Helminthosporoside, a Host-specific Toxin from Helminthosporium sacchari*

SUMMARY Helminthosporoside has been isolated from the sugar cane fungus pathogen Helminthosporium sacchari. This compound is the first host-specific plant toxin to have a structure proposed for it. The toxin produces symptoms only on those to Based on its spectral and chemical properties, the proposed structure of helrninthosporoside is Z-hydroxycyclopropyl-a-D-galactopyranoside. A biological assay for its quantifica-tion based on the degree of symptom expression on sugar cane leaves is described. HelminthosporosideJ4C can be isolated from cultures of H. sacchari that have been incubated with galactose-lJ4C. plant-pathogenic affect rlone$ wit,hin

proposed for it. The toxin produces symptoms only on those clones of sugar cane that are susceptible to the fungus. Based on its spectral and chemical properties, the proposed structure of helrninthosporoside is Z-hydroxycyclopropyla-D-galactopyranoside.
A biological assay for its quantification based on the degree of symptom expression on sugar cane leaves is described.
HelminthosporosideJ4C can be isolated from cultures of H. sacchari that have been incubated with galactose-lJ4C.
Host-specific toxins have been reported for a number of plantpathogenic sl)ecit?s of Nelminthosporizdm and a few other fungi, and there are numerous review articles on these toxins (l-4). These eoml)ounds have been termed host specific since they affect wly those rlone$ wit,hin a plant qwks t,hat, are susceptible to the fungus that produces them. Furthermore, these compounds do not ljroduce symptoms on any plant species not otherwise attacked by the fungus. Several investigators have indicated that characterization of these toxins has been difficult since they are unstable in a homogeneous condition. The published data, however, indicate that they are low molecular weight peptides (2,3). Steiner rind Bythcr (5) reported that a host-specific toxin was produced i)y lIelmint?wsporium sacchari (Van Breda de Haan) Butler.
H. sacclzari is the causal organism of eye spot disease of sugar cane which occurs in most of the sugar cane-growing areas of the world (6). The fungus causes eye-shaped lesions on the leaves, followed by the development of reddish brown streaks or "runners" extending from the lesions toward the tip of the leaf. The fungus can be isolated from the lesion but not from the runner areas. This observation suggested that a toxic compound from the fungus was causing the runner. Steiner and Byther (5) partially purified a substance that was capable of causing runners only on susceptible clones of sugar cane. Thus, since the toxin had the same host range as the fungus, it could be termed host specific. Their report indicated that the toxin had a low molecular weight and was stable to high temperatures. The partially purified substance is now used to screen clones of sugar cane for resistance to the eye spot disease (5). Because the toxin appeared to be stable, it was a likely candidate for isolation and characterization. This paper has a 4-fold purpose: (a) to present a method for purification of a host-specific toxin of H. sac&am'; (b) to report on the characterization of the toxin based on spectroscopic and chemical analyses; (c) to describe some of the biological properties of the toxin, and (d) to suggest the trivial name helminthosporoside for the toxin, based on its source and structure.
EXPERIMENTAL PROCEDURE Culturiv-The culture of II. sac&& used in this investigation was originally isolated from naturally infected sugar cane in Hawaii. The organism was maintained on agar made from sugar cane leaf extract (7). For toxin production, the fungus was grown for 18 to 20 days in l-liter Roux bottles containing 160 ml of Fries medium supplemented with 0.1% yeast extract dialysate (8). The cultures were grown at 22-24" under stationary conditions.
Biol.ogical Assuy-Assays were made on a susceptible clone of sugar cane 51 NG,97, in a manner similar to Steiner and Byther (5). Leaves of the cane were cut into l&cm sections and 1 ~1 of the test solution was placed on a needle puncture spot near the base of the leaf. Inoculated leaves were placed into a moist chamber at room temperature and measurements of runner length made after 22 hours unless otherwise indicated. The length of the runner was taken as an indication of the toxic potency of the preparation and a relationship was established between runner lengths and the amount of toxin applied to the leaf. Specific biological activity was arbitrarily taken as the amount of material required to produce a 5-cm runner.
The toxin appeared as a reddish spot with antimony trichloricle and as a yellow spot with antimony pentachloride.
Both reaction products with the antimony compounds fluoresced under ultraviolet light. was removed by centrifugation at 26,000 x g for 5 min and tliscnrded.
Acetone was removed by flash evaporation, and the remaining solution was partitioned against 3 volumes of chloroforn, The aqueous phase was concentrated to about 0.02 of t,h(J original volume.
At this state of purification, the toxin coultl be ,st,ored for extended periods of time without loss of a,&\-ity.
