Dipeptide Nitrile CD34 with Curcumin: A New Improved Combination Strategy to Synergistically Inhibit Rhodesain of Trypanosoma brucei rhodesiense

Rhodesain is the main cysteine protease of Trypanosoma brucei rhodesiense, the parasite causing the acute lethal form of Human African Trypanosomiasis. Starting from the dipeptide nitrile CD24, the further introduction of a fluorine atom in the meta position of the phenyl ring spanning in the P3 site and the switch of the P2 leucine with a phenylalanine led to CD34, a synthetic inhibitor that shows a nanomolar binding affinity towards rhodesain (Ki = 27 nM) and an improved target selectivity with respect to the parent dipeptide nitrile CD24. In the present work, following the Chou and Talalay method, we carried out a combination study of CD34 with curcumin, a nutraceutical obtained from Curcuma longa L. Starting from an affected fraction (fa) of rhodesain inhibition of 0.5 (i.e., the IC50), we observed an initial moderate synergistic action, which became a synergism for fa values ranging from 0.6 to 0.7 (i.e., 60–70% inhibition of the trypanosomal protease). Interestingly, at 80–90% inhibition of rhodesain proteolytic activity, we observed a strong synergism, resulting in 100% enzyme inhibition. Overall, in addition to the improved target selectivity of CD34 with respect to CD24, the combination of CD34 + curcumin resulted in an increased synergistic action with respect to CD24 + curcumin, thus suggesting that it is desirable to use CD34 and curcumin in combination.


Introduction
Human African Trypanosomiasis (HAT) is a parasitic disease widespread in sub-Saharan Africa, where it represents an important death cause [1]. Two subspecies of Trypanosoma are able to induce HAT: T. brucei gambiense, typical of western and central Africa and responsible for the chronic form of HAT, and T. b. rhodesiense, widespread in eastern and southern Africa and able to induce the acute form of HAT, which has a higher mortality rate [2].
HAT is characterized by two main stages: in the first stage, also known as the hemolymphatic stage, a bloodstream invasion by the parasite induces fever and muscle aches; if untreated, the first stage can evolve into the neurological stage, in which the penetration of the central nervous system by the Trypanosoma induces neurological disturbances, sleep disorders, and finally death [3].

Chemistry
The dipeptide nitrile CD34 was obtained by coupling reactions between the carboxylic acid 1 and amine 2 (Scheme 1), using ECDI and HOBt as coupling reagents, where acid 1 mimics the P3-P2 fragment while amine 2 represents the P1 synthon, both synthesized in accordance with our established synthetic procedure [19,29]. In the present study, we developed a novel dipeptide nitrile CD34 (Figure 1), 3,5difluoro-substituted on the P3 aromatic nucleus. In our previous studies, we found that the incorporation of a difluoro phenyl ring at the P3 site resulted in more effective selectivity towards rhodesain with respect to the mono-fluoro-substitution [18]. Concerning the P2 site, the leucine present in CD24 was switched with a phenylalanine in analogy to K11777, one of the most potent rhodesain inhibitors, with whom rhodesain has been co-crystallized (Protein Data Bank ID: 2P7U) [36].
Overall, the new fluorinated dipeptide nitrile was synthesized with the aim of investigating the reactivity of the novel inhibitor, the selectivity towards the target protease, and the possibility of using it in combination with curcumin in order to reduce the toxicity for the human host by reducing the individual dose [37]. The results of this investigation will now be presented and discussed.

Chemistry
The dipeptide nitrile CD34 was obtained by coupling reactions between the carboxylic acid 1 and amine 2 (Scheme 1), using ECDI and HOBt as coupling reagents, where acid 1 mimics the P3-P2 fragment while amine 2 represents the P1 synthon, both synthesized in accordance with our established synthetic procedure [19,29].

