A new UBE3A overexpressing mouse for ubiquitin proteomic studies
We have previously introduced the bioUb mice, an in vivo tool to identify ubiquitinated proteins, reporting both the liver physiological ubiquitome [44] and how this ubiquitome is altered upon liver fibrosis [47]. Upon overexpression of an E3 ligase, more molecules of their substrates become ubiquitinated, and therefore these substrates become easier to identify and quantify by MS [46]. We generated a new UBE3A overexpressing mouse strain coding for the dual isoform UBE3A gene under the CAG promotor that is also used for the bioUb mice (Fig. 1A), and combined it with the bioUb mice. Unless mentioned otherwise, all through this work we have used this combined mouse (bioUb + UBE3A), which we call UBE3A-OE, and is homozygous for both pCAG-bioUb and pCAG-Ube3a. As described elsewhere [19] both UBE3A isoforms (long isoform II [O08759-1], and shorter isoform III [O08759-2], Supplementary Fig. 1A) should be expressed from their respective start codons, as the shorter isoform’s ATG codon has not been removed in this construct. Based on immunoblotting results on whole brain extracts of homozygous mice, the amount of total UBE3A in the brain of the newly generated mice was 4 to 5-fold times greater than in control mice (Fig. 1B and Supplementary Fig. 1B). Based on immunohistochemistry results on primary neurons, overall UBE3A localization remained unchanged (Supplementary Fig. 1C). Behavioural analysis of these mice at 1–7 weeks of age indicated very minor differences compared to the control mice (Supplementary Fig. 2).
The first step in our molecular analysis of this new mouse model was to define the global proteome changes caused by the increased UBE3A levels in the brain. One whole brain was collected for each of the three experimental replicates from both control and UBE3A overexpressing mice. After trypsin digestion, TMT labelled peptides of each sample type were analysed by LC-MS/MS. Highly reproducibly TMT data (Supplementary Fig. 3A) allowed the identification and quantification of 5244 protein groups (Supplementary Table 1A). Gene Ontology (GO) analysis of the total identified proteins determined by g:Profiler showed a highly significant enrichment of proteins within the cerebral cortex, cerebellum and hippocampus (respective p-values of 10− 248, 10− 215 and 10− 197), as expected. Furthermore, quantification with MaxQuant and Perseus software confirmed a significant increase of UBE3A peptides in the UBE3A overexpressing mouse brain (Supplementary Table 1B).
Isolation and quantification of the ubiquitinated proteome in the mice brain
In order to confirm that the UBE3A overexpressing mice could be used for ubiquitin proteomic studies with the bioUb strategy [44, 46, 48, 49], we first compared the brain samples from bioUb pull-downs on birA, bioUb and UBE3A-OE mice using several markers. The BirA antibody confirmed that the bioUb precursor was appropriately digested both in bioUb and in UBE3A-OE mice brains, as only the band corresponding to free BirA was detected (Supplementary Fig. 3B). Biotin immunoblotting (Supplementary Fig. 3C) on the BirA control mice only detected endogenous carboxylases, while bioUb containing mice displayed both free and conjugated bioUb, in addition to the carboxylases. Monitoring all ubiquitinated material showed no distinct pattern between control and bioUb mouse whole brain extracts, indicating that the amount of overexpressed ectopic biotinylated-ubiquitin did not significantly alter total ubiquitin levels (Fig. 1C).
Using the bioUb-based pull-down, we also confirmed that ubiquitinated material could be properly isolated from both bioUb containing brain samples (Fig. 1C). Importantly, no enrichment of either endogenous free ubiquitin or proteins conjugated with endogenous ubiquitin is observed in the elution of the control sample without bioUb, underlining the high specificity of the bioUb strategy. Similarly, immunoblotting with antibodies to polyubiquitinated proteins confirmed that overexpression of UBE3A did not significantly alter the ubiquitinated material (Supplementary Fig. 3D). The anti-ubiquitin antibody identified an intense band of about 10 KDa corresponding to free bioUb in the elutions of both bioUb expressing mice, but not in the whole cell lysate (Fig. 1C, indicated with an arrowhead). This further supports the idea that ectopically expressed bioUb in the brain is a minority form relative to the endogenous ubiquitin (shown with an arrow in Fig. 1C, only detectable in input samples). This minor amount of ectopically expressed bioUb is however sufficient to isolate a valuable fraction of the ubiquitinated proteome, as illustrated by the silver staining in Supplementary Fig. 3E.
