Determination of the physiological and pathological roles of E2F3 in adult tissues

While genetically engineered mice have made an enormous contribution towards the elucidation of human disease, it has hitherto not been possible to tune up or down the level of expression of any endogenous gene. Here we describe compound genetically modified mice in which expression of the endogenous E2f3 gene may be either reversibly elevated or repressed in adult animals by oral administration of tetracycline. This technology is, in principle, applicable to any endogenous gene, allowing direct determination of both elevated and reduced gene expression in physiological and pathological processes. Applying this switchable technology to the key cell cycle transcription factor E2F3, we demonstrate that elevated levels of E2F3 drive ectopic proliferation in multiple tissues. By contrast, E2F3 repression has minimal impact on tissue proliferation or homeostasis in the majority of contexts due to redundancy of adult function with E2F1 and E2F2. In the absence of E2F1 and E2F2, however, repression of E2F3 elicits profound reduction of proliferation in the hematopoietic compartments that is rapidly lethal in adult animals.


Results
To generate mice in which the endogenous E2f3 gene may be reversibly switched on or off at will, a heptameric tetracycline-response element (TRE) derived from the pTRE2 vector (Clontech) was targeted via homologous recombination to the promoter of the endogenous E2f3 locus in mouse embryonic stem cells (mESCs), from which E2f3 TRE mice were derived. The TRE was placed in close proximity to the E2f3a promoter yet avoiding the critical Myc and E2F binding sites described previously 19 (Fig. 1a). Insertion of this TRE then allows for ectopic control of E2f3 expression by serving as a binding site for tetracycline-regulatable transcriptional activators or repressors. Accordingly, E2f3 TRE mice were then crossed into mice ubiquitously expressing the reverse tetracycline-dependent transactivator (rtTA), for reversible enforced endogenous E2f3 induction, or into mice ubiquitously expressing either the tTS(kid) tetracycline regulated transcriptional repressor or the reverse tetracycline-controlled transcriptional silencer (rtTS), both potentially enabling reversible repression of endogenous E2f3. TRE insertion into the E2f3 promoter does not perturb normal E2f3 regulation. The great majority of E2f3 knock-out mice die between E13.5 and P2 dpc 20 , exhibiting fatal cardiac abnormalities and typical signs of congestive heart failure. By contrast, homozygous E2f3 TRE/TRE mice were born at expected Mendelian ratios, indicating that TRE insertion does not critically interfere with control of E2f3 expression during development. In normal somatic cells, the kinetics and levels of E2f3 expression, specifically the E2f3a isoform, are tightly regulated in a cell cycle-dependent manner by multiple transcription factors, including Sp1, E2F and Myc families 19 . To assess any impact of the TRE insertion on normal E2f3 cell-cycle expression, quiescent wild-type and E2f3 TRE/TRE MEFs were serum-stimulated and E2f3a expression monitored over time. Kinetics and levels of E2f3a expression were essentially identical in E2f3 TRE/TRE and wild-type MEFs, in both cases peaking at 16 hours post serum-stimulation (Fig. 1b). We similarly saw no statistically significant differences in either E2f3a or E2f3b mRNA expression in adult E2f3 TRE/TRE versus wild-type mouse spleen, lung, liver, or heart (Fig. 1c). Insertion of the TRE into the E2f3 promoter also had no discernible impact on the expression of genes located to either side of E2f3 on mouse chromosome 13 -Mboat1 (50 kb telomeric) and Cdkal1 (151 kb centromeric) -in the tissues examined, save for a modest decrease in the liver (Fig. 1d). Taken together, these data indicate that insertion of the 300 bp TRE into the E2f3 promoter has negligible impact on normal expression and regulation of either E2f3 or its flanking genes.
