|Home | About | Journals | Submit | Contact Us | Français|
The derivation of the primitive endoderm layer from the pluripotent cells of the inner cell mass is one of the earliest differentiation and morphogenic events in embryonic development. GATA4 and GATA6 are the key transcription factors in the formation of extraembryonic endoderms, but their specific contribution to the derivation of each endoderm lineage needs clarification. We further analyzed the dynamic expression and mutant phenotypes of GATA6 in early mouse embryos. GATA6 and GATA4 are both expressed in primitive endoderm cells initially. At embryonic day (E) 5.0, parietal endoderm cells continue to express both GATA4 and GATA6; however, visceral endoderm cells express GATA4 but exhibit a reduced expression of GATA6. By and after E5.5, visceral endoderm cells no longer express GATA6. We also found that GATA6 null embryos did not form a morphologically recognizable primitive endoderm layer, and subsequently failed to form visceral and parietal endoderms. Thus, the current study establishes that GATA6 is essential for the formation of primitive endoderm, at a much earlier stage then previously recognized, and expression of GATA6 discriminates parietal endoderm from visceral endoderm lineages.
After fertilization, mouse eggs are activated and subsequently go through cell division, compaction, and hatching, to form blastocysts composed of epiblast, trophectoderm, and primitive endoderm cells when reaching the uterine endometrium (Gardner, 1989; Beddington and Robertson, 1999; Rossant, 2004; Rossant and Tam, 2004; Pfister, et al., 2007). Immediately after implantation onto uterine wall, a morphologically recognizable primitive endoderm layer covering the epiblast is present (Gardner, 1983). Later at approximately E4.75, a parietal endoderm is extended to cover the blastocoel and a visceral endoderm, that is morphologically distinct from the prior primitive endoderm, is established to cover what is now ectoderm. Based on morphological observations, a scheme was suggested that visceral and parietal endoderm were derived from a common origin, the primitive endoderm (Gardner, 1982). However, several observations from experimenting with isolated inner cell mass (ICM) grown in vitro led to the conclusion that the primitive ectoderm retains the ability to differentiate into a visceral endoderm-like lineage (Dziadek, 1979). Thus, an alternative model argues that two distinct phases of extraembryonic endoderm differentiation take place, and visceral and parietal endoderm are derived separately without sharing the primitive endoderm as a common precursor.
A recent addition to the understanding of primitive endoderm formation is that primitive endoderm cells are first derived within the interior of ICM and subsequently migrate and position on the surface to form the primitive endoderm layer (Rossant et al., 2003; Chazaud et al., 2006; Rula et al., 2007). This sorting model significantly differs from the original view that the superficial cells of the ICM are designated to become primitive endoderm cells (Gardner, 1983). Moreover, the sorting model was further supported by studies showing that surface positioning of primitive endoderm cells requires Dab2, an endocytic adaptor protein, because both primitive and parietal endoderm cells are trapped in the interior of Dab2 null embryos (Yang et al., 2002, 2007). Simulation of cell movement and migration in embryoid bodies suggests that both primitive endoderm cells and undifferentiated ICM cells move dynamically without directionality in the ICM, though differentiated cells eventually become fixed on surface upon their arriving (Rula et al., 2007). The positioning of the primitive endoderm cells is due to the ability of the endoderm cells to generate an apical–basal polarity mediated by Dab2 (Yang et al., 2007). The presence of parietal endoderm cells in the mound of disorganized cells of Dab2-null embryos suggests that the parietal endoderm cells can be derived without attaching to the blastocoel surface (Yang et al., 2007). The primitive endoderm cell sorting model provides a parallel scenario of what may be possible for the scheme proposed by Dziadek (1979), that differentiation and subsequent sorting to form primitive and parietal endoderm occurs first, and the primitive endoderm further matures to become visceral endoderm, while the parietal endoderm cells migrate to cover the blastocoel.
