Minimal Evidence of Transdifferentiation from Recipient Bone Marrow to Parenchymal Cells in Regenerating and Long-Surviving Human Allografts


*Corresponding author: A. J. Demetris,


Liver, small intestine, and heart allografts in residence for 4 days to 16 years were analyzed by simultaneous XY fluorescent in situ hybridization to search for evidence of the recently described process of transdifferentiation of recipient bone marrow stem cells to allograft parenchymal cells. These studies were carried out in an effort to find conditions associated with maximal levels of engraftment or expansion of the recipient parenchymal cells. Despite prolonged survival up to 16 years, regeneration after severe preservation injury or use of split livers, only rare, isolated and tentatively identified recipient hepatocytes were detected in liver allografts. In intestinal allografts, despite survival of up to 8 years and extensive mucosal regeneration because of severe damage from acute rejection, there was no crypt replacement by recipient epithelial cells. In cardiac allografts, no recipient myocytes were detected despite recipient survival for 2–3 days and 3–4 weeks after myocardial infarcts at 5 and 8 years after transplantation.

Parenchymal cell transdifferentiation from recipient bone marrow stem cells was rare to nonexistent in severely injured, regenerating, and long-surviving allografts. The rare isolated recipient parenchymal cells tentatively identified did not appear to behave as stem cells: they did not form clusters and did not increase with time after transplantation. Because of the extremely low frequency, interpretation was difficult. Regardless of these results, a more vigorous search for conditions that promote transdifferentiation is warranted.


Recent exciting experimental animal and human studies suggest that embryonic and hematopoietic stem cells can transdifferentiate into a variety of adult cell types including hepatocytes (1–6), cardiac myocytes (7–9), intestinal epithelium (3), neurons and microglia (10–12), renal tubular epithelium (13), and skeletal muscle (14). Exploiting this phenomenon could revolutionize solid organ and cell transplantation. Chronic rejection and the lifelong immunosuppression needed to prevent it might be avoided if treatment strategies could be devised to promote gradual replacement of donor parenchymal cells with recipient ones. Instead of making the immune system tolerant to the allograft, the allograft could be made tolerant to the recipient.

Sex differences between human allograft donors and recipients have been exploited to document the appearance of recipient hepatocytes (5,6), renal tubular epithelium (13), and cardiac myocytes (9) amidst donor parenchymal cells in human liver, kidney, and heart allografts or in native organs of bone marrow allograft recipients. Theise et al. showed that 1 month to 2 years after bone marrow or liver transplantation, 4–43% of hepatocytes were of male origin in two female recipients of male bone marrow allografts and four male recipients of female orthotopic liver allografts (5). Similar results were obtained by Alison et al. (6) who found 0.5–2% male hepatocytes in the liver of nine female patients who received male bone marrow and 11 male recipients of female orthotopic liver allografts. Poulsom found up to 20% of the renal tubular epithelium had been replaced in kidney allografts less than 1 year after transplantation (13). Quaini et al. (9) found that 4–552 days after transplanting a female donor heart into a male recipients, 7–10% of cardiac myocytes, vascular smooth muscle cells, and endothelial cells carried the Y-chromosome.

The number of recipient-derived parenchymal cells reported in these humans allografts is astonishingly high considering the relatively short time after transplantation when the studies were conducted (5,6,9). Both the human and experimental animal studies suggest that two factors might favor the transdifferentiation process: (1) parenchymal cell turnover because of injury and proliferation (3,5,7–9); and (2) ongoing selection pressures that favor the growth of recipient-derived parenchymal cells (2–5,15). As recipient-derived cells should not be immunologically recognized as ‘foreign’ and thereby enjoy a selective growth advantage, we reasoned that prolonged time after transplantation, reduced-size donor livers that undergo hyperplasia after transplantation, and allografts that are severely damaged, but recover, should show easily detectable clusters of recipient parenchymal cells within at least some of these allografts.