Inhibition of Rhodesain and Target Selectivity
CD34 was tested against recombinant rhodesain by using Cbz-Phe-Arg-AMC as a fluorogenic substrate. First, a preliminary screening at a fixed inhibitor concentration of 20 µM was performed. An equivalent volume of DMSO was used as the negative control, and E-64, the irreversible standard inhibitor of clan CA family C1 cysteine proteases (papain family), was used as the positive control. Since CD34 showed full inhibition of rhodesain at 20 µM, Tian continuous assays were performed at seven different concentrations, in the range 0.001-20 µM (from minimal to full enzyme inhibition), to determine the dissociation constant of the non-covalent complex enzyme-inhibitor Ki (nM) ( Table 1), since CD34 reversibly inhibited rhodesain. An analysis of the obtained data clearly established that both CD24 and CD34 inhibited rhodesain at the nanomolar level (Table 1), with a slight increase in activity for CD24. However, when both CD24 and CD34 were tested against human cathepsin L to check the selectivity of the two dipeptide nitriles towards the trypanosomal protease, CD34 showed an improved selectivity with respect to CD24 (SI = 7.26 vs. 2.12 for CD34 and CD24, respectively). Scheme 1. Reagents and conditions: (a) EDCI, HOBt, DIPEA, dry DCM, 0 • C to rt., on.

Inhibition of Rhodesain and Target Selectivity
CD34 was tested against recombinant rhodesain by using Cbz-Phe-Arg-AMC as a fluorogenic substrate. First, a preliminary screening at a fixed inhibitor concentration of 20 µM was performed. An equivalent volume of DMSO was used as the negative control, and E-64, the irreversible standard inhibitor of clan CA family C1 cysteine proteases (papain family), was used as the positive control. Since CD34 showed full inhibition of rhodesain at 20 µM, Tian continuous assays were performed at seven different concentrations, in the range 0.001-20 µM (from minimal to full enzyme inhibition), to determine the dissociation constant of the non-covalent complex enzyme-inhibitor K i (nM) ( Table 1), since CD34 reversibly inhibited rhodesain. An analysis of the obtained data clearly established that both CD24 and CD34 inhibited rhodesain at the nanomolar level (Table 1), with a slight increase in activity for CD24. However, when both CD24 and CD34 were tested against human cathepsin L to check the selectivity of the two dipeptide nitriles towards the trypanosomal protease, CD34 showed an improved selectivity with respect to CD24 (SI = 7.26 vs. 2.12 for CD34 and CD24, respectively).

Calculation of the Combination Index
For the calculation of the combination index (CI), curcumin was also tested against rhodesain in the range 1-100 µM and the IC 50 values were obtained using GraphPad Prism 5.0.3 (GraphPad software Inc., San Diego, California) from dose-response curves as shown in Figure 2 (0.35 ± 0.035 µM for CD34 and 24.5 ± 1.74 µM for curcumin).