GATA4 and GATA6 transcription factors are markers of extraembryonic endoderm lineages (Arceci et al., 1993; Morrisey et al., 1996), and GATA6 was reported to be essential for visceral endoderm development (Morrisey et al., 1998; Koutsourakis et al., 1999). In cell culture studies, transfection and expression of either GATA4 or GATA6 are sufficient to down-regulate Oct-3/4 expression and induce primitive endoderm differentiation of embryonic stem (ES) cells (Fujikura et al., 2002). GATA4 and GATA6 also can induce the expression of each other in ES cells (Fujikura et al., 2002; Capo-Chichi et al., 2005). Although GATA4-null ES cells fail to undergo spontaneous endoderm differentiation upon aggregation in embryoid bodies, addition of retinoic acid is sufficient to induce the GATA4-null cells to differentiate into endoderm (Soudais et al., 1995; Bielinska and Wilson, 1997). Correspondingly, extraembryonic endoderm is present in GATA4-null embryos (Kuo et al., 1997; Molkentin et al., 1997). Nevertheless, the extraembryonic endoderm is defective in GATA4-null embryos: over a third of the GATA4−/− embryos arrest before or during gastrulation, before heart formation (Molkentin et al., 1997). The rest of the GATA4-deficient embryos die between embryonic day (E) 8.5 and E10.5, displaying severe defects in heart development, which are a secondary consequence of extraembryonic deficiencies that disrupt gastrulation. GATA4-positive extraembryonic endoderm can rescue cardiogenesis in GATA4-deficient embryos, presumably through a non–cell-autonomous mechanism (Narita et al., 1997; Watt et al., 2004).
The analysis of the differentiation properties of GATA-deficient ES cells led to the conclusion that GATA4 is required for cell aggregation-induced endoderm differentiation, and GATA6 is necessary for both aggregation and retinoic acid-induced endoderm differentiation (Capo-Chichi et al., 2005). Overall, the inability of GATA6-null ES cells to undergo endoderm differentiation in vitro in monolayer culture and in embryoid body model also suggests a critical role for GATA6 in the formation of extraembryonic endoderm lineages. In embryoid bodies derived from GATA6-null ES cells, endoderm-like cells are completely absent (Capo-Chichi et al., 2005); however, the affected lineage in GATA6-deficient embryos, whether the formation of primitive, visceral, or parietal endoderm, has not been clearly worked out (Morrisey et al., 1998; Koutsourakis et al., 1999). Further more, even the expression pattern of GATA factors in early embryonic lineages is not fully documented (Pfister et al., 2007), especially the information is lacking in implanted blastocysts. In the current report, we further analyzed the expression pattern and role of GATA6 in lineage determination of early mouse embryos.
GATA transcription factors are important in the derivation of the early lineages in blastocysts, but their expression patterns have not been unambiguously determined, especially in blastocysts implanted in uterine. Using immunostaining with a GATA6 antisera developed recently, we investigated the GATA4 and GATA6 expression in newly implanted blastocysts at the proximity of E4.5 stages. In immediately implanted blastocysts defined by that only one side of the blastocyst is in contact with the uterine epithelium, a primitive endoderm has formed and covers the ICMs (Fig. 1A). These primitive endoderm cells express both GATA4 and GATA6 strongly (Fig. 1). In some E4.5 embryos, a gap consisting of GATA4- and GATA6-negative cells was observed in the primitive endoderm epithelial structure covering the surface of the ICM, indicating the progressive formation of cell–cell adhesion and the epithelium, as also observed in embryoid bodies (Rula et al., 2007). Nevertheless, a complete epithelial structure was consistently seen at a slightly later stage, E4.75, which is judged by the enclosing of the blastocysts into the uterine linings over all sides (Fig. 1B). At E4.75, the GATA4-and GATA6-positve cells from the completed lining of primitive endoderm start to extend outside the boundary of the ICM area onto the surface of the blastocoel (Fig. 1B). Subsequently, at around E5.0, the early embryos exhibit completed GATA4- and GATA6-positive visceral and parietal endoderms (Fig. 1C). Thus, primitive endoderm cells, and later, visceral and parietal endoderm cells in peri-implantation embryos appear to express both GATA4 and GATA6.
At E5.5, E6.5, E7.0, the egg cylinder stages of the mouse embryos, the extra-embryonic endoderms are matured (Lu et al., 2001; Tam and Loebel, 2007). The parietal endoderm cells are stretched and flatten to cover the surface of the blastocoel and are responsible for the production of the thick layer of Reichert's membrane. The primitive endoderm layer is replaced with a morphological distinctive visceral endoderm, which is expanded progressively to surround the developing ectoderm and mesoderm (Fig. 2, arrow). The columnar visceral endoderm cells could be identified by the presence of abundant large apical vacuoles. Parietal endoderm cells can be distinguished by strong positive staining for laminin (Fig. 2A,C), while visceral endoderm cells can be identified based on positive Dab2 staining (Fig. 2C). Both the matured parietal (arrowhead) and visceral endoderm (arrow) cells are positive for GATA4 staining (Fig. 2). Without exception, the visceral endoderm cells are GATA6-negative (or very weak) (Fig. 2, arrow). In contrast, the parietal endoderm cells are positive for both GATA4 and GATA6 at all stages (Fig. 2, arrowhead). Thus, we conclude that primitive and parietal endoderm cells express both GATA4 and GATA6, and matured visceral endoderm cells express GATA4 but contain no (or very little) GATA6 protein.
