Bone marrow-derived stem cells appear surprisingly multipotent in experimental settings, but the physiological significance of such plasticity is unclear. We have used sex-mismatched cattle twins with stably chimeric hematopoietic systems to investigate the general extent of integration of circulating cells to the nonhematopoietic cell lineages in an unmanipulated large mammal. The donor-derived (Y+) nonhematopoietic cells in female recipient tissues were visualized by Y-chromosome specific in situ hybridization combined with pan-leukocyte labeling. Y+ leukocytes were frequent in all tissues, but in 11 of 12 animals, average contribution to nonhematopoietic lineages was in any tissue below 1% (in brain <0.001%). Significantly higher integration rate was detected in regenerating granulation tissue. Also, one animal showed a high frequency of nonhematopoietic Y+ cells in several tissues, including intestinal epithelium and mammary gland stroma. In conclusion, circulating cells do not appear significant in the development and maintenance of nonhematopoietic bovine tissues, but may be important in regeneration and other special conditions.
The differentiation potential of stem cells (SCs) in adult mammals is currently a topic of intense debate. Many recent reports have challenged the traditional dogma of strictly restricted differentiation options of adult cells. Bone marrow (BM) cells, hematopoietic stem cells (HSCs), or mesenchymal stem cells (MSCs) have been shown to differentiate into cells of the skeletal muscle [1, 2], liver [3, 4], epithelia of skin, lung, and gastrointestinal tract [5–7] and even to cells of the central nervous system (CNS) [8–10]. However, the physiological significance of the findings is uncertain. The cell culture techniques and transplantation methods applied are likely to induce changes both in the cells  and recipient tissues [12, 13]. Also, heterogeneity of the transplanted cell populations , potential cell fusion events [15, 16], and problems in reproducing some of the results [17, 18] complicate interpretation of transdifferentiation experiments.
Circulating cells derived from BM or other sources may have a role in tissue repair . A traditional model of regenerating tissue is a subcutaneously implanted cellulose viscose sponge . The sponge is invaded by irregular, richly vascularized granulation tissue similar to that forming in a healing wound. Granulocytes and macrophages first enter the sponge, followed by fibroblasts, and a rapidly increasing collagen synthesis and angiogenesis in a few days [20–22]. The origin of the granulation tissue producing cells is unclear, with several contradictory studies supporting either a local or systemic source [23, 24].
We have used freemartin cattle to monitor the contribution of prenatally circulating cells to the development, physiological maintenance, and repair of bovine tissues. Freemartin is a female calf born as a twin to a bull. About 90% of cattle twins are chimeric due to placental anastomoses that form early in the development and permit the exchange of blood and other cells between the fetuses [25, 26]. The twins permanently share identical composite blood types  and are mutually tolerant to allografts from each other [28, 29]. The proportions of donor-derived nucleated cells in blood and BM of different animals are randomly distributed between less than 5% and more than 95% . The freemartins develop normally, apart from the inner sex organs, which are masculinized to a variable degree probably due to hormonal influences . Donor-derived cells in freemartin tissues can be readily identified by their Y chromosomes. Thus, the freemartin is a suitable model for studies of SC transfer in an unmanipulated large mammal, and also unique in allowing the investigation of prenatal events in a chimeric animal.
Freemartin tissues have formerly been analyzed by cytogenetic techniques , but little is known about the contribution of donor-derived cells to the development, growth, and cell dynamics of the non-hematopoietic organs. We have previously characterized the freemartin immune system by Y-chromosome-targeted in situ hybridization (Y-ISH) [32, 33], which allows the identification of bull-derived cells in intact tissue sections. Y-ISH can be combined with immunohistochemistry or other staining methods to investigate the phenotypes of cells. In this paper, we describe an extensive mapping of freemartin tissues using Y-ISH combined with CD45 immunohistochemistry and mistletoe lectin I (ML-I) histochemistry to exclude cells of the hematopoietic lineage. CD45 is a cell surface antigen exclusively expressed by all nucleated cells of the hematopoietic lineage . ML-I labels bovine microglia .
Materials and Methods
The calves used in this study were identified as freemartins by Y-ISH analysis of blood samples as described previously .
