Introduction:

Endoplasmic reticulum (ER) is the organelle where membrane-bound and secretory proteins are synthesized. To ensure the proper folding and assembly of newly synthesized proteins, there exists a rigorous quality-control system consisting of a variety of folding enzymes and molecular chaperons. Calnexin (CNX) and calreticulin (CRT) are two homologous chaperones having sugar-binding ability. They comprise the so-called CNX/CRT cycle (Fig. 1), together with glucosidase I (GI) and glucosidase II (GII), both of which remove glucose from N-glycans, and UDP-glucose:glycoprotein glucosyltransferase (GT), which adds back glucose. GII mediated deglucosylation and GT-catalyzed reglucosylation cycle continues until proper folding of ptotein is achieved.

GII is a heterodimeric complex consisting of a catalytic α subunit (GIIα) and a β subunit (GIIβ). It controls the import and export of newly synthesized glycoproteins within CNX/CRT cycle by trimming two α1,3-linked glucose residues from N-glycans (Fig. 1). Though past studies have made great progress in understanding this glucose trimming process, there are still some issues that need to be addressed. First, GIIβ contains a domain (MRH) with homology to mannose 6-phosphate receptors that specifically bind to phosphorylated mannose residues on acid hydrolases. This raises a possibility that GIIβ may possess a sugar-binding activity. However, whether GIIβ possesses sugar-binding activity and, if so, what role this activity plays in the function of GII has not been demonstrated. Second, recently, an ER resident protein, malectin, is found to selectively recognize Glc2-N-linked glycans in Xenopus laevis. The capacity of malectin to bind to G2M9, one of the substrates of GII, suggests malectin may have some effects on the trimming of G2M9. To provide further insights into the glucose-trimming process by GII, my present study focuses on the investigation of the sugar-binding ability of human GIIβ and malectin, and the functions mediated by their sugar-binding activity.

Results

1.GIIβ-MRH binds to cell surface glycans

MRH domain of GIIβ (GIIβ-MRH) with a C-terminal biotinylation sequence was expressed in E.coli BL21(DE3)pLysS, refolded and purified. After biotinylation with biotin ligase BirA, purified GIIβ-MRH was incubated with R-phycoerythrin-labeled streptavidin (PE-SA) to form PE-labeled GIIβ-MRH tetramer, which was then used to investigate the capacity of GIIβ-MRH to bind sugars on the cell surface. GIIβ-MRH did not bind to the HeLaS3 cells, but treatment of the cells with either of the two α-mannosidase I inhibitors, kifunensine (KIF) and deoxymannojirimycin (DMJ), but not the Golgi α-mannosidase II inhibitor, swainsonine (SW), caused the binding of GIIβ-MRH (Fig. 2A). High-mannose type glycans accumulate on the cell surface in KIF- or DMJ-treated cells, which can be cleaved by endo-β-N-acetylglucosaminidase H (endo H). I treated the DMJ-treated HeLaS3 cells with endo H and examined the binding of GIIβ-MRH (Fig. 2B). Endo H treatment of the cells abolished the binding of GIIβ-MRH to the cells almost completely, confirming the binding of GIIβ-MRH to DMJ-treated HeLaS3 cells is through the cell surface high mannose-type glycans. Given that KIF or DMJ specifically inhibit the cleavage of α1,2-linked mannose residues from high mannose-type glycans (Fig. 3A), these results suggest that the α1,2-linked mannose may be important for recognition by GIIβ-MRH.

2. The terminal α1,2 linked mannose, especially that on the C-arm, of high mannose-type glycans, is required for the binding of GIIβ-MRH

The sugar-binding specificity of GIIβ-MRH was investigated by frontal affinity chromatography (FAC). As shown in Fig. 3B, M9 and M8A, both of which contain three terminal α1,2-linked mannose, exhibited the highest affinity to GIIβ-MRH. G1M9, M8B, M8C, M7C, and M7B, each of which possesses two terminal α1,2-linked mannose residues, showed lower affinities, and the affinities of the oligosaccharides with a single terminalα1,2-linked mannose were further impaired. These results suggest that the terminalα1,2-linked mannose of N-glycan is the major determinant for the binding of GIIβ-MRH. Interestingly, although each of G1M9, M8B, M8C, M7C, and M7B possesses two terminalα1,2-linked mannose residues, the affinity of M8C and M7B is quite lower than those of the other three. Comparing their structures indicates that M8C and M7B are distinguished from G1M9, M8B, M7C by the absence of the α1,2-linked mannose on the C-arm. These data suggest that theα1,2-linked mannose on the C-arm is essential for the strong binding of GIIβ-MRH.

