Mammalian high temperature requirement A3 (HtrA3) is a serine protease of the HtrA family. It is an important factor for placental development and a tumor suppressor. The biochemical properties of HtrA3 are uncharacterized. One critical step in biochemical characterization is overexpressing and purifying the full-length recombinant protein. However, utility of cell-based expression systems is limited for a protease because of autocleavage. The wheat-germ cell-free translation system is highly efficient at producing “difficult” eukaryotic multidomain proteins and is easily modifiable for protein synthesis at different temperatures. In this study, we evaluated the potential of the wheat-germ cell-free translation system for producing human HtrA3. HtrA3 underwent autocleavage when synthesized at 17°C. When the synthesis temperature was lowered to 4°C, full-length HtrA3 was successfully produced and proteolytically active. Catalytic site serine substitution with alanine (S305A) stabilized HtrA3 while abolishing its protease activity. This mutant was readily synthesized and stable at 17°C. When used with glutathione S-transferase (GST) pull-down assay, S305A HtrA3 was a valuable bait in searching for endogenous HtrA3 binding proteins. Thus, we demonstrated the unique utility of the wheat-germ cell-free translation system for producing and characterizing human HtrA3. These strategies will be likely applicable to a wide range of proteases.

The high temperature requirement A (HtrA) proteases are a family of serine proteases identified in organisms ranging from bacteria to mammals (1). They are typically characterized by the presence of a serine protease domain and one or two carboxy (C)-terminal PDZ (postsynaptic density of 95 kDa, discs large, and zonula occludens) domains (2). To date, four mammalian HtrAs are identified in the genome (2). The first three members (HtrA1, HtrA2/Omi, HtrA3) have been cloned, and their expression patterns reported (3); the fourth one (HtrA4) is known only by the cDNA sequence deposited in GenBank (2).

While bacterial HtrAs act as proteases and chaperones to participate in key aspects of protein quality control (1), the mammalian HtrAs have evolved to exert diverse functions, including cell proliferation, migration, and apoptosis, and their altered expression is associated with severe diseases, including cancer, arthritis, neurodegenerative and neuromuscular disorders, and age-related macular degeneration (1,3). This highlights the need to specifically characterize each mammalian HtrA.

HtrA3 was initially identified in the developing placenta both in the mouse and human as a serine protease associated with pregnancy (4-7). HtrA3 is now known to negatively regulate trophoblast invasion during placental development (8,9), and abnormal levels of HtrA3 during early pregnancy in women are associated with risks of developing preeclampsia (a severe pregnancy-specific disorder) (10,11). HtrA3 is also down-regulated in a number of cancers (ovary, uterus, lung) (12-15) and promotes etoposide- and cisplatin-induced cytotoxicity in lung cancer cell lines (15). HtrA3 is thus proposed to be a tumor suppressor and a potential therapeutic target in cancer treatment (3,15). It has been suggested to inhibit the transforming growth factor-β (TGF-β) signaling (16).

However, to date, the biochemical properties of HtrA3 are largely unknown. Unlike other HtrAs, HtrA3 has two isoforms [long (HtrA3-L) and short (HtrA3-S)] resulting from alternative splicing (4,5). Both isoforms contain a signature serine protease domain following an N-terminal insulin-like growth factor binding domain and a Kazal protease-inhibitor domain. They differ only by the presence in HtrA3-L and absence in HtrA3-S of a C-terminal PDZ domain (4,5). HtrA3-S thus presents a unique naturally occurring HtrA lacking the C-terminal PDZ domain (2). While HtrA3-L is predominantly expressed in the mouse, both HtrA3-L and HtrA3-S are expressed in human tissues especially in the placenta (5,6). It is unknown whether HtrA3-L and HtrA3-S are biochemically distinct.

One critical step in biochemical characterization of a protein is to overexpress and purify the full-length recombinant protein. Cell-based expression systems present challenges for protease expression because of autocleavage of the product. Indeed multiple bands were obtained when human HtrA1 was expressed in either insect or mammalian cells (17), while no distinct bands were detected for HtrA2 expressed in either Escherichia coli or insect cells due to autodegradation during overexpression (18,19). Thus the truncated forms of HtrA1 and HtrA2 are often used for functional analysis (19-22).

In recent years, significant advances have been made in the area of cell-free in vitro transcription/translation systems to express “difficult” proteins (23). One such approach is the wheat-germ cell-free translation system, which is highly efficient in producing eukaryotic multidomain proteins in a folded state for functional characterization (23-25).

In this study, we evaluated the potential of the wheat-germ cell-free translation system for producing human HtrA3-L and HtrA3-S in both wild-type (WT) and mutant (MT) forms and explored their potential for further characterizing human HtrA3.

Materials and methods

Production of human HtrA3 protein and its MT using the wheat-germ cell-free translation system

The long (HtrA3-L) and short (HtrA3-S) isoforms of human HtrA3 were produced using wheat-germ cell-free technology (23,24,26). In brief, the open reading frames of the human HTRA3-L and HTRA3-S cDNA (4) were cloned into the pEU-E01-GST expression vector (27) (CellFree Sciences, Matsuyama, Japan) using standard methods. The engineered constructs for producing N-terminal glutathione S-transferase (GST)-tagged HtrA3 were transformed into DH5α cells (Invitrogen, Mulgrave, Vic, Australia), and the plasmids were purified using Qiagen Plasmid Maxi kit (Qiagen, Doncaster, Vic, Australia).

