15
ACS SYMP0 SIUM SERIES 589 Computer-Aided Molecular Design Applications in Agrochemicals, Materials, and Pharmaceuticals Charles H. Reynolds, Editor Rohm and Haas Company M. Katharine Holloway, Editor Merck Research Laboratories Harold K. Cox, Editor Zeneca Ag Products Developed from a symposium sponsored by the Division of Computers in Chemistry and the Division of Agrochemicals at the 207th National Meeting of the American Chemical Society, San Diego, California, March 13-17, 1994 American Chemical Society, Washington, DC 1995

ACS 589 Computer-Aided Molecular Design Applicationsin Agrochemicals, Materials, and Pharmaceuticals

  • Upload
    zjelcic

  • View
    216

  • Download
    0

Embed Size (px)

Citation preview

Page 1: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

A C S S Y M P 0 S I U M S E R I E S 589

Computer-Aided MolecularDesign

Applications in Agrochemicals, Materials,and Pharmaceuticals

Charles H. Reynolds, EditorRohm and Haas Company

M. Katharine Holloway, EditorMerck Research Laboratories

Harold K. Cox, EditorZeneca Ag Products

Developed from a symposium sponsoredby the Division of Computers in Chemistry

and the Division of Agrochemicalsat the 207th National Meeting

of the American Chemical Society,San Diego, California,

March 13-17, 1994

American Chemical Society, Washington, DC 1995

Page 2: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

Supplied by U.S. Dept. of AgricultureNational Center for Agricultural

Utilization Research, Peoria, Illinois

Chapter 14

'7327

Insect Aggregation Pheromone ResponseSynergized by "Host-Type" Volatiles

Molecular Modeling Evidence for Close Proximity Bindingof Pheromone and Coattractant in Carpophilushemipterns (L.)

(Coleoptera: Nitidulidae)

Richard J. Petroskil and Roy Vaz2

lBioactive Constituents Research, National Center for AgriculturalUtilization Research, Agricultural Research Service,U. S. Department of Agriculture, Peoria, IL 61604

2Marion Merrell Dow, Cincinnati, OR 45242

The driedfruit beetle, Cmpophilus hemipterus (L.) is a worldwide pest ofavariety of fruits and grains, both before and after harvest. Attractivenessof the male-produced aggregation pheromone is enhanced by the presence ofa "host-type" volatile coattractant. A set of 26 compounds was used toexplore relationships between pheromone structure and activity by 3D­QSAR/CoMFA methods. Significant differences in aggregation pheromoneCoMFA-coefficient contour maps were observed in the presence and absenceof the "host-type" volatile coattractant.

The driedfruit beetle, Cmpophilus hemipterus (L.) (Coleoptera: Nitidulidae), attacksa large number of agricultural commodities in the field, during storage after harvestor in transport (1). It js also able to vector microorganisms responsible for thesouring of figs (1) and mycotoxin production in corn (2).

Both sexes of C. hemipterus respond to a male-produced aggregation pheromone(3). A wind tunnel bioassay guided the isolation of eleven all-E tetraenehydrocarbons, two Z-isomer tetraene hydrocarbons and one all-E triene hydrocarbon(3,4). The pheromone components were tentatively identified by spectroscopicmethods then the assigned structures were proven by synthesis (3-5). Structures ofthe synthesized compounds are shown in Figure 1. Compounds A to N have beenidentified in the C. hemipterus pheromone blend (3,4); the additional compoundswere prepared to explore structure activity relationships (4).

Previous studies have shown that aggregation pheromone activity may beenhanced when the pheromone is used in combination with attractive chemicalsproduced by the host plant or associated microorganisms, termed host-type volatilesor host-type coattractants (6-9). In order to investigate relationships between thestructure of the pheromone molecule and biological activity, as well as explorepossible additional relationships between the coattractant and pheromone structure­activity relationships, all compounds (A to Z) were tested for activity both with and

0097-6156/95/0589-0197$12.00/0© 1995 American Chemical Society

Page 3: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

198 COMPUTER·AlDED MOLECULAR DESIGN

~p

Q~

Figure 1. Hydrocarbon structures used in the Carpophilus hemipterus data set.

