There has been uncertainty regarding the precise role that the pocket protein Rb1 plays in murine melanocyte homeostasis. It has been reported that the TAT-Cre mediated loss of exon 19 from a floxed Rb1 allele causes melanocyte apoptosis in vivo and in vitro. This is at variance with other findings showing, either directly or indirectly, that Rb1 loss in melanocytes has no noticeable effect in vivo, but in vitro leads to a semi-transformed phenotype. In this study, we show that Rb1-null melanocytes lacking exon 19 do not undergo apoptosis and survive both in vitro and in vivo, irrespective of the developmental stage at which Cre-mediated ablation of the exon occurs. Further, Rb1 loss has no serious long-term ramifications on melanocyte homeostasis in vivo, with Rb1-null melanocytes being detected in the skin after numerous hair cycles, inferring that the melanocyte stem cell population carrying the Cre-mediated deletion is maintained. Consequently, whilst Rb1 loss in the melanocyte is able to alter cellular behaviour in vitro, it appears inconsequential with respect to melanocyte homeostasis in the mouse skin.
As RB1 is often considered to be the PP with the strongest tumour suppressive function, a number of mouse and mouse-cell-based models have been developed that have allowed the evaluation of how Rb1-loss affects melanocyte homeostasis. The results of these studies have been paradoxical. In one case, deletion of Rb1 in all cell types within the skin of Rb1F19/F19 mice using a TAT-Cre system resulted in apoptosis of melanocytes and depigmentation, and this appeared to be a cell autonomous effect (Yu et al., 2003). In contrast, chimaeras composed of Rb1-null and Rb1 wild-type cells do not exhibit pigmentation defects (Maandag et al., 1994; Williams et al., 1994). Similarly, the overexpression in melanocytes of a mutant E2F-1 that sequesters Rb1, rendering them ‘pseudo Rb1-null’, does not result in apoptosis. Rather, the melanocytes proliferated more readily than wild-type cells, even in reduced mitogen conditions (Halaban et al., 1998a). Similar results were observed when we performed tissue-specific ablation of Rb1 in mouse melanocytes using the Cre/loxP system. In this case, Tyrosinase expressing Cre (TEC) mice expressing Cre recombinase under the control of Tyrosinase gene transcriptional elements (Tonks et al., 2003) were crossed with mice carrying an Rb1 allele in which exon 2 was flanked with unidirectional loxP sites. These mice lacked any pigmentation defects and the melanocytes behaved in a semi-transformed fashion when cultured in vitro (Tonks et al., 2005).
To decipher what might underpin these observed differences in melanocyte homeostasis, we obtained the Rb1F19/F19 mice used by Yu et al. (2003), which harbour Rb1 alleles whose exon 19 is flanked with loxP sites and we subsequently performed Cre-mediated melanocyte-specific deletion of Rb1 at different developmental stages. To accomplish this, we used the TEC1 (Tonks et al., 2003) and Tyr::CreER(T2) (4 Hydroxy Tamoxifen (4-OHT)-inducible Cre driven by Tyrosinase transcriptional elements) systems (Bosenberg et al., 2006). Further, the studies using Rb1F2/F2 mice were extended to determine the long-term in vivo effects on melanocyte homeostasis of tissue-specific post-natal ablation of Rb1 function.
Results and discussion
To address whether the phenotypic differences observed upon Rb1-ablation in different studies was attributable to the exon targeted for ablation we bred and analysed Rb1F19/F19:TEC1+/− mice. The TEC1 transgene facilitates the high frequency knockout of floxed alleles within melanocytes from as early as 11.5 days post coitum (dpc) (Tonks et al., 2003). We observed that the pigmentation profile of Rb1F19/F19 and Rb1F19/F19: TEC1+/− mice was identical in flank skin (Figure 1A), belly (Figure 1B) paws (Figure 1C) and ears (Figure 1D). The distribution of melanocytes within the ear and flank skin of Rb1F19/F19: TEC1+/− and Rb1F19/F19 mice was also shown to be identical (Figure S1).
