Lack of a Fusion Requirement for Development of Bone Marrow-Derived Epithelia

See allHide authors and affiliations

Science  02 Jul 2004:
Vol. 305, Issue 5680, pp. 90-93
DOI: 10.1126/science.1098925


Analysis of developmental plasticity of bone marrow–derived cells (BMDCs) is complicated by the possibility of cell-cell fusion. Here we demonstrate that epithelial cells can develop from BMDCs without cell-cell fusion. We use the Cre/lox system together with β-galactosidase and enhanced green fluorescent protein expression in transgenic mice to identify epithelial cells in the lung, liver, and skin that develop from BMDCs without cell fusion.

Cells marked as bone marrow–derived (BMDCs) can be found as mature cells of many nonhematopoietic tissues, including lung, liver, kidney, skin, and muscle. The BMDCs are often identified by the presence of the Y chromosome in sex-mismatched bone marrow transplants (male into female) or by detection of gene products such as enhanced green fluorescent protein (EGFP) that are present in the donor but not the recipient. Interpretation of these results has been complicated by observations that in vitro coculture of embryonic stem cells and somatic cells can result in spontaneous cell fusion (1, 2), giving rise to cells of mixed phenotype and genotype. The in vivo appearance of marrow-derived hepatocytes, cardiomyocytes, and Purkinje cells is due, at least in part, to fusion of BMDCs with these cell types (17). However, because some of these cell types are known to form heterokaryons in settings of profound tissue injury (810), the incidence of this process must be examined in nonfusogenic organs under physiologic conditions.

This study was designed to evaluate for fusion events, including those that may have been masked by reductive division. We did this using the Cre/lox recombinase system to examine whether fusion occurs between BMDCs and host cells after bone marrow (BM) transplantation. We used mice of the Z/EG Cre-reporter strain (11, 12) (fig. S1) as marrow donors for transplantation into mice that ubiquitously express Cre. In this model, any cell resulting from fusion of a BMDC with a host cell should express EGFP.

We transplanted lethally irradiated female mice that ubiquitously expressed Cre recombinase with BM from male Z/EG donor mice, β-actin–Cre (Cre) donor mice (13), or Z/EG and β-actin–Cre F1 (Z × C F1) donor mice. The transplants from Z × C F1 mice into Cre mice served as positive controls for EGFP expression in donor-derived cells, and the Cre-into-Cre transplants served as negative controls. Tissues from the recipients were analyzed 8 to 12 weeks after transplantation for the presence of BM-derived (Y chromosome–positive) epithelial cells and EGFP expression.

For analysis of the lungs, single-cell suspensions were analyzed by flow cytometry and fluorescence microscopy. For fluorescence-activated cell sorting (FACS), the M1 gate was determined by the fluorescence intensity greater than 99% of the cells in an EGFP-negative population (Fig. 1A, horizontal black line). Given this gate, 66% of all cells from the lungs of Z/EG × Cre F1 mice were EGFP-positive (Fig. 1A, green). The EGFP-negative population mostly comprised blood and endothelial cells (14). In the EGFP-positive control transplants (Fig. 1A, blue), 10% of the cells were EGFP-positive. Subsequent immunocytochemistry on FACS-sorted EGFP-positive cells showed that 6% of these were both EGFP- and cytokeratin-positive (Fig. 1, B and C). The high percentage of EGFP-positive, cytokeratin-negative cells was due to contaminating blood cells in the lung digests. From these data, we conclude that at least 0.6% of total lung cells were EGFP-positive epithelial cells. In experimental transplants from Z/EG mice into Cre mice (Fig. 1A, orange and red), ∼0.8% of the experimental cells had a fluorescence intensity greater than the EGFP-negative baseline population, which is equal to the percentage in negative-control β-actin–Cre (Fig. 1A, black) animals. Immunofluorescence analysis indicated that these cells were uniformly negative for EGFP (14).

Fig. 1.