The fact that matured visceral endoderm cells express only GATA4 but not GATA6 contradicts the in vitro observation that GATA4 can induce GATA6 expression and vice versa (Fujikura et al., 2002; Capo-Chichi et al., 2005). Thus, we analyzed further the co-expression of GATA4 and GATA6 in populations of differentiated ES cells in vitro. ES cells were differentiated with retinoic acid in cultures, upon examining of the expression of GATA4 and GATA6 by immunofluorescence double labeling, we found a fraction (30–50%) of differentiated cells expressed only GATA4 but not (or, very weakly) GATA6, as shown by two examples (Fig. 3, arrows indicate GATA4 positive and GATA6 negative cells).
Consistently, in embryoid bodies, all surface endoderm cells express GATA4, but a fraction of these endoderm cells shows little or no GATA6 expression (Fig. 4A, arrows). In further histological examination, the GATA4-positive and GATA6-negative endoderm cells exhibit a morphology resembling visceral endoderm: these cells are columnar and exhibit extensive apical vacuoles (Fig. 4B, arrow); while the double positive cells are flat and associated with a thick, PAS-stained basement membrane underneath, resembling parietal endoderm cells (Fig. 4B, arrowhead). Therefore, in monolayer cultures and embryoid bodies derived from ES cells, expression of GATA6 distinguishes two types of endoderm cells: visceral endoderm like cells are GATA6-negative, and parietal endoderm like cells are GATA6-positive. The results also indicate that GATA4 can be expressed independently without inducing GATA6 expression.
It has been shown that GATA6-deficient ES cells fail to undergo extraembryonic endoderm differentiation in embryoid bodies or in monolayers treated with retinoic acid (Capo-Chichi et al., 2005). Thus, we sought to test whether GATA6-null ES cells could be converted to visceral endoderm-like cells that are GATA6-negative. Transfection and expression of GATA4 in GATA6-null ES cells did not induce significant expression of Dab2, an extraembryonic endoderm marker (Capo-Chichi et al., 2005). Nevertheless, when these transfected cells were allowed to aggregate to form spheroids, Dab2 induction was drastically increased as detected by Western blot (Fig. 5A) and immunofluorescence microscopy (Fig. 5B). Thus, under the circumstance of GATA4 transfection/expression and cell aggregation, GATA6 is not required for differentiation of ES cells into visceral endoderm-like cells, that are Dab2-positive. These experiments support the observation that GATA6 is absent in matured visceral endoderm cells of the mouse embryos.
Given the expression pattern of GATA6 in the early lineages of mouse embryos we observed, we next addressed the potentially different roles of GATA6 on primitive, parietal, and visceral endoderm formation. Thus, we revisited the phenotype of GATA6 null embryos by systematically analyzing E4.5 to E6.5 embryos from timed matings of GATA6 (+/−) mice. Previous studies had used in situ hybridization as the genotyping assay for GATA6 in early mouse embryos from intercrosses between GATA6 (+/−) parents (Morrisey et al., 1998). In the current study, we distinguished GATA6 null from wild-type and heterozygous E4.5 to E6.5 embryos using immunostaining, which we believe is more convenient and accurate. Out of the approximately forty E5.5 embryos from timed matings of GATA6 (+/−) mice, 7 putative GATA6-null E5.5 embryos were identified, with all the rest of the littermates identified as wild-type or heterozygous because of positive GATA6 staining in the parietal endoderm cells (and GATA4 staining in the visceral endoderm cells). As shown by an example (Fig. 6A), these mutant embryos exhibit Oct-3/4–positive embryonic cells, but show no recognizable endoderm structure or any cell organization pattern, contain no GATA4-positive parietal and visceral endoderm cells, and show essentially no expression of the endoderm marker Dab2. All of the putative mutant embryos were sectioned through entirely to ensure that the middle and main portion of the embryos were represented. Several presumptive GATA6-null E6.5 embryos were also observed, as an example shown in Figure 6B. These abnormal E6.5 embryos appear slightly bigger in size than the E5.5 GATA6-null embryos, but are similar to the E5.5 embryos for the lack of GATA4 and Dab2 expression and lack of cell organization.
