Introduction

Glucagon-like peptide-1 (GLP-1) is an incretin secreted by intestinal L-cells in response to food ingestion. Its biological effects include glucose-dependent enhancement of insulin secretion, inhibition of glucagon secretion, slowing of gastric emptying and reduction of food intake (Arulmozhi and Portha 2006). However, GLP-1 action has a short half life due to its degradation by dipeptidyl peptidase-IV (DPP-IV). DPP-IV is a homodimeric glycoprotein with an Mr of 110–150 kDa per subunit and is expressed in many tissues (Ludwig et al. 2003). Inhibition of DPP-IV prevents GLP-1 degradation in vivo, stimulates insulin secretion and improves glucose tolerance (Mest and Mentlein 2005). Several DPP-IV inhibitors, such as Sitagliptin (Merck Co.) and Vildagliptin (Novartis Co.), are effective in the treatment of Type 2 diabetes.

Previous methods for screening DPP-IV inhibitors utilized DPP-IV isolated from pig liver or kidney, or Caco-2 cell lysates (Kim et al. 2005b). However, the identity of amino acid sequence of DPP-IV between human and pig or mouse is 85% (Misumi et al. 1992). Moreover, the 3D structure of native DPP-IV is a key factor in regulating the interaction with other components (Engel et al. 2003; Thoma et al. 2003). Therefore, the differences between DPP-IVs from human and other species might influence the specificity to their inhibitors. On the other hand, although human colon cancer cell line Caco-2 has been used to screen DPP-IV inhibitors, the components in crude cell lysates may greatly influence its specificity and it is not economical to culture Caco-2 cells for large-scale screening. In this report, we developed a new strategy by cloning and expressing human DPP-IV in Pichia pastoris. Partially purified recombinant human DPP-IV was then applied in a new screening protocol to search for inhibitors. By high throughput screening, a novel compound (W5188) was identified.

Materials and methods

Cloning, expression and purification of recombinant human DPP-IV (rhDPP-IV)

Human colon cancer cell line Caco-2 (from the Peking Union Medical College, Beijing, China) were cultured in MEM medium containing 1% (v/v) non-essential amino acids (Gibco) and 10% (v/v) FBS at 37°C and in a humidified 5% CO2 environment for 12 days to induce differentiation of the cells (Baricault et al. 1995). DPP-IV cDNA was obtained from total RNA of differentiated Caco-2 cells with ThermoScript RT-PCR Kit (Invitrogen Co.). PCR primers were: FW1 (5′-GGAATTCATGAAGACACCGTGGAAG-3′) with an EcoRI site (bold letters); RV1(5′-GGCGGCCGCTCAGTGGTGGTGGTGGTGGTGAGGTAAAGAGAAACATTG-3′) with a NotI site (bold letters) and 6 × his tag (italics). Then the cDNA was subcloned into pPIC9 K plasmid (Invitrogen Co.) resulting in the recombinant plasmid pPIC9 K-rdpp (Fig. 1).

Fig. 1
figure 1

Schematic diagram of recombinant plasmid pPIC9K-rdpp construction. The recombinant plasmid contains 5′-AOX1 promoter region, DPP-IV coding region, 6×his-tag, 3′-AOX1 transcription termination region (TT), kanamycin and ampicillin resistant gene, etc

Full size image

The recombinant plasmid pPIC9 K-rdpp was transformed into Pichia pastoris GS115 (Invitrogen Co.) by electroporation. The original pPIC9 K plasmid was also transformed resulting in mock transformant as control. Twenty positive transformants with phenotype Mut+ were screened for the expression of rhDPP-IV. Transformants were grown at 30°C in YPD liquid medium (1% yeast extract, 2% peptone and 2% glucose) for 24 h, then in YPM liquid medium (1% yeast extract, 2% peptone and 2% methanol) in baffled shake-flasks at 30°C and 280 rpm for 5 days. Methanol at 2% (v/v) was added to YPM medium every day. The culture filtrate was centrifuged at 15,000g for 20–30 min. The supernatant was freeze-dried and dialyzed in 50 mM Tris/HCl (pH 8.0). After (NH4)2SO4 precipitation, the 40–80% fraction was purified on gel chromatography (Sephacryl S-200 HR column) eluting with 50 mM Tris/HCl (pH 8.0) containing 100 mM NaCl. The fractions of first protein peak were collected and concentrated as rhDPP-IV stock solution.

The rhDPP-IV was verified by 9% (v/v) SDS-PAGE and Western blot analysis (the first antibody was anti-his-tag monoclonal antibody (Novagen Co.).

