Identification of small molecule inhibitors against SecA of Candidatus Liberibacter asiaticus by structure based design

European Journal of Medicinal Chemistry 54 (2012) 919e924

Contents lists available at SciVerse ScienceDirect

European Journal of Medicinal Chemistry journal homepage: http://www.elsevier.com/locate/ejmech

Short communication

Identification of small molecule inhibitors against SecA of Candidatus Liberibacter asiaticus by structure based design Nagaraju Akula a, Pankaj Trivedi a, Frank Q. Han b, Nian Wang a, * a b

Citrus Research & Education Center, Department of Microbiology and Cell Science, University of Florida, 700 Experiment Station Rd., Lake Alfred, FL 33850, USA Structure Based Design, Inc., 6048 Cornerstone Court West, Suite D, San Diego, CA 92121, USA

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

< Homology model of SecA was built and optimized for structure based design. < Molecular docking methods were used to identify 20 structures for activity studies. < Five compounds were found at nano molar concentrations against SecA ATPase activity. < Five compounds have shown antimicrobial activity against Agrobacterium tumefaciens. < Identified small molecules could be used as leads for development of antimicrobial compounds.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 5 March 2012 Received in revised form 20 May 2012 Accepted 23 May 2012 Available online 2 June 2012

Huanglongbing is the most devastating disease of citrus caused by Candidatus Liberibacter asiaticus (Las). In the present study, we report the discovery of novel small molecule inhibitors against SecA ATPase of Las by using structure based design methods. We built the homology model of SecA protein structure of Las based on the SecA of Escherichia coli. The model was used for in-silico screening of commercially available compounds from ZINC database. Using the glide flexible molecular docking method, twenty structures were chosen for in vitro studies. Five compounds were found to inhibit the ATPase activity of SecA of Las at nano molar concentrations and showed antimicrobial activities against Agrobacterium tumefaciens with MBC ranging from 128 to 256 mg/mL. These compounds appear to be suitable as lead compounds for further development of antimicrobial compounds against Las. Ó 2012 Elsevier Masson SAS. All rights reserved.

Keywords: Huanglongbing Candidatus Liberibacter asiaticus SecA ATPase Homology modeling Molecular docking

1. Introduction

Abbreviations: HLB, Huanglongbing; Las, Candidatus Liberibacter asiaticus; HTVS, High Throughput Virtual Screening; IC, Inhibitory Concentration; MBC, Minimum Bactericidal Concentration. * Corresponding author. Tel.: þ1 863 956 8828; fax: þ1 863 956 4631. E-mail address: nianwang@ufl.edu (N. Wang). 0223-5234/$ e see front matter Ó 2012 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.ejmech.2012.05.035

Huanglongbing (HLB), or citrus greening, is the most destructive citrus disease worldwide [1,2]. HLB is associated with a fastidious, Gram-negative, phloem-limited bacterium (Candidatus Liberibacter spp.) [3], and recent attempts had limited success in culturing the organism [4]. In Florida, Ca. L. asiaticus (Las) is the causal agent of