Twelve freemartin calves (ages: 20 days - 20 months) were anesthesized and exsanguinated. Samples were collected from most tissues throughout the body (about 70 different locations from most of the animals) and processed as described previously .
All animal experiments were approved by the local animal welfare authorities.
Cellulose Viscose Sponge Implants
Block-shaped Cellspon cellulose viscose sponges (Cellomeda; Turku, Finland; http://www.abo.fi/∼klonnqvi) measuring 5 × 10 × 20 mm were implanted subcutaneously in the prescapular region of a freemartin calf (age: 70 days at first implantation). Five sponges were implanted at one week intervals in alternating sides, one sponge at a time. Operations were performed under general anesthesia induced by intramuscular injections of ketamine and xylazine. The calf was sacrificed and implants recovered one week after the last operation. Granulation tissue in implants incubated for 1, 2, and 4 weeks was selected for double staining analysis.
The Y-chromosomal probe was a cocktail of 28mer oligonucleotides [5′-TT(A/C/T) TCA GCC CTG TGC C(C/T)T GG(A/C/G/T) (A/G)A(C/T) TGT G-3′] corresponding to the repetitive bovine btDYZ locus . As a control, a previously described probe (5′-TTT ACC TTA GAA CAA ACC GAG GCA C-3′)  recognizing the bovine autosomal 1.709 satellite sequence  was applied.
Genomic In Situ Hybridization
Genomic in situ hybridization (GISH) was performed as described previously . No false positive Y chromosome signals have been detected in the more than 1,000 samples from different tissues from normal cows hybridized. Thus, of several parallel stainings for each sample, those with highest Y+ frequency represent the most reliable estimate of the real frequency of bull-derived cells.
CD45 Immunohistochemistry and Mistletoe Lectin I Histochemistry
CD45 immunohistochemistry was performed using the anti-bovine CD45 antibody CACTB51A (VMRD; Pullman, WA; http://www.vmrd.com) and tyramide signal amplification. Briefly, 4 μm paraffin sections of ethanol-fixed tissues were permeabilized, digested with 10 μg/ml protease P6911 (Sigma; St. Louis, MO; http://www.sigmaaldrich.com) in 10 mM Tris-HCl pH 7.4, 0.5 mM EDTA for 30 min at 37°C where necessary for antigen retrieval, blocked, and incubated in the primary antibody overnight. The sections were incubated in a biotinylated antimouse Ig antibody (DAKO; Glostrup, Denmark; http://www.blackwell-synergy.com), and the tyramide signal amplification was performed using either the TSA Biotin System (NEN Life Science Products; Boston, MA; http://las.perkinelmer.com) essentially as recommended by manufacturer or using biotinylated tyramide  in combination with horseradish peroxidase avidin D (Vector Laboratories; Burlingame, CA; http://www.vectorlabs.com).
As the CD45 staining intensity of microglia was unsatisfactory, ML-I histochemistry was used instead for investigating the CNS. The lectin stains bovine microglia and other macrophage lineages as well as most endothelia . Paraffin sections of paraformaldehyde-fixed tissues were stained using biotinylated ML-I (obtained from Institut für Phytochemie; Witten/Herdecke University; Witten, Germany; http://www.uni-wh.de) and the tyramide signal amplification method as described above.
In combined GISH and CD45 immunohistochemistry, the CD45 detection was performed first to avoid antigen degradation by the GISH procedure. After the tyramide reaction, GISH was performed, and following that, the immunohistochemistry was finished. Diaminobenzidine (Vector Laboratories) was used as the peroxidase substrate. Sections were counterstained with Mayer's hematoxylin and mounted with Faramount (DAKO).
In combined GISH and lectin histochemistry, the lectin staining was performed after GISH, as the GISH procedure was found to function as an effective pretreatment for the lectin staining.
Double stained tissue sections with most Y chromosome positive (Y+) cells (and strong leukocyte staining) were selected for cell counting from sets of parallel stainings with slightly varying prehybridization treatments, as no false positive signals have been detected in Y-ISH (see genomic in situ hybridization above). Each tissue was counted from five different animals (one section per tissue per animal; not the same animals for different tissues) with the exception of the mammary gland, with samples containing ducts available from four animals, and the regenerating tissue in implants, with one sample per time point available.