Determination of rhDPP-IV enzymatic activity

DPP-IV standard protein (isolated from pig kidney) and its substrate Gly-Pro-p-nitroanilide hydrochloride were from Sigma. Each well contained 150 μl substrate (0.26 mM, in 100 mM Tris/HCl, pH 8.0, buffer) and 50 μl rhDPP-IV stock solution (50 mU ml−) in a 96-well plate. The plate was incubated at 37°C for 60 min. The rhDPP-IV activity was defined as the increase in absorbancy between 0 and 60 min (at 405 nm).

Activities of standard DPP-IV dilutions with a series of concentrations were also measured and plotted as a standard curve (activity/concentration). The rhDPP-IV concentration was determined from the standard curve.

Validation of screening method in 96-well plate

(S)-1-Cyclohexyl-2-oxo-2-(thiazolidin-3-yl)-ethylamine, INDP-2 (a homologue of Sitagliptin, see Table 1) was used as reference compound. Its synthetic route was provided by Dr. Jinyou Xu, Merck Co., USA. Each well contained 125 μl substrate (0.26 mM), 50 μl rhDPP-IV stock solution (50 mU ml−1), and 25 μl INDP-2 solution. INDP-2 was from 8 nM to 8.4 μM. A concentration-activity curve was obtained for the calculation of its IC50.

Table 1 Inhibitory effects of compounds against rhDPP-IV
Full size table

Calculation of Z′ factor and S/N ratio

The reaction rate of DPP-IV at 405 nm was linear at 37°C with up to 0.3 mM product for up to 120 min (Nagatsu et al. 1976). To assess the robustness of this screening method, the parameter of Z′ factor was used:

$$ {Z}^{\prime } = {1} - \left[ {{{\left( {{3}\sigma_{\text{c}} + {3}\sigma_{{{\text{c}} - }} } \right)} \mathord{\left/ {\vphantom {{\left( {{3}\sigma_{\text{c}} + {3}\sigma_{{{\text{c}} - }} } \right)} {\left( {\mu_{\text{c + }} - \mu_{{{\text{c}} - }} } \right)}}} \right. \kern-\nulldelimiterspace} {\left( {\mu_{\text{c+}} - \mu_{{{\text{c}}-}} } \right)}}} \right] $$

wherein, σ is the standard deviation of the data, μ is the mean of a sample and the subscripts “c+” and “c−” mean positive and negative controls respectively (Zhang et al. 1999). Data of 80 positive control wells (INDP-2, 100 μg ml−1) and 80 negative control wells (100 mM Tris/HCl buffer without test compound) were used to calculate the Z′ factor.

The signal/noise ratio is defined as: S/N = (mean value of negative controls – mean value of backgrounds)/mean value of backgrounds. All the data were expressed as mean ± SD and statistical analysis was performed by Microsoft Excel software.

Results

Production and validation of rhDPP-IV

Human DPP-IV cloned from Caco-2 cells was 2.3 kb and with 100% homogeneity to the reported sequence (GenBank BC065265). Compared with the culture filtrate of mock transformant, rhDPP-IV was identified as a 110-120 kDa protein by SDS-PAGE and Western blot (Fig. 2). According to concentration-activity curve of pig DPP-IV, the rhDPP-IV stock solution was made at 50 mU ml−1. The yield of rhDPP-IV was calculated to be about 1.2 mg l−1 of growth medium.

Fig. 2
figure 2

Silver-stained SDS-PAGE (left) and Western blot (right) of culture filtrates of transformants. 9k: mock transformant. rdpp: rhDPP-IV transformant. M: MW marker. The band was identified around 110–120 kDa (arrow)

Full size image

Establishment of screening system

After studying the influence of substrate concentration and enzyme amounts on enzymatic activity (Figs. 3 and 4), we establish the screening system in 96-well plates and 384-well plates.

Fig. 3
figure 3

Influence of the substrate concentration on rhDPP-IV activity. Each well in a 96-well plate contains 150 μl substrate solution and 50 μl rhDPP-IV stock solution (50 mU ml−1). The substrate concentrations were: (a) 0.26 mM (symbol ×), (b) 0.13 mM (symbol ▲), (c) 0.065 mM (symbol ■), and (d) 0.0325 mM (symbol ♦)

Full size image
Fig. 4
figure 4

Influence of the rhDPP-IV concentration on rhDPP-IV activity. The total volume was 200 μl and the substrate concentration was 0.26 mM. The rhDPP-IV amount is 25 μl (symbol ▲), 50 μl (symbol ■) and 100 μl (symbol ♦) (corresponding to 6.25, 12.5 and 25 mU ml−1, respectively)

Full size image

Each test well in 96-well plate (total volume of 200 μl) contained: 50 μl rhDPP-IV (50 mU ml−1); 125 μl Gly-Pro-p-NA hydrochloride (0.26 mM); 25 μl test compound (0.1 mM ~ 0.1 nM). Reagents used in 384-well plate (total volume of 40 μl) were decreased proportionally.

