**4. Discovery of biocontrol agents amenable to commercial production**

Main objectives driving the development of our techniques to discover beneficial biological control agents for dry rot suppression involved two phases: (1) rapid screening of large numbers of microbes using enrichment techniques to concentrate desirable populations and a crop-relevant bioassay to identify useful biological control agents; (2) rating potentially useful agents based on the challenges of manufacturing and delivery.

### **4.1 Rapid isolation from large populations via enrichment techniques**

Ideally biological control agent isolation should begin in areas where biological control is naturally occurring in the field, as opposed to areas where it is not. Evaluating a maximal number of putative biocontrol agents increases the chance of discovering an effective strain. Isolating prospective biocontrol agents from appropriate tissues and under appropriate environmental conditions helps to insure that the microbial antagonists isolated will be well adapted to survival and activity on the specific tissues requiring protection. Application of these concepts resulted in our rapid isolation of 18 putative biological control agents for suppression of Fusarium dry rot. The steps of our method are illustrated in Figure 1 (Schisler & Slininger, 1994). Specifically, gamma irradiation-sterilized field soil samples were first enriched with potato tuber periderm, inoculated with a small amount of field soil obtained from potato fields with low dry rot disease incidence, and incubated for 1 week at 15°C. The microorganisms most adept at rapid growth on the nutrients found in potato periderm and at wound sites would make up the majority of microbes in each recolonized soil sample. Next, conidia of *G. pulicaris* were added to the microbially recolonized soils, and 2 days later, aqueous soil pastes of each soil were applied to wounded potato tubers to initiate a realistic disease bioassay. After incubation 4 weeks at 15°C, tubers were scored for

Biological Control Agents for Suppression

(1994).

1994).

of Post-Harvest Diseases of Potatoes: Strategies on Discovery and Development 145

synthetic chemical processes, fermentation processes are relatively expensive, a circumstance which has largely limited the exploration and development of biotechnology to the food and pharmaceutical fields (Van Brunt, 1986). Primary cost factors include raw materials, utilities, labor, and capital investment. Since the culture medium is central to fermentation process design and economics, our selection of the most commercially promising strains was based on their ability to grow rapidly and to high yield on a variety of liquid culture media and then to accomplish biocontrol upon harvest and delivery to potatoes. These considerations were combined to select efficacious dry rot antagonists that could be produced with reduced fermentor volume and cultivation process costs. The steps in this screening process are illustrated in Figure 2 based on the procedure of Slininger et al.

Fig. 2. Two dimensional liquid culture method of ranking commercial development potential of biological control strains using relative performance index, RPI (Slininger et al.

Strains with the nutritional flexibility to grow rapidly and achieve large bioefficacious populations were sought by challenging with glucose media ranging in richness from a minimal medium (with nitrogen supplied by urea) to a semidefined complete medium (with casamino acids and growth factors) to an undefined medium (with added yeast extract, peptone, and tryptone). Such flexibility is very desirable because it allows process optimization choices to be driven by materials cost and convenience rather than by the fastidiousness of the microorganism. Consistent with utilities considerations, shake-flask cultures were provided a low oxygen transfer coefficient (Kla~0.5 min-1) and moderately warm temperature (25°C) without pH control since most soil-borne microorganisms survive and grow with temperatures ranging from 7 to 30°C and within a fairly broad pH range

dry rot disease development. Those wounds that developed inconsequential disease were highly likely to contain microbial communities able to survive on potato periderm, to colonize potato tissue, and to suppress disease development. Consequently, clear wounds were excavated and dilution plated on nonselective media that allowed growth of bacteria, fungi, actinomycetes, and yeasts to allow isolation of broad microbial diversity. Using this process, over 350 isolated colonies were obtained from clear wounds receiving microbial communities transferred via live soil samples from 35 locations of low disease incidence. To screen out only those strains participating in dry rot suppression, each isolate was suspended in buffer with conidia of the pathogen and inoculated to a fresh potato wound. After 3 weeks at 15°C, tubers were checked for the presence of disease and only 18 of the 350 isolates demonstrated significant dry rot suppression relative to controls inoculated only with pathogen. It is notable that all of the 18 beneficial isolates were identified as Gramnegative bacteria.

Fig. 1. Isolation of microbial antagonists effective in suppressing Fusarium dry rot of potatoes (Schisler & Slininger, 1994).

