin
- cells are, as verified in the present study, indeed virtually devoid of long term repopulation potential. In addition, we have recently shown that ALDH
hiLin
- sorted cells from human BM contained populations of functionally primitive mesenchymal progenitor populations[
26]. UCB, as used in the present study, is, however, known to contain lower numbers of mesenchymal progenitors in comparison to BM[
17]. We cultured the cells overnight under conditions that promote retention of primitive hematopoietic phenotypes[
17]. The present AMI xenotransplantation study thus predominantly reflects the regenerative potential of highly purified hematopoietic stem and progenitor cells. Gentry et al. have previously shown that ALDH
hi sorted cells contain subsets of primitive stem and progenitor cells of non-hematopoietic lineages, including mesenchymal stem cells and endothelial progenitor cells[
6]. Although we did not assess the proportion of these non-hematopoietic cells in the present study, due to the cell source and isolation and culture method, it is unlikely that they contributed to the observed results in a substantial way. We found no evidence of a direct contribution of the transplanted cells to regenerated infarcted tissue although down regulation of β2m expression by the donor cells as discussed above may have rendered some donor-derived cells types undetectable by our present methods. Engrafting human cells were predominantly of a hematopoietic phenotype, although non-hematopoietic cells were also identified. These CD45 negative cells rarely appeared in the infarcted tissue and it is therefore unlikely that they represent primitive cardiomyocytes. We were unable to precisely determine if the engrafting cells were tissue resident cells or circulating hematopoietic cells retained in the microvasculature. Although none of the donor cells appeared to reside in large caliber vessels we did, however not analyze peripheral blood samples to confirm the presence of a circulating pool of donor derived cells. Moreover, although we recently reported that fusion of human donor UCB ALDH
hiLin
- cells and host murine hepatocytes could generate hybrid cells that only retained minimal amounts of human DNA in a NOD/SCID/MPSVII liver injury model, this was indeed a very rare event[
23]. The present results are thus more in line with our previous results and recent reports on the role of donor hematopoietic cells in the regeneration of damaged tissue[
17,
26‐
28]. In a recent study we also failed to detect any long term human myocardial engraftment or functional improvement following intramyocardial injection of human CD34
+ sorted mobilized peripheral blood progenitors in athymic nude rats with AMI[
29]. In the present study we were similarly unable to detect an improvement in cardiac function as a result of cell treatment in either the ALDH
loLin
- or ALDH
hiLin
- treated groups. We did, however detect a significantly better vascularization of the central infarct area in the ALDH
hiLin
- treated group as compared to the ALDH
loLin
- and PBS treated groups. The fact that the ALDH
loLin
- cells also appeared to improve vascular density compared to PBS when correcting for outliers suggested that this population, although devoid of long term repopulating cells, may include a transiently present population of cells with angiogenic potential.
The most well described larger randomized clinical study of cell-based regenerative therapy for AMI reports a modest 2.5% increase in left ventricular EF following intra-coronary infusion of BM MNCs[
30]. We were unable to detect an improvement in cardiac function as a result of cell treatment in either the ALDH
loLin
- or ALDH
hiLin
- treated groups. It should, however, be noted that the study was not powered to detect small improvements in cardiac function and modest improvements as reported in clinical trials would thus go unnoticed in the present study. The fact that we found a superior vascularization in the ALDH
hiLin
- treated group but no improvement in cardiac function may indeed be due to the relatively large variation in the echocardiographic data. The lack of a detectable functional improvement can, alternatively, be explained by the early end point of functional evaluation. It is indeed at this point not clear whether the vascular structures that we detected in the central infarct area are patent and thus represent mature and functional blood vessels. These questions may be resolved in future studies by both including a more direct measure of blood flow to the infracted area as well as extending the evaluation period to eight weeks and beyond. Nonetheless, a long term benefit is not likely to depend on a direct contribution of the transplanted cells to the regenerating myocardium, since we found no evidence of a substantial donor derived population in the central infarct area or in the blood vessels. These results are in agreement with our recent findings that human BM derived ALDH
hiLin
- cells improve perfusion to the ischemic hind limb of NOD/SCID β2m null mice and improve vascular density as compared to ALDH
loLin
- or MNC control treated mice[
26]. Moreover, using a similar labeling strategy as the one employed in the present study, we found that the human donor cells only transiently engrafted the ischemic tissue. Only few cells were detected at 21 to 28 days post transplant in animals receiving ALDH
hiLin
- cells while animals receiving ALDH
loLin
- cells were devoid of engrafting donor cells at the endpoint. Although there are obvious differences with respect to the cell source and the details of the purification protocols employed in our hind limb ischemia study and the present study, the lack of long term engraftment of the ALDH
loLin
- cells as shown in the hind limb ischemia model is corroborated by the present immunofluorescent and PCR data. The sensitivity of our PCR assay may however allow for a non-detected low level of engraftment to persist although we have previously been able to detect ~2 human cells per 10.000 murine cells in a related PCR system[
31]. Even though we found similar results using UCB and BM in the present cardiac infarction models and in our previously reported hind limb ischemia model, respectively, in a direct comparison of BM and UCB derived human CD133
+ purified cells, Ma
et al found that only BM derived cells induced functional recovery as measured by improved shortening fraction at four weeks post intramyocardial transplantation of 5 × 10
5 human donor cells in a NOD/SCID cryo-injury model of AMI[
32]. Interestingly, in spite of the significant difference in functional recovery between UCB and BM treated animals, no difference was observed in infarct size and capillary density between the two cell treatment groups.
In conclusion, we found that a larger proportion of human UCB cells selected according to high expression of the cytosolic enzyme aldehyde dehydrogenase specifically distributed to the infarcted tissue as compared to cells with low aldehyde dehydrogenase activity. ALDHhiLin- cells also had a superior global engraftment potential in multiple organs including the infarcted heart at four weeks post transplantation. Although no significant improvement in cardiac performance was detected at four weeks post transplantation, the superior engraftment potential was associated with an increased vessel density in the infarct zone, as compared to controls. The significant increase in vessel density in the stem cell-injected mice, as compared to the injured but non-transplanted, or committed progenitor - transplanted controls, is interesting, and the mechanism responsible is not yet known. The increased density of large-caliber vessels could be caused by an enlargement in size and function of pre-existing tiny vessels, or could be caused by neovascularization into the infarct zone. Future studies will examine those possibilities.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CSS and DH conceived of the study and carried out its design and coordination. DM and DPW were responsible for imaging studies. CW performed LAD ligation to promote cardiac injury. IR assisted in stem cell isolation and Flow cytometry. MC provided umbilical cord blood samples discarded form the St. Louis cord blood bank and reviewed data. AK performed functional cardiology studies in the murine recipients of the human stem cells. LP and JAN funded the study, approved of its design, reviewed and interpreted the data. CSS and JAN wrote the manuscript and performed editorial revisions. All authors read and approved the manuscript.