1. Introduction

Trypanosoma brucei gambiense (T. b. gambiense) and Trypanosoma brucei rhodesiense (T. b. rhodesiense) are generally referred to as African human trypanosomes. They are blood parasites that cause the disease called Human African Trypanosomiasis (HAT), also known as sleeping sickness. HAT is a potentially fatal but neglected disease that is endemic in sub-Saharan Africa. In this region, about 70 million people are at risk; 300,000 people are presently infected, and about 30,000 humans are annually infected. This disease is racially unbiased, as over 200 cases of imported HAT have been reported in various European countries. The parasites are transmitted by insects called tsetse flies, during their blood meal. The insect form is known as procyclic forms (PCFs), while the animal form is called blood stream forms (BSFs).

Upon infecting the human host, African trypanosomes live extracellularly in the blood stream, and are able to escape the host defense mechanisms by a strategy that is known as antigenic variation. Using this mechanism, they easily and repeatedly change their surface coat, hence loose recognition by the host antibody. This evasion of the adaptive immune responses contributes to parasite virulence, and has frustrated all efforts towards vaccine development; hence, management of the disease by chemotherapy remains the only reasonable hope towards a solution to this devastating disease. Unfortunately, only four drugs (suramin, pentanidine, melarsoprol, and eflornithine) are available for treatment of HAT. All of them are no longer safe and effective because problems such as narrow spectrum, treatment failures due to resistance, high cost, and toxicities. Therefore, there is urgent need to discover and design new, safer, and affordable more effective drugs with a broader action spectrum.

. To design a good drug, it is important to identify and validate molecular targets in the parasites that are essential for their growth but absent, not essential, or structurally different from the host molecule. In other words, taking advantage of differences between the parasite and the humans will lead to the discovery of a good drug that is not toxic to the host. Looking closely at their energy generation mechanisms reveal some differences:

Our laboratory found ascofuranone to be so far, the most excellent inhibitor of AOX; however, its full potential at curing animals that were experimentally infected with trypanosomes requires co-administration with 5mM glycerol. This amount of glycerol is physiologically too high and toxic to the animals. Our group has identified and validated GK to be the target of glycerol. Although GK, in conjunction with AOX, is thus a promising target for chemotherapy, an effective and selective trypanosome GK inhibitor has yet to become available. In fact, there is no known inhibitor of any GK.

It has long been recognized that knowledge of the three-dimensional (3-D) structures of protein targets using X-ray crystallography, has the potential to accelerate the discovery and design of new drugs. It is in view of all the above that I design this work to utilize structure-based (in silico screening) and activity-based (high-throughput screening) approaches for the design of good compounds that will discriminatorily inhibit parasite GK.

2. Result and Discussion

GK gene (gk) from both T. b. gambiense and T. b. rhodesiense were cloned into pET151/D-TOPO vector and sequenced. The nucleotide sequence, which our group has deposited to the GenBank (accession Nos. AB517984 and AB517985, respectively) revealed that GKs of T. b. gambiense and T. b. rhodesiense are same at amino acid level. The cloned gk of T. b. gambiense was transformed into, JM109 (DE3 + pRARE2) E. coli strain for protein expression. The expressed His6-tagged protein was purified to homogeneity by a combination of affinity chromatography on Ni-NTA, and gel filtration on superdex 200. Approximately 150mg of active (31.7 μmol/min/mg) pure protein (single band on SDS-PAGE) was obtained from a 10 L culture of transformants. The enzyme was characterized, and used for crystallization experiments.

A total of 576 crystallization conditions from commercially available screening kits were tested by sitting drop vapour diffusion with protein sample of 5mg/ml concentration. Tiny crystals were obtained by a conditions that contain 2.5-5% PEG 6000, at 293 K. This condition was optimized, and single crystals suitable for X-ray diffraction experiments were obtained with 30% (w/v) PEG 400 in HEPES buffer, pH 7.0. X-ray diffraction data were collected at λ= 1.000A under cryocooled conditions (100 K) on BL41XU beam line at Spring-8 (Harima, Japan) and BL17A beam line at Photon Factory (Tsukuba, Japan). The cryoprotectant used was 40% (w/v) PEG 400. Data were collected for ligand-free and ligand-bound GK. The ligands used were glycerol, glycerol 3-phosphate, ATP, ADP, and 4-nitrophenylphosphate; whose crystals diffracted X-ray to resolutions of 2.4, 2.8, 2.3, 2.7, and 2.7 A, respectively. The ligand-free crystals diffracted to a resolution of 2.9 A. Diffraction data were processed and scaled with HKL-2000 software package. Analyses of the symmetry and systematic absences in the recorded diffraction patterns revealed that while the crystals of glycerol, ATP, and ADP bound TbgGK belong to the monoclinic P21 space group, those of ligand-free, and g3p- bound TbgGK belongs to the orthorhombic space group P212121.

