Cultivar was more influential than bacterial strain and other experimental factors in recovery of Escherichia coli O157:H7 populations from inoculated live Romaine lettuce plants

ABSTRACT The varied choice of bacterial strain, plant cultivar, and method used to inoculate, retrieve, and enumerate Escherichia coli O157:H7 from live plants could affect comparability among studies evaluating lettuce–enterobacterial interactions. Cultivar, bacterial strain, incubation time, leaf side inoculated, and sample processing method were assessed for their influence in recovering and quantifying E. coli O157:H7 from live Romaine lettuce. Cultivar exerted the strongest effect on E. coli O157:H7 counts, which held up even when cultivar was considered in interactions with other factors. Recovery from the popularly grown green Romaine “Rio Bravo” was higher than from the red variety “Outredgeous.” Other modulating variables were incubation time, strain, and leaf side inoculated. Sample processing method was not significant. Incubation for 24 hours post-lettuce inoculation yielded greater counts than 48 hours, but was affected by lettuce cultivar, bacterial strain, and leaf side inoculated. Higher counts obtained for strain EDL933 compared to a lettuce outbreak strain 2705C emphasized the importance of selecting relevant strains for the system being studied. Inoculating the abaxial side of leaves gave higher counts than adaxial surface inoculation, although this factor interacted with strain and incubation period. Our findings highlight the importance of studying interactions between appropriate bacterial strains and plant cultivars for more relevant research results, and of standardizing inoculation and incubation procedures. The strong effect of cultivar exerted on the E. coli O157:H7-lettuce association supports the need to start reporting cultivar information for illness outbreaks to facilitate the identification and study of plant traits that impact food safety risk. IMPORTANCE The contamination of Romaine lettuce with Escherichia coli O157:H7 has been linked to multiple foodborne disease outbreaks, but variability in the methods used to evaluate E. coli O157:H7 association with live lettuce plants complicates the comparability of different studies. In this study, various experimental variables and sample processing methods for recovering and quantifying E. coli O157:H7 from live Romaine lettuce were assessed. Cultivar was found to exert the strongest influence on E. coli O157:H7 retrieval from lettuce. Other modulating factors were bacterial incubation time on plants, strain, and leaf side inoculated, while sample processing method had no impact. Our findings highlight the importance of selecting relevant cultivars and strains, and of standardizing inoculation and incubation procedures, in these types of assessments. Moreover, results support the need to start reporting cultivars implicated in foodborne illness outbreaks to facilitate the identification and study of plant traits that impact food safety risk.

L eafy vegetables and their growing and processing environments have long been recognized for their potential to harbor pathogenic Escherichia coli, including E. coli O157:H7 (1,2).Based on the foodborne outbreak data between 2016 and 2020, the highest percentage of foodborne disease cases caused by E. coli O157:H7 in the United States were attributed to vegetable row crops (3).More than 40 outbreaks were linked to leafy greens contaminated with Shiga toxin-producing E. coli in the United States and Canada from 2015 to 2019 (4).Among the cases where information regarding the source of contamination was available, Romaine lettuce (54%) was more frequently implicated compared to other varieties of leafy greens (4)(5)(6)(7)(8).Several of these outbreaks were linked to environmental strains isolated from implicated farms or proximal areas (9)(10)(11), providing strong evidence that Romaine lettuce crops were contaminated at the pre-harvest stage.Therefore, there is a clear necessity to investigate this pathogen-com modity pair using live plants to identify the most important factors affecting this human pathogen plant association.Published studies are not standardized, however, and the specific variables that affect bacterial recovery from lettuce are not well understood.
Previous studies have investigated strain and incubation time as factors influencing the survival of E. coli O157:H7 on Romaine lettuce plants but reports are variable.Inoculation of E. coli O157:H7 strain ATCC 700728 and EC4045 onto live Romaine lettuce (cultivar "Parris Island") yielded a 2-3 log reduction in the culturable population within the first 24 hours and more than 6 log, 7 days post-inoculation, with equivalent cell density between different strains (12).Following the inoculation of E. coli O157:H7 strain ATCC 700728 onto live Romaine lettuce (cultivar "Parris Island Cos"), the recovered population size showed a decrease of less than 2 log during the first 24 hours and 4-5 log, 7 days post-inoculation (13).The same strain on live lettuce plants (cultivar "longifolia") resulted in a decline of less than 1 log within 48 hours of inoculation (14).Although these varying results may be due to variations in additional experimental conditions, the lack of consensus among studies might also be due to differences in the methods used to inoculate, retrieve, and enumerate cells on live plants, and in the differential interactions governed by the specific characteristics of the given pathogen strain and lettuce cultivar.
This study, therefore, was designed to systematically compare the interaction between two cultivars and two strains of E. coli O157:H7 to evaluate different experi mental setups and the efficiency of different sample processing methods in recovering and quantifying the pathogen from live Romaine lettuce.We used the commonly used strain CDC EDL 933 and a strain from a 2019 Romaine lettuce-associated outbreak (9) to inoculate a commercially appropriate Romaine lettuce cultivar "Rio Bravo" and the red Romaine cultivar "Outredgeous" for comparison.The two lettuce cultivars were chosen to represent the popular varieties (green and red Romaine lettuce) on the market.Results from the study distinguish variables that impact this pathogen-commodity association and also elucidate which factors yield maximal pathogen recovery from lettuce.This can help design, optimize, and standardize future studies evaluating bacteria-lettuce interactions.