#### **4.2 Screening potential strains for commercial suitability**

Researchers involved in the discovery and development of biological control agents may speed biocontrol agent commercialization by using an end-process-oriented screening approach. This concept refers to designing the screen to select strains based on their performance under conditions simulating key challenges typically posed by mass production (Standbury & Whitaker, 1984). Since the U.S. industry standard for the manufacture of microbial products is batch liquid cultivation, it was chosen as the method of producing the 18 dry rot antagonistic bacteria for further evaluations. Compared with

dry rot disease development. Those wounds that developed inconsequential disease were highly likely to contain microbial communities able to survive on potato periderm, to colonize potato tissue, and to suppress disease development. Consequently, clear wounds were excavated and dilution plated on nonselective media that allowed growth of bacteria, fungi, actinomycetes, and yeasts to allow isolation of broad microbial diversity. Using this process, over 350 isolated colonies were obtained from clear wounds receiving microbial communities transferred via live soil samples from 35 locations of low disease incidence. To screen out only those strains participating in dry rot suppression, each isolate was suspended in buffer with conidia of the pathogen and inoculated to a fresh potato wound. After 3 weeks at 15°C, tubers were checked for the presence of disease and only 18 of the 350 isolates demonstrated significant dry rot suppression relative to controls inoculated only with pathogen. It is notable that all of the 18 beneficial isolates were identified as Gram-

Fig. 1. Isolation of microbial antagonists effective in suppressing Fusarium dry rot of

Researchers involved in the discovery and development of biological control agents may speed biocontrol agent commercialization by using an end-process-oriented screening approach. This concept refers to designing the screen to select strains based on their performance under conditions simulating key challenges typically posed by mass production (Standbury & Whitaker, 1984). Since the U.S. industry standard for the manufacture of microbial products is batch liquid cultivation, it was chosen as the method of producing the 18 dry rot antagonistic bacteria for further evaluations. Compared with

negative bacteria.

potatoes (Schisler & Slininger, 1994).

**4.2 Screening potential strains for commercial suitability** 

synthetic chemical processes, fermentation processes are relatively expensive, a circumstance which has largely limited the exploration and development of biotechnology to the food and pharmaceutical fields (Van Brunt, 1986). Primary cost factors include raw materials, utilities, labor, and capital investment. Since the culture medium is central to fermentation process design and economics, our selection of the most commercially promising strains was based on their ability to grow rapidly and to high yield on a variety of liquid culture media and then to accomplish biocontrol upon harvest and delivery to potatoes. These considerations were combined to select efficacious dry rot antagonists that could be produced with reduced fermentor volume and cultivation process costs. The steps in this screening process are illustrated in Figure 2 based on the procedure of Slininger et al. (1994).

Fig. 2. Two dimensional liquid culture method of ranking commercial development potential of biological control strains using relative performance index, RPI (Slininger et al. 1994).

Strains with the nutritional flexibility to grow rapidly and achieve large bioefficacious populations were sought by challenging with glucose media ranging in richness from a minimal medium (with nitrogen supplied by urea) to a semidefined complete medium (with casamino acids and growth factors) to an undefined medium (with added yeast extract, peptone, and tryptone). Such flexibility is very desirable because it allows process optimization choices to be driven by materials cost and convenience rather than by the fastidiousness of the microorganism. Consistent with utilities considerations, shake-flask cultures were provided a low oxygen transfer coefficient (Kla~0.5 min-1) and moderately warm temperature (25°C) without pH control since most soil-borne microorganisms survive and grow with temperatures ranging from 7 to 30°C and within a fairly broad pH range

Biological Control Agents for Suppression

RPI1

Efficacy Growth

not significantly different are designated with the same group letter.

kinetics

Isolate number (NRRL-)

agar-grown antagonists.

in liquid culture (Slininger et al. 1994).

of Post-Harvest Diseases of Potatoes: Strategies on Discovery and Development 147

B-21050 67.3 65.0 66.2± 4.9 A 1 (18) B-21128 66.3 64.9 65.6± 5.6 A 2 (13) B-21133 67.2 62.3 64.8± 3.6 A 3 (11) B-21134 66.3 60.3 63.3± 4.4 A 4 (15) B-21132 56.9 69.4 63.1± 5.6 AB 5 (14) B-21102 62.1 60.8 61.4± 7.4 ABC 6 (16) B-21136 58.9 57.6 58.2± 6.9 BC 7 (8) B-21101 56.9 58.8 57.9± 4.9 BC 8 (11) B-21103 58.4 55.7 57.0± 4.4 C 9 (5) B-21053 59.2 54.2 56.7± 5.7 C 10 (6) B-21135 58.7 53.8 56.2± 6.0 C 11 (17) B-21129 53.0 56.5 55.1± 11.6 CD 12 (8) B-21104 63.2 35.9 49.5± 10.5 DE 13 (4) B-21048 47.1 45.4 46.2± 11.1 DE 14 (1) B-21137 42.8 46.7 44.7± 11.5 DE 15 (6) B-21051 60.9 25.1 43.0± 12.2 E 16 (1) B-21105 38.4 28.3 33.3± 7.1 F 17 (10) B-21049 31.2 23.8 27.5± 9.6 F 18 (1) 1RPIEfficacy and RPIKinetics each indicate the average of six RPI values determined from two independent productions of cells on minimal defined, semi-defined, and undefined liquid media. 2Commerical potential groupings were arrived at by applying a two-tailed t analysis to determine the 95% confidence intervals associated with each mean RPIEff,Kin as indicated by ± values. Means that are

