DEREPLICATION OF PENICILLIUM BIOACTIVES USING MULTIDETECTOR HPLC AND BIOACTIVITY PROFILES Thomas G. McCloud,* Paul Klausmeyer,* and David J. Newman# *Natural Products Support Group, SAIC Frederick Cancer Research and Development Center

and #Developmental Therapeutics Program, NCI, Frederick, MD 21702 USA

Presented at the American Society of Pharmacognosy Annual Meeting – July 22-26, 2000, Seattle, WA

ABSTRACT

The NCI drug discovery program has, among >150,000 natural products extracts, over 18,000 extracts obtained from fungal fermentations. Over 80 different species of Penicillium are included in the collection. Members of the diverse genus Penicillium are known producers of bioactive substances, such as the natural penicillins, penicillic acid, wortmannin, and others, which will be repeatedly detected in screening programs. Thus, the natural products chemist in a drug discovery program will be presented with the same bioactive extracts, as detected with different bioassays, and faced with the challenge of determining whether the active substance is a known molecule or a new chemical entity.

For example, 28 toxic extracts from Penicillium sp. were “fingerprinted” using a multidetector HPLC system which included diode array, laser light scattering, fluorescence, and mass spectrometer. Column eluate was collected in microtiter format for bioassay. Comparisons of both physical and biological data can then be made to assess the presence or absence of known bioactive compounds in the crude extract.

INTRODUCTION

The NCI drug discovery program at FCRDC includes a search for new lead compounds from plant, marine and fungal extracts as anticancer, antiviral and antimicrobial agents. Among the library of >150,000 crude natural products extracts are >18,900 extracts of fungal origin, and among these are 1,987 extracts from cultures identified as genus = Penicillium. Ninety different viable cultures identified as Penicillium are represented in the DTP repository.

The genus Penicillium has about 150 described species, and these various organisms are well known as producers of a variety of bioactive substances (i.e., the natural penicillins, penicillic acid, wortmannin, gliotoxin, duclauxin, cyclopiazonic acid, etc.) that will be detected by many different sorts of bioactivity screens. The Berdy Antibiotic Database lists 602 bioactive compounds with Penicillium as the producing organism. But even when taxonomy has determined that two fungal isolates are identical to the species level, they do not necessarily produce the same secondary metabolites. Possible genetic differences within the species, variable culture conditions, and composition of the growth media all affect the mixture of metabolites produced by the fungus. Thus, the natural products chemist in a drug discovery program will be repeatedly presented with bioactive extracts from Penicillium cultures, and be faced with the challenge of determining whether the active substance is a known molecule or a new chemical entity. Previously, chemists have utilized UV profiles of bioactive substances (1), HPLC retention tables (2), and chromatographic plus bioactivity (3,4) methods to address the identification of compounds found in fungal extracts.

The technique we have been developing for dereplication of extracts is construction of a “fingerprint” of the extract composed of both spectral and biological data through use of a multidetector HPLC system which includes diode array, evaporative laser light scattering, fluorescence and mass spectrometer. Physical data relating to retention time, UV absorption profile, weight of component in the extract, and mass of the molecule are collected simultaneously. Acquiring these measurements consumes 5% of the column eluate with the remaining 95% being collected into microtiter plates. Microtiter plates are replicated with each well (fraction) tested for biological activity. Biological and physical information are then correlated and determinations can be made involving the presence (or absence) of known bioactive compounds in these extracts.

MATERIALS AND METHODS

a) PRODUCTION OF FUNGAL EXTRACTS

In the NCI screening program fungal cultures are typically grown in three media: potato dextrose broth (PDB) stationary, soy-glucose-starch (SGSM) agitated, and glucose-sucrose-fructose (GSF) agitated. Whole broth is homogenized and extracted by partitioning against dichloromethane. Following drying and weighing, the extract is prepared for screening in microtiter plate format, with either 50 µg/well (turbidometric measurement of growth inhibition) or 500 µg (filter paper disc) for disc diffusion bioassay.

b) CHROMATOGRAPHIC ANALYSIS

The multidetector HPLC system consists of a Waters 600 pump controlled by Waters Millennium 32 software and a Waters 2700 autosampler which allows for samples to be loaded in either vials or microtiter plate format. Eluate from the column is split initially 115:1 with a high precision LCP splitter which directs 0.8% volume to a Kratos fluorescence detector, a Waters 996 Photodiode Array Detector, and a Waters Micromass API Electrospray Mass Spectrometer. The bulk of the flow goes to a 25:1 LCP splitter, with the major flow directed to an ISCO Foxy fraction collector which collects in microtiter plate format. The remainder of the flow is directed into a Varex Evaporative Light Scattering Detector. This arrangement gives 95% sample recovery.

