Abstract

High-temperature requirement protease A4 (HtrA4) is a secretary serine protease whose expression is up-regulated in pre-eclampsia (PE) and hence is a possible biomarker of PE. It has also been altered in cancers such as glioblastoma, breast carcinoma, and prostate cancer making it an emerging therapeutic target. Among the human HtrAs, HtrA4 is the least characterized protease pertaining to both structure and its functions. Although the members of human HtrA family share a significant structural and functional conservation, subtle structural changes have been associated with certain distinct functional requirements. Therefore, intricate dissection of HtrA4 structural and functional properties becomes imperative to understand its role in various biological and pathophysiological processes. Here, using inter-disciplinary approaches including in silico, biochemical and biophysical studies, we have done a domain-wise dissection of HtrA4 to delineate the roles of the domains in regulating oligomerization, stability, protease activity, and specificity. Our findings distinctly demonstrate the importance of the N-terminal region in oligomerization, stability and hence the formation of a functional enzyme. In silico structural comparison of HtrA4 with other human HtrAs, enzymology studies and cleavage assays with X-linked inhibitor of apoptosis protein (XIAP) show overall structural conservation and allosteric mode of protease activation, which suggest functional redundancy within this protease family. However, significantly lower protease activity as compared with HtrA2 indicates an additional mode of regulation of the protease activity in the cellular milieu. Overall, these studies provide first-hand information on HtrA4 and its interaction with antiapoptotic XIAP thus implicating its involvement in the apoptotic pathway.

Introduction

High-temperature requirement protease A (HtrA) is a family of oligomeric serine proteases that are conserved from prokaryotes to humans. The Escherichia coli counterparts DegP, DegQ, and DegS have dual regulatory functions; they act as molecular chaperones at lower temperature and exhibit proteolytic activity with an increase in temperature. In humans, four HtrA homologs (HtrA1–4) have been identified [13]. These proteins share a common domain architecture that comprises an N-terminal region, a serine protease domain (SPD) for enzymatic activity and a PDZ domain (post-synaptic density of 95 kDa, disk large, and zonula occludens-1 domain) for protein–protein interactions. Apart from these common domains, all except HtrA2, harbor a Kazal-like domain that acts as an inhibitor for serine protease activity, insulin-like growth factor-binding protein (IGFBP) domain, and a signal peptide [4,5]. While HtrA1, HtrA3, and HtrA4 share very similar domain organizations, the uniqueness of HtrA2 is manifested through the absence of Kazal and IGFBP domains and existence of a distinct N-terminal region, which includes a prominent IAP (inhibitor of apoptosis)-binding motif (IBM) and a transmembrane domain [6].

In humans, these proteases are involved in numerous cellular processes ranging from the maintenance of mitochondrial homeostasis to cell death. Therefore, deregulation of their normal functions leads to various diseases that include neurodegeneration, neuromuscular disorders, arthritis, age-related macular degeneration, pregnancy-specific pre-eclampsia (PE), and cancer [714]. Available literature suggests that these serine proteases form a functional oligomeric assembly that is primarily mediated through their N-terminal region. While hexameric and higher-order oligomeric complexes are observed in bacterial HtrAs, their human counterparts are mostly trimeric. For example, in E. coli HtrA (DegP), the trimeric units further oligomerize to form higher-order oligomers upon substrate binding that range from trimeric to 24-mer ensembles [15]. Therefore, despite overall structural conservation, subtle conformational differences and distinctness in oligomeric assembly might define diversity in cellular and pathophysiological functions in HtrA family members.

Among human HtrA homologs, HtrA1 and HtrA2 are well characterized with available high-resolution crystal structures. The secreted trimeric HtrA1 exhibits endo-proteolytic activity, including autocatalytic cleavage [16,17]. It binds, cleaves, and activates a variety of targets, which are involved in critical physiological processes such as maintenance of cell death and cell signaling. HtrA2, the most elaborately studied family member, has the unique characteristic of being allosterically activated through its N- as well as C-terminal regions [18,19]. Upon apoptotic induction, the first 133 residues that include the mitochondrial localization signal get cleaved, leading to the formation of the active mature form of HtrA2 with the exposure of the tetrapeptide IBM that comprise AVPS residues at the N-terminus. Therefore, the involvement of HtrA2 in apoptosis and other cellular processes has implicated it in various diseases including neurodegeneration and cancer. Although some studies have been done on HtrA3, its partial low-resolution structure might not be sufficient for providing a detailed snapshot of its intricate structural complexities [20]. However, HtrA4 is the least characterized human HtrA protease concerning both structure and its functions. It has been found to be up-regulated in PE and hence has been suggested to be used as a biomarker of PE [14]. Meta-analyses of available microarray data suggested that HtrA4 is up-regulated in glioblastoma multiforme and breast carcinoma, whereas, down-regulated in metastatic prostate cancer; making it an emerging therapeutic target [12,21]. Therefore, an intricate dissection of HtrA4 structure with an understanding of its functional reciprocity becomes imperative to harness its properties with desired characteristics. In this study, in silico, biochemical and biophysical characterizations of HtrA4 have been performed to delineate the role of different domains (individually and in various combinations) in defining substrate specificity, oligomerization, stability, and allosteric properties if any. Our study highlights the importance of the short N-terminal region in maintaining stability and oligomerization of HtrA4. The observation that PDZ domain positively influences protease activity suggests the requirement of allosteric modulation and intricate inter-molecular interaction network in HtrA4 functions. This information coupled with enzymology studies provides crucial information regarding the mechanism of activation of HtrA4.

Full-length HtrA4 consists of a signal peptide (1–31 residues), an IGFBP N-terminal (36–99 residues), a Kazal-like (88–154 residues) domain, a serine protease domain or SPD (202–362 residues), and one PDZ (382–474 residues) domain. The active site residue S326 along with H218 and D248 form the catalytic triad. Comparison of our modeled HtrA4 structure with other human HtrA counterparts shows overall sequence and structural conservation suggesting functional redundancy within this protease family. For example, the N-terminal IBM motif as observed in HtrA2 is conserved in HtrA4 (AVPV), implicating its possible role in apoptosis. This observation has been validated through binding and cleavage studies with X-linked inhibitor of apoptosis protein (XIAP), suggesting its involvement in programmed cell death. Overall, this study for the first time provides a fundamental understanding of the structural organization of HtrA4, its mechanism of action and hence a lead toward delineating its role in normal cellular processes as well as pathogenicity.

Materials and methods

Plasmid constructions

Human ORF (open reading frame) sequence of HTRA4 (1–476 amino acids) cloned in the pBluescriptR vector was obtained from the Mammalian Genome Collection (MGC, NIH, U.S.A.). Using appropriate primers as shown in Supplementary Table S1, different constructs of HTRA4, i.e. SPD-PDZ, N-SPD, SPD, and PDZ were PCR-amplified and further sub-cloned into pET20b (+) expression vector (New England Biolabs, MA, U.S.A.) with a C-terminal His6 tag using Nde-1 and Xho-1 as restrictions sites. All HtrA4 constructs were confirmed by sequencing.

Recombinant protein expression and purification

Rosetta (DE3) pLysS cells (New England Biolabs, MA, U.S.A.) were transformed with pET20b (+) plasmid containing HtrA4 variants. Cells were grown at 37°C until the OD600 reached 0.6 followed by induction with 0.6 mM IPTG. Cells were further cultured at 16°C for 18-h post-induction. Cells were lysed by sonication in lysis buffer containing 20 mM Na2HPO4/NaH2PO4 and 200 mM NaCl at pH8.0 (Buffer A) with 10 mM imidazole and 0.2% (v/v) Triton X-100. Lysates were centrifuged at 14 000 rpm for 30 min, and the supernatant was incubated with His60 Ni Superflow resin (DSS Takara Bio India Pvt. Ltd., New Delhi, India) for 1 h at 4°C. Unbound protein was washed with wash buffer (Buffer A with 10 mM imidazole and 2% (v/v) glycerol) and eluted with an imidazole gradient (20–500 mM) in elution buffer (Buffer A with 2% (v/v) glycerol). Eluted protein (>95% purity) was dialyzed overnight in Buffer A and stored at −80°C.

Full-length XIAP clone in pGEX-4T vector with N-terminal glutathione S-transferase (GST)-tag was obtained from Addgene (Cambridge, MA, U.S.A.). Protein with GST-tag (GST-XIAP) was purified by affinity chromatography using GST-sepharose resin (Novagen, MA, U.S.A.) in buffer 20 mM Na2HPO4/NaH2PO4, pH 7.8 containing 100 mM NaCl, 25 mM β-mercaptoethanol (β-ME), and 50 µM ZnCl2. The fractions with the highest percentage of purity as estimated by SDS–PAGE were pooled and stored at −80°C.