(A) FACS histogram showing EGFP fluorescence (X axis) of cells derived from dissociated lungs, for Z/EG × Cre-positive controls (green), β-actin Cre-negative controls (black), Z/EG × Cre F1–into-Cre transplants (blue), and two separate Z/EG-into-Cre transplants (red and orange). These data are representative of seven Z/EG-into-Cre transplants. (B and C) Immunofluorescence for EGFP (green) and cytokeratin (red) on cytospins from EGFP-positive cells from lungs of Z/EG × Cre F1–into-Cre mice that were sorted by FACS with the M1 gate. The same image is shown (B) merged and (C) with the green channel removed. (D to G) Immunofluorescence staining for EGFP (green) and cytokeratin (red) on unsorted lung cytospins from (D) Z/EG × Cre F1 mice and from (E) Z/EG-into-Cre (representative of 12 mice), (F) Cre-into-Cre, and (G) Z/EG × Cre F1–into-Cre transplants. (H) Immunofluorescence staining for keratin-18 (dark green) and FISH for the X (red) and Y (pale green) chromosomes, showing both XY donor-derived (arrow) and XX recipient epithelial cells. (I) Lung cytospin from a Z/EG-into-Cre transplant. Donor-derived β-galactosidase–positive cells (green) and Pro-SPC–positive cells (red) were identified (arrow). (J) The image shown in (I) with the red channel removed. Scale bars, 20 μm.

We also looked for coexpression of cytokeratin with EGFP by immunofluorescence (Fig. 1, D to G) and for coexpression of cytokeratin with the X and Y chromosomes by fluorescence in situ hybridization (FISH) (Fig. 1H) in cytospins of unsorted cells from the lung digests. EGFP was expressed in the Z/EG × Cre (Fig. 1D) and Z × C F1–into-Cre control mice (Fig. 1G), but not in Z/EG-into-Cre experimental (n = 12 mice) or Cre-into-Cre negative-control recipients (Fig. 1, E and F). In all cases, cytokeratin-positive lung cells contained either two X chromosomes or one X and one Y chromosome, further suggesting that fusion had not occurred (Fig. 1H). The proportion of donor-derived epithelia in this population was 0.2%. From 12 recipient mice, a combined total of 80,000 single and sectioned lung epithelial cells was examined.

To prove that the transgene can be expressed in Z/EG BM-derived pneumocytes, we stained cytospins of lung digests for coexpression of β-galactosidase and Pro-Surfactant C (Pro-SPC) using immunofluorescence. In experimental Z/EG-into-Cre transplants, donor-derived cells were identified that coexpressed β-galactosidase and Pro-SPC (Fig. 1, I and J). Lung digests from control animals (from Z/EG × Cre into Cre and from Cre into Cre) showed no β-galactosidase expression (14).

Although EGFP expression is high after Cre-mediated recombination in hepatocytes of Z/EG × Cre F1 animals, the β-galactosidase transgene is expressed at very low levels in hepatocytes of Z/EG mice (11). Therefore, the Y chromosome alone was used as a marker of BM-derived hepatocytes and the absence of EGFP as evidence for absence of fusion in the liver. By immunohistochemistry, Cre mice transplanted with Z × C F1 BM had EGFP-positive hepatocytes, and Cre mice transplanted with Z/EG BM did not (Fig. 2, A and B). Y chromosome-containing, cytokeratin-positive hepatocytes were identified in all of the transplants, proving that BM-derived hepatocytes developed in the Z/EG-into-Cre transplant recipients (Fig. 2C). Of the more than 36,000 hepatocytes examined in this manner, 0.05% contained the Y chromosome and none expressed EGFP. Therefore, in Cre recipients of Z/EG BM, donor-derived hepatocytes form without undergoing Cre-mediated excision of the Z/EG β-galactosidase cassette. In the positive-control transplants from Z/EG × Cre F1 into Cre mice, ∼0.05% and 0.1% of hepatocytes were marrow-derived according to Y chromosome FISH and EGFP expression, respectively. When estimating engraftment levels by Y FISH, we made no correction for partial sectioning of nuclei and the resulting potential loss of Y chromosomes, which explains the difference in engraftment levels assessed with Y FISH versus EGFP expression.

