Characterization of exposure to molds and actinomycetes in agricultural dusts by scanning electron microscopy, fluorescence microscopy and the culture method.

K, MALMBERG P. Characterization of exposure to molds and actinomycetes in agricul tural dusts by scanning electron microscopy, fluorescence microscopy and the culture method. Scand J Work Environ Health 1989;15:353-359. Air samples from 79 farms with 10 5 to 1011 rnicroorganisms/m' were analyzed by scanning electron microscopy (SEM), fluorescence microscopy (FM), and thc culture method. The total exposure to microorganisms (particularly actinomycetes) was underestimated when as sessed as colony-forming units (cfu). The average cfu count was one-sixth of the total count according to SEM or FM, and the individual variability was great. This occurrence was partly explained by the aggre gation of spores. Single spores accounted for 2-65 % of all spores in 35 samples. There was an average of three spores/particle, and 93 (range 67-100) 0/0 of the spores were single or in aggregates of respirable size. Aggregation was more pronounced for actinomycetes and at high spore counts. Actinomycetes and bacteria could not be distinguished by FM. Bacteria (other than actinomycetes) were not detected by SEM, yet the lotal count of microorganisms was similar for FM and SEM. Most particles were spores from actinomycetes and fungi of the genera Aspergillus or Penicillium.

Allergic alveolitis is associated with exposure to high concentrations of spores of molds and actinom ycetes. Thi s relat ionship can be illustrated by the association with climati c and enviro nmental factors promot ing growth of microorganisms (1, 2) and by direct measurements (3). Exposure to very high concentrations of microorgani sms may ca use a to xic febrile reacti on (I , 4, 5, 6) even in apparently non sensitized subjects (organic dust tox ic syndrome).
Man y different method s have been used to measure exposure to microorganisms. Sampling methods include impaction on gel, use of cyclones and impingers, and collection on a filter. Microorganisms ha ve been identified with the use of cultures or optical microscopy (7). A multi stage impactor (Andersen sampler) which allows fractioning accord ing to the aerodynamic size of the particles (8) has been extensively used in the study of agricultur al dusts. Recently a method has been developed which involves dust collection on a filter followed by extraction , staining with acridine orange dye, refiltering on a black filter , and counting with fluorescence micro scop y (FM) (the FM method) (9). Part of the eluted collection can be used for culturing.
All method s based on the use of cultures record particles which form colonies on the media and the temperatures used (colony forming units), but the relevant exposure measure includes spores which will not form colonies under these cond itions. Earlier stud ies which employed the FM method to record a total count of microorganisms noted much higher total counts than 1 National Institute of Occupational Health , Solna, Sweden.
Reprint requ ests to: Dr P Malmber g, Nat ional Institut e of Occupation al Health , S-171 84 Solna, Sweden . the count s of colony-forming units (cfu) obtained with cultures. The variability was also great for the ratio of the cfu count to the total count (10, 1I). This difference may be due to an aggregation of the particles or to a lack of viability.
In the present study scanning electron microscopy (SEM) was used to characterize exposure from the farming environment. This method allows direct characterization of how airborne microorganisms appear, ie, as single spores, in aggregates, or as spores bound to other particles. The size of the particle s and the shape of individual spores can also be determined. Since spores of many species have a characteristic appearance, taxonomic cha racteri zation is possible to a certain degree with morphological criteria. All spore s can be counted regardless of viability or aggregation.
This study comprises a comparison of the total spore count according to SEM and FM, the cfu count , the " particle count" of SEM , characterization of spore aggregat es in different samples, and a descript ion of spore types in different environments based on cultur ing and morphological classification with the use of SEM.