To confirm that the skin harboured Cre-expressing/Rb1-ablated melanocytes, the flank skin DNA of Rb1F19/F19:TEC1+/− and control animals was subjected to PCR using Rb212/Rb18 primers (Marino et al., 2000). This amplifies a 260-bp PCR product from Cre-deleted Rb1 alleles, which was readily detected in the flank skin DNA of Rb1F19/F19:TEC1+/− mice (Figure 1E; lane 2) and was correlated with the presence of the Cre transgene (Figure 1E; lane 4). In the case of Rb1F19/F19 controls, no PCR products corresponding to either the Cre-deleted Rb1 allele (Figure 1E; lane 3) or Cre transgene were detectable. To ensure that the melanocytes in the skin possessed a homozygous deletion of Rb1, we performed PCR analysis on three independent melanocyte cultures established from Rb1F19/F19:TEC1+/− neonates. The PCR used the Rb212/Rb18/Rb19E primer combination, which detects deleted, wild-type and floxed alleles within a given sample (Marino et al., 2000) and resulted only in the amplification of the 260-bp product corresponding to the deleted allele (Figure 2A, lanes 2, 3 and 4). In contrast, the control Rb1F19/F19 cells only amplified the 283-bp product corresponding to the floxed allele (Figure 2A; lane 5). This correlated with protein expression as Western blot analysis with an anti-Rb1 antibody indicated that the control cultures (Figure 2B; lanes 4 and 5) all harboured detectable protein, which was absent in the Rb1F19/F19:TEC1+/− melanocytes (Figure 2B; lanes 1, 2, and 3).
The morphology and pigmentation of cultured Rb1F19/F19 (Figure 2C, E, G) and Rb1F19/F19:TEC1+/− (Figure 2D, F, H) (Rb1ΔX19/ΔX19) melanocytes were examined and, by all criteria tested, the cells appeared identical. Furthermore, no signs of apoptosis were apparent in these cultures irrespective of passage, like Rb1ΔX2/ΔX2 cells (Tonks et al., 2005).
Consequently, our data indicated that deletion of exon 19 from Rb1 in early-stage melanocytes causes a loss of Rb1 but does not affect in vivo melanocyte homeostasis. Further, there is no sign of a dominant negative effect imparted by the exon 19-ablated allele as it recapitulates the phenotype observed in melanocytes of other Rb1 knockout systems (Halaban et al., 1998a; Maandag et al., 1994; Tonks et al., 2005; Williams et al., 1994). Thus, the ablation of either exon 2 or exon 19 from Rb1 at an early developmental stage dictates an identical phenotypic outcome to melanocytes and indicates that any differences observed between studies on Rb1 are attributable to other causes.
In some cell types, such as fibroblasts, the developmental point at which Rb1 ablation occurs can have a profound effect on cellular phenotypes (Sage et al., 2003). Consequently, we next sought to establish whether there is a developmental stage-dependent effect that could explain the variable melanocytic phenotypes noted in different Rb1-ablation studies. We used a melanocyte-specific 4-OHT-inducible Cre system [Tyr::CreER(T2)] (Bosenberg et al., 2006) to delete exon 19 specifically in mouse cutaneous melanocytes in vivo and in vitro at later developmental stages. To induce the ablation of floxed alleles both Rb1F19/F19 and Rb1F19/F19:Tyr::CreER(T2)+/− neonates were subjected to three daily topical applications of 4-OHT to the flank skin and ears from 1-day postpartum (dpp). The pigmentation profile of the cohort was then followed for 6 months. At no stage was any evidence of hypopigmentation or hyperpigmentation observed compared to untreated controls (Figure 3A). The 4-OHT treatment regimen appeared effective in ablating the floxed Rb1 allele in vivo as the 260-bp Rb18/Rb212 PCR product was readily amplified from the skin of treated Rb1F19/F19:Tyr::CreER(T2)+/− animals (Figure 3B; lane 2) but not untreated Rb1F19/F19:Tyr::CreER(T2)+/− controls (Figure 3B; lane 3). No Cre-mediated recombination of Rb1 was detected in the skin of Rb1F19/F19 controls, irrespective of 4-OHT treatment status (Figure 3B; lanes 4 and 5).
To ensure that these melanocytes carried a homozygous deletion of Rb1, the recombination status of the floxed allele was determined in four individual Rb1F19/F19:Tyr::CreER(T2)+/− melanocyte lines within 2 weeks of initial explanting from 4-OHT-treated neonates. The proportion of PCR products amplified from the primary melanocyte cultures using the Rb18/Rb19E/Rb212 primer combination indicated that only a minor amount of the undeleted Rb1 allele was present in the melanocytes of 4-OHT-treated Rb1F19/F19:Tyr::CreER(T2)+/− neonates (Figure 3C; lanes 2, 3, 4 and 5). Quantitative PCR (qPCR) was performed with the Rb18/Rb212 and Rb18/Rb19E primer combinations and this showed that an average of 89.8% (SEM = 1.39) of the floxed Rb1 alleles in the culture were correctly recombined (Figure 3C). Further subculturing of these melanocyte lines yielded cultures that amplified only the product corresponding to the deleted allele when PCR was performed with the Rb18/Rb19E/Rb212 primer combination (Figure 3C; lanes 6, 7, 8 and 9).