Reversible repression of endogenous E2f3 expression in E2f3 TRE/TRE mice. We used two complementary approaches to repress E2f3 expression. In the first, we crossed E2f3 TRE/TRE mice to a transgenic strain that expresses the tetracycline regulated repressor, tTS 22 driven from the broadly active β-actin promoter, as described previously 23 . tTS is a fusion protein combining the bacterial tet repressor (TetR) with the KRAB-AB domain of the Kid-1 transcriptional repressor. In the absence of antibiotic (tetracycline or doxycycline), tTS binds to its cognate TRE-element, whereupon it represses transcription of proximal genes (Fig. 3a). E2f3 TRE/TRE ; β-actin-tTS pups were born at normal Mendelian frequency from mothers administered with 100 mg/L doxycycline throughout pregnancy to inactivate the tTS repressor (Fig. 3b). Furthermore normal levels of both E2F3a and b proteins were expressed in MEFs derived from embryos developed in the continuous presence of doxycycline (Fig. 3c). By contrast, neither E2F3a nor E2F3b proteins was detectable in MEFs derived from embryos continuously deprived of doxycycline throughout development and cell culture (Fig. 3c). Moreover, no E2f3 TRE/TRE ; β-actin-tTS pups were born in the absence of doxycycline (Fig. 3b) and anatomical analysis of antibiotic-deprived promoter with relative positions of known transcription factor binding sites, the transcription start site, and the TRE insertion site. Blue box-Myc binding site; green box-E2F binding site; grey box-E2f3a exon 1a; arrowtranscription start site; red box-TRE insertion site. (b) Quantitative RT-PCR analysis of E2f3a expression in quiescent wild type and E2f3 TRE/TRE MEFs following serum stimulation. E2f3a expression is normalized to HPRT and data normalized to the quiescent sample within each genotype. n = 3 independent MEF lines; error bars, s.d. (c) Quantitative RT-PCR analysis of E2f3a and E2f3b mRNA expression in adult wild type and E2f3 TRE/TRE mouse tissues. Expression is normalized to HPRT and relative to the mean of the wild type tissues. n = 3 mice; error bars, s.e.m. (d) Quantitative RT-PCR analysis of Mboat1 and Cdkal1 expression in adult wild type and E2f3 TRE/TRE mouse tissues. Expression is normalized to actin and relative to the mean of the wild type samples. n = 3 mice; error bars, s.e.m. Two-tailed t-tests; Cdkal1: liver, **P = 0.004.
Doxycycline, like other members of the tetracycline family of antibiotics, accumulates in developing bones 24 . Such accumulation serves as an antibiotic reservoir that delays its clearance after it is withdrawn from E2f3 TRE/TRE ; β-actin-tTS pups born to mothers previously treated with doxycycline throughout gestation and weaning (unpublished observations and 25,26 ). To circumvent the impact of this delay on switchably repressing E2f3 expression in E2f3 TRE/TRE mice, we replaced doxycycline with tetracycline, which is less lipophilic and has a shorter biological half-life 27 . Maintenance of pregnant mothers on drinking water containing tetracycline concentration at 50 mg/L proved sufficient to maintain normal embryonic development of E2f3 TRE/TRE ; β-actin-tTS pups (Fig. 3b) and culminated in neonate and adult mice indistinguishable from their E2f3 wild-type littermates. This effective level of tetracycline was then used for all subsequent studies unless otherwise noted.
