Structure dereplication natural products information

Using this approach for natural products dereplication, data are routinely obtained from 40 gg of crude extract. Performance examples include the identification of 16 analogs of teicoplanin and 12 analogs of phenelfamycin from separate samples. The summary of results obtained for phenelfamycin is shown in Table 6.4. The correlation of fraction, retention time, and molecular weight provides the essential information for rapid dereplication and identification. The time required to dereplicate natural product samples is about 1 week with this LC/MS-based method compared to several weeks by previous methods that involve traditional isolation steps. The use of this LC/MS-based methodology results in greater clarity and confident decisions for proceeding with the full structural study of an active component derived from a culture. 

When a new impurity is encountered during chemical process research, retention time and molecular weight information are compared to the database for rapid identification. This approach is similar to the procedure described for natural product dereplication. If the compound is not contained in the structure database, then the corresponding LC/MS/MS analysis is performed to obtain substructural detail and the proposal of a new structure. 

Bradshaw et al. have disclosed a rapid and facile method for the dereplication of a purified natural product library [58]. The method integrates the molecular weight from low resolution MS data with the exact count of the number of methyl, methylene and methine groups obtained from ]H-13C NMR correlation data. Those structure properties were converted into a searchable text file that could be downloaded into a customized software program with chemical structure information in a specific format - SMILES. In the program, more than 120,000 unique structures were derived from commercial databases, such as Dictionary of Natural Products and Beilstein. 

There are numerous natural product databases that are used in dereplication such as Chemical Abstracts Service (CAS), Berdy s Bioactive Natural Products Database, Dictionary of Natural Products, MarinLit23 and AntiBase.24 Integration of structure-based and chemical shift searches, such as NAPROC-13,25 make these databases even more valuable. AntiMarin is a result of a recent merger between AntiBase and MarinLit 26 MarinLit and AntiMarin are the first commercially available databases that contain structural information (functional groups) as searchable fields.27... 

The study of natural products in plant extracts is an interesting challenge to LC-MS. Generally, the relevant compounds must be detected as minor components in complex mixtures. A combination of LC separation, especially to resolve isomeric stractures, and MS detection is needed. Furthermore, structural information is needed for the identification and dereplication of the unknown plant constituents. Because of the complexity of the sample pretreatment procedures involved in the isolation, MS in most cases is the only applicable spectrometric technique too much of a purified component would be needed for IR and NMR analysis. On-line analysis in relatively erode samples is obligatory for the detection of minor constitnents. When electrospray ionization (ESI) or atmospheric-pressnre chemical ionization (APCl) are applied for analyte ionization, structural information mnst be obtained by application of colhsion-indneed dissociation (CID), either via in-sonree CID or preferably via MS-MS or MS". LC-MS and LC-MS-MS have proved to be extremely snccessful in this area. 

Since the Golden Age of Antibiotics in the 1950s, natural products chemists have faced the steadily increasing problem of how to maximize the discovery of new compounds and minimize the re-evaluation of natural products already described in the literature. If a compound has been isolated, identified, and reported, it should be possible to use the published information to identify the compound when it appears again, without having to repeat the entire isolation and structure-determination process. In addition, in many instances, the questions being asked in a study can be answered simply by partial identification of the unknown structure. These complementary processes of rapid identification of known compounds from a partially purified mixture and identification of enough of an unknown structure to prioritize or conclude an isolation, have come to be termed dereplication by the natural products community. 

In order to carry out dereplication procedures, a number of general characterization techniques are routinely employed to build information about the natural product that can be compared with other characterized compounds or sets of compounds. These techniques include chromatographic and other separation methods, chemical, spectroscopic, and biological approaches. Essentially, the process involves a series of questions at each stage of the extraction based on characteristics of the compoimd that allow it to be assigned to either a known compound, a general class of compounds, or an unknown structure. No single piece of data is... 

To illustrate how useful the unit-molecular-weight information is in natural product dereplication, Fig. 3A shows the unit-molecular-weight distribution of the approx 78,000 known natural products contained in the Chapman and Hall database (32). It can be seen that, for the reported natural products, determination of the molecular weight will narrow the reported structural possibilities to <800, and the largest number of known natural products that share a common unit molecular weight is only 759 (mol wt = 318). For that molecular weight, there are known natural products that exhibit 45 different empirical (or molecular) formulas, each of which adds up to approx 318 Dalton. 

LC/NMR in various combinations with LC/UV-DAD, LC/MS, LC/MSMS, LC/IR, and/or LC/CD has been used in many applications related to the online identification of natural products. In this field, the challenge for hyphenated techniques is important since often the characterization of completely unknown molecules is required in very complex biological matrices. In this case, LC hyphenated techniques are used for the chemical evaluation of biologically active fractions or extracts and for dereplication purposes. As full structure assignment is often needed, all online spectroscopic data are taken into consideration. Most applications are performed in the stop-flow mode and extensive 2D NMR correlation experiments are measured. For unknown online determination the need for data is often mandatory. This type of information can be deduced from HSQC and HMBC indirect measurements and very recently it has been demonstrated that even direct measurements were possible in a crude plant extract. For this application the LC peak of interest was preconcentrated by trapping on SPE and the measurement was performed on a cryogenic flow... 

---