Fig. 2.

(A and B) Immunohistochemical staining for EGFP on livers of (A) Z/EG × Cre–into-Cre and (B) Z/EG-into-Cre transplants. (C) Immunofluorescence staining with cytokeratin (green) and Y chromosome FISH (red) in the same Z/EG-into-Cre transplant, showing the presence of a donor-derived Y-positive hepatocyte (arrow). (D and E) Immunofluorescence staining of (D) untransplanted positive control (Z/EG × Cre F1) and (E) experimental (Z/EG into Cre) epidermal sections with antibodies against EGFP (green) and cytokeratin AE1/AE3 (red). (F) Immunofluorescence staining with cytokeratin AE1/AE3 (red) and Y chromosome FISH (green), showing the presence of a donor-derived Y-positive keratinocyte (arrow) in the epidermis of a Z/EG-into-Cre transplant recipient.

Engraftment of BMDCs as keratinocytes in the skin was assessed after we induced full-thickness wounds, which were allowed to heal by secondary intention. After 10 and 21 days, the wounds were excised and examined for BMDC engraftment by simultaneous immunofluorescence for cytokeratin, CD45, and EGFP, as well as by Y chromosome FISH (Fig. 2, D to F). Keratinocytes that stained positively for cytokeratin and the Y chromosome and negatively for CD45 were present in the transplanted, wounded animals at a proportion of 0.1% of all keratinocytes. In these cells, no EGFP expression was detected (Fig. 2E).

In order to assess for EGFP expression in a larger number of cells than could be analyzed on tissue sections, reverse transcription–polymerase chain reaction (RT-PCR) for EGFP was performed on a wide range of tissues from Z/EG-into-Cre experimental animals. RT-PCR on RNA from EGFP-positive Z/EG × Cre F1 mice that had been serially diluted into RNA from EGFP-negative Cre animals showed that this technique has the sensitivity to detect one EGFP-positive cell in 106 total cells (12) (fig. S2), which is well within the calculated range of donor-derived epithelial cells of at least 5 in 105 cells. Using this approach, we detected no EGFP mRNA in any of the organs analyzed from the Z/EG-into-Cre transplants (n = 12 mice) (Fig. 3A). The RNA of striated muscle and myocardium from Z/EG-into-Cre animals also did not reveal fusion, which is consistent with observations that fusion in skeletal muscle is negligible when the tissue is not damaged (15).

Fig. 3.

(A) RT-PCR for β-actin (top) and EGFP (bottom) on a range of different tissues from a Z/EG-into-Cre transplant, showing the absence of EGFP expression in any organ. EGFP can be detected in all organs of a Z/EG × Cre–into-Cre transplant (14). Data are representative of 12 transplants. GI, gastrointestinal tract. Pos. Cont., positive control; Neg. Cont., negative control. (B) RT-PCR for β-actin (top) and EGFP (bottom) on various tissues from a ZEG-into-Cre recipient that had received notexin injection to the right TA. The injected muscle contains message for EGFP, indicating that a fusion event occurred. Ctr, control. (C) Immunohistochemistry for EGFP on a negative control (left) and on a ZEG-into-Cre recipient (right) that had received a double injection of snake venom. In the right panel, the binucleate cell in the lower left corner stains positive for EGFP, indicating that it derived from cell-cell fusion.