Bacterial cultures and growth condition
E. coli O157:H7 CDC EDL933 (ATCC 43895) and FDA#MD2019EC2705C (2705C hereafter) were used in this study.The E. coli 2705C strain was isolated from a Romaine lettuce sample during the 2019 lettuce outbreak associated with E. coli O157:H7 (9).Strains were adapted to rifampicin in Tryptic Soy Broth (Difco, BD, Franklin Lakes, NJ, USA) and grown on Trypticase soy agar (TSA) (Difco, BD) amended with 50 µg/mL rifampicin (TSAR) (MP Biomedicals, Solon, OH, USA).To prepare the inoculum for lettuce inoculation, a single colony on TSAR plates for each strain was individually streaked onto a fresh TSAR plate and incubated at 37°C for 18 hours.After 18 hours, separate colonies were transferred to 5 mL of sterilized 0.1% peptone water (PW) (Difco, BD) and diluted to 7.2 ± 0.1 log CFU/mL.Exact inoculum concentrations were confirmed by serial plating on TSAR.

Lettuce growth and inoculation
Seeds of two Romaine lettuce (Lactuca sativa L. var.longifolia) cultivars, a green variety, "Rio Bravo" that is commonly commercially grown in the Salinas Valley, CA, and a red Romaine, "Outredgeous, " were planted in Sunshine Professional Growing Mix #1 (Sun Gro Horticulture Distribution, Bellevue, WA) and grown in a greenhouse for 4 weeks under a 16 hours photoperiod at 22°C-24°C.Potted plants aged 21-28 days were transferred to a biosafety cabinet and allowed to acclimate for 2 hours prior to inoculation.Aliquots of 100 µL (10-15 drops) of the freshly prepared E. coli O157:H7 EDL933 or 2705C inoculum were spot-inoculated onto the abaxial or adaxial side of the third to sixth true leaves of live plants and left inside the biosafety cabinet with lights and blower turned off for 24 or 48 hours at 20°C.The relative humidity was measured at 30%-40%.

Sample processing and enumeration of E. coli O157:H7 on lettuce leaves
At 24 or 48 hours post-inoculation (hpi), the four inoculated leaves per plant were excised and individually transferred to sterile sampling bags with mesh (Whirl-Pak, Nasco, Modesto, CA, USA) with 5 mL of sterile 0.1% PW.E. coli O157:H7 cells were recovered via either stomaching in a Stomacher 400 laboratory blender (Seward, Westbury, NY, USA) at 250 rpm for 5 minutes or by manual homogenizing to a watery paste with a pestle passed over the bag for 1 minute.The filtered wash or homogenate was serially diluted in 0.1% PW, drop-plated onto TSAR and incubated overnight at 37°C before counting.Microbial plate counts from sample homogenates were log transformed (log CFU/mL) and the microbial load per leaf (log CFU/leaf ) calculated.Results from individual leaves on the same plant were averaged to generate a single biological replicate.The limit of detection in sample homogenates was 1.7 log CFU/mL, which equaled to 2.4 log CFU/leaf.All the results in this study were above the limit of detection.