Circular dichroism spectroscopy

HtrA4 and HtrA4(S326A) proteins were dialyzed in Buffer A, and the concentration was estimated by Bradford assay (Bio-Rad, California) as well as absorbance (A280) measurement (NanoDrop ND-1000 spectrophotometer). Far-UV circular dichroism (CD) spectra (190–250 nm) of HtrA4 and its variants were recorded with a protein concentration of 10 µM at 37°C using a JASCO J815 spectropolarimeter (Jasco, Easton, MD, U.S.A.). Each experiment was baseline-corrected and was done in triplicate. Secondary structural characteristics were analyzed using K2D2 online software [22]. Thermal stability of proteins was assessed by monitoring the CD spectrum as a function of temperature (30–95°C). Ellipticity corresponding to 222 nm at different temperatures was obtained for calculation of melting temperature (Tm). Plots were drawn using GraphPad Prism7 software (GraphPad Software, Inc., CA, U.S.A.).

Size exclusion chromatography

The molecular masses of HtrA4 and its variants were estimated by size exclusion chromatography. Post-affinity purification, all protein constructs were concentrated to 2–3 mg/ml, and 1 ml of samples were injected into a Superdex 200 10/300 HR column (GE Healthcare, Uppsala, Sweden), pre-equilibrated with Buffer A. Proteins were eluted with the same buffer at a flow rate of 0.6 ml/min and molecular masses were calculated using commercially available gel filtration standards (Sigma, St. Louis, MO, U.S.A.). Elution volume/void volume (Ve/V0) versus the log of molecular masses of standards were plotted to calculate molecular masses of HtrA4 protein and its variants. HtrA4 oligomerization status in the presence of generic substrate β-casein (Sigma, St. Louis, MO, U.S.A.) was analyzed by pre-incubating inactive HtrA4(S326A) and β-casein for 12 h in 1:2 molar ratio, respectively, at 4°C. This HtrA4–β-casein complex was subjected to analysis by using size exclusion chromatography as described above.

Enzyme activity assays

To determine the protease activity of different constructs of HtrA4, generic serine protease substrate β-casein was used. For each 30 µl reaction mixture, 6 µg of β-casein was incubated with 4 µg of different HtrA4 variants in Buffer A at 37°C for 10 h. Protease assays were also performed at a temperature range of 30–45°C for 10 h using β-casein as substrate. The reaction was stopped by adding 2× SDS sample loading buffer followed by boiling for 5 min. All reaction products were analyzed on 12% Tris–tricine gel and bands were quantified using ImageJ software (1.47V, NIH, U.S.A.). For all quantitative enzyme kinetic studies, fluorescein isothiocyanate (FITC)-labelled casein (Sigma, St. Louis, MO, U.S.A.) was used as a substrate. Fluorescent substrate cleavage was determined by incubating 30 µM of HtrA4 enzyme with increasing concentration (0–20 mM) of β-casein at 37°C in cleavage buffer (Buffer A with 0.1 mM DTT). Using an excitation wavelength of 485 nm and emission at 545 nm, fluorescence was monitored in a multi-well plate reader (Berthold Technologies, Germany). The steady-state kinetic parameters were obtained by fitting data to the Hill form of the Michaelis–Menten equation: , where Vmax is the maximum velocity and KM is the substrate concentration at half-maximal velocity, using KaleidaGraph software (Synergy Software, Reading, PA, U.S.A.). Assays were done in triplicate, and the average data were plotted. To determine the optimal temperature for HtrA4 enzymatic function, protease activity study over a temperature range of 30–45°C with an interval of 5°C was performed. For each reaction mixture, 6 µg of β-casein and 4 µg of different constructs of HtrA4 in Buffer A were incubated for 10 h at the respective temperature. The reaction was stopped by adding 2× SDS sample loading buffer and boiling for 5 min. All reaction products were analyzed in 12% Tris–tricine gel and bands were quantified using ImageJ (1.47V, NIH, U.S.A.). Graphs were plotted using GraphPad Prism 7.

N-terminal sequencing

To determine the substrate specificity of HtrA4, generic serine protease substrate β-casein was used. HtrA4 (4 µg) was incubated with β-casein (10 µg) in Buffer A at 37°C for 10 h. The proteolytic products were analyzed on 12% Tris–tricine gel and transferred onto PVDF membrane (Millipore Corporation, Billerica, MA, U.S.A.) using Bio-Rad wet transfer apparatus in 1× transfer buffer (25 mM Tris, 192 mM glycine, 20% (v/v) methanol, 0.025–0.1% SDS, pH 8.3) at a constant voltage of 20 V for 10 h. Transferred proteins were stained with 0.1% amido black, 25% isopropanol and 10% acetic acid solutions, followed by destaining in 40% methanol and 10% acetic acid. The stained fragments were excised from the membrane and sent to protein sequencing facility for identification of five to seven N-terminal amino acids by ABI 494 sequence pro Protein sequencers (Tufts University Core Facility, Boston, MA, U.S.A.).

Pull-down studies

HtrA4 and XIAP binding studies were performed using in vitro pull-down assays where GST-tagged XIAP was used as bait and HtrA4 (S326A) as prey. 10 µg of GST-XIAP cell lysate was incubated with 20 µl of GST-sepharose resin (Novagen) for 3 h at 4°C. To prevent nonspecific protein binding to GST resin, beads were incubated with 5% BSA for 10 h followed by extensive washing with Buffer A containing 0.5 mM DTT and 0.1% TritonX-100 (pH 7.2). After washes, 100 µg of purified recombinant HtrA4 was incubated with GST-bound XIAP for 2 h at 4°C. Unbound HtrA4 was removed by washing the beads thrice with Buffer A followed by boiling in 30 µl Laemmli buffer. The bound proteins were separated on 12% SDS–PAGE and were probed with an anti-His antibody (Abcam, Cambridge, MA, U.S.A.) against His6-HtrA4 protein.

Molecular modeling of HtrA4

Molecular modeling of HtrA4 was performed using homology modeling. Human HtrA4 sequence starting from 144AVPV (UniProt ID: P83105) was retrieved from UniProt database and submitted to I-TASSER server for three-dimensional (3D) structure prediction [23]. This server prepares models either by comparative modeling or de novo threading methods depending on the availability of the templates for the particular model. The resultant HtrA4 model given by I-TASSER was built using other three HtrA homologs, namely, HtrA1 (PDB ID: 3NUM), HtrA2 (PDB ID: 1LCY), and HtrA3 (PDB ID: 4RI0) as templates [20,24,25]. Quality of the predicted model was checked by Ramachandran plot, which showed few residues lying in the energetically disallowed regions of the plot [26]. Those outlier residues were refined using ModLoop online server, and the refined structure was further validated using Ramachandran plot [27]. The stability of the model was assessed using molecular dynamics simulation (MDS) method to generate the best conformer. This final model was later used for SiteMap analysis and docking.

SiteMap analysis and docking

The binding sites of HtrA4 (PMDB ID: PM0081859) model were predicted using SiteMap (SiteMap, Schrödinger, LLC, New York, NY, 2018) that generated a list of five binding pockets (Supplementary Table S3). These pockets were scored based on their volume, hydrophobic and hydrophilic characters, the degree to which ligand might donate or accept hydrogen bonds and exposure to solvent. Hence, the top-ranked site based on the site score was selected for the analysis and named as the selective binding pocket (SBP_HtrA4). The SBP_HtrA4 includes residues from both SPD and PDZ domain along with linker region residues.

For docking analysis, the generic substrate β-casein (UniProt ID: P02666) was used. Based on our earlier studies, C-terminal oligopeptide of β-casein (199DMPIQAFLLYQEPVLGPVRGPFPIIV224) was retrieved and divided into 23 heptamer peptides [28]. An 11-mer peptide having the sequence LNQPKNNPKNN that does not match the consensus SBP_HtrA4-binding peptide pattern was used as a negative control. All these peptides were built in silico using 3D builder (Schrödinger, LLC, New York, 2018). They were further pre-processed and energy minimized using the protein preparation wizard (Schrödinger, LLC, New York, 2018) and subsequently docked with HtrA4. Energy-minimized HtrA4 structure was prepared for docking and grid file was generated using the protein structure and SiteMap output (SiteMap, Schrödinger, LLC, New York, NY, 2018). Docking was performed using peptide docking tool in Bioluminate platform (Bioluminate, Schrödinger, LLC, New York, NY, 2018) and was scored using MM-GBSA (Molecular Mechanics using the Generalized Born Surface Area continuum) scoring method [29]. A subset of best possible peptides for further studies was selected based on docking score.

Molecular dynamics simulation and analysis

The HtrA4–β-casein complex (PMDB ID: PM0081893), which scored the highest with respect to binding affinity and glide score, was given for MDS run using GROMACS 2018 [30]. AMBER99sb-ILDNP force field (OPLS-AA) was used to generate topology and parameter files [31]. The complex was solvated using TIP4P water molecules and neutralized by adding an adequate number of positive (Na+) ions [32]. Similarly, unbound HtrA4 structure (PMDB ID: PM0081859) was also prepared to use it as a control for the simulation analysis. Neutralization was completed by adding four Na+ ions for unbound HtrA4 and three Na+ ions for HtrA4–β-casein complex (PMDB ID: PM0081893). Two rounds of steepest–descent minimization were performed for each system where all the electrostatic and van der Waals interactions were treated using the particle-mesh Ewald (PME) method [32,33].