3Numbers in parenthesis indicate rank based on traditional screen of efficacy of one-fifth trypticase soy

Table 1. Use of relative performance indices (RPI) to accomplish a two-dimensional assessment of isolate commercial potential based on growth and efficacy of cells produced

**4.3 Multi-dimensional screens to assess commercial potential and robustness** 

The concept of early commercial-process-oriented screening brings us closer to rapid development of marketable biocontrol agents; however, it is likely that liquid cultivation of biocontrol agents will be followed by formulation, drying, storage, and reconstitution prior to potato application. These steps are necessary to preserve cells for convenient storage and handling in the time between production and application, and represent other features or "challenges" that could be built into an expanded multi-dimensional strategy for selecting the most commercially promising strains. Furthermore, in the natural potato storage environment, many different strains of *Fusarium sambucinum* pathogen are present to challenge biocontrol strains, and in addition, the biocontrol agent will be expected to perform well on many different potato cultivars, different crop field histories, and different wound environments. Schisler et al. (2000) examined performance variability as a function of pathogen and cultivar, and in addition to nutritional flexibility to support robustness, biocontrol strains with better overall performance against multiple strains of pathogen on

Overall2 RPIEff.Kin Commercial potential group2

Rank3

from 5 to 8. Harvested bacteria were then bioassayed using the wounded potato assay described above to assess efficacy.

For each bacterium, a relative performance index (RPI) was calculated based on each kinetic parameter, such as specific growth rate and cell yield. Given parameter values normally distributed across the isolate group tested, the value of F = (X - Xavg)/s ranges from –2 to +2. Here, X designates a single value observed per bacterium, and Xavg and s are the average and standard deviation, respectively, of all values observed for the isolate group. Using the formula RPI = (F + 2) x 100/4, data corresponding to each parameter type were translated to dimensionless indices, scaled from 0 to 100, which reflected relative bacterial performance. For a given production trial, overall relative kinetic performance indices were calculated for each bacterium: RPIkinetics = (RPIgrowth rate+ RPIcell yield)/2. Similarly, a relative performance index based on biocontrol efficacy was calculated for each bacterium using log (disease rating) data: RPIefficacy = (2-F) x 100/4. Note that the term (2-F) is used instead of (2+F) because efficacy improves as disease rating decreases. Thus, RPIefficacy and RPIkinetics are provided a common dimensionless 0-100 scale that allows both data types equal weight in the overall performance assessment. As a result of this screening method (Fig. 2), referred to as "two-dimensional liquid culture focusing" (2DLCF), the 18 bacterial dry rot antagonists were ranked with respect to potential for commercial development, 6 strains being in statistical significance group A (Table 1).

Dogma calls for screening the efficacies of prospective biocontrol agents grown under conditions as similar as possible to what is expected to be encountered in nature. Thus, "promising" isolates have been traditionally selected based on efficacy following growth on solidified media. Only after extensive laboratory, greenhouse, and field tests of these promising isolates has mass production in liquid culture become a concern. Indeed our data have shown that the traditional one-dimensional screen based on bioefficacy of agargrown isolates selects a different set of top-performers than does the commercial processoriented 2DLCF screen, and the traditional screen is likely to miss selection of the most commercially useful biocontrol agents because it fails to recognize that liquid culture competency varies widely among microbes. Our experiments have illustrated this by showing that the top-performing strains selected via the 2DLCF screen were often ranked the worst performing strains in the traditional one-dimensional screen of one-fifth tryptic soy broth agar-grown isolates (Table 1) (Slininger et al., 1994). If our goal is to develop bacteria with a commercial future as biocontrol agents, then early screening strategies must reflect the production requirements of the commercial setting. Since liquid culture is the industrial standard for microbial production, liquid cultivation should be the method of biocontrol agent production during early screening. In addition, a two-dimensional assay examining liquid culture growth kinetics as well as product biocontrol efficacy is needed because, our results have shown that isolate performance ranking based on kinetics is not necessarily reflective of the performance ranking based on the biocontrol efficacy, yet both of these features are critical to process economics and commercial success. The processes shown in Figures 1 and 2 have resulted in identification of strains able to suppress dry rot under commercial storage conditions (Slininger et al., 1996a; Schisler et al., 1998a; Schisler et al. 2000b), and recently have been similarly applied to find additional novel bacterial strains with commercial potential for post-harvest biocontrol of pink rot (Adiyaman et al., 2011).

