A broad pH range indicator-based spectrophotometric assay for true lipases using tributyrin and tricaprylin.

A continuous assay is proposed for the screening of acidic, neutral, or alkaline lipases using microtiter plates, emulsified short- and medium-chain TGs, and a pH indicator. The lipase activity measurement is based on the decrease of the pH indicator optical density due to protonation which is caused by the release of FFAs during the hydrolysis of TGs and thus acidification. Purified lipases with distinct pH optima and an esterase were used to validate the method. The rate of lipolysis was found to be linear with time and proportional to the amount of enzyme added in each case. Specific activities measured with this microplate assay method were lower than those obtained by the pH-stat technique. Nevertheless, the pH-dependent profiles of enzymatic activity were similar with both assays. In addition, the substrate preference of each enzyme tested was not modified and this allowed discriminating lipase and esterase activities using tributyrin (low water solubility) and tricaprylin (not water soluble) as substrates. This continuous lipase assay is compatible with a high sample throughput and can be applied for the screening of lipases and lipase inhibitors from biological samples.

robust HTS method for screening lipase activities should cover a wide range of pHs to become universal.
In this work, six different pH indicators were used to continuously monitor the lipase activity on emulsified TG(4:0) and TG(8:0), at pH values ranging from 5.0 to 9.2. This pH indicator-based lipase assay (PHIBLA) was validated using acidic, neutral, and alkaline lipases and the lipase inhibitor tetrahydrolipstatin (THL).

Lipases
Recombinant human pancreatic lipase (rHPL) was expressed in Pichia pastoris and purifi ed from culture media as described by Belle et al. ( 33 ). Porcine pancreatic extract, also named pancreatin (P7545; 8× USP ), was purchased from Sigma-Aldrich. Porcine pancreatic lipase (PPL) was purifi ed according to Verger et al. ( 34 ). Porcine colipase was partly purifi ed from lipid-free pancreatic powder using the procedure described in Fernandez et al. ( 35 ). Rabbit gastric extract and purifi ed rabbit gastric lipase (RGL) were produced according to Moreau et al. ( 36 ). Pure recombinant dog gastric lipase (rDGL) was a generous gift of Meristem Therapeutics (Clermont-Ferrand, France). The purifi ed Thermomyces lanuginosus lipase (TLL) was a generous gift from Dr. S. Patkar (Novozymes, Denmark). LIP2 lipase from Yarrowia lipolytica (YLLIP2) was produced and purifi ed according to Aloulou et al. ( 37 ). Recombinant feruloyl esterase A (rAnFaeA) from Aspergillus niger was produced and purifi ed from culture media as described by Record et al. ( 38 ).

Lipase activity measurements using PHIBLA
Activities of rHPL, PPL, RGL, rDGL, TLL, YLLIP2, and rAnFaeA were determined according to a HTS method implemented in our laboratory, adapted from previous work ( 32 ), by measuring the apparition of FFAs upon hydrolysis of TG(4:0) or TG(8:0). Each samples. Furthermore, the catalytic turnover number of lipases on p -nitrophenyl esters ( p NPEs) is usually several orders of magnitude lower than that obtained with TGs ( 16 ). Some true lipases, like pancreatic lipase, do not display enzyme activity on these substrates, but nonspecifi c titration of some surface amino acid residues by p NPEs has often been confused with a true enzyme activity ( 17,18 ). This pseudo-esterase activity is in the same order of magnitude as the hydrolysis of p NPEs by serum albumin ( 19 ), and in that case, it was shown that the pseudo-esterase activity was the result of irreversible acylation of 82 residues and was not the result of turnover ( 20 ).