Tissue sections were examined in an Olympus BH2 microscope equipped with a ColorView 12 digital camera (Soft Imaging System; Münster, Germany; http://www.soft-imaging.de). Cells were counted using the touch count tool of AnalySIS 3.0 image analysis software (Soft Imaging System). All relevant CD45 (or mistletoe lectin I in the brain) negative cells in a field of 230 μm × 185 μm (photographed with a 40× objective, or with a 20× objective in the brain) were counted, and the process was repeated for additional fields until the cumulative percentage of Y+ cells no longer fluctuated more than 0.2 percentage units (10–25 fields, representing 345–2,880 total cells per sample). If no non-leukocyte Y+ cells were found, the whole section was viewed and the total number of screened cells (approximately 4,700–110,000 cells per sample) was estimated by measuring the area viewed.
Numbers were corrected with counts obtained from similar bull tissues: cerebrum 45% Y+ (used for cerebrum and hippocampus), olfactory bulb (OB) 66%, dermis (used for dermis, mammary gland, connective tissue, and granulation tissue) 42%, epidermis 37%, renal epithelia 58%, liver 69%, salivary gland epithelia (used for mammary gland epithelium) 58%, cardiac muscle 60%, skeletal muscle 57%, smooth muscle 62%, thyroid gland epithelium 52%, intestinal epithelium 66%. All statistical analyses were performed in SPSS 10.0 for Windows. Mann-Whitney U test was used for estimating statistical significance of differences in cell frequencies. Similarity of the implanted freemartin to the others was analysed by comparing scatterplots of cell counting results from different animals in each tissue.
We first performed a wide Y-ISH screening of tissues from 12 freemartins (ages: 20 days to 20 months). Samples from 20 to 74 different body sites per animal were analyzed. Y-chromosome positive (Y+) bull-derived cells were detected in all tissues examined in all 12 animals. The highest frequencies of Y+ cells were found in blood, BM, and lymphatic tissues. These tissues also exhibited considerable variation between individual animals. Among blood mononuclear cells, the proportion of Y+ cells varied from approximately 10% to 90%. In other organs, the variation was less pronounced.
The distribution of Y+ cells suggested that a significant proportion of them were infiltrating leukocytes, which often are difficult to distinguish reliably from parenchymal cells in a standard histological staining. To label the leukocytes residing in the tissues, we combined Y-ISH and CD45 immunohistochemistry or, in the case of CNS, ML-I histochemistry for bovine microglial cells . We re-examined all major tissue types and derivatives of all embryonic layers. The double staining confirmed that most Y+ cells indeed were of the hematopoietic lineage (Fig. 1A–1G). CD45/ML-I−Y+ cells were identified in most tissues, but in 11 of 12 calves (208 succesfully double-stained sections from 14 different tissues) they were mostly sporadic. No groups of CD45/ML-I−Y+ cells indicating local clonal expansion were seen in these calves. However, one of the calves (FM10) was a striking exception, with high numbers of CD45/ML-I−Y+ cells in several tissues. This case is discussed more thoroughly below.
The donor contribution to the nonhematopoietic cells in various tissues was estimated by counting the proportion of CD45/ML-I−Y+ cells in 4 to 5 calves per tissue (including the exceptional case). Altogether, several tissues were analyzed from all 12 freemartins. Apart from FM10, the calves did not differ markedly in the relative frequency of CD45/ML-I−Y+ cells, which constituted less than 1% of all CD45/ML-I− cells in most tissues investigated (Fig. 2). There was no obvious correlation between the frequency of Y+ cells in blood and in nonhematopoietic tissues, or between different nonhematopoietic tissues in an animal. Donor-derived nonhematopoietic cells were markedly rare in the ectoderm-derived epidermis and CNS. Otherwise, no consistent differences between derivatives of different embryonic layers were seen. In 11 of 12 calves, highest numbers of CD45/ML-I−Y+ cells were detected in the connective tissue around ducts of mammary glands (0.94 ± 0.67%), liver (0.85 ± 0.41%, primarily not hepatocytes), and skeletal muscle (0.83 ± 0.26%). In cardiac and smooth muscle, renal epithelia, and epidermis the average frequency of CD45−Y+ cells was 0.1% or less (in epidermis the interdigitating Langerhans cells make the identification of CD45− cells difficult). In the brain, very few ML-I− donor-derived cells were found, although donor contribution to the microglia was in most cases substantial (we note that in addition to microglial cells, the lectin labels most endothelia). In the cerebral cortex and the underlying white matter, 0.0020 ± 0.0023% of ML-I− cells were Y+. In the hippocampus and the OB, two regions involved in the generation of new neurons in adult mammals [40–44], no ML-I−Y+ cells were detected in approximately 2 × 105 cells screened in double-stained sections of each tissue. No Y+ cells with clearly neuronal morphology were seen in about 720 successfully hybridized sections from freemartin CNS, representing approximately 28 × 106 total cells and including samples from cerebral cortex, cerebellum, brain stem, corpus callosum, hippocampus, OB, thalamus, ventricular walls, hypophysis, and spinal cord.