We next assessed the efficacy and reversibility of E2f3 repression in mouse tissues in vivo using the tTS system. We first confirmed that administration of tetracycline throughout development maintains expression of (spleen, lung), n = 5(liver)), E2f3 TRE/TRE ; Rosa26 rtTA/+ (n = 4) and E2f3 TRE/TRE ; Rosa26 rtTA/rtTA (n = 3) mouse tissues following administration of doxycycline for three days. Expression is normalized to HPRT and relative to the mean of the control samples. Error bars, s.e.m. Two-way ANOVA with a Tukey's multiple comparisons test; *P < 0.5, ***P < 0.001. (c) Immunoblot analysis of E2F3 protein levels in adult control, E2f3 TRE/TRE ; Rosa26 rtTA/+ and E2f3 TRE/TRE Rosa26 rtTA/rtTA mouse tissues following administration of doxycycline for three days. Replicate samples are generated from tissues isolated from independent mice. Expression of GAPDH is included as a loading control. Refer to Supplementary Fig. 11 for full images. (d) Quantitative RT-PCR analysis of E2f3 expression in adult control (n = 3) and E2f3 TRE/TRE ; Rosa26 rtTA/+ tissues isolated from mice following administration of doxycycline for three days only (+3, n = 2) or for three days with subsequent withdrawal for seven days (+3/−7, n = 5). Expression is normalized to HPRT and relative to the mean of the control samples. Error bars, s.e.m. Two-tailed t-test: control vs. +3/−7, P = 0.82 (spleen), 0.10 (liver), 0.52 (lung); n.s., not significant. Control genotypes include E2f3 TRE/TRE in the presence and absence of doxycycline and E2f3 TRE/TRE ; Rosa26 rtTA/+ and E2f3 TRE/TRE Rosa26 rtTA/rtTA maintained in the absence of doxycycline. E2f3 at ostensibly normal levels in all tested adult tissues (Fig. 3d). However, withdrawal of tetracycline from adult mice hitherto maintained on tetracycline induced ubiquitous and profound repression of E2f3, evident within one week and thereafter sustained (Fig. 3e), with concomitant reduction of both E2F3a and E2F3b proteins (Fig. 3f). Expression of E2f3's two neighbouring genes, Mboat1 and Cdkal1, was unaffected except for a modest decrease in Cdkal1 in spleen ( Supplementary Fig. 3). Subsequent restoration of tetracycline to the drinking water of E2f3-repressed E2f3 TRE/TRE ; β-actin-tTS mice rapidly restored normal E2f3 expression (Fig. 3f).
We also reconfigured the tTS repression system from one where tetracycline inactivates the repressor to one where the antibiotic activates it. Such a configuration has clear advantages in situations requiring long periods of normal E2f3 expression punctuated by acute periods of E2f3 repression. To do this, we generated a novel Rosa26 CAG-rtTS strain that ubiquitously expresses a chimeric protein formed from the fusion of the reverse tetracycline regulated region of rtTA 28 with the robust transcriptional silencer derived from the KRAB domain of Kid-1 protein 22,29,30 , similar to the one previously described 31 . The resulting rtTS repressor is dependent on tetracycline or doxycycline for its repressing activity (Fig. 3g). Administration of doxycycline to E2f3 TRE/TRE ; Rosa26 CAG-rtTS/+ mice for one week repressed E2f3 transcription to nearly undetectable levels in all tissues tested; this repression was reversible upon withdrawal of doxycycline (Fig. 3h). Together these two systems, tTS and rtTS, allow for unprecedentedly rapid and flexible switching of target gene activity in adult tissues in vivo.

Consequences of differential expression of E2F3 in adult tissues.
Because switchable genetic systems obviate the complications of both embryonic lethality and adaptive compensation induced by absence or constitutive over-expression of genes during development, we could uniquely use our switchable E2f3 TRE/TRE mice specifically to determine the biological functions of E2F3 in adult tissues in mice that had developed in the presence of normal E2F3 expression.
Consistent with its role in driving cell cycle, enforced expression of endogenous E2f3 in E2f3 TRE/TRE ; Rosa26 rtTA/rtTA mice induced known E2F3 target genes, Ccna1, Cdc2, and E2f1 ( Fig. 4a) in all tissues examined (spleen, liver, and lung). Increased proliferation (Ki67-or p-H3-positive cells) was clearly evident within 3 days of enforced E2f3 expression (Fig. 4b,c) in two relatively quiescent adult tissues, lung and liver while in spleen, which is already highly proliferative, a modest increase was also evident. The increase in proliferation, which was rtTA level-dependent (i.e. greater in 2 rtTA alleles versus 1) (Fig. 4c), is consistent with the increase in proliferation induced by classical transgenic over-expression of ectopic E2f3 expression in other mouse models [16][17][18]32 . Interestingly, E2F3-induced proliferation in liver initially peaked and then fell back to a lower, although still elevated, level that was thereafter sustained for at least 60 days, so long as E2f3 induction was maintained (Fig. 4d). Similarly, peripheral leukocyte populations showed a trend towards a transient increase immediately following ectopic E2f3 expression ( Supplementary Fig. 4). Hence, acute induction of enforced E2f3 expression drives ectopic proliferation in multiple somatic tissues of adult mice.