920

N. Akula et al. / European Journal of Medicinal Chemistry 54 (2012) 919e924

HLB [5] and is transmitted by Asian citrus psyllid, Diaphorina citri. The disease is widespread in most areas of Asian countries that grow citrus, Africa, Brazil, and U.S [4,6]. The current management methods of HLB are to chemically control psyllids and to scout and remove infected trees. However these approaches have not been able to stop the spreading of the HLB disease [7]. The efficacy of current strategies for management of HLB is limited and no conventional measure has shown to provide consistent and effective suppression of the disease. High cost of frequent insect control and tree removal will eventually render citrus groves nonprofitable. In addition, large scale application of insecticides will disrupt the eco-system and pollute the environment [8]. Frequently, insecticides will become non-effective due to the acquisition of resistance. Insecticides could also kill non-target beneficial insects which disrupt the biological control currently in place. Development of alternative or complementary approaches for effective management of the disease is highly desirable and will greatly help the citrus industry due to the difficulty to control the HLB disease. Considering the highly destructive nature of HLB disease and the lack of control measures, there is a huge potential to develop antimicrobial small molecules against the causal agent thus to suppress the population of Las in planta and to reduce the inoculum for psyllid transmission. Development of antimicrobial small molecules may provide economic and ecological benefits by reducing producing costs, decreasing insecticide application, preserving the natural habitat and populations of beneficial insects, and enhancing productivity of citrus in the presence of HLB. In this study, SecA was used as a target for our molecular modeling studies to identify potential lead antimicrobial compounds. SecA was selected as a promising antimicrobial agent [9e11] because this protein is more conserved and essential in all bacteria and is absent in humans. SecA is the protein translocase ATPase subunit and is involved in pre-protein translocation across and integration into the cellular membrane in bacteria. It is one essential component of the Sec machinery which provides a major pathway of protein translocation from the cytosol across or into the cytoplasmic membrane [12]. SecA cooperates with the SecB chaperone to target pre-proteins to SecYEG as an active ATPase to drive protein translocation across the bacterial membrane when it is bound to the SecYEG complex [13]. SecA is the peripheral membrane ATPase, which couples the hydrolysis of ATP to the stepwise translocation of pre-proteins [14]. The crystal structures of SecA are available for other bacteria such as Escherichia coli [15] and the ATPase active site has been clearly defined. This structural information had been utilized earlier for structure based design to identify antimicrobial compounds with IC50 value up to 2.5 mM against SecA of Las [16]. In this study, we further expanded our previous study in identifying lead antimicrobial compounds with higher activities by targeting SecA using various computational techniques like homology modeling, virtual screening, molecular docking & minimization. Due to the uncultivable nature of Las, we tested the potential inhibitory effect of the selected compounds against Agrobacterium tumefaciens, which is phylogenetically related to Las. The identified active compounds provide lead compounds for further optimization to develop antimicrobial compounds to control the HLB disease. 2. Results & discussion 2.1. Homology modeling & optimization of ATP binding site In our earlier report [16] we used only one PDB structure (2FSG) as template to build the SecA homology model. In the current study, three PDB structures (2VDA, 2FSF and 2FSG) were used as templates to further optimize the homology model of SecA. The precise

positions, orientations/conformations of the helices in the homology modeled SecA varied slightly from the reported templates with RMS ranging from 0.6 to 0.8 Å. The model structure was also validated by PROSA2003 [17] analysis, and the Z-score was 11.58 which indicates the reliability of the model. The homology model of SecA was used to superimpose on the reported PDB (2FSG) structure to identify the ATP binding site on SecA protein of Las (Fig. 1). The ATP-binding site is located between NBD and IRA2-VAR. ATP was extracted from 2FSG and was manually placed on SecA at the similar receptor site of the model and merged with the proteineligand complex (Fig. 1A). Then the proteineligand complex was minimized using OPLS5 force field with 5000 minimization cycles until the minimization was terminated once the RMSD reached a maximum value of 0.05 Å. In the optimized complex, ATP retained its orientation without losing the

Fig. 1. Homology model of SecA of Las (A) Three-dimensional homology model of SecA ATPase of Las and Escherichia coli SecA (2FSG.pdb) were superimposed. The RMS deviation between these two proteins is 0.8 Å. Various domains of the protein structure were represented in different colors and defined as: NBD (Nucleotide binding domain), IRA (1 or 2) Intramolecular Regulator of ATPase, VAR (Variable region of IRA 2), CTD (Carboxy-terminal domain also called C-domain), SD (scaffold sub-domain) and WD (wing sub-domain). Identified ATP (Adenosine Triphosphate) binding site is in between NDB and IRA2-VAR domains colored in blue and green. (B) Intermolecular interactions of ligandeprotein (ATP e SecA of Las) complex after molecular minimization. H-bond interactions are observed between ATP and SecA active site residues R344, G79, K82, T83, L84, Q61 & R56 and pp interactions with F58 are also present. ATP is represented as stick model with dot surface and the remaining active site residues are in sticks.

N. Akula et al. / European Journal of Medicinal Chemistry 54 (2012) 919e924

H-bond interactions between ATP and SecA active site residues (R344, G79, K82, T83, L84, Q61 and R56). The observed intermolecular pp interactions with F58 were similar to the reported crystal structure (Fig. 1B).

structures were selected based on the docking scores against SecA. The resulting poses of the docked compounds were ranked according to their Glide standard and extra precision docking scores. In the case of the library of commercial compounds, the topscoring 500 structures were then evaluated using the extra precision (XP) mode. For the individual ligand studies, the XP mode was used. Compounds detected by 2D and 3D ligand-based searches were analyzed with the binding modes, intermolecular interactions, and with our chemical intuition. Twenty compounds were selected for biological activity studies (Fig. 2).