Further purification was accomplished by extracting the nqueoux phase (5.0 ml) with three IO-ml volumes of watersntur:ttled I-butanol. After removing the butanol by flash eVa]llJr;ttiOll, the material was resuspended in 1 .O ml of Hz0 and applied to :I c~olumn (1.5 x 48 cm) of Sephadex G-15 and eluted with distilled 1IsO. Fractions (I ml) were collected.
The tubes (35 through 4x1, after the void volume containing biological activity, were pooled, flash evaporated to dryness, and chromatographed ou water-washed Whatman No. 541 filter paper in Solvent a. The area containing biological activity was eluted with water ant1 rechromatographed in Solvent b. The toxin was elut,rtl wit,h water and stored in a desiccat,or over P&,.
The yitld of pure toxin from 1 liter of culture medium was approximately 7 to 9 mg. Small weight measurements were made on a C'ahn eleczi robnlance.
Gas-Liqltid Chromatography-Sugars were determined quantitat'ivelp ant1 qualitatively by gas-liquid chromatography after acid hydrolysis and neutralization of the samples. After 0.95 ml of a misturo containing hexamethyldisilane, trimethylchlorosilanc, and pyridine, 3:l :Q v/v/v, was added to the sample, the misture w:rs incubated at room temperature for 30 min in a capped vinl (10). The silyl ether derivatives of the sugars were subject.ed t.o gas chromatography on an F and M gas chromatogralrh equil)petl with a column of 370 SE-30 on Gas Chrom Q (0.4 x 180 cm).
The column oven temperature was run isothermally :I t 175" with a detector temperature of 250" and a carrier g:is flow rat,e of 30 to 50 cc per min.
The silyl ether derivatives of the standard sugar were chromatographed for quantitative and reference purposes. Other products of acid hydrolysis of the toxin were detected by injecting 5-~1 aliquots of the aqueous-acid hydrolysate directly on to a column (0.4 x 180 cm) of Poropak Q. The column oven temperature was run isothermally at 190" and all other conditions of gas chromatography were as described above.

Instrumental
An&sea---The infrared spectrum was obtained on a Beckman microspec using a micropellet of KBr. Mass spectral data were obtained from a Varian CH-5 with 100 pamp on the filament and the probe heated to 195". Ultraviolet analyses were performed on a Beckman DU spectrophotometer. Specific optical rotation of the compound was determined in a Zeiss circle polarimeter with 0.76 mg of the sample dissolved in 0.1 ml of H20 and placed in a tube with a light path of 50 mm at 30". Amino acid analyses were performed on a Technicon automatic analyzer.
The total carbon in each tube from samples obtained from Sephadex column chromatography was measured directly in a Beckman CO? analyzer. All compounds used were either reagent or spectroscopic grade. Fig. 1 is a flow chart illustrating the procedure used to isolate the toxin.

RESULTS
The effectiveness of each step in removing contaminating materials is shown in Table I. The specific biological activity increased with purification.
The toxin could not be demonstrated with the biological assay in either of the precipitates following the steps of the addition of acetone or chloroform to the culture filtrate.
Trace amounts of the toxin were detectable in the water phase after partitioning against butanol. Although fractionation with Sephadex G-15 removed a large amount of contaminating materials, tubes 35 through 48 containing most of the toxin still had additional contaminants (Fig. 2).
Purification of the toxin was completed by chromatography in Solvent a, followed by elution and chromatography in Solvent b. However, after chromatography in Solvent a, two bands possessing toxic activity were observed on the chromatogram, confirming previous results that two toxins were present after Sephadex column chromatography ( Fig. 2) (5). The toxin (helminthosporoside) in Peak 1 (Fig. 2) was the substance chosen for investigation in this study since it was the more biologically potent of the two.  a For each step in the purification procedure, the total dry weight and specific biological activity were determined. The data are based on 1.0 ml of crude culture filtrate as the starting material.
* This is defined as the amount of material required to produce a runner on a susceptible sugar cane leaf 5 cm long 22 hours after inoculation.
A 5-cm lesion was arbitrarily set as a standard lesion length.
The length of a lesion obtained from a given test was extrapolated from Fig. 5 to yield the amount of toxin present in the preparation.
Paper and thin layer chromatography of purified helminthosporoside in all of the solvent systems listed revealed that it traveled as a single band.
Acid hydrolysis of the compound in 6 N HCl for 20 hours in a sealed tube,.followed by analysis did not yield any amino acids indicating that the toxin was not a peptide.
Furthermore, the compound gave no absorption bands in the ultraviolet region where copper complexes of peptides absorb (11). The compound gave negative tests with 0.3% alco- holic ninhydrin and with the reducing sugar test of Trevelyan,Procter,and Harrison (12), respectively. In all of the chromatographic systems tested (a through k), the biological activity eluted at the place corresponding to the RF on the chromatogram, giving a positive test with the antimony trichloride reagent.
