From plant or marine organism to purified bioactive compound: liquid-liquid extraction, flash chromatography, HPLC isolation, structural elucidation by NMR and MS
Natural products — secondary metabolites produced by plants, bacteria, fungi, and marine organisms — have been the source of over half of all approved drugs since 1981. These chemically diverse molecules represent millions of years of evolutionary optimization for biological activity: binding enzyme pockets, disrupting membrane structures, blocking protein-protein interactions. Today, natural product extraction has evolved from crude decoctions to a sophisticated analytical science integrating metabolomics, bioassay-guided fractionation, and AI-powered structure prediction.
Natural products are chemically categorized into four major biosynthetic classes:
1. Alkaloids (nitrogen-containing): • Biosynthesis from amino acid precursors: tryptophan (indole alkaloids: vinblastine, quinine, strychnine), tyrosine (morphine, colchicine, ephedrine), lysine (anaferine), ornithine (atropine, cocaine) • Ecological role: anti-herbivory, anti-microbial, allelopathic (competitive exclusion of plants) • Drug examples: - Vinblastine/vincristine (Catharanthus roseus): tubulin depolymerization → first vinca alkaloids (1958) - Morphine (Papaver somniferum): opioid receptor agonist → analgesic (1804, first pure plant alkaloid) - Quinine (Cinchona bark): heme polymerization inhibition → antimalarial (1820) - Colchicine (Colchicum autumnale): gout treatment, anti-mitotic (500 BC Ebers Papyrus)
2. Terpenoids/Terpenes (isoprenoid pathway): • Built from C5 isoprene units via MEP (plastid) or MVA (cytoplasm) pathway • Monoterpenes (C10): menthol, limonene; Sesquiterpenes (C15): artemisinin; Diterpenes (C20): taxol; Triterpenes (C30): steroids, saponins; Tetraterpenes (C40): carotenoids • Drug examples: - Taxol/paclitaxel (Taxus brevifolia): tubulin hyperstabilization → antineoplastic (1992) - Artemisinin (Artemisia annua): sesquiterpene endoperoxide → antimalarial (Tu Youyou Nobel 2015) - Camptothecin (Camptotheca acuminata): topoisomerase I inhibitor → irinotecan derivative
3. Polyketides (malonyl-CoA pathway): • Assembled by polyketide synthase (PKS) enzyme megacomplexes • Include tetracyclines, macrolides (erythromycin), antifungals (amphotericin B), lovastatin, rapamycin • Rapamycin (Streptomyces hygroscopicus, Easter Island soil): mTOR inhibitor → transplant immunosuppression + cancer • Epothilones (myxobacteria): clinical compounds ixabepilone, patupilone — taxol-like mechanism
4. Non-ribosomal peptides (NRPSs): • Cyclic or branched peptides assembled by NRP synthetase enzyme complexes • Vancomycin, penicillin, cyclosporin, daptomycin, bleomycin • Often contain non-standard amino acids, D-amino acids, N-methylations — resistant to proteases
Sequential extraction strategy for alkaloid isolation: 1. Drying (50°C, forced air) → kill enzymes, reduce weight, concentrate 2. Grinding (2mm mesh) → surface area increase → extraction efficiency 3. Hexane extraction: removes waxes, chlorophyll a/b, non-polar terpenoids 4. EtOAc extraction: mid-polarity — extracts most alkaloids at neutral pH 5. Acid-base partition (pH 3 acid, pH 10 base): alkaloids (basic) → ionized at pH 3 (aqueous), free base at pH 10 (organic) — purifies away neutral compounds 6. MeOH:H₂O 80:20 — polar glycosides, phenolic acids, saponins
Flash chromatography (Still et al., 1978) uses positive pressure to accelerate silica gel column chromatography by a factor of 10–20 × gravity-driven systems. A procedure that took 12 hours under gravity now completes in 20 minutes. Modern systems (Teledyne ISCO, Biotage Isolera) are fully automated with fraction collection, UV/ELSD detection, and gradient programming — enabling gram-to-kilogram scale isolation of natural products.