Energy minimization was followed by system equilibration to establish a favorable orientation of water molecules and Na+ ions around unbound HtrA4 and HtrA4–β-casein complex. The system was subjected to equilibration performed under NVT (N = number of particles, V = system's volume, T = absolute temperature) or isothermal–isochoric ensemble for 500 ps and was followed by NPT (N = number of particles, P = system's pressure, T = absolute temperature) or isothermal–isobaric ensemble equilibration for 2000 ps [34]. For both phases of equilibration, LINCS (LINear Constraint Solver) constraint algorithm was used to apply position restraining force on all the heavy atomic bonds in the protein [35]. Finally, a 30 000 ps MD simulation (production run) was conducted for each system under NPTensemble, and the resultant trajectories were analyzed. To conduct a comparative analysis, the best structure was chosen from each simulation using a clustering method [36]. Clusters were generated at 100 ps intervals throughout the 30 ns simulation with an output of 300 structures for each simulation. Among those, the structure representing the largest cluster was taken as the best conformer for comparative analysis. Comparisons were based on their overall as well as domain-wise RMSD (root mean square deviation) and were plotted using XmGrace (ftp://plasma-gate.weizmann.ac.il/pub/grace/).

Results

Purification and secondary structural characterization of HtrA4 and its variants

Different variants of HtrA4 were generated using polymerase chain reaction (PCR) and site-directed mutagenesis (SDM) to understand the role of different domains and their combinations in regulating HtrA4 stability and functions (Figure 1A). Pairwise sequence alignment of HtrA4 with mature HtrA2 showed 54% sequence similarity and also displayed a conserved N-terminal 144AVPV motif similar to that of 133AVPS motif (IAP-binding motif) in HtrA2 (Supplementary Figures S1–S4). Based on this observation, we cloned HtrA4 starting from 144AVPV residues, and this construct is referred it to as HtrA4 in the manuscript. Several literature reports highlighted the importance of the N-terminal region in regulating the enzymatic activity of HtrA family proteins [15,37,38]. Therefore, to understand the role of the N-terminal region in oligomerization, stability, and functions of HtrA4, we generated an N-terminal deleted construct (SPD-PDZ). Similarly, only the SPD variant was also generated to understand its function as a single domain. Similarly, different domain combinations of HtrA4 with active site (S326A) mutation were purified using nickel affinity chromatography and characterized.

Purification and secondary structural properties of HtrA4.

Figure 1.
Purification and secondary structural properties of HtrA4.

(A) Schematic representation of different HtrA4 variants used for the study. Different domains are color-coded and are self-explanatory. Solid lines indicate N-terminal region (144–202 residues) and the linker region (362–383 residues). (B) Far-UV CD spectra of HtrA4 and its inactive variant. (C) Thermal denaturation curves for HtrA4 and its inactive variant using circular dichroism study. Signal at 222 nm has been recorded within a temperature range of 30–90°C.

Figure 1.
Purification and secondary structural properties of HtrA4.

(A) Schematic representation of different HtrA4 variants used for the study. Different domains are color-coded and are self-explanatory. Solid lines indicate N-terminal region (144–202 residues) and the linker region (362–383 residues). (B) Far-UV CD spectra of HtrA4 and its inactive variant. (C) Thermal denaturation curves for HtrA4 and its inactive variant using circular dichroism study. Signal at 222 nm has been recorded within a temperature range of 30–90°C.

To check for structural perturbation if any in the inactive (S326A) HtrA4 variant, far-UV CD of the protein was performed and compared with its active counterpart. The similarity in secondary structure as shown in Figure 1B suggests that the mutation is not structurally perturbing [39]. Melting curves of HtrA4 and the catalytically inactive (S326A) HtrA4 implicate that the proteins have similar thermal stability with a Tm ∼58°C as shown in Figure 1C. However, Tm for the rest of the individual domains could not be calculated as they precipitated beyond 45°C suggesting that the stability of the protein is significantly compromised with deletion of domains.

Oligomerization status of various HtrA4 domains and mutants

Prokaryotic HtrA family members are capable of forming higher-order oligomeric complexes consisting of almost 24 molecules. While DegS is a stable trimer, active DegQ and DegP form higher-order oligomers comprising 12 and 24 monomeric subunits, respectively [4042]. Human HtrA homologs (HtrA1, HtrA2, and HtrA3), however, have been reported to be in the trimeric form [20,25,43], while no report on HtrA4's oligomeric property has been obtained so far. Therefore, to understand the oligomerization status of HtrA4, gel filtration chromatography was performed using a Superdex 200 gel filtration column. Elution profiles of HtrA4 variants (N-SPD, SPD-PDZ, and SPD) were also monitored, and their respective molecular masses were calculated from the calibration curve generated from protein standards. The estimated molecular masses of these proteins have been shown in Table 1, which demonstrate that HtrA4 and N-SPD are trimers. Interestingly, unlike HtrA2 [38], some amount of monomeric population is also observed in HtrA4 suggesting lesser stability of trimeric HtrA4. However, upon incubation of inactive HtrA4 with its generic substrate β-casein, only the trimeric peak was observed with complete disappearance of the monomeric population. N-SPD showed solely a trimeric ensemble implicating that the PDZ along with the linker region might provide more conformational flexibility to the protease. As expected, SPD-PDZ and SPD were found to be monomers reiterating the importance of N-terminal region in trimerization. The oligomeric properties of HtrA4 and its domain combinations are shown in Table 1.

Table 1
Oligomeric properties of different HtrA4 constructs: molecular masses of different HtrA4 constructs and their oligomerization status are determined by gel filtration chromatography using a Superdex 200 column
 Protein Theoretical molecular mass (kDa) Calculated molecular mass (kDa) Oligomeric status 
HtrA4 (Inactive) 109 120 ± 10 Trimer (∼80%) 
36 ± 8 Monomer (∼20%) 
N-SPD 71.4 66 ± 5 Trimer 
SPD-PDZ 30 28 ± 7 Monomer 
SPD 19 24 ± 8 Monomer 
HtrA4 + β-casein NA 140 ± 12 Trimer 
 Protein Theoretical molecular mass (kDa) Calculated molecular mass (kDa) Oligomeric status 
HtrA4 (Inactive) 109 120 ± 10 Trimer (∼80%) 
36 ± 8 Monomer (∼20%) 
N-SPD 71.4 66 ± 5 Trimer 
SPD-PDZ 30 28 ± 7 Monomer 
SPD 19 24 ± 8 Monomer 
HtrA4 + β-casein NA 140 ± 12 Trimer 

Protease assays

To compare enzymatic properties of HtrA4 primarily with HtrA2 as well as to understand the role of different domains and their oligomerization status in regulating its protease activity, gel-based protease assays of HtrA4 and its variants were performed using β-casein. In these studies, the active site mutant (S326A) was used as a negative control. Different HtrA4 variants were incubated with β-casein at 37°C for 10 h (Figure 2A). Uncleaved β-casein band was semi-quantified using ImageJ software and subsequently cleaved β-casein percentage was calculated and plotted as shown in Figure 2B. It was observed that both the trimeric protein HtrA4 and N-SPD cleaved β-casein although the activity of N-SPD found to be less when compared with HtrA4, particularly at higher temperatures. However, the monomeric variants, i.e. SPD-PDZ and SPD, did not show any protease activity, highlighting the importance of trimerization for HtrA4 activity.

Proteolytic activity of HtrA4 and its variants with β-casein as a substrate.

Figure 2.
Proteolytic activity of HtrA4 and its variants with β-casein as a substrate.

(A) Different constructs of HtrA4 were incubated with β-casein at 37°C for 10 h. Reaction samples were resolved by 15% Tris–tricine SDS–PAGE and visualized with Coomassie Brilliant blue staining. (B) Intensity of the substrate remained after 10 h was semi-quantified using ImageJ software and the cleaved β-casein percentage was calculated and plotted. (The error bars are the representation of SEM, n = 3). (C) Steady-state kinetic parameters of HtrA4. The steady-state kinetic parameters were calculated from the reaction rates by fitting the data to the Hill form of the Michaelis–Menten equation. Results are represented as mean ± SEM, n = 3.

Figure 2.
Proteolytic activity of HtrA4 and its variants with β-casein as a substrate.

(A) Different constructs of HtrA4 were incubated with β-casein at 37°C for 10 h. Reaction samples were resolved by 15% Tris–tricine SDS–PAGE and visualized with Coomassie Brilliant blue staining. (B) Intensity of the substrate remained after 10 h was semi-quantified using ImageJ software and the cleaved β-casein percentage was calculated and plotted. (The error bars are the representation of SEM, n = 3). (C) Steady-state kinetic parameters of HtrA4. The steady-state kinetic parameters were calculated from the reaction rates by fitting the data to the Hill form of the Michaelis–Menten equation. Results are represented as mean ± SEM, n = 3.