There are differences in the division rates of Rb1-null (22 h) and wild-type (30 h) melanocytes in established cultures (Tonks et al., 2005) and this may have the potential to misrepresent the initial rates of Cre-mediated recombination in the skin. Consequently, to determine the percentage of melanocytes that were homozygously recombined at Rb1 at the point closest to culture derivation, we performed immunofluorescence analysis on Rb1F19/F19:Tyr::CreER(T2)+/− melanocyte cultures within 38 h of the final in vivo 4-OHT treatment and initial explant. The use of anti-Mitf and anti-Rb1 antibodies in immunofluorescence analysis allowed the percentage of Rb1-expressing and Rb1-null melanocytes to be determined whilst Rb1-positive:Mitf-negative cell types, which represent keratinocytes and fibroblasts, could be excluded from the analysis (Figure S2). Cell counts performed on 3 different 4-OHT-treated Rb1F19/F19:Tyr::CreER(T2)+/− explants found that approximately 63, 78 and 88% of melanocytes lacked Rb1-expression, indicating that in vivo an average of 76% of melanocytes had undergone Cre-mediated recombination. Collectively, the qPCR and immunofluorescence data indicate that the majority of melanocytes present in the 4-OHT-treated skin of Rb1F19/F19:Tyr::CreER(T2)+/− neonates are Rb1-null.
As we have previously shown that melanocytes cultured in vitro may manifest different cellular characteristics when released from in vivo regulatory constraints (Tonks et al., 2005) Rb1F19/F19:Tyr::CreER(T2)+/− melanocytes were cultured, subjected to 4-OHT-induced Cre-mediated deletion and examined for both levels of apoptosis and the deletion status of the floxed allele 48 h later to see if in vitro deletion had any influence on phenotypic outcomes. The Rb1F19/F19:Tyr::CreER(T2)+/− primary cultures that were not treated with 4-OHT only amplified the PCR product corresponding to the floxed allele when tested with the Rb18/Rb19E/Rb212 primer combination (Figure 3D, lanes 2, 3 and 4). In contrast, the proportion of different PCR products amplified from the three 4-OHT-treated Rb1F19/F19:Tyr::CreER(T2)+/− primary cultures indicated that the majority of the floxed alleles in the culture had undergone Cre-mediated recombination (Figure 3D, lanes 5, 6 and 7). More precise, analysis of the 4-OHT-treated primary cultures by qPCR showed that an average of 85.6% (SEM = 4.92) of floxed alleles were deleted in the three different cell lines examined, indicating that the majority of cells are homozygously deleted (Figure 3D).
To evaluate whether the Cre-mediated deletion of Rb1 triggered any detectable apoptosis, the melanocyte cultures were assayed 48 h after the addition of 4-OHT with an antibody (ASP175) that detected active caspase 3. Whilst the positive control readily showed the ability of the assay to detect apoptotic cells (Figure 3E), the morphologies and levels of apoptosis of 4-OHT-treated Rb1F19/F19:Tyr::CreER(T2)+/− cells (Figure 3F) appeared similar to untreated controls (Figure 3G), indicating that the loss of Rb1 did not appear to trigger apoptosis at any detectable level.
Consequently, irrespective of whether the allele is deleted in vivo or in vitro, or the developmental stage at which deletion occurs, the resultant Rb1ΔX19/ΔX19 melanocytes do not appear to undergo apoptosis but continue to proliferate. This indicates that the developmental point at which Rb1 ablation occurs in the melanocyte is ineffectual in modulating phenotypic outcomes.
