V.1. Stability and Aggregation Distinctness among LC3 Variants
The fusion MBP-ULD could be treated with the Factor Xa at room temperature for 3 days without showing the significant appearance of the over-digested ULD as examined by the mass spectrometry (Fig. 24). This observation, together with the denaturation curves of the ULD (Figs 31 & 32) showing the notably delayed denaturations in these denaturants, suggests that the ULD itself should be a compact entity.
By contrast, in the Factor Xa treatment of the MBP-LC3 T12D, the time for the reaction at 4°C (i.e., 3 days) should not be longer than a half for that in the wild type or in the T6D (i.e., 7 days, also at 4°C), otherwise the LC3 T12D could be over-digested into smaller peptides or proteins which had been observed in the final check by MS showing a mass peak 11734 Da, corresponding to the T12D losing the first 19 amino acid residues on the NHD (data not shown). This suggests that the phosphomimetic aspartate, or probably phosphorylated Thr12, could render the NHD more accessible to the surrounding proteolytic agents, and implies that the NHD might bind less compactly with the ULD in the LC3 T12D. This idea is also supported in the denatuaration curves (Figs 31 & 32), which showed the much earlier or larger decrease of CD222 in T12D than in the wild type, suggesting lower structural stability in the phosphomimetic
47
mutant. Taking the ULD denaturation curve as reference, if this drastic decease of the CD222 could be accounted mostly by the denaturation of the α-helices, the lower stability on the T12D mutation might suggest changed dynamics on the NHD.
In this study, the ULD was more prone to aggregation than any other LC3 variants under high concentrations at the higher temperature (e.g., the high-speed centrifugation before the FPLC injection at room temperature, p. 30-31), and this might be related to the proposed function of the ULD in the LC3 multimerization;5 thus the NHDs on the LC3s are suggested to be regulatory in the play.
The pH change was reported during the progress of some autophagic activities (e.g., starvation-induced autophagy) and could range between 7.1-7.7.32 Thus the CD spectrum and the denaturation curve of the LC3 T12D, which showed the higher susceptibilities to the denaturant perturbations (Figs 31 & 32), were also taken at pH 7.9.
The results did not show significant difference both in the CD spectra (Fig. 29) and the urea denaturation curves (Fig. 33), and they thus far could not tell possible effects of the changing pH on the LC3 phosphorylation.
48
V.2. The Limitations in Using the Phosphomimetic Mutants and Presenting the CD Denaturation Curves
In this study, the phosphomimetic mutants of the LC3 with the phosphorylable threonines replaced by aspartates were utilized instead of using the phosphoproteins, since the point mutageneses provide faster probes to examine whether these phosphorylaitons could have structural significance; nevertheless, the charge distributions and the steric effects of these aspartates on LC3 cannot be identical to those of the real phosphorylations, but the results in this study give us an idea that the pure phosphoproteins, which could be less readily obtained, would be worth preparing and studying on further.
Circular dichroism can only give the ensemble of the secondary structures of a protein and be used to estimate their compositions,30 but it cannot show the detailed transitions of any given structures during protein denaturations. In other words, if we were to see exactly how the NHD and the ULD behave and whether the existence of one domain influences the unfolding process of the other during LC3 denaturation, other techniques such as nuclear magnetic resonance (NMR), X-ray crystallography, etc. may be needed.
For the economic reuse and reducing the random error of the protein concentrations in preparing the samples for the denaturation curves, I prepared the
49
protein stock solutions in either the phosphate/NaCl or 8 M urea (or 6 M GuHCl)/phosphate/NaCl having the same protein concentration, and obtained the samples with different denaturant concentrations by incremental replacements of the protein sample under lower denaturant with that under the 8 M urea (or 6 M GuHCl) (Fig. 30, Tables 3 & 4). One of the drawbacks in this method could be that since the protein concentration of the later data point depends on the former data point, one inaccuracy in the sample mixing (e.g., aspirating and adding a given volume for the sample replacement) might introduce a systematic error into the expected denaturant concentrations of all the data points afterward. Except repeating the experiment for the average, the ways to examine the extent of this possible concentration discrepancy could be taking the denaturation curve from the data point with the highest denaturant to the lowest, like taking a renaturation curve, to see if the CD signals were identical, or at least close, to those recorded in my study. Since the ensemble of a protein’s secondary structures depends only on its final condition, the CD of the protein sample under a given denaturant concentration at equilibrium should be constant regardless of from
“denaturation” or “renaturation.”
