Automatical Identification of Terpenoids

8/22/2019 Automatical Identification of Terpenoids

1/20

Analytica Chimica Acta 429 (2001) 151170

Automatic identification of terpenoidskeletons through 13C nuclear magnetic resonance

data disfunctionalization

Marcelo J.P. Ferreira a, Antnio J.C. Brant a,Gilberto V. Rodrigues b, Vicente P. Emerenciano a,

a Instituto de Qumica, Universidade de So Paulo, Caixa Postal 26077, 05513-970 So Paulo, Brazilb Departamento de Qumica, ICEx, Universidade Federal de Minas Gerais,

30161-000 Belo Horizonte, Brazil

Received 11 July 2000; accepted 25 October 2000

Abstract

The proposal of this paper is to present a procedure that utilizes 13C NMR for terpenoid skeletons identification. By thisreason, a novel program named REGRAS was developed for the specialist system SISTEMAT. This program carries out ananalysis of the 13C NMR data from a given compound and, from ranges of chemical shifts, recognizes the chemical functionsexisting on specific positions of carbon skeletons. At the end of this procedure, the program matches the types of carbonatoms obtained against a database, displaying as analysis results the likely skeletons of the questioned substance. The programREGRAS wastested on skeleton elucidation of 35 compoundsfrom the most varied classes of terpenoids, exhibiting excellentresults in skeleton prevision precesses. 2001 Elsevier Science B.V. All rights reserved.

Keywords: Terpenoids; Natural products; 13C NMR; Skeletons elucidation; Expert system

1. Introduction

The nuclear magnetic resonance (NMR) techniquesfor 1H or 13C and their bidimensional variant are cur-rently the most powerful existing techniques referringto furnishment of data about an organic substance.These techniques are relevant when one treats struc-ture elucidation of new natural products, due to thegreat diversity and the structural complexity foundwithin these classes of substances [13].

Structural determination of natural products isrealized by experienced spectroscopists who, in their

Corresponding author. Fax: +55-11-38155579.E-mail address: [email protected] (V.P. Emerenciano).

majority, by analyzing a set of spectral data, trackusually the reasoning logic as follows:1. identification of the natural product class, i.e. if the

questioned substance is a terpenoid, an alkaloid, aflavonoid among others;

2. identification of the substance subclass, for exam-ple, if it is a terpenoid, if this subclass belongsto monoterpenes, sesquiterpenes, diterpenes, triter-penes or steroids;

3. determination of the substance carbon skeleton;4. functionalization of the substance skeleton, that is,

determination of the organic functions and their

respective positions on the carbon skeletons;5. determination of the relative stereochemistry of the

molecule chiral centers.

0003-2670/01/$ see front matter 2001 Elsevier Science B.V. All rights reserved.P I I : S 0003- 2670( 00) 01268- X


2/20

152 M.J.P. Ferreira et al. / Analytica Chimica Acta 429 (2001) 151170

The steps described in a and b are simple stages ofthe process, for the conditions and the solvents utilizedin the extraction process lead to restrict the problem[4]. The fundamental stage of the procedure previouslydescribed is the determination of the substance car-bon skeleton, since this forms the basic unit to whichthe substance belongs. However, the skeleton elucida-tion is not an easy task towards a great diversity ofcarbonic rearrangements which determined groups ofsubstances can exhibit, for example, in sesquiterpenesor diterpenes, where there are some hundreds of pos-sible skeletons [1].

In order to help the user to resolve these problems,numerous expert systems were developed and testedfor automatic identification of substructures [512].Throughout out last decade, we have developed anexpert system SISTEMAT [1315], whose main pur-pose is to support this user in processes of structuraldetermination of natural products. Thus, in the devel-

oped system was implanted the carbon skeleton defi-nition, so that the system by itself, during a research,can find the skeleton of the analyzed substance. Atthat rate, some classes of substance, e.g. monoter-penes, sesquiterpenes, diterpenes and triterpenes[1620] were successfully analyzed, so that mostdiverse natural product researchers could have onemore available tool for making easier the processes ofstructural determination of new compounds from theseclasses.