At E7.0, following analysis of five litters of embryos obtained from GATA6 (+/−) intercrosses, all the embryos were found to contain GATA6 positive parietal endoderm. No identifiable GATA6-null embryos were recovered, and remains of embryos that were undergoing re-absorption were found in the approximate ratio of 1:4, as expected to be the GATA6-deficient embryos.
These studies substantiate that GATA6 is essential for the generation of both parietal and visceral endoderms, and no differentiation and endoderm structure are present in GATA6-null embryos.
Because we did not detect the presence of parietal or visceral endoderms in GATA6-null embryos, we further examined the impact of GATA6 deletion on primitive endoderm in an earlier stage, at E4.5. Uteri were collected from time-mated of GATA6 (+/−) parents, and the uteri were sectioned through to identify all the potential E4.5, immediately implanted blastocysts. Out of 40 immediately implanted blastocysts (E4.5) collected from matings between GATA6+/− mice, we identified 6 putative GATA6-null embryos that contain no GATA6-or GATA4-positive cells in the embryos, as an example shown here (Fig. 7A). Each putative mutant embryo was found together with morphological wild-type, GATA4- and GATA6-positive cell-containing blastocysts in the same uterus, indicating the absence of GATA4 and GATA6 staining in the putative mutant embryos was not a technical irregularity. The putative GATA6-null implanting blastocysts also contain Oct-3/4–positive cells without a covering GATA4- and GATA6-posititve primitive endoderm layer (Fig. 7A). Most of the remaining implanted blastocysts of the same litters, however, contain a complete or partial GATA4- and GATA6-positive primitive endoderm layer covering several Oct-3/4-positive cells of the ICM (Fig. 7B). A few embryos were not unambiguously identified to be either wild-types or mutants due to limited materials available or the poor qualities of the sections.
Thus, no primitive endoderm cells or recognizable primitive endoderm layer are present in GATA6-null embryos. We conclude that GATA6 is essential for the derivation of primitive endoderm cells and formation of the primitive endoderm structure.
In a detailed analysis of the expression of GATA4 and GATA6 in a large number of early embryos of E4.5 to E7.0 covering small incremental developmental stages after implantation, we resolved with high confident a pattern and sequence of GATA4 and GATA6 expression in early mouse embryonic development. At the E4.5 stage, newly originated primitive endoderm cells are strongly positive for both GATA4 and GATA6. Subsequently, GATA4- and GATA6-positive endoderm cells migrate and form the parietal endoderm layer covering the blastocoel. Throughout these stages, parietal endoderm cells continue to be GATA4- and GATA6-positive. At E4.75 to E5.0 stages, the endoderm cells in contact with cells of the ICM exhibit strong GATA4 expression but a weakened GATA6 expression. By E5.5 and after, the matured visceral endoderm cells become GATA6-negative, though they remain positive for GATA4. Thus, we conclude that the expression of GATA6 is a distinguishing marker for parietal (GATA6-positive) and visceral (GATA6-negative) endoderm.
From detailed analysis of GATA6 knockout embryos, we also conclude that GATA6 is required for the generation of primitive endoderm, and GATA6-null embryos do not exhibit any extraembryonic endoderm markers or structure, which is consistent with the inability of GATA6-null ES cells to undergo endoderm differentiation in vitro.
GATA6 deficient mice were reported to exhibit a defective visceral endoderm without gross morphological abnormality; however the details were not characterized (Morrisey et al., 1998; Koutsourakis et al., 1999). This current study shows that all extraembryonic lineages—primitive, parietal, and visceral endoderm, fail to develop and no recognizable structure is present in GATA6-null embryos. Particularly, the defect is obvious as early as the formation of the primitive endoderm at E4.5 stage, which is a stage earlier than the previously noted for the GATA6 knockout phenotype (Morrisey et al., 1998; Koutsourakis et al., 1999). We attribute that the current use of immunostaining as a genotyping assay rather than the more technically challenging in situ hybridization assay used in previous studies is more accurate and enables us to easily identify GATA6-null embryos with more certainty.
