The experimental conditions used in the first set of experiments, the corresponding antibacterial activities of cycle I and II, their differences and average values are presented in Table 1. The error limits, effects and the change in the mean effect were calculated and the results are shown in Table 2. The extraction temperature, extraction time and ethanol concentration of the central point in first set (E₁₀, E₂₀) were 50 °C, 14 h and 50%, respectively.

In the first set, the error limits for average, effects and changes in mean were 0.1739, 0.1235 and 0.1096, respectively. The change in mean effect was −0.2105. According to the decision-making procedure, after calculating the change in the mean effect and error limit, an examination was necessary to determine whether any change in the control (search level) experimental conditions would help to improve the objective function [10]. The optimum condition was achieved when the effect was smaller than the error limit while the change in the mean effect was large. Moreover, because the dependent variables are the population of P. aeruginosa in which growth was suppressed by addition of P. frutescens var. acuta leaf extract, the optimum point was reached when the code of mean effect was negative.

The determination of the magnitude of the change in mean effect, which is negative and large, compared to the error limit, is a requirement in order to confirm the achievement of the optimum condition. Such a situation, where some of the effects are larger in comparison to the error limit, does not ensure that the conditions in the search region (E₁₀, E₂₀) of the first set is the actual optimum and a second set of experiments is called for.

In the second set, the search level (E₁₀, E₂₀) was fixed at the best condition of Set I, at a level of E₂₁, in which the number of P. aeruginosa was 4.955 log CFU/mL. The extraction temperature, extraction time and ethanol concentration of the central point in the second set (E₁₀, E₂₀) were 65 °C, 20 h and 75%, respectively. The experimental conditions and the results of the Set II experiments are presented in Table 3 and the effects and error limits are shown in Table 4. In the second set, the error limits for average, effects and changes in mean was 0.1612, 0.1145 and 0.1016, respectively. The change in mean effect was −0.0685. The most effective antibacterial activity (4.095 log CFU/mL) was obtained at E₁₄. The extraction temperature, extraction time and ethanol concentration of the E₁₄ point in the second set were 80 °C, 26 h and 50%, respectively. In this case, not all of the effects were smaller than the error limit, and the change in mean effect was smaller compared to the error limit even though it is positive. It has been reported that if all or any of the effects are larger than the error limits, the change in the experimental conditions may yield better results [10].

From the above situation, the third set of experiments was designed in which the best condition of Set II (E₁₄) was selected as the search level (E₁₀, E₂₀) for Set III. The experimental conditions and the results of Set III are shown in Table 5 and the calculated effects and error limits are presented in Table 6. In the EVOP-factorial design, the effects remain smaller than the error limits while the changes in the mean effect remain larger and positive so as to reach the optimum level. Thus, in the experiments of third set, we were able to arrive at the proper optimum condition, in which all effects were smaller than the error limit and the change in mean effect is large and positive. As shown in Figure 1, the population of P. aeruginosa decreased from 6.660 log CFU/mL in the initial set to 4.060 log CFU/mL in the third set.

In this study, it was shown that higher antibacterial activity was achieved at a higher extraction temperature of 80 °C (R = −0.800^**) and in a longer extraction time of 26 h (R = −0.731^**). However, antibacterial activity of P. frutescens var. acuta leaf extract against P. aeruginosa was not affected by the presence of different ethanol concentrations in the extraction solvent (R = −0.076), as shown in Figure 2. Therefore, the maximum antibacterial activity of P. frutescens var. acuta leaf against P. aeruginosa, determined by the EVOP-factorial technique, was obtained at 80 °C extraction temperature, 26 h extraction time and 50% ethanol concentration. These conditions were also found in our previous study conducted on Coptidis rhizoma against Streptococcus mutans [17].

Although the optimal condition of P. frutescens var. acuta on antibacterial activity is well described in this study, the extraction yield and bioactive compounds in P. frutescens var. acuta are highly variable depending on environmental conditions. Therefore, further studies are needed to compare the effects of various seasonal samples of P. frutescens var. acuta leaf on antibacterial activity using EVOP-factorial design technique.

Further, a SEM study of was carried out to visualize the effects of the ethanolic extract of P. frutescens var. acuta leaf on the morphology of P. aeruginosa and demonstrated altered cell morphology as compared to a control group (Figure 3). Control cells in the absence of the extract showed a regular, smooth surface (Figure 3A). In contrast, cells inoculated with the ethanolic extract of P. frutescens var. acuta leaf revealed severe detrimental effects on the morphology of the cell membrane, showing disruption and lysis of the membrane integrity (Figure 3B). Exposure of P. frutescens leaf extract to P. aeruginosa revealed extensive surface collapse and wrinkled abnormalities on the morphology of the cells along with some small clefts formation (Figure 3C, 3D) and these finding are in strong agreement with a previous report [16]. Besides, several researchers have reported the effects of various plant extracts on the morphology of pathogenic bacteria including P. aeruginosa [18,19].