As the consequences of exon 2 and 19 ablation are phenotypically equivalent with respect to melanocyte homeostasis, we exploited an ongoing study using Rb1F2/F2:Tyr::CreER(T2)+/− mice to determine the long-term effects of Rb1 loss on melanocytes in vivo. The Rb1F2/F2:Tyr::CreER(T2)+/− and control Rb1F2/F2 mice were subjected to three daily topical applications of 4-OHT from 1 dpp. Culture studies similar to those performed for the Rb1F19/F19:Tyr::CreER(T2)+/− cohort had shown that this 4-OHT treatment regime was effective in driving melanocyte-specific Cre-mediated ablation of floxed Rb1 exon 2 alleles in melanocytes in vivo (Figure S3).
Mice were subsequently observed for a minimum of 14 months up to a maximum of 16 months. Amongst the 4-OHT painted Rb1F2/F2:Tyr::CreER(T2)+/− (n = 16) and Rb1F2/F2 cohorts (n = 26), no perturbations were observed in melanocyte homeostasis, with the pigmentation of treated conditional tissue-specific knockout mice being equivalent to controls (Figure 4A). To assure that Rb1ΔX2/ΔX2 melanocytes were still present at the completion of the experiment, the skin was genotyped using the Rb5/Rbamp3 primer and Rbamp2/Rbamp3 combinations that detect deleted and floxed Rb1 alleles (Tonks et al., 2005), respectively, as well as primers that detected the Tyr::CreER(T2) transgene (Bosenberg et al., 2006). Only the skin of the 4-OHT-treated Rb1F2/F2:Tyr::CreER(T2)+/− mice displayed the 329-bp band corresponding to the Cre-deleted Rb1 allele (Figure 4B lane 2). In contrast, the 652-bp band corresponding to the floxed allele was readily amplified from both knockout (Figure 4B lane 5) and control (Figure 4B lane 6) skins. This indicated that the application of 4-OHT had induced Cre-mediated deletion of the Rb1 floxed alleles in cutaneous melanocytes only in Rb1F2/F2 mice that possessed the Tyr::CreER(T2) transgene (Figure 4B lane 8). Further, this result implies that cells containing the initial recombination event were still present in follicular melanocyte stem cells numerous hair cycles later, as has been reported elsewhere for this system (Bosenberg et al., 2006).
This indicated that irrespective of whether Rb1-loss occurred during embryonic development (Tonks et al., 2005) or post-natally, there were no overt effects on melanocyte homeostasis over the short term or long term (up to 16 months).
Collectively taken, our data raises the question of whether the apoptotic phenotype noted in the Yu et al. (2003) study is attributable to some other experimental variant and not Rb1-loss per se. For example, the incidence of Cre toxicity is a well-acknowledged and documented complication of the Cre/loxP system. Recently, Schmidt-Supprian and Rajewsky suggested that, when speaking of Cre toxicity, ‘Whilst mice or cells expressing the Cre transgene alone should be a routine control in conditional gene-targeting experiments, such control experiments unfortunately do not completely resolve the issue. Cre toxicity may become apparent only through the introduction of a targeted mutation, which by itself may not have an effect. Thus, Cre-mediated inactivation of an antiapoptotic gene or of a gene involved in DNA repair may lead to apoptosis of the mutant cells because of their inability to deal with Cre toxicity’ (Schmidt-Supprian and Rajewsky, 2007). In this context, the observed apoptosis of Rb1ΔX19/ΔX19 melanocytes shown by Yu et al. (2003) may be attributable to a dosage of TAT-Cre that induces apoptosis when Rb1, a known anti-apoptotic protein, is ablated. However, the potential for complications may not be merely limited to a Cre dosage effect but to the effect of the TAT domain on the fusion protein as it is highly charged and its presence in Rb1-ablated melanocytes may induce an apoptotic effect.
In conclusion, the relevance of RB1 to melanoma is not in question and this is highlighted by a recent CGH study that found that RB1-loss is an early and defining change in melanoma (Vincek et al., 2009). However, our data shows that Rb1 ablation alone does not lead to a short-term or long-term loss of melanocyte homeostasis and apoptosis either in vivo or in vitro, irrespective of the developmental stage at which the recombination event driving deletion occurs. This indicates that the effects of RB1-loss, whilst integral to the process of melanomagenesis, must be potentiated by further mutational events. The identity of other melanoma-predisposing mutations indicates that loss of collective PP function could be one such event driving melanoma.
This work was supported by a grant from the Cancer Council Queensland (442956). NKH is supported by a National Health and Medical Research Council fellowship. GW and GFK are supported by Cancer Council Queensland Fellowships. We wish to thank Dr Susan Woods for the generous gift of the Mitf antibody.