Chromatographic separation is governed by differential adsorption between the stationary phase (silica) and the mobile phase (solvent):
Normal Phase Silica Gel Chromatography: • Stationary phase: porous silica (SiO₂) with surface silanol groups (Si-OH), polar • Mobile phase: non-polar to moderately polar solvents (hexane, DCM, EtOAc, MeOH) • Retention: more polar compounds → more H-bonding with silanol → longer retention • Elution order: non-polar first → polar last • Gradient: hexane 100% → EtOAc → MeOH (increasing polarity) • Silica particle size: irregular 40–63 μm for flash; spherical 15–40 μm for high-res prep
Reverse Phase C18 Chromatography (HPLC): • Stationary phase: silica with C18 alkyl chains (octadecylsilyl, ODS) — hydrophobic • Mobile phase: water + organic modifier (MeOH, ACN) — polar • Retention: more hydrophobic compounds → more Van der Waals with C18 → longer retention • Elution order: polar/hydrophilic first → hydrophobic last • Gradient: 10% organic → 95% organic (increasing hydrophobicity) • HPLC particle: spherical 3–10 μm (flash); 1.7–3 μm (UHPLC)
Detection during chromatography: • UV at 254nm: aromatic compounds (most alkaloids, flavonoids, coumarins) • UV at 280nm: phenolics, indoles (Trp-containing alkaloids) • ELSD (evaporative light scattering): universal — detects non-UV-absorbing compounds (steroids, saponins, carbohydrates) • Dereplication: LCMS in real-time identifies known vs. new compounds by m/z + UV profile comparison to Dictionary of Natural Products
Bioassay-guided fractionation: • Each collected fraction tested for biological activity (e.g., MTT cytotoxicity assay, antimicrobial MIC) • Active fractions merged and subjected to next chromatographic step • GNPS (Global Natural Products Social Molecular Networking): molecular network of all MS/MS spectra in fractions → clusters compounds by structural similarity → identifies which cluster contains bioactivity • FBMN (Feature-Based Molecular Networking): links GNPS clusters to chromatographic features → direct structure-activity correlation during fractionation
Thin Layer Chromatography (TLC) monitoring: • Silica gel 60 F₂₅₄ plates (Merck): quick check of fraction composition • UV 254nm: aromatic compounds quench fluorescence (dark spots) • UV 365nm: fluorescent compounds apparent • Dragendorff reagent (BiI₃/KI/acetic acid): alkaloids → orange precipitate (positive) • Cerium sulphate/H₂SO₄: universal charring reagent (all organic compounds → black spots after heating) • Rf value = distance spot / distance solvent front: identifies compound by comparison to standards
Preparative high-performance liquid chromatography (prep-HPLC) is the gold standard for obtaining pure natural products at milligram to gram scale for biological testing, structural elucidation, and preclinical studies. A modern prep-HPLC system can separate 100mg–5g of crude material per injection and generate >99% pure fractions ready for NMR and bioassay — the critical step between a chromatographic fraction and a validated bioactive compound.
Preparative HPLC system components and optimization:
1. Column selection: • C18 reverse phase: most universal for natural products (alkaloids, terpenoids, polyketides) • Column ID: 21.2mm (semi-prep, ~100mg capacity), 30mm (prep, ~1g), 50mm (large prep, ~5g) • Column length: 150–250mm (longer = better resolution, slower) • Particle size: 5–10 μm (prep); 10 μm optimal balance of resolution and back-pressure • Brand examples: Phenomenex Luna C18, Waters XBridge, Agilent Zorbax
2. Mobile phase optimization: • Solvent A: water (+ 0.1% FA or TFA for basic compounds protonation) • Solvent B: MeOH or ACN (ACN = lower viscosity, lower backpressure, but higher UV cutoff at 210nm) • Acid modifier: 0.1% TFA → protonates basic alkaloids, improves peak shape, prevents tailing • Ammonium format/acetate: pH 3–6 buffers for better retention reproducibility • Resolution (Rs) = 1.18 × (tR2 - tR1) / (W1 + W2); Rs > 1.5 = baseline separation
3. Example gradient optimization for vinblastine: • Initial: 40% MeOH (vinblastine retention too long at 60–70% MeOH) • Gradient: 40% → 95% MeOH over 45 min at 30 mL/min • Vinblastine elutes at 28.3 min (0.1% TFA modifier), leurosine at 32.1 min • Without TFA: both alkaloids tail severely (basic amine sticks to silanol residuals) • UV detection: 254nm primary (indole chromophore), 280nm secondary