UBE3A is a HECT-type E3 ligase, which conjugates ubiquitin through its active site cysteine before transferring it to its substrates [14]. Indeed, UBE3A was isolated in our bioUb-based pull-down, as the thioster bond between ubiquitin and the cysteine of UBE3A active site is not disrupted during the purification procedure. However, prior to running the SDS-PAGE gel, DTT is used to elute the samples from the beads, by disrupting the thioester bond between ubiquitin and UBE3A. Hence, UBE3A migrates as an apoenzyme, and a single band is observed corresponding to its own Mw (Fig. 1D). According to the intensity of that band, the amount of active UBE3A in the mouse brain appears to be about 6-fold higher in the UBE3A-OE mice than in the control mice (Fig. 1D and Supplementary Fig. 3F), similar to the increase in total UBE3A (Fig. 1B).
Having confirmed that overexpressed UBE3A is active in the UBE3A-OE mouse brain, and that the ubiquitinated material could be properly isolated from both genotypes, triplicate samples were prepared from P1 mice for both UBE3A-OE and control mice. In brief, 12 brains (~ 1 g in total) were collected for each experimental replica, homogenised, and subjected to the bioUb pull-down to isolate biotinylated and hence, ubiquitinated proteins [44]. Similar levels of ubiquitinated material were enriched in all replicas of both experimental conditions as observed by western blotting and silver staining analysis of the pull-down samples (Supplementary Fig. 4A and 4B). In order to detect by MS the proteins isolated by Neutravidin pull-down, samples were first fractionated by SDS-PAGE, Coomassie-stained and all gel lanes cut into four slices as illustrated in Supplementary Fig. 4C. Next, proteins were in-gel digested with trypsin, and the resulting peptides analysed by shotgun proteomics in a Q Exactive HF-X LC-MS/MS system. Finally, MS-derived raw data was processed using MaxQuant and the obtained label free quantitative (LFQ) intensities output analysed with Perseus software.
A total of 1344 protein groups were identified with at least 1 unique peptide and quantified by this LFQ proteomics approach (Supplementary Table 1A). Venn diagrams showed a very high overlap (in the range of 87–92%) between proteins identified across replicas of the same experimental condition (Supplementary Figs. 4D). Multi-scatter plots correlating measured LFQ values for all 6 samples (Supplementary Fig. 4E) indicate that the similarity between distinct replicas of the same experimental condition is higher (Pearson correlation coefficient (r) 0.986–0.992) than between distinct datasets (r ranging from 0.887–0.893). Imputation of missing values was performed as earlier described [46]; these imputations allow for a correct statistical analysis of the data [50].
Multiscatter plots across whole brain samples quantified by TMT (Supplementary Fig. 3A) or the brain ubiquitome quantified by LFQ (Supplementary Fig. 4E) indicate that UBE3A overexpression did not cause major changes in the overall correlation of quantified intensities. However, PCA analysis of the whole brain samples (Fig. 2A) as well as the brain ubiquitome (Fig. 2B) indicate a clear clustering of the samples overexpressing UBE3A as compared to the control samples.
Comparison of the whole brain proteome and its ubiquitome
When comparing across both proteomic datasets, more than 85% of the proteins detected in the ubiquitome analysis had also been detected in the whole mice brain extracts (Fig. 2C). When analysing for enrichment (relative to the TMT dataset) by g:Profiler, several molecular functions were enriched in the brain total ubiquitome dataset, the most predominant term being cell adhesion molecule binding with a p value of 10− 17. Furthermore, we noted the cellular compartment terms vesicle, membrane and synapse to be predominantly enriched (p values of 10− 34, 10− 16 and 10− 8, respectively) in the brain ubiquitome relative to the brain proteome. These findings support previous work indicating that ubiquitination in the brain might particularly regulate synaptic function by acting on endocytic and lysosomal pathways [51].