In FM10, the distribution of CD45/ML-I−Y+ cells differed dramatically from the rest of the calves. Higher numbers of CD45−Y+ cells were found in the epithelium of small intestine (19%, Fig. 1H), connective tissue of mammary gland (15%), liver (4.3%), epithelium of mammary gland (4.0%), skeletal muscle (2.3%), cardiac muscle (1.0%), and smooth muscle (0.5%), whereas in brain and in renal epithelia the frequency did not differ notably from the other calves. In the villi of the intestinal mucosa, strands of CD45−Y+ cells were seen (Fig. 1H). The hematopoietic system of FM10 differed from the majority of calves in that the donor contribution to B cells was clearly weaker than to the total leukocyte pool (not shown).
To investigate the role of donor-derived cells in tissue regeneration, we implanted cellulose viscose sponges  in the subcutis of a 70-day-old freemartin and stained the invading granulation tissue for Y-chromosome and CD45. The regenerating tissue contained relatively high numbers of CD45−Y+ cells, with increasing numbers within the incubation period of 4 weeks (Figs. 1I, 2). At 4 weeks, the frequency of CD45−Y+ cells in the regenerating tissue was significantly higher than in most other nonhematopoietic tissues of the same animal (p = 0.12 for liver, p < 0.02 for any other tissue investigated). The frequencies of CD45/ML-I−Y+ cells in the tissues of the recipient did not differ markedly from those in the 10 other calves of the majority type. No clear Y+ cell clusters indicating local expansion were seen.
SC transfer experiments with manipulated or diseased recipients assess the limits of cell differentiation potential in special conditions. The results may be influenced by cell purification and in vitro cultivation methods , and in typical transplantation experiments by radiation damage to the recipient tissues, especially to various SC niches [12, 45]. The chimeric freemartin is a relevant model system for investigating the fate of circulating cells in an unmanipulated large mammal. The blood in the bovine twins is effectively mixed for the last 8 months of the 280-day gestation beginning at the 10–15 mm crown-rump stage [25, 46, 47], which corresponds to approximately 30–35 days p.c. [47, 48] or Carnegie stages 15–18 . As the vascular anastomosis forms before the establishment of the immunological self, the engraftment of the hematopoietic system by the sibling's circulating SCs is successful, and the recipient BM is permanently populated by donor cells . The progeny of donor cells in the freemartin tissues can be readily identified by Y-ISH. In this study, we examined the general extent of integration of donor-derived cells into the nonhematopoietic cell lineages.
We found high numbers of bull-derived cells in most freemartin tissues, such as intestinal mucosa, liver, and brain presented in Figure 1. However, in 11 of 12 calves most Y+ cells exhibited a leukocyte phenotype. The proportion of Y+ cells among all nonhematopoietic cells (negative for the pan-leukocyte marker CD45 or bovine microglial marker ML-I) in different tissues was in most cases clearly below 1% (Figs. 1 and 2). Notably, we saw very few nonhematopoietic Y+ cells in the brain, although donor-derived microglia were abundant. In the cerebral cortex, less than 0.005% of all nonhematopoietic cells were donor-derived; in the hippocampus or the OB, none were seen. These findings suggest a minor influence of circulating cells for the development, growth and physiological turnover of bovine nonhematopoietic tissues in normal conditions. As the oldest animals examined were more than 1 year in age, a significant contribution to the physiological turnover of tissues should have been detected.