Calculation of the Combination Index
For the calculation of the combination index (CI), curcumin was also tested against rhodesain in the range 1-100 µM and the IC50 values were obtained using GraphPad Prism 5.0.3 (GraphPad software Inc., San Diego, California) from dose-response curves as shown in Figure 2 (0.35 ± 0.035 µM for CD34 and 24.5 ± 1.74 µM for curcumin). In a subsequent experiment, six data points were selected for both compounds (1/32 × IC50, 1/4 × IC50, 1/2 × IC50, IC50, 2 × IC50, and 4 × IC50, Table 2) with the aim of evaluating if a synergistic, additive, or antagonistic effect occurs in the combination study of the two inhibitors. In this assay, the combination of CD34 and curcumin (molar ratio 1:70) provided an IC50 value of 8.06 ± 0.68 µM. We then converted each dose-response curve into a Median Effect Plot, which was obtained by plotting the log (fa/fu) on the y-axis vs. the log (D) on the x-axis ( Figure 3). In the median effect plot, the maximum response corresponds to 1 instead of 100 in the doseresponse curve. Therefore, fa + fu = 1, where fa corresponds to the "affected fraction", i.e., the percentage of enzyme that has been inhibited, while fu is the unaffected fraction, i.e., the residual enzyme activity. The slope of the straight line of each Median Effect Plot is the "m value". In detail, CD34 had an m1 value of 1.0006, curcumin had an m2 value of 0.9455, and the combination assay had an m1,2 value of 2.8377 with a CD34/curcumin molar ratio of 1:70.
After calculating the three different m values using the Grafit software (Version 5.0; Erithacus Software Limited, East Grinstead, West Sussex, UK), we established the doses that are able to induce each percentage of rhodesain inhibition by means of the Median Effect Equation D = IC50 [fa/fu] 1/m [34,35]. In a subsequent experiment, six data points were selected for both compounds (1/32 × IC 50 , 1/4 × IC 50 , 1/2 × IC 50 , IC 50 , 2 × IC 50 , and 4 × IC 50 , Table 2) with the aim of evaluating if a synergistic, additive, or antagonistic effect occurs in the combination study of the two inhibitors. In this assay, the combination of CD34 and curcumin (molar ratio 1:70) provided an IC 50 value of 8.06 ± 0.68 µM. We then converted each dose-response curve into a Median Effect Plot, which was obtained by plotting the log (f a /f u ) on the y-axis vs. the log (D) on the x-axis ( Figure 3). In the median effect plot, the maximum response corresponds to 1 instead of 100 in the dose-response curve. Therefore, f a + f u = 1, where f a corresponds to the "affected fraction", i.e., the percentage of enzyme that has been inhibited, while f u is the unaffected fraction, i.e., the residual enzyme activity. The slope of the straight line of each Median Effect Plot is the "m value". In detail, CD34 had an m 1 value of 1.0006, curcumin had an m 2 value of 0.9455, and the combination assay had an m 1,2 value of 2.8377 with a CD34/curcumin molar ratio of 1:70. With the aim of determining the inhibitory effect given by the combination of CD34 and curcumin, we used the Chou-Talalay method to evaluate the multiple drug effects [34,35]. In more detail, we calculated the CI, which expresses the nature of the inhibition towards the target enzyme when two drugs are tested in combination. In particular, it is   [34,35].
With the aim of determining the inhibitory effect given by the combination of CD34 and curcumin, we used the Chou-Talalay method to evaluate the multiple drug effects [34,35]. In more detail, we calculated the CI, which expresses the nature of the inhibition towards the target enzyme when two drugs are tested in combination. In particular, it is well known that CI > 1, CI = 1, and CI < 1 generally correspond to antagonistic, additive, and synergistic effects, respectively [34,35]. The CI for mutually non-exclusive drugs that act independently was calculated as follows: where (IC 50 ) 1 and (IC 50 ) 2 are obtained from the dose-response curves and D 1 and D 2 , are the concentrations that are able to induce a specific percentage of rhodesain inhibition, as determined by the Median Effect Equation.
The Grafit software was used to determine the CI ranging from 50% to 100% rhodesain inhibition ( Figure 4). With the aim of determining the inhibitory effect given by the combination of CD34 and curcumin, we used the Chou-Talalay method to evaluate the multiple drug effects [34,35]. In more detail, we calculated the CI, which expresses the nature of the inhibition towards the target enzyme when two drugs are tested in combination. In particular, it is well known that CI > 1, CI = 1, and CI < 1 generally correspond to antagonistic, additive, and synergistic effects, respectively [34,35]. The CI for mutually non-exclusive drugs that act independently was calculated as follows: where (IC50)1 and (IC50)2 are obtained from the dose-response curves and D1 and D2, are the concentrations that are able to induce a specific percentage of rhodesain inhibition, as determined by the Median Effect Equation.
The Grafit software was used to determine the CI ranging from 50% to 100% rhodesain inhibition (Figure 4).
Starting from IC50 (fa = 0.5), which is the first value to take into consideration when we describe the activity of a novel inhibitor, we observed an initial moderate synergistic action, which became a synergism for fa ranging from 0.6 to 0.7. Interestingly, at 80-90% inhibition of rhodesain activity, a strong synergism was observed, resulting in a very strong synergism at 100% enzyme inhibition.  Starting from IC 50 (f a = 0.5), which is the first value to take into consideration when we describe the activity of a novel inhibitor, we observed an initial moderate synergistic action, which became a synergism for f a ranging from 0.6 to 0.7. Interestingly, at 80-90% inhibition of rhodesain activity, a strong synergism was observed, resulting in a very strong synergism at 100% enzyme inhibition.