To replace nonspecifi c chromogenic substrates, continuous methods for measuring lipase activity with real lipase substrates were developed using fl uorescent long-chain TGs from Parinari glaberrimum ( 21 ) and UV-absorbing TGs from Aleurites frodii seeds or tung oil ( 22 ). TGs from Parinari glaberrimum are, however, very sensitive to oxidation. The TGs extracted from tung oil and used in the HTS method reported by Serveau et al. ( 22 ) are less sensitive to oxidation when they are coated on the surface of UV microtiter plate wells. Tung oil contains ␣ -eleostearic acid ( 23,24 ), which is a conjugated triene giving absorption in the UV. However, this method required special UV microtiter plates. TGs with fl uorescent pyrene acyl chains have been employed to measure lipase activity using a continuous and sensitive (moles of product per minute) assay ( 25 ), but these substrates are not real lipase substrates and are very expensive.
The short-chain tributyrin [TG(4:0)] substrate offers several advantages as a substrate for lipases compared with natural long-chain TGs. It is readily dispersed without the need for emulsifi ers like gum Arabic used with olive oil, the products formed on hydrolysis are water-soluble, and can be titrated directly in a large range of pHs. This is a major advantage for setting up continuous assays at various pH values, while the direct and continuous titration of longchain fatty acids can only be made at alkaline pH. Synthetic TG(4:0) substrate has thus been used in many studies of lipases ( 16,(26)(27)(28)(29)(30)(31), although it has no physiological relevance, because all known lipases are active on this substrate. However, due to its partial water solubility, it can be hydrolyzed by some esterases that are not active on insoluble TGs. The use of tricaprylin [TG(8:0)] as a totally insoluble medium-chain TG substrate is thus more appropriate to detect and assay a true lipase activity, as demonstrated with various microbial and mammalian lipases ( 28 ). Moreover, the production of the soluble caprylic acid confers advantages for direct titration compared with long-chain fatty acids.
In a previous work ( 32 ), a spectrophotometric HTS protocol for the rapid and reliable determination of lipase/esterase activity was validated using short-chain [TG(4:0)] and medium-chain [TG(8:0)] emulsifi ed TGs and a pH indicator. The principle of the method is the indirect quantifi cation of fatty acid released by lipase through protonation of a pH indicator, p -nitrophenol, and loss of its yellow coloration followed by reading absorbance at 410 nm. This method is restricted, however, to assays performed at pH 7.2 due to the chemical properties of p -nitrophenol, and some lipases, like acid lipases, could not be tested under these conditions . A TG(8:0), that are some of the most common substrates used to measure lipase activity, apart from natural longchain TGs. The principle of the microtiter plate lipase assay is shown in Fig. 1 . It consists of monitoring the progress of lipolysis, which releases a proton, through the use of a pH indicator which changes color as it is protonated.
The most important variables in using a pH indicator are the buffer and the ionization properties of the pH indicator ( 40 ). Both the buffer and the indicator must have the same affi nity for protons (pKa values within 0.1 units of each other) so that the relative amounts of protonated buffer and protonated indicator stay constant as the pH shifts during the reaction. Also, the pKa of the indicator has to be close to the pH of the reaction mixture to ensure that changes in pH give a large and linear color change ( 41 ). The choice of the pH indicator therefore determines what type of lipase activity can be tested (acid, neutral, or alkaline).
In order to develop a spectrophotometric method that allows measuring lipolytic activities over a broad pH range, six different indicator-buffer pairs were chosen here for measuring activity at pH 5.0, 5.5, 6.0, 7.2, 8.2, and 9.2 ( Table 1 ). The pKa of each indicator is similar to the pKa of the respective buffer used for the assay (difference <0.2 units of pKa). It is important to note that the rate of lipolysis in an acidic environment is diffi cult to measure using pH indicators, because the pKa values of the FFA released by the lipase (BtA and OcA) are close to 5 (pKa 4.82 and 4.89, respectively) and protons would be hardly detectable. Nevertheless, the use of short-and medium-chain TG substrates allows the assay to be performed at lower pH values (down to 5) than long-chain TG substrates containing fatty acids with much higher pKa values ranging from 6.5 to 9 ( 42 ). Moreover, long-chain FFA tends to remain adsorbed at the interface interacting with neighboring molecules, which have an important effect on pKa ( 43 ). ␤ -CD can be useful to desorb the fatty acids ( 44 ) and consequently to improve linearity of the assay. Using TG(4:0) and TG(8:0) provides the advantage of the formation of FFA soluble into water at the assay temperature. TG substrates with a longer chain, like tridodecanoin, were also tested, but the release C12:0 fatty acid was found to be crystallized in solution and thus could not be used for spectrophotometric assays (supplementary Fig. 1).