from 5 to 8. Harvested bacteria were then bioassayed using the wounded potato assay

For each bacterium, a relative performance index (RPI) was calculated based on each kinetic parameter, such as specific growth rate and cell yield. Given parameter values normally distributed across the isolate group tested, the value of F = (X - Xavg)/s ranges from –2 to +2. Here, X designates a single value observed per bacterium, and Xavg and s are the average and standard deviation, respectively, of all values observed for the isolate group. Using the formula RPI = (F + 2) x 100/4, data corresponding to each parameter type were translated to dimensionless indices, scaled from 0 to 100, which reflected relative bacterial performance. For a given production trial, overall relative kinetic performance indices were calculated for each bacterium: RPIkinetics = (RPIgrowth rate+ RPIcell yield)/2. Similarly, a relative performance index based on biocontrol efficacy was calculated for each bacterium using log (disease rating) data: RPIefficacy = (2-F) x 100/4. Note that the term (2-F) is used instead of (2+F) because efficacy improves as disease rating decreases. Thus, RPIefficacy and RPIkinetics are provided a common dimensionless 0-100 scale that allows both data types equal weight in the overall performance assessment. As a result of this screening method (Fig. 2), referred to as "two-dimensional liquid culture focusing" (2DLCF), the 18 bacterial dry rot antagonists were ranked with respect to potential for commercial development, 6 strains being in

Dogma calls for screening the efficacies of prospective biocontrol agents grown under conditions as similar as possible to what is expected to be encountered in nature. Thus, "promising" isolates have been traditionally selected based on efficacy following growth on solidified media. Only after extensive laboratory, greenhouse, and field tests of these promising isolates has mass production in liquid culture become a concern. Indeed our data have shown that the traditional one-dimensional screen based on bioefficacy of agargrown isolates selects a different set of top-performers than does the commercial processoriented 2DLCF screen, and the traditional screen is likely to miss selection of the most commercially useful biocontrol agents because it fails to recognize that liquid culture competency varies widely among microbes. Our experiments have illustrated this by showing that the top-performing strains selected via the 2DLCF screen were often ranked the worst performing strains in the traditional one-dimensional screen of one-fifth tryptic soy broth agar-grown isolates (Table 1) (Slininger et al., 1994). If our goal is to develop bacteria with a commercial future as biocontrol agents, then early screening strategies must reflect the production requirements of the commercial setting. Since liquid culture is the industrial standard for microbial production, liquid cultivation should be the method of biocontrol agent production during early screening. In addition, a two-dimensional assay examining liquid culture growth kinetics as well as product biocontrol efficacy is needed because, our results have shown that isolate performance ranking based on kinetics is not necessarily reflective of the performance ranking based on the biocontrol efficacy, yet both of these features are critical to process economics and commercial success. The processes shown in Figures 1 and 2 have resulted in identification of strains able to suppress dry rot under commercial storage conditions (Slininger et al., 1996a; Schisler et al., 1998a; Schisler et al. 2000b), and recently have been similarly applied to find additional novel bacterial strains with commercial potential for post-harvest

described above to assess efficacy.

statistical significance group A (Table 1).

biocontrol of pink rot (Adiyaman et al., 2011).


1RPIEfficacy and RPIKinetics each indicate the average of six RPI values determined from two independent

productions of cells on minimal defined, semi-defined, and undefined liquid media. 2Commerical potential groupings were arrived at by applying a two-tailed t analysis to determine the 95% confidence intervals associated with each mean RPIEff,Kin as indicated by ± values. Means that are not significantly different are designated with the same group letter.

3Numbers in parenthesis indicate rank based on traditional screen of efficacy of one-fifth trypticase soy agar-grown antagonists.

Table 1. Use of relative performance indices (RPI) to accomplish a two-dimensional assessment of isolate commercial potential based on growth and efficacy of cells produced in liquid culture (Slininger et al. 1994).