The leaves of P. frutescens var. acuta were obtained from Yakrung market, Daegu, Republic of Korea, in July 2008. The specimen was lyophilized for 48–72 h after storage at −70 °C. Freeze dried samples were pulverized with a blender (HJM-7000, Hanil, Korea). Extra pure grade ethanol solvent was purchased from Daemyung Scientific Co., Daegu, Republic of Korea. Chemical reagents were obtained from Sigma Co. (St. Louis, MO, USA), unless otherwise stated.

Pseudomonas aeruginosa KCTC 2004 was used in the antibacterial assay. The strain was obtained from the Korea Food and Drug Administration (KFDA), Daegu, Republic of Korea. Active cultures for experimental use were prepared by transferring a loopful of cells from stock cultures to flasks and inoculatino in Luria-Bertani (LB) broth medium at 37 °C for 24 h. A culture of the bacterial strain was maintained on LB agar medium at 4 °C.

To design an experiment for establishing proper extraction conditions, 20 g of sample was hydrolyzed and extracted in a reflux extraction apparatus under different extraction temperatures (35, 50, 65, 80 and 95 °C), extraction times (8, 14, 20, 26 and 32 h) and ethanol concentrations (25, 50, 75 and 100%) under EVOP factorial design and then freeze dried in order to establish the proper extraction conditions. The concentration of P. frutescens var. acuta extract used for the determination of the microbial population was 0.1%. Samples were homogenized for 30 seconds and serially diluted with peptone water (Difco, USA), as needed for the determination of microbial populations.

In this study, the optimization of antibacterial activity of Perilla frutescens var. acuta leaf was investigated using the EVOP-factorial design technique, which was first proposed for the optimization of different operational parameters in a functioning chemical processing plant. The main advantage of EVOP methodology is its clear-cut decision making procedure, which indicates the change in influencing variables toward the desired maxima or minima [20]. The EVOP-factorial design technique was applied to select the optimum conditions of three extraction factors (temperature, time, and ethanol concentration) in different experiments [9]. Firstly, the control or search level experimental conditions (E₁₀, E₂₀) were selected based on the results of earlier investigation on the effect of individual extraction conditions on the antibacterial activity of the ethanolic extract of P. frutescens var. acuta leaf [21]. Secondly, the new experimental conditions (E_be) were selected with lower and higher levels of inducers compared to the initial search level (E_b0). Antibacterial activities of P. frutescens var. acuta leaf extract were estimated following the given assay procedure and recorded for cycle I and II. Differences in the antibacterial activities between cycle I and II, and average antibacterial activities were calculated to estimate the effects and error limits. The magnitudes of effects, error limits and change in mean effect were examined as per the decision making procedure to arrive at the optimum level. When the experimental results of the first set did not reach to the satisfactory level of optimum conditions, a second set of experiments was planned, selecting the best condition of the first set as the new search level for the second set. This procedure was repeated until the optimum condition was obtained.

Further, to determine the antibacterial activity of P. frutescens var. acuta leaf extract, enumeration of viable counts on LB plates was monitored as follows: 1 mL of the re-suspended culture was diluted into 9 mL buffer peptone water, thereby diluting it by 10-fold. In total, 0.1 mL sample of each treatment was diluted and spread on the surface of LB agar. The colonies were counted after 24 h of incubation at 37 °C.

To determine the efficacy of the extract of P. frutescens var. acuta leaf on the morphology of P. aeruginosa, SEM studies were performed using the optimum extraction conditions of the ethanolic extract of P. frutescens var. acuta leaf. Controls were prepared without extract. Further, to observe the morphological changes, the method of SEM was modified from Kockro et al.’s method [22]. The bacterial sample was washed gently with 50 mM/L phosphate buffer solution (pH 7.2), fixed with 2.5 g 100/mL glutaraldehyde and 1 g/100 mL osmic acid solution. The specimen was dehydrated using sequential exposure to ethanol concentrations ranging from 30–100%. The ethanol was replaced by tertiary butyl alcohol. After dehydration, the specimen was dried with CO₂. Finally, the specimen was sputter-coated with gold in an ion coater for 2 min, followed by microscopic examinations (S-4300; Hitachi, Japan).