It is often assumed that ubiquitination’s main endpoint is proteasome-driven degradation, with a recent study again searching for substrates of UBE3A by identifying downregulated proteins in various AS models [52]. We therefore tested this widespread assumption that ubiquitination enriched proteins would negatively correlate with their total enrichment in the brain (Fig. 2D). The complete lack of correlation (r = 0.08, p = 0.008) is astonishing. Only UBE3A itself stands out with a direct correlation, as expected from being both overexpressed at the whole brain extract level and enriched in the ubiquitinated material as an active ubiquitin carrier. However, no negative correlation was observed as it would be predicted by the hypothesis that increased ubiquitination driven by UBE3A would translate into a corresponding decrease on protein abundance levels. Indeed, a thorough analysis of available literature proteomic data on ubiquitomes identified by diGly antibodies and their corresponding whole proteome datasets indicate that the expected negative correlation has never yet been seen (Osinalde, manuscript in preparation).
Whole brain proteome changes induced by UBE3A
Given the fold-change compression of TMT-based MS acquisition, instead of predefining a fold change threshold in the volcano plot to categorize proteins as enriched, reduced or non-changed, using Perseus software [50] we applied a Student´s t-test with permutation based false discovery rate (FDR) correction (FDR cutoff = 0.01, s0 fuzz factor = 0.1). A total of 52 proteins appeared significant enriched in the UBE3A-OE mouse brain, while another 52 proteins appeared depleted (Fig. 3A, Supplementary Table 1B). Relative to the total proteome, g:Profiler analysis identified a very significant enrichment of synaptic proteins in the brain extracts of UBE3A-OE mice, as well as of proteins regulating calmodulin binding and of proteins involved in chorea and status epilepticus, with a concomitant depletion of nucleic acid binding proteins being identified (Fig. 3B). The proteins belonging to these enriched GO terms are shown respectively with filled blue and orange circles in Fig. 3A and are shown highlighted in Supplementary Table 1B. A correlating enrichment of membrane and nuclear protein abundances was also observed (Fig. 3C).
Proteomic analysis of the brain ubiquitome
A detailed analysis of the list of 1344 putative ubiquitin conjugates isolated from the mouse brain at P1 via the bioUb system revealed that, as noticed also in earlier studies [46, 48, 49, 53], many enzymes of the ubiquitin proteasome pathway were present. We detected as active ubiquitin carriers two E1 enzymes, as well as 17 conjugating E2 enzymes and at least 20 HECT-type and RBR-type ligating E3 enzymes (Supplementary Table 1C). The simplest explanation is that these enzymes are enriched through their transiently thioester-bound ubiquitin, which these enzymes load prior to its transfer to the next enzyme in the cascade or to their substrate. Additionally, and due to the tight regulation of the ubiquitin proteasome system itself via ubiquitination [54], up to 26 subunits of the proteasome were identified as ubiquitinated in the brain, as well as 27 other (non-HECT) E3 ligases and 21 DUB enzymes (Supplementary Table 1D).
As expected, endogenous biotinylated carboxylases PCB, PCCase subunit alpha, ACC1 and MCCase subunit alpha were also isolated in addition to ubiquitinated proteins. The volcano plot revealed these proteins to have LFQ ratios near to one, indicating that similar amounts of biological material were extracted for both experimental conditions (Fig. 4A). In regards to endogenous Ub and bioUb sequences (combined under the headings of Ubb and Ub, respectively), a ratio near to 1 was also observed, confirming again that a similar amount of total ubiquitinated material was isolated from both UBE3A-OE and control mice. This was also the case for UBA1, the main ubiquitin activating enzyme, which was also purified as a ubiquitin carrier (Supplementary Fig. 4F).