Chemistry
All reagents and solvents were obtained from commercial suppliers and were used without further purification. Elemental analyses were carried out on a C. Erba Model 1106 (Elemental Analyzer for C, H, and N) instrument, and the obtained results were within ±0.4% of the theoretical values. Merck silica gel 60 F254 plates were used for analytical TLC; flash column chromatography was performed on Merck silica gel (200-400 mesh). 1 H and 13 C NMR spectra were recorded on a Varian 500 MHz spectrometer equipped with a ONE_NMR probe, operating at 499.74 and 125.73 MHz for 1 H and 13 C NMR spectra, respectively. We used the residual signal of the deuterated solvent as an internal standard. Splitting patterns are described as singlet (s), doublet (d), doublet of doublet (dd), triplet (t), quartet (q), multiplet (m), or broad singlet (bs). 1 H and 13 C NMR chemical shifts (δ) are expressed in ppm, and coupling constants (J) are given in Hz.

Rhodesain Inhibition Assays and Calculation of the Combination Index
Curcumin was purchased from Sigma Aldrich. Since the range of active concentrations for curcumin was known by us [33], a preliminary screening with rhodesain was performed only for CD34 at 20 µM; since it inhibited rhodesain by more than 80%, it passed the initial screening and was characterized in detail. An equivalent amount of DMSO was used as negative control. Product release from substrate hydrolysis (Cbz-Phe-Arg-AMC, 10 µM) was determined continuously over a period of 10 min at room temperature. The assay buffer contained 50 mM sodium acetate, pH = 5.5, 5 mM EDTA, 200 mM NaCl, and 0.005% Brij 35 to avoid aggregation and false-positive results. The enzyme buffer contained 5 mM DTT rather than Brij 35. Inhibitor solutions were prepared from stocks of DMSO.
Fluorescence of the product AMC of the substrate hydrolyses was measured using an Infinite 200 PRO microplate reader (Tecan, Männedorf, Switzerland) at room temperature with a 380 nm excitation filter and a 460 nm emission filter. Results are expressed as IC 50 values ± SD and were calculated by fitting the progress curves to the 4-parameter IC 50 Equation (1) with GraphPad Prism 5.0.3 (GraphPad software Inc., San Diego, CA, USA): where y (∆F/min) is the substrate hydrolysis rate; y max is the maximum value of the doseresponse curve measured at an inhibitor concentration of I = 0 µM; y min is the minimum value obtained at high inhibitor concentrations; and s is the Hill coefficient. The inhibitory constants (K i ) were calculated according to the Cheng-Prusoff Equation (2): where S is the substrate concentration and K M is the Michaelis-Menten constant (K M = 0.8 µM for Cbz-Phe-Arg-AMC).
Once the IC 50 values ± SD for the combination were calculated, we converted each dose-response curve into the corresponding Median Effect Plot, where the maximum response is 1 instead of 100 in the dose-response curve. The fraction of enzyme that is inhibited is called the "affected fraction" (f a ), while the fraction of enzyme that is not inhibited is called the "unaffected fraction" (f u ), where f a + f u = 1. The Median Effect Plot is obtained by plotting the log (f a /f u ) vs. the log (D) on the x-axis. This allowed us to calculate the "m value", which represents the Hill-type coefficient, which means the sigmoidal trend (S-shaped) of the dose-response curve.
Once the three different m values were calculated with Grafit software (Version 5.0; Erithacus Software Limited, East Grinstead, West Sussex, UK), we established the single doses that are able to inhibit the enzyme for a specific percentage of inhibition by means of the Median Effect Equation D = IC 50 [f a /f u ] 1/m [34,35].
Then, considering that CD34, in analogy to CD24 [27], acts as a competitive inhibitor and that curcumin acts as a non-competitive inhibitor, the CI for mutually non-exclusive drugs that act independently was calculated as follows: where (IC 50 ) 1 and (IC 50 ) 2 were obtained by dose-response curves, and D 1 and D 2 are the concentrations able to induce a specific percentage of rhodesain inhibition, as determined by the Median Effect Equation.