Page 4: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

14. PETROSKI & VAZ Insect Aggregation Pheromone Response 199

without adding a host-type coattractant (propyl acetate) to the bioassay treatments(4). The results of this previous work are summarized in Table 1.

Individual compounds are capable of eliciting the pheromonal response, asopposed to an obligate requirement for a blend of compounds (4). This observationis consistent with a hypothesis that all the structures interact with a singlerecognition site or a family of component recognition sites having conservation ofthe required bioactive conformation of the ligands at the recognition sites. Anobligate requirement for a blend of compounds would indicate the action ofdistinctly different recognition sites, each with its own structural and conformationalrequirements.

Recent advances in computational chemistry enabled us to probe quantitativestructure-activity relationships (QSAR) in three-dimensional space. We report 3DQSAR studies with the aid of Comparative Molecular Field Analysis (CoMFA)methodology (10-13). With CoMFA, a suitable sampling of the steric andelectrostatic fields surrounding a set of ligand molecules might provide all theinformation necessary for understanding their observed biological properties (13).

Materials and Methods

Data. Chemical structures (Figure 1) and the corresponding bioassay data with andwithout coattractant (Table 1) were taken from Bartelt et al. (4). The datacorresponded to a counting of the number of beetles alighting on pieces of filterpaper in a wind tunnel bioassay; two treatment preparations to be compared(pheromone versus control or pheromone plus coattractant versus control) wereapplied to pieces of filter paper, and those were hung side by side in the upwind endof the wind tunnel. The coattractant (propyl acetate) alone vs a blank filter papercontrol was only minimally attractive to C. hemipterus; relative bioassay activity wasless than 5 percent (4). Two CoMFA analyses were done, one for the data withouta coattractant (propyl acetate) and one for the data with the coattractant.

Establishing the Conformation of Each Molecule. A computation using theMOPAC (14) program and the AMI Hamiltonian was done on the sequence ofmodel structures shown in Figure 2 which shows the optimal geometries as well asthe bond orders. The doubly substituted structure is twisted more than the singlysubstituted structure. The amount of delocalization decreases as substituent methylgroups are introduced in the progression. Some conformational searching is requiredto find the low energy conformations. Hence, the single bonds in structure A wereassumed to be rotatable with a reasonable energy barrier in terms of all states beingpopulated at room temperature.

The 3D structures represented in Figure 1 were constructed using structure Afrom the figure as a template. Structure A was subjected to conformationalsearching about the rotatable bonds using the Tripos 5.2 Molecular Mechanics ForceField (10).

The minima encountered in the conformational search of compound A wereoptimized with the AMI Hamiltonian and the minimum conformation was used asthe template. If there were any extensions made, in terms of adding rotatable bondsto structure A such as in structure B, a molecular mechanics force field conforma-

Page 5: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

200 COMPUTER·AlDED MOLECULAR DESIGN

f<ORlHOGONAL VlPN-r-..,"'~

~-r" ..~JJO

Figure 2. A sequence of model structures (2E,4E,6E, 8E,-tetradecenes having 0,1, or 2methyl substituents on carbons 5 and 7) showing lower delocalization and thuslowering the rotational barrier for the single bond between carbons 5 and 6 with lessersubstitution.

Page 6: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

14. PETROSKI & VAZ Insect Aggregation Pheromone Response 201

tional search was again done on the additional rotatable bond and the minimaobtained, optimized using the AMI Hamiltonian and the energies compared,choosing the minimum energy conformation again. Conformations described hereare vapor phase. Since all the compounds examined in this study are onlyunsaturated hydrocarbons, "solvent" effect on conformations at the putativepheromone recognition site located on an insect antenna should be minimal.