The compound was considered homogenous. Biological Activity Studies-Sugar cane leaves inoculated with the pure toxin caused symptoms that were identical with those produced by the fungus (Fig. 3). The first visible indication of toxicity to the treated leaf was the development of a light green area surrounded by the dark green area of the nonaffected portion of the leaf. Eventually the light green area developed into a reddish brown runner.
The length of visible runners on susceptible sugar cane leaves as a function of time and concentration of toxin was determined (Fig. 4). Based on these results, subsequent assays relied on a 22-hour incubation period in order to allow for full symptom development.
After 22 hours of incubation, runner length, as a function of toxin concentration, was virtually a linear relationship when plotted in a semilog scale (Fig. 5). Using this inoculation technique, symptoms on leaves were observable in amounts of helminthosporoside as low as 5 x lo-l2 moles. Helminthosporoside was tested on 11 clones of sugar cane differing in their susceptibility to H. succhcrti. In all cases, the plants responded in the same manner to the fungus as they did when helminthosporoside was placed on the leaf (Table II). In addition, the toxin did not produce symptoms on several varieties of wheat, corn, sorghum, and native grass species that were tested.
The toxin caused no visible symptoms in mice receiving  The stability of helminthosporoside was determined by placing 150 pg into each of many vials. Two sets of vials were sealed under vacuum and one set stored a't -15", the other set a,t room temperature, respectively.
Two other sets of vials were covered with parafilm and also stored in the same fashion.
Sa'mples from the four sets of vials were tested periodically.
Xo loss of biological activity was noted in any of the treatments up to 57 days, at which time the experiment was terminated.
Properlies of Helminthosporoside-.4t room temperature and humidity the purified compound was a yelloIvish syrup.
How-  5 Substance from Peak 8, Fig. 2. This toxin was located on chromatograms by elution with water and bioassaying on sugar cane leaves.
b Not determined.
FIG. 6. Infrared spectrum of helminthosporoside. The toxin was dried over P106 and redried after the preparation of a mixture with KBr.
A micropellet was prepared and the sample was run in a Beckman Microspec IR. The X axis is wave length in micra. ever, it became crystalline after desiccation in a partial vacuum over P&, indicating that it was highly hygroscopic. Table III shows the Rp values of both helminthosporoside and the other toxin (Peak 2 in Fig. 2) in several solvents by both thin layer and paper chromatography.
Helminthosporoside moved well in polar solvents, suggesting that it was a polar compound.
The infrared spectrum of the toxin revealed a strong absorption band at 3.0 M indicating the presence of -OH groups on the molecule (Fig. 6). The lack of a strong band around 6.0 /.L assured the absence of a carbonyl group either as an acid or an aldehyde.
The remainder of the spectrum initially suggested that the toxin could be a glycoside, which would be in keeping with the chromatographic data.
The ultraviolet spectrum revealed no absorption bands above 210 rnp, eliminating the presence of aromatic and conjugated systems in the molecule. The soecific optical rotation of the compound was [a]; = -49.0.
Mass spectroscopy of the toxin gave a discernible molecular ion peak of 236 m/e (Fig. 7). That the peak of 236 represented the molecular ion was confirmed by a molecular weight determination by membrane osmometry near 250. The peaks at 218 and 200 each represent the loss of 1 water molecule, respectively (Fig. 7).
Elemental analysis of helminthosporoside yielded the following percentages Based on a molecular weight of 236, an empirical formula CeHn,Or could be written for this compound. Only a trace of nitrogen was found in the samples.
Glycone Portion of Helmin-sporoside-Helminthosporoside (300 pg) was refluxed in 0.3 ml of 0.8 N H&04 for 2 hours at which time an excess of BaCOa was added to the solution to neutralize the acid.
Hz0 (2 ml) was added to the suspension and it was centrifuged to remove the precipitate. The supernatant liquid was dried in a stream of warm air and the residue was treated with the silylation reagent and subjected to gas chromatography. Two peaks with the same retention times as OL-and /3-galactopyranoside were observed. Theoretically, 228 pg of galactose would be the amount recoverable from this experiment, whereas the actual value was 160 pg. The identity of galactose was confirmed by paper chromatography in Solvent a and in a system consisting of ethylacetate-pyridine-HZO, 8 :2 : 1 v/v. In both systems the unknown had Rp values identical with authentic galactose.
After neutralization, acid-refluxed solutions of helminthosporoside were not phytotoxic. The proton magnetic resonance spectrum of helminthosporoside run in Dr.0 is shown in Fig. 8. Table IV summarizes the integrals for each of the resonance peaks.