Fig. 1. Example of calculation of carbon atom types present in a skeleton.

In order to quicken carbonic skeleton determinationof a substance in the specialist system SISTEMAT,one has elaborated a program termed REGRAS,which carries out the analysis of 13C NMR datathrough automatic recognition of certain functionalgroups and further count of types of carbon atoms(methyl, methylene, methine and quaternary carbons)present in a substance. This process is denominateddisfunctionalization, and it is worthwhile to point outhere that this type of analysis is often mentally real-ized by experimented spectroscopists to obtain likelycandidate skeletons. The advantage of the develop-ment and utilization of this program is its speed whencompared with the specialists one, in addition to itspossible usage by unskilled people at defined classesof substances.

The objective of this paper is to introduce theprogram REGRAS, developed for the expert systemSISTEMAT, its performance methodology and the

results achieved for unprecedented substancesrecently published in the literature.

2. The program REGRAS

2.1. Methodology

At this first stage of the project SISTEMAT, wehave directed our studies to a natural product class


3/20

M.J.P. Ferreira et al. / Analytica Chimica Acta 429 (2001) 151170 153

Table 1Chemical shift ranges utilized for disfunctionalization of the 13CNMR data

Chemicalfunction

Initialmultiplicity

Finalmultiplicity

Chemical shiftranges

C=O 1 3 190.0250.0CHO 2 4 190.0250.0

RCOOR

1 4 165.5190.0C=C 1 2 113.0165.5

2 3 104.0167.03 4 100.0167.0

COH 1 2 61.0100.02 3 54.090.03 4 54.090.0

C(OR)2 1 3 100.0113.0

CH(OR)2 2 4 90.0104.0

that possesses a great variety of structures as well asa great skeleton diversity: the terpenoids. By this rea-son, we have collected from the literature thousands ofsubstances bearing the most varied types of skeletons[12,1621]. So, from these skeletons, we calculatedfor each one the number of kinds of carbons present,i.e. the whole number of carbons encompassing qua-ternary, methine, methylene and methyl carbons, asoutlined in Fig. 1.

These data were obtained for all the skeletonsobserved in the bibliographic review. From 13C NMRdata of thousands of filed substances, we have created13C NMR chemical shifts ranges usually noted for a

given functional group. These chemical shifts rangesare shown in Table 1, together with the initial multi-plicity they exhibit in a 13C NMR spectrum, and thefinal multiplicity they must present after data disfunc-tionalization. This information together with the typesof carbon extant on each skeleton were recorded ina database allied to the program REGRAS, which iswritten in PASCAL languague. In Fig. 2 is presenteda flow chart of the program. This program is availableat the ftp site address ftp://143.107.53.186/PUB andruns in PC/DOS environment.

2.2. Program performance

To typify operation and utilization of the programREGRAS, we have selected from the literature the

Fig. 2. Flowchart of the REGRAS program.

steroid whose skeleton is a withanolide, isolated fromAjuga parviflora [22], shown in Fig. 3. After databasecreation with 800 types of skeletons gathered in the lit-

erature and their respective carbon atom types, as wellas another database of defined chemical shift rangesrepresented in Table 1, we initiated data from 13CNMR entry, that is, chemical shifts and multiplicitiesof the substance in Fig. 3.

After entry of spectral data, we started the researchon skeletons by the program REGRAS, which carriesout the disfunctionalization of13C NMR data throughthe ranges displayed in Table 1, for further compari-son with the data generated around the skeleton typesobserved in the literature.

The program REGRAS, at the beginning, shows ascreen (Table 2) with the chemical shift correspond-

ing to the carbon that will be disfunctionalized and therange wherein the latter is fitted with the respectivemultiplicity; it also shows the extant function before


4/20


Fig. 3. Example of a withanolide employed for testing the program REGRAS.

the process disfunctionalization and the sort of carbonwhich will be generated after finishing the procedure.In Table 2, this step corresponds to disfunctionalizingC=O into CH2. At the end of the screen, the programstill exhibits the occurrence number of each multiplic-ity in the original spectrum and the total number ofcarbons. This process is repeated until the chemicalshifts which are found in the ranges shown in Table 1are completely analyzed.