For routine analysis, 50 mg of crude extract is dissolved in acetonitrile/water (1:1) at 10 mg/ml and injected onto a 250 mm x 21 mm Dynamax 60A C-18 phase bonded column eluted with acetonitrile/20 mM ammonium acetate pH 4.0 at 15 ml/min under gradient conditions 0-5 min (30:70), 25 min (50:50), 45-60 min (100:0). A typical total volume of eluate is 645 mls, which is collected into 4 x 24 well microtiter plates. These are reformatted by a TECAN High Speed Robotic Liquid Handling System into two 96 well master microtiter plates with 2 ml in each well, plus 15 replicate daughterplates at 250 µl in each well. Three of these daughter plates are 96 well polypropylene flat bottom containing one 5 mm filter paper disc in each well. All plates are dried for several hours in a Savant Speed Vac to remove the majority of the organic solvent, then placed into dry ice, and when solidly frozen, into a freeze drier. Once thoroughly dried, these fractions are suitable for biological evaluation, while the bulk of the sample mass is retained in 96 well plate format for further chromatography or spectroscopy. All the data is captured electronically, and following biological evaluation, a combined physical/biological data chart is assembled.

c) BIOASSAY

1) Of 297 organic solvent extracts from fermentations of Penicillia which were evaluated during a high throughput screen against an azole resistant strain of Candida albicans, 44 showed toxicity. A subset (20) of these were selected for chromatographic analysis to test the applicability of HPLC/bioactivity fingerprinting to the task of dereplication.

2) As stated above, an aliquot of the column eluate is transferred directly into microtiter plates. Thus, the exact mass applied to each test disc is not known. Disc diffusion assays were performed by transferring each 5 mm filter disc from the well of the vacuum dried polypropylene plate onto the surface of a 245 x 245 mm Petri dish to which one of three wild type organisms had been added prior to pouring: Candida albicans (fungus), Bacillus subtilis (gm+), or Escherichia coli (gm–). Typically, the paper discs from one 96 well microtiter plate were placed onto one 245 x 245 agar plate in the same 8 x 12 format as found on the plate. Appropriate antibiotic containing standard discs were also applied. After 24 hrs incubation at 30°C, the zones of inhibition were scored manually, and normalized to the standards. Additionally, selected fractions were evaluated in an azole resistant strain of Candida albicans 99-778 and a methicillin resistant strain of Staphlococcus aureus grown in liquid culture in microtiter plate format, with inhibition of growth determined by optical density after ~20 hrs.

Toxicity of Standard Compounds Isolated from Genus Penicillium Minimum Inhibitory Dose (µg)

Test Organism

Compound MF/MW C. albicans 99-778 C. albicans W E. coli W B. subtilis W

Brefeldin A C16H24O4MW280 50 >50 Citreoviridin C23H30O6MW402 >50 >50 >50 6 Citrinin C13H14O5MW250 12 >50 >50 50 Cyclopiazonic Acid C20H20N2O3MW336 50 >50 >50 50 Duclauxin C29H22O11MW546 50 >50 >50 Emodin C15H10O5MW270 >50 >50 >50 >50 Gliotoxin C13H14N2O4S2MW326 50 >50 >50 Patulin C7H6O4MW154 25 >50 12 12 Penicillic Acid C8H10O4MW170 50 >50 25 25 Penicillin G C16H18N2O4SMW333 >50 >50 >50 50 >50 >50 >50 Secalonic acid D C32H30O14MW638 12 >50 >50 6 Wortmannin C23H24O8MW428 50 2,4 DNP* C6H6N4O4MW198 50 >50 >50 >50

*Internal Standard, not a Penicillium compound

STANDARDS

The chromatographic and biological profiles of several frequently encountered bioactive compounds known to be produced by the genus Penicillium were assembled. Always included in the analysis of extracts from microbial fermentations is a chromatogram of the extract obtained from un-inoculated medium.