4. Fraction collection strategies: • Time-based: collect fixed time windows — simple, works if peaks well-separated • Peak-triggered: detector threshold triggers fraction collector → collects exactly peak envelope • Mass-directed (LC-MS): real-time MS identifies target m/z → triggers collection for that compound only Most advanced: no UV chromophore required; LCMS confirms identity simultaneously
5. Fraction analysis & pooling: • Analytical HPLC of each fraction: inject 5μL, run 10 min gradient → determine purity % • Pool fractions with >95% purity → concentrate on rotary evaporator • Freeze-drying (lyophilization): for aqueous fractions + heat-sensitive compounds • Final QC: analytical HPLC purity, HRMS confirmation, residual solvent NMR check
6. Chiral HPLC (for enantiomers): • Polysaccharide chiral stationary phases (Chiralpak IA, IB, IC, AD, AS) • Many natural products: single enantiomer (correct absolute configuration biologically critical) • Example: taxol (2S,3R) vs. epimer: 1000× difference in tubulin binding • Chiral separation by prep-HPLC preparative scale: 10–50g/day possible
Modern structure elucidation of a natural product typically requires only 1–5 mg of material and takes 1–3 days with a 600 MHz NMR spectrometer and high-resolution mass spectrometer. The combination of ¹H and ¹³C NMR, 2D correlation experiments (COSY, HSQC, HMBC, NOESY), and HRMS can determine the complete connectivity and relative stereochemistry of even complex polycyclic molecules with molecular weights up to ~2000 Da.
NMR structure elucidation protocol for vinblastine (MW 811, C₄₆H₅₆N₄O₉):
1. ¹H NMR (600 MHz, CDCl₃): • Chemical shifts (δ): protons in different environments resonate at characteristic frequencies • Multiplicity: coupling to adjacent H (J constants: geminal 2J, vicinal 3J) • Integration: proportional to number of protons • Key signals in vinblastine ¹H NMR: - δ 9.12 (s, 1H): indole N-H - δ 7.82, 7.20, 7.10 (indole aromatic protons) - δ 3.58 (s, 3H): methyl ester (OCH₃ on vindoline portion) - δ 3.43 (s, 3H): N-methyl on vindoline tertiary N - δ 0.84, 0.73 (t, 3H): ethyl group CH₃
2. ¹³C DEPT-135: • CH and CH₃ point up; CH₂ point down; quaternary C absent • Identifies each carbon as CH₃, CH₂, CH, or C • 46 carbons in vinblastine: 9 CH₃, 8 CH₂, 18 CH, 11 quaternary
3. COSY (Correlated Spectroscopy): • H-H coupling through bonds (3J mainly) • Maps connectivity between adjacent protons → identifies spin systems • Example: ethyl group CH₃ CH₂ at δ 0.84 coupled to CH₂ at δ 1.82 COSY shows connectivity
4. HSQC (Heteronuclear Single Quantum Coherence): • Directly bonded H-C correlations (1J_CH) • Assigns each ¹H signal to its carbon → builds H-C pairs • Combined with DEPT: each peak labeled as CH, CH₂, or CH₃
5. HMBC (Heteronuclear Multiple Bond Coherence): • Long-range H-C correlations (2J and 3J) • Connects protons to carbons 2–3 bonds away — CRITICAL for: - Connecting spin systems separated by quaternary C - Identifying carbonyl C (no H attached) - Confirming ring junctions and bridgehead connectivities • Example: methyl ester OCH₃ (δ_H 3.58, δ_C 51) shows HMBC correlation to ester C=O (δ_C 171)
6. NOESY (Nuclear Overhauser Effect Spectroscopy): • Through-space correlations: protons within 5Å show cross-peaks • CRITICAL for relative stereochemistry determination • Axial H shows strong NOE to axial neighbors; equatorial to equatorial • For vinblastine: distinction of α/β face; C3′H vs. C5'H orientation → absolute configuration from known
7. HRMS (High-Resolution Mass Spectrometry): • ESI-Orbitrap: [M+H]⁺ = 812.4167 measured • Theoretical: C₄₆H₅₆N₄O₉ + H = 812.4172 (error: 0.6 ppm) • Unambiguous molecular formula determination: eliminates isomers at same nominal mass • MS/MS fragmentation: diagnostic ions at m/z 355.2 (catharanthine portion), 297.1 (vindoline portion) → "dimer" structure of catharanthine + vindoline confirmed by diagnostic fragments
The complete de novo structure elucidation of a complex new natural product — a fully unknown compound never seen before — typically requires 10–50 mg of material, 1–2 weeks of NMR work, and team collaboration between the isolation chemist and a computational NMR specialist. The real challenge is relative and absolute stereochemistry: a molecule like taxol (11 stereocenters = 2048 possible stereoisomers) requires NOESY analysis, single-crystal X-ray diffraction (SCXRD), or electronic circular dichroism (ECD) comparison to unambiguously confirm the correct absolute configuration.