After the whole analysis of 13C NMR data,the program displays the occurrence number of

Table 2Initial screen of the REGRAS programa

Disfunctionalization

Multiplicity Occurrence

1 102 63 84 4

No. of carbons 28a Chemical shift: 190 210.5 250; multiplicity: 1; disfunc-

tionalizing C=O into CH2.

each multiplicity in the disfunctionalized spectrum(Table 3). After realizing the research in the database,the program shows the probability of the substancebelonging to a determined skeleton; in this case, theprogram showed withanolide skeleton with a 100.0%probability. Through this result, one can verify thatthe program REGRAS furnished correctly the skele-ton of the substance exhibited in Fig. 3.

Another example that can be given is the com-plex case of structural determination of limonoids.Limonoids are highly oxidized and degraded tetra-

Table 3Final screen of the REGRAS programa

Disfunctionalization

Multiplicity Occurrence

1 22 83 124 6

No. of carbons 28a Chemical shift: 54 58.1 250; multiplicity: 3; disfunc-

tionalizing COH into CH3.


5/20


Fig. 4. Some limonoid skeletons.


6/20


Fig. 4. (Continued).

cyclic triterpenes, which constitute true challengeson process of their structural determination, dueto their high oxidation degree and skeleton diver-sity [23]. In Fig. 4 are shown the most commontypes of limonoid skeletons. To test the efficiencyof REGRAS program on this class of structures, wechoose from the literature the Elegantin A limonoid(Fig. 5), isolated from Trichilia elegans ssp. elegans[24], and put its 13C NMR data into the program.After data disfunctionalization process, the programREGRAS proposed the likely skeleton of the sub-

stance, the limonoid of type IX, according to Fig. 4,which represents the correct skeleton of the referredcompound.

3. Results and discussion

In order to evaluate the program REGRAS, we se-lected randomly from the literature five unprecedentedsubstances pertaining to each terpenoid class (total-izing 35 substances) and from their 13C NMR datatests were realized. The structures of the utilized sub-stances for testing the program REGRAS are exhib-ited in Fig. 6. The results obtained with this programare in Table 4, where one can find the number of thesubstance corresponding to the structure in Fig. 6, 13C

NMR data, the actual skeleton of the substance and theskeleton proposed by the program, and the respectivereferences. The structures of skeletons suggested by


7/20


8/20


9/20


10/20


Fig. 5. Limonoid used to test the program REGRAS.

the program REGRAS are shown in Fig. 7. It is worthnoting here to mention that the chemical shifts con-cerning the substituents are not exhibited in Table 4,once these are previously excluded by the programMACRONO, that identifies the kinds of substituentsas well as the signals belonging to the same. This pro-gram has widely been tested and evaluated in other

papers [25,26].Tests of 35 samples were realized by the programREGRAS, where, from the results displayed in Ta-ble 4, we verified that in all the cases the programproposed the right skeleton of the substance, so thatin 62.9% of the cases, this was the unique proposedskeleton, and in the remaining cases the right skeletonwas proposed together with other skeletons, being themajor proposal number observed in the substance oftest XXVII, for which were proposed as a total sevenprobable skeletons.

One can verify that in the cases in which the pro-gram REGRAS has proposed other skeletons besides

the correct one, only the types of carbon atoms presentin the substance skeleton are not enough to define onlya skeleton proposal, however, it is important to point

out that at this point of the analysis, we have a mean-ingful reduction of the likely skeletons. For example,if we analyze the substance XXVII for which the pro-gram has shown seven skeleton proposals, we can notethat there has been an extremely significant reductionof proposals, for there are 180 likely skeleton types fortriterpenes [20], therefore, resulting a clear reduction

of analysis time.This kind of analysis will furtherly be integratedwith the system PICKUP [16], which allows searchand obtaining of characteristic 13C NMR chemicalshift ranges of the skeletons, in order that, from theproposals generated by the program REGRAS, thereis a consultation, what reduces even more the skeletonprobability offered by the programs.