In accordance with what we observed by western blotting (Fig. 1D), we also detected by MS enriched active UBE3A in the UBE3A-OE mice compared to the control brains. In this case, the LFQ intensity ratio calculated over 43 unique peptides indicated a 3.8-fold enrichment in the UBE3A-OE brain (Table 1), similar to the increase of active UBE3A determined by WB (Fig. 1D). In contrast to this, 18 other HECT and RBR E3 ligases as well as 13 ubiquitin E2 conjugating enzymes isolated as active ubiquitin carriers in the brain samples had no significant changes relative to the control samples (Supplementary Table 1C). However, two other E2 enzymes appeared less abundant in the UBE3A brains: UBE2N and UBE2L3 (Fig. 4A). Based on mass action law, it could be expected that active E2 enzymes would become depleted of their carried ubiquitin moieties in the presence of an excess of their client E3, pointing towards those two enzymes as being ubiquitin providers for UBE3A. Indeed, UBE2L3 is well established as the E2 enzyme of reference for UBE3A [13]. Our observation of UBE2N being similarly depleted in the presence of excess UBE3A suggested it would be very likely for UBE2N to act as an additional E2 enzyme for UBE3A, so we tested this possibility using a standard in vitro ubiquitination assay [55]. Using UBE2L3 as a positive control, we found UBE2N by itself not to be able to support UBE3A autoubiquitination activity (Supplementary Fig. 4G). However, UBE2N has been described to require the support of the cofactor UBE2V1 in order to be active [56]. Indeed, in the presence of both UBE2N and UBE2V1, UBE3A autoubiquitination was detected (Supplementary Fig. 4G), confirming UBE2N/ UBE2V1 as a novel E2 for UBE3A in mice brain samples and in vitro.
Brain ubiquitome changes induced by UBE3A
We initially applied a standard LFQ ratio threshold of at least 2-fold (with a p-value < 0.05) to categorize ubiquitinated proteins as enriched, reduced or non-changed [46]. A total of 137 proteins appeared enriched (top-right quadrant of Fig. 4A and Supplementary Table 1E and 1F), with 122 proteins fulfilling the following criteria: to be identified with at least 2 unique peptides, and quantified in at least 4 of the 6 experimental replicas, or else in the 3 replicas of one of the experimental conditions (Supplementary Table 1F and 1G).
GO analysis of these 122 putative UBE3A substrates using g:Profiler reveals an enrichment in the membrane-localized category when compared to the whole brain ubiquitome (Figs. 4B and 4C), with a significant enrichment of active transmembrane transporter proteins being as well observed. Nearly all the proteins identified as UBE3A substrates have been described to be synaptic proteins (Supplementary Table 1G), according to a recent literature analysis providing a list of proteins that conform the synaptic proteome [57]. Furthermore, a significant enrichment of seizure related proteins amongst the putative UBE3A substrates was also detected (Fig. 4C and Supplementary Table 1G). Indeed, an automatized search on ClinVar for neurodevelopmental disorder (NDD)-associated terms [see 58] identified that genes coding for 36 proteins out of the queried 122 cause monogenic NDDs (Supplementary Table 1G). Given the prominent role of UBE3A in LTP induction and maintenance [25, 31], we also performed a bibliographic search on the 122 identified UBE3A-induced ubiquitinated proteins, and found that at least 28 of them are involved in the molecular mechanism of LTP (Supplementary Table 1G).
Interestingly, of the 122 putative UBE3A substrates, 82 had LFQ values in both control and overexpression samples with the other 40 being detected as ubiquitinated only upon UBE3A overexpression (Supplementary Table 1G). A graphic representation of their LFQ intensity increase (Fig. 4D) suggests that these UBE3A substrates went missing on the control sample due to their values being too close to the detection limit threshold in the control brains. And indeed, PRM allowed the detection in the control sample (see Supplementary Table 1G and Fig. 5A) of a number of these proteins not detected in the LFQ experiments. In order to further challenge the list of 122 candidate UBE3A substrates, a permutation-based FDR correction was applied using Perseus to this dataset, with standard parameters of FDR cutoff and an s0 of 0.5, resulting in only 5 proteins being excluded (Supplementary Table 1H). A more stringent Benjamini-Hochberg test with a 0.01 FDR cutoff excluded 41 further proteins (Supplementary Table 1H). A list of selected UBE3A substrates complying with all three statistical tests and identified in both control and UBE3A samples is given in Table 1 and Supplementary Table 1I. The candidate UBE3A substrates that complied with all three tests but were only identified in the UBE3A samples are given in Supplementary Table 1J.