2.2. Molecular docking: high throughput virtual screening/SP/XP High Throughput Virtual Screening (HTVS) by molecular docking was performed and the docking scores were calculated to filter the structures, and the filtrates were subjected to standard (SP) and extra (XP) precision docking to evaluate the scoring functions. 20,000 structures identified from the physicochemical properties were subjected to glide HTVS to filter the low scored ones. The initial step to screen the small molecule database is grid generation. Receptor grid file was generated by Glide [21] to search for favorable interactions between the ligand and ATP binding site region. The shape and properties of the receptor are represented on a set of grids for positioning and scoring ligand poses. From HTVS, 5000

The selected twenty compounds were tested for their inhibition against SecA ATPase activity. Among the twenty compounds, only five of them showed greater than 50% inhibition at 1 mM (Fig. 3A). The IC50 values of the five compounds C16, C17, C18, C19, and C20 NChiral

N

S N O N

O

2.3. Inhibitory assay against SecA ATPase activity of Las and antimicrobial assay against A. tumefaciens

H Chiral

O

O

-

921

N

O

H

N

S

S

O

O

N

N N

N

(C1)

N

O

N

S

S

(C2)

(C3)

(C4)

O

Cl

N

N

N

S

N

S

O N

N

O

O

CH3

O

O NH

N

N

N

N Br

(C5)

Cl

(C7)

O

(C6)

(C8)

O O

H N

O N

N

S

N NH

S NH O O

O

O

N O

N

O

N

O

S N

S

(C10)

O

N

N N

S N

O

N

H H

O

O

N N

(C12)

N NH

N

O S NH

O

O

O CH3

NH

O

S N

Chiral

O N

S

O

O O

O

S

N N

N

N

N

N N

S N

O N

S

N

N N

O

O O

O (C17)

N (C16)

(C15)

(C14)

(C13)

N O

O

N

H

H

S

(C11)

N Chiral N

N

N N N

N

N

N

(C9)

N O

N

O N

CH3

(C18)

(C19)

+

O

N O

Fig. 2. Structural information of the identified 20 inhibitory compounds (C1eC20) against SecA of Las.

(C20)

922

N. Akula et al. / European Journal of Medicinal Chemistry 54 (2012) 919e924

SecA ATPase activity (%)

A

Compounds concentration at 20 µM 2 µM 1 µM

% of SecA ATPase activity

B

Concentration of the compounds in nM

Fig. 3. Inhibitory activities of identified compounds against SecA of Las. A) Inhibitory activities of compounds 1e15 at 20 mM, 2 mM & 1 mM concentrations against SecA of Las. B) Inhibitory activities of compounds C16eC20 at 1 mM, 750 nM, 500 nM, 250 nM, 200 nM, 100 nM & 50 nM concentrations against SecA of Las.

are 0.25, 0.92, 0.48, 0.64, and 0.44 mM respectively. These values are calculated based on the 50% remaining ATPase activity (Fig. 3B). These five compounds were further tested against A. tumefaciens for antimicrobial activity since Las has not been cultivated in vitro. The minimal bactericidal concentrations (MBCs) of the five identified compounds and streptomycin were determined by broth microdilution method. The MBC values of five compounds C16, C17, C18, C19, C20 and streptomycin are 256, 256, 256, 128, 256 and 64 mg/mL, respectively. The MBC values of all compounds are 2e4 folds higher than streptomycin with C19 showing better activity, which indicates they could probably act as potential antimicrobial agents.