This methodology adapted for analysis of ter-penoids should be revised for analysis of aromaticcompounds, for instance, flavonoids, in that the 13CNMR chemical shift ranges accepted for disfunc-tionalization in program REGRAS do not allow the

treatment of two organic functions present on thesame carbon atom, what is an often observed case inaromatic compounds.


11/20


Fig. 6. Substances utilized to test the program REGRAS.


12/20




13/20




14/20




15/20


Fig. 7. Skeletons proposed by the program REGRAS.


16/20




17/20




18/20



4. Conclusions

From the results obtained with the program RE-GRAS, one can conclude that this is introduced as avaluable tool for the process of terpenoid skeleton de-termination, in that it reduces considerably the numberof skeleton proposals for a determined substance. This

program in next future, if coupled with the charac-teristic 13C NMR ranges already obtained for severalterpenoids [1619], could be utilized as a restriction

module for the structure generator that is being de-veloped for the expert system SISTEMAT, i.e. insteadof the generator working with all the skeletons of aclass of substance to start the process of generationof likely structures, it will have to initiate the processby utilizing only the skeletons previously proposed bythe program REGRAS. The immediate consequence

of the utilization of this program on to the structuregenerator will be the reduction of the computationaltime and the number of displayed structural propos-


19/20


als, what avoids the combinatorial explosion problemobserved in other specialist systems developed up tonow [6266].

Acknowledgements

This work was supported by grants from the Fun-dao de Amparo Pesquisa do Estado de So Paulo(FAPESP) and by the Conselho Nacional de Pesquisa(CNPq).

References

[1] J.D. Connolly, R.A. Hill, Dictionary of Terpenoids, 1stEdition, Chapman & Hall, London, 1991.

[2] R.A. Hill, D.N. Kirk, H.L.J. Makin, G.M. Murphy, in: F.M.MacDonald (Ed.), Dictionary of Steroids: Chemical Data,Structures and Bibliographies, 1st Edition, Chapman & Hall,

London, 1991.[3] I.W. Southon, J. Buckingham, in: G.A. Cordell, J.E. Saxton,M. Shamma, G.F. Smith (Eds.), Dictionary of Alkaloids,Chapman & Hall, London, 1989.

[4] F.J.A. Matos, in: C. Cincia (Ed.), Introduo FitoqumicaExperimental, UFC, 1988.

[5] R.K. Lindsay, B.G. Buchanan, E.A. Fergenbaum, J.Lederberg, Applications of Artificial Intelligence for OrganicChemistry: The DENDRAL Project, McGraw-Hill, New York,1980.

[6] C. Shelley, M.E. Munk, Anal. Chem. 54 (1982) 516.[7] N.A.B. Gray, Computer-Assisted Structure Elucidation,

Wiley, New York, 1986.[8] M.E. Munk, M. Farkas, A.H. Lipkis, B.D. Christie,

Mikrochim. Acta II (1986) 199.

[9] B.D. Christie, M.E. Munk, Anal. Chim. Acta 200 (1987) 347.[10] M. Carabedian, I. Dagane, J.-E. Dubois, Anal. Chem. 60(1988) 2186.

[11] K. Funatsu, S. Sasaki, J. Chem. Inform. Comput. Sci. 36(1996) 190.

[12] M. Will, W. Fachinger, J.R. Richert, J. Chem. Inform.Comput. Sci. 36 (1996) 221.

[13] J.P. Gastmans, M. Furlan, M.N. Lopes, J.H.G. Borges, V.P.Emerenciano, Qumica Nova 13 (1990) 10.

[14] J.P. Gastmans, M. Furlan, M.N. Lopes, J.H.G. Borges, V.P.Emerenciano, Qumica Nova 13 (1990) 75.