Natural Products Support Group Natural Pr oducts Support GroupHPLC Trace /Biological Activity Fingerprint for a Natural Products Crude Extract HPLC Trace /Biological Activity Fingerprint for a Natural Products Crude Extract

F223889-Mixture of Bioactive Standards F204891-PDB Media Blank UV240 UV240

0.030Secalonic0.20 DuclauxinAcid B 0.025

0.15 0.020Penicillic Gliotoxin Brefeldin A Wortmannin0.10 Acid 0.015 0.05 0.010 0.00 0.0050 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0.000Deep Well #-0.05 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90-0.005 Deep Well # Ca W Ca W

40Zone 4030 0,0Zone 3020(mm) * 2010 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 * * (mm) 100 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Bs W Bs W40

* 4030 30 0,020 20 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 10 10 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Ec W Ec W40 40

20 30 30 0,02010

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 100 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Ca99-788 Ca99-788100 100%% 75 75 * * * * 0,050Inh Inh 50

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 25 25

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Sa-385 Sa-385 100 10075 75 0,050 50 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

25 25

TIC+ TIC+3500000 25000003000000 2000000 2500000

M+H M-36+H =245

=171 15000002000000 1000000

1000000 1500000

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 500000 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

TIC 800000 TIC500000 700000 400000

M-H =545 600000

M-2S-H 500000

200000

300000 =261 M-H =427

200000 300000 400000

0100000 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0100000 5 10 15 20 25 30 35 40 45 50 55 60 65 70 7575 80 85 90

0 50300006000090000120000

150000 197

10 15 20 25 30 35 40 45

2,4-DNP

50 55 60 65 70 75 80 85 90 0 5050000100000

150000200000250000 197

10 15 20 25 30 35 40

2,4-DNP

45 50 55 60 65 70 75 80 85 90 Deep Well # Deep Well #

EXTRACTS CONTAINING KNOWN TOXIC COMPOUNDS

Selected for special analysis were several extracts obtained from fungal strains known to be producers of toxic substances. Penicillic acid is a compound frequently detected in large quantity. An example is shown of Penicillium duclauxi, a producer of wortmannin.

Natural Products Support Group Natural Products Support Group HPLC Trace /Biological Activity Fingerprint for a Natural Products Crude Extract HPLC Trace /Biological Activity Fingerprint for a Natural Products Crude Extract

F217993-Penici l l ium sp. F217231-Penici l l ium duclauxi UV240 UV240

Penicillic Acid 225 nm 0.25 0.203.0

2.0

258 1.5OOMM ee 0.152.5 0.20 296 Wortmannin0.10OOHH 1.02.0 0.15 0.05 0.5

0.00200 300 400 500 6001.5 0.10

OO 0.0 -0.05 200 300 400 500 6001.0 0.05OO

0.5 0.00 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 900.0 -0.05 Deep Well #0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Deep Well #Ca W Ca W

40Zone 30 Zone

40 0,0 30 20,0

10 20 (mm) 20(mm) 10 * 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Bs W Bs W 40 40 30 30 0,0 10

* 14,11 2020 10

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Ec W Ec W 40 40 30 0,0 10

11,0 30 10

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

20 20

Ca99-788 Ca99-788 % 100 % 100 Inh 50

75 7525

95,96 Inh 50 * 104,8825

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Sa-385 Sa-385 100 100

50 0,075 7520,7 50

25 250 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

TIC+ 153 TIC+8000000 Penicillic Acid 125 8000000

6000000 7000000 M+H=171 7000000 5000000 171 6000000 4000000 5000000 3000000 4000000 2000000 3000000 1000000 2000000

0 10000000 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90TIC TIC2500000 2000000 427

2000000 Wortmannin1500000 1500000 428.1M-H-=427

399.1

367.11000000 459.1141.0 201.1 355.1

417.1249.1 443.1385.1137.0 317.1143.0 460.2 495.11000000 171.0 202.1 311.2 323.2 444.1187.1 497.1 522.1121.0 241.2 265.2 478.2115.7 213.2 271.1 549.2 573.2 599.4 m/z

80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600

500000

0 500000

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 197- 19730000 2,4-DNP 20000025000 2,4-DNP1500002000015000 10000010000 5000050000 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Deep Well # Deep Well #

TOXIC EXTRACTS OF UNDETERMINED COMPOSITION

Chromatographic resolution of extract components has often revealed multiple zones of bioactivity. In the example below a large amount of patulin is responsible for the observed toxicity to E. coli, while a second bioactive component in well 46 has a separate activity against C. albicans.

In the second example, the culture had tentatively been identified as a Penicillium, but the retention time, spectra, and bioactivity of a major component did not correspond to a known Penicillium toxin. Following isolation the structure was determined to be 2,3 dimethoxy 5,6 dimethyl benzophenone, which had been reported from Gliocladium roseum. Re-examination of the culture confirmed that it is a Gliocladium.