Once a pure natural product is in hand, it enters a systematic biological evaluation pipeline. The NCI 60-cell panel, established in 1990, screens every compound against 60 cancer cell lines from 9 tissue types in a single 48-hour assay — generating a tumor-specific activity fingerprint that can identify mechanism of action by pattern recognition. Semi-synthesis then allows selective modification of the natural scaffold to improve potency, reduce toxicity, and establish SAR without total synthesis.
NCI 60-cell panel (Developmental Therapeutic Program):
Panel composition (9 tumor types × ~7 cell lines each): • Leukemia: CCRF-CEM, K-562, MOLT-4, HL-60, SR, RPMI-8226, U266 • Lung (NSCLC): A549, EKVX, HOP-62, HOP-92, NCI-H226, NCI-H322M, NCI-H460 • Colon: COLO-205, HCC-2998, HCT-116, HCT-15, HT29, KM12, SW-620 • CNS: SF-268, SF-295, SF-539, SNB-19, SNB-75, U251 • Melanoma: LOX-IMVI, MALME-3M, M14, SK-MEL-2, SK-MEL-5, SK-MEL-28, UACC-257 • Ovarian: IGROV1, OVCAR-3, OVCAR-4, OVCAR-5, OVCAR-8, NCI/ADR-RES, SK-OV-3 • Renal, Prostate, Breast: similarly 6–7 lines each
Assay protocol: • Cells seeded at ~10,000/well in 96-well plates; 24h attachment • Compound: 5-point dose series (10 pM – 100 μM); 48h incubation • Endpoint: SRB (sulforhodamine B) protein staining → OD at 540nm • GI₅₀: 50% growth inhibition; TGI: total growth inhibition; LC₅₀: 50% lethality • COMPARE algorithm: correlates activity pattern to all 100,000 screened compounds → identifies MOA
FINGERPRINT PATTERN interpretation: • Vinblastine pattern: most potent in leukemia, lymphoma cells; relatively spared solid tumors • Taxol pattern: most potent in breast, ovarian, lung — distinct from vinblastine • If new compound matches vinblastine pattern → tubulin depolymerization mechanism predicted • Matches camptothecin → TopoI inhibitor predicted; DNA alkylator pattern → distinct
Semi-synthesis of vinca alkaloids: Vinblastine → semi-synthetic analogs via selective chemical modification:
1. N-Desformylation + N-acylation of catharanthine C-16/17 double bond: → Vinorelbine (Navelbine): anhydrovinblastine intermediate + peroxide oxidation → Less neurotoxicity than vinblastine (different ring system), better oral bioavailability → FDA approved 1994, still standard-of-care NSCLC
2. C-23 functionalization: → Vindesine: amide nitrogen modification → lymphoma activity pattern shifts → Less myelosuppression than vincristine
3. Key SAR findings from vinca alkaloid series: - N-methyl on vindoline (C-3 N): ESSENTIAL for anti-tubulin activity (demethyl = 500× less potent) - C-16 ester group: tolerated modifications (desacetyl = 4× less); acetyl optimal - Catharanthine C-20 → C-20 hydroxyl: 10× potency increase (20-OH vincristine) - Ring contraction (vindorosine scaffold): maintains activity, different selectivity
Total synthesis considerations: • Vinblastine total synthesis (Fukuyama 2009; 31 steps): elegant but not commercially viable • Commercial: extraction from C. roseus remains the only scalable source • Metabolic engineering: yeast expression of vindoline/catharanthine biosynthetic pathway (Mikanagi 2019: full 31-enzyme pathway in S. cerevisiae → mg/L titers, demonstration scale) • Plant cell culture: Catharanthus roseus cell fermentation → lower yield than whole plant but sustainable