Structure of glycerol-bound TbgGK was solved by molecular replacement method with the MOLREP program from the CCP4 suite using the refined coordinates of GK from P. falciparum (PDB code: 2w41, 40% amino acid sequence identity). A promising solution with a homodimer structure was obtained. Using the solution of the molecular replacement, the structure of the various ligand-bound GK has been solved; adjustments and refinements of the various structures are done by repeated cycles of COOT and REFMAC programs.

The solved structure of TbgGK revealed that the enzyme is a homodimer, which is formed by a somewhat strong association of 2 monomer chains A and B where the dimer interface is made up of an antiparallel beta sheet and 3 alpha helices that are contributed by each of the monomers (Fig 1). Each monomer is made up of 2 domains (I and II), glycerol binding site is in domain I.

When the various forms of TbgGK were superposed, a difference is active site conformation was observed between the ligand-free and glycerol bound forms. In the ligand-free form a loop closes up the active site, the loop conformation was maintained by a disulphide linkage between cysteine residues at positions 278 and 319 (C278 and C319). Since the disulphide bond was absent in the glycerol-bound form, it was probably reduced during glycerol binding. To investigate the relevance of this disulphide bond and its reduction on the biological activity of TbgGK, I performed activity measurements in the presence and absence of 5mM reducing agents (Dithionite or Tricarboxyethylphosphine, TCEP). The enzyme activity was 2 and 7 folds higher respectively, in the presence of reducing agents. This result was confirmed by site-directed mutagenesis, during which C278, C319, or both were mutated to alanine or serine. Activity of serine mutants were also about 7-8 folds higher than the wild type enzyme, and they were unresponsive to the presence of reducing agents. The alanine mutant activity was not significantly higher than wild type enzyme, suggesting a strict positional requirement for polar residues.

Notable distinctive feature of the TbgGk is the presence of 2 loops at the active site, where they seem to affect the active site conformation. And also the presence of an unusual ADP / ATP binding site that is about 30 A away from the active site groove. Interestingly, the unique loops and ADP / ATP binding sites are formed by amino acid residues that are not found at the corresponding segment of human GK therefore, the design of compounds that bind this loop may be potent TbgGK-specific inhibitors. The residues that bind ADP in the active site are also different from that of other GKs (Fig. 2), this unique binding result in orienting the ADP proximally (3.4 A) to phosphoryl group of g3p. This could be the reason for the reverse-ability of trypanosome GK. In P. falciparum where GK lacks reverse capability, the second phosphoric acid group of ADP is over 8 A away from phosphate of g3p, hence the lack of proximity necessary for catalysis.

Also herein, I developed a new high throughput screening assay method for trypanosome GK by using 4-nitrophenylphosphate; its phosphate group is cleaved by TbgGK, and probably transferred to ADP. This assay method has been optimized for the screening of inhibitors that were predicted by in silico screening with the solved structures. The experimentally screened compounds were from The University of Tokyo's faculty of pharmaceutical sciences chemical compound library established by Professor Nagano. So far, about 50 hits have been found out of 2000 compounds. The hits were compounds that inhibited 50% of TbgGK activity at 20μM. Representatives of the hits were compounds with UT ID: T-103825, T-103826, and T-103843; which showed 70, 60, and 92% inhibitory effects respectively. They contain diazinane ring. The best poses from the docking of these compounds revealed that they are bound to the active site cleft, at or close to the ADP binding pocket (Fig. 3). Since the ADP binding mode in TbgGK is different from the human hosts, optimization of these compounds by structure-activity relationships will help us to obtain drug-like specific inhibitors of TbgGK.

Figure1: Structure of glycerol-bound TbgGK coloured by chains. Arrows show glycerol in the active site of the enzyme. N and C represent the terminals of each monomer.