Statistical analysis
This study evaluated two processing methods to recover two E. coli O157:H7 strains inoculated onto either side of lettuce leaves of two cultivars and incubated for two different durations.Each experiment was repeated twice, where each replicate consisted of four biological replicates (four individual lettuce plants) for each tested condition or factor.Data from the two replicates were pooled.A Standard Least Squares model was fit to the data and Tukey's Honestly Significant Difference test was employed post-hoc in JMP (Version 15.2, SAS Institute, NC, USA).

RESULTS
The standard least square model fit to the data was significant (P < 0.001) with an acceptable fitness to the observed values (RMSE = 0.89).The model explained 29.2% of the variability.The tested factors of cultivar, incubation time, leaf side inoculated, and strain had significant effects on the model predictions, while sample processing method did not (Tables 1 and 2).The recovered levels of the two E. coli O157:H7 strains applied to leaves of two lettuce cultivars are represented in Fig. 1.

Effect of sample processing method on the recovery of E. coli O157:H7 on lettuce plants
The overall population levels of E. coli O157:H7 recovered from lettuce leaves were not significantly affected by the sample processing method used (Table 1).The average recovery from the leaf homogenizing method (4.9 ± 0.1 log CFU/leaf ) was not signifi cantly different from stomaching (4.8 ± 0.1 log CFU/leaf ) (Table 1).No difference was observed between stomaching and homogenizing when data for the two bacterial strains, the inoculated leaf sides, the incubation times, and the lettuce cultivar were considered separately (Table 2).While the sample processing method had little effect on the overall model, interactions with some other factors were detected and are described below.
Higher average counts from "Rio Bravo" when compared to "Outredgeous" were also observed when the effects of processing method, inoculated leaf side, or incubation time were differentiated (Table 2).The recovery level from stomaching (0.68 log higher; P < 0.001) or homogenizing (0.67 log higher; P < 0.001) of "Rio Bravo" was higher than "Outredgeous, " indicating that cultivar, not method, was the driver in this interaction.Counts retrieved from "Rio Bravo" were always higher than "Outredgeous, " regardless of leaf side inoculated; by 0.71 log (P < 0.001) on the abaxial side of leaves and by 0.64 log (P < 0.001) on the adaxial side of leaves.Differences in bacterial counts between the two cultivars were detected at both incubation times, with "Rio Bravo" yielding higher counts than "Outredgeous" at 24 (0.84 log higher; P < 0.001) and 48 hpi (0.5 log higher; P < 0.01).

Effect of incubation time on the recovery of two E. coli O157:H7 strains from lettuce plants
Incubation time had a significant effect on E. coli O157:H7 counts retrieved from lettuce leaves (P < 0.001) (Table 1).The overall recovery at 24 hpi (5.0 ± 0.1 log CFU/leaf ) was higher than at 48 hpi (4.6 ± 0.1 log CFU/leaf ) (P < 0.001) (Table 1); but this was attributa ble only to counts from cultivar "Rio Bravo" at 24 hpi (5.5 ± 0.1 log CFU/leaf ), which were higher than at 48 hpi (4.9 ± 0.1 log CFU/leaf ) (P < 0.001) (Table 2).No incubation time effect was detected on the cultivar "Outredgeous." Although an incubation time effect was only detected for samples processed by stomaching (5.1 ± 0.1 log CFU/leaf at 24 hpi   2), suggesting that this method was superior to homogenizing, this difference was driven by cultivar; all the counts from the homogenizing/stomaching method after 24-hour incubation time from "Rio Bravo" were higher than all combinations from "Outredgeous" (P < 0.05).The effect of incubation time on bacterial recovery was also affected by strain type.Higher counts of EDL933 (5.2 ± 0.1 log CFU/leaf ) and 2705C (4.9 ± 0.1 log CFU/leaf ) were recovered from lettuce incubated for 24 hours when compared to 2705C counts from a 48-hour incubation (4.5 ± 0.1 log CFU/leaf ) (P < 0.01), but no significant difference between the different

