Fusarium mycotoxins as a problem in Finnish feeds and cereals

In the past mycotoxins have been only an occasional minor problem for domestic animals in Finland. In 1982 a large number of intoxicated animals were suspected of being affected by mycotoxicosis. Later on imported maize was found to be the cause of the illness. After 2 years of investigations we concentrated our attention on Fusarium toxins, and in the autumn of 1984 we were able to detect the presence of trichothecenes in feeds. The rainy summer of 1984 created very favourable conditions for the growth of moulds. The production lines of some feed factories were contaminated by Fusarium fungi. The number of intoxicated animals increased drastically. Two strains of Fusaria were isolated at a feed factory. Since the trichothecenes have strong dermotoxic and cytotoxic effects, biological tests were used in this investigation. Capillary gas chromatography and mass spectrometry were used to identify and quantify the trichothecenes. Qualitative and quantitative results are presented together with the symptoms observed in the intoxicated animals. The reliability and the significance of the results have been discussed. Despite what has earlier been believed, it is evident from our results that trichothecenes are unstable in cereals and feeds. In fact, samples which originally proved to be tpxic were found to be almost toxinfree after 3 to 6 months storage at +4°C.


Introduction
Fusarium fungi are common contaminants of Finnish crop plants (Ylimäki 1970, Uoti  and Ylimäki 1974), and it has long been known that they cause diseases in domestic animals when they grow as parasites on wild and cultivated plants or other feedstuffs (Rai-  nio 1932), (Korpinen et  ai. 1972), (Roine et  ai.1971).However, theFusarium toxins have not been a serious trouble in Finland in the Index words: mycotoxins, trichothecenes, Fusarium toxins   past.The most important Finnish report on this subject deals only with zearalenone.Con- tamination by Fusarium fungi was excep- tionally heavy in 1972, although it did not create serious problems to animals (Kukkula et ai. 1975).An extensive study carried out in 1976 and 1977 on crops collected in different parts of Finland (Ylimäki et ai.1979) showed that 24 of the 230 samples analyzed contained zearalenone, and 4 samples T-2 toxin at con- centrations ranging from 0.1 to 40.0 mg/kg and 10 to 50 ug/kg respectively.Suspected mycotoxicosis was very unusual in Finland up until 1982, when a large number of cases of intoxication in mink, farm-foxes, calves and pigs were reported.The symptoms pointed to mycotoxicosis.In the autumn of 1982 our laboratory was contacted by some of the feed producers who asked us to carry out an investigation on this subject since the materials used to produce feed for several different animal species were suspected as the cause of the illness.The raw-material in question was imported maize.Samples of the maize and the commercial feeds were analyzed for possible poisonous compounds, with negative results.The results of the analyses for zearalenone and several other mycotoxins were negative, too.
Part of the imported maize was used up and part destroyed, but new cases of suspected mycotoxicosis broke out in 1983 and 1984.In autumn 1984 Fusarium toxins, trichothecenes, which are very toxic for animals and humans were detected in the feeds.The summer and autumn of the same year were rainy and wet in Finland and favourable conditions were created for the growth of moulds.The number of intoxicated animals drastically increased.
In all of the cases we considered the animals had received commercial feed.Inspections carried out at the feed factories showed that the production lines were contaminated.Water had condensed in some places and colonies of Fusarium fungi, which were iso- lated and identified as F. graminearum and F. poae were found growing there.

Mycological investigation 3700 colonies of Fusarium poae (Peck)
Wollenweber and 1700 colonies of Fusarium graminearum Schwabe were isolated per g feed on potato dextrose agar (Difco 0013-01-4) after incubation for one week in the light at + 25°C.The taxonomy of Booth (Booth  1971) was used in identification.
Fusarium graminearum rapidly filled the petri dish, producing floccose brownish rose- coloured aerial hyphae.The reverse side of the colony turned vinaceous red.Few macroconidia were seen on the aerial hyphae.The size of the sickle-shaped conidia was 4-5 x 30-50 um and they had well marked foot cells.Septation varied from 3 to 5. Chlamy- dospores were not found.F. poae grew as a powdery, reddish colony, which had a red reverse side.Great numbers of globose to ampulliform microconidia were seen on the lateral phialides.Their size was sxB and 5x5 um.Macroconidia were not de- tected, but a few chlamydospores were found in the cultures.The colonies were dry owing to the lack of slime production.