Another important variable in the pH indicator assay is the concentration ratio of indicator and buffer. Maximum sensitivity is obtained at maximum concentration of substrate was prepared as follows: one volume of the substrate (50 mM dissolved in tert -butanol), also containing the pH indicator, was mixed vigorously on a vortex with nine volumes of buffer solution to reach a fi nal substrate concentration of 5 mM that was found to be optimum for lipase activity measurements ( 32 ). Stock solutions (10×) of pH indicators and buffer solutions were prepared at the concentrations indicated in Table 1 . Buffer solution included (HPL, PPL, RGL, rDGL, TLL, YLLIP2, rAnFaeA) 150 mM NaCl, (HPL, PPL, RGL, rDGL, TLL, YLLIP2, rAnFaeA) 3 mg/ml ␤ -CD, (HPL, PPL, TLL, YLLIP2, rAnFaeA) 6 mM CaCl 2 , (HPL, TLL, rAn-FaeA) 0.5 mM NaTDC, (PPL, RGL, rDGL) 2 mM NaTDC, (YLLIP2) 4 mM NaTDC, (RGL, rDGL) 1.5 M BSA, (HPL, PPL) colipase at an excess of approximately 5 eq. Then, 20 l of enzyme solution at an appropriate dilution in buffer was added in each microplate well and 100 l of substrate emulsion was quickly added using an eightchannel pipette. Subsequently, the plate was placed in a microtiter plate scanning spectrophotometer (x-Mark TM , Bio-Rad) and shaken for 5 s before each reading. The decrease in absorbance at a wavelength corresponding to the max of the pH indicator ( Table 1 ) was recorded every 30 s at 37°C. Blanks without enzyme were performed and data were collected at least in triplicate for 15 min. Different standard curves of BtA or OcA were established for each condition. One unit corresponds to 1 mol of fatty acid released per minute in the assay conditions.

Study of lipase inhibition using PHIBLA
The enzyme-inhibitor preincubation method was used ( 39 ) to test whether any direct interactions might occur between the lipase and the inhibitor in aqueous medium and in the absence of substrate. Recombinant HPL (0.2 mg/ml in 25 mM MOPS buffer, pH 7.2) or RGL (0.2 mg/ml in 25 mM succinate buffer, pH 5.0) were preincubated at 25°C with THL (10 mM in DMSO) at an enzyme:inhibitor molar ratio of 1:100 (4% DMSO fi nal concentration) in the presence or absence of 4 mM NaTDC. The residual enzyme activity was then measured at various incubation times using the microtiter plate assay with TG(4:0) as substrate.
Alternatively, the inhibition-during-lipolysis method ( 39 ) was used to test whether any inhibition reaction occurred in the presence of substrate during the lipolysis reaction. Lipolysis of TG(4:0) was performed in a microtiter plate using rHPL (1.3 nM, fi nal concentration) or RGL (6.7 nM, fi nal concentration). THL (at an enzyme/inhibitor molar ratio of 1:100) was injected 5 min after the lipolysis reaction started. Lipase activity was continuously recorded.

Principle of the PHIBLA using emulsifi ed TGs in microtiter plates
We developed a continuous spectrophotometric lipase assay in microtiter plates using emulsifi ed TG(4:0) and Indicator concentrations were chosen as a function of their solubility and their responsiveness to pH changes (sensitivity).
To correlate FFA levels with absorbance, the wavelength that allows detection of the highest change in the color of each indicator was fi rst determined. Spectra of buffer solution containing indicator with and without OcA (deprotonated or protonated indicator, respectively) were done. Figure 2A shows an example of maximum absorption wavelength ( max ) determination of BCG solution in malate indicator and minimum concentration of buffer ( 41 ). The concentration of indicator is limited by its solubility, and the concentration of buffer should be as low as possible, consistent with the condition that the pH does not change enough to cause a change in kinetic constants.