In contrast to the apparently limited roles of circulating cells in normal physiological conditions, our results suggest that they or their derivatives may represent an important reserve for tissue regeneration, as also implied by several recent papers concerning plasticity of BM cells [1–3, 5]. The marked and increasing proportion of Y-chromosome positive CD45− cells in the cellulose viscose sponges (Figs. 1 and 2) suggests that donor-derived cells are recruited either directly from circulation or from dormant precursors in adjacent tissues, originating from a population with a higher proportion of Y+ cells than the intact subcutaneous tissue. No extensive proliferation of the invading cells probably occurs in the regenerating tissue, as no clear clusters of CD45−Y+ cells were seen. Cell migration may obscure such patterns, if proliferation is slow. We note that as the original population is unknown, it is difficult to accurately estimate the donor contribution to it. This complicates the estimation of the total impact of circulation-derived CD45− cells to tissue repair. Generally, the origin of granulation tissue producing cells is controversial. Numerous earlier studies provide evidence for a local origin [23, 50], while others suggest a systemic source . Peripheral blood cells expressing fibroblast markers vimentin, fibronectin, and collagen, as well as CD45 and the SC marker CD34 have been shown to mediate tissue repair [24, 52]. These are excluded in this study due to the CD45 expression, unless it is downregulated after extravasation.
Our results are in general agreement with data obtained from sex-mismatched BM, peripheral blood SC, and organ transplantations in humans [5, 53–56]. Low levels of nonhematopoietic cells derived from donor BM or blood are found in various tissues, and recipient cells integrate into organ transplants. The process is remarkably enhanced by tissue damage. However, the reported integration rates in transplant recipients are typically higher than seen in the unmanipulated freemartin tissues. This is most pronounced in the CNS, with markedly stronger engraftment reported in several BM transplant studies in mice [8, 10] and humans . These disparities may reflect effects of transplantation procedures such as radiation damage to the blood-brain barrier, variable criteria for defining the donor-derived cells, or species differences.
No cell type specific antibodies were applied in the study. Thus, the exact phenotypes of the CD45/ML-I−Y+ cells were not determined. The morphology and spatial organization of donor-derived CD45/ML-I−Y+ cells, however, suggest that they are primarily fibroblast-like cells, pericytes, possibly myocytes and/or associated satellite cells, and occassional epithelial cells. No cells with unambiguously endothelial phenotype were found. A majority of the donor-derived nonhematopoietic cells may represent interstitial cell populations, and integration into parenchymal populations is apparently rare.
The migration and differentiation pathways of the donor-derived nonhematopoietic cells in the freemartins are unclear. We are at present unable to differentiate between the plasticity of circulating cells and heterogeneity of circulating cell populations. Two major SC types are known to be circulating in fetal and adult blood in mice and humans: the HSCs and MSCs [58–61]. Both are currently intensely investigated in the context of adult SC plasticity. Significant numbers of HSCs are exchanged between bovine twins, as the freemartin hematopoietic system is permanently chimeric  and can be up to 96% donor derived . Thus, our results suggest that the contribution of HSCs to the nonhematopoietic tissues is very limited, despite recent reports to the contrary [4, 6]. MSCs, also known as stromal SCs, reside in the BM and other tissues, and are able to differentiate into various mesenchymal lineages [62–64]. The heterogenous MSC pool includes so-called multipotential adult progenitor cells (MAPCs) that can differentiate also into endodermal and neuroectodermal cell types, at least after prolonged culture [7, 65, 66]. Interestingly, postnatally injected MAPCs contribute little to brain, skin, and kidney , tissues with especially few donor-derived cells in freemartins. We are currently unable to reliably estimate the rate of MSC exchange between bovine twins. Potential postnatal contribution from the MSC pool to freemartin tissues would not be detected, if the cells were predominantly of host origin. MSC engraftment after BM transplantation is inefficient [67, 68]. In fetal life, however, MSCs circulate during the establishment of the hematopoiesis-supportive stromal environment in liver, spleen, and BM [58, 69]. MSCs are therefore likely to be exchanged between the twins, but based on our results, they too are rarely integrated into undamaged nonhematopoietic bovine tissues.