Effect of bacterial strain on the recovery of E. coli O157:H7 from lettuce plants
The E. coli O157:H7 strain used affected its recovery from lettuce plants (P < 0.01) (Table 1).The overall recovery of E. coli O157:H7 from lettuce leaves varied by strain with higher counts obtained for strain EDL933 (5.0 ± 0.1 log CFU/leaf ) compared to strain 2705C (4.7 ± 0.1 log CFU/leaf ) (P < 0.01) (Table 1).When assessed by cultivar, however, EDL933 was higher than 2705C on cultivar "Rio Bravo" (0.6 log difference; P < 0.001) but not on "Outredgeous" (Table 2).Moreover, a higher EDL933 (4.9 ± 0.1 log CFU/leaf than 2705C (4.5 ± 0.1 log CFU/leaf ) (P < 0.01) retrieval was only detected on adaxial side-inoculated plants, but not abaxial side-inoculated plants (Table 2).Results in Table 2 also suggested that the effect of strain difference was not observed when the effect of incubation time was differentiated, indicating that incubation time was a stronger driver than strain type.

Effect of leaf side inoculated on the recovery of two E. coli O157:H7 strains from lettuce plants
The leaf side inoculated affected the recovery of E. coli O157:H7 from lettuce plants (P < 0.01) (Table 1).Although the average E. coli O157:H7 counts recovered from plants inoculated on the abaxial side were higher than the adaxial side-inoculated plants, the difference was small (0.35 log; P < 0.01) (Table 1) and strain dependent.While higher recovery was observed on lettuce inoculated with 2705C on its abaxial side (4.9 ± 0.1 log CFU/leaf ) compared to adaxial inoculation (4.5 ± 0.1 log CFU/leaf ) (P < 0.01), no difference between leaf side was detected with EDL933 (Table 2).Similarly, although the average recovery of E. coli O157:H7 inoculated onto the abaxial side (5.3 ± 0.1 log CFU/ leaf ) was higher than the cells inoculated onto the adaxial side (4.8 ± 0.1 log CFU/leaf ) when a 24-hour incubation period was applied (P < 0.01), the effect was lost at 48 hpi.The attenuating effect of adaxial leaf side was most marked on "Outredgeous, " which indicates that the effect was overcome when lettuce cultivar was differentiated (Table 2).

