The use of microplate bioassays, or broth microdilution assays, to measure the activity of biological and synthetic compounds against fungal pathogens has increased in recent years; this technique has been identified as the most promising in vitro bioassay for quantifying antifungal activity (1). Quantification of fungal growth by spectrophotometric methods can be imprecise, however, because mycelial growth by nature is filamentous and therefore heterogeneous in liquid media. Studies using these methods have shown that there is variability in mycelial growth within a single well but such variability has not been quantified (2,3). Sample variability among replicate wells has been reported, but variability of absorbance readings within a single well has not been taken into account (4,5,6). Despite the growing popularity of this type of high-throughput assay, and possible limitations because of high within-well variance, sample variability and the introduction of sampling error have received little attention.

This study presents an improvement of a high-throughput screening method presented earlier (7) and based upon a previously published method (8). The assay was used to quantify the antifungal activity of a Bacillus subtilis biocontrol product (BCP) against plant pathogenic fungi of the genus Fusarium. The new method allows for more precise quantification of antifungal activity by accounting for variation in optical density within a single well of a 96-well plate.

B. subtilis strain TrigoCor 1448 (accession no. 202152; ATCC, Manassas, VA, USA) (9) was grown in solid state fermentation (SSF) bioreactors as described previously (7). Methanol extraction of the active compounds and subsequent solid phase extraction (SPE) for sample purification were also carried out as described (7). Briefly, fermentation samples were suspended in approximately 2 volumes (v/w) methanol. Methanol extracts were dried via rotoevaporation and partially purified using a C-18 SPE column (Burdick & Jackson™; Honeywell, Muskegon, MI, USA).

For assays comparing extract inhibition of cultures of Fusarium oxysporum and Fusarium graminearum, SSF biocontrol products were dried to <15% moisture content (wet basis) before extracting with methanol. Crude methanol extracts were diluted 1:1 with water and applied directly to the SPE column; the active fractions were eluted with 100% methanol and dried via rotoevaporation and under nitrogen to obtain extract weights. Assays verified no loss of activity using this process (data not shown).

The microplate bioassay to measure inhibition of F. oxysporum f. sp. melonis (isolate no. 81 of F. oxysporum race 2) (10), as described in Reference 7, was altered to reduce sample variability between triplicate wells. The method was also used to quantify control of F. graminearum (isolate Gz014NY98). Fungal cultures were cultivated for 1–2 weeks on quarter-strength potato dextrose agar plates. Plates were flooded with 10–15 mL sterile distilled water and gently scraped to form a spore suspension. Spore suspensions were diluted to 1×10⁴ spores/mL for use in bioassays.

Extract inhibition was tested by adding 100 µL potato dextrose broth, 10 µL BCP extract (7) or solvent control, and 90 µL fungal spore suspension to each well. All treatments were measured in triplicate. A Synergy™ HT plate reader (Bio-Tek Instruments, Winooski, VT, USA) was used to read plates at an absorbance of 620 nm before and after incubating for 48 h at 25°C. The heterogeneous nature of mycelial growth even in relatively small wells led to high sample variability between replicate wells. The KC4™ software (Bio-Tek Instruments) controlling the plate reader allowed the user to read absorbance from a single point at the center of each well or scan the well across a user-defined grid. The standard software only allowed the minimum 3 × 3 scanning grid for the small well size in a 96-well plate. This scanning protocol, however, led to high absorbance readings at the corners of the grid because of meniscus and sidewall effects occurring near the well edge.

To reduce this problem, the micro-plate geometry profile was adjusted using the KC4 software so that the well diameter throughout the plate was reduced from 6860–4600 µm. Although the actual plate well size was not changed, the reader scanned the plate as if a plate with such geometry were being used. At such a small well diameter, the software will not allow for the full 3 × 3 grid scan. The corners of the 3 × 3 scanning grid are not included, but readings are still recorded at the center and all four sides of each well for a total of five readings per well. These absorbance readings are all far enough away from the side of the well to diminish interference occurring near the well edge. Final absorbance in each well was measured as the mean absorbance of the five readings within each well. Extract mean inhibition and standard deviation were calculated as described in Reference 7,.