Materials and Methods

Fluorescence in situ hybridization for X and Y chromosomes

Archived tissues from 21 male recipients of female liver, small bowel, and heart allografts were analyzed. Studies were carried out under the exempt RB protocol #010259. One section containing a 2 × 2-cm tissue fragment of all long-term (>2 years) resected whole liver allografts were examined. In the longest survivor (L5: 16 years), two sections were examined. Formalin-fixed paraffin-embedded sections were serially sectioned at 4–5-μm intervals. The slides were deparaffinized in xylene twice for 10 min, dehydrated twice with 100% ethanol. In a representative case (the longest survivor), an indirect immunolabeling procedure was used to simultaneously localize LCA protein expression in the formalin-fixed sections [leucocyte common antigen, 1 : 100 dilution (M701 Dako, 1 : 100)] before in situ hybridization for the sex chromosomes. Briefly, samples were blocked with Blue Block (Shandon, Pittsburgh, PA) for 15 min, and then incubated with the primary antibody for 1 h at room temperature. After washing with phosphate-buffered saline (PBS) plus 0.05% Tween 20, slides were incubated at room temperature for 30 min with biotinylated horse antimouse secondary antibody (1 : 350 dilution, BA 2000; Vector Laboratories, Burlingame, CA). The samples were then developed with Vectastain-Elite ABC (Vector Laboratories), stained with liquid diaminobenzidine (DAB) + (DAKO), and placed in PBS/Tween for 5 min.

The slides were then pretreated using the Vysis Paraffin Pretreatment Kit and digested for 28 min in protease solution (0.5 mg/mL) at 37 °C. Dual color FISH was performed using the X/Y direct labeled probe (Vysis, Inc., Downers Grove, IL). The target slide was denatured at 75 °C for 5 min in 70% formamide (Intergen, Purchase, NY) and dehydrated in solutions of 70%, 85%, 100% ethanol and then incubated with probe overnight at 42 °C in a humidified chamber. The posthybridization wash used 0.4 × SSC at 72 °C for 2 min. The slides were then air-dried in the dark, counterstained with DAPI and viewed and photographed with a ×100 Oil plan fluor objective using a Nikon Optiphot-2 and Quips Genetic Workstation equipped with Chroma, the Technology 83000 filter set with single band exitors for Texas Red/Rhodamine, FITC, DAPI (uv 360 nm) and a dual filter for spectrum Orange/FITC, and a triple filter for spectrum of Orange, FITC, and DAPI. All analyses were performed using dual-color FISH to increase the sensitivity and specificity of the assay. Positive and negative controls for X and Y chromosomes were performed using the tissues from nonallograft male and female patients: two orange signals (X chromosomes) in female cells, and an orange and a green signal (X and Y chromosomes) in male cells. No male parenchymal cells were found in control female tissues, and no female parenchymal cells were found in male control tissues. The presence of female parenchymal cells and male inflammatory cells in the analyzed allografts served as additional internal positive controls.

The slides were first reviewed and evaluated for quality by assessing the presence of appropriate signals in the positive internal controls (inflammatory cells). A diligent search for any male parenchymal cells (hepatocytes, bile duct cells, intestinal epithelium, cardiac myocytes, and smooth muscle cells) was made by five pathologists and an experienced cytogenetic technician. Any apparently Y positive parenchymal cells were photographed. Then an average of 1000 hepatocytes (except L4 because of the small amount of tissue available for analysis), 500 intestinal epithelial cells, and 200 myocytes were ‘officially counted’ in each case, starting with the bias of including any equivocally positive cells. Cytoplasmic autofluorescence (green/yellow) was used to aid in identification of hepatocytes; nuclei were devoid of autofluorescence, and stained blue with the DAPI counter stain. Areas of possible replacement were evaluated using a series of filters that enabled viewing of only the X or Y chromosomes (single bandpass filters), as well as both the X and Y chromosomes (dual bandpass filters), and viewing the X and Y chromosomes with DAPI counter stains (triple bandpass filters). Digital images were taken of every equivocal cell and discussed at each stage.

Immunohistochemistry for p21, Ki-67, LCA and Desmin

An indirect immunolabeling procedure after microwave antigen retrieval in 10 mmol/L citrate buffer (pH 6.0) was used to localize p21WAF1/Cip1 and Ki-67 protein expression in the formalin-fixed sections (Waf-1, 1 : 60 dilution, Cat No. OP64; Oncogene Research, Cambridge, MA; Ki67 1 : 25 #M7187 Dako). LCA (M701 Dako, 1 : 100) and Desmin (M760 Dako, 1 : 100) were performed with no antigen retrieval. Briefly, after antigen retrieval in the required antibodies, samples were blocked with Blue Block (Shandon, Pittsburgh, PA) for 15 min, and then incubated with the primary antibody for 1 h at room temperature. After washing with phosphate-buffered saline (PBS) plus 0.05% Tween 20, the slides were incubated at room temperature for 30 min with biotinylated horse antimouse secondary antibody (1 : 200 dilution, BA 2000; Vector Laboratories, Burlingame, CA), and then developed with Vectastain-Elite ABC (Vector Laboratories), stained with liquid Dab + (DAKO) and counterstained with hematoxylin. An immunoglobulin class-matched nonimmune antibody was substituted for the primary antibody in the negative controls.