Moreover, to understand the rate of substrate binding and catalysis, kinetic parameters of HtrA4 were determined fluorometrically using FITC-labeled β-casein as a substrate (Figure 2C). Further to gain insights into the HtrA4 active site orientation and conformational dynamics, these kinetic parameters were compared with HtrA2 [38]. From this study, the observed Vmax (1.9 × 10−12 M s−1), kcat (0.06 × 10−6 s−1), and kcat/KM (0.0123 M−1 s−1) were relatively low when compared with HtrA2, which indicates HtrA4 is catalytically less competent than HtrA2 with lower enzymatic reaction rate and turnover number under identical experimental conditions in vitro. However, the KM value of HtrA4 has been found to be almost similar to that of HtrA2 (HtrA4-KM 4.84 ± 0.8 µM and HtrA2-KM 4.60 ± 0.5 µM [38]) suggesting similar substrate-binding affinity of the two proteases. The Hill coefficient of 1.3 for HtrA4 is indicative of allosteric activation like other members of the HtrA family [18,38].

Literature suggests that HtrA2 and its homologs show a significant increase in activity with temperature. This heat activation is associated with considerable plasticity at the PDZ–protease interface similar to activation via substrate binding at PDZ [37,38]. Since HtrA4 is a close homolog of HtrA2, we speculated that it might also exhibit similar dynamic behavior and conformational changes in the presence of temperature. Therefore, protease assays of HtrA4 were performed as a function of temperature. HtrA4 variants were incubated with β-casein in the temperature range of 30–45°C for 10 h and analyzed using Tris–tricine gels (Figure 2A). Uncleaved β-casein band after 10 h was semi-quantified using ImageJ software and subsequently cleaved β-casein percentage was calculated and plotted as shown in Figure 2B. The activity of HtrA4 was found to be high at higher temperatures (>30°C). However, for N-SPD, a subsequent decrease in activity was observed beyond 35°C, which might be due to lack of necessary conformational changes in the absence of the regulatory PDZ domain and protease destabilization.

Substrate specificity of HtrA4

Cleavage specificities of HtrA1, HtrA2, and HtrA3 have been well documented in the literature [6,20,24,4345]; therefore, to understand whether HtrA4 shares similar substrate specificity as its human counterparts, we determined the substrate specificity of HtrA4 using β-casein. Proteolytically cleaved fragments (∼23, 19, 13, 7, and less than 5 kDa) of β-casein were subjected to N-terminal sequencing by Edman degradation. The specificity profile of N-terminal sequence analysis of these fragments identified three preferred cleavage sites on β-casein (Figure 3A,B and Table 2). The first cleavage site is between 15A–16R, second between 108M–109G and third cleavage site is between 170V–171M amino acids, respectively. Cleavage at single or multiple sites resulted in a combination of different peptide fragments as shown in Figure 3C.

Generic substrate β casein cleavage byHtrA4.

Figure 3.
Generic substrate β casein cleavage byHtrA4.

(A) Protease assays of HtrA4 with β-casein: peptide fragments sent for the N-terminal sequence are indicated as B1, B2, etc. (B) Residues identified by N-terminal sequencing are highlighted in red. Preferred cleavage sites are indicated by blue arrows. (C) Schematic representation of cleavage products observed on the gel after proteolysis. The 25 kDa β-casein is cleaved into various fragments after incubating with HtrA4, initially, cleavage at site 1-generated fragments B1 (∼23 kDa) and B2 (∼19 kDa), upon further incubation, cleavage at site 2-generated B3 (∼13 kDa). Further incubation, cleavage at site 3-generated B4 (∼7 kDa) and B5 (<5 kDa) bands.

Figure 3.
Generic substrate β casein cleavage byHtrA4.

(A) Protease assays of HtrA4 with β-casein: peptide fragments sent for the N-terminal sequence are indicated as B1, B2, etc. (B) Residues identified by N-terminal sequencing are highlighted in red. Preferred cleavage sites are indicated by blue arrows. (C) Schematic representation of cleavage products observed on the gel after proteolysis. The 25 kDa β-casein is cleaved into various fragments after incubating with HtrA4, initially, cleavage at site 1-generated fragments B1 (∼23 kDa) and B2 (∼19 kDa), upon further incubation, cleavage at site 2-generated B3 (∼13 kDa). Further incubation, cleavage at site 3-generated B4 (∼7 kDa) and B5 (<5 kDa) bands.

Table 2
Substrate specificity of HtrA4: prime (P) and nonprime (P′) residues at the cleavage sites were determined by N-terminal sequencing. B1, B2, B3, B4, and B5 are the peptide fragments sent for the N-terminal sequencing as shown in Figure 3A 
 P4 P3 P2 P1 – P1′ P2′ P3′ P4′ P1′ residue 
B1 – Arg16 
B2 – Arg16 
B3 – Arg16 
B4 – Gly109 
B5 – Gly109 
 Met171 
 P4 P3 P2 P1 – P1′ P2′ P3′ P4′ P1′ residue 
B1 – Arg16 
B2 – Arg16 
B3 – Arg16 
B4 – Gly109 
B5 – Gly109 
 Met171 

Our observations suggest that HtrA4 has a strong preference for aliphatic residues at the P1 (A, V, and M) position similar to the other three human HtrA homologs. The P2 position is occupied by aliphatic (L and V) and polar (T) residues, HtrA1 and HtrA2 also preferred similar aliphatic residues (L) at this position. At P3 and P4 positions, HtrA4 preferred aliphatic (A and L) and polar residues (P and E), which also matches with HtrA1 and HtrA3 specificity. At the P1′ position, HtrA4 preferred aliphatic (G and M) and polar amino acids (R) that are similar to other family members. At P2′ position like other family members, HtrA4 also preferred aromatic (F) and aliphatic (V) residues. At P3′ position, HtrA4 preferred polar (S and P) and aliphatic (L) residues. At the same position, HtrA2 and HtrA3 showed a preference for similar polar residues (S and P) only. At P4′ position, other family members showed a preference for mostly aromatic (F, Y, and W) and polar (S), while HtrA4 preferred polar (P, E, and K) residues [6,20,24,4345]. Overall, these results suggest that HtrA family members have similar but not identical substrate specificities (Table 3).

Table 3
Substrate specificity of HtrA family members
Positions HtrA1 [24,43HtrA2 [44,45HtrA3 [20HtrA4 
P1 A, M, I, V, L M, I, V A, I, V, L A, M, V 
P2 L, T L, R E, S, Q, P L, V, T 
P3 Q, L K, R, Y E, P, L, S, V E, A, P 
P4 A, P L, P, I L, P, Q L, P 
P1′ R, S, T S, A S, A, E R, G, M 
P2′ L, P, W Y, F V, Q, L, G E, V, F 
P3′ V, M, D S, P, Y S, P, V S, P, L 
P4′ F, Y, S L, G, W, S E, K, P 
Positions HtrA1 [24,43HtrA2 [44,45HtrA3 [20HtrA4 
P1 A, M, I, V, L M, I, V A, I, V, L A, M, V 
P2 L, T L, R E, S, Q, P L, V, T 
P3 Q, L K, R, Y E, P, L, S, V E, A, P 
P4 A, P L, P, I L, P, Q L, P 
P1′ R, S, T S, A S, A, E R, G, M 
P2′ L, P, W Y, F V, Q, L, G E, V, F 
P3′ V, M, D S, P, Y S, P, V S, P, L 
P4′ F, Y, S L, G, W, S E, K, P 

Binding and cleavage of XIAP

Identification of substrates and binding partners of HtrA4 is important for providing insight into its specific role in physiological processes. It has been reported that HtrA1, HtrA2, and HtrA3 cleave XIAP to induce apoptosis, which has been very well characterized in HtrA2 [4648]. XIAP is an antiapoptotic protein that interacts with downstream caspase-3 and prevents activation of the caspase cascade [49,50]. HtrA4 shares a very high sequence similarity with other human HtrAs (Supplementary Table S2) that implies functional overlap within the family. Our sequence alignment demonstrated that HtrA4 shares an N-terminal IAP-binding tetrapeptide 144AVPV similar to 133AVPS motif of HtrA2. Based on this observation, we performed in vitro binding and proteolytic assays of HtrA4 using XIAP (Supplementary Figures S2 and S3).

For in vitro binding studies, pull-down assays were performed with GST-fused XIAP as bait and inactive HtrA4 (S326A) as prey. Pull-down demonstrated that HtrA4 interacts with GST-XIAP but not with GST-tag alone. To determine the role of the N-terminal 144AVPV in binding to XIAP if any, we performed pull-down studies with ΔAVPV construct of HtrA4. The absence of AVPV abrogated the interaction thus highlighting its importance in HtrA4–XIAP complex formation. This interaction was further confirmed by Western blot analysis using anti-His antibody. The observed result indicates that XIAP is an N-terminal binding partner of HtrA4 (Figure 4A).

Binding and protease assays of HtrA4 using GST-XIAP as a putative binding partner and substrate.

Figure 4.
Binding and protease assays of HtrA4 using GST-XIAP as a putative binding partner and substrate.