[15] V.P. Emerenciano, G.V. Rodrigues, P.A.T. Macari, S.A.Vestri, J.H.G. Borges, J.P. Gastmans, D.L.G. Fromanteau,Spectroscopy 12 (1994) 91.

[16] M.J.P. Ferreira, V.P. Emerenciano, G.A.R. Linia, P. Romoff,P.A.T. Macari, G.V. Rodrigues, Progress Nucl. Magnet.Reson. Spectrosc. 33 (1998) 153.

[17] V.P. Emerenciano, A.C. Bussolini, M. Furlan, G.V. Rodrigues,D.L.G. Fromanteau, Spectroscopy 11 (1993) 95.

[18] F.C. Oliveira, M.J.P. Ferreira, C.V. Nez, G.V. Rodriguez,V.P. Emerenciano, Progress Nucl. Magnet. Reson. Spectrosc.37 (2000) 1.

[19] S.A.V. Alvarenga, J.P. Gastmans, G.V. Rodrigues, V.P.Emerenciano, Spectroscopy 13 (1997) 227.

[20] P.A.T. Macari, J.P. Gastmans, G.V. Rodrigues, V.P.Emerenciano, Spectroscopy 12 (1994/1995) 139.

[21] C.M.B.F. Maia, R. Braz-Filho, V.P. Emerenciano, An. Acad.

Bras. Cincias 62 (1990) 119.[22] H.R. Nawaz, A. Malik, P.M. Khan, S. Ahmed, Phytochemistry

52 (1999) 1357.[23] M.F.G.F. Silva, O.R. Gottlieb, Biochem. Syst. Ecol. 15 (1987)

85.[24] F.R. Garcez, W.S. Garcez, M.T. Tsutsumi, N.F. Roque,

Phytochemistry 45 (1997) 141.[25] G.V. Rodrigues, I.P.A. Campos, V.P. Emerenciano,

Spectroscopy 13 (1997) 191.[26] M.J.P. Ferreira, F.C. Oliveira, S.A.V. Alvarenga, P.A.T.

Macari, G.V. Rodrigues, V.P. Emerenciano, Chemometr.Intelligent Lab. Syst., submitted for publication.

[27] M. Cueto, J. Darias, Tetrahedron 52 (1996) 5899.[28] A.A. Ahmed, A.A. Mahmoud, Phytochemistry 45 (1997) 533.[29] K. Machida, R. Onodera, K. Furuta, M. Kikuchi, Chem.

Pharm. Bull. 46 (1998) 1295.[30] N. Zhu, N. Sharapin, J. Zhang, Phytochemistry 47 (1998)265.

[31] J. Kitajima, C. Okamura, T. Ishikawa, Y. Tanaka, Chem.Pharm. Bull. 46 (1998) 1596.

[32] W. Zhao, G. Yang, R. Xu, G. Qin, Phytochemistry 41 (1996)1553.

[33] Y. Takeda, Y. Morimoto, T. Matsumoto, C. Ogimi, E. Hirata,A. Takushi, H. Otsuka, Phytochemistry 39 (1995) 829.

[34] Y. Chen, S. Lin, C. Chein, Phytochemistry 44 (1997) 891.[35] R.X. Tan, L.D. Kong, H.X. Wei, Phytochemistry 47 (1998)

1223.[36] H. Sudo, T. Ide, H. Otsuka, E. Hirata, A. Takushi, Y. Takeda,

Phytochemistry 46 (1997) 1231.[37] M. Tori, M. Yoshida, M. Yokoyama, Y. Asakawa,

Phytochemistry 41 (1996) 977.

[38] K. Siems, F. Weigt, E. Wollenweber, Phytochemistry 41(1996) 1119.

[39] N. Gren, S. Kirmizigl, C. Zdero, Phytochemistry 44 (1997)311.

[40] K. Nabeta, S. Ohkubo, R. Hozumi, Y. Fukushi, H. Nakai, K.Katoh, Phytochemistry 43 (1996) 83.