Histological evaluation

Pathological diagnoses of the removed allograft tissue were based on the original pathological report and re-examination of the tissue sections by two pathologists (T.W and A.J.D).


Study patients, tissues, and circumstances

The type of transplant, donor and recipient ages, recipient original disease, duration of allograft survival at the time of sampling, the type of sample, and the histopathologic diagnosis, are shown in Table 1. Specimens were first selected on the basis of female to male recipients and then by circumstances such as duration of engraftment, allograft injury, and/or regeneration. These studies were carried out according to UPMC Institutional Review Board Protocols #010259, 000966 and Children's Hospital of Pittsburgh exempt protocol of 6/25/2001.

Table 1.  Demographic characteristics, original disease, specimen types, time after transplantation and cause of allograft failure or diagnosis
CaseTxRec. ageDonor ageOriginal diseaseSpecimenTime after TxHistopathologic findingsGenotype of hepatocytes (% based on count of at least 1000 hepatocytes)
  1. AS = angiosarcoma; Au = autopsy; BA = biliary atresia; CAD = coronary artery disease; CM = cardiomyopathy; ETOH = alcoholic cirrhosis; HBV = hepatitis B virus; HCV = hepatitis C virus; ICH = idiopathic chronic hepatitis; LG×= liver allograft resection; NA = not applicable; NB×= needle biopsy; ND = not done; OHT = orthotopic heart transplant; OLT = orthotopic liver transplant; PCD = polycystic disease; PSC = primary sclerosing cholangitis; RMS = rhabdomyosarcoma; SBG×= small bowel allograft resection; SBS = short bowel syndrome; SBT = orthotopic small bowel transplant.

L1OLT-whole4047HBVLGx9.5 yearsCirrhosis because of recurrent HBV11.878.
L2OLT-whole4255HBVLGx3.5 yearsCirrhosis because of recurrent HBV1979.
L3OLT-whole5151HCV/HBVLGx11 daysHA thrombosis with centrizonal necrosis and regeneration22.
L4OLT-whole3654HCV/ETOHLGx4 daysPrimary non-function with multi-focal necrosis16.183.9000
L5OLT-whole2658ICH/ETOHLGx16 yearsRecurrent idiopathic chronic hepatitis and early cirrhosis3.
L6OLT-whole34NAHBVLGx298 daysFibrosing Cholestatic Hepatitis, viral type B099.6000.4
L7OLT-whole54NAPCDNBx3 monthsDiffuse steatosis11.777.
L8OLT-whole1NAPSCNBx2 monthsCryptosporidiosis and CMV infection13.385.70.900.1
L9OLT-split1NABANBx22 daysDiffuse steatosis1286.
L10OLT-split4NABANBx3 monthsChanges suggestive of biliary obstruction11.1821.75.20
L11OLT-split5NAASNBx4 monthsChanges suggestive of biliary obstruction25.869.63.21.40
L12OLT-split1NAAlagille's syndromeNBx1 monthChanges suggestive of biliary obstruction14.984.
L13OLT-whole8NARMSNBx3 monthsChanges suggestive of biliary obstruction15.983.70.300.1
S1SBT104SBSSBGx3 yearsAcute and chronic rejectionNDNDNDND0
S2SBT3136SBSSBGx2 yearsAcute and chronic rejectionNDNDNDND0
S3SBT2730SBSPartial8 yearsSerosal fibrosis SBGxNDNDNDND0
S4SBT2932SBSSBGx7 monthsSerosal fibrosisNDNDNDND0
S5SBT3718SBSSBGx7 monthsSerosal fibrosisNDNDNDND0
H1OHT3619CADAu5 yearsChronic rejection and myocardial infarct (2–3 days old)NDNDNDND1 cell
H2OHT5639CMAu8 yearsChronic rejection and myocardial infarct (3–4 weeks old)NDNDNDND0
H3OHT6754CADAu1 yearNo significant pathological changesNDNDNDND0