Superimposing the Molecules Within a Region. Once an optimal conformationwas obtained for all the structures, the structures were then overlapped via an RMSfit using the atoms labeled with an asterisk in Figure 3. The overlapped structuresare depicted in Figure 4. A region, as shown in Figure 4, was then constructed suchthat all structures fell at least 2 N away from the region extents. The region onlyhad carbon atoms used as probes and the lowest and highest points had thecoordinates of 9.6170, 7.2766, -5.0496) and (10.4965,4.2695, 6.5532) respectively.The points were separated at intervals of 2 AO along each axis.

Comparative Molecular Field Analysis. This region containing the superimposedstructures was utilized in a Comparative Molecular Field Analysis (CoMFA)experiment (Figure 5). Normally in a CoMFA, two probes are used. One probe isa carbon atom with no charge and the other probe is a positive charge (with no massor van der Waals radius). The energy of the probe at each point of the region iscalculated using the Tripos 5.2 force field (10). The two terms of interest in theforce field are the 6-12 van der Waals terms which account for the Londondispersion forces, and the coulombic terms representing coulombic forces arisingfrom point charges. The positive charge probe was not used in this study becausethe 1t electron clouds of the unsaturated hydrocarbons in the analysis are notreflected by the charge on the carbon atoms. Points in the region where the energyof the carbon probe exceeded 30 kcal/mol were dropped from the analysis. Thebiological activities used were those listed in Table 1. CoMFA columns whosestandard deviation was less than 2.0 kcal/mol were ignored in the calculation. Thisreduced the number of columns involved in the Partial Least Squares (PLS)statistical analysis (15) substantially. Also, changing the dropped columns to thosehaving a standard deviation of less than 1.0 kcal/mol did not have any significantimpact on the statistics.

The predictive ability of both models (3D QSAR with and without coattractant)were evaluated using cross-validation in which the cross-validation was done usingas many groups as there were rows except as noted. Cross-validation involvespretending that one of the rows does not have experimental data. The resultingequation is used to predict the experimental measurement for the omitted compound.The cross-validation cycle is repeated, leaving out one different compound until eachcompound has been excluded and predicted exactly once. The resulting individualsquared errors of prediction are accumulated. The result of the cross-validation isthe sum of squared prediction errors, sometimes termed the PRESS (PredictiveResidual Sum of Squares). In PLS, the iterations are continued until the PRESS nolonger decreases significantly. Substituting PLS, which operates on all independentvariables simultaneously, for regression, which operates on one independent variableat a time, reduces the probability of accepting a chance correlation (13).

Page 7: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

202 COMPUTER·AlDED MOLECULAR DESIGN

*

Figure 3. The carbons marked by an asterisk are used to match the other structuresafter they are optimized.

Figure 4. All optimized structures aligned using the atoms marked in Figure 3.

Page 8: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

14. PETROSKI&VAZ Insect Aggregation Pheromone Response 203

point i point jActivity

anargy fait by probe at point i

A

.. .. .. .. .. + ... .. .... .. ... .. ... + + ++ .. + ..-

.. + + + + .. + + "i- ..... + .. ... .. + .. + .. ++ + +.. • + structura A ++ .. + ... + .. + ++ .. .. + + + + ... ..... + ++ ... .... + .. .. .. .. ..- .. ....

~ + ... .. .. ... ... + ff- + + ... ..+ .. .. .. ++ + + .. ... + + + .. .. ..- + + +

+ + ... .. + .. .. + .. ... ++ + ... ...+ .. + .. of

.. .. ... ... ... + + +1 + .. ... ... + + .. +f- ... + ... ... .. ..+ + .. + ... + .. ...+ + .. ... ... + .. ...... + ... + + +

B

c

+Partial Least Squares (PLS)

Activity; const .. C1·VDW1 .. C2·VDW2 + .

Figure 5. Schematic of the CoMFA sterlc field for structure A.

Page 9: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

204 COMPUTER·AlDED MOLECULAR DESIGN

Conventional R-squared values (regression range from 0 to 1; however, the cross­validated R-squared values (PLS) range from negative infInity to 1.0.