2.4. Molecular docking and minimization: comparative study of ATP binding of high and low activity compounds The built and optimized homology model was used for molecular docking study. To validate our generated grid model and molecular docking methodology, we docked the ATP structure to the active site of SecA of Las. Glide extra precision method was used to rigidly dock the ligand at the binding site of the Las SecA without any constraints and water molecules. The obtained results indicated that the binding mode of ATP inside the receptor was similar and in agreement with the experimental mode of the crystal structure (Fig. 4A). This result indicates that our receptor grid file could be used to dock rigidly with the twenty structures. The

N. Akula et al. / European Journal of Medicinal Chemistry 54 (2012) 919e924

923

Fig. 5. Intermolecular interactions of SecA with ligand after molecular minimization. Intermolecular (H-bond, hydrophobic & pp) interactions between ligandeprotein complexes. (A) C16-SecA active site interactions with R344, D347, T78, G79, K82, T83, R112 & F58 (binding energy: 73878 k cal/mol); (B) C4-SecA active site interactions with D347, T83, L84 & F58 (binding energy: 73039 k cal/mol).

observed docking results of the twenty identified structures showed similar binding orientations and pp interactions with Phe58 as ATP. The structural difference and binding mode between high & low activity structures were analyzed and compared with ATP in order to understand the differences between the activities among the compounds. Specifically ATP is involved in several HBond interactions in the active site due to presence of more nucleophilic phosphate groups. Also the adenine moiety has pp stacking with the aromatic ring of residues Phe58 and forms HBond with Gln64 (Fig. 4A). The conformational orientation of rigidly docked ATP at Las SecA binding sites is slightly different from its original minimized ATP-SecA complex. Nevertheless, the critical interactions with Arg344, Arg112, Thr83, Gly79 and Phe58 remained the same in both models (Figs. 1B and 4A). The high activity structure C16 displayed similar interactions like ATP, particularly in H-Bond interactions (involving residues Gln64, Lys82, Thr83 & Arg344) and pp stacking with Phe58 (Fig. 4B). The low activity structure C4 lacked of interactions with the inside binding site residues Gly81, Lys83 & Arg344 which are more important to keep the ligand inside the binding site to inhibit the ATPase activity. Although the low activity structure has pp stacking with its aromatic ring and Phe58 (Fig. 4C), the lack of internal interactions might be responsible for the lower activity of C4 compound. In addition, the docking scores also correlated well with the in vitro activities of C16 & C4. These experimental and theoretical results indicated the importance to study residue interactions for selecting compounds with higher activity against selected proteins. To further explain the difference between high and low activity structures, energy minimization studies were performed after the molecular docking analysis. While the docking methodology allowed us to study the binding of the Las SecA inhibitors with flexible and rigid biding modes, it does not allow the protein mobility. In order to give free molecular motion of SecA protein along with ligands we minimized the proteineligand complexes and calculated the binding energies (Fig. 5). Interestingly, it was observed that the high activity structure C16 has hydrophobic

Fig. 4. Molecular docking interactions of SecA with different ligands including ATP, high & low activity structure and intermolecular H-Bond interactions, between ligandeprotein complexes. (A) ATP-SecA active site interactions with R344, R112, T78, T83, Q64 & P57 (dock score: 8.7412 k cal/mol); (B) C16-SecA active site interactions with R344, T83, K82 & Q64 (docking score: 7.2140 k cal/mol); (C) C4-SecA active site interactions with T83 & Q64 (docking score: 5.6561 k cal/mol).

interactions with active site residues (Thr78, Gly79 & Arg344) along with other contacts after energy minimization (Fig. 5A), whereas the low activity structure C4 is missing many of these interactions in energy minimized complex (Fig. 5B). This demonstrates that the electronic effects of these substituents have an impact toward higher affinity. The need for a strong H-bond donor/acceptor in the ligand is critical for the binding interactions between the ligand and active site residues. The lack of these interactions in compounds resulted in lower activity, in terms of ATPase inhibition, compared with high activity compounds. 3. Conclusions Structure based virtual screening was performed to identify the novel SecA inhibitors of Las. Five compounds were found to inhibit the ATPase activity of SecA of Las at nano molar concentrations. These compounds showed antimicrobial activities against A. tumefaciens with MBC ranging from 128 mg/mL to 256 mg/mL. Molecular docking and minimizations studies showed the correlation between the experimental and theoretical studies. These compounds appear to be suitable as lead compounds for further development of antimicrobial compounds against Las. 4. Experimental protocols 4.1. Computational methodology Homology model of SecA was built with Prime structure prediction of Schrodinger software [18]. Reported X-ray crystal structures in Protein Data Bank (PDB ID: 2VDA, 2FSF & 2FSG) [15,19] of E. coli SecA homodimer bound with ATP was used to build SecA homology model of Las. This model was further verified by PROSA2003 analysis. The coordinates for all reported proteins were obtained from the PDB. Structures were prepared using the Maestro [20] software package and aligned using the Protein Structure Alignment module in Prime. If a PDB structure was missing side-chain atoms, Prime was used to predict their locations. Water molecules in all structures were removed. A brief relaxation was performed on each starting structure using the Protein Preparation module in Maestro with the “Refinement Only” option. The modeled protein structure was prepared with appropriate bond orders and formal charges by protein preparation wizard of Maestro module. Then ATP ligand was manually docked as reported