By contrast, systemic repression of E2f3 for one or two weeks in adult mice had no discernible impact on the proliferative indices of any tested tissues, including constitutively proliferating tissues such as small intestine, spleen and thymus (Fig. 5a). Likewise, expression of the common E2f target genes Ccna1, Cdc2 or E2f1 in multiple tissues was unperturbed by E2F3 repression (Supplementary Fig. 5). Furthermore, long-term systemic repression of E2f3, over many weeks, had no discernible impact on the health of animals, whose tissues remained histopathologically indistinguishable from those of wild-type controls (data not shown). These data are consistent with the notion that the function of E2F3 in adult tissue homeostasis is largely dispensable, presumably due to functional redundancy with other activator E2F family members 3,5,33 . To determine whether such functional redundancy remains sufficient to compensate for a lack of E2F3 activity under conditions of increased proliferative burden, we assessed the impact of E2F3 repression in acutely regenerating tissues. We saw no measurable inhibitory impact of E2F3 repression on liver regeneration after CCl 4 injury or intestinal crypt regeneration after radiation injury (Fig. 5b,c). By contrast, mammary gland expansion and elaboration in pregnant mice was significantly retarded, as evidenced by decreased proliferation and reduced terminal budding (Fig. 5d). We conclude that, while E2F3 clearly contributes to the proliferative capacity of normal proliferating and regenerating tissues, its repression in most adult tissues can be complemented by alternate mechanisms -presumably E2F1 and E2F2 -even in extremis (see below).
Determining real-time dependency on activator E2F activity in adult tissues in vivo. The common transcriptional activity of the three mitogenic E2Fs is essential for somatic cell cycle progression and proliferation of diverse cell types 7, 33-36 . However, the redundant functionalities of the mitogenic E2Fs -E2F1, E2F2, and normalized to HPRT and relative to the mean of the samples maintained on tetracycline. Error bars, s.e.m. Oneway ANOVA with a Tukey's multiple comparisons test, ***P < 0.001. (f) Immunoblot analysis of E2F3 protein levels in E2f3 TRE/TRE ; β-actin-tTS mice maintained on tetracycline throughout (0), removed from tetracycline for 2 weeks (−2), or removed from tetracycline for 2 weeks followed by re-addition of tetracycline for 1 week (−2/+1) or 2 weeks (−2/+2). Replicate samples are generated from tissues isolated from independent mice. GAPDH is included as a loading control. Refer to Supplementary Fig. 14 for full images. (g) Schematic of rtTS driven repression. Grey box-exon, blue box-TRE element, red oval-reverse tetracycline trans silencer, orange triangle-doxycycline. (h) Quantitative RT-PCR analysis of E2f3 in control E2f3 TRE/TRE (n = 12 (spleen, liver), 11 (lung)) and E2f3 TRE/TRE ; Rosa26 CAGrtTS/+ mouse tissues in the presence of 2 g/L doxycycline for seven days ( + 7, n = 5) or for seven days and a subsequent seven days post withdrawal ( + 7/−7, n = 4). Expression is normalized to HPRT and relative to the mean of the control samples. Error bars, s.e.m. One-way ANOVA with a Tukey's multiple comparisons test, **P < 0.01, ***P < 0.001, n.s. = not significant.