Materials and methods
One or two air samples were obtained from 79 farms in the cour se of three studies reported elsewhere (1, 3,12,13). Eighteen farms were from a random sample of farm s on which th e farmer had no respiratory problems (reference farm s). On most of the remaining farm s the farmer had reported respi ratory problems. Ninety-seven samples were analyzed by the cul- . , ' .
.., ". . Occupational Health , Umea . Th e results fro m the two parallel samples were averaged . Th e third filter was ana lyzed for the total numb er of spores of molds or actinomycetes with a scanning electron microscope at the National Institute of Occupational Health , Solna .
The enumeration and classification of micro or ganisms were performed within 1-2 d after the sampling . Th e spores were eluted from th e filters with the use of 0. 1 l TJo (weight /volume) pept one water with 0.0 1 % Tween 80@ (polyoxye thylene so rbitanmo nooleate). One par t of the suspensio n was used for dete rmining viable microorganisms by the plate count method. Malt agar plates containing penicillin and streptomycin were used as a funga l med ium to inhibit bacterial grow th. Bacteria and actino mycetes were grow n on nu trien t agar with actidione to pr event funga l grow th . The number of cfu was recorded after 4 d on plates cultured at an elevated temperature and after 7 d on plates cultured at ro om temperature (21-24°C). Fun gi were grow n at room temperature an d at 45°C, and actinomycetes a nd ot her bacteria were grown at roo m tempe ra ture a nd at 55°C. Th e highest counts of fungal, act ino mycete, and bacterial colo nies were added 9 10 11 10 10 10 Spores/m-' of air ture met hod , and 42 of these same samples were also ana lyzed by electron microscopy. The du sts were sampled during work with different far m materials, which ra nged from no rma l to extremely mo ldy. Most of the samples were of th e "worst case" type and were collected for 5-15 min while the farmer was handling mat erial which had been associated with symptoms or was believed to cause maximum expo sure to mold dust. Other samples were obtained during 1-2 h of feeding, milking, an d the distribut ing of bedd ing materials to th e cows in t he cow barn (background samples).
A nalysis FM was used for enumera ting the tot al number of spores fro m fungi and actinomycetes an d/or bacteria. In addition, the num ber of cfu was counted, and viable microorganisms were characterized as described earlier (9) at th e Department of Microbiology, University of Agriculture, Uppsala , or at the Nation al Institute of

Samp ling
Th e samples were co llected with person al samplers.
The farmer or the investigator carried three filter cassett es directed obliqu ely downw ards on the chest. Th e cassettes were equipped with polycarbonate filters with a diam eter of 37 mm and a pore size of 0.4 11m (Nuclepore Corp, Pleasanton, California, United States). The air flow was I l/min, a nd the cassettes were closedfaced .
Wh en we compared p aired samples, we used Student's t-te st or Wilcoxon's signed rank. In addition , simple or multiple regression was used (Statview'", Mac intosh'"), The logarithm of the spore counts wa s used in all the calculations.

Results
dations of a working gr oup on th e harmonizat ion o f samp ling and analyzin g mold spores (14).   Colony forming to give th e total number of cfu . Pure cultures of differe nt fun gal co lonies were sent to Cent raalbureau voo r Schimmelc ultures in H oll an d for identificat ion. Ano ther part of the suspensio n was used for enumerati ng the total number of spores . The spores were fixed with forma lin , stai ned with acridine or an ge, and filtered through a polycarbonate filter dyed with Sudan black. The total number of spores (single spores and spores in aggre gat es) was determined with an epifluor escence micro scope from a cou nt of 40 view fields o r abo ut 200 microo rgani sm s per filter.
T he filter s were prepared for electro n micro scop y by gold-plating in a lEOL FC -I 100 sputter. T he to tal number o f spor es (single spo res and spores in agg rega tes) wa s counted with a lEOL lSM-840A scanning electro n microscop e. Seventy to eigh ty viewing fields evenly distr ibuted on a qu ad rant of the filter were counted at a magnification o f 2000 to 6000 X . The mo st common types of fungal spores were cla ssified, and the proportions of actinomycetes and fun gal spo res were ca lculat ed on the basis o f morphological criteria . Th e procedures co nformed to the recornm en- 9 10 11 10 10 10 Spore containing particles/m' of air  SEM and cfu count (lOlogarithm) was 0.38, corresponding to a ratio of 2.4 ( figure 3). There was no significant relationship between the proportion of actinomycetes in the samples with SEM and the total spore count (figure 4). Air samples from 14 reference farms had a mean of 44 (SD 37) 070 for the actinomycetes content of the total spore count with the SEM. This value varied considerably between samples (figures 4 and 5). The percentage of colonies of actinomycetes out of the total actinomycete plus fungal cfu count was significantly lower in the cultures (10 %) than the percentage of actinomycetes out of the total fungi plus actinomycetes with the SEM (49 %, N = 10, P < 0.05).
There was an average of three spores per sporecontaining unit with the SEM, but the variation was great, as illustrated by the variability in the distribution of aggregate sizes between samples (figure 6). High actinomycete and total spore counts predisposed for aggregation. The percentage of single spores was 2-65 % in 35 samples according to the following equation: % single spores = 100 -% actinomycetes X 0.20-7.6 x IOI0g(spores/ m' ), R2= 0.54.
Most spores were single or in aggregates with a minimum diameter of less than 10 11m (mean 93 %, range 67-100 %).
The most common fungus species (according to the culture method) are shown in table 1. A total of approximately 50 species were detected, but most of these were found only in small numbers. There were only small differences in species composition, depending on the spore concentration (table I) or materials which had generated the dust (table 2). Most samples were completely dominated by molds from the taxonomically related genera Aspergillus and Penicillium or by actinomycetes or by combinations of these three types of spores (reference farms, figure 5). Spores from the Aspergillus glaucus group, which were larger than the other Aspergi!!us spores and had characteristic surface structures, were found in 16 % of the samples. Aspergi!!us fumigatus were smaller and had surface markings which were easily recognizable. The appearance of many Aspergillus and Penici!!ium species overlapped, and these taxonomically related fungi therefore could often not be differentiated from each other by SEM. Otherwise the appearance of different genera of deuteromycetes were sufficiently distinct to allow classification with SEM. A collection of SEM photos of different spores can be obtained from the authors.

