Structural And Functional Insights Into The Key Proteins In Organic Nitrogen Degradation And Light Harvesting From Marine Bacteria

Substance-recycling and energy-flowing are two basic processes in marine ecosystem. In marine substance-recycling, nitrogen-recycling is very important to marine ecosystem. Photosynthesis provides energy for marine ecosystem, and algae fix the most of solar energy. Generally, microorganisms play a significant role in marine substance-recycling and energy-flowing. It is important to understand the underlying mechanisms for how microorganisms, especially the key proteins, take part in marine nitrogen-recycling and solar energy fixation, which would give new insights into the mechanism of how marine ecosystem works. In this thesis, we studied two proteases from a deep-sea bacterium which may participate in marine nitrogen-recycling, and also some key proteins from cyanobacteria responsible for solar energy fixation. Our study provides an important theoretical base for analyzing how microorganisms take part in marine substance-recycling and energy-flowing.1. The autoprocessing mechanism for Thermolysin-like proteasesThermolysin family is a very large and important family in metalloproteases, most of which are secreted by kinds of microorganisms. Thermolysin family proteases play an important role in the earth organic nitrogen degradation and nitrogen-recycling. A Thermolysin-like protease (TLP) is firstly synthesized as a zymogen with a long propeptide, and then undergoes an autoprocessing process for maturation. Due to the lack of structural analysis for maturation intermediates, the underlying mechanism for how TLPs mature by autoprocessing is still unclear.Many marine bacteria can secrete TLPs to degrade extracellar organic nitrogen, and these TLPs facilitate the recycling of marine nitrogen. Previous study shows that the deep-sea cold-adapted bacterium Pseudoalteromonas sp. SM9913secretes MCP-02, a typical TLP. In this section, with MCP-02as a model enzyme, we studied the underlying mechanism for how TLPs mature.We first determined the crystal structure of mature MCP-02to1.7A. The overall structure of MCP-02is very similar to that of thermolysin, the protype of Thermolysin family. The entrance to the catalytic zinc in MCP-02is a deep and narrow cleft between the N terminal domain (NTD) and the C terminal domain (CTD).Several site-directed mutations were made to stabilize MCP-02maturation intermediate states. Finally, the crystal structures for the E346A complex and the E369A complex were successfully determined with an RMSD of only0.190A. The propeptide in the MCP-02autoprocessed complex forms a large "C" shape surrounding the CTD of the catalytic domain. The interface between the propeptide and the catalytic domain is rather large, with a buried surface area up to1541.98A2. Most of the forces connecting the propeptide with the catalytic domain result from the complicated network of hydrogen bonds and salt bridges between the (3-sheets of the FTP domain and the CTD. The C-terminus of the propeptide inserts into the catalytic cleft, and the carboxyl group of the last residue (His204) replaces the activated water molecule in the mature enzyme, acting as a monodentate ligand to zinc. The loss of this water molecule, the nucleophile during peptide hydrolysis, results in the inhibition of protease activity of the catalytic domain.In the autoprocessed complex, Ala205shifts upwards by33A from the previously covalently-linked residue, His204. Residues between Ala205to Ser212form a new β-strand and insert into the catalytic domain. To further clarify these conformational changes, we simulated the structure of the unautoprocessed zymogen, whose free energy is much higher than that of the autoprocessed complex. The MCP-02unautoprocessed zymogen had an apparent Tm of50℃, while the MCP-02autoprocessed complex had a higher apparent Tm of70℃. The rather low denaturation temperature of the unautoprocessed zymogen indicates that it is kinetically trapped in a high-energy state, and that conformational changes in MCP-02from the unautoprocessed zymogen to the autoprocessed complex may arise from free-energy transformation.By N-terminal sequencing of the propeptide fragments derived from its degradation by the catalytic domain, the two cutting sites were determined, which are located at Ser49-Va150and Gly57-Leu58, respectively. Structural analysis and biochemical