E2F3 -obscure the impact of blocking any individual member (see above and ref. 4,5), necessitating inhibition of all three to understand their shared role in vivo. Unfortunately, E2f1-3 triple knock-out mice display embryonic lethality, as well as aberrant dynamics in multiple embryonic tissues -most notably, constitutively high levels of apoptosis 34, 36, 37 -which precludes meaningful analysis of their shared role in adult tissue homeostasis and pathology. Therefore, to determine directly the role of global E2F1-3 activity in adult mice, we generated E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice in which, due to deficiency of E2F1 and E2F2, all activator E2F function is funnelled through the regulatable E2f3 gene. E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice maintained on tetracycline (tet repressor inactive) were grossly and histopathologically indistinguishable from control E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE animals (data not shown) and exhibited similar life-spans (median survival: 172.5 and 159 days, respectively). However, withdrawal of tetracycline (tet repressor active) rapidly triggered profound pathologies. By three weeks post tetracycline withdrawal, E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice exhibited severe leukopenia and thrombocytopenia (Fig. 6a). Correlating with these reduced hematopoietic cell numbers in the periphery, reduced proliferation was evident in the bone marrow, spleen, and thymus (Fig. 6c) and reduced cellularity was observed in the bone marrow (Fig. 6d). The three main hematopoietic populations of the bone marrow -B-cells, T-cells, and myeloid cells -were all markedly reduced by three weeks post tetracycline withdrawal (Supplementary Fig. 6). At this time, the internal structures and relative geographical locations of differentiated cell populations within the spleen remained similar in both E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice and appropriate controls ( Supplementary Fig. 7), but both the spleen and thymus appear macroscopically smaller in E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice than their controls (data not shown). By five weeks' post tetracycline withdrawal, all mice were moribund and severely cytopenic, including erythrocytopenia ( Fig. 6b and Supplementary Fig. 8). In the small intestine, acute repression of E2f3 in E2f1/2 doubly-deficient mice triggered a progressive increase in enteroendocrine cells and decrease in goblet and Paneth cells (Fig. 6e and Supplementary Fig. 9). Repression of E2f3 in E2f1 −/− ; E2f2 −/− doubly deficient mice had no discernible impact on non-proliferating tissues such as liver, lung and kidney (data not shown). Hence, acute and long-term homeostatic dependency on functional E2F1-3 is restricted to proliferative organs and most marked in hematopoietic tissues.
Given the dramatic and rapidly lethal impact of E2f1-3 repression in E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice, we next asked whether expeditious reversal of E2f3 repression could reverse the blood and intestinal phenotypes and preserve viability. Indeed, re-administration of tetracycline to leukopenic E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice deprived of tetracycline for two weeks rapidly and completely reversed both cytopenic and intestinal phenotypes, leading to full recovery (Fig. 7a,b and Supplementary Fig. 9). By contrast, re-administration of tetracycline to near moribund E2f1 −/− ; E2f2 −/− ; E2f3 TRE/TRE ; β-actin-tTS mice deprived of tetracycline for three weeks failed to rescue the animals (data not shown). These data define a tight temporal window during which E2F1-3-deprived mice can recover if E2F function is restored.

Discussion
We describe a novel class of switchable genetic mouse model, using the E2f3 gene as an exemplar, which allows for reversible up or down regulation of an endogenous gene in adult mice at will. Such regulation is achieved by inserting a TRE into the endogenous E2f3 promoter, allowing external modulation of the endogenous E2f3 gene via either a tetracycline-dependent transcriptional activator (rtTA) or a tetracycline-dependent repressor tTS(kid) or rtTS. Insertion of the TRE by itself has no significant effect on the normal regulation of E2f3 aside from a very mild decrease of E2f3 expression in the livers of E2f3 TRE/TRE mice. Nonetheless, there is an  obvious requirement to avoid disrupting as many cis-regulatory sequences as possible when choosing the site for TRE insertion, as illustrated by the instance of TRE insertion into the Hoxa2 5′ untranslated region, which resulted in markedly decreased expression of Hoxa2 38 . In our E2f3 case, TRE insertion into the promoter proved highly effective for ectopic E2f3 regulation. However, as indicated by other studies involving ectopic regulation of endogenous genes 39 , each targeted gene will need to be considered individually, most especially when seeking rtTA-dependent transcriptional activation.
