In this thesis, we tried to construct a fluorescence based method that can give us a brief and clear result without ambiguity. From previous studies, researchers took advantage of Western blot to detect Pol II ubiquitination on Rpb1, and they could identify mono- or poly-ubiquitination [28-30]. However, when we looked at our Western blot results, we found that there are many non-specific binding bands.
Anti-ubiquitin antibody bound not only to impurities in E2 and E3 enzyme but also to Pol II itself (figure 3.14B). This is a serious problem, since it would be difficult for us to distinguish monoubiquitinated Rpb1 from unmodified ones. Clearly, this was the drawback that pushed us to seek for alternative methods in ubiquitination research.
Fluorescent dye Alexa-488, which has a high quantum yield and higher stability [39], was used in ubiquitin conjugation. Our result shows that ubiquitin reacted almost completely in excess amoun of Alexa-488 dye (figure 3.8 B&D, no free ubiquitin peak shown in the results). In following ESI-TOF analysis, we confirmed that samples of R72C and R72C-A488 were indeed the expected products (figure 3.10 A&B), whereas samples of M[C]Q underwent N-terminal methionine excision (NME) [37, 38] and the N-terminal methionine of M[C]Q was excised (figure 3.9 A-C).
Since ubiquitination reaction involves several enzymes and reagents, many of them
50
should be freshly prepared to ensure a good experiment quality. In our experience, purifying E2 and E3 without an ideal procedure led to inactivity of these enzymes and unsuccessful ubiquitination (figure A1). In ubiquitination assay, we showed that Ubc5 (E2) and Rsp5 (E3) with SUMO tag could also catalyze ubiquitination (figure 3.13 and 3.15) as the tag-removed enzymes did (figure 3.14 and 3.16), though using the enzymes without SUMO tags are considered closer to the conditions in vivo.
In our works, we used fluorescent-dye-labeled ubiquitin as a probe to identify ubiquitinated proteins. When M[C]Q-A488 and D39C-A488 were used, Rpb1 band could be detected under fluorescent scanner (figure 3.13C, 3.14C, 3.15C and 3.16B), indicating that these two dye-labeled ubiquitin mutant were able to work similarly to wild-type ubiquitin in Pol II ubiquitination, and maybe also in ubiquitination of other substrates. On the other hand, when using R72C-A488, there was no fluorescent signal on Rpb1 band (figure 3.17C), indicating that R72C-A488 could not conjugate normally to the substrate. It might be attributed to the bulky molecule Alexa-488, which is conjugated at Cys72 of Ub R72C and is near C-terminal. Since Gly76 would be conjugated to the substrate directly in ubiquitination, C-terminal tail plays an important role in ubiquitination, and the appearance of a bulky group near C-terminal might hinder the normal function of ubiquitin. R72C-A488 could still be labeled onto E1 and E2, though the intensity is lower than M[C]Q-A488 and D39C-A488. In conclusion,
51
two out of three dye-labeled ubiquitin mutants could work, representing a positive result at the first step.
Besides, another thing we should also take note of is the signs of E3-independent ubiquitination. Our fluorescent images indicates that Pol II could also be monoubiquitinated in the absence of E3 (figure 3.13C, 3.14C, 3.15C and 3.16C), as long as E2 is in presence. The phenomenon of E3 independent ubiquitination has been stated previously [40, 41], but no studies before has reported this occurs on Pol II. The accidental discovery further marks the value of our method.
Nevertheless, considering the type of ubiquitination (mono- or poly-), we didn’t observe clear polyubiquitination signal (marks by a smeared band over 250 kDa) in our work. From our result, it seems that only one D39C-A488 molecule was conjugated to Rpb1 of Pol II (figure 3.15C); M[C]Q-A488 probably forms polyubiquitin chain on Pol II, but the evidence is not clear enough (figure 3.13C). Using a K63R mutant ubiquitin (e.g. M[C]Q/K63R-A488 or D39C/K63R-A488) in the assay, or reducing the amount of enzymes and ubiquitin to lower the interference from other proteins, might be feasible ways to clarify it. However, in our Western blot result, we could hardly find a smeared band when M[C]Q-A488 or D39C-A488 was used in ubiquitination assay (the leftmost lanes of figure 3.13B and 3.15B). In contrast, those lanes which used wild-type ubiquitin to carry out ubiquitination assay show bands at high molecular weight (over
52
200 kDa), and it possibly marks polyubiquitinated Pol II.
Therefore, there are two questions remained unanswered: (1) Was mono- or poly-ubiquitination (from previous studies, it should be a K63 chain) really formed in our system? (2) Why didn’t our Western blot show a similar result as wild-type protein, when dye-labeled ubiquitin was used in the assay? Trying to answer those questions, we looked into our previous trials, in which ubiquitin D39C and D39C/K63R were used and labeled by another fluorescent dye, Cy3 (GE Healthcare, UK). In those trials, Ub D39C-Cy3 and D39C/K63R-Cy3 were used in ubiquitination assay, and fluorescent image was also taken after that (figure A2C). In this figure, we can see D39C/K63R-Cy3 fluorescent signal only appeared as a single band at the top, indicating that only monoubiquitinated Pol II was formed; while D39C-Cy3 signal formed a smeared band at the same place, indicating that polyubiquitination might exist. When comparing D39C-Cy3 and D39C-A488, the only difference between them is the conjugated dye at position 39. It should be noted that when wild-type ubiquitin, Ub K29R and Ub K48R were used, smeared black lanes appeared on Western blot result, and Ub K63R didn’t show this feature (figure A2B). This result is in agreement with our previous study and previous research from other group [30]. However, in figure A3B, few antibodies were bound to dye-labeled ubiquitin, and no signal was seen at the position of Pol II Rpb1 (over 200 kDa) on lane 5 and lane 6 in Western blot. It is
53
contradictory, since fluorescent imaging shows that dye-labeled ubiquitin was indeed conjugated to Pol II, but Western blot does not show the same result. The possible explanation is the antibody we used in Western blot (rabbit anti-ubiquitin antibody;
Sigma-Aldrich, USA) is not that specific to dye-labeled ubiquitin as it is to wild-type and K-to-R-mutant ubiquitin. It probably points out another drawback of Western blot:
besides non-specific interactions, lacking affinity to desired targets might also be a problematic issue (even though we used polyclonal antibody in our study). Actually, this problem could already be seen in ubiquitin conjugated E2: in ubiquitination assay, large amount of wild-type ubiquitin was conjugated to E2 enzyme, which could be confirmed by silver staining (figure 3.13A, 3.14A, 3.15A and 3,17A), but few of them could be seen on Western blot (none could be seen on figure 3.13B, 3.15B and 3.17B, and only some could be ssen on figure 3.14B). Maybe it is because the epitopes of ubiquitin were covered by E2 and could not be identified and bound by antibody. If we also take non-specific binding, which we have mentioned above, into account, it might further reduce the credibility of Western blot results.
Above all, though dye-labeled protein is required, fluorescent imaging gives a more reliable result (e.g. figure 3.13C and 3.14C, where E2-Ub signals are clearly shown).
The availability of a commercial kit for detecting ubiquitination using fluorescent imaging and FRET (LanthaScreen™ Ubiquitin and SUMO Assay Reagents, Invitrogen,
54
Life Technologies, USA) also supports our hypothesis that fluorescent-dye-labeled ubiquitin could be used as a probe for high sensitivity and accuracy.
55