assays suggested that these cuts result in the structural disassembly of the propeptide and the release of the propeptide fragments from the catalytic domain, which would finally activate the protease MCP-02. Most of TLPs have very similar sequences and structures like MCP-02, so based on the results above, we suggested a new model for the mechanism of TLPs maturation.2. Molecular insights into collagen recognition and degradation by the catalytic domain of the S8serine collagenase MCP-01Collagen is one of the most important organic nitrogen in marine ecosystem, and microorganism-drived collagen degradation is a key step for marine nitrogen-recycling. Previous study indicates that MCP-01secreted by Pseudoalteromonas sp. SM9913is a serine collagenase of the S8family. Although several S8serine collagenases have been reported, the underlying mechanism for how these proteases recognize and degrade collagen is still unknown.In order to clarify the underlying mechanism of collagen recognition by the catalytic domain (CD) of MCP-01, we firstly determined the crystal structure of the MCP-01CD. Through a detailed structural analysis, sequence alignment and biochemical assay, it was found that three loops surrounding the substrate-binding pocket of the MCP-01CD play an important role in collagen recognition, and that there are lots of acidic and aromatic residues on these three loops. Mutation experiments showed that the acidic and aromatic residues on the three loops play an important role in collagen binding. Moreover, sequence alignment analysis indicated that it is a common feature for the increasing of acidic and aromatic residues in these three loops in the S8family collagenases. To reveal the degradation mechanism of MCP-01CD on collagen, we studied the specificity of MCP-01CD to synthetic peptides and the cleavage sites of MCP-01CD on collagen. The result showed that MCP-01CD did not have strick substrate specificity, which might facilitate collagen degradation by MCP-01CD. In summary, these results give structural and mechanical insights into collagen recognition and degradation of the S8collagenases for the first time.3. Mechanistic insights into the assembly of the phycobilisome rod in cyanobacteriaPhycobilisomes are the light-harvesting complexes for both marine and fresh water cyanobacteria. Phycobilisomes are composed of phycobiliproteins (PBPs) and linker proteins. Linker proteins play an important role in the phycobilisome assembly and energy-transformation modification. Generally, all kinds of linker proteins are composed of three different domains:Pfam01383domain with known structure, Pfam00427domain and Pfam00502domain, both of which are lack of structural and biochemical studies.Here we recombinantly expressed Pfam00427from the model cyanobacterium Synechocystis sp. PCC6803, and determined its crystal structure at1.9A by using a newly developed method with3-iodo-L-tyrosine. Basically, the pfam00427domain consists of six a-helices (al to a6). The six a-helices of pfam00427are arranged into a point-symmetric architecture. An exhaustive search of the Protein Data Bank using a DALI server failed to identify any entry that is significantly homologous to the intact six-a-helix bundle structure of the pfam00427domain. This finding indicates that the structure of pfam00427probably represents an uncharacterized fold in all-alpha proteins.To investigate how the pfam00427domain binds into the inner face of the C-PC (αβ)6hexamer cavity, we simulated the pfam00427-C-PC complex by using the program AutoDock. Simultaneously, a conserved "C-PC-binding patch" on the Pfam00427was identified by a GST pull-down assay. Based on these results, we further chose a candidate model to run a molecular dynamics simulation for structural optimization, and obtained the final resulting model.Based on our pfam00427-C-PC complex model and the previous APC-LC7.8crystal structure, we proposed a precise model for the PBS rod. First, we generated a structural model for LRT (pfam01383) by homology modeling using LC7.8as a template. We therefore matched LRT into the central hole of the opposite C-PC (αβ)3according to the APC-LC7.8crystal structure. Then, this packing mode is applied to the whole PBS rod, as the domain organization in the different linkers is repetitive moving from distal to proximal to the core. In our assembly model, the LR spans and connects two adjacent C-PC (αβ)6with its linking region (～60residues) being located between the two C-PC (αβ)6and accessible to the solvent. The PBS rod is terminated by an LRT at the distal end and is joined to the PBS core through an LRC at the proximal end. Our simulated model for the molecular structure of phycobilisome rod reveals the mechanism for how phycobilisome rod assembles, which also modified the phycobilisome rod model previously proposed. 