Expression of the rtTA transcriptional activator in E2f3 TRE mice resulted in dramatically elevated and persistent doxycycline dependent ectopic E2f3 expression, which was reversible upon withdrawal of doxycycline. Encouragingly, analysis of E2f3's two neighbouring genes, Mboat1 and Cdkal1, showed them to be unaffected by either TRE insertion itself or by the rtTA-dependent upregulation of E2f3, indicating the high precision of E2f3 gene activation. Of note, the extent of E2f3 induction was around two-fold higher in animals carrying two versus one rtTA alleles, further embellishing the tunability of the experimental system. Such doxycycline-dependent upregulation resembles that seen in a previous study involving upregulation of a TRE-modified Mlc1 allele and an alternative tTA system 40 .
Conversely, expression of the tTS(kid) or rtTS repressors in E2f3 TRE cells and tissues allows for profound, doxycycline-dependent, reversible repression of endogenous E2f3. Hence, the combined iterations of these mouse models allow for the reversible toggling of endogenous E2f3, up and down, at will, in adult tissues. Of note, in both of our E2f3 activation and repression models, E2f3a and E2f3b isoforms are both co-regulated. Hence, in this study we cannot attribute any phenotype to a particular E2F3 isoform but rather to total E2F3. However, while we acknowledge that E2F3a and E2F3b possess some distinct properties, past studies indicate that each can significantly complement for the absence of the other 14 .
We generated two iterations of the TRE-dependent E2f3 repression system, one (E2f3 TRE/TRE ; β-actin-tTS) repressing E2f3 upon tetracycline withdrawal and the other (E2f3 TRE/TRE ; Rosa26 CAG-rtTS/+ ) repressing E2f3 upon doxycycline addition. Each iteration has its own advantages in terms of rapidity of switching, building up of antibiotic reservoirs in bone and teeth over extended antibiotic administration, and minimizing perturbation of animals. Either addition of doxycycline to E2f3 TRE/TRE ; Rosa26 CAG-rtTS/+ mice or withdrawal of tetracycline from E2f3 TRE/TRE ; tTS adult mice triggered rapid repression of endogenous E2f3 expression in all tested tissues. E2f3 TRE/TRE ; β-actin-tTS embryos born to mothers maintained on doxycycline were born with Mendelian frequency. However, the embryos of the same genotype developed in mothers deprived of doxycycline failed to survive to birth and exhibited the same heart trabeculation deficit as that seen in classical E2f3 knockout embryos. Essentially complete E2f3 repression in doxycycline-deprived E2f3 TRE/TRE ; β-actin-tTS embryos was confirmed by analysis of E2F3 protein expression isolated from E2f3 TRE/TRE ; β-actin-tTS MEFs; E2F3 protein is normally expressed in doxycycline-treated E2f3 TRE/TRE ; β-actin-tTS MEFs but completely absent from doxycycline-deprived fibroblasts. Intriguingly, subsequent in vitro addition of doxycycline to MEFs in which E2f3 had been repressed throughout development failed to restore E2F3a/b expression ( Supplementary Fig. 10), indicating that the E2f3 gene is permanently silenced if actively repressed through embryogenesis. The mechanism behind this is unclear but may involve methylation, as observed elsewhere 41 .
In adult tissues of E2f3 TRE/TRE ; β-actin-tTS mice, E2f3 repression was effectively complete and maximal by 1 week of antibiotic withdrawal, persisting thereafter for as long as tetracycline was withheld. Despite the profound repression of E2f3 expression in tetracycline-deprived E2f3 TRE/TRE ; β-actin-tTS mice, we nonetheless observed little impact on proliferation rates of tissues, their architecture and histology, or expression of key E2F1-3 transcriptional targets, such as E2f1, Cdc2 and Cyclin A. This lack of overt phenotype was true even after extended (9 week) absence of tetracycline. In contrast to our inability to reverse constitutive repression through embryonic development, E2f3 repression acutely imposed in adult tissues proved rapidly reversible upon re-administration of tetracycline.