Peak A, equivalent to 1 proton had the correct chemical shift for the proton at the anomeric carbon at a glycosidic linkage (13). Peak B (4 protons) was assigned to protons on carbon atoms having a secondary alcohol function (13). Three of these protons would be contributed by the gala&se moiety of the toxin (13). Two protons on the carbon atom with a primary alcohol function on the C-6 of gala&se, and the proton on C-5 of the hexopyranose ring were collectively attributed to Peak Area C which integrated for 3 protons and had the correct chemical shift for this assignment (13). The protons on the -OH of sugars are readily exchanged for deuterium and thus not seen in a sample that is dissolved in DzO.
Helminthosporoside-i4C with a specific activity of 1.43 $Zi per mmole was isolated from a culture of H. succhati that had been administered galactose-"C. After acid hydrolysis of 40 pg of the toxin and chromatography of the neutralized hydrolysate in Solvent a, at least 73% of the radioactivity was recovered as gala&se.
This lends further support for gala&se as the glycone moiety and is an effective way of preparing labeled helminthosporoside for physiological studies. Ag1ycon.e Moiety-The molecular weight of the aglycone moiety minus the glycosidic oxygen atom must be 57 if galactose was contributing 179 to the total molecular weight.
The base peak at 73 in the mass spectrum was attributed to the aglycone fragment carrying the glycosidic oxygen atom, and the large peak at 57 was assigned to the aglycone fragment alone (Fig. 7). The chemical shift shown in this spectrum assumes a shift of 4.758 for the HDO resonance.

4355
Cleavage on either side of an oxygen atom involved in a linkage of this type is a well established phenomenon (14). Based on the empirical formula, a mass of 57 leaves only CaHsO for the aglycone fragment.
The chemical shift position and the complex nature of absorption bands F and G in the NMR spectrum suggested that these protons were associated with a cyclopropane ring (Fig. 8). This assignment was in close agreement with those assignments reported for other cyclopropane compounds such as phenylcyclopropane and I-methyl-1-phenylcyclopropane (15). Furthermore, each peak integrated for only 1 proton, indicating the presence of a functional group on the remaining carbon atom.
This functional group could only be an -OH for reasons relating to empirical formula and spectral considerations. It was suggested that the proton on the carbon carrying the secondary alcohol functional group on the cyclopropane ring was located under Peak B, accounting for the 4th proton in that peak (Table IV).
The proton on the carbon atom of the cyclopropane ring linked to the glycosidic oxygen was assigned to AR (Fig. 8). Peak Area D integrated for 1 proton (Table IV) :n~d was heavily split by the protons on the adjacent carbon atoms of the cyclopropane ring as would be expected (15). The splitting of the spectrum of this proton was in agreement with that occurring in the spectrum of methyl-cyclopropylcarboxylate (15) ; however, Peak D was shifted downfield because of an adjacent oxygen atom.
The spike Peak E, integrating for 1 proton, was assigned to the proton on the -OH on the cyclopropane ring.
After prolonged periods in DzO, the intensity of this peak diminished, which was consistent with the idea that this proton is exchangeable, but not readily so. Paper chromatography of the residue, after hydrolysis of helminthosporoside (0.8 mg) in 1.0 N HCl for 2 hours in Solvent a only yielded a spot, corresponding to galactose when the chromatogram wan treated with alkaline silver nitrat,e.
In addition, no spots became evident on a similar chromatogram treated with the ant~imony trichloride reagent, indicating that hydrolysis was complete and the aglycone and products thereof were volatilized.
In addition to the NMR evidence for the presence of a cyclopropane ring as the carbon skeleton of the aglycone, the infrared spectrum showed a weak but detectable band in the region of 9.8 ~1 (16), further supporting this contention. The toxin reacted with the antimony trichloride reagent producing a reddish spot that fluoresced under ultraviolet light. However, this same reagent did not react with galactose and other free sugars, and several niethyl glycosides tested produced a slightly yellowish reaction product.
Furthermore, isopropanol and propanol did not react with this reagent and acrolein gave a brown reaction product.
Ilowcvrr, cyclopropyl derivatives, including cycloprol)a~ie carljo.\ylic acid, cyclopropyl nitrile, and cyclopropyl ethanol, all produced pink to reddish spots with this reagent and fluorcsccd under ultraviolet in a manner similar to that of the intact toxin.
That the c*yclopropyl ring possessed a hydroxyl function was supported by the detection of a substance in acid hydrolysates with the same retention time as acrolein.
Samples (5 ~1) of the sulfuric acid hydrolysate of the toxin were directly subjected to gas chromatography on Poropak Q. Compounds with retention times of 2.9, 5.2, and 7.4 min were detected.
The acrolein peak (2.9 mm) accounted for approximately 25% of the volatile products detected.
The presence of acrolein was expected as a product of acid hydrolysis of the toxin. Its presence was explained by an clirnination of the hydroxyl proton on the cyclopropyl ring resulting in bond reorganization and eventual cleavage of the glvcositlic bond.
The other volatile products of acid hydrolysis \verc not identified.