50
V.3. Exploring the “Open-Closed” Hypothesis of LC3 with the Technique Other Than CD
In the previous studies an “open-closed” model have been proposed for the functioning of the LC3, primarily based on the structural homology with its paralog GABARAP,5,25,26,27 and the mutational analysis on the ULD’s role in the membrane dynamics.5 However, this possible conformational change of the LC3 has not yet been directly observed, such as by X-ray crystallography. In the study, the more susceptible CD222 from the phosphomimetic mutations to the denaturations might be correlated to the destabilizations of the α-helices in the NHD, whereas the pure ULD behaves relatively stable (Figs 31 & 32). These observations draws out a hypothesized model of the denaturation, which assumes that a flexible NHD associates with a more rigid ULD (Fig. 36).
To further test this idea, other approaches are already available to furnish the model with more practical information. One is to establish the fluorescent-based denaturation curves of the LC3 using the fluorescent mutants constructed in the study (e.g., D106W, F108W, or Y110W). Since there is no tryptophan in the LC3 wild type, the fluorescent spectrophotometry might be able to monitor the interface between the NHD and the ULD. Using the fluorescent mutations of the residues previously reported to be buried in the interface and essential for the LC3 functions (e.g., D106, F108, and Y110),5 (if
51
these mutations per se would not significantly affect the protein’s native structure, which could be examined by CD) the fluorescent denaturation curves of these potential reporters might probe when the NHD dissociates from the ULD.33
In the study, the early decrease of the CD222 in the denaturation curves are attributed, at least partially, to the denaturation of the NHD. If the fluorescent denaturation curves described above could tangibly show the midpoints in the fluorescent intensity shifts close to the denaturant concentrations where the early denaturations began in the CD denaturation curves, the model and the “open-closed”
idea behind would further be supported.
52
V.4. The PKA as the Candidate to Directly Phosphorylate Human LC3
The phosphorylation of LC3 by PKA is observed in mouse,28 and in this study the human PKA was used to perform the phosphorylation assay on the human LC3. The phosphorylation of the Kemptide by the PKA (Fig. 34) suggested that our PKA can be active in the buffer condition (p. 21); although having the PKA’s consensus recognition sequence (i.e., Arg-Arg-Thr-Phe) on the NHD,28 the kinase assay did not show a positive result on the LC3 (Fig. 35). The possible reasons could have several including:
there needs other unknown regulatory factor essential for coordinating the interaction between the LC3 and the PKA; other buffer condition where the LC3 could assume more accessible conformations; or an alternative PKA paralog only which does so, since multiple PKA catalytic subunit paralogs exist in human, such as catalytic subunits α (including isoform 1 & 2; in the study, the isoform 1 was used), β, and γ.
53
V.5. Possible Biological Effects of the Phosphorylations Using Other Biophysical Methods
Susceptibility of the LC3 multimerization after the possible phosphorylations could further be tested biophysically. One way is to utilize analytical ultracentrifuge, which could examine the binding equilibrium of the LC3s at different phophorylation states (or the different phosphomimetic mutants), since multimerization makes particles with greater mass and size.
To see if phosphorylation could change LC3’s affinity with other key players in autophagy (e.g., the protein p6234), the technique isothermal titration calorimetry (ITC), which could give the binding constant between molecules, would be helpful to address this question using the phosphoproteins or the phosphomimetic mutants of LC3.
54