While the lack of any phenotype associated with acute E2F3 repression is unsurprising in non-proliferating tissues like lung and liver, we also saw no measurable impact in proliferative tissues like intestine, spleen and thymus. Equally surprisingly, E2f3 repression had no inhibitory impact on proliferation in tissues regenerating after damage -specifically liver after CCl 4 injury and intestine after irradiation. Only in the mammary tissue of pregnant mice was any impact of acute E2f3 repression evident, which manifested as the significant retardation of terminal end bud proliferation and elaboration. This observation may be relevant given the previously published potentiating role of E2f3 in experimental mammary cancers of mice 11,12 .
in all of our tTS repression studies: tetracycline has shorter plasma half-life and proved effective at doses as low as 50 mg/L. To circumvent accumulation of the antibiotic, we also reconfigured the tTS(kid) repressor, which is inactivated by tetracycline, to the corresponding reverse tetracycline rtTS(kid) variant, which is dependent on the drug for its activity. This Rosa26 CAG-rtTS strain is ideal for studies in which normal E2F3 expression is needed during development, followed by acute E2F3 repression in adult tissues. Furthermore, the new Rosa26 CAG-rtTS allele exists in a conditional form in which rtTS protein expression is dependent upon Cre activity, allowing repression to be manifest only within specific target cell-types or tissues. Similar conditional rtTA alleles are readily available for complementary overexpression studies. These further refinements to the experimental system significantly enhance its utility and flexibility.
Repressing E2f3 in E2f1 −/− ; E2f2 −/− doubly-deficient mice triggered profound collapse of bone marrow, together with the onset of leukopenia, thrombocytopenia and progressive erythrocytopenia, with corresponding decreases in lymphocyte and myeloid cell populations. These data are consistent with attrition of each blood cell type according to its natural lifespan. Mice were moribund following 3 weeks post tetracycline withdrawal. However, re-administration of tetracycline two weeks after its withdrawal, at a time of rapidly declining blood counts, reversed all pathologies and all animals completely recovered. Such complete rescue of all animals by timely restoration of activator E2F activity defines a potentially useful therapeutic window for any treatment modality that works by blocking activator E2F activity.

Materials and Methods
Generation of E2f3 TRE mice. A vector containing 15 kb of E2f3 genomic sequence was a kind gift from Jacqueline Lees. The TRE element was isolated from pTre2 (Clontech) using KpnI and XhoI, and then cloned into a KpnI site 500 bp upstream of the E2f3a translation start site along with a LoxP flanked Neomycin resistance cassette. No E2f3 sequences were removed or altered at the insertion site, although silent mutations were incorporated into the first exons of E2f3a and E2f3b. The final sequences encompassing these mutations is as follows: E2f3aWT: cggtggcccaccg/E2f3aMut: TggAggAccTccg; E2f3bWT: cggaaatgcccttacagcagcag/E2f3bMut: cggaaatgc-cACtTcaAcagcag. Finally, a DTA cassette was added to this vector to allow negative selection in mouse embryonic stem cells. This vector was then linearized using NotI and introduced into Sv/129 mouse embryonic stem cells. Correct targeting was confirmed by Southern blot, and correctly targeted clones were transiently transfected with a plasmid expressing Cre-recombinase in order to remove the conditional Neomycin resistance cassette. The final sequence inserted to the KpnI site of mESCs used to generate chimeric animals is below; the residual single LoxP site is in bold: G  The SA70b pROSA-CAGGS-attP50/B53-hygro-NLSLacZ vector was obtained from ARTEMIS Pharma-ceuticals. A fragment containing the CAGGS promoter and an intron was removed from SA70b and cloned