Results and Discussion

Some synthetic analogs, which were never detected from the beetles (compounds 0to Z in Table 1), showed activity in the bioassay (e.g. compounds 0, P, V and W).This observation led us to the conclusion, shared by others (16), that insectpheromone communication systems are not as rigid as once thought.

Some generalizations can be made about structure-activity relationships (4): (1)The left-hand terminal alkyl group (as drawn in structure A) should be methyl;substitution of ethyl for methyl renders the compound inactive or nearly so.Compounds E, H, J, K, and X have low or no activity. (2) The left-hand alkylbranch should be methyl, but the one example with an ethyl group at that position(compound R) did have slight activity. (3) An ethyl group as the middle alkylbranch (e.g., the 5-ethyl group in compound D) also renders the compound inactive.(4) The right-hand alkyl branch (e.g., the 7-position of compound B) can be methyl,ethyl (as in compound F), or propyl (as in compound V) and still have activity;however, only a hydrogen in that position (compounds S and T) renders thecompound inactive. (5) The right-hand terminal alkyl group can be methyl(compound A), ethyl (compound B) or propyl (compound W) and still have activity,but the ethyl group seems most consistent with high activity. (6) Alkyl groups inthe 9-position (compound Y) or in the 2-position (compound Z) greatly reduceactivity. (7) The presence of cis double bonds at any position reduces activity(compounds M, N, 0, P and Q).

Another important general feature was evident from the results shown in Table1. Relative activity was often enhanced when each unsaturated hydrocarbon wasseparately tested in the presence of the coattractant but the proportion ofenhancement varied from hydrocarbon to hydrocarbon tested. In some cases (e.g.compounds A, N, and 0), activity decreased in the presence of the coattractant.These observations revealed a relationship between the structure of the hydrocarbontested and the role of the coattractant; maximal activity in the presence of thecoattractant was observed when the right-hand terminal alkyl group was ethyl.Compounds A, N, and 0 all have methyl as the terminal alkyl group. Beyond thisobservation, it is hard to imagine a more precise role for the coattractant without useof modem computational tools.

Although insights can be acquired by looking at two-dimensional representationsof structures as are shown in Figure 1, the compounds are actually three­dimensional. A more refmed examination is gained by using modem 3D QSARmethods.

The predicted versus actual plots for the 3D-QSAR analyses with and withoutthe coattractant show that both CoMPA models are workable predictors of biologicalactivity (Figure 6). The R-squared values and other relevant statistics for bothanalyses are reasonable (Table 2). Thus, the CoMPA results also support thehypothesis of either a single pheromone recognition site or (less likely) a family of

Page 10: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

14. PEfROSKI & VAZ Insect Aggregation Pheromone Response

Prod.

28

205

-

++++

++

+ + ++

+ +++

+ + +·2 *;+

Ace

-2 28

Predicted vs actual for structures without coattractant

Prado

+.5

*

+

$~

++ +

'5Predicted vs actual for structures with coattractant

Ace

Figure 6. Plots of predicted versus actual biological activity values for structures withand without coattractant.

Page 11: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

206 COMPUTER-AIDED MOLECUlAR DESIGN

Table 1Bioassay Activity for Individual Hydrocarbons a

Relative Activity (%)

Without WithHydrocarbon Coallractant Coallractant

ABCoEFGHIJKLMNoPQASTUVWXyZ

242921o2

162ooo455

111113

76oo8

3514oo3

186041

23

498

171o55338

1746o11

411211

84

'Data from Bartelt et al., J. Chern. Eeol. 18(3) 379-402 (1992).The coattractant was propyl acetate (20 pi, 10% In minerai oil).

Table 2Partial Least Squares (PLS) Analysis of CoMFA Data

(only steric field included)

Statistics

Number of componentsStandard error of estimateA-squaredA-squared (crossvalidated')Standard error of prediction

Compounds dropped from analysis

WithoutCoallractant

43.7270.8510.4906.891

V,K,W

WithCoallractant

32.9570.9640.8116.730

U,J,X,I

'The A-squared is related to the "PRESS" via the equation:(S.D.• PRESS)/S.D. where S.D. is the sum over all moleeilesof squared deviations of each biol09ical parameter from themean and PAESS (Predictive Sum of Squares) is the sum overall molecules of the squared differences between the actualand predicted biological parameters (range is neg infinity to 1).