3. The sugar-binding activity of GIIβ is important for GII to trim glucose efficiently from N-glycans.

Glucosylated high mannose-type oligosaccharides (G1M9, G2M9) are known substrates of GII. The capacity of GIIβ-MRH to bind high mannose-type glycans suggests that GIIβ is possibly involved in glucose-trimming process. Two GIIβ mutants myc-GIIβ(Y410A) and myc-GIIβ(Q420E) that do not have the ability to bind sugars were expressed in 293T cells. We found that these point mutations have no effects on the ability of GIIβ to make complex with GIIα and that endogenous GIIβ can be substituted with over-expressed myc-tagged GIIβ mutants. To investigate possible dominant negative effects of the GIIβ mutations on enzymatic activity of GII, the myc-GIIβ or its mutants were over-expressed in 293T cells and the glucosidase activity in the cell lysates was measured using p-nitrophenyl-α-glucopyranoside (pNP-αGlc) or methotrexate (MTX)-derivatized G1M9 and G2M9 as substrates. As shown in Fig. 4A, similar amounts of the wild type GIIb and the GIIb mutants were expressed. Over-expression of either the wild type GIIb or the GIIb mutants slightly increased the hydrolysis of pNP-aGlc to the similar extent (Fig. 4B). However, when MTX-derivatized G1M9 (Fig. 4C) or G2M9 (Fig. 4D) were used as substrates, expression of the wild type GIIb and that of the GIIb mutants showed opposite effects on the glucosidase activity. Over-expression of the wild type GIIb slightly increased the removal of glucose from G1M9 and G2M9, while over-expression of the either of the two GIIb mutants significantly decreased the removal of glucose from G1M9 and G2M9 compared to mock transfected cells. These data suggest that the sugar-binding activity of GIIβ-MRH is not required for the hydrolysis of pNP-αGlc, but important for hydrolysis of glucosylated high mannose-type glycans.

4. Human malectin selectively binds to G2M9

The sugar-binding activity of human malectin was investigated using malectin tetramer prepared as similar to that of GIIβ-MRH. Human malectin did not bind to the cells treated with castanospermine (CST), KIF, or SW, but selectively bind to deoxynojirimycin (DNJ)-treated cells (Fig. 5A). Given that DNJ preferentially inhibits the activity of GII and causes the accumulation of Glc2-N-glycans, these results suggest human malectin binds to Glc2-N-linked glycan. The detailed sugar-binding specificity of human malectin was analyzed by FAC using a series of high mannose-type glycans resident in the ER. Among these oligosaccharides, G2M9 was found to be the only oligosaccharide recognized by malectin (Fig. 5B). Given that M9, G1M9 and G3M9 could not bind to malectin, these results indicate that terminal Glcα1,3Glc is the major determinant for the binding of malectin.

5. The expression of malectin was significantly induced during tunicamycin-induced ER stress

The remarkable selectivity of malectin to Glc2-N-linked glycan points to a role for malectin in the early glucose trimming process (Fig. 1). By analogy with CNX/CRT, malectin may function as a lectin chaperon that specifically recognizes G2M9 on folding glycoproteins, slowing the trimming of G2M9 (Fig. 1, step 2) to increase protein folding efficiency. To address this possibility, I first analyzed the effects of malectin on the trimming of G2M9. Unexpectedly, the trimming of G2M9 in vitro was not influenced by malectin. Though the effect of malectin on the glucose trimming process is still not clear, I found that the expression of malectin was significantly induced by tunicaymcin-induced ER stress (Fig. 6). Tunicamycin is the inhibitor of GlcNAc-phosphotransferase. It blocks the synthesis of all N-linked glycoproteins and is well-known inducer of ER stress. These results suggest that malectin is likely to involve in the folding of glycoproteins, though its detailed function remains unknown.

Conclusions

To provide further insights into the glucose trimming process in the ER, I have investigated the sugar-binding activity and function of GIIβ and malectin. My present study demonstrated the sugar-binding activity of GIIα, and the importance of the activity for efficient glucose trimming. In addition to previous known roles of GIIβ, ER localization and assisting folding of GIIβ, our study revealed that GIIβ also participates in the glucose trimming process. My study also showed that human malectin selectively binds to G2M9. Though the effects of malectin on the trimming of G2M9 are not clear, I found the induced expression of malectin during tunicamycin-induce ER stress. These studies suggest that malectin may functions as a lectin chaperon in the ER to promote the folding of glycoproteins

Fig. 1 Scheme of the CNX/CRT cycle. There are 5 main steps in the CNX/CRT cycle. (1) trim-ming of the outermost glucose by GI, (2) trimming of the middle glucose by GII allowing the entry of glycoproteins to CNX/CRT cycle, (3) association of glycoproteins with CNX/CRT, (4) trimming of the innermost glucose by GII leads to release of glycoproteins from CNX/CRT, (5) reglucosylation of incompletely folded glycoproteins by GT allows entry to another cycle.

Fig. 2. GIIβ-MRH binds to cell suface glycans (A) the binding of GIIβ-MRH tetramer (filled histogram) or PE-SA as a control (thin line) to HeLaS3 cells that were treated with DMJ, KIF or SW. (B) the binding of Gllβ-MRH tetramer (filled histogram) or PE-SA (thin line) to DMJ-treated HeLaS3 cells that were pretreated with endo H.

Fig. 3 Sugar-binding specificity of GIIβ-MRH analyzed by FAC (A) Structure of the N-linked glycan precursor, G3M9. (B)The affinity of each PA-labeled oligosaccharide to GIIβ-MRH was de-termined by the FAC analysis.