Evaluation of the assay reproducibility

Z′ factor and S/N ratio were used to evaluate the reproducibility. The mean values of △ A over 60 min of positive control and negative control were 0 ± 0.01 and 0.383 ± 0.025 (Fig. 5). Therefore, the Z′ factor for this assay was calculated to be 0.73. Z′ > 0.5 indicates that the assay is applicable, and the closer the Z′ factor is to 1, the higher its reliability is (Zhang et al. 1999). The S/N ratio of HTS screening was calculated as 6.89. In general, a screening method is reliable when the S/N ratio is more than 5. Therefore, Z′ factor and S/N ratio both showed the stability and reliability of our method.

Fig. 5
figure 5

Z′ factor assay for 384-well method. 80 positive control wells (c+, INDP-2 compound solution (100 μg ml−1), symbol ■) and 80 negative control wells (c−, 100 mM Tris/HCl buffer with no test compound, symbol ♦) were tested

Full size image

New discovery of DPP-IV inhibitors screening

In high throughput screening, all addition were performed robotically. The test compound was at 100 μg ml−1. Plates were incubated at 37°C for 60 min. Each plate contains one positive control well (INDP-2, 100 μg ml−1) and one negative control well (100 mM Tris/HCl buffer). The inhibition percentage was calculated according to the formula below. Only compounds with inhibition percentage more than 50% were selected as candidates.

$$\hbox{Inhibition} (\%)=[(\Updelta \hbox{A}_{405} \hbox{ of negative control} -\Updelta \hbox{A}_{405} \hbox{ of test compound})/ \Updelta \hbox{A}_{405} \hbox{ of negative control}]\times 100\%.$$

A total of 75,000 compounds in our chemical library were screened with HTS. Two compounds (W5186 and W5188, see Table 1) were confirmed as active inhibitors. Compound W5188 was more effective than W5186 and was selected for further study. W5188 inhibited DPP-IV in concentration dependent manner (Fig. 6). The IC50 values of W5188 and INDP-2 were 6.5 and 0.46 μM, respectively (Table 1).

Fig. 6
figure 6

Dose-dependent inhibitory effects of W5188. Symbol ♦ represents the inhibition percentage (n = 3 of separate experiments). Each well contains 25 μl substrate (0.26 mM), 10 μl rhDPP-IV stock solution (50 mU ml−1) and 5 μl W5188 compound at 12.5, 6.25, 3.125, 1.56, 0.78, 0.39, 0.195 and 0.098 μg ml−1

Full size image

Discussion

Recently, several in vitro methods based on DPP-IV from pig, rat or mouse tissues or cultured human cells (e.g. from Caco-2 cells) have been used to identify DPP-IV inhibitors (Kim et al. 2005a, b). However, inhibitors of non-human DPP-IV might lack specificity when used against human targets. On the other hand, protocols using raw human cell lysate can have reduced sensitivity and S/N ratio due to disturbance of other unknown substances in the extract. In our new strategy, we solved these problems by expressing human DPP-IV gene in P. pastoris and purifying secreted rhDPP-IV protein from the media. The rhDPP-IV leads to higher S/N ratio of 6.89. Furthermore, large-scale screening demands a sufficient supply of DPP-IV protein. Our protocol is therefore more economical, sensitive and convenient than the protocols based on DPP-IV from non-human tissues (Fukasawa et al. 1978, 1981; Kyouden et al. 1992) or even crude human cell lysates.

So far, there are three classes of DPP-IV inhibitors: reversible product analogue inhibitors (e.g. pyrrolidines and thiazolidines), covalently-modified product analogues (e.g. cyanopyrrolidines) and reversible non-peptidic heterocyclic inhibitors (e.g. xanthines and aminomethylpyrimidines). Some of them have a five- or six-member ring derived from proline. Though compound W5188 (see Table 1) had a five-member ring structure, our findings suggested a novel class of rhDPP-IV inhibitors (such as W5188) which have a completely different heteroatom (P, O and N) in the ring structure as compared with other inhibitors. This new class of inhibitors has not been reported previously and is presumed to be a covalently modified product analogue. Moreover, the selectivity of DPP-IV inhibitors against other isoenzymes appears to be essential for the development of safe agents. So the structure-activity relationship research of its derivatives and study of their selectivity to DPP-IV isoenzymes are now under progress.