Indicator and buffer concentrations are shown in Table 1 . Based on the pKa values of BtA and OcA, we chose a buffer concentration of 1.5 mM for assays at pH 5.0 and 5.5, and a buffer concentration of 2.5 mM for assays at pH 6 to 9.2.  dilution factor of the sample . One unit equals 1 mole of FFA per minute.

Substrate preparation and emulsion stability
Emulsion quality is the main factor affecting substrate availability and reproducibility of the method. Moreover a great and stable interfacial area is needed to obtain measurable kinetic reads. This can be achieved by adding a detergent ( 32 ). NaTDC, a bile salt that is widely used in standard tests of lipase activity (45)(46)(47), can be used for this purpose at a minimum concentration of 0.5 mM. This small amount of detergent allows for the formation of a TG(8:0) emulsion that is stable for at least 60 min with a mean particle size of 2-3 m [supplementary Fig. 2(3), (4)]. TG(4:0) emulsion (mean particle size about 25 m) was found to be less stable and had to be used during a time interval of 15 min after being prepared [supplementary Fig. 2(1), (2)].
On the other hand, insoluble TGs, like TG(8:0), produce turbid emulsions which may interfere with the spectroscopic measurements. An overall substrate concentration of 5 mM was selected for the assays. At this concentration, turbidity may cause background noise across the visible spectrum, which increases with the concentration of detergent, but this does not affect the determination of reaction rate.
To achieve reliable and reproducible results, substrate emulsion should always be prepared just before the assay and in the same manner, considering speed and time of stirring, test tube dimensions, and the size of pipette tip used to dispense substrate drop wise into the microtiter plate wells.

Kinetics of lipases using PHIBLA
To validate PHIBLA, lipases with different pHs optima in the hydrolysis of TGs were used . RGL and rDGL are known to display an optimal activity at an acidic pH ( 36,46 ). rHPL and PPL prefer to catalyze the hydrolysis of TG at neutral or slightly alkaline pH ( 48,49 ). TLL is known to work optimally at alkaline pH ( 45 ). A particular case is YL-LIP2, whose optimum pH of TG hydrolysis depends on the amount of detergent (NaTDC) present in the assay ( 45 ). An esterase acting at acidic pH, rAnFaeA, was also buffer (pH 5.0). Absorption spectra (350-700 nm) of deprotonated and protonated indicator were observed showing two peaks at 440 and 610 nm, corresponding to yellow and blue colors, respectively. The larger difference in the extinction coeffi cients of the protonated and deprotonated forms was observed at 610 nm, which gives good sensitivity in optical density (OD) variations. It was therefore decided to measure FFA release indirectly through the disappearance of the blue color. Similar absorption spectra were recorded for the other indicators and max for observing the largest changes in absorbance upon protonation was determined for each indicator. The max and color followed during the reaction for each indicator is shown in Table 1 .
In order to develop a quantitative method, calibration curves of absorbance variations with BtA or OcA amounts were established for each experimental condition. Figure 2B shows an example of a standard curve obtained with BtA and OcA at pH 5.0 with BCG. Linear variations in absorbance were observed while increasing fatty acid concentration, and the experimental slopes were 0.711 mM/ OD and 0.6121 mM/OD for BtA and OcA, respectively. A steeper slope means more sensitivity of the method. Using linear regression, very good correlation factors (R 2 ) of 0.99 and 0.98 were obtained for BtA and OcA, respectively. The linearity in the standard curve implies the linearity of the enzyme assay. Table 2 presents a variety of different slopes and R 2 values obtained from standard curves at different pH values. All standard curves were linear in experimental conditions assayed up from 0.1 to 1.5 mM FFA (fi nal concentration), demonstrating that the quantifi cation of enzymatic activity can be done. Each assay had to be tested against a blank sample (medium or buffer only) and lipase activity (units per milliliter) using the microtiter plate assay was calculated according to the following equation: Activity (units per milliliter) = [(OD/min Reaction Ϫ OD/min Blank )/S] × (reaction volume/sample volume) × DF; where OD/ min Reaction and OD/min Blank are the rates for enzyme catalyzed reaction and blank sample, respectively; S is the standard curve slope in millimolar per OD; a ratio of six between the total reaction volume and the sample volume in the microtiter plate well that was chosen here; and DF is the the steady-state reaction rate and linear kinetics were recorded at least for the fi rst 5 min of the reaction. It is worth noticing that the background noise measured in blanks without enzyme increased with the pH, probably because of some spontaneous hydrolysis of substrate occurring at alkaline pH. Blank assays were therefore performed for all pH values and subtracted from the assays with enzyme.
