Ubiquitylation of p53 by the E3 Ligase LNX1

4.6.1 LNX1 induces reduction of p53 levels in cellulo

LNX1 is a member of the LNX family of proteins which contain a RING domain and PDZ domains (for details see 1.3.1). LNX1 was identified as interaction partner of Numb and it was shown that LNX1 catalyzes ubiquitylation of Numb and targets it for proteasomal degradation and thereby regulating the NOTCH pathway (Nie et al., 2002). Multiple other interaction partners of LNX1 have been identified including the high risk HPV E6 proteins (Weber, 2009). However, we were not able to elucidate the function of this potential interaction, since LNX1 does not appear to be a ubiquitylation substrate of the E6/E6AP complex within cells and is not an E3 ligase for E6 ubiquitylation (Weber, 2009).

Furthermore, no effect of E6 on LNX1 ligase activity was observed (Schätzle 2011, under supervision). When we used Flag-LNX1 in a p53 degradation assay in H1299 cells as control, we observed that levels of p53 are decreased in the presence of Flag-LNX1 (data not shown).

The tumor suppressor protein p53 has critical functions including regulation of cell death, senescence and proliferation (reviewed in Efeyan and Serrano, 2007). p53 is tightly regulated by post-translational modifications including ubiquitylation and subsequent proteasomal degradation.

Thus, the obtained result pointed to an unexpected role of LNX1 in the degradation of p53. To further address this issue, we tested whether the RING E3 ligase activity is needed to observe the decrease in p53 levels. Therefore, we analyzed p53 protein levels in the presence of wild-type Flag-LNX1 or Flag-LNX1 C48A, which has a mutation in the RING domain and is inactive as an E3 ligase. Indeed, coexpression of LNX1 leads to reduced p53 levels and reduction of p53 levels by LNX1 is dependent on LNX1 E3 ligase activity (figure 37A).

Moreover, we investigated the effect of coexpression of Mdm2 and wild-type Flag-LNX1 or Flag-LNX1 C48A on p53 levels. Coexpression of wild type Flag-LNX1 and Mdm2 lead to a further reduction of p53 levels compared to samples with only one E3 ligase overexpressed, whereas coexpression of Flag-LNX1 C48A and Mdm2 did not show an additive effect on p53 degradation (figure 37A).

Figure 37: p53 levels are reduced in the presence of LNX1

A: p53 degradation assay was performed in H1299 cells. Cells were transfected with the respective expression vectors for p53, Mdm2, Flag-LNX1wt and Flag-LNX1 C48A. Cells were lysed under native conditions and equalized for transfection efficiencies. Lysates were analyzed by SDS-PAGE, followed by western blot analysis using anti-Flag and anti-p53 (DO-1) antibodies (Nicole Richter-Müller, under supervision). B: Schematic presentation of DHRF-HA-ubiquitin-p53 system. (DHFR: dihydro folat dehydrogenase). C: Degradation assay was performed in H1299 cells. Expression vectors coding for the respective proteins were transfected in H1299 cells. DHFR-HA-ubiquitin-p53 is expressed as one polyprotein and cleaved by ubiquitin specific proteases into p53 and DHFR-HA-ubiquitin. Cells were lysed under native conditions equalized for transfection efficiencies and analyzed by SDS-PAGE, followed by western blot analysis using anti-Flag, anti-HA and anti-p53 (DO-1) antibodies. Levels of p53 and ubiquitin were quantified (ImageJ) and p53 levels were adjusted to DHFR-HA-ubiquitin signals (Dolde, 2013, under supervision).

The above results indicate that the E3 ligase activity of LNX1 is necessary for reduction of p53 levels and that coexpression of Mdm2 and Flag-LNX1 has an additive effect on p53 levels.

To exclude that these results are due to differences in transfection efficiencies and to show that the decrease occurs on the posttranscriptional level, we repeated the degradation assay by transfecting a construct encoding DHFR-HA-ubiquitin-p53 (figure 37B), which is expressed as one polyprotein. This polyprotein is cotranslationally cleaved by ubiquitin

specific proteases into p53 and DHFR-HA-ubiquitin resulting in same amounts of both proteins (figure 37B and C). Hence, a clear conclusion about p53 degradation within cells can be drawn by determining the ratios between DHFR-HA-ubiquitin and p53. Note that DHFR-HA-ubiquitin is a rather stable protein (Varshavsky, 2005); quantification of p53 levels showed a dose-dependent reduction of p53 levels by Flag-LNX1 dependent on its E3 ligase activity (figure 37C).