into a modified pROSA26 plasmid 48 , creating the pROSA26CAG-PAS vector. The rtTS cDNA above was cloned into pBigT, and then into pROSA26CAG-PAS 48 to create the final Rosa26 CAG-LSL-rtTS targeting vector. This vector was then linearized introduced into mouse embryonic stem cells. Germ-line transmission was confirmed by PCR and verified by Southern analysis. A Rosa26 CAG-LSL-rtTS/+ male was bred to a Pgk-Cre 49 female which removed the Lox-stop-lox cassette in the germline to produce Rosa26 CAG-rtTS/+ mice, which were then intercrossed with E2f3 TRE/TRE . by the presence of multiple cell layers and a lack of a ductal lumen. These structures were counted in at least three independent biological repeats per genotype by two independent researchers. Blood cell analysis. Peripheral blood was collected into EDTA or heparin-coated tubes and 40 µl was analyzed on a Sysmex haematology analyser. For estimation of total cellularity of bone marrow, the total number of nucleated cells was counted per field of view from a hematoxylin and eosin stained section. For the B220, CD3, and Cd11b quantifications of cells within the bone marrow, the number of positively stained cells were counted per field of view. For both cellularity and cell surface marker stainings in the bone marrow, 3 images per organ/mouse were taken at 40x magnification and quantified; the mean of 3 raw counts was calculated and represents one data point per graph.

C G A A G G C C T G A C G A C A A G G A A A C T C G C T C A A A A G C T G G G A G T T G A -G C A G C C T A C C C T G TA C T G G C A C G T G A A G A A C A A G C G G G C C C T G C T C G AT G C C C -T G C C A A T C G A G A T G C T G G A C A G G C A T C A T A C C C A C T T C T G C C C C C T G G A
Image acquisition tools and image processing software packages. Immunoblots were developed on Fuji RX X-ray film 18 × 24 cm and then scanned on an Epson Perfection V500 Photo flatbed scanner. Images were cropped using Adobe Photoshop, but were otherwise unprocessed. Immunohistochemical staining was imaged on a Zeiss Axio Imager using the Zeiss AxioVision 4.8 software using the AutoLive setting and interactive white balance. Quantification was performed by counting number of cells per field of view for 5 images per organ/ mouse, the mean of 5 raw counts was calculated and represents one data point per graph. Immunohistochemical analysis. Immunohistochemistry was performed on 4.5 µm sections. Sections were de-paraffinized and rehydrated by passing through xylene and a series of ethanols to water. Antigen retrieval was performed by boiling in 10 mM citrate buffer (pH 6.0) for 10 minutes. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 30 minutes. Sections were then subject to rabbit VECTASTAIN Elite ABC horseradish peroxidase kit following the manufacturers protocols. Sections were developed in DAB for 5 minutes, counterstained in hematoxylin, dehydrated with ethanol, cleared in xylene and mounted in DPX. Primary antibodies; Anti-Ki67 (Thermo Scientific, Fremont, CA, USA; clone: SP6; 1:200), Anti-p-H3 (Merck Millipore, Germany; 06-570: Anti-phospho-Histone H3 (Ser10) Antibody; 1:500), Anti-lysozyme (Life Technologies; A0099), Anti-Chromogranin A (Abcam; ab15160), Anti-CD11b antibody (Abcam; ab133357; 1:4000), Anti-CD3 (Themo Scientific, RM-9107-R7; prediluted), Anti-CD45R (Thermo Scientific, RA3-6B2, 1:100). Alcian blue staining was performed following standard procedures in 1% Alcian blue in 3% acetic acid, pH 2.5.
Tissue regeneration studies. Liver damage was induced in female and male mice (age 8-12 weeks) by a single intraperitoneal injection of CCl 4 (0.5 ml/Kg in corn oil). Livers were collected 3 days following the CCl 4 injections and proliferative hepatocytes were quantified. Intestinal damage was induced by 14 Gy of gamma irradiation from a Cs 137 source and small intestine collected 3 days later.
Data Availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information files).