924

N. Akula et al. / European Journal of Medicinal Chemistry 54 (2012) 919e924

[16] against the homology model and the proteineligand complexes were energy minimized. The receptor grid file was generated by excluding ATP and defining 8 Å radius from ATP without any constraints. Structure based virtual screening was used to screen putative SecA inhibitors from approximately 11 million small molecule compounds available from the ZINC11 database [21]. To reduce the workload into the pipeline, twenty thousand structures were selected based on the physicochemical properties e.g. Net charge (1 to þ1), H-Bond donor/acceptor (2e6), and molecular weight (300e600 Da). Glide program [22] was used to build receptor grid file and for the docking studies. Molecular minimization of ATP, highly active ligands with protein complexes were performed by Macromodel suite program [23]. OPLS [24] force field and “distance-dependent” dielectric constant were used during minimization. All the minimizations were carried out by means of 5000 iterations of PolakRibiere conjugate gradient method and followed by converge on gradient until a convergence threshold is 0.05 kJ/mol Å. All the molecular modeling studies have been performed on HP ProLiant, RedHat Linux operating system and docking postures were taken by PYMOL program [25]. 4.2. Inhibition assay against SecA of Las Cloning, expression, and purification of SecA were conducted as described previously [16]. Quantichrom ATPase/GTPase kit (Bioassay Systems, Hayward, CA) was used to test the inhibitory effect of different compounds on SecA of Las as described previously [26,27]. All these assays were done in triplicate and repeated three times with similar results. The OD values are measured from Benchmark plus ELISA micro plate spectrophotometer (Bio-Rad Hercules, CA). ATPase activities were determined by the release of phosphate ion (Pi) detected spectrophotometrically, using malachite green and inhibition was calculated by showing the percentage (%) of the remaining ATPase activities [16]. 4.3. Antimicrobial activity against A. tumefaciens A. tumefaciens was obtained from the culture collection of citrus associated bacteria initially isolated from the roots and rhizosphere of citrus and maintained in our laboratory as glycerol stock stored at 80  C [28]. Minimum bactericidal concentrations (MBC) of the selected compounds against A. tumefaciens were determined by the broth microdilution method [28,29]. Streptomycin was used as a positive control. Each small molecule compound was dissolved in DMSO and two-fold serially diluted with LB medium to give final concentrations of 8 mg/mL to 512 mg/mL. Aliquots of 100 mL were poured into 96-well microplates. A. tumefaciens was grown overnight and diluted with LB medium to give final concentrations 5  105 colony-forming unit (CFU) mL1. Samples of 100 mL were inoculated onto LB microdilution plates containing the tested compounds and incubated at 28  C for 24 h. After 24 h of incubation at 28  C, 100 mL samples from different dilutions were transferred onto antibiotic-free LB agar plates, which were then incubated at 28  C. After incubation for 2 days, the number of colonies recovered was counted. Survival rate was determined as the percentage of recovered cells compared to viable cells in the initial inoculum. MBC was defined as the lowest drug concentration producing