Table 1. Identified ubiquitination substrates of UBE3A. Protein Uniprot IDs, gene names and their Mw are provided for the UBE3A substrates complying with all three statistical tests performed and present in both UBE3A-OE and control samples. The complete table is provided in Supplementary Table 1. Two E2 enzymes depleted of their active carrier form in the UBE3A overexpressing brains are also shown. Actual fold change, -log10(p-value) and number of unique peptides are shown for both whole brain lysates (TMT experiment) and biotin-based pull-downs (LFQ experiment). Proteins with empty values are those not identified in the TMT experiment. Previously described involvement in LTP (as per number of papers), seizures (as being identified per GO terms in g:Profiler, or via Pubmed, Pm) or NDDs (as per number of unique mutations recorded in the ClinVar database) are indicated in the Functional Evidence column.
|
|
|
|
Whole brain lysate
|
Biotin pulldown
|
Functional evidence
|
Uniprot IDs
|
Gene names
|
MW [kDa]
|
Cover [%]
|
Fold change
|
-log p value
|
Unique peptides
|
Fold change
|
-log p value
|
Unique peptides
|
LTP papers
|
Seizures
|
Clin
Var
|
O08759
|
Ube3a
|
99,8
|
50,5
|
1,45
|
2,51
|
30
|
3,7
|
4,4
|
43
|
12
|
GO
|
11
|
Q6PIC6
|
Atp1a3
|
112
|
41,2
|
1,19
|
1,48
|
29
|
5,0
|
5,4
|
21
|
|
GO
|
83
|
Q9DBG6
|
Rpn2
|
69,1
|
40,1
|
0,98
|
0,13
|
18
|
2,0
|
2,6
|
16
|
|
|
|
P48962
|
Slc25a4
|
32,9
|
37,2
|
1,06
|
0,45
|
7
|
3,9
|
4,2
|
5
|
|
GO
|
|
O55143-2
|
Atp2a2
|
110
|
37
|
|
|
|
5,7
|
5,5
|
22
|
1
|
GO
|
|
Q6PIE5
|
Atp1a2
|
112
|
37
|
1,15
|
1,16
|
21
|
3,5
|
4,7
|
18
|
|
GO
|
2
|
Q9EPN1-4
|
Nbea
|
326
|
35,7
|
1,14
|
0,93
|
28
|
2,5
|
4,7
|
72
|
2
|
GO
|
13
|
Q91YQ5
|
Rpn1
|
68,5
|
33,7
|
0,93
|
0,46
|
26
|
2,2
|
5,7
|
14
|
|
|
|
Q8VDN2
|
Atp1a1
|
113
|
33,3
|
1,11
|
0,88
|
19
|
3,9
|
4,9
|
19
|
|
GO
|
3
|
Q99PV0
|
Prpf8
|
274
|
32,4
|
0,96
|
0,18
|
16
|
2,1
|
3,7
|
62
|
|
GO
|
|
Q5SQX6
|
Cyfip2
|
146
|
29,6
|
1,04
|
0,28
|
18
|
2,5
|
5,8
|
14
|
|
GO
|
5
|
Q6ZQ08
|
Cnot1
|
267
|
28,8
|
|
|
|
3,9
|
3,4
|
51
|
|
GO
|
2
|
P55096
|
Abcd3
|
75,5
|
26,4
|
1,13
|
0,89
|
10
|
7,8
|
6,4
|
12
|
|
|
|
Q8QZY1
|
Eif3l
|
66,6
|
25,5
|
0,94
|
0,38
|
19
|
2,7
|
3,1
|
13
|
|
|
|
Q8K2C9
|
Hacd3
|
43,1
|
24,3
|
1,06
|
0,35
|
6
|
8,6
|
3,8
|
7
|
|
|
|
Q9Z1G4-3
|
Atp6v0a1
|
95,6
|
24,2
|
1,11
|
0,93
|
18
|
9,8
|
2,7
|
18
|
|
|
|
Q99104
|
Myo5a
|
216
|
23,9
|
1,05
|
0,27
|
23
|
3,3
|