Liver allografts

Thirteen liver allograft specimens (six failed allografts and seven needle biopsies) were examined for male parenchymal cells between 4 days and 16 years after transplantation (Table 1). All seven adult and 2/6 pediatric liver allograft recipients were given female whole cadaveric livers, while 4/6 pediatric patients received split liver grafts. Needle biopsies from the split livers were included to determine whether the compensatory hyperplasia associated with transplanting a liver fragment (16) resulted in enhanced incorporation of donor parenchymal cells during the peak period of hepatocyte proliferation (16). Failed allograft specimens were included because of the large amount of tissue available for evaluation. Several of the failed allografts also offered the opportunity to examine tissues after prolonged engraftment, during which they were exposed to continual hepatocyte turnover because of recurrent chronic viral hepatitis type B and/or C (3.5–9.5 years) or recurrent chronic idiopathic hepatitis (16 years). Other failed allografts were chosen because they showed brisk regeneration of hepatocytes or progenitor cells (17) after severe acute injury because of primary nonfunction (one case, 4 days after transplantation) or infarcts from hepatic artery thrombosis (one case, 11 days after transplantation).

After the entire slide was reviewed, a formal count of at least 1000 hepatocytes was undertaken, starting with the bias including any putatively positive cells. The range of percentages of hepatocytes with the following genotypes were detected: XO (3.2–25.8%); XX (69.6–99.6); XXX (0–6.1), and XXXX (0.1–5.2%) (Table 1). XY positive cells seen within the hepatocyte plates and bile ducts were carefully assessed by several pathologists and a cytogeneticist using appropriate filters for visualizing the various combinations of one, two or all three labels (X = orange; Y = green; and DAPI nuclear counter stain = blue). Without prior discussion, each of the investigators who reviewed the slides (TW, KC, MN, PR, GM, RJ, AJD) independently reached the same conclusion. Most, if not all, of the XY positive cells were infiltrating inflammatory cells. This interpretation was supported in the longest survivor by simultaneous LCA labeling and XY FISH (Figure 1). However, rare isolated XY positive cells in the hepatocyte plates could not be confidently assigned to a particular cell type in spite of extensive analysis, and were labeled as ‘ambiguously positive cells’. These putative recipient-derived hepatocytes accounted for 0.1–0.4% of the total number of hepatocytes (n = 1–4 hepatocytes/per 1000 hepatocytes). They did not localize to periportal cholangioles or canals of Herring, and did not form clusters or colonies. In addition, the putative recipient hepatocytes were surrounded by female hepatocytes, did not increase with time after transplantation (Table 1, Figure 2), and were not more common in allografts with regeneration because of split liver operations or severe injury (Table 1).

Figure 1.

Example of the difficulties encountered in analyzing liver allograft tissues. (A) In this example, note the male X (orange signal)/Y (green signal, →) recipient sinusoidal lymphoid cells (→), characterized by scant cytoplasm lacking autofluorescence and small nuclei with dense homogeneous nuclear staining. (B) Occasionally, XY positive cells appeared to be within the hepatic plates (→). However, on closer examination with the use of double labeling for leukocyte common antigen, two overlapping nuclei could be identified. The smaller arrowheads outline the smaller LCA + lymphoid cell, which also shows red cytoplasmic labeling from diaminobenzidine. Behind the lymphoid cell is the larger nucleus of a hepatocyte (P = hepatic plate; S = sinusoids). Note also that most of the surrounding hepatocytes are XX + 0. (C) Similar overlaps were seen in a cross-section with another apparent XY + hepatocyte (→). (D) However, using the DAPI filter alone, note the dense nuclear staining of the putative recipient hepatocyte and the difference in level of the tissue plane between the surrounding more typical hepatocytes. Note also the ‘cell’ in question is actually two closely aligned cells separated by a cleft.

Figure 2.

Failed liver allograft (L5) removed more than 16 years after transplantation because of recurrent idiopathic chronic hepatitis and early cirrhosis. As is typical for the development of cirrhosis from chronic hepatitis there was significantly increased hepatocyte injury and proliferation. Immunohistochemical staining for the proliferation marker Ki-67 showed that up to 2% of the hepatocytes were actively dividing (bottom right inset). FISH analysis for X and Y chromosomes showed that 96.6% of hepatocytes were of XX, XXX, or XXXX genotype (donor origin), whereas 3.2% were XO and 0.2% (2/1050) were putative male recipient hepatocytes. One such cell is shown in the bottom left inset (→). As in all of the liver specimens, the putative recipient hepatocyte in this case was found as an isolated single cell and not within a group or small cluster. Note that most of the surrounding hepatocytes have at least two XX chromosomes and no Y chromosomes.