The current findings that GATA6 knockout embryos fail to generate primitive, parietal, and visceral endoderm are consistent with the inability of GATA6-null ES cells to undergo endoderm differentiation in embryoid body models. Thus, the embryoid body model emulates well the requirement of GATA6 in extraembryonic endoderm development.
The new data might appear to contradict the model that parietal and visceral endoderm cells are derived from a common parental lineage, the primitive endoderm cells, because the parietal and visceral endoderm cells have different GATA4 and GATA6 expression patterns. One possibility is that primitive endoderm cells differentiate into parietal endoderm cells that migrate to cover the blastocoel, and visceral endoderm is generated from the ICM to replace the primitive endoderm (Dziadek, 1979). However, a problem of this model is to explain how the deficiency of GATA6 effects the differentiation of GATA6-negative visceral endoderm cells. A second possibility is that the requirement of GATA6 for visceral endoderm differentiation is non–cell-autonomous: the GATA6-positive primitive or parietal endoderm cells may be required to induce the differentiation of visceral endoderm cells. To test this idea, we have designed a “priming” experiment in which we mixed a small fraction (range of 10–30% were tested) of retinoic acid differentiated wild-type ES cells with GATA6-null ES cells to form spheroids. We then determined if GATA6-null ES cells might differentiate into visceral endoderm-like cells, presumably induced by the differentiated wild-type ES cells. In multiple trials, all the differentiated cells were of wild-type origin, and we did not detect any differentiation of GATA6-null ES cells (data not shown). Thus, these experiments did not support a non–cell-autonomous activity of GATA6 in inducing visceral endoderm differentiation.
The other possibility is that the primitive endoderm gives rise to both parietal and visceral endoderm (Gardner, 1982): a portion of the GATA4- and GATA6-positive primitive endoderm cells further differentiate and migrate to become parietal endoderm; the rest of the primitive endoderm cells remain stationed but gradually lose GATA6 expression while maintain GATA4 expression and convert into visceral endoderm cells. Indeed, we did observe that in a fraction of the E5.0 embryos, the visceral endoderm cells exhibit a variable level of GATA6 expression (Fig. 8). An example of an E5.5 embryo shows the presence of both stronger (arrow) and weaker (arrowhead) GATA6 staining of the visceral endoderm cells in the same embryos (Fig. 8). These embryos may represent intermediates during the maturation of visceral endoderm cells when GATA6 is gradually reduced.
The additional characterization of the expression pattern of GATA factors and the phenotype of GATA6-null embryos permit a new understanding of the lineage derivation in the early embryos. We conclude that visceral endoderm cells have a different expression pattern of GATA6 from that of primitive endoderm and parietal endoderm. Consistently, we showed that GATA6-null ES cells could be converted to a visceral endoderm like cell lineage, supporting that visceral endoderm cells can be maintained in the absence of GATA6. However, GATA6-null ES cells and embryos fail to develop any extraembryonic endoderm lineages because that differentiation of primitive endoderm is GATA6 dependent. Thus the GATA6-positive primitive endoderm is likely a precursor of both parietal and visceral endoderm. Specifically, the formation of visceral endoderm requires GATA6 because the cells develop from the GATA6-positive precursors, the primitive endoderm cells.
Thus, a scheme for the lineage derivation in postimplantation blastocysts can be postulated (Fig. 9). First, induction of GATA4 and GATA6 expression leads to the origination of primitive endoderm in the ICM, and subsequently the cells are sorted to the surface to form the primitive endoderm structure. Afterward, a fraction of the primitive endoderm cells migrate to cover the blastocoel where these cells further differentiate into parietal endoderm, retaining the expression of both GATA4 and GATA6. Lastly, the remaining primitive endoderm cells covering the ICM gradually lose GATA6 but retain GATA4 expression and mature into visceral endoderm. The model may enable us to explore further the roles of GATA factors in defining characteristics of each early endoderm lineages, which have distinct instructive signaling and regulatory mechanisms in the sequential lineage derivation and pattern formation of early embryos.