[41] A.A. Ahmed, M.A.A. Elela, Phytochemistry 51 (1999) 551.[42] M.R. Kim, B.R. Suh, J.G. Kim, Y.H. Kim, D.K. Kim, D.C.

Moon, Phytochemistry 52 (1999) 113.[43] A.M. Helal, N. Nakamura, H. El-Askary, M. Hattori,

Phytochemistry 53 (2000) 473.[44] P. Torres, J. Ayala, C. Grande, J. Anaya, M. Grande,

Phytochemistry 52 (1999) 1507.[45] G.M.S.P. Guilhon, A.H. Mller, Phytochemistry 49 (1998)

1347.

[46] B. Muckensturm, F. Diyani, D.L. Noun, S. Fkih-Tetouani,J. Reduron, Phytochemistry 44 (1997) 907.[47] C. Che, T. Zhou, Q. Ma, G. Qin, I.D. Williams, H. Wu, Z.

Shi, Phytochemistry 52 (1999) 117.


20/20


[48] M. Ono, Y. Ito, T. Nohara, Phytochemistry 48 (1998) 207.[49] P. Tuntiwachwuttikul, N. Boonrasri, J.B. Bremner, W.C.

Taylor, Phytochemistry 52 (1999) 1335.[50] J. Jakupovic, F. Jeske, T. Morgenstern, F. Tsichritzis, J. A

Marco, W. Berendsohn, Phytochemistry 47 (1998) 1583.[51] Q. Shi, T. Oritani, T. Sugiyama, T. Yamada, Phytochemistry

52 (1999) 1571.[52] Y. Kimura, T. Akihisa, K. Yasukawa, S. Takase, T. Tamura,

Y. Ida, Chem. Pharm. Bull. 45 (1997) 415.[53] N.D. Tommasi, L. Rastrelli, M.R. Lauro, R. Aquino,

Phytochemistry 49 (1998) 1123.[54] S. ksz, S. Serin, Phytochemistry 46 (1997) 545.[55] J.O. Neto, M.F.G.F. Silva, E.R. Fo, J.B. Fernandes, P.C.

Vieira, A.L. Pinheiro, Phytochemistry 49 (1998) 1369.[56] K. Takeya, H. Kobata, A. Ozeki, H. Morita, H. Itokawa,

Phytochemistry 48 (1998) 565.[57] A. Rahman, M. Shabbir, M. Yousaf, S. Qureshi, D. e-Shahwar,

A. Naz, M.I. Choudhary, Phytochemistry 52 (1999) 1361.

[58] A. Inada, H. Murata, Y. Inatomi, T. Nakanishi, D. Darnaedi,Phytochemistry 45 (1997) 1225.

[59] K. Vokv, M. Budesinsky, J. Harmatha, J. Kohoutova,Phytochemistry 49 (1998) 2109.

[60] V.U. Ahmad, S.M. Khaliq-uz-Zaman, S. Shameel, S. Perveen,Z. Ali, Phytochemistry 50 (1998) 481.

[61] Y. Mimaki, T. Satou, M. Kuroda, Y. Sashida, Y. Hatakeyama,Phytochemistry 51 (1999) 567.

[62] R.E. Carhart, D.H. Smith, N.A.B. Gray, J.G. Nourse, C.Djerassi, J. Org. Chem. 46 (1981) 1708.

[63] R. Attias, J. Chem. Inform. Comput. Sci. 23 (1983) 102.[64] M.E. Munk, J. Chem. Inform. Comput. Sci. 38 (1998)

997.[65] M. Jaspars, Nat. Prod. Rep. 16 (1999) 241.[66] M. Badertscher, A. Korytko, K.P. Schulz, M. Madison,

M.E. Munk, P. Portmann, M. Junghans, P. Fontana,E. Pretsch, Chemometr. Intelligent Lab. Syst. 51 (2000)73.

Automatical Identification of Terpenoids

Documents