Structure of human aspartyl aminopeptidase complexed with substrate analogue: insight into catalytic mechanism, substrate specificity and M18 peptidase family
- First Online:
- Cite this article as:
- Chaikuad, A., Pilka, E.S., De Riso, A. et al. BMC Struct Biol (2012) 12: 14. doi:10.1186/1472-6807-12-14
- 4.1k Views
Aspartyl aminopeptidase (DNPEP), with specificity towards an acidic amino acid at the N-terminus, is the only mammalian member among the poorly understood M18 peptidases. DNPEP has implicated roles in protein and peptide metabolism, as well as the renin-angiotensin system in blood pressure regulation. Despite previous enzyme and substrate characterization, structural details of DNPEP regarding ligand recognition and catalytic mechanism remain to be delineated.
The crystal structure of human DNPEP complexed with zinc and a substrate analogue aspartate-β-hydroxamate reveals a dodecameric machinery built by domain-swapped dimers, in agreement with electron microscopy data. A structural comparison with bacterial homologues identifies unifying catalytic features among the poorly understood M18 enzymes. The bound ligands in the active site also reveal the coordination mode of the binuclear zinc centre and a substrate specificity pocket for acidic amino acids.
The DNPEP structure provides a molecular framework to understand its catalysis that is mediated by active site loop swapping, a mechanism likely adopted in other M18 and M42 metallopeptidases that form dodecameric complexes as a self-compartmentalization strategy. Small differences in the substrate binding pocket such as shape and positive charges, the latter conferred by a basic lysine residue, further provide the key to distinguishing substrate preference. Together, the structural knowledge will aid in the development of enzyme-/family-specific aminopeptidase inhibitors.
KeywordsAspartyl aminopeptidaseDodecameric tetrahedronM18 peptidaseMetalloproteaseDomain swapping
Aminopeptidases (APs) catalyze the sequential removal of amino acids from the unblocked N-termini of protein or peptide substrates, a process necessary for intracellular metabolism  and implicated in several human diseases . Most APs are metalloproteases and are classified based on substrate preference towards an acidic, basic or neutral amino acid at the P1 position of the scissile peptide bond. Very few acidic APs are known to date, the most extensively studied being the membrane-bound glutamyl aminopeptidase (ENPEP, also known as aminopeptidase A; EC 184.108.40.206) . ENPEP, a membrane-bound Ca2+-activated enzyme, is involved in the renin-angiotensin system (RAS) by catalysing the conversion of angiotensin II to angiotensin III, a key regulator of blood pressure [4, 5]. A second, cytosolic acidic AP has been reported in yeast, fungi and mammals, and termed aspartyl aminopeptidase (DNPEP, also known as DAP; EC 220.127.116.11) due to its preference for aspartate over glutamate at the P1 position [6–8]. In mammals, DNPEP is preferentially expressed and has high enzymatic activity in neurons and neuroendocrine tissues [6, 9, 10]. Its reported conversion of angiotensin I to angiotensin 2–10 , and of angiotensin II to angiotensin III in vitro, implicates a role in RAS and regulation of blood pressure. Moreover, a mild antagonist effect of DNPEP towards the bone morphogenetic protein signalling pathway has recently been reported .
DNPEP is the sole mammalian entry for the M18 metallopeptidase family, which contains ~600 putative members from bacteria and eukaryotes . The M18 family, together with the M20, M28 and M42 families, are classified into the metalloprotease H (MH) clan of proteases on the basis of active site sequence conservation according to the MEROPS database [12, 13]. Only a handful of M18 enzymes have been biochemically characterized in any detail; these include yeast vacuole aminopeptidase I (API, also known as Lap4) with a broad substrate specificity for non-polar amino acids , as well as yeast yhr113w (also known as Ape4)  and mammalian DNPEP which prefer an acidic amino acid. These M18 enzymes are shown to homo-oligomerize, reminiscent of the self-compartmentalization strategy in the well-characterized proteasomes to confer specificity towards unfolded polypeptides and not folded proteins . However, the reported dodecameric form in yeast Lap4 and Ape4 [7, 14] contrasts with the proposed octameric form in DNPEP .
Little is known about the structure-function relationship of DNPEP and other M18 members, which contain a binuclear metal centre in the active site but lack the signature Zn2+-binding sequence motif (HExxH + E) found in other metalloproteases such as ENPEP . Although several conserved histidines essential for catalysis have been identified in human DNPEP , their roles are yet to be elucidated. In this study we determined the crystal structure of human DNPEP (hDNPEP) complexed with catalytic Zn2+ and substrate analogue L-aspartate-β-hydroxamate (ABH), and confirmed its dodecameric architecture by electron microscopy (EM). The bound ABH ligand highlights the importance of a domain-swapped loop in constructing the active site and provides a structural basis for hDNPEP’s catalytic mechanism and substrate specificity. By comparison with available bacterial M18 structures we further develop a family-wide description of this unannotated peptidase family and suggest unifying catalytic features across the MH clan.
