BMC Genomics 2008, 9:591http://www.biomedcentral.com/1471-2164/9/591Identification <strong>of</strong> differentially regulated transcriptionfactors within the <strong>mammary</strong> epitheliumTranscriptional regulators are key cell-intrinsic factors inlineage selection and cell fate decisions in stem – differentiatedcell hierarchies [52,53]. Therefore, the gene expressionarray data set was interrogated to identifydifferentially expressed factors which may regulate transcription[see Additional file 19]. The expression <strong>of</strong> a subset<strong>of</strong> these was analysed by qPCR (Figure 6B). As with thepotential paracrine interactions, the basal/myo<strong>epithelial</strong>cells differentially expressed many more transcriptionalregulators (fifty-two) than the other two populations. TheER - luminal cells differentially expressed twenty-one transcriptionfactors but only seven transcription factors weredifferentially expressed in the luminal ER + cells. Threetranscription factors were found to be expressed in boththe basal/myo<strong>epithelial</strong> and luminal ER - populations, onewas common to both the basal/myo<strong>epithelial</strong> and luminalER + cells and two were found in all three populations.In common lineage progenitors, functional mutualrepression and auto-stimulation by transcription factorscan facilitate bilineage cell fate decisions [54]. Once thecell fate decisions have been executed, continued function<strong>of</strong> different subsets <strong>of</strong> the factors active in the progenitorsare required to maintain the different differentiated celllineages in stable fates [53]. Thus, modelling interactionsbetween lineage-specific transcription factors can elucidatecell fate decision processes occurring in progenitorcells. Therefore, to begin to understand how interactionsbetween lineage specific transcription factors may influencecell fate choices in the <strong>mammary</strong> epithelium, aninteraction network was built. Transcription factors identifiedas lineage-specific but for which no interaction dataexists were excluded. The resulting network (Figure 7) wascentred around two main hubs, namely Myc and Esr1, suggestingthat they are key regulators <strong>of</strong> <strong>mammary</strong> cell fate[55,56]. Clearly, a great deal <strong>of</strong> information exists on theirinteractions and this would bias any interaction networkaround these two key nodes. Nevertheless, the number <strong>of</strong>(indirect) interactions that Myc makes with the basal-specificfactors in particular suggests that it is a key factor inregulating <strong>mammary</strong> cell fate determination. The smallnumber <strong>of</strong> luminal ER + population specific factors and thefact that three <strong>of</strong> them, Esr1, Pgr and Foxa1 [57], areclosely connected also supports a key role for the estrogenreceptor and its transcriptional network in regulatingluminal ER + cell fate. As described above, however, theluminal ER + transcriptional pr<strong>of</strong>ile is not, in general, anestrogen responsive pr<strong>of</strong>ile. Therefore, the activity <strong>of</strong> theluminal ER + specific transcription factors most likely leadsto long term stable cell fate changes and not changeswhich are responsive to short-term fluctuations in estrogensignalling. Mutual repression and auto-stimulationloops may be one way to achieve this long term stability.Sox6 is a determinant <strong>of</strong> luminal cell fateTo demonstrate that the lineage specific transcription factorswe identified can indeed influence cell fate and differentiation,we chose to investigate in more detail thefunction <strong>of</strong> Sox6 [52], the expression <strong>of</strong> which was onlydetectable in one population, the luminal ER - cells (Figure6B). Primary mouse <strong>mammary</strong> <strong>epithelial</strong> cells were transducedin three independent experiments with either acontrol lentivirus carrying GFP only or a virus carrying theSox6 gene plus GFP. The transduced cells were then splitbetween in vitro and in vivo assays.For in vitro <strong>analysis</strong>, cells were cultured for one week andthen harvested, GFP + cells were separated by flow cytometryand RNA isolated from them. The expression <strong>of</strong> Sox6,Krt14, Krt18 and Esr1 was examined by qPCR and comparedto expression levels in cultured primary cells whichhad not been transduced with virus. The data (Figure 8A)demonstrated that Sox6 over-expression (approximately800-fold over-expressed in Sox6 GFP virus-transducedcells compared to non-transduced cells) did not significantlyalter Krt14 gene expression. However, Krt18 geneexpression was significantly increased in Sox6 overexpressingcells compared to non-transduced cells or cellstransduced with the control virus (an approximately 3-fold increase in expression