Materials and methods
167 samples have been analyzed in our laboratory.About 75 per cent of these were commercial feed and the rest consisted of different species of cereal.We were interested in detecting the six trichothecenes which, as re- ported by Ueno (Ueno 1983) are natural con- taminants in cereals.The only difference was that we used the HT-2 toxin in our mixture of standards instead of acetyldeoxynivalenol.The structure of the six trichothecenes, which are sesquiterpenoids with a tricyclic skeleton containing an epoxide ring, a double bond in position 9-lo and which are characterized as 12,13-epoxytrichotecenes, are shown in Fig. 1.Ueno divided these compounds into two groups according, among others, to their chemical characteristics: type A (T-2 toxin, HT-2 toxin and diacetoxyscirpenol) and type B (deoxynivalenol, nivalenol and fusarenon-X).
Since the trichothecenes are both strongly cytotoxic and dermotoxic, it is convenient to screen all of the samples by biological test in order to detect for the presence of contaminated materials.
Cell culture, rabbit skin and brine shrimp (Artemia salina) tests (Korpinen 1974) were used in the investigation.Those samples which gave negative results in the biological test were not analyzed by chemical methods.A large number of domestic animals suspected of having died as a result of mycotoxicosis have been investigated in the department of histo- patology, microbiology and virology at our institute in order to exclude any other possible causes of death.Meat and liver extracts of the carcasses were analyzed for the presence of toxic compounds using the cell culture test.

Biological tests
Cell culture RK-I3 (Rabbit kidney cell line 13) a) 10 g of meat or liver homogenate were extracted with 50 ml of chloroform.The solvent was evaporated on a Rotavapor and the residue transferred to a test tube with diethylether.After evaporation of the solvent the re- sidue was suspended in 0.8 ml of nut oil and 1.2 ml of a 1 % aqueous solution of Tween 20.The suspension was mixed for 20 seconds on a cyclomixer and then 70 ul of the sample were added to 1 ml of the cell culture at the time of seeding.Eegle's minimum essential Medium (MEM) supplemented with 10 % foetal calf serum, penicillin and streptomycin, was used throughout the work.b) 1 g of meat or liver homogenate was ex- tracted with 10 ml of water, filtered through filter paper and 70 ul of the sample were analyzed as in point a).The cell cultures were inspected after 24 and 48 hours.
Preparation of the feed sample for biological tests and chemical analysis The procedure used in extracting and purifying the sample is a modification of the method described by Dohi et al. (Dohi 1984).Extraction 20 g of feed or cereal were milled to a fine homogeneous powder and extracted 2 times with 150 ml of methanol-water (95 : 5) on a flask shaker for 1 hour.The extract was fil- tered through filter paper, concentrated to a small volume on a Rotavapor, transferred to a test tube by washing 3 times with 3 ml of toluene-acetone (4 : 2) and evaporated to dryness.The residue was redissolved in 2 ml of toluene-acetone (4 : 2) and purified on a chromatographic column containing 7 g of silica gel, with 4 g of anhydrous sodium sul- fate on the top and 2 g on the bottom.Hexane was used to pack the column which had been washed with 60 ml of tolueneand eluted with 100ml of acetone.The eluates were evaporated on a Rotavapor and redissolved in 500 ul of acetone (solution »A»).

Rabbit skin test
Young Albino rabbits were shaved on the dorsal region one day before the test.The skin was accepted for use in the test if there were no signs of wounds or irritation.25 ul of solution »A» were seeded on the skin in the dor- sal region in as small a spot as possible.The rabbit's dorsum was inspected every day for one week, and compared with a control area on the same animal.

Artemia salina (Brine shrimps)
225 ul of solution »A» were evaporated using a nitrogen stream in a test tube. 2.5 ml of an aqueous solution of sea-salt were added to the tube and also to a control tube.The shrimp's eggs had been grown in a special hatcher in 6 % saline water for 48 hours.The larvae separate themselves from the eggs because they are phototrophic.A spoonful (about 30 larvae) was transferred to each tube and inspected after 17 and 24 hours to deter- mine the number and condition of the surviving larvae.

Chemical analysis
Purification 250 ui of solution »A» were evaporated, the residue was dissolved in 35 ml of methanol-water (5 : 30) and extracted two times with 30 ml of hexane.The upper phase was discarded and the lower phase evaporated to dryness and redissolved in 10ml of chloroform-methanol (9:1) (solution »B»), A chromatographic column was prepared as follows: Florisil was treated with hot methanol for about 1 hour, filtered through a sintered glass and activated overnight in an oven at 105°C. 2 g of anhydrous sodium sul- fate (at the bottom), 10 g of Florisil and 5 g of sodium sulfate (at the top) were packed with chloroform and the column was washed with 30 ml of chloroform-methanol (9 : 1).
Solution »B» was transferred to the Florisil column.The trichothecenes were eluted with 150 ml of chloroform-methanol (9:1) and the solvent was evaporated off.
The residue was transferred to a Sep-Pack C lg cartridge using 150 ul of methanol 40 %.The trichothecenes were eluted with 40 ml of the same solution and the solvent evaporated to dryness.This purification is usually suffi- cient for cereals, grains and some of the feeds, but additional purification is needed in the case of fish feed, for example.A plug of glass- wool, 0.7 g of neutral alumina and 0.3 g of activated charcoal were placed inside a Pas- teur pipette.The residue was dissolved up in 2 ml of acetonitrile-water (84 : 16), trans- ferred to the column, eluted with 20 ml of the same solution and evaporated to dryness.

