Depletion of CSN5 in HEK293 cells is not lethal
To analyze the effect of loss of deneddylation in human cells, we utilized the doxycyline-inducible shRNA system developed by Clevers and colleagues  to conditionally down-regulate Csn5 protein levels. Eight days of doxycycline treatment of cells carrying the inducible Csn5-specific shRNA resulted in a drastic reduction in both Csn5 mRNA and protein levels when compared to induced cells expressing a scrambled shRNA (Figures 1A and 1B). By contrast, the levels of other CSN subunits showed little or no change (Figure 1C). Despite this significant drop in Csn5 protein levels, few morphological changes were observed, the cells continued to grow and divide, and little change in steady-state cell cycle distribution was detected by FACS (data not shown). This is consistent with prior observations on normal diploid BJ1 fibroblasts depleted of Csn5 . Given that loss of CSN function results in accumulation of the neddylated form of cullins in yeast, Drosophila, and C. elegans, we analyzed the endogenous cullins 1–4 by western blot in human cells depleted of Csn5. Loss of Csn5 resulted in enhanced neddylation of all four cullins, although to varying extents (Figure 1D). The failure to observe complete conversion of all cullins to the neddylated species – as was seen for Cul1 in csn1Δ S. pombe cells  – may be due to residual Csn5 activity, an alternative cullin deneddylase, limiting Nedd8 attachment activity, or the sequestration of deneddylated cullins into complexes with CAND1 . Despite the change in Nedd8 modification, the total levels of cullins 1–4 were largely unaltered in Csn5-depeleted human cells, except for a modest reduction in the level of Cul2.
Loss of deneddylation causes a reduction in F-box protein levels
It has been proposed that loss of deneddylation could render F-box adaptor proteins unstable [6,15]. To test this hypothesis, we analyzed the levels of several endogenous F-box proteins in cells depleted of Csn5. In this experiment and throughout this report, we focused our effort on endogenous proteins to avoid potential artifacts associated with overexpression of recombinant proteins. Concomitant with loss of Csn5, we observed significant decreases in the steady-state protein levels of Skp2, cyclin F, Fbw7, Fbx4 and Fbx7 (Figure 2A). To test whether this decrease in protein levels was post-transcriptional, we performed RT-PCR on the corresponding mRNA transcripts. Despite the significant drop in protein levels, there was little change in mRNA levels, suggesting that for 4 of the 5 F-box proteins examined (Fbx4 being the exception) the loss of protein was due to reduced translation or increased degradation (Figure 2B). Interestingly, two F-box proteins, β-TrCP and Emi1, showed little or no change in protein level when Csn5 was suppressed (data not shown).
To determine if the decrease in F-box protein levels was specific and due to loss of JAMM isopeptidase activity, we ectopically expressed mouse Csn5 in cells depleted of endogenous Csn5 by shRNA. Transfection of wild-type mouse CSN5 cDNA, which differs by two nucleotides from human CSN5 in the region that is targeted by the silencing shRNA, restored wild-type levels of both Skp2 and cyclin F (Figure 3B). In contrast, expression of two different JAMM point mutants of mouse Csn5 failed to restore Skp2 and cyclin F protein levels, despite normal levels of expression of the mutant proteins (Figure 3B). The protein levels of Fbx4 could not be rescued by overexpression of wild-type mouse CSN5, suggesting that loss of Fbx4 arose from a promiscuous effect of the CSN5-directed siRNA and was not due to Csn5 knockdown (data not shown), Therefore, the reduction in Skp2 and cyclin F levels in Csn5-depleted cells is caused by the loss of CSN isopeptidase activity.
Loss of F-box proteins is dependent on the ubiquitin-proteasome system
We next examined whether the 26S proteasome was involved in the reduction of F-box protein levels. Treatment of Csn5-depleted cells with the proteasome inhibitor MG132 largely restored normal levels of Fbx7, cyclin F, and Fbw7, whereas Skp2 levels were only partially rescued (Figure 3A). By contrast, the loss of Fbx4 was not reversed by proteasome inhibition, which is consistent with our prior observation that Csn5 shRNA non-specifically diminished the levels of Fbx4 transcripts. Similar results were obtained for all proteins when we used the proteasome inhibitor LLnL. Therefore, the loss of F-box proteins in Csn5-depleted cells appears to depend largely upon the 26S proteasome.
The role of the proteasome in mediating the loss of F-box proteins in Csn5-depleted cells suggested that Csn5 might normally protect F-box proteins within SCF complexes from being 'autoubiquitinated' by the associated E2 enzyme and then degraded. To address this possibility we sought to examine the levels of a representative F-box protein in cells in which Csn5 was depleted and SCF activity was simultaneously inhibited. A C-terminal truncation mutant of Cul1 [Cul1 (1–428)] possesses dominant-negative activity . To determine if the loss of cyclin F in Csn5-depleted cells was SCF dependent, we transiently expressed either wild-type or Cul1 (1–428) in doxycycline-induced cells. Whereas expression of full length Cul1 did not alter cyclin F accumulation, expression of Cul1 (1–428) restored cyclin F to levels comparable to those seen in untreated cells (Figure 3C). These data indicate that loss of cyclin F was dependent upon an intact SCF complex.
Given the reduction in Fbw7 levels in Csn5-depleted cells, we analyzed two substrates of SCFFbw7 – cyclin E and c-myc – to see if they were stabilized. Down-regulation of Csn5 resulted in accumulation of both c-myc and cyclin E (Figure 4A) as well as cyclin E-associated histone H1 kinase activity (Figure 4B). Moreover, depletion of Csn5 resulted in a loss of cell cycle regulation of cyclin E protein abundance (Figure 4C), which is consistent with a defect in cyclin E turnover.
Overexpression of a non-degradable form of cyclin E results in chromosome instability . To examine chromosome stability in Csn5-depleted cells, we induced cells for 21 days with doxycycline and examined them for chromosome abnormalities by FISH and karyotyping. Despite a significant increase in cyclin E, we could not detect any chromosomal abnormalities (data not shown).