Small intestine allografts

Small intestinal epithelium is unequivocally derived from stem cells located in the crypts (18), and the crypt epithelium is specifically targeted for immunologic destruction in acute and chronic rejection (19,20). Two small bowel allografts (S1 and S2) seemed to be ideal candidates for documenting the transdifferentiation process. During the post-transplant course, both of these allografts underwent extensive mucosal sloughing because of severe acute rejection, which was successfully treated. The epithelium completely regenerated in both cases, from stem cells. However, both allografts also eventually failed from chronic rejection approximately a year or more after the episodes of acute rejection (Table 1). Several other small intestine allograft specimens were examined because of intestinal obstruction 8 years after transplantation (S3), or during ileostomy closure 7 months after transplantation (S4 and S5).

In all cases, XY-chromosome positive inflammatory cells were easily identified in the lamina propria, submucosa, and intravascular spaces, and served as internal positive controls. Isolated XY positive cells were also located within the epithelial layers, interpreted as either part of the rejection process, or normal small intestinal physiology. These intraepithelial cells did not have the morphologic features of intestinal epithelial cells, and most were LCA-positive by double labeling for LCA and XY FISH. In addition, despite a normal pattern of replication and differentiation in the intestinal epithelium (Figure 3), the XY positive cells within the epithelial layer were not adjacent to each other as might be expected in an expanding epithelial cell clone within a crypt (18,21).

Figure 3.

Analysis of an allograft small intestinal allograft (S1). Immunohistochemical staining for the proliferation marker Ki-67 (A) and the cyclin-dependent kinase inhibiter p21 (B) in a resected small intestine allograft (S1) showed the expected distribution of positive staining for nuclear Ki-67 at the base of the crypts and nuclear p21 expression in the surface epithelium. (C) and (D) FISH analysis of the same allograft (S1) showed that the crypts were comprised almost totally of XX positive cells with only occasional XY chromosome lymphocytes interspersed within the epithelial layer (→). This particular allograft experienced extensive mucosal sloughing from moderate to severe acute rejection before allograft failure. However, after successful treatment with immunosuppression, the epithelium had largely regenerated, presumably from residual donor stem cells. Unfortunately, the allograft eventually failed approximately 1 year later because of chronic rejection. In intestine allografts such as this one, at least occasional crypts should be entirely of recipient origin given the selection pressure of immune elimination from rejection and the need for stem cell activation and expansion to populate the mucosa. However, no recipient crypts were identified.

Heart allografts

Studies in experimental animals suggest that hematopoietic stem cells incorporate into the damaged myocardium during repair of infarcts where they differentiate into myocytes, smooth muscle cells of arteries, and vascular endothelial cells (7,8). Two such heart allograft specimens were studied. One patient (H1) died 5 years after transplantation because of chronic rejection and a superimposed acute left ventricular subendocardial myocardial infarct that occurred 2–3 days before his demise. Another patient (H2) died 8 years after transplantation because of Staph aureus bronchopneumonia. He experienced a subacute left ventricular and anterior-septal myocardial infarct estimated to have occurred 3–4 weeks or longer before his demise (Figure 4). A third specimen was available from a recipient who died 1 year after transplantation because of sepsis and right-sided heart failure secondary to chronic tricuspid regurgitation.

Figure 4.

Analysis of allograft heart tissues. (A) H&E section of an allograft heart (H1) from a recipient 8 years after transplantation. The subendocardial myocyte dropout and early fibrosis are the result of a myocardial infarct estimated to have occurred approxiately 3–4 weeks or more before the patient died. (B) Interface zone of an acute myocardial infarct occurring 5 years after transplantation. Based on clinical symptoms and histologic findings, the infarct is estimated to have occurred 3–4 days before the recipient died. (C,D) FISH analysis showing the presence of XY positive lymphocytes and interstitial cells in both samples. However, no cardiac myocytes, characterized by copious cytoplasm showing lipofuschin autofluorescence and delicate nuclear staining were detected. As in the other tissues, the inflammatory cells are characterized by scant cytoplasm lacking autofluorescence and dense homogeneous nuclear staining.