Results & discussion
Overall structure of hDNPEP
Structural comparison of M18 hDNPEP with members of other MH clan families (M20, M28 and M42) (Figure 1D) shows that the proteolytic domains of all four families can be superimposed well (pairwise rmsd ~2.3 Å), particularly in the core β-sheet and the binuclear metal centre. This structural homology suggests an evolutionarily-conserved strategy for metal coordination and metal-assisted catalysis . Away from the proteolytic domain, however, the four families diverge structurally in the dimerization domain, with M28 members lacking this domain altogether (Figure 1D, right), a fact that is reflected in their different oligomeric states. The hDNPEP dimerization domain exhibits closer topology and orientation to the dodecameric M42 enzymes (Figure 1D, bottom) [18–20], but has distinct fold and tertiary arrangements compared to the counterpart domain in M20 members (Figure 1D, top) that are known monomers or dimers [21, 22]. This observation suggests a closer structural relationship of M18 with M42 enzymes, than with M20 or M28 members, a feature not apparent from sequence-based comparisons. This is further supported by M18 and M42 members sharing similar oligomeric assembly and active site architecture (see following sections).
hDNPEP dodecameric tetrahedron
The hDNPEP dodecamer contrasts with an octameric arrangement previously deduced from native PAGE analysis . As an independent verification we performed EM image analysis, revealing one homogeneous particle population on micrographs (Figure 2D) with characteristic patches of density surrounding a hole in the middle, corresponding to the 3-fold symmetrical view down the wide channels at a facet of the tetrahedron complex on the 2D classification (Figure 2E). The tetrahedron shape and dimensions from the EM projection are in excellent agreement with the crystallographic dodecamer (Figure 2F), lending support to its physiologically relevance. While the oligomeric state of the bacterial M18 homologues is not reported, their crystal structures suggest the formation of dodecameric tetrahedrons similar to hDNPEP (Additional file1, Figure S2), pointing towards a common self-compartmentalization strategy for catalysis.
Architecture of wide and narrow channels
The narrow channels (Figure 3C) are located at the interface of three monomers that are constituents of different dimers, giving rise to an inner helical bundle with a β-barrel-like outer casing (Figure 3D and Additional file1, Figure S3). The essential nature of this channel has been demonstrated for some tetrahedron aminopeptidases . In hDNPEP, we observed water, glycerol molecules and a hydrated Mg2+ ion within this channel (Figure 3C and D), suggesting a possible route for small molecules such as cleaved amino acids to exit after hydrolysis . The narrow channel may also provide a path for the translocation of metal ions (e.g. catalytic zinc), mediated by layers of charged residues within the channel. However, to achieve either transit function, slight conformational changes may be required to open up the channel pore considering its narrow width (~3 Å)(Figure 3C).
Metal-dependent active site
Additional coordination to the binuclear zinc is provided by the bound ABH molecule, a competitive inhibitor of hDNPEP [6, 17]. ABH binds to the active site with the hydroxamate moiety towards the binuclear metal centre to contribute its carbonyl and hydroxyl oxygen atoms for zinc coordination (Figure 4C), while its amino-acid backbone protrudes into a cavity often known as the P1 substrate pocket (Figure 4B). ABH engages in a number of direct or water-mediated hydrogen bonds to Glu301 and Asp346 via the hydroxamate moiety, and to Tyr381, Lys374 and His349 via the amino acid backbone (Figure 4D). Of particular interest is an interaction between the hydroxamate carbonyl oxygen and His170 from the opposing subunit (His170b) of a dimer (Figure 4D). His170b sits at the tip of the β8-β9 loop from the neighbouring subunit that crosses over to complete the active site (Figure 4B and Additional file1, Figure S1). Such loop swapping to translocate a distant ligand-binding residue into the active site is crucial to hDNPEP catalysis, as evident by a complete abolishment of activity in a His170Phe mutant . This histidine residue is also conserved in M42 enzymes, although in the available M42 structures the equivalent loops are disordered or partially disordered. This disorder could be due to the lack of bound substrate/analogue, suggesting a substrate-induced conformational reorientation is necessary to complete the catalytic centre. Conservation of this histidine therefore implies that the loop-swapped active site is a common structural feature among M18 and M42 dodecamers built from dimeric units.
A possible catalytic mechanism for M18 hDNPEP is proposed (Figure 4E), on the assumption that the hydroxylamine nitrogen and carbonyl oxygen of the ABH hydroxamate represent where the amine and carbonyl groups of the substrate peptide would be coordinated by Zn2 and Zn1, respectively. A nucleophilic water molecule could feasibly occupy the position of the ABH hydroxyl oxygen and would be activated by Glu301 to attack the scissile bond. His170b can function to bind the peptide carbonyl oxygen and stabilize the tetrahedral intermediate. This mechanism is consistent with that proposed for other metallopeptidases .
