Transgenic mice that uniformly express readily detectable markers in all cells of all tissues are an essential tool for marking cells for transplantation studies, whole animal chimera experiments, and cell lineage analyses. Several lines of mice created for this purpose partially fulfill the criteria of pancellular expression in some but not all tissues, making it potentially difficult to identify the origin of all cells in transplant recipients. We have generated ROSA26EGFP transgenic mouse lines expressing EGFP under control of a 187-kb BAC containing the murine ROSA26 locus that have several advantages over existing models. In our experience, using homologous recombination in Escherichia coli
is an efficient and rapid approach for generating transgenic mice, as it enables generation of BAC transgenes ready for pronuclear microinjection in 3 weeks. The identified murine BAC appears to contain sequences sufficient to recapitulate the generalized expression directed by the endogenous ROSA26 locus as originally reported (Friedrich and Soriano, 1991
; Zambrowicz et al., 1997
The ROSA26-EGFP transgene is efficiently and uniformly expressed in the differentiated progeny of transplanted HSCs, making their identification unambiguous, and this expression is much more faithful than that conferred by HSCs from CAG-EGFP donors. It is difficult or impossible to establish the origin of leukocytes in recipients of CAG-EGFP hematopoietic stem cells, owing to the much lower percentage of cells expressing EGFP above the background level. Only a fraction of the major lymphoid lineages were EGFP positive in the recipients of CAG-EGFP bone marrow, which is consistent with the variegated EGFP expression in splenic tissue of CAG-EGFP mice. The presence of high percentages of CD11b+
EGFP expressing cells 4 months after engraftment in ROSA26-EGFP recipients suggests that EGFP is expressed in long-term hematopoietic stem cells that give rise to a relatively uniform population of descendant lineages (Lagasse and Weissman, 1994
; Morrison and Weissman, 1994
; Wright et al., 2001
In contrast to recipients of CAG-EGFP mice, most (95%) myeloid cells in recipients of ROSA26-EGFP mice showed at least 10- to 100-fold more EGFP fluorescence than control donor cells. Analysis of donor ROSA26-EGFP lymphocytes also showed a small fraction of EGFP−
cells (Supplemental Fig. 1
). For each of the lineages examined, the fraction of EGFP+
cells was slightly higher in donor mice than the BMT recipients. Likewise, a much higher percentage of CAG-EGFP lymphocytes were EGFP−
compared with the ROSA26-EGFP mice (38% vs. 14%, not shown). Thus the small fraction of lymphoid and myeloid cells in ROSA26-EGFP recipients showing a lower level EGFP expression may result from a combination of factors including low level variegated expression of EGFP, rare cells that survived irradiation, rare nondonor cells from the recipient that bind to lineage specific antibodies used for FACS, or a moderate engraftment defect related to histoincompatibility. Although the ROSA26-EGFP donors were not fully backcrossed to the recipient background, any hybrid resistance would favor engraftment and stable expression by the CAG-EGFP donor cells, which are congenic with their recipient. Therefore, the ROSA26-EGFP mice are superior donors for any experiments that track hematopoietic cell fate in transplant recipients.
The inability to achieve the desired pattern of expression for a given transgene is a common occurrence. In some cases, relatively short transgenes failed to include distant yet essential transcription regulatory sequences required for cell-type specific expression, leading to either misexpression in inappropriate cells or loss of expected expression in other cells. A second problem encountered with short transgenes is their susceptibility to the effects of chromosomal integration, where enhancers or silencers in neighboring genes modify expression (Ogbourne and Antalis, 1998
). Integration site effects often lead to marked differences in expression patterns between independent pedigrees because of insufficient sequence to insulate the transgene from effects of nearby genes. Position effect variegation, with variable rather than uniform expression in similar cells within a single tissue, represents another effect of integration site on expression patterns. The commonly used CAG-EGFP mouse, which expressed EGFP in a short transgene, exhibits extensive variegation. Another transgene containing only 0.8 kb of the ROSA26 locus (Kisseberth et al., 1999
) directed widespread EGFP expression that was uniform in a number of tissues. However, some tissues, including liver and intestine, showed considerable position effect variegation in these mice.
Insertion of a reporter into the ROSA26 locus by homologous recombination in ES cells to generate a “knock in” mouse is one approach that ensures widespread expression because of the presence of the native chromosomal sequences extending millions of bp flanking the locus. Several groups have knocked-in genes expressing fluorescent proteins, such as EGFP, EYFP, and ECFP, into the ROSA26 locus to improve the general utility of this reporter line (Mao et al., 2001
; Srinivas et al., 2001
). Anecdotal reports suggest that knocking EGFP into endogenous loci frequently results in insufficient EGFP expression for direct visualization in tissue sections (Gong et al., 2003
). A single copy of the ROSA26-EGFP BAC transgene universally expresses EGFP at levels allowing visualization of endogenous EGFP fluorescence in organ whole mounts as well as in thin tissue sections and live cells. The ROSA26-EGFP transgenic mice described here show an approximately 100-fold increase in EGFP fluorescence in lymphocytes compared with nontransgenic mice with minimal overlap in intensity between the two. FACS analysis of EGFP expression in lymphocytes from a mouse with EGFP knocked-in to the ROSA26 locus showed a less than one log increase in fluorescence with substantial overlap between transgenic and nontransgenic lines (Mao et al., 2001
). These results suggest that EGFP expression in lymphocytes of the ROSA26-EGFP BAC mice is considerably higher and easier to distinguish from background than lymphocytes from the previously described “knock in” mice (Mao et al., 2001
). The high level of EGFP fluorescence seen with both pedigrees, compared with the knock-in, suggests that distant inhibitory elements in the ROSA26 locus that reduce transcription of endogenous ROSA26 gene were not included in the transgene. The loss of these regulatory elements appears to enhance the expression of EGFP while enough sequence is still present in the BAC to accurately reproduce the generalized expression of the endogenous gene (Friedrich and Soriano, 1991
In addition to the use of the ROSA26 locus as a reporter line for transplantation and cell lineage analysis, mice with conditional ROSA26 alleles represent a powerful approach to activate expression of any gene in specific cell types at different stages of development. A gene inserted into the ROSA26 BAC with upstream floxed transcription termination sequences can be expressed as desired with the appropriate Cre deleter strain. Rapid production of transgenes in the ROSA26 BAC by recombineering as described here is potentially a new, highly efficient approach for generating lines of mice expressing a variety of proteins in any tissue desired.