The effects of lipase amounts on the steady-state reaction rate were tested using the same lipases ( Fig. 3 ). Whatever the substrate [TG(4:0) or TG(8:0)], enzyme activity used to verify the specifi city of the method. This esterase can hydrolyze the soluble fraction of TG(4:0), while it has no activity on TG(8:0).
Typical kinetics of the decrease in OD during the hydrolysis of TG(4:0) and TG(8:0) at different pHs are shown in Fig. 3 for RGL ( Fig. 3A ; 39.6 ng of RGL/microwell; pH 5.0), rHPL ( Fig. 3C ; 7.68 ng of rHPL/microwell; pH 7.2), and TLL ( Fig. 3E ; 3.7 ng of TLL/microwell; pH 9.2). A greater sensitivity of the assay was observed at alkaline pH. Lag times were not observed in any case before reaching the FFAs released upon substrate hydrolysis by NaOH, thus keeping the pH at a constant end point value.
The pH-dependent activity profi les of rHPL on TG(4:0) and TG(8:0) were recorded with PHIBLA and were compared with those obtained with the pH-stat technique at the same pH values ( Fig. 4 ). The maximum specifi c activities on TG(4:0) were obtained at pH 7.2 using both PHIBLA (1,343 ± 80 U/mg) and the pH-stat technique (10,474 ± 654 U/mg). Similarly, the maximum specifi c activities on TG(8:0) were obtained at the same pH value of 8.2 using PHIBLA (922 ± 76 U/mg) and the pH-stat technique (8,949 ± 284 U/mg). Despite differences in the two methods (e.g., substrate concentration, agitation, presence of alcohols), similar pH-dependent activity profi les were obtained. Hydrolysis rates, however, were found to be around 10 times higher with the pH-stat technique than with the PHIBLA microtiter plate assay. Table 3 shows a comparison of specifi c activities of pure enzymes on TG(4:0) and TG(8:0) obtained by means of PHIBLA and the pH-stat technique at distinct pH values. Although the specifi c activities obtained with PHIBLA were 2-to 20-fold lower than those recorded with the was found to increase linearly with enzyme amounts up to 40 ng/well for RGL ( Fig. 3B ; R 2 = 0.99 and 0.97, respectively; pH 5.0) and rDGL (R 2 = 0.99 and 0.99, respectively; pH 5.5), 8 ng/well for rHPL ( Fig. 3D ; R 2 = 0.99 and 0.99, respectively; pH 7.2), 12 ng/well for PPL (R 2 = 0.99 and 0.99, respectively; pH 8.2), 6 ng/well for TLL ( Fig. 3F ; R 2 = 0.99 and 0.99; pH 9.2), and YLLIP2 (R 2 = 0.99 and 0.97, respectively; pH 6.0) and 200 ng/well for rAnFaeA (R 2 = 0.99; pH 5.0). It is noteworthy that the rAnFaeA esterase was active only on TG(4:0), and the amount of enzyme required was much higher than for a lipase.