Two mating pairs of GATA6 (+/−) mice obtained from the University of Pennsylvania (Morrisey et al., 1998) was introduced to and maintained by inbreeding at the animal facility of Fox Chase Cancer Center since 2005. To genotype the mice, DNA was extracted from tail fragments and analyzed for GATA6 gene by Southern blot as described (Morrisey et al., 1998) or for PCR amplification of neo transgene. Neo positive mice are identified as GATA6 (+/−) and neo negative as wild-types, For early embryos (earlier than E6.5) from timed mating of GATA6 (+/−) mice, genotyping was performed by examining the morphology and/or GATA6 immunostaining of the embryos. GATA6 (+/+) or (+/−) and (−/−) genotypes were distinguished by immunostaining of GATA6 in neighboring sections. GATA6-positive littermates in the same uterus were used as positive controls. Suspected mutant embryos were sectioned through and a slide from every three sections was stained for GATA6 to ensure the accuracy of identification as GATA6 (−/−).
Generally, cells were cultured on glass coverslips or in chamber slides and analyzed using CCD or laser-scanning confocal microscopy. DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) or propidium iodide were used as nuclear markers. Immunofluorescence staining was described in details previously (Capo-Chichi et al., 2005; Yang et al., 2007). A Nikon Eclipse E 800 epifluorescence microscope with ×60 oil immersion objective linked to a Roper Quantix CCD (charged coupled device) camera was used to examine the slides. A Nikon Eclipse E800 fluorescence microscope with ×60 water immersion objective linked to a Bio-Rad Radiance 2000 LSCM (laser scanning confocal microscope) camera was also used for observation and image acquisition.
Lines of RW-4 wild-type, GATA6 null, or GATA4 null embryonic stem cells were maintained in ES medium with LIF on a layer of irradiated mouse embryonic fibroblasts as previously described (Capo-Chichi et al., 2005). Embryoid bodies were made by culturing ES cells in suspension in nonadhesive Petri dishes for 5–7 days, with daily changes of fresh medium by a brief centrifugation to collect cell aggregates. At the end of the experiments, the embryoid bodies were harvested by brief centrifugation and used for biochemical or histological analysis.
After timed matings, uteri containing embryos at different developmental stages, E4.5 to E7.5, were collected. The tissue samples were formalin-fixed and paraffin-embedded. Sections (5 μm) were cut and adhered to positively charged slides (Fisher). Antibodies used were: anti-GATA4 and anti-GATA6, polyclonal rabbit antibodies from Santa Cruz Biotechnology, Inc.; anti-Dab2, monoclonal mouse antibodies from BD Transduction Laboratories; anti-Oct-3/4, monoclonal mouse antibodies from Santa Cruz Biotechnology, Inc.; anti-laminin, polyclonal rabbit antibodies from Sigma. An anti-GATA6 rabbit polyclonal antiserum was also generated and used. The antibodies were generated against a peptide sequence SQGPA AYDGA PGGFV H from human GATA6 transcription factor (amino acids 10 to 25 of accession U66075.1 or amino acids 156 to 171 of accession NM_005257.3) that is common to both human and mouse. The anti-serum was characterized and shown to be specific for human or mouse GATA6 protein. In Western blot, the anti-GATA6 antibodies recognized a single GATA6 protein in lysate of retinoic acid differentiated mouse embryonic stem cells. The antibodies also recognized GATA6 protein in 293 cells transfected with mouse GATA6 but not GATA4 cDNA. Standard hematoxylin–eosin, PAS staining, and immunohistochemical staining, Western Blotting, etc, were applied in experiments described in this manuscript as detailed in previously published papers (Capo-Chichi et al., 2005; Yang et al., 2002, 2007).
We deeply appreciate the generosity of Drs. Edward Morrisey and Michael Parmacek of the University of Pennsylvania for providing the GATA6 knockout mouse line and agreeing to let us re-examined the mutant embryonic phenotype of the mice. We acknowledge the excellent technical assistance from Jennifer Smedberg and Cory Staub. We appreciate the assistance and contribution of Xiang Hua of the Fox Chase Cancer Center transgenic mouse facility, Tony Lerro and Jackie Valvardi of the Fox Chase Cancer Center animal facility, Sharon Howard from cell culture facility, Cass Renner and Fangping Chen of the Fox Chase Cancer Center pathology facility, and Dr. Sandra Jablonski for her help with immunofluorescence microscopy. We appreciate very much of our colleagues Drs. Elizabeth Smith and Robert Moore for their suggestions, comments, reading, and editing during the course of the experiments and preparation of the manuscript. X.X.X. was funded by NCI, NIH.
Grant sponsor: NCI, NIH; Grant number: R01 CA095071; Grant number: R01 CA79716; Grant number: R01 CA75389.