Figure 2: Active site ADP binding amino acid residues in TbgGK are different from the P. falciparum GK residues. (B) TbgGK amino acid residues that bind ADP, they are different from those for ADP binding in P. faciparum GK (C)

Figure 3: Best poses for the docked structures of some TbgGK inhibitors showing the proposed binding mode for (A) T-103825, (B) T-103826, and (C) T-103843. The compounds are bound to the ADP binding site. Diazinane-trione ring (marked "D3OR") is common to them and seems to mediate their binding.

本研究はアフリカトリパノソーマ症の薬剤標的タンパク質であるトリパノソーマのグリセロールキナーゼ(GK)と様々な化合物との複合体の結晶構造をX線結晶構造解析により明らかにして下記の結果を得ている。

1.トリパノソーマのGKのapo型及びグリセロール, グリセロール-3-リン酸(g3p), ATP, ADP, 4-nitrophenylphosphateとの複合体の結晶構造をそれぞれ2.9, 2.4, 2.8, 2.3, 2.7, 2.7Å分解能で決定した。これは、真核生物で生存に必須なGKの初めての構造である(真核生物の構造解析の最初の例は、マラリア原虫のGKであるが、これは生存に必須な酵素ではない)。

2.トリパノソーマのGKは、結晶中でサブユニット間の逆平行のβ-シートで強固なダイマーを形成していた。この結果は、ゲルろ過クロマトグラフィーでの溶出位置から計算される分子量がダイマーという結果に一致している。

3.apo型と他の化合物との構造を比較すると、apo型のみでC278とC319間でジスルフィド結合が形成されている。このジスルフィド結合が活性に関与しているかを還元剤存在下とシステインをセリンまたはアラニンに置換した変異体を作成して酵素活性を測定すると、還元剤存在下及びセリンに置換した変異体で活性が約8倍になることを明らかにした。

4.グリセロール及びg3pは、N末ドメインに結合していることが明らかになった。グリセロール及びg3pの共通の認識には、Glu84, Glu85, Trp104, Asp254, Phe279が関与していることを明らかにした。

また、g3pのリン酸基の認識には、Thr11, Asp254, Gln255, Thr276が関与していた。

5.トリパノソーマのGKには、2つのADP結合部位があることが分かった。1つ目は、N末ドメインとC末ドメインの間の活性部位に結合していた。このADPの認識には、4つのスレオニン残基が重要な役割を果たしていることが明らかになった。2つ目は、活性部位から約30Å離れた分子の表面に2つのADPがマグネシウムイオンを介して結合していた。このADPの認識には、分子表面のアルギニン残基が重要な役割を果たしていた。また、ATPは、活性部位には結合しておらず、2つ目の活性部位から離れた分子表面のみに結合していた。

6.唯一構造が明らかになっている真核生物のマラリア原虫のGKとトリパノソーマのGKを比較すると、グリセロールの認識については、ほとんど同じアミノ酸残基が関与していたが、活性部位のADPの認識についてはトリパノソーマでは、多くのスレオニン残基が関与していたが、これらのアミノ酸残基は、マラリア原虫のものには保存されていないことが明らかになった。

また、トリパノソーマのGKでADPとg3pの構造を重ね合わせると、リン酸基間の距離は3.4Åであるが、マラリア原虫では8.2Åと離れており、このことがトリパノソーマのみで逆反応(ADP+g3p→ATP+グリセロール)が進行する理由であることを示すことができた(逆反応で結合する2つのリン酸基の間の距離が近い)。

7.4-nitrophenylphosphateを用いたトリパノソーマのGKの活性のアッセイ法を開発し、それを用いて、インシリコスクリーニングで決定した約2500の化合物のスクリーニングを行なったところ、約50種類の阻害剤の候補を決定することができた。

以上、本論文は、トリパノソーマのグリセロールキナーゼの立体構造を様々な化合物との複合体で解析し、また重要であると思われるアミノ酸の変異体を作成するなど、構造と機能相関に関する重要な知見を得ることができた。また、薬剤のスクリーニングを行う系を開発し、実際に有用であると思われる薬剤を見いだしたことはアフリカトリパノソーマ症の薬剤開発に十分貢献するものと考えられ、学位の授与に値するものと考えられる。