(A) GST pull-down assay. Pull down was performed using GST-XIAP as bait and HtrA4 as prey. Interaction was further validated by Western blotting using anti-His antibody. (B) Protease assays of HtrA4 variants with GST-XIAP. Different constructs of HtrA4 were incubated with GST-XIAP at 37°C for 0–24 h. Reaction samples were resolved by 15% SDS–PAGE and decrease in the intensity of the full-length GST-XIAP was observed as a function of time. (C) Quantitative analysis of GST-XIAP cleavage calculated from the intensity of the full-length protein band as a function of time. Intensity of the residual substrate after 24 h was quantified using ImageJ software. Error bars represent mean ± SEM of four independent experiments. P-value (0.0002) was calculated using a paired t-test.

Figure 4.
Binding and protease assays of HtrA4 using GST-XIAP as a putative binding partner and substrate.

(A) GST pull-down assay. Pull down was performed using GST-XIAP as bait and HtrA4 as prey. Interaction was further validated by Western blotting using anti-His antibody. (B) Protease assays of HtrA4 variants with GST-XIAP. Different constructs of HtrA4 were incubated with GST-XIAP at 37°C for 0–24 h. Reaction samples were resolved by 15% SDS–PAGE and decrease in the intensity of the full-length GST-XIAP was observed as a function of time. (C) Quantitative analysis of GST-XIAP cleavage calculated from the intensity of the full-length protein band as a function of time. Intensity of the residual substrate after 24 h was quantified using ImageJ software. Error bars represent mean ± SEM of four independent experiments. P-value (0.0002) was calculated using a paired t-test.

Further to understand whether XIAP is a substrate for HtrA4, protease assays were performed. Since XIAP was less stable, GST-tagged XIAP was used as a substrate for our study. Interestingly, HtrA4 cleaved GST-XIAP but not GST alone in a time-dependent manner suggesting XIAP might be a novel substrate of HtrA4 (Figure 4B). Further to investigate the rate of protease activity semi-quantitatively, time-based degradation of GST-XIAP was monitored. The rate of hydrolysis was quantitated by measuring the intensity of the proteolytically degraded GST-XIAP band at 24 h relative to uncleaved GST-XIAP using ImageJ software (Figure 4C). HtrA4 cleaved GST-XIAP in a time-dependent fashion with 80% of the substrate being hydrolyzed in 24 h.

Further to corroborate our binding studies, we performed protease assays with ΔAVPV construct (Figure 4B). No cleavage of GST-XIAP was observed with this deletion construct of HtrA4 as expected. These observations suggest binding to 144AVPV might be a pre-requisite for subsequent cleavage of XIAP by HtrA4. In all the protease assays, inactive HtrA4 (S326A) was used as a negative control. We could not do further analysis of the cleaved products as the cleaved fragments of XIAP were very small and remained undetected even in Tris–tricine gel.

Structural analysis and identification of binding pocket of HtrA4

An HtrA4 model (PM0081859), generated from I-TASSER server, was validated using Ramachandran plot (Supplementary Figure S5) and analyzed using PSIPRED tool to understand the orientation of the secondary structures [23,51]. PSIPRED showed a total of 5 α helices, 20 β-strands, and 8 major loops traversing throughout the SPD and PDZ domain of HtrA4 model. Helices and β-strands were numbered sequentially starting from the N-terminus towards the C-terminal end of the model. SPD contains 12 β-strands (β1–β12) connected by evolutionarily conserved loops which were named according to the chymotrypsin nomenclature [52]. Among these loops, LB (β3–β4), LC (β6–β7), and L1 (β9–β10) accommodate the catalytic triad residues H218, D248, and S326, respectively. These loops synergistically interact with LA (β1–β2), LD (β7–β8), and L3 (β8–β9) loops, which are found to be essential for proteolytic activity and its regulation for various serine proteases (Figure 5) [15,19].

In silico structural analysis of HtrA4 model.

Figure 5.
In silico structural analysis of HtrA4 model.

(A) PSIPRED results of HtrA4 model showing secondary structures in ribbon representation; helices (pink), beta strands (yellow), and loops (black). Numbering of the secondary structures is done sequentially starting from the N-terminal. Catalytic triad residues are highlighted by green-colored boxes. (B) Cartoon representation of HtrA4 model generated by I-TASSER. Important regions such as N-terminal, SPD, linker, YLGL motif and PDZ are labeled. SBP is marked by a mesh surface.

Figure 5.
In silico structural analysis of HtrA4 model.

(A) PSIPRED results of HtrA4 model showing secondary structures in ribbon representation; helices (pink), beta strands (yellow), and loops (black). Numbering of the secondary structures is done sequentially starting from the N-terminal. Catalytic triad residues are highlighted by green-colored boxes. (B) Cartoon representation of HtrA4 model generated by I-TASSER. Important regions such as N-terminal, SPD, linker, YLGL motif and PDZ are labeled. SBP is marked by a mesh surface.

Prior to docking analysis of HtrA4 with β-casein, the binding pocket was identified using SiteMap. The best binding pocket (named as SBP_HtrA4) was selected on the basis of site score that is calculated using parameters such as effective volume, solvent-accessible surface area (SASA), and nature of residues in terms of the donor to acceptor ratio. SBP_HtrA4 has the highest number of available hydrogen donor and acceptor groups that are important for interacting with the peptides. It is also large enough to accommodate the heptameric peptides generated from the peptide library analysis. Among the top five resulted sites, SBP_HtrA4 was the only site to represent residues from both SPD and PDZ domains (Figure 5B and Supplementary Table S3). Moreover, with stronger hydrophobicity as compared with the other four sites, SBP_HtrA4 was considered the best interaction site and was chosen for further docking and MDS studies (Supplementary Figure S6).

Docking analysis reveals the important interacting residues in SBP_HtrA4

To find leads toward critical interacting residues, the affinity of the interaction and the allosteric changes associated upon peptide binding, we prepared peptide library of β-casein (library preparation and docking discussed in the Methods section) and docked them with HtrA4. Among 23 β-casein peptides present in the library, 199DMPIQAF205(β-casein peptide-199DMPIQAFLLYQEPVLGPVRGPFPIIV224) showed the highest binding affinity with a docking score of −10.126 kcal/mole. The negative control peptide LNQPKNNPKNN, as expected, did not dock at the SBP_HtrA4 site. Interaction analysis showed residues L245, D248, D309, L383, and S394 of HtrA4 forming hydrogen bonds, whereas A217, K308, F361, K379, V447, and D451 making van der Waals contacts with DMPIQAF peptide. A217 and L245 belong to the LA and LB loops, respectively, where the catalytic triad residues, H218, and D248 reside. Apart from that, L383 that is present in the conservative YLGL motif (corresponding to GLGF motif in the PDZ domain of HtrA family) formed hydrogen bond with the DMPIQAF peptide [41]. V447 and D451 that are located at the SPD–PDZ interface of SBP_HtrA4 also showed van der Waals interactions with the same peptide. Hence, the complex formed by β-casein (DMPIQAF) peptide after binding with HtrA4 was chosen for further analysis using MDS.

MDS analysis representing structural reorientations of modeled HtrA4 upon binding β-casein peptide

The RMSD plot generated from MDS analysis of HtrA4–β-casein complex (PMDB ID: PM0081893) showed fluctuation in the N-terminus, linker region and PDZ domain when compared with the unbound protease. Among these three regions, N-terminus showed the maximum deviation of 0.2 nm (2 Å) as compared with the unbound HtrA4. The linker region started fluctuating after 5 ns that got stabilized beyond 10 ns of simulation. The highest deviation for the linker region was 0.3 nm (3 Å) in a bound form that is 0.1 nm (1 Å) less than the same in the apo-form. Subtle changes in the overall RMSD value for PDZ domain was also observed upon binding the β-casein peptide (Figure 6A,B).

MDS analysis of HtrA4–β-casein complex.

Figure 6.
MDS analysis of HtrA4–β-casein complex.

RMSD plot for (A) unbound HtrA4 (B) bound HtrA4 showing the difference in fluctuations for important domains and catalytic triad residues. (C) Unbound HtrA4 (light-blue) and bound HtrA4 (blue-white) are aligned to show the movement of the loop and helix regions. Important fluctuating regions are marked as maroon for unbound and yellow for bound. Location of each catalytic triad residue is indicated by an asterisk.

Figure 6.
MDS analysis of HtrA4–β-casein complex.

RMSD plot for (A) unbound HtrA4 (B) bound HtrA4 showing the difference in fluctuations for important domains and catalytic triad residues. (C) Unbound HtrA4 (light-blue) and bound HtrA4 (blue-white) are aligned to show the movement of the loop and helix regions. Important fluctuating regions are marked as maroon for unbound and yellow for bound. Location of each catalytic triad residue is indicated by an asterisk.