Comparison of the lipase-specifi c activities determined by PHIBLA in microtiter plates and by the pH-stat technique
The pH-stat method is generally used as a reference lipase assay ( 50,51 ). This is a convenient technique for characterizing lipase activity and specifi city, and it has been used with partly water-soluble substrates for studying the interfacial activation of lipases ( 9,29,52,53 ). The lipase activity is determined using a mechanically stirred emulsion of a natural or synthetic TG, by continuously titrating Fig. 4. pH profi le of rHPL activity during hydrolysis of TG(4:0) and TG(8:0) using PHIBLA (A) and pHstat technique (B). Reactions were carried out at 37°C. Buffer solutions contained 2.5 mM MOPS, 150 mM NaCl, 6 mM CaCl 2 , 0.5 mM NaTDC, and 5-fold molar excess of colipase. ␤ -CD at a concentration of 3 mg/ml was used for PHIBLA. TG(4:0) and TG(8:0) concentrations were 5 mM for both substrates using PHIBLA and 114 mM and 68 mM, respectively, using the pH-stat. Stirring was orbital for PHIBLA and with an impeller for pH-stat. Maximum velocity was taken into account for activity determination. also found to increase (35-fold) the rate of TG(4:0) hydrolysis by rAnFaeA measured with the pH-stat technique (data not shown).

Study of the effect of pH on enzyme activity using PHIBLA
The effect of pH on the lipase-catalyzed hydrolysis of TG(4:0) and TG(8:0) was determined using PHIBLA ( Fig. 5 ). Overall, the pH-dependent activity profi les were similar to those obtained by the pH-stat technique. PHIBLA is however limited to activity measurements in the pH 5.0-9.2 range with intervals of 0.5-1 unit, and therefore may not be adapted for determining the pH optimum of some enzymes pH-stat technique, the ratios of activities on TG(4:0) and TG(8:0) were found to be similar for both methods and each lipase.
The fact that purifi ed lipases were found to be more active with the pH-stat assay than with the PHIBLA microplate method is probably related to a better emulsifi cation of substrate with the mechanical stirring of the pH-stat device, thus generating a larger accessible surface for the lipase ( 54 ). The opposite phenomenon was observed for rAnFaeA esterase , which was 15 times more active in catalyzing the hydrolysis of TG(4:0) using PHIBLA than using the pH-stat technique. This might be attributed to the addition of cosolvent, because addition of tert -butanol to the emulsion was extracts containing PPL, and the culture medium of a recombinant strain of Pichia pastoris producing TLL. This enabled us to confi rm that highly purifi ed RGL ( Fig. 5B ), PPL ( Fig.5D ), TLL ( Fig. 5F ), and their corresponding crude enzyme preparations showed similar pH-dependent activity profi les (supplementary Fig. 3). No major interference of other compounds found in crude enzyme preparations could therefore be identifi ed.

Study of lipase inhibition using PHIBLA
In order to validate a possible application of the PHI-BLA assay in the screening of lipase inhibitors, THL, a well-known covalent inhibitor of digestive lipases ( 56,57 ), was used. Because the inhibition by rHPL and RGL by THL has been extensively studied with other methods (pH-stat technique, monomolecular fi lms), these lipases were chosen for testing the PHIBLA assay.
Pre-incubation of either rHPL or RGL with THL (at a molar ratio of 1:100) was performed in the absence or in the presence of bile salts at 25°C and pH 7.2 or 5.0, respectively. The residual lipase activity was measured with PHIBLA like rAnFaeA esterase ( Fig. 5A ) and TLL ( Fig. 5F ) that have optimum pH below pH 5 and above pH 9, respectively.
The maximum activity was observed at pH 5.5 for RGL using TG(4:0) ( Fig. 5B ), similarly to what was reported by Moreau et al. ( 36 ), at pH 5.5 and 6 for rDGL using TG(4:0) and TG(8:0), respectively ( Fig. 5C ), similarly to what was reported by Carrière et al. ( 46 ), at pH 6.0 for YLLIP2 using both TGs with 4 mM NaTDC ( Fig. 5D ), similarly to what was reported by Aloulou and colleagues ( 37,45 ), at pH 8.2 for PPL using both TGs ( Fig. 5E ), similarly to what was reported by Borgström and Erlanson ( 49 ), at pH 9.2 for TLL ( Fig. 5F ; in both TGs), similarly to what was reported by Aloulou et al. ( 45 ). It is worth noting that all lipases were active on both substrates [TG(4:0) and TG(8:0)], and, as expected, rAnFaeA esterase activity was observed only in the emulsion of shortchain TG [TG(4:0)], with a maximum activity at pH 5.0 ( Fig. 5A ). The effect of pH on the activity of rAnFaeA on TGs has not been reported so far. Only the optimum pH in the hydrolysis of methyl ferulate was found to be pH 5.0 ( 55 ).