#### **4.3 Multi-dimensional screens to assess commercial potential and robustness**

The concept of early commercial-process-oriented screening brings us closer to rapid development of marketable biocontrol agents; however, it is likely that liquid cultivation of biocontrol agents will be followed by formulation, drying, storage, and reconstitution prior to potato application. These steps are necessary to preserve cells for convenient storage and handling in the time between production and application, and represent other features or "challenges" that could be built into an expanded multi-dimensional strategy for selecting the most commercially promising strains. Furthermore, in the natural potato storage environment, many different strains of *Fusarium sambucinum* pathogen are present to challenge biocontrol strains, and in addition, the biocontrol agent will be expected to perform well on many different potato cultivars, different crop field histories, and different wound environments. Schisler et al. (2000) examined performance variability as a function of pathogen and cultivar, and in addition to nutritional flexibility to support robustness, biocontrol strains with better overall performance against multiple strains of pathogen on

Biological Control Agents for Suppression

control agents.

**6.2 Late blight** 

**6.1 Sprout inhibition** 

of Post-Harvest Diseases of Potatoes: Strategies on Discovery and Development 149

regulatory hormones of isolated antifungal compounds suggested the fruitfulness of exploring the spectrum of use as a means of improving the market draw for our biological

Current practices for reducing sprouting in storage could also benefit from microbial alternatives. Because of processing demands, over 54% of the annual potato harvest must be stored at 7º to 13ºC, a temperature range above that needed for ideal sprout control (ASAE, 1990). Chemical sprout inhibitors are applied to over 50% of the potato harvest to extend storage time. The potato industry has become very dependent on CIPC (1-methylethyl-3 chlorophenylcarbamate) as the most efficient sprout inhibitor with fewest detrimental sideeffects on process potato quality (Lewis et al., 1996). However, recently, the tolerance for residues of CIPC has been reduced to 30 mg/kg (EPA 738-R-96-023, 1996) because of CIPC's persistence in the environment and potato tissue and concerns about its toxicity (Mondy et al., 1992). In the U.S.A., CIPC is the only synthetic chemical registered for post-harvest sprout control of stored potatoes, and it is the most widely used sprout inhibitor worldwide. Due to environmental and health safety concerns, the use of CIPC has become more restricted--opening a potential market for alternative sprout control methods. Consequently, six of our bacteria strains, exhibiting superior dry rot suppressiveness in previous research, were grown in two different liquid culture media and sprayed on Russet Burbank potatoes to assay sprout suppresiveness (Slininger et al., 2000, 2003). In growth chamber and pilot experiments repeated at two storage sites in two successive years, all six isolates demonstrated significant sprout control capabilities when applied after growth on at least one of the culture media supplied. Of the six strains tested, *Pseudomonas fluorescens* S11:P:12 (NRRL B-21133) and two strains of *Enterobacter* sp., S11:T:07 (NRRL B-21050) and S11:P:08 (NRRL B-21132), exhibited highest relative performance levels with sprout control being

statistically similar to that of 16.6 ppm CIPC thermal fog after 4-5 months storage.

Several of our top six dry rot suppressive strains have now also been found to significantly reduce late blight infection of stored potatoes (Slininger et al., 2007). Consistent with our observations of indoleacetic acid (IAA) as a major antifungal product produced by one of our dry rot suppressive strains (Slininger et al., 2004), Martiniez Noel et al. (2001) also previously showed that IAA attenuates disease severity in potato-*Phytopthora infestans* interactions and inhibits pathogen growth *in vitro*. *Phytopthora infestans*, the causative agent of the potato late blight disease, infects tubers through eyes or wounds, primarily via zoospores washed into soil from sporangia on infected leaves. Harvested tubers can become infected during washing (Fairclough et al., 1997) and during storage and handling (Lambert et al., 1998). *Phytopthora infestans* is considered to be the most significant pathogen of the crop worldwide (Fry et al., 2001) and historically was the cause of the Great Potato Famine of the late 1840's. The introduction of US-8 genotypes of *P. infestans* has coincided with an increase in severity of potato late blight in North America. As alternatives to chemical fungicides, our 18 bacterial strains patented as biological control agents of both sprouting and Fusarium dry rot were cultivated in 3 liquid media and screened in wounded potato bioassays for their ability to suppress late blight incited by *P. infestans* (US-8, mating type A2) (Slininger et al., 2007). Washed or unwashed stationary-phase bacteria were mixed with

multiple cultivars could be selected using the dimensionless relative performance index concept. The ability of biocontrol agents to solve multiple pest control problems is another potential screening dimension. For example, our dry rot antagonistic bacteria have also been shown to be able to suppress late blight (Slininger et al., 2007), pink rot (Schisler et al. 2009), and sprouting of stored potatoes (Slininger et al., 2000, 2003). The ability to expand the market to multiple pest control applications is expected to enhance commercial development potential of a given biocontrol product and is a recurring theme influencing the progression of our research as will be discussed at various points later in this account.