Movement in the H3 helix was observed upon H-bond interaction between S394 in H3 with the β-casein peptide (Figure 6C). In addition, L383 of 382YLGL385 motif in PDZ domain displayed interaction with the peptide resulting in deviation of both the linker region and 382YLGL385 motif. The 382YLGL385 motif moved away to accommodate the heptameric β-casein peptide. Moreover, loop movement in the SPD domain contributed to the reorientation of the catalytic triad, where LC and L1 loops came closer to each other bringing H218 and S326 residues in close proximity. Moreover, the LC loop moved away from L4 causing an increment in the distance between H218 and D248 (Figure 6C). These arrangements when compared with other serine proteases have been found to be conducive for substrate binding and catalysis. For example, similar movements are observed in L1, L2, and LD loops of HtrA2 when an activator peptide binds to it, resulting in reorientation of the catalytic triad residues [28].

Reorientation of the catalytic triad residues was also observed in HtrA4–β-casein complex (PMDB ID: PM0081893) along with changes in the oxyanion hole residues. Distance analysis showed that in unbound HtrA4 model (PMDB ID: PM0081859), the distance between Nε2 atom of H218 and Oγ atom of S326 is 8.5 Å and between Nδ1 atom of H218 and Oδ1 atom of D248 is 5.6 Å. Upon binding β-casein peptide, the distance between H218 and S326 decreased to 7.5 Å, whereas there is an increase in the distance to 6.2 Å between H218 and D248 (Figure 7A). The overall angular distance between Nδ1 atom of H218, Oδ1 atom of D248 and Oγ atom S326 is reduced to 56.4° from 59.3° upon binding of a β-casein substrate (Figure 7B). Moreover, side-chain movements in the oxyanion hole residues (Y323, G324, and N325) near S326 are observed in the bound HtrA4 model (PMDB ID: PM0081893). The most notable mention among them is Y323, which flipped by a rotating angle of 70° to stabilize the β-casein bound HtrA4 complex (Figure 7C).

Structural comparison of catalytic triad and oxyanion hole residues.

Figure 7.
Structural comparison of catalytic triad and oxyanion hole residues.

Comparison of the (A) linear distances and (B) angular distances between unbound (blue) and bound (pink) catalytic triad residues (Ser 326, Asp 248, and His 218) of HtrA4. (C) Oxyanion hole orientation in peptide bound HtrA4 (pink) and unbound HtrA4 (blue) forms. Oxyanion hole residues namely, N325, G324, and Y323 are labeled as −1, −2, and −3, respectively, where S326 is considered as 0.

Figure 7.
Structural comparison of catalytic triad and oxyanion hole residues.

Comparison of the (A) linear distances and (B) angular distances between unbound (blue) and bound (pink) catalytic triad residues (Ser 326, Asp 248, and His 218) of HtrA4. (C) Oxyanion hole orientation in peptide bound HtrA4 (pink) and unbound HtrA4 (blue) forms. Oxyanion hole residues namely, N325, G324, and Y323 are labeled as −1, −2, and −3, respectively, where S326 is considered as 0.

Discussion

HtrA4 shows differential expression in glioblastoma multiforme, breast carcinoma, and metastatic prostate cancer, which makes it an emerging therapeutic target [12,21]. Moreover, it is up-regulated in PE and hence has been suggested to be used as a biomarker of PE [14]. In this study, we have for the first time performed biophysical and biochemical characterization of HtrA4 using a multidisciplinary approach including protein engineering, molecular docking, spectroscopy, and functional enzymology tools.

Pairwise sequence alignment in quest of conservative motifs in HtrA4 resulted in identifying an N-terminal 144AVPV moiety that is similar to the 133AVPS motif of HtrA2. Based on these observations, we cloned and bacterially expressed HtrA4 starting from144AVPV residues. The protein was found to cleave β-casein in the gel-based assay, which was further substantiated with quantitative enzyme kinetic and spectroscopic studies.

Prokaryotic HtrA family members have the ability of forming higher-order complexes consisting of up to 24 molecules, while human HtrAproteins (HtrA1–3) have been found to be primarily trimeric [19]. Size-exclusion chromatographic studies demonstrated that HtrA4 is primarily a trimeric protein with (∼20%) monomeric population. The presence of residual monomeric population might be due to lesser stability of the trimeric complex which has been substantiated by thermal stability studies. Surprisingly, unlike HtrA3, where PDZ is essential for oligomerization, the N-SPD construct showed primarily a trimeric ensemble implicating conformational flexibility in the full-length protease [20]. Moreover, our studies have also highlighted the role of the N-terminal region in trimerization, which is a pre-requisite for the formation of a functionally active protease as observed in other human counterparts [15,40].

Both gel-based and quantitative assays showed that HtrA4 is less active when compared with HtrA2. From the quantitative studies, the activity of HtrA2 has been found to be 3.6 × 105 fold higher than HtrA4. HtrA4 also shows allosteric behavior like HtrA2 although with lesser cooperativity (Hill coefficients for HtrA4 and HtrA2 are 1.3 and 2.8, respectively). This might be as a consequence of the difference in the nature of residues and size of the putative binding site as observed with in silico tools. Moreover, ligand docking studies in combination with MDS demonstrated different conformational plasticity upon substrate binding. Comparative analysis with HtrA2 shows that the distance between nitrogen (Nε2) atom of histidine and oxygen (Oγ) atom of serine is nearly twofold and distance between nitrogen (Nδ1) atom of histidine and oxygen (Oδ1) atom of aspartate is nearly threefold higher in unbound HtrA4 (Figure 7C) [28]. Apart from this, fluctuations (RMSD) of the main loop regions (L1, L4, LC, and linker) of bound HtrA4 also differ from those seen in the loop regions of bound HtrA2 [28]. An increase in distance between different pairs of catalytic triad residues might be one of the reasons for lower basal activity as observed in vitro. Moreover, temperature-based activity and stability studies show lesser stability of HtrA4 as compared with HtrA2 [38]. These observations indicate that the underlying conformational changes induced by both temperature and substrate might be different in both of these proteases. Therefore, subtle structural variations in the dynamic loop regions and around the active site along with lesser overall protease stability might be one of the primary reasons for lower protease activity of HtrA4. In addition, the enzyme kinetics study shows a strikingly lower cooperativity index (Hill coefficient 1.3 as opposed to 2.8 in HtrA2) with β-casein as a substrate. Similar cooperativity was also observed in bacterial periplasmic protease Deg S (∼1.6) that required an activator molecule (OMP) for cleavage of RseA protein [53]. This has also been reflected in lower basal activity of DegS (∼1000-fold less as compared with HtrA2) suggesting a complex regulatory process for its activation. Thus, low activity of HtrA4 also hints toward the involvement of a specific allosteric mechanism for precise control of HtrA4's protease activity in the cellular milieu. This distinctive mode of allosteric regulation offers promising opportunities to specifically modulate its proteolytic activity with desired characteristics.

Overall, this study for the first time presents an extensive biophysical and enzymatic characterization of serine protease HtrA4 and proposes a working model of its mode of action (Figure 8). We propose that HtrA4 exists in both trimeric and monomeric forms; trimeric one is the dominant and most favorable native oligomeric state. The addition of substrate enhances the conversion of the residual population of monomers to trimers. The model also suggests that initial binding of the β-casein peptide at SBP_HtrA4 leads to conformational rearrangement of the N-terminal region and loops at and around the catalytic triad. The significant role of the N-terminal region in fostering HtrA4's activity and maintaining its trimeric form is evident from the fact that both SPD-PDZ and SPD monomers, which are devoid of the N-terminal region are inactive. Hence, it further emphasizes the importance of trimerization to acquire a catalytically competent structure and indicates that the SPD alone is not sufficient for substrate catalysis. Moreover, our study is also the first one to show antiapoptotic XIAP as a binding partner cum substrate of HtrA4 in vitro. Thus, with the help of the comprehensive model proposed by our study, further identification of natural substrates of HtrA4 would provide a clue for devising strategies for therapeutic intervention against diseases it is associated with [49,54].

A proposed simplistic model for HtrA4 mechanism of action.

Figure 8.
A proposed simplistic model for HtrA4 mechanism of action.

The model illustrates the existence of HtrA4 primarily as a trimer with a small amount of monomeric population. The trimeric population is stabilized upon substrate binding. (A) In trimeric HtrA4, initial substrate binding or increase in temperature leads to conformational changes in regulatory loops and makes protease active and cleaves substrates into shorter fragments. (B) In trimeric N-SPD with partially exposed active site and no regulatory PDZ domain, activity is less. (C) Removal of N-terminal makes protease monomer, and hence no catalytic activity is observed. (D) Removal of both N-terminus and PDZ also generates monomeric protease, which lacks proteolytic activity.

Figure 8.
A proposed simplistic model for HtrA4 mechanism of action.

The model illustrates the existence of HtrA4 primarily as a trimer with a small amount of monomeric population. The trimeric population is stabilized upon substrate binding. (A) In trimeric HtrA4, initial substrate binding or increase in temperature leads to conformational changes in regulatory loops and makes protease active and cleaves substrates into shorter fragments. (B) In trimeric N-SPD with partially exposed active site and no regulatory PDZ domain, activity is less. (C) Removal of N-terminal makes protease monomer, and hence no catalytic activity is observed. (D) Removal of both N-terminus and PDZ also generates monomeric protease, which lacks proteolytic activity.