levels). Esr1 expression levelswere also increased in Sox6 over-expressing cells, althoughmore modestly (1.7 fold; P < 0.05) [58]. Next, culturedvirus-transduced primary cells were stained for keratin 14(K14) and keratin 18 (K18) expression. When primarymouse <strong>mammary</strong> cells are isolated and grown in shorttermculture, the majority <strong>of</strong> cells which proliferate arederived from the luminal epithelium. Within 48 hours inculture, these luminal origin cells (which are K14 - K18 + invivo) begin to promiscuously express K14 and acquire aK14 + K18 + phenotype [59-61]. It is thought that this representsa de-differentiation event resulting from the cellsbeing removed from their normal environment [62].Unsurprisingly, therefore, when non-transduced (GFP - )and control transduced primary cells were stained for K14and K18 expression, they were found to be both K14 + andK18 + . However, Sox6 transduced cells were K18 + butshowed only weak K14 staining in occasional cells (Figure8B) [see Additional file 20]. Therefore, Sox6 maintained<strong>mammary</strong> <strong>epithelial</strong> cells in the luminal <strong>epithelial</strong> lineageand blocked promiscuous K14 protein expression in vitro.However, as Sox6 over-expression did not block Krt14gene expression, this cannot be a direct effect on Krt14transcription.For in vivo <strong>analysis</strong>, cells transduced with either GFP-onlyor Sox6-GFP virus were transplanted into cleared mouse<strong>mammary</strong> fat pads. After eight weeks, the transplanted fatpads were examined (Figure 9A). In nine out <strong>of</strong> eighteenfat pads transplanted in three independent experimentsPage 15 <strong>of</strong> 28(page number not for citation purposes)
BMC Genomics 2008, 9:591http://www.biomedcentral.com/1471-2164/9/591basal Figure Interaction and 7 luminal mapping differentiation <strong>of</strong> differentially expressed transcriptional regulators suggests Esr1 and Myc control the balance betweenInteraction mapping <strong>of</strong> differentially expressed transcriptional regulators suggests Esr1 and Myc control thebalance between basal and luminal differentiation. Interaction map generated using in-house ROCK database and thenmanually curated. Genes are colour-coded as indicated. Interaction data based on physical interactions (solid lines) and transcriptionalinteractions (dashed lines).with cells transduced with the GFP-only virus, extensive<strong>mammary</strong> <strong>epithelial</strong> outgrowths were seen in which bothducts and alveolar buds were GFP labelled. However, nooutgrowths were observed in twenty fat pads transplantedin three independent experiments with cells carrying theSox6-GFP, although in two cases, cyst-like GFP-labelledstructures were observed.It was not possible to determine at the wholemount levelwhether fat pads which did not have extensive GFP + outgrowthscontained scattered GFP-labelled cells incorporatedinto non-GFP epithelium. Therefore, transplantedfat pads were processed to single cells, labelled with anti-Sca-1 and anti-CD24 antibodies and analysed by flowcytometry to identify GFP + cells and determine which <strong>of</strong>the <strong>mammary</strong> <strong>epithelial</strong> cell populations they segregatedwith (Figure 9B, C). Transplanted fat pads from two experimentswere pooled for this <strong>analysis</strong> whilst fat pads fromthe third transplant experiment were analysed independently.As expected, GFP + cells could be detected in thepreparations derived from these control fat pads (Figure9B). These control GFP + cells mainly segregated with theCD24 +/High Sca-1 - luminal ER - population, although theycould also be found in the CD24 +/Low Sca-1 - basal/myo<strong>epithelial</strong>and CD24 +/High Sca-1 + luminal ER + cells (Figure9C). However, GFP + cells could also be detected in preparations<strong>of</strong> cells from Sox6 GFP virus transplants (Figure9B). Although only low numbers <strong>of</strong> GFP + cells werepresent, their distribution was shifted towards the CD24 +/High Sca-1 + luminal ER + population (Figure 9C).Page 16 <strong>of</strong> 28(page number not for citation purposes)
- Page 1: This is an author produced version
- Page 4 and 5: BMC Genomics 2008, 9:591http://www.
- Page 6 and 7: BMC Genomics 2008, 9:591http://www.
- Page 8 and 9: BMC Genomics 2008, 9:591http://www.
- Page 10 and 11: BMC Genomics 2008, 9:591http://www.
- Page 12 and 13: BMC Genomics 2008, 9:591http://www.
- Page 14 and 15: BMC Genomics 2008, 9:591http://www.
- Page 18 and 19: BMC Genomics 2008, 9:591http://www.
- Page 20 and 21: BMC Genomics 2008, 9:591http://www.
- Page 22 and 23: BMC Genomics 2008, 9:591http://www.
- Page 24 and 25: BMC Genomics 2008, 9:591http://www.
- Page 26 and 27: BMC Genomics 2008, 9:591http://www.
- Page 28 and 29: BMC Genomics 2008, 9:591http://www.