Page 12: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

14. PETROSKI & VAZ Insect Aggregation Pheromone Response 207

pheromone component recognition sites having bioactive conformation conservationin C. hemipterus.

The main analysis tools, in terms of computer aided molecular design, are thecoefficient plots as shown in figures 7 and 8. These plots are actually contours ofthe standard deviation times PLS coefficient [(std dev)*(coefficient)] at each pointin the region that fall in a particular range. The field is created as the point bypoint product of the PLS coefficient and the standard deviation of energies at thepoint among all compounds in the study. The view of this field is preferred to theview of only the PLS coefficients field because it reduces the visual cluster ofmoderately large coefficients that arise by chance association with larger scaletrends.

The contours are centered at -0.7 (light gray, both Figures) and 0.14 (black,Figure 7) or 0.19 (black, Figure 8) with structure A embedded in the contour plots.These contours have the same meaning as the plots in reference 3 viz. if the contouris for a region corresponding to a negative value, in that case, that region in spacewould need lower van der Waals interaction energies if a carbon probe atom wereplaced in that region or at the very most, no change would be made in that regionfor increased activity. Similarly, a positive region would prefer increased van darWaals interaction energies for a carbon probe for increased activity. This can bederived from the equation in Figure 5.

The positive and negative coefficient regions show that extending the structureA by a methylene in the direction as in structure B puts the methylene in a positivecoefficient region and similarly extending structure A by a methylene in the otherdirection, as in structure E, puts this latter methylene in a negative coefficientregion. Also, extending molecule A by a methylene such as in structure B ordifferently such as in structure C, even though both regions have positivecoefficients, their relative values are different and thus the structural extensions havedifferent consequences on the activity.

The contour plot from the analysis of the structures with a coattractant is quiteinteresting. A new, sharp, and very well-defined most-negative region seems tohave been created which could possibly be attributed to the coattractant occupyingthis region on the putative receptor. The region corresponding to the most negativecoefficient region for the analysis without the coattractant is still present in theanalysis with the coattractant.

The field times coefficient field where the field value represents the product ofthe molecule's field energy and the PLS coefficient from the appropriate analysisfrom which they were dropped between the two analyses did not lead to any activityin the new negative region for the appropriate dropped molecules, therebyeliminating this new region as arising from the outliers not used in the analysis.The (field)*(coefficient) plot represents the contribution of this field for thismolecule to its predicted activity.

The significant change in the CoMFA contour plot with coattractant versus thecorresponding CoMFA contour plot without coattractant suggests a close proximityin binding sites for the pheromone and the coattractant near the la-position of thepheromone (e.g., compound A in Figure 8) If compound B were pictured in thefigure, the new most negative area would still reside at the 10-position, which wouldbe over the methylene portion of the right-hand terminal ethyl group. An

Page 13: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

208

.'

...........

COMPUTER·AlDED MOLECUlAR DESIGN

MostNegative

Coefficient

Figure 7. The std. dey. * coefficient Co MFA contour plot for the structures withoutcoattractant showing only structure A.

MostNegative

Coefficient

\ , ..... w;.

\\

.,, .

Figure 8. The std. dey. * coefficient CoMFA contour plot for the structures withcoattractant showing only structure A.

Page 14: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

14. PETROSKI & VAZ Insect Aggregation Pheromone Response 209

alternative, but less likely, interpretation of our CoMFA data would be binding ofthe coattractant at a separate binding site but affecting the binding of the pheromone(allosterism). This placement of the coattractant would not have been possiblewithout a 3D analysis.

If C. hemipterus pheromone recognition sites have enough fluidity then it istheoretically possible to develop pheromone analogs that serve as species-specificinsect control agents. Based on our CoMFA results, it might be possible that anoxygen atom from the coattractant (propyl acetate) resides at this new most negativesite. It would be interesting to place an oxygen atom in a pheromone analog at thisposition in 3D space, but this has yet to be tested experimentally.