Cathepsin L Inhibition Assays
The determination of the activity of the inhibition against human Cathepsin L (hCatL) was performed in fluorescence-based assays [39]. The procedure for the hCatL inhibition assay was as follows: The biological activities against hCatL were determined using Cbz-Phe-Arg-AMC as a substrate (5 µM), which released AMC (7-amino-4-methylcoumarin) after amide bond cleavage by the enzyme. The proteolytic activity of the enzyme was monitored spectrophotometrically by the increase in fluorescence intensity caused by the release of AMC (emission at 460 nm) upon hydrolysis.
Inhibitor solutions were prepared from stocks of DMSO. An initial screening at an inhibitor concentration of 20 µM was performed to identify ligands (CD24 and CD34) with an inhibition of hCatL higher than 80%.
CD34 and CD24 were then tested separately by Tian continuous assays, twice in duplicate in 96-well plates, for a total volume of 200 µL. An equivalent amount of DMSO was used as negative control. In more detail, we used 0.001 µM, 0.01 µM, 0.05 µM, 0.1 µM, 0.5 µM, 1 µM, and 10 µM for CD34 and CD24. The residual enzyme activities were determined continuously over a period of 10 min at room temperature.
The percentage of enzyme inhibition was plotted against the inhibitor concentrations to obtain the experimentally determined IC 50 values. The inhibitory constants (K i ) were calculated according to the Cheng-Prusoff Equation (1), with K M = 6.5 µM for Cbz-Phe-Arg-AMC.

Rhodesain Expression and Purification from P. pastoris and E. coli
The expression of rhodesain from P. pastoris was performed as described in [40]. For purification from E. coli Rosetta2 DE3 pLysS (Novagen), the T. b. rhodesiense rhodesain sequence (uniprot-ID.: Q95 PM0), followed by a TEV cleavage site and a hexahistidine-tag, was cloned between the NdeI and BamHI restriction sites of a pET-11a vector by Genscript using codon optimization. The plasmid was amplified in XL-10 cells. A GFP sequence was additionally introduced between the TEV site and the 6xHis-tag by Gibson assembly following extensive construct and purification optimization. The rhodesain plasmid and the GFP gene were amplified in a PCR using the primers 5 -CACCACCATCACCACCACTAAGG-3 and 5 -GCTACCTTGAAAATACAGATTCTCCGGG-3 for the vector and 5 -CTGTATTTTCAAG GTAGCGTGAGCAAGGGCGAGGAGC-3 and 5 -GGTGGTGATGGTGGTGCTTGTACAG CTCGTCCATGCCG-3 for the insert.

Statistical Analyses
The statistical analysis of the data was performed using the one-way test (ANOVA) with Dunnett's multiple comparison test, considering significant differences of p < 0.05 with respect to the percentage of rhodesain inhibition of curcumin, CD34, and curcumin + CD34. The analyses were performed with GraphPad Prism 5.0.3. Results were expressed as the arithmetic mean ± standard deviation (SD).

Conclusions
Besides the slightly higher activity of CD24 compared to CD34 (IC 50 = 0.2 and 0.35, respectively), CD34 showed increased target selectivity, which is fundamental considering the high degree of homology between rhodesain and human cathepsin L, both belonging to the papain family.
If we compare the obtained values of CI with those obtained for the combination CD24 + curcumin [33] (Table 3), it is possible to observe an initial additive effect at the IC 50, followed by a slight-to-moderate synergism at 60-70% and a synergistic action from 80% to 100% rhodesain inhibition. Thus, it is clear that the combination of CD34 + curcumin resulted in a much stronger synergistic action, especially at 100% enzyme inhibition, where a very strong synergism was observed, thus assessing an improved combination strategy for CD34 + curcumin with respect to that of CD24 + the same nutraceutical.
Considering that both CD24 and CD34 bind to the active site of rhodesain, while curcumin acts as a non-competitive inhibitor, a possible explanation for the increased synergistic action can be attributed to the different structural requirements of CD34, i.e., the presence of a phenyl alanine residue instead of a leucine, which would probably better establish van der Waals contacts with the adjacent lipophilic residues of the S2 pocket (M68, A138, L160, and A208).
In contrast, the presence of the fluoro-substituted aromatic nucleus at the P3 site, in agreement with our previous studies [27], should be able to establish a parallel-displaced ππ interaction with F61 of the rhodesain S3 pocket, forming also additional charge-transfer interactions with the π-faces of G65 and G66 backbone peptide bonds. The presence of an additional fluorine atom at position 5 of the P3 aromatic nucleus, with regard to CD24, could further enhance the strength of these charge-transfer contacts. Overall, both the improvement in the strength of interactions with rhodesain and presumably the different disposition in the catalytic site of CD34, with regard to CD24, could favor the interaction with curcumin, which is believed to bind to an allosteric site close to the catalytic site.