ECD-Gas chromatography
The residue of the eluates was silylated with 100 ul of trimethylsilylimidazole in a 2 ml glass stoppered tube for 1 hour at 100°C.After the reaction the sample was diluted with 1 ml of hexane and washed with 1 ml of water. 1 ul of the upper phase was injected in the splitless mode.The gas chromatograph was a Va- rian 3700 equipped with a fused silica column (25 meters, inner diameter 0.3 mm), bonded phase OV-1, 63 Ni electron capture detector operated at 290°C, injector temperature 260°C.The carrier gas was helium at a flow rate of 2 ml/min, the make up gas nitrogen at a flow rate of 30 ml/min.The sample was injected at 40°C; after 1 minute the tempera- ture was programmed to 160°C at 45°C/min and then to 250°C at 2°C/min.The amount of trichothecenes was obtained using an inte- grator (Varian model 4270) by comparing with standard solutions.The samples were injected at intervals of 2-3 hours during the same day in order to correct the variations in detector response.The recovery and detection limits of the trichothecenes were calculated from spiked samples, and their coefficient of variation from five replications of the same analysis.The detection limits for the trichothecenes analyzed by GC-ECD, their recovery and coefficient of variation are reported in Table 1.The derivatized trichothecenes remain stable when kept in a freezer, otherwise it is

GC-Mass spectrometry
The sample was first analyzed by ECD-gas chromatography.In cases peaks were detected with the same retention time as some of the trichothecenes, the sample was concentrated to a volume of 100-50 ul and analyzed again by GC-MS.
The gas chromatograph was a DANI 3800 HR equipped with a 17-meter-long fused silica column and bonded phase OV-1.The mass spectrometer was a JEOL DX-300 equipped with a data system (JMA-2000).The instru- ment was used in the electron impact mode.
The ion source temperature was 200°C, electron energy 70 eV, ionization current 200 uA, resolution 1000, and interface tempera- ture 250°C.If the concentration of the tri- chothecenes was high enough the sample was analyzed in the low resolution mode.The components were located on the basis of the mass chromatograms and their spectra used for the identification of the mycotoxins.This was done by comparing the spectra with reference data and with trimethylsilyl deriv- atives of standard compounds.In most of the cases the instrument was operated in the selecting ion monitoring mode.Two ions were monitored for each mycotoxin (see Table 2) and the results were compared with those ob- tained by GC-ECD in order to confirm the presence of the trichothecenes in the sample.
In the case of uncertain results the identity was confirmed by monitoring the eight most prominent ions in the spectrum of the tri- chothecene in question.

Results and discussion
The cytotoxicity of the six trichothecenes analyzed in the study are shown in Table 3.
The trichothecenes of type B (DON, NV and F-X) are much less cytotoxic than those of type A (T-2, HT-2, DAS).The concentra- tion of the trichothecenes has to be at least 100 ug/kg and 0.5 ug/kg for DON and T-2 respectively, for the sample results to by cytotoxic.
This method is very suitable for testing cer- tain types of food samples when very low con- centration of trichothecenes or some of their metabolites have to be detected.not cytotoxic; + = slightly cytotoxic; + + = 50 % of cells dead, + + + = 100 % of the cells dead.
Rainbow trouts were the kind of animals which had the largest losses because of the very many ill and dead fishes.Several hundred fishes had died in about 20 fish-farms and, in addition, a very large number had been se- verely injured.The cell toxicity test gave pos- itive results mainly for the water extract of their flesh and the chloroform extract of their livers.However, when this test gives a posi- tive result for a liver extract it does not de- finitely mean that trichothecenes are present.The liver from fish cought in the sea some- times also gave positive results.

Rabbit skin test
The effects and concentrations of the six trichothecenes applied to the rabbit skin are reported in Table 4.Although being very sen- sitive, the evaluation of toxicity using this method is not as accurate as the cell culture test and the results are not always reliable.Furthermore, the skin may suffer from a bac- terial infection some days after the start of the test, thus making the inflammed area appear worse than it really is.However, the results show that DON and NY have a weaker der- motoxic effect than toxins of the T-2 type.The exception to this is F-X which appeas to be very dermotoxic.Four different mixtures of randomly selected trichothecenes were tested.When the effect of mixture 1 is compared over a period of seven days with that of DON, NY or DAS alone, it is evident that the effect is much stronger when more than one toxin is present.This may be true not only as regards the dermotoxic effect, but also as general in- toxication when contaminated feed is consumed.Our samples frequently contained more than one trichothecene.
The smallest amounts of T-2 and HT-2 toxins detectable by the rabbit skin test are 0.025 ug and 0.250 ug respectively (Ylimäki et ai. 1979).