There are several reasons why EGFP would not be expressed after a fusion event. This could represent loss of the genomic DNA containing the transgene in a reductive division event, failure of adequate Cre expression, or inaccessibility of the Z/EG transgene to Cre recombinase. Therefore, several additional control experiments were performed. To prove that a fusion event would be detected using this approach, we induced fusion in vivo. Cre recipients of Z/EG BM received intramuscular injection of notexin in the tibialis anterior (TA) to induce muscle damage and force a fusion event. Two months after injection, RT-PCR on the injected muscle showed EGFP expression, whereas the contralateral undamaged muscle and other organs showed no evidence of EGFP (Fig. 3B). One mouse received cardiotoxin to one TA, followed with a contralateral notexin injection one month later. One week later, this mouse became ill and was killed. At this early time point, the muscle from the dually injected mouse showed no evidence of myogenic fusion, but it did exhibit rare EGFP-positive liver cells, as detected by both RT-PCR (14) and immunohistochemistry (Fig. 3C). Of ∼8750 hepatocytes examined in this mouse, two were positive for EGFP. One cell was binucleate; the other contained only one nucleus. This proportion of multinucleate cells (50%) was somewhat higher than that noted for wild-type, untransplanted liver (20%), but the number of fused cells (2 of 8750) is too small to allow any definitive conclusion regarding reductive division versus stable fusion. The finding of EGFP-expressing fused cells in this injury model is consistent with the known ability of snake venom to cause profound myocyte (16) and hepatocyte (17, 18) damage. No EGFP was found in any other organ from this mouse.

Adenoviral vectors were used to confirm that (i) the genomic DNA of Z/EG mice is accessible to the Cre recombinase in mature epithelial cells and (ii) functional Cre recombinase is expressed in mature epithelial cells of Cre mice. First, when 4- to 6-week-old Z/EG mice were infected either intranasally or intravenously with an adenovirus-expressing Cre recombinase (19), EGFP-expressing epithelial cells were present in the liver (Fig. 4, A and B) and lung (14). RT-PCR demonstrated EGFP expression in the liver, lung, gastrointestinal tract, skin, and muscle of these mice (14). Second, when adult Cre mice were infected with a “stoplight” adenovirus (20) that will express Ds Red in the absence of and GFP in the presence of Cre, GFP-positive epithelial cells were present in the lung and liver (12) (fig. S3).

Fig. 4.

Immunohistochemistry for EGFP on livers from (A) Z/EG and (B) Cre animals that had been intravenously injected with adenovirus encoding Cre recombinase.

Reports exist supporting (17) and showing absent (1820) or rare (21) cell fusion. There are several possible reasons for the observed inconsistencies. First, severe tissue injury appears to promote cell fusion. This is shown by the presence of EGFP expression only in the mice that were exposed to muscle and/or liver injury other than radiation. The organs of mice that received only the dose of radiation required for myeloablation before BM transplantation did not have detectable fusion events. Second, unlike the other recently published papers in this area (which used whole bone marrow), we performed an immunodepletion that removes mature T-cells. This may remove a fusogenic progenitor that can only form heterokaryons after full marrow engraftment has occurred. Third, variations in detection methods could also account for these differences. Depending on the conditions used, direct fluorescence to detect EGFP to confirm the presence or absence of donor-derived cells can be both insensitive (21) and nonspecific (22, 23), which is why we employed rigorous RNA analysis and immunofluorescence for our EGFP detection protocols. Overall, using FACS, RT-PCR, immunofluorescence, and FISH for chromosomal content, we observed no examples of in vivo cell fusion in organs that were not secondarily injured and that were able to undergo Cre-mediated recombination of the stoplight construct.

The absence of fusion in this model does not necessarily imply that transdifferentiation (a change in phenotype of one mature cell type to that of another mature cell type) has occurred. It may be that an as-yet-unidentified, multipotent epithelial precursor exists in the BM that can differentiate into epithelial cells (24). Further work in this field will help to elucidate these issues. However, on the basis of the data outlined here, we conclude that (notwithstanding pathological states that may induce heterokaryon formation) the epithelial cells that differentiate from BMDCs after irradiation of the recipient are not the result of cell fusion.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article