In 1 of 12 freemartins (FM10), markedly higher numbers of CD45−Y+ cells were detected in several tissues, especially in the mucosa of the small intestine and in the mammary gland. It is at present difficult to explain these findings. FM10 was of the Ayrshire breed like the majority of the freemartins in the study. It was the oldest animal examined (20 months), but not markedly older than the closest cases (16 months and 14 months). An exceptionally early formation of anastomoses could possibly result in an exchange of more multipotential cells than those available later, but this is unlikely to occur. The anastomoses are generally thought to form at 30–35 days p.c.[25, 46, 47]. This is only possible after the vascularized allantois has fused with the avascular chorion during days 28–33 . Thus, the window is probably too narrow to allow significant variation.
In the FM10 intestinal villi, strands of CD45−Y+ cells were seen (Fig. 1I), suggesting that some of the epithelial SCs residing in the crypts were donor-derived. This is surprising, as the intestinal epithelium is thought to originate from the local primitive gut endoderm [70, 71]. The conversion from an undifferentiated stratified epithelium to a mucosa with villi covered with simple columnar epithelium, associated with the establishment of the crypt SC system, occurs at a fairly late stage, when donor cells should be available in all freemartin fetuses . Therefore, a subtle difference in the timing of the anastomosis does not alone explain the differences in the donor contribution. A similar intestinal engraftment has been reported after intravenous injection of BM-derived multipotential progenitor cells into adult non-obese diabetic/severe combined immunodeficient mice . Circulating cells thus appear to integrate to the intestinal epithelium in certain conditions, but not generally.
Mammary glands develop from sprouts of local ectoderm into underlying mesenchyme. A small ductal tree develops prenatally and expands later in puberty [72, 73]. Donor contribution to the mammary connective tissue is perhaps less unexpected than to the duct epithelium. MSCs are known to circulate in the fetal blood , and abundant vasculature is seen around the proliferating mesenchyme in the early mammary gland anlage . In FM10 the frequency of Y+ cells in the basal myoepithelial layer intriguingly appears to be several times higher (and comparable to the surrounding connective tissue) than in the luminal layer (unpublished observations). Yet, both epithelial populations are thought to have a common ectodermal origin and seem to be derived from common multipotential SCs [74, 75].
The abundant Y+ signal in FM10 may also result from reasons other than circulating cells. With current tools we are unable to completely exclude the possibility of chromosomal abnormalities in part of the cells of this freemartin. Another theoretically feasible explanation is the fusion of early embryos followed by recleavage, which would be impossible to distinguish from a solely vascular connection afterwards. Such events are rare, but the sample size in this study does not allow reliable estimation of the frequency of such freemartins in the general cattle population.
Taken together, our findings support the view that contribution of circulating cells (including HSCs) to development, growth and physiological turnover of nonhematopoietic tissues is in normal conditions very limited, but they may contribute effectively to the regeneration of damaged tissues. The donor-derived cells probably exit the circulation in a differentiated state, or their proliferation rate in target tissue is low in relation to migration rate.
The freemartin is a unique model for studying regenerative processes in a large mammal. It will be interesting to dissect the functions of donor-derived cells in damaged and pathological freemartin tissues. As the level of inherent nonhematopoietic chimerism is low, transplantation and follow-up of immunologically compatible Y+ cells or tissues from the male twin are feasible. Thus, the cattle twins provide an exciting tool for studying the plasticity of various cell types without manipulation of the recipient or the transplanted cells.
We thank M. Georges (University of Liège, Belgium) for the btDYZ clone, U. Pfüller (Witten/Herdecke University, Germany) for the biotinylated mistletoe lectin I, staff at the Saari unit (Dept. Clinical Veterinary Medicine, University of Helsinki) for providing blood samples of potential freemartins, K. Lahti, T. Pankasalo, H. Valtonen, and A. Koivunen for expert technical assistance, and L.C. Andersson and L.A. Lindberg for valuable comments about the work.