Carefully designed pheromone analogs, or blends thereof, might surpass thenatural pheromone in biological activity, ease of preparation, or stability under fieldconditions. Such analogs would iinprove our ability to monitor pest populations,lower pest populations by mass trapping, or lower pest populations by use ofcombinations of pheromone and either insecticides or biological control agents. Itis also theoretically possible that pherolnone perception inhibitors (antagonists) couldbe developed against C. hemipterus. Pheromone perception inhibitors could be usedfor the protection of commodities during storage or transport.

Literature Cited

1. Hinton, H. E. A Monograph of the Beetles Associated with Stored Products;Jarrold and Sons: Norwich, U.K., 1945, 443 pp.

2. Wicklow, D. T. In Phytochemical Ecology: Allelochemicals, Mycotoxins andInsect Pheromones and Allomones; Chou, C. H.; Waller, G. R., Eds.; Instituteof Botany, Academia Sinica Monograph Series No.9: Taipei, ROC., 1989,p263.

3. Bartelt, R. 1.; Dowd, P. F.; Plattner, R. D.; Weisleder, D. J. Chem. Eco!. 1990,16, 1015.

4. Bartelt, R. J.; Weisleder, D.; Dowd, P. F.; Plattner, R. D. J. Chem. Eco!. 1992,18, 379.

5. Bartelt, R. 1.; Weisleder, D.; Plattner, R. D. J. Agric. Food Chem. 1990, 18,2192.

6. Walgenbach, C. A.; Burkholder, W. E.; Curtis, M. 1.; Khan, Z. A. J Econ.Entomo!. 1987, 80, 763.

7. OeWscWager, A. C.; Pierce, A. M.; Pierce H. D. Jr.; Bprden, J. H. J. Chem.Eeo!. 1988, 14, 2071. .

8. Birch, M. C. In Chemical Ecology ofInsects; Bell, W. J.; Carde, R. T., Eds.;Sinauer Assoc.: Sunderland, Massachusetts, 1984; Chapter 12.

9. Bartelt, R. 1.; Schaner, A. M.; Jackson, 1. 1. Physiol. Entomo!. 1986,11,367.10. Clark, M.; Cramer III, R. D.; Van Opdenbosh, N. J. Compo Chem. 1989,10,

982.11. Cramer III, R. D.; Patterson, D. E.; Bunce, 1. D. J. Am. Chem. Soc., 1988,

110, 5959.12. Cramer III, R. D., DePriest, S. A., Patterson, D. E., Hecht, P. in "3D QSAR

in Drug Design: Theory and Applications"; Kabinyi, H, Ed; ESCOM, TheNetherlands, 1993, p 443.

Page 15: ACS 589 Computer-Aided Molecular  Design  Applicationsin Agrochemicals, Materials,  and Pharmaceuticals

210 COMPUTER·AIDED MOLECUlAR DESIGN

13. Cramer III, R. D., Simeroth, P., Patterson, D. E. in "QSAR: Rat ion a IApproaches to the Design ofBioactive Compounds"; SHipo, C. and Vittoria,A., Eds; Elsevier Science Publishers B. V., Amsterdam

14. MOPAC 5.0 is available from QCPE, Indiana University, Bloomington, IN.15. Cramer III, R. D., Bunce, J. D., Patterson, D. E., Frank, 1. E. Quant. Struct.­

Act. Relat. Pharmacol., Chern. Bioi. 1988, 7, 18.16. Carlson, D. A.; McLaughlin, J. R. Experientia 1982, 38, 309.

RECEIVED January 31, 1995

Reprinted from ACS Symposium Series No. 589Computer-Aided Molecular Design: Applications in Agrochemicals, Materials, and PharmaceuticalsCharles H. Reynolds, M. Katherine Holloway, and Harold K. Cox, EditorsCopyright © 1995 by the American Chemical SocietyReprinted by permission of the copyright owner