The heart allografts with infarcts described earlier showed predominantly neutrophilic (H1) or mild lymphocytic (H2) inflammation in areas of recent or remote myocyte necrosis, respectively (Figure 4). Cardiac tissue away from the infarcts was relatively unremarkable. The heart allograft from case H3 revealed no significant pathological changes. Scattered XY positive neutrophils and lymphocytes in the interstitium and vascular lumina were more prominent within and near the infarcts (Figure 3), but the cardiac myocytes in and around the infarcts were entirely comprised of XX positive female myocytes. Although occasional XY chromosome positive cells were seen in the myocyte parenchyma, careful examination, as with the liver, suggested that most of these cells are inflammatory cells rather than cardiac myocytes. A single XY positive cell in the myocyte parenchyma was difficult to assign to either myocytes or inflammatory cells (H1); it was considered as ‘ambiguously positive’. This type of cell with an uncertain origin was not seen in any of the other cases.


The results of this study suggest that transdifferentiation of recipient bone marrow stem cells into parenchymal cells is an extremely rare event in liver, small intestinal, and heart allografts, and the process does not increase with time after transplantation or with acute or persistent injury or hyperplasia, or with the chronic selection pressure of immune-mediated injury. Moreover, the equivocal recipient parenchymal cells detected did not appear to behave as stem cells. The apparently transdifferentiated cells did not localize to traditional stem cell compartments and did not give rise to cell clusters, as is typical for parenchymal cell colonization in chimeric organs (18,21–23) or as seen in convincing experimental models of bone marrow to hepatocyte transdifferentiation (2,15).

Some of our findings in the liver are similar to those of Theise et al. (5) and Alison (6), but other findings and the interpretation are significantly different, and more in agreement with Fogt et al. (24). As curative replacement of large clusters or clones of hepatocytes derived from flow cytometry-sorted hematopoietic stem cells has been described in an experimental animal model of tyrosenemia (2), we expected to find clusters or clones of recipient parenchymal cells, considering the long periods and conditions of engraftment and tissue samples examined. In the experimental animal studies, hepatocyte functional activity and bone marrow-specific phenotypic/morphologic markers provided two independent methods of verifying that the bone marrow cells had indeed transdifferentiated into hepatocytes (2). The liver allografts examined in this study were exposed to severe acute or chronic injury or hyperplastic stimuli under the selection pressure of alloimmunity. However, similar to Theise et al. (5) we found only single isolated hepatocytes, albeit at a much lower frequency (5). We found no small clusters or clones of recipient hepatocytes, as described by Alison et al. (6).

The method of analysis used to identify transdifferentiated hepatocytes in this study differed from analyses performed by other groups. We used simultaneous X and Y FISH followed by ×100 oil magnification to search for transdifferentiated cells, as well as for photographic documentation. In comparison, Theise et al. (5) used ×60 for analysis and ×20 for photographic documentation. At these lower magnifications, we found it difficult to discern infiltrating cells from parenchymal cells and dot-like concentrations of blue or turquoise DAPI signals that closely mimicked the authentic green Y signals. In addition, cell morphology is more accurate at ×100 magnification using the single band pass filter for DAPI. These combinations enable one to distinguish inflammatory cells or overlapping cells from the parenchymal cells. Many cells initially thought to be equivocal were determined to be nonparenchymal cells using these approaches.

Although we used simultaneous LCA and XY FISH for further analysis in the most interesting case, caution is urged in the interpretation of such double-labeled samples. First, additional manipulation of the tissue needed for double labeling further distorts the morphology and introduces other artifacts that make interpretation more difficult (25). Second, it is our opinion that double labeling does not completely circumvent the problem of overlapping cells. Some cells are so closely aligned that double labeling actually contributes to a false sense of certainty, particularly at lower magnifications. Lastly, we reasoned that some obvious clusters of recipient parenchymal cells should have been detected in at least some of these allografts, and double-labeling determinations should not be needed.