Database Depositions

Model data are available in the PMDB database under the accession numbers PM0081859 and PM0081893.

Abbreviations

     
  • Deg

    degradation of periplasmic proteins

  •  
  • FITC

    fluorescein isothiocyanate

  •  
  • GST

    glutathione S-transferase

  •  
  • HtrA

    high-temperature requirement protease A

  •  
  • IAP

    inhibitor of apoptosis proteins

  •  
  • IBM

    IAP-binding motif

  •  
  • IGFBP

    insulin-like growth factor-binding protein

  •  
  • IPTG

    isopropyl β-d-1-thiogalactopyranoside

  •  
  • I-TASSER

    iterative threading assembly refinement

  •  
  • MDS

    molecular dynamics simulations

  •  
  • ORF

    open reading frame

  •  
  • PDZ

    post-synaptic density protein, Drosophila disc large tumor suppressor, zonula occludens-1 protein

  •  
  • PMDB

    protein model database

  •  
  • PSIPRED

    PSI-BLAST-based secondary structure prediction

  •  
  • RMSD

    root mean square deviation

  •  
  • SBP

    selective binding pocket

  •  
  • SPD

    serine protease domain

  •  
  • XIAP

    X-linked inhibitor of apoptosis protein

Author Contribution

R.K. and K.B. conceived and designed the study. S.D. performed in silico studies and analyzed Figures 57. L.K.C. analyzed Figure 2C and Table 2. R.K. performed and analyzed all other experiments. With inputs from S.D. and L.K.C., K.B. and R.K. wrote the manuscript. All authors read and approved the manuscript.

Funding

This work was supported by institutional funding from ACTREC (Grant No: ACTREC - 42).