4.6.2 GST-LNX1 binds to p53 in vitro

As we observed reduction of p53 levels by Flag-LNX1 in cells dependent on its E3 ligase activity we speculated that this is due to ubiquitylation of p53 by LNX1 and its subsequent degradation by the 26S proteasome. If this is the case, both proteins should interact with each other. Therefore, we performed an in vitro binding assay using bacterially expressed GST, GST-LNX1 or GST-Mdm2 and in vitro translated p53. We used as positive control in vitro translated Numb, which is a known interaction partner of LNX1 and Mdm2 (Nie et al., 2002;

Yogosawa et al., 2003). This assay revealed that GST-LNX1 is able to interact with p53 (figure 38). However, GST-Mdm2 binds stronger to p53 than GST-LNX1. In contrast, binding of Numb to GST-LNX1 was more efficient than to GST-Mdm2 (figure 38).

Figure 38: GST-LNX1 binds to p53 in vitro

GST pulldown assays were performed with GST-fusion proteins and in vitro translated p53 and Numb.

Bound proteins were detected by SDS-PAGE and subsequent fluorography (left panel). Coomassie staining shows input levels of GST proteins used for pulldown experiments (right panel) (Nicole Richter-Müller, under supervision).

4.6.3 p53 is a ubiquitylation substrate of LNX1

Since we could show that p53 and LNX1 bind to each other, we next investigated whether LNX1 is able to ubiquitylate p53 in vitro. To do so, we performed an in vitro ubiquitylation assay using baculovirus expressed His-p53 as substrate. GST-Mdm2 was used as positive control. As expected, we detected ubiquitylation of p53 by GST-Mdm2 shown by the reduction of levels of unmodified p53 and the appearance of a high molecular weight smear in samples with GST-Mdm2 and ubiquitin (figure 39). Moreover, similar effects were obtained in the presence of GST-LNX1 and ubiquitin showing that GST-LNX1 is able to ubiquitylate p53 in vitro. Furthermore, ubiquitylation of p53 by LNX1 occurs in a dose-dependent manner (figure 39).

Figure 39: GST-LNX1 ubiquitylates p53 in vitro

In vitro ubiquitylation assay was performed with baculovirus expressed His-p53 as substrate and GST-LNX1 or GST-Mdm2 as E3 ligase. Samples were analyzed by SDS-PAGE and subsequent western blot analysis using anti-p53 antibody (DO-1) (Dolde, 2013, under supervision).

In the next step, we investigated whether LNX1 is able to ubiquitylate p53 in cellulo. In addition we analyzed if LNX1 is able to neddylate p53 in cellulo, as from our experience in cellulo neddylation assays are more efficient than ubiquitylation assays (Scheffner group, unpublished data). To do so, we performed a His-ubiquitin/NEDD8 assay in the presence and absence of HA-HR23A-TUBEs (see figure 17) in H1299 cells (figure 40). TUBEs serve as stabilization tool for ubiquitylated proteins (for details see 4.3.2)

In contrast to overexpression of Mdm2, where an enhanced neddylation of p53 was observed (figure 40, comparing lane 3 and 11), overexpression of Flag-LNX1 did not have a significant effect on p53 neddylation (figure 40, comparing lane 3 and 7). Moreover, expression of HA-TUBEs did not have an effect on p53 neddylation. Ubiquitylation of p53 by Flag-LNX1 could not be detected in the absence of HA-TUBEs even though lower p53 levels were found in the input samples (figure 40, lane 6). However, ubiquitylation of p53 by Mdm2 which

should have served as positive control could also not be shown in this assay (figure 40, lane 10). In the presence of HA-TUBEs a weak high molecular weight smear was detected dependent on the expression of Flag-LNX1 with an intact RING domain (figure 40, comparing lane 4, 8, 12). This result may indicate that LNX1 is able to ubiquitylate p53 in cellulo.

Figure 40: Analysis of ubiquitylation and neddylation of p53 within cells

H1299 cells were transfected with the respective expression constructs. Input samples were lysed under native conditions and equalized for transfection efficiencies. Pulldown samples (Ni²⁺PD) were lysed under denaturing conditions and PD was performed with Ni-NTA agarose to bind ubiquitylated or neddylated proteins. Input and PD samples were analyzed by SDS-PAGE followed by western blot analysis using anti-HA, anti-Flag and anti-p53 (DO-1) antibodies as indicated (Nicole Richter-Müller, under supervision).