In the hepatic allografts examined, from 73 to 99% of the hepatocytes contained at a least two sex chromosomes. This figure compares favorably with the percentage of Y chromosome loss in FISH studies in tissue sections (26,27), but is somewhat lower than Theise reported in his male controls (5). Although the percentage of XO positive cells reached as high as 26% in one case, the XO hepatocytes were distributed evenly and randomly throughout the sections. Thus, if expanding clones of recipient hepatocytes were present, it is unlikely that they would have been missed by the techniques employed. In addition, all of the putative recipient cells identified were of the XY genotype and were surrounded by hepatocytes with at least two X chromosomes. None of the Y positive hepatocytes were XXY or XXXY, as described by Alison et al. (28). This suggests, but does not prove, that cell fusion is unlikely to account for the rare recipient hepatocytes identified. However, if these hepatocytes were derived from transdifferentiation, they did not behave normally and clonally expand under the proliferative stimuli and the alloimmune selection pressures examined in this study.

The intestinal allografts provided probably the most stringent test of transdifferentiation in allografts. Crypt epithelial cells are specifically targeted for injury in rejection (20,29) and the crypt and villous epithelium is clearly derived clonally from stem cells (18,30). Two of the allografts examined experienced extensive mucosal sloughing because of acute rejection. After successful treatment, the epithelium regenerated, presumably from residual stem cells, but these allografts eventually failed from chronic rejection. Despite showing normal proliferation and differentiation kinetics in the resected allo-grafts, we were unable to find clonal crypt replacement by recipient epithelial cells, as might be expected (18,21,31) under these circumstances. Our findings are similar to those reported by Tryphonopoulos et al. (25), although their interpretation was different. They concluded that isolated recipient cells in the epithelium represented transdifferentiated cells that had incorporated into the crypts.

In heart allografts we found no convincing evidence of bone marrow to myocyte transdifferentiation after myocardial infarcts, in contrast to the high percentage of transdifferentiation documented by Quaini et al. (9) in heart allografts without infarcts. Unlike Quaini et al. (9) however, we chose to avoid atrial tissue and tissue near anastomoses because of the potential for confusion between the donor and recipient boundaries.

There are several possible explanations for our findings and differences with other studies: (1) selection pressures on the transdifferentiated cells in the organ allografts examined may not be strong enough to elicit clonal expansion, as seen in the experimental models (15); (2) transdifferentiated parenchymal cells might be abnormal or fragile, or occur as a result of cell fusion (10,32), such that they have limited replicative capacity and/or a short lifespan; (3) local differences in immunosuppression regimens or other treatment policies might interfere with transdifferentiation and/or growth; (4) optimal conditions for promoting transdifferentiation or sufficient tissue volumes may not have been examined in this study; (5) transdifferentiation may not be as common as originally described (33,34); or (6) various combinations of the above.

Still other possibilities exist. First, long-term ‘accepted’ allografts might harbor more transdifferentiated cells, and in fact, might contribute to allograft acceptance. Examination of operationally tolerant human liver (35,36) and kidney allografts (37) that had survived up to 29 years after transplantation using the Y chromosome and/or anti-MHC monoclonal antibodies to distinguish donor and recipients cells in humans came to the same general conclusion as this report. The vast majority, if not all, of the allograft parenchymal cells had the sex chromosomes and MHC phenotype of the donor. It is very unlikely therefore that recipient bone marrow stem cell transdifferentiation to allograft parenchymal cells contributes significantly to mechanisms of allograft acceptance in humans. Second, in liver allografts, one cannot be entirely certain that all recipient hepatocytes are removed at the time of transplantation. It is possible that hepatocytes or circulating liver stem cells might move back to the liver environment.

In summary, apparent transdifferentiation from bone marrow stem cells to hepatocytes has been shown in some experimental systems (2,15). However, we did not find clusters of transdifferentiated parenchymal cells in sex-mismatched solid organ allografts despite examining organs exposed to a variety of conditions where it might be expected, such as injury, repair, hyperplasia, long-term engraftment, all occurring under the persistent selection pressure of alloimmune destruction. Because the frequency of apparently transdifferentiated cells was so low, interpretation was fraught with pitfalls. Nevertheless, as ambiguous cells were detected and the biologic and therapeutic potential of this process is so promising, a more vigorous search for evidence of significant transdifferentiation in organ allografts is warranted.


Although controversy continues in the field of transdifferentiation of adult hematopoietic stem cells to organ parenchymal cells, one of the most convincing models of transdifferentiation (2) has recently been shown to occur by cell fusion (38).


This work was supported by NIH grant DK49615.