Structure basis for hDNPEP substrate specificity
The bound ABH provides a template to build dipeptide models of Asp-Ala and Glu-Ala into the active site in order to rationalize hDNPEP substrate specificity. For both peptides, the Asp and Glu sidechains fit into the P1 substrate pocket without steric constraints, while the mainchain is modeled onto the hydroxamate group of ABH in a position optimal for hydrolysis. The P1 substrate pocket (Additional file1, Figure S4A) is created by strand β15 and the β16-α12 and β17-α13 loops, with the β17-α13 loop lining the wall and restricting the dimensions of the pocket. This limited space disfavours bulky hydrophobic residues, as illustrated by a structural comparison with the P1 pockets in M28 neutral aminopeptidases where the equivalent loop is displaced away from the P1 pocket thereby generating a large cavity for bulky residues such as Phe and Met (Additional file1, Figure S4B and C).
In summary, we provide a structural annotation of the M18 metallopeptidase family, highlighting common catalytic residues and oligomeric properties. In particular, a loop-swapped active site utilizing a residue from an adjacent subunit for catalysis is likely a common characteristic among M18 and M42 dodecameric aminopeptidases. Furthermore, the bound substrate analogue in the active site provides insight into the reaction mechanism and substrate specificity for hDNPEP, facilitating the next steps in the development of family-specific small-molecule binders to further probe its cellular role in metabolic pathways and disease.
Cloning, expression and protein purification
A DNA fragment encoding hDNPEP aa 1–468 (Uniprot ID: Q9ULA0) was sub-cloned into the pNIC-CTHF vector, incorporating a C-terminal His6-tag and TEV protease site. The recombinant protein was expressed in E. coli BL21(DE3)-R3 by induction with 0.1 mM IPTG overnight at 18°C. Cells were harvested and homogenized in lysis buffer (50 mM HEPES pH 7.5, 500 mM NaCl, 5 mM imidazole, 5% glycerol). Protein was purified by affinity (Ni-Sepharose) and size exclusion chromatography (Superdex 200). The affinity tag was removed by His-tagged TEV protease and the TEV-cleaved protein was passed over Ni-Sepharose resin. Purified protein was stored at −80°C in 10 mM HEPES, pH 7.5, 500 mM NaCl, 5% glycerol and 0.5 mM TCEP.
Crystallization and data collection
hDNPEP (10 mg/ml) was pre-incubated with 5 mM L-aspartate-β-hydroxamate (ABH) and crystallized by sitting drop vapour diffusion at 20°C in a 150-nl drop by mixing protein and reservoir solution (15% w/v PEG 3350, 0.25 M MgCl2 and 0.1 M Tris–HCl pH 8.0) in a 2:1 ratio. Crystals were cryo-protected with mother liquor supplemented with 25% glycerol and flash-cooled in liquid nitrogen. Diffraction data were collected at Diamond Light Source beamline I03, and processed and scaled with MOSFLM and SCALA from the CCP4 suite .
Data Collection and Refinement Statistics
PDB accession code
Diamond Light Source, I03
Resolution rangea (Å)
56.11 – 2.20 (2.32 – 2.20)
Unit cell dimensions
a = b = c = 224.60 Å; α = β = γ = 90.0°
No. unique reflectionsa
Rmergea (%); Rpima (%)
17.9 (83.3); 5.5 (29.7)
Wilson B factor (Å2)
No. atoms in refinement (P/L/M/O)c
Bf (P/L/M/O)c (Å2)
rms deviation bond lengthb (Å)
rms deviation bond angleb (°)
hDNPEP at ~0.7 μM was applied to EM grids and stained with 2% uranyl acetate. Electron micrographs were recorded (x 45,000) using a FEI-Phillips CM120 EM. Images were digitized on a Nikon Super Coolscan 9000 (step size of 12.5 μm with a pixel size of 2.78 Å). The WEB and SPIDER software  were used for image processing. 4,736 particles were windowed, subjected to reference-free alignment, and sorted into classes using the K-means clustering method . Manual fitting of the hDNPEP crystal structure into the 2D map was achieved using CHIMERA .
The atomic coordinates and structure factors have been deposited in the Protein Data Bank (http://www.rcsb.org/) with accession number 4DYO.
We thank staff at the Diamond Light Source for help with diffraction data collection. The Structural Genomics Consortium is a registered charity (number 1097737) that receives funds from the Canadian Institutes for Health Research, the Canadian Foundation for Innovation, Genome Canada through the Ontario Genomics Institute, GlaxoSmithKline, Karolinska Institutet, the Knut and Alice Wallenberg Foundation, the Ontario Innovation Trust, the Ontario Ministry for Research and Innovation, Merck & Co., Inc., the Novartis Research Foundation, the Swedish Agency for Innovation Systems, the Swedish Foundation for Strategic Research and the Wellcome Trust.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.