2,8
|
34
|
1
|
GO
|
3
|
P28660-2
|
Nckap1
|
119
|
23,2
|
|
|
|
3,9
|
4,3
|
21
|
|
|
3
|
P53995
|
Anapc1
|
216
|
22,5
|
1,02
|
0,11
|
2
|
2,2
|
3,0
|
33
|
|
GO
|
|
G5E829
|
Atp2b1
|
135
|
22,2
|
1,14
|
0,99
|
19
|
2,5
|
4,4
|
11
|
|
|
|
P61027
|
Rab10
|
22,5
|
22
|
1,04
|
0,27
|
11
|
5,5
|
3,6
|
3
|
|
|
|
Q9R0K7
|
Atp2b2
|
133
|
21,6
|
1,15
|
0,96
|
19
|
3,9
|
3,6
|
12
|
|
|
|
Q9Z0E0-2
|
Ncdn
|
77,3
|
20,9
|
1,14
|
1,06
|
9
|
4,9
|
5,0
|
12
|
3
|
Pm
|
4
|
Q9JLN9
|
Mtor
|
289
|
20,2
|
0,98
|
0,12
|
7
|
18
|
2,6
|
36
|
91
|
GO
|
21
|
E9PVA8
|
Gcn1l1
|
293
|
20,1
|
0,99
|
0,09
|
35
|
2,1
|
3,8
|
42
|
|
|
|
Q7TPB0
|
Lppr3
|
76,7
|
19,1
|
0,97
|
0,24
|
14
|
2,5
|
3,5
|
11
|
|
|
|
Q04690-2
|
Nf1
|
317
|
18,2
|
|
|
|
3,1
|
2,9
|
40
|
11
|
GO
|
9
|
Q3UUQ7
|
Pgap1
|
105
|
17,7
|
0,94
|
0,43
|
12
|
11
|
3,0
|
10
|
|
GO
|
21
|
P56564
|
Slc1a3
|
59,6
|
17,5
|
1,16
|
1,27
|
8
|
2,3
|
3,9
|
6
|
4
|
GO
|
|
Q9CY27
|
Tecr
|
36,1
|
16,6
|
1,02
|
0,10
|
8
|
4,7
|
4,3
|
5
|
|
GO
|
1
|
Q9QYS2
|
Grm3
|
99,1
|
15,6
|
1,11
|
0,94
|
14
|
3,0
|
4,0
|
10
|
1
|
GO
|
|
Q3UH60
|
Dip2b
|
171
|
15,4
|
0,94
|
0,37
|
7
|
3,4
|
3,9
|
16
|
|
GO
|
2
|
Q920I9-2
|
Wdr7
|
160
|
12,6
|
|
|
|
2,8
|
3,5
|
13
|
|
|
|
E9Q8I9
|
Fry
|
339
|
12,3
|
|
|
|
2,2
|
5,7
|
30
|
|
|
|
Q91V92
|
Acly
|
120
|
11,7
|
0,97
|
2,19
|
37
|
2,1
|
3,4
|
9
|
|
|
|
P43006
|
Slc1a2
|
62
|
11,2
|
1,17
|
4,83
|
1
|
3,1
|
3,6
|
4
|
6
|
GO
|
|
P42859
|
Htt
|
345
|
11,0
|
|
|
|
6,0
|
3,2
|
26
|
12
|
GO
|
|
O70228
|
Atp9a
|
119
|
10,9
|
1,14
|
3,03
|
10
|
25
|
2,1
|
9
|
|
|
|
Q9ES97-3
|
Rtn3
|
25,4
|
9,7
|
|
|
|
3,2
|
3,5
|
2
|
|
|
|
B2RQC6
|
Cad
|
243
|
9,4
|
1,04
|
0,73
|
13
|
2,5
|
2,7
|
16
|
3
|
GO
|
|
E9Q3L2
|
Pi4ka
|
237
|
9,3
|
|
|
|
5,0
|
3,1
|
15
|
|
GO
|
|
Q8R1A4
|
Dock7
|
241
|
7,6
|
|
|
|
2,3
|
3,5
|
14
|
|
GO
|
2
|
Q8CIQ7
|
Dock3
|
233
|
7,5
|
1,09
|
1,30
|
2
|
5,9
|
2,9
|
13
|
|
GO
|
5
|
Q5SSH7-2
|
Zzef1
|
324
|
7,4
|
|
|
|
2,0
|
2,8
|
17
|
|
|
|
Q6VNB8
|
Wdfy3
|
392
|
7,3
|
|
|
|
2,2
|
3,5
|
20
|
|
GO
|
5
|
P61620
|
Sec61a1
|
52,3
|
6,3
|
1,08
|
0,73
|
3
|
22
|
2,3
|
2
|
|
GO
|
|
Q60714
|
Slc27a1
|
71,3
|
3,4
|
0,89
|
0,58
|
2
|
3,9
|
2,2
|
2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
P61089
|
Ube2n
|
17,1
|
55,3
|
0,93
|
0,63
|
9
|
0,42
|
4,7
|
7
|
|
|
|
P68037
|
Ube2l3
|
17,9
|
43,5
|
0,95
|
0,39
|
6
|
0,33
|
4,4
|
5
|
|
|
|