PHIBLA was also tested with crude lipase preparations like rabbit gastric extracts containing RGL, porcine pancreatic BLA could therefore be used to discriminate lipase and esterase activities. PHIBLA also provides a useful tool for HTS of lipase and lipase inhibitor in various biological samples. It can also be used as a quick determination of enzyme activity during purifi cation of lipases/esterases (data not shown). In addition, the principle of PHIBLA for detecting the production of FFA at various pHs should allow the use of other substrates with a carboxylic ester function (i.e., methyl esters, phospholipids, etc.) and this paves the way to future developments of PHIBLA. and TG(4:0) emulsions at 37°C, using samples taken from the incubation medium.
In the presence of 4 mM NaTDC and colipase in the incubation medium, a rapid and nearly complete reduction of rHPL activity (initial velocity) was observed, whereas, in the absence of NaTDC in the incubation medium, THL partially inhibited HPL and a plateau value at around 50% of the initial activity was rapidly reached when using TG(4:0) as substrate ( Fig. 6A ). Similar kinetics were observed by Tiss et al. ( 26 ) in experiments carried out using the pH-stat technique at pH 8.0.
On the other hand, after a few minutes of incubation with THL, the RGL initial activity measured on TG(4:0) ( Fig. 6C ) was drastically reduced either in the presence or absence of NaTDC. Furthermore, RGL was completely inactivated after 30 min. Gargouri et al. ( 58 ) observed that the presence of NaTDC at a fi nal concentration of 4 mM in the incubation medium did not affect RGL inhibition in assays carried out at pH 6.0 with the pH-stat technique. However, they observed a remaining activity of 30% after 2 h of incubation.
PHIBLA was also used to study lipase inhibition during lipolysis. The hydrolysis of TG(4:0) was recorded at pH 7.2 with rHPL and at pH 5.0 with RGL. Five minutes after the start of the reaction, THL was added at a molar ratio of 1:100. As can be seen from the kinetics shown in Fig. 6 , TG(4:0) hydrolysis decreased rapidly by around 70% in reactions catalyzed by rHPL ( Fig. 6B ) and was completely stopped in reactions catalyzed by RGL ( Fig. 6D ), when adding THL. Control experiments were conducted with DMSO only in order to see the effect of the solvent in which THL solution was prepared. The rate of TG(4:0) hydrolysis decreased by around 25% in reactions catalyzed by RGL ( Fig. 6D ), whereas there was no effect on the reactions catalyzed by rHPL.
PHIBLA may be a convenient assay for the fi rst line screening of lipase inhibitors before performing a more detailed study on the specifi city and mechanism of action of these inhibitors.

CONCLUSION
The PHIBLA in microtiter plates is a convenient and rapid assay to measure lipase activity on short-and medium-chain TGs, by performing continuous kinetic recordings at different pH values ranging from 5.0 to 9.2. Despite the fact that the lipase-specifi c activities measured with this method were two to twenty times lower than those obtained with the pH-stat technique, the limit of detection was highly satisfactory and the method could be used to detect amounts as low as 1 g of FFA, and 1 ng of enzyme. Substrate emulsion was easy to prepare and was stable for at least 15 min, the period of time needed for recording the reaction kinetics. In addition, PHIBLA was a convenient assay for studying the substrate preference of lipases between short [TG(4:0)] and medium [TG(8:0)] acyl chain TGs. Well-known lipases were able to hydrolyze both substrates while a model esterase like rAnFaeA could only hydrolyze TG(4:0), which is partly soluble in water. PHI-