Fig. 4 Enzymatic activity assay of GII.(A) Expression of Gila and GIIβ by Western blotting. (B) The activity of Gil in the cell lysates prepared in (B) using pNP-aGlc as substrate. (C) the activity of Gil in the cell lysates prepared in (B) using G1M9-MTX (D) and G2M9-MTX (E) as subtrate

Fig. 5 Sugar-binding specificity of Malectin (A) Binding of Malectin tetramer to cells treated with CST, DNJ, KIF, or SW (B) The affinity of each PA-labeled oligosaccharide to Malectin was determined by the FAC analysis.

Fig. 6. Increased expression of malectin during tunicamycin-induced ER stress. HEK293 cells were treated with 5 μg/ml tun-ciamycin for 12 and 24 hours. The expressin of malectin was an-lyzed by real time PCR

本論文は大きく2つから構成されており、前半はマンノース6リン酸レセプターホモロジードメイン(MRHドメイン)をもつglucosidase IIβサブユニット、後半はマレクチンについて、それぞれレクチンとしての糖結合活性の有無の検討、糖結合特異性の決定、それぞれの機能に及ぼすレクチン活性の意義について述べられている。さらに、これら解析の結果を踏まえて、細胞内の小胞体における糖タンパク質品質管理機構における、これら2つの分子の意義、および協調的な機能について議論されている。

本論文第1章では、glucosidase IIβサブユニットのMRHドメインを可溶型四量体として大腸菌で発現させ、細胞表面糖鎖に対する結合を指標に糖結合活性の検討を行った。その結果、高分子量の高マンノース型糖鎖に強い親和性を示すことを明らかにした。さらに、フロンタルアフィニティクロマトグラフィーによる解析によって、非還元末端にα1,2Man残基をもつ側鎖に特異性を持っていること、また、これら側鎖の数が多いほど強い結合を示し、特にC-armという側鎖に最も強い親和性を示すことを明らかにした。このことは、C-armのα1,2Manがglucosidase IIの作用を受ける鍵となる残基であり、これが除去されるともはやglucosidase IIの作用を受けなくなることを示唆していた。次に、このMRHドメインの糖結合に関与するアミノ酸残基を特定し、それらの変異体を作成することにより、糖結合活性が消失すること、さらにこの変異体を用いてglucosidase IIβサブユニットのMRHドメインのレクチン活性がglucosidase 活性に及ぼす影響について調べた。glucosidase IIは酵素活性を担う触媒サブユニットであるα、および機能未知のβサブユニットのヘテロ二量体である。βサブユニットのMRHドメインの糖結合活性を消失させた変異体は、野生型と同様にαサブユニットと複合体を形成すること、また、人工基質であるp-ニトロフェニルαグルコシドを基質として酵素活性を測定すると、wild-typeのβサブユニットと変異体のβサブユニットとの間に、glucosidase活性の差は見られなかった。一方、内在性の基質であるGlc1Man9GlcNAc2(G1M9)を基質として酵素の活性を測定すると、その結果は人工基質とは異なり、βサブユニットのレクチン活性を消失した酵素は、glucosidase活性が顕著に低下していた。この結果は、従来の仮説では説明できない結果であり、MRHドメインのレクチン活性がその酵素活性に大きく寄与していることを示す重要な発見であった。

第2章では、マレクチンという小胞体内に局在する新規のレクチン様分子について検討を行った結果をまとめている。この分子はアフリカツメガエルでその存在が報告されている以外は、何の情報もない。胡丹はこのヒトホモログ遺伝子を探索し、この分子を大腸菌で発現させて、さまざまな解析を試みた。まず、糖結合活性の有無を検討するために、上記で確立した可溶型四量体を作成し、細胞表面との結合を調べた。その結果、デオキシノジリマイシンを細胞に処理しαグルコシダーゼIおよびIIを阻害することにより、細胞表面にG2M9、G3M9糖鎖の発現を誘導すると、マレクチンが結合するようになることを見出した。この結果は、フロンタルアフィニティクロマトグラフィーによる解析によっても詳細に検証され、マレクチンがG2M9特異的レクチンであることを示した。これまで、このような特異性をもつレクチンの存在が知られていなかっただけでなく、N型糖鎖前駆体が、全ての生物種において、G3M9という構造をもつ生物学的意義を解明するための有力な手がかりを得た意味で大きな成果であった。また、マレクチン変異体を作成し、糖結合に寄与するアミノ酸を同定、これに変異を導入しレクチン活性が消失すること、グルコシダーゼとの競合によりカルネキシンサイクルへの制御が行われている可能性についても検討した。これらの結果は、小胞体内のレクチン分子の解析を通して、N型糖鎖をタグとして新生糖タンパク質の品質管理を行っていることを明確にした点で、大きな意義のある成果である。

なお、本論文の第1章は、神谷由紀子、戸谷希一郎、神谷大貴、山口大介、松尾一郎、松本直樹、伊藤幸成、加藤晃一、山本一夫との共同研究であるが、論文提出者が全面的に主体となって実験・解析および考察を行ったものであり、提出者の寄与が十分であると判断する。

従って、博士(生命科学)の学位を授与できると認める。