Transcription and translation were performed using a CFS-TRI-1240G kit (CellFree Sciences) as per the manufacturer's instructions. The transcription was carried out at 37°C for 6 h with 250 µL containing 100 ng/µL plasmid, 1 U/µL SP6 RNA polymerase, 1 U/µL RNase inhibitor, 2.5 mM NTPs, and 1×transcription buffer. To test different synthesis conditions, small-scale translation was carried out in standard flat-bottomed 96-well plates as a bilayer reaction. Briefly, 206 µL 1×SUB-AMIX containing all 20 amino acids (300 µM each) in a proprietary buffer were transferred into the well, and the transcription mixture (10 µL transcription product, 0.8 µL 1 mg/mL creatine kinase, 10 µL WEPRO1240G) and 2 µL ¹⁴C-Leu (50 µCi) were then carefully pipeted into the bottom of the same well to form a bilayer. The plate was sealed with Parafilm and incubated at 4°C or 17°C for 16 h. The resultant translation products (6 µL) were analyzed by denaturing SDS-PAGE, stained with Coomassie Blue, and scanned with a phosphorimaging system (Storm Molecular Imager, Molecular Dynamics, Sunnyvale, CA, USA). To produce MT HtrA3 proteins, the catalytic site serine (S305) was mutated into alanine by site-directed mutagenesis with a PCR overlap extension strategy (28). The MT constructs were confirmed by sequencing, and the corresponding proteins were produced as for the WT counterparts.

To purify HtrA3, the translation was scaled up as per the manufacturer's instructions using 6-well plates. Briefly, 5.5 mL 1× SUB-AMIX was transferred into the well, the transcription mixture (250 µL transcription product, 1 µL 20 mg/mL creatine kinase, and 250 µL WEPRO1240G) was then pipeted into the bottom of the well, and the plate was incubated at 4°C (WT) or 17°C (MT) for 16 h. The synthesized proteins were purified at 4°C (WT) or room temperature (MT) using Glutathione Sepharose 4B gel (GE Healthcare Bio-Sciences, Rydalmere, NSW, Australia) and analyzed by SDS-PAGE to confirm purity.

In vitro protease activity assay

Protease activities of both HtrA3-L and HtrA3-S and their S305A MT forms were detected using casein labeled with fluorescein isothiocyanate (FITC) as a generic substrate as published (29) with minor modifications. In brief, the reaction contained 10 µL HTRA3-L or HTRA3-S (WT or MT; 500 pg), 20 µL assay buffer (20 mM sodium phosphate buffer with 150 mM sodium chloride, pH 7.6, at 37°C), and 20 µL 0.01% (w/v) FITC-casein (both from Sigma-Aldrich, St. Louis, MO, USA) in a microcentrifuge tube. After mixing, the solution was incubated at 37°C (0–120 min), acidified by gently adding 150 µL 0.6 N trichloroacetic acid solution (Sigma-Aldrich), and incubated for an additional 60 min at 37°C. The mixture was then centrifuged (1 min at 100× g), a 60-µL aliquot of the supernatant was neutralized to 400 µL with 500 mM Tris buffer (pH 8.5), and 200 µL of this was transferred to a 96-well plate for fluorescence measurement (490/525 nm, Wallac Victor2 spectrophotometer; PerkinElmer, Boston, MA, USA). In every assay, 0.001% (w/v) trypsin and ultrapure water were used as positive control and blank, respectively.

Using MT HtrA3 as bait in a GST pull-down assay to identify HtrA3 binding proteins

A GST pull-down assay (ProFound pull-down GST protein-protein interaction kit; Pierce, Rockford, IL, USA) was modified to include the MT HtrA3 as bait to search for endogenous HtrA3 binding proteins. In brief, MT HtrA3-L (containing an N-terminal GST-tag, ^GSTHtrA3-L-MT) or GST alone (negative control) were immobilized to glutathione affinity resin as per the manufacturer's instructions and served as the bait. Prey protein lysates were from a day (d) 10.5 mouse placenta (which expresses high levels of HtrA3; approval was obtained from the Animal Ethics Committee at Monash Medical Centre, Melbourne, Australia). The bait and prey were incubated at 4°C for 16 h with gentle rocking, the resulting mixture was washed five times, and the binding proteins were eluted with 100 mM glutathione elution buffer.

The eluted proteins were analyzed by SDS-PAGE, bands of interest were excised from the gel, and protein identifications were determined by mass spectrometry as previously described (30). For further validation, the elution products from ^GSTHtrA3-L-MT or GST resins were analyzed by Western blot analysis using specific antibodies (myosin-9 antibody, Santa Cruz Biotechnology, Santa Cruz, CA, USA; HtrA3 antibody as published, Reference 11). Coimmunoprecipitation (co-IP) was used to further validate the identified proteins as endogenous HtrA3 binding proteins. The candidate proteins were immunoprecipitated from the placenta, and the resulting IP products were analyzed by Western blot analysis for HtrA3 co-IP.

Results and discussion

Synthesis of full-length HtrA3 requires low temperature

We first attempted to synthesize human (HtrA3-L) and short (HtrA3-S) at 17°C using a standard protocol and including GFP as a control. While a single band corresponding to the expected size of GFP was detected, multiple bands of smaller than expected HtrA3 proteins were consistently observed (Figure 1A). This suggests that HtrA3-L and HtrA3-S were unstable at 17°C. Presentation of discrete bands for both HtrA3-L and HtrA3-S indicates autocleavage rather than random degradation, consistent with previously reported features of human HtrA1 (17).