When looking more closely to the size of the identified proteins in the ubiquitome we identified a significant shift on the molecular weight of the proteins enriched upon UBE3A overexpression (Fig. 4E). A one-way ANOVA performed on the lognormal weight distributions revealed a significant difference in the size of the identified proteins with proteins with UBE3A-induced ubiquitination having bigger molecular weight when compared to the whole proteome, ubiquitome or to proteins whose levels of ubiquitination were reduced in the presence of UBE3A.
For the purpose of validating the identified UBE3A substrates in independent biological replicas obtained from mouse brains subjected to the same bioUb protocol as earlier described, we tested two possible strategies: (i) western blotting and (ii) targeted proteomics by Parallel Reaction Monitoring (PRM). The latest proved particularly useful for validation of proteins with larger Mw-s or for which antibodies were not available. We monitored in new biological replicates from both UBE3A and control mice an average of 3 unique peptides per protein corresponding to eleven potential UBE3A substrates (MTND4, IP3R 1, mTOR, MED23, Tuberin, CACNA1B, Huntingtin, BIG1, DOCK3, PI4KA, INTS1). Additionally, we monitored endogenously biotinylated PCB carboxylase as a negative control (UBE3A-OE/Ctr ratio 1.13 ± 0.05) and UBE3A as a positive control (UBE3A-OE/Ctr 6.9 ± 2.7). UBE3A displayed a similar value to the increase determined by WB (about 6-fold higher in the UBE3A-OE mice than in the control mice, Fig. 1D), and also in close agreement with the MS intensity values for UBE3A peptides based on LFQ values (about 4-fold, Supplementary Table 1).
Analysis of the PRM data confirmed that all the above mentioned 11 proteins are indeed more abundant on the ubiquitinated material enriched upon UBE3A overexpression (Fig. 5A and Supplementary Table 2), in agreement with the enrichment identified by LFQ. Among these, WB confirmed MTND4 ubiquitination being mediated by UBE3A. Despite the less reproducible nature of WBs, when combining with the bioUb approach it allows to detect the amount of ubiquitin molecules conjugated to the substrate of interest [44, 48, 49]. Half of the validated proteins (MTND4, MTCH1 and NCDN) display a polyubiquitination pattern by western blotting (Fig. 5B and Supplementary Fig. 6), while another two (RAB3A and FAR-1) appear as mostly with one single band corresponding to the shift caused by a single ubiquitin molecule.