Acknowledgements

We thank Revathy Nair, Rashi Sharma, and Ankita Bhavsar for help with protein purification, as well as the members of Bose lab for helpful discussions. The authors also thank biophysics facility and BTIS, ACTREC for providing infrastructure for the necessary experiments.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Clausen
,
T.
,
Southan
,
C.
and
Ehrmann
,
M.
(
2002
)
The HtrA family of proteases: implications for protein composition and cell fate
.
Mol. Cell
10
,
443
455
2
Gray
,
C.W.
,
Ward
,
R.V.
,
Karran
,
E.
,
Turconi
,
S.
,
Rowles
,
A.
,
Viglienghi
,
D.
, et al.  (
2000
)
Characterization of human HtrA2, a novel serine protease involved in the mammalian cellular stress response
.
Eur. J. Biochem.
267
,
5699
5710
3
Nie
,
G.-Y.
,
Hampton
,
A.
,
Li
,
Y.
,
Findlay
,
J.K.
and
Salamonsen
,
L.A.
(
2003
)
Identification and cloning of two isoforms of human high-temperature requirement factor A3 (HtrA3), characterization of its genomic structure and comparison of its tissue distribution with HtrA1 and HtrA2
.
Biochem. J.
371
,
39
48
4
Pallen
,
M.J.
and
Wren
,
B.W.
(
1997
)
The HtrA family of serine proteases
.
Mol. Microbiol.
26
,
209
221
5
Runyon
,
S.T.
,
Zhang
,
Y.
,
Appleton
,
B.A.
,
Sazinsky
,
S.L.
,
Wu
,
P.
,
Pan
,
B.
et al.  (
2007
)
Structural and functional analysis of the PDZ domains of human HtrA1 and HtrA3.
Protein Sci.
16
,
2454
2471
6
Chaganti
,
L. K.
,
Singh
,
N.
and
Bose
,
K.
(
2015
) Cathepsins and HtrAs – multitasking proteases in programmed cell death. In
Proteases in Apoptosis: Pathways, Protocols and Translational Advances
(
Bose
,
K.
, ed.), pp.
95
141
,
Springer International Publishing
,
Cham
7
Jones
,
J.M.
,
Datta
,
P.
,
Srinivasula
,
S.M.
,
Ji
,
W.
,
Gupta
,
S.
,
Zhang
,
Z.
, et al.  (
2003
)
Loss of Omi mitochondrial protease activity causes the neuromuscular disorder of mnd2 mutant mice
.
Nature
425
,
721
727
8
Grau
,
S.
,
Baldi
,
A.
,
Bussani
,
R.
,
Tian
,
X.
,
Stefanescu
,
R.
,
Przybylski
,
M.
, et al.  (
2005
)
Implications of the serine protease HtrA1 in amyloid precursor protein processing
.
Proc. Natl Acad. Sci. U.S.A.
102
,
6021
6026
9
Grau
,
S.
,
Richards
,
P.J.
,
Kerr
,
B.
,
Hughes
,
C.
,
Caterson
,
B.
,
Williams
,
A.S.
et al.  (
2006
)
The role of human HtrA1 in arthritic disease
.
J. Biol. Chem.
281
,
6124
6129
10
Coleman
,
H.R.
,
Chan
,
C.-C.
,
Ferris
,
F.L.
and
Chew
,
E.Y.
(
2008
)
Age-related macular degeneration
.
Lancet
372
,
1835
1845
11
Milner
,
J.M.
,
Patel
,
A.
and
Rowan
,
A.D.
(
2008
)
Emerging roles of serine proteinases in tissue turnover in arthritis
.
Arthritis Rheum.
58
,
3644
3656
12
Chien
,
J.
,
Campioni
,
M.
,
Shridhar
,
V.
and
Baldi
,
A.
(
2009
)
HtrA serine proteases as potential therapeutic targets in cancer
.
Curr. Cancer Drug Targets
9
,
451
468
13
Dynon
,
K.
,
Heng
,
S.
,
Puryer
,
M.
,
Li
,
Y.
,
Walton
,
K.
,
Endo
,
Y.
et al.  (
2012
)
HtrA3 as an early marker for preeclampsia: specific monoclonal antibodies and sensitive high-throughput assays for serum screening
.
PLoS ONE
7
,
e45956
14
Inagaki
,
A.
,
Nishizawa
,
H.
,
Ota
,
S.
,
Suzuki
,
M.
,
Inuzuka
,
H.
,
Miyamura
,
H.
et al.  (
2012
)
Upregulation of HtrA4 in the placentas of patients with severe pre-eclampsia
.
Placenta
33
,
919
926
15
Singh
,
N.
,
Kuppili
,
R.R.
and
Bose
,
K.
(
2011
)
The structural basis of mode of activation and functional diversity: a case study with HtrA family of serine proteases
.
Arch. Biochem. Biophys.
516
,
85
96
16
Risør
,
M.W.
,
Poulsen
,
E.T.
,
Thomsen
,
L.R.
,
Dyrlund
,
T.F.
,
Nielsen
,
T.A.
,
Nielsen
,
N.C.
et al.  (
2014
)
The autolysis of human HtrA1 is governed by the redox state of its N-terminal domain
.
Biochemistry
53
,
3851
3857
17
Campioni
,
M.
,
Severino
,
A.
,
Manente
,
L.
,
Tuduce
,
I.L.
,
Toldo
,
S.
,
Caraglia
,
M.
, et al.  (
2010
)
The serine protease HtrA1 specifically interacts and degrades the tuberous sclerosis complex 2 protein
.
Mol. Cancer Res.
8
,
1248
1260
18
Singh
,
N.
,
D'Souza
,
A.
,
Cholleti
,
A.
,
Sastry
,
G.M.
and
Bose
,
K.
(
2014
)
Dual regulatory switch confers tighter control on HtrA2 proteolytic activity
.
FEBS J.
281
,
2456
2470
19
Zurawa-Janicka
,
D.
,
Wenta
,
T.
,
Jarzab
,
M.
,
Skorko-Glonek
,
J.
,
Glaza
,
P.
,
Gieldon
,
A.
et al.  (
2017
)
Structural insights into the activation mechanisms of human HtrA serine proteases
.
Arch. Biochem. Biophys.
621
,
6
23
20
Glaza
,
P.
,
Osipiuk
,
J.
,
Wenta
,
T.
,
Zurawa-Janicka
,
D.
,
Jarzab
,
M.
,
Lesner
,
A.
et al.  (
2015
)
Structural and functional analysis of human HtrA3 protease and its subdomains
.
PLoS ONE
10
,
e0131142
21
Zurawa-Janicka
,
D.
,
Skorko-Glonek
,
J.
and
Lipinska
,
B.
(
2010
)
HtrA proteins as targets in therapy of cancer and other diseases
.
Exp. Opin. Ther. Targets
14
,
665
679
22
Perez-Iratxeta
,
C.
and
Andrade-Navarro
,
M.A.
(
2008
)
K2d2: Estimation of protein secondary structure from circular dichroism spectra
.
BMC Struct. Biol.
8
,
25
23
Yang
,
J.
,
Yan
,
R.
,
Roy
,
A.
,
Xu
,
D.
,
Poisson
,
J.
and
Zhang
,
Y.
(
2014
)
The I-TASSER suite: protein structure and function prediction
.
Nat. Methods
12
,
7
8
24
Eigenbrot
,
C.
,
Ultsch
,
M.
,
Lipari
,
M.T.
,
Moran
,
P.
,
Lin
,
S.J.
,
Ganesan
,
R.
, et al.  (
2012
)
Structural and functional analysis of HtrA1 and its subdomains
.
Structure
20
,
1040
1050
25
Li
,
W.
,
Srinivasula
,
S.M.
,
Chai
,
J.
,
Li
,
P.
,
Wu
,
J.-W.
,
Zhang
,
Z.
et al.  (
2002
)
Structural insights into the pro-apoptotic function of mitochondrial serine protease HtrA2/Omi
.
Nat. Struct. Biol.
9
,
436
441
26
Sheik
,
S.S.
,
Sundararajan
,
P.
,
Hussain
,
A.S.Z.
and
Sekar
,
K.
(
2002
)
Ramachandran plot on the web
.
Bioinformatics
18
,
1548
1549
27
Fiser
,
A.
and
Sali
,
A.
(
2003
)
Modloop: automated modeling of loops in protein structures
.
Bioinformatics
19
,
2500
2501
28
Bejugam
,
P.R.
,
Kuppili
,
R.R.
,
Singh
,
N.
,
Gadewal
,
N.
,
Chaganti
,
L.K.
,
Sastry
,
G.M.
et al.  (
2013
)
Allosteric regulation of serine protease HtrA2 through novel non-canonical substrate binding pocket
.
PLoS ONE
8
,
e55416
29
Suenaga
,
A.
,
Okimoto
,
N.
,
Hirano
,
Y.
and
Fukui
,
K.
(
2012
)
An efficient computational method for calculating ligand binding affinities
.
PLoS ONE
7
,
e42846
30
Abraham
,
M.J.
,
Murtola
,
T.
,
Schulz
,
R.
,
Páll
,
S.
,
Smith
,
J.C.
,
Hess
,
B.
et al.  (
2015
)
GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers
.
SoftwareX
1
2
,
19
25
31
Aliev
,
A.E.
,
Kulke
,
M.
,
Khaneja
,
H.S.
,
Chudasama
,
V.
,
Sheppard
,
T.D.
and
Lanigan
,
R.M.
(
2014
)
Motional timescale predictions by molecular dynamics simulations: case study using proline and hydroxyproline sidechain dynamics
.
Proteins
82
,
195
215
32
Marquardt
,
D.W.
(
1963
)
An algorithm for least-squares estimation of nonlinear parameters
.
J. Soc. Ind. Appl. Math.
11
,
431
441
33
Kawata
,
M.
and
Nagashima
,
U.
(
2001
)
Particle mesh Ewald method for three-dimensional systems with two-dimensional periodicity
.
Chem. Phys. Lett.
340
,
165
172
34
McDonald
,
I.R.
(
1972
)
NpT-ensemble Monte Carlo calculations for binary liquid mixtures
.
Mol. Phys.
23
,
41
58
35
Hess
,
B.
,
Bekker
,
H.
,
Berendsen
,
H.J.C.
and
Fraaije
,
J.G.E.M.
(
1997
)
LINCS: a Linear Constraint Solver for molecular simulations
.
J. Comput. Chem.
18
,
1463
1472
36
Shao
,
J.
,
Tanner
,
S.W.
,
Thompson
,
N.
and
Cheatham
,
T.E.
(
2007
)
Clustering molecular dynamics trajectories: 1. Characterizing the performance of different clustering algorithms
.
J. Chem. Theory Comput.
3
,
2312
2334
37
Zurawa-Janicka
,
D.
,
Jarzab
,
M.
,
Polit
,
A.
,
Skorko-Glonek
,
J.
,
Lesner
,
A.
,
Gitlin
,
A.
, et al.  (
2013
)
Temperature-induced changes of HtrA2(Omi) protease activity and structure
.
Cell Stress Chaperones
18
,
35
51
38
Chaganti
,
L.K.
,
Kuppili
,
R.R.
and
Bose
,
K.
(
2013
)
Intricate structural coordination and domain plasticity regulate activity of serine protease HtrA2
.
FASEB J.
27
,
3054
3066
39
Lees
,
J.G.
,
Miles
,
A.J.
,
Wien
,
F.
and
Wallace
,
B.A.
(
2006
)
A reference database for circular dichroism spectroscopy covering fold and secondary structure space
.
Bioinformatics
22
,
1955
1962
40
Hansen
,
G.
and
Hilgenfeld
,
R.
(
2013
)
Architecture and regulation of HtrA-family proteins involved in protein quality control and stress response
.
Cell. Mol. Life Sci.
70
,
761
775
41
Wilken
,
C.
,
Kitzing
,
K.
,
Kurzbauer
,
R.
,
Ehrmann
,
M.
and
Clausen
,
T.
(
2004
)
Crystal structure of the DegS stress sensor: how a PDZ domain recognizes misfolded protein and activates a protease
.
Cell
117
,
483
494
42
Clausen
,
T.
,
Kaiser
,
M.
,
Huber
,
R.
and
Ehrmann
,
M.
(
2011
)
HTRA proteases: regulated proteolysis in protein quality control
.
Nat. Rev. Mol. Cell Biol.
12
,
152
162
43
Truebestein
,
L.
,
Tennstaedt
,
A.
,
Mönig
,
T.
,
Krojer
,
T.
,
Canellas
,
F.
,
Kaiser
,
M.
et al.  (
2011
)
Substrate-induced remodeling of the active site regulates human HTRA1 activity
.
Nat. Struct. Mol. Biol.
18
,
386
388
44
Vande Walle
,
L.
,
Van Damme
,
L.
,
Lamkanfi
,
P.
,
Saelens
,
M.
,
Vandekerckhove
,
X.
,
Gevaert
,
J.
et al.  (
2007
)
Proteome-wide identification of HtrA2/Omi substrates
.
J. Proteome Res.
6
,
1006
1015
45
Martins
,
L.M.
,
Turk
,
B.E.
,
Cowling
,
V.
,
Borg
,
A.
,
Jarrell
,
E.T.
,
Cantley
,
L.C.
et al.  (
2003
)
Binding specificity and regulation of the serine protease and PDZ domains of HtrA2/Omi
.
J. Biol. Chem.
278
,
49417
49427
46
He
,
X.
,
Khurana
,
A.
,
Maguire
,
J.L.
,
Chien
,
J.
and
Shridhar
,
V.
(
2012
)
HtrA1 sensitizes ovarian cancer cells to cisplatin-induced cytotoxicity by targeting XIAP for degradation
.
Int. J. Cancer
130
,
1029
1035
47
Suzuki
,
Y.
,
Imai
,
Y.
,
Nakayama
,
H.
,
Takahashi
,
K.
,
Takio
,
K.
and
Takahashi
,
R.
(
2001
)
A serine protease, HtrA2, is released from the mitochondria and interacts with XIAP, inducing cell death
.
Mol. Cell
8
,
613
621
48
Wenta
,
T.
,
Glaza
,
P.
,
Jarząb
,
M.
,
Zarzecka
,
U.
,
Żurawa-Janicka
,
D.
,
Lesner
,
A.
et al.  (
2017
)
The role of the LB structural loop and its interactions with the PDZ domain of the human HtrA3 protease
.
Biochim. Biophys. Acta Proteins Proteom.
1865
,
1141
1151
49
Salvesen
,
G.S.
and
Duckett
,
C.S.
(
2002
)
Apoptosis: IAP proteins: blocking the road to death's door
.
Nat. Rev. Mol. Cell Biol.
3
,
401
410
50
Duckett
,
C.S.
,
Li
,
F.
,
Wang
,
Y.
,
Tomaselli
,
K.J.
,
Thompson
,
C.B.
and
Armstrong
,
R.C.
(
1998
)
Human IAP-like protein regulates programmed cell death downstream of Bcl-xL and cytochrome c
.
Mol. Cell. Biol.
18
,
608
615
PMID:
[PubMed]
51
McGuffin
,
L.J.
,
Bryson
,
K.
and
Jones
,
D.T.
(
2000
)
The PSIPRED protein structure prediction server
.
Bioinformatics
16
,
404
405
52
Perona
,
J.J.
and
Craik
,
C.S.
(
1997
)
Evolutionary divergence of substrate specificity within the chymotrypsin-like serine protease fold
.
J. Biol. Chem.
272
,
29987
29990
53
Sohn
,
J.
and
Sauer
,
R.T.
(
2009
)
OMP peptides modulate the activity of DegS protease by differential binding to active and inactive conformations
.
Mol. Cell
33
,
64
74
54
Cilenti
,
L.
,
Kyriazis
,
G.A.
,
Soundarapandian
,
M.M.
,
Stratico
,
V.
,
Yerkes
,
A.
,
Park
,
K.M.
et al.  (
2005
)
Omi/HtrA2 protease mediates cisplatin-induced cell death in renal cells
.
Am. J. Physiol. Renal Physiol.
288
,
F371
F379

Author notes

*

The author is deceased and is contributing posthumously.