Brine shrimp test
The results of our experiments show that the brine shrimp test is not suitable for test-T a b l e  ing commercial feed and grains owing to the and quantitative determination of small very high number of incorrect positive results.
If the sample still contains pigments and other impurities after the preliminary purification, then the larvae will die and the test results be positive.The result for the rabbit skin test on the same sample would be negative.Prior has also come to the same conclusion (Prior 1979).

Chemical analysis
This method involving a combination of three different techniques was found to be very suitable for the simultaneous detection of the trichothecenes (DON, NV, F-X, DAS, T-2, HT-2), and can be used for the qualitative amounts of mycotoxins in grain and feed.One of the advantages of this method is that be- cause the biological tests are done first, none of the uncontaminated samples will be sub- jected to the whole analytical procedure.Par- ticular attention was paid to the purification and clean-up steps in order to use high sen- sitivity and to avoid contamination of the instruments and capillary columns.
It is necessary to check the purity of all the materials used in the analyses in order to avoid adding further impurities to the sample.Despite what has been reported (Rosen and  Rosen 1984), Sep-packs alone are not suffi- cient to achieve an effective purification of samples of this type owing to their limited capacity.Trichothecenes of the type »A» and »B» are often present in the same feed.They can be detected if sufficient material is analyzed and if the sensitivity of the method is high enough.
The toxins of type B, when analyzed by BCD, give a response which is about 40 times higher than that for type A. This is due to the presence of a conjugated carbonyl group in position 8 on their molecule.It should be pointed out here that the detection limit, as well as the good repeatability of the results, are affected by the amount of impurities in the sample.Large amounts of impurities will interfere in the derivatization step and possibly act as a negative catalyst in the injection cham- ber, thus decreasing the detectability of tricho- thecenes.
The results of our analyses on some of the feed samples and the symptoms observed in the animals which had consumed the feed are reported as an example in Table 5.
The case of the horses fed with domestic oats reported in Table 5 is the only case of animals fed with crops harvested in Finland that we know for certain were contaminated by trichothecenes.
Table 5. Amount of trichothecenes detected in the feeds and some of the symptoms observed in the animals.It is difficult to obtain representative sam- ples of cereals and feeds because the myco- toxins are not distributed homogeneously and very large amounts of material have to be mixed together before a sample can be taken.

Kind of feed
Our results concerning the toxicity of the trichothecens cannot be compared with those reported in the literature for the toxicity of individual mycotoxins (Mirocha 1983) because our samples contained more than one.The feedstuffs were analyzed between 3 to 6 months after they had caused intoxication in animals.The samples had been stored at + 4C in plastic containers.We believe that the re- sults shown in Table 5 represent the amount of trichothecenes at the time the analyses were carried out, and not the amount when the animals were given the feed.We have analyzed a second time some of the feeds which proved to be very toxic six months earlier by bio- logical methods.The results of the test on the second occasion were negative and the chemi- cal analysis showed only trace amounts of DON.
The T-2 and HT-2 toxins were still present in the samples in February, but no longer in April of the same year.DAS, on the other hand, was still detected in some of the feed- stuffs.Most of the samples still contained trace amounts of DON in April.These obser- vations indicate that trichothecenes are degraded in the samples over time.The tricho- thecene problem has occurred in three con- secutive years in Finland.Each year, however, the problem disappears by spring and no fur- ther complaints reported.One possible reason for this could be the lability of trichothecenes during storage.Scott et al. report that the level of DON in wheat decreases.The concentra- tion of DON in their samples decreased in July 1983 from 1.56 to 0.21 ug/g over a period of one week, and from 0.21 to 0.11 ug/g in four days.The wheat was stored for one month in air-tight bags at + 5C prior to analysis.The possible explanations reported by Scott et al.  are: reaction with plant components, metabo- lism by host plant enzymes, degradation of DON by microorganisms (Scott et al. 1984).

Fig
Fig. I. Structure of six Mycotoxins produced by Fusarium species

Fig. 3 .
Fig. 3. ECD gas chromatogram of an extract of naturally contaminated fish feed

Fig. 4 .
Fig. 4. Mass chromatogram analysis (A) of a sily- lated extract of maize.The electron impact mass spectrum (B) confirms the presence of DON in the sam- ple.

Fig. 5 .
Fig. 5. GC/MS selected ion monitoring analysis of a derivatized extract of naturally contaminated fish feed.

Table 1 .
Quantitative determination of trichothecenes in feeds by GLC-ECD.

Table 3 .
Cytotoxicity of the trichothecenes to cultured cells.