We read with interest the article by Heinrich et al1 on apparent erythroid-specific expression of green fluorescent protein (GFP) and Cre-recombinase as a result of a “knock-in” mutation into erythropoietin receptor (EpoR) gene. Currently, evaluation of erythroid progenitors relies on the colony-forming assay that is often hampered by difficulty in interpretation, as it depends on both the proliferation and differentiation of the progenitors. In addition, there is a shortage of surface antigens suitable for detecting erythroid progenitors.
To overcome these problems, several lines of mice, in addition to that described by Heinrich et al,1 have been established that express fluorescent proteins in the erythroid lineage. The fluorescent protein transgenes are under different regulatory influences, but all are active in the erythroid lineage. First, the 24-kilobase (kb) β-globin gene fragment was used to direct the expression of a cyan fluorescent protein transgene in the erythroid lineage.2 Second, we exploited the 8-kb GATA-1 gene hematopoietic regulatory domain (G1-HRD) to express GFP and found that erythroid progenitors are efficiently labeled with this method.3 In these studies, through the analysis of GFP-positive cells in the transgenic mice in combination with several known surface markers, we delineated 2 classic erythroid progenitors, CFU-E and BFU-E (erythroid colony-forming unit and burst-forming unit, respectively), as 2 discrete subfractions (late erythroid progenitor [LEP] and early erythroid progenitor [EEP]) separable by the flow cytometer.3 Third, as noted above, Heinrich et al established a knock-in mouse line that expresses GFP-Cre fusion protein under the EpoR gene regulatory influence.1
How distinctly do these reporter transgenes mark the erythroid lineage? We believe that detailed assessments of their expression profiles are crucial for the use of these transgene products as internal erythroid markers. For instance, GATA-1 is expressed in the erythroid, megakaryocytic, eosinophilic, and mast cell lineages as well as Sertoli cells.4 Our analyses also unveiled that G1-HRD is active in the erythroid and megakaryocytic lineage cells, but not in Sertoli cells.4 In contrast, Heinrich et al reported that GFP-Cre transgene from the EpoR locus is expressed specifically in the erythroid lineage,1 even though the endogenous EpoR gene is expressed in a variety of tissues.5 In addition, whereas EpoR is expressed in both embryonic and hematopoietic stem cells,6 the cells harboring Cre-mediated gene recombination were detected only in the erythroid lineage and embryonic vasculature, suggesting that valid Cre expression may start from the erythroid progenitor stage.1 The precise reason for the apparent acquirement of erythroid specificity is unknown as expression of EpoR has been noted in other tissues.1,7 While we recognize that the GFP-Cre mouse line could be very useful for the study of erythropoiesis, the precise expression profiles of GFP-Cre, especially those in megakaryocytes and embryonic cells, need to be clarified in more detail before claiming strict erythroid specificity. With progress in these areas, we may be able to approach novel concepts on the mechanisms of erythropoiesis in the near future.
Tissue specificity and transgene expression
The letter by Suzuki et al summarizes different mouse lines that have been established to visualize cells of the erythroid lineage. Most of these lines contain randomly integrated transgenes that may lead to unstable or incomplete integration of transferred DNA, site-of-integration effects, and founder strain-dependent variations in transgene expression. Recently, an analysis of 2 mouse lines independently generated by random integration of a GATA-1-controlled Cre-recombinase transgene revealed conflicting results regarding tissue specificity of transgene expression.1,2 Position-effect variegation and transgene regulation were shown to differ in a strain- and background-specific manner, dependent on the chromosomal boundaries, their epigenetic regulation, and chromatin structure.3-5 In addition, multicopy insertions of lacZ-transgenes transcriptionally regulated by α-globin gene fragments hampered efficient transgene expression due to mosaic expression and silencing effects.6-8
Unlike the mouse models that use randomly integrated transgenes under transcriptional control of GATA-1 or β-globin gene fragments,9,10 our ErGFP-Cre model uses a defined knock-in strategy to direct fluorescent protein expression to the erythroid lineage. The defined knock-in of a GFP-Cre-encoding transgene into the endogenous erythropoietin receptor (EpoR) gene locus ensures reliable EpoR promoter-controlled transgene expression and, in addition, allows the generation of EpoR knock-out mice that accumulate green fluorescent protein (GFP)-positive erythroid progenitor cells of probably erythroid colony-forming unit (CFU-E) stage.11
In all models, including ours, tissue specificity was analyzed by flow cytometry and fluorescence microscopy, leading to a snapshot analysis of fluorescent protein expression at a distinct time point of mouse development. We examined GFP-Cre expression in different hematopoietic subpopulations of adult and embryonic ErGFP-Cre mice by flow cytometry, using an antibody cocktail staining nonerythroid hematopoietic cells11 or, in a more precise analysis, using single antibodies, and we observed a strictly erythroid-specific expression pattern of GFP-Cre.
To claim strict tissue specificity, the analysis of the transgene expression should also include tissue-specific expression over time and embryogenesis. An advantage of our ErGFP-Cre mouse model is that by crossing the mice with R26R-reporter mice, GFP-Cremediated LacZ expression can be induced and used as an indicator for previous or persistent GFP-Cre expression. The spatial and temporal analysis revealed that nonhematopoietic expression of GFP-Cre is restricted to the vascular system and confirmed that within the hematopoietic system GFP-Cre expression is limited to the erythroid lineage.
Correspondence: e-mail: U.Klingmueller@dkfz-heidelberg.de.