Discussion
On the dairy farms on which respiratory problems had occurred, the vast majority of particles in the air samples were microorganisms such as spores of fungi and actinomycetes ( figure 7). These could be present in numbers exceeding IOlO spores/m'. This dominance aggregate size 10 15 20 spores/aggregate 5 50 80 4 lOa -'--"""T"'<"""'--_
• P <0 .05.  of microbial spores was also found in a random sample of farms on which the farmer had no symptoms of disea se. Since these symptom-free farmers had evidence of immune stimulation and airway inflammation (12,15), efforts to control and monitor exposure are warranted. Air samples collected during work with cotton, corn, soybean, and nonmoldy grain may be dominated by particles other than microorganisms, including starch grain (16). The air samples from swine confinement buildings are dominated by large organic particles of different origin, and the microorganisms are less conspicuous. Many of th e microorganisms are small, and therefore indicate the presence of either actinomycetes or bacteria, and they ma y stick to other organic particles (17).
In SEM a high magnification is used; therefore it is desirable to have a high density of spores on the filter. The SEM method employs a vacuum which causes collapse of most bacteria and yeasts unless special fixation techniques are used . Mold and actinomycete spores are, however, less susceptible to collapse and may be identified in samples which have not been specially prepared. The method is therefore particularly useful in environments with high exposure to spores from molds and actinomycetes, such as in dairy farming or in wood trimming departments of sawmills (18). In these environments most impaction samplers risk overloading, and sampling times must be reduced to seconds with these devices. A modification of the slit sampler method , which partly circumvents this problem, has, however, been described (19).
The FM method has been advocated as a useful tool with which to monitor spo re exposure in swine confinement buildings (10). The dye may be selectively attached to microorganisms so that the y are visible also in the presence of other particles.
It is not possible to distinguish between actinomycetes and other ba cteria with optical microscop y or FM . On the other hand, with the routine preparation method used in the present study, bacteria (other than actinomycetes) cannot be ob served in SEM . This phenomenon did not appear to influence the total spore count significantly, as judged from the comparison between SEM and FM. In one-fourth of the samples, bacteria (other than actinomycetes) constituted more than 50 0J0 of the total number of colonies formed at culture. Since the total cfu count averaged only onesixth of the total count in the SEM, this finding does not necessarily show that bacteria dominated in these samples.
The results of the present study show that spores often exist in aggregates. Preliminary reports have indicated similar findings from a limited number of dust samples (10,11,20). The tendency of spores to aggregate is especially pronounced in samples with a high proportion of actinomycetes and a high total spore count. This situation helps explain part of the difference between the cfu count and the total spore count of SEM or FM , but the large variability indicates that many spores do not form colonies in the cultures due to a lack of viability, inhibition by other microorganisms , or inadequate culturing conditions.
The aggregation of spores is probably best studied by inspection of the original collection filter, either with SEM or optical microscopy, rather than on material which has been washed from the original filter holder, dyed , and collected on a second filter, as in FM .
The air samples were often dominated by a few groups, usually Aspergillus, Penicillium, or actinomycetes. It is not known if there are differences in the potential of different microorganisms to cau se allergic alveolitis or toxic febrile reactions. Mold s from the Aspergillus glaucus group have been implicated as important in this respect (21), and spores from this group can be distinguished from other spores by their morphological appearance in SEM.
The choice of method for collecting and evaluating microorganisms in air samples should be governed by the information desired and the environment studied. For research purposes, in formation from various methods can be combined. Detailed morphological classification may be obtained with cultures, but the cfu count is less informative and ma y seriously underestimate the total spore exposure. The proportion of actinomycetes appears to be particularly underestimated with the culture method, either due to slow growth or a high aggregating tendency. Optical microscopy and FM offer a relatively inexpensive means of estimating total exposure to microorganisms, but actinomycetes and other bacteria cannot be separated with the se methods. The present study indicates that SEM is a useful tool for the study of exposure to microorganisms in dairy farming. The method allows evaluation of the total exposure to fun gal and ac-