Natural Products Support Group Natural Products Support Group HPLC Trace /Biological Activity Fingerprint for a Natural Products Crude Extract HPLC Trace /Biological Activity Fingerprint for a Natural Products Crude Extract

F214055-Penici l l ium monovert ic i l late F221757-Gliocladium roseumUV240 Patulin UV2400.5

OO 2,3-Dimethoxy-5,6-dimethyl-benzoquinoneOO 0.80.4 * OOO 0.70.3 0.6

0.5 O OOHH0.2 0.4* 0.3 O0.1 0.2

O0.10.0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0.0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90Deep Well # -0.1Ca W

Deep Well # 30 40Zone

0,0 Ca W 10

(mm) 20 Zone 40 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 30*

20 10(mm) *0Bs W 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 9011,640

30 Bs W20 40

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 30 10

20 10

0Ec W 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 40 30 0,020 Ec W 10 40

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 30 20

0Ca99-788 10 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90* 100

75% 34,37 * Ca99-788Inh 50 25 100

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 75%0 0,0

25 Inh 50

0Sa-385 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 100 0,0

75 Sa-38550 25 100

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 750 0,0 25 50

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90TIC+ 8000000 7000000 TIC+ 120000006000000 5000000 10000000 4000000 8000000 3000000 2000000 6000000 1000000 4000000

0 2000000 0

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

100 311TIC- Patulin1200000 M-H=153

% TIC- M-H1000000 141 600000 800000 329 * =1950 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 m/z 500000Unknown600000

400000400000 200000 300000

00 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 200000

0 50 50000

100000 150000 200000

197

10 15 20 25 30 Deep Well #

35 40 45

2,4-DNP

50 55 60 65 70 75 80 85 90

0 5100000

0 503000006000009000001200000

1500000 197

10

10

15

15

20

20

25

25

30

30

35 40 45

35 40 45

2,4-DNP

50

50

55

55

60

60

65

65

70

70

75

75

80

80

85

85

90

90 Deep Well #

CONCLUSIONS

The HPLC/multidetector/bioactivity fingerprint can confirm the presence of a number of known Penicillium-produced toxic natural products from a crude extract in a single chromatographic step without additional purification. In one instance detection of a known compound not anticipated from a Penicillium provided a key bit of chemotaxonomic information which aided in the identification of the culture as a Gliocladium sp.

More interestingly, the presence of two or more zones of bioactivity, sometimes with toxicities toward different test organisms, is often detected with this approach to analysis. The chemist can then focus on the specific bioactivity of interest.

As hundreds of spectral plus bioactivity fingerprints are acquired and added to a searchable database, the value of that data in speeding the dereplication process will increase.

REFERENCES 1. Neutral, alkaline and difference ultraviolet spectra of secondary metabolites from Penicillium and other fungi, and comparisons to

published maxima from gradient high-performance liquid chromatography with diode-array detection. R. Russell, M. Patterson and Carlos Kemmelmeier, Journal of Chromatography 511 (1990):195-221.

2. High performance liquid chromatographic determination of profiles of mycotoxins and other secondary metabolites. Jens C. Fristvad, Journal of Chromatography 392 (1987):333-347.

3. Integrated biological physiochemical system for the identification of antitumor compounds in fermentation broths. Derek J. Hook, Carolee E. More, Joseph J. Yacobucci, George Dubay and Sean O’Connor, Journal of Chromatography 385 (1987): 99-108.

4. Rapid detection and subsequent isolation of bioactive constituents of crude plant extracts. K. Hostettmann, J.L. Wolfender and S. Rodriguez, Planta Medica 63 (1997):2-10.

“Thi

s pr

ojec

t has

bee

n fu

nded

in w

hole

or

in p

art w

ith F

eder

al f

unds

fro

m th

e N

atio

nal C

ance

r In

stitu

te, N

atio

nal I

nstit

utes

of

Hea

lth, u

nder

Con

trac

t No.

N01

C

O-5

6000

.”

AC

KN

OW

LE

DG

ME

NT

S

We

wou

ld li

ke to

than

k D

r. M

ike

Cur

rens

, Ms.

Cat

hy H

artl

and

and

Ms.

Pat

rici

a S

tale

y of

the

In V

itro

Scr

eeni

ng T

echn

olog

ies

Bra

nch

for

data

gen

erat

ed in

th

eir

drug

res

ista

nce

mic

roor

gani

sm s

cree

n. W

e w

ould

als

o li

ke to

than

k M

r. L

ucas

Bar

r fo

r hi

s si

gnif

ican

t hel

p in

pre

pari

ng th

e de

repl

icat

ed H

PLC

fra

ctio

ns

for

bioa

ssay

.