4. Mechanism of the auto-chromophorylation of LP502and the molecular model of phycobilisome terminal energy acceptorLcm is the most important linker protein in the phycobilisome core section, belonging to the trimeric (αβ)(LCMβ18) terminal energy acceptor (TEA). The N-ternimal domain of LCM, Pfam00502(LP502), is the only bililyzed domain through auto-chromophorylation in all linker proteins. However, little is known regarding the underlying mechanisms of the auto-chromophorylation of LP502and the detailed structure of TEA.All of known PBPs possess a conserved phycobilin-attached cysteine, and this cysteine is replaced by Ser152in LP502. In contrast, LP502contains a unique chromophore attachment site Cys190. Biochemical assay showed that the chromophorylation site on LP502can not be shifted onto Cys152. We also experimentally detected the interaction between PCB and apo-LP502by measuring the absorbance at667nm. Interestingly, we found that the S152A and C190S mutants bound non-covalently to PCB to give an absorbance at667nm. These results reveal that LP502can recognize the PCB molecule by itself, which makes a reasonable explanation for why LP502can auto-chromophorylate.In order to find out the PCB binding pocket on LP502, we created several mutations on LP502, and our mutational results suggested that the PCB binding pocket on LP502is the same as that on the APC a-subunit, and that Arg154and Asp155play an important role in the PCB binding.To further substantiate the mechanism for PCB binding to apo-LP502, we generated a rational PCB-LP502complex model by computational simulation. Notably, the PCB binding pocket of LP502is much deeper and larger than the binding pockets of the APC α-and β-subunits. In addition, the PCB conformation in the distinct binding pocket of LP502is distorted compared to that in the pocket of the APC a-subunit. The PCB Ring A is rotated by70°to be covalently linked to Cys190, and in particular, the Ring D is rotated by almost90°. Detailed structure analysis of the PCB-LP502complex model reveals that Ser152forms a hydrogen bond with the Ring A of PCB in the PCB-LP502structure. This hydrogen bond forces the methylene carbon of Ring A to be orientated towards Cys190and further facilitates the thioether bond formation. Interestingly, when Ser152was mutated to cysteine, LP502could still auto-catalytically add PCB; however, the mutated bilin-LP502complex conferred an apparent blue shift from667nm to646nm in the absorption spectrum. These results reveal that the special conformation of the PCB Ring A is another key element contributing to the formation of the667nm absorption maximum of LP502.X-ray crystallography studies with APC trimers show that the two PCB molecules at the interface between two APC (αβ) monomers form strongly coupled pairs or dipoles. All of the three dipoles are different in the LCM mediated (αβ)2(LCMβ18) TEA trimer. Dipole1in the TEA is composed of the β18-subunit and a-subunit₂. According to sequence alignment of β-and β18-subunits, interactions between β18-subunit and the adjacent PCB must be lost in the Dipole1as a result of two amino acid substitutions (Tyr to Phe and Thr to Ile). Dipole2in the TEA is the same as that in the APC (αβ)3trimer. The most extraordinary dipole in the TEA, Dipole3, is composed of LP502and β-subunit₁. Our analyses shows that LP502contains a distorted phycobilin PCB in its deep binding pocket and bears a unique absorbance peak at667nm, which is longer than the absorbance peak (650nm) of the APC trimer. The PCB molecule in LP502adopts a distorted conformation and Ring A and Ring D rotate more largely compared to the PCB in APC α-/β-subunits. As the absorption maximum of LP502and TEA is quite similar (～670nm), it is plausible to conclude that the distorted PCB molecule in LP502play a major role in the formation of the absorption spectrum of TEA. Based on these analyses described above, a detailed molecular structure for the TEA in cyanobacterial phycobilisome was proposed, in which the three dipoles are coordinated.