a survival rate of <0.1%. Experiments were carried out in triplicate and repeated three times. Acknowledgments We thank Dr. Aswathy Sreedharan and Dr. Jinyun Li for their technical help. This research has been supported by grant from the Citrus Research and Development Foundation. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.ejmech.2012.05.035. These data include MOL files and InChiKeys of the most important compounds described in this article. References [1] J.V. da Graca, Citrus greening disease, Annu. Rev. Phytopathol. 29 (1991) 109e136. [2] S.E. Halbert, K.L. Manjunath, Entomology 87 (2004) 330e353. [3] J.M. Bové, J. Plant Pathol. 88 (2006) 7e37. [4] A. Sechler, E.L. Schuenzel, P. Cooke, S. Donnua, N. Thaveechai, E. Postnikova, A.L. Stone, W.L. Schneider, V.D. Damsteegt, N.W. Schaad, Phytopathology 99 (2009) 480e486. [5] U.S. Sagaram, K.M. De Angelis, P. Trivedi, G.L. Andersen, S.E. Lu, N. Wang, Appl. Environ. Microbiol. 75 (2009) 1566e1574. [6] T.R. Gottwald, Annu. Rev. Phytopathol. 48 (2010) 119e139. [7] Y. Duan, L. Zhou, D.G. Hall, W. Li, H. Doddapaneni, H. Lin, L. Liu, C.M. Vahling, D.W. Gabriel, K.P. Williams, A. Dickerman, Y. Sun, T. Gottwald, Mol. Plant Microbe. Interact. 22 (2009) 1011e1020. [8] L. Jun, J. Xing-Vao, J. For. Res. 16 (2005) 339e342. [9] W. Chen, Y.J. Huang, S.R. Gundala, H. Yang, M. Li, P.C. Tai, B. Wang, Bioorg. Med. Chem. 18 (2010) 1617e1625. [10] M. Li, P.C. Tai, B. Wang, Biochem. Biophys. Res. Commun. 368 (2008) 839e845. [11] M.Y. Jang, S. De Jonghe, K. Segers, J. Anné, P. Herdewijn, Bioorg. Med. Chem. 19 (2011) 702e714. [12] E.H. Manting, A.J. Driessen, Mol. Microbiol. 37 (2000) 226e238. [13] A. Economou, W. Wickner, Cell 78 (1994) 835e843. [14] B. Van den Berg, W.M.J. Clemons, I. Collinson, Y. Modis, E. Hartmann, S.C. Harrison, T.A. Rapoport, Nature 427 (2004) 36e44. [15] Y. Papanikolau, M. Papadovasilaki, R.B. Ravelli, A.A. McCarthy, S. Cusack, A. Economou, K. Petratos, J. Mol. Biol. 366 (2007) 1545e1557. [16] N. Akula, H. Zheng, F.Q. Han, N. Wang, Bioorg. Med. Chem. Lett. 15 (2011) 4183e4188. [17] M. Wiederstein, M.J. Sippl, Nucleic Acids Res. 35 (2007) W407eW410. [18] Schrödinger, LLC, Schrödinger Suite 2006. Induced Fit Protocol, Prime Version 1.5, Schrödinger, LLC, New York, 2005. [19] I. Gelis, M.A. Bonvin, D. Keramisanou, M. Koukaki, G. Gouridis, S. Karamanou, A. Economou, C.G. Kalodimos, Cell 131 (2007) 756e769. [20] Schrödinger, LLC, Maestro, Version 7.5, Schrödinger, LLC, New York, 2005. [21] J.J. Irwin, B.K. Shoichet, ZINC  A free database of commercially available compounds for virtual screening, J. Chem. Inf. Model. 45 (2005) 177e182. [22] R.A. Friesner, J.L. Banks, R.B. Murphy, T.A. Halgren, J.J. Klicic, D.T. Mainz, M.P. Repasky, E.H. Knoll, M. Shelley, J.K. Perry, D.E. Shaw, P. Francis, P.J. Shenkin, J. Med. Chem. 47 (2004) 1739e1749. [23] F. Mohamadi, N.G.J. Richards, W.C. Guida, R. Liskamp, M. Lipton, J. Comput. Chem. 11 (1990) 440e467. [24] W.L. Jorgensen, D.S. Maxwell, J. Tirado-Rives, J. Am. Chem. Soc. 118 (1996) 11225e11236. [25] L.L. DeLano, The PyMOL Molecular Graphics System (2002). http://www. pymol.org/. [26] L. Denis, G.P. Guy, R.B.Andrien, 85 (1978) 86e89. [27] D.H. Richard, L.V. John, A.W. Richard, Anal. Biochem. 169 (1998) 312e318. [28] F. Kavanagh, Methods Enzymol. 43 (1975) 55e69. [29] J.P. Anhalt, L.D. Sabath, A.L. Barry, Special tests: bacterial activity of antimicrobics in combination, and detection of b-lactamase production, in: E.H. Lennette (Ed.), Manual of Clinical Microbiology, third ed., American Society for Microbiology, Washinton DC, 1980, pp. 478e484.