DISCUSSION
The dynamics of bacterial interactions with live plants are influenced by a multitude of factors, making it challenging to establish comparability across various human pathogens on plant studies.E. coli O157:H7 studies with Romaine lettuce present a relevant model for such studies, but the diversity in bacterial strains, lettuce cultivars, and techniques employed for inoculating, retrieving, and enumerating E. coli O157:H7 cells from lettuce confound comparative analysis.Garnering a clear understanding of the magnitude of influence of various variables, that can be manipulated during experimental design, on the recovery of E. coli O157:H7 from inoculated lettuce plants is essential for data interpretation.In this study, cultivar imposed the strongest influence on bacterial counts recovered from lettuce, followed by incubation time, bacterial strain, and the leaf side inoculated.Lettuce cultivar had the strongest effect on bacterial counts in this study.The higher average recovery level of E. coli O157:H7 from the green Romaine cultivar "Rio Bravo" compared to the red variety "Outredgeous" point to intrinsic differences between the two lettuce cultivars that impacted the attachment, survival, and recoverability of E. coli O157:H7.The varying surface characteristics (leaf traits), such as wax deposition, roughness, stomata, and trichomes can influence how well bacteria attach to and are retained on the leaf surface, which is mainly dictated by the species, cultivar, and age of plants (15,16).Previous studies have investigated the correlation between bacterial activity and the intraspecific phenotypic variability of produce.Significant variation in bacterial attachment for Salmonella enterica Typhimurium and E. coli O157:H7 was reported among nine lettuce genotypes and attributed in part to a differential plant immune response among cultivars (17).Truschi et al. also observed differences in E. coli attachment among two Romaine lettuce cultivars, where cultivar "Maraichere" was less susceptible to contamination than cultivar "Bionda degli Ortolani, " as affected by their different leaf roughness and water content (15).The colonization by E. coli O157:H7 on spinach cultivar "Tyee, " which had more prominent ridges and crevices on its leave surface, was higher than on cultivar "Space" or "Bordeaux" (18).Leaf surface topography was not examined in the cultivars tested in our study, but "Outredgeous" leaves had a visually smoother appearance than "Rio Bravo." It was reported that red Romaine lettuce cultivar "Outredgeous" had a mean surface roughness of 2.5 µm on both adaxial and abaxial leaf sides (19); while "Bionda degli Ortolani, " a green romaine lettuce cultivar similar to "Rio Bravo, " had a much higher roughness of 71.56 µm (15).Higher stomatal density and lower hydrophobicity have been suggested to promote the adhesion of microorganisms and hinder the detachment of E. coli O157:H7 from produce surface during subsequent washing treatment (15,16,20,21).Other studies, however, found no correlation between stomatal density and bacterial attachment (17,22).The chemical composition on the leaf surface of lettuce cultivars can also differ, affecting the availability of nutrients for bacteria.Differences in the types and amounts of sugars and other organic compounds on lettuce leaves have been reported in previous studies to affect the ability of E. coli O157:H7 to adhere and persist (23)(24)(25).In addition, it is possible that differential levels of specialized metabolites in various lettuce types (26,27) could result in leaf cuticle and surface differences that modulate bacterial associations.Although the exact mechanism remains unknown, it has been reported that the red leaf lettuce cultivar "New Red Fire, " characterized by higher phenolic content, exhibited reduced survival rates for S. enterica Typhimurium and E. coli O157:H7 compared to the green Romaine lettuce cultivar "Coastal Star" (28), and the red oak-leaf lettuce "Mascara" with higher flavonoid content than the green Romaine "Parris Island Cos" was also less favorable for Salmonella (27).Despite the strong evidence for plant genotype as a strong driver of the E. coli O157:H7-lettuce association, lettuce cultivar data pertaining to recalls and outbreaks are rarely made publicly available.This information could be useful for identifying characteristics that may make a cultivar more susceptible to contamination and could inform food safety plans.
Bacterial residence time on the lettuce surface led to reduced recovery of E. coli O157:H7, with the difference depending on the lettuce cultivar, bacterial strain, and leaf side inoculated.Although a growing lettuce plant offers microsites for bacterial attachment and colonization, E. coli O157:H7 adhering to the leaf are subject to challenging environmental conditions, including low humidity, lack of nutrients, and competition from native microorganisms (24).These restrictive factors could all adversely affect the survival of E. coli O157:H7 on the leaves of a growing lettuce plant.Previous studies have explored the survival of non-pathogenic E. coli and E. coli O157:H7 on Romaine lettuce grown in the field and in controlled conditions and most report a very rapid (1-4 log) population decline during the initial 2-4 days post inoculation (13,(29)(30)(31).The outcomes of the current study exhibited a comparable pattern.Interestingly, an effect of incubation time (between 24 and 48 hpi) was detected in the cultivar "Rio Bravo" but not "Outredgeous." This difference could be linked to the differential favorability of the two cultivars."Outredgeous" appeared to present a less favorable microniche and hence bacterial population decline could have been more immediate.On the other hand, more amenable conditions on "Rio Bravo" would have slowed the decay rate in the initial 24 hours before the gradual inactivation during the subsequent incubation.Indeed, a biphasic decline for enteric pathogens on lettuce, with a more rapid rate of decay in the first phase and a transition to a slower decay rate in the second phase, commencing within the first 24 hours was reported for field studies in a multi-regional study (31).
The higher overall recovery of E. coli O157:H7 EDL933 was unexpected, as we had hypothesized that the outbreak strain 2705C would exhibit superior persistence on lettuce.Similar to our findings, Xicohtencatl-Cortes et al. reported that when compared to another clinical strain E. coli O157:H7 86-24, strain EDL933 was recovered at a higher level from inoculated lettuce and spinach leaves after 9 hours (32).Enhanced attachment of E. coli O157:H7 EDL933 to leaf surfaces has been attributed to flagella and the type III secretion system and differential expression levels of adhesion and resistant genes (32).The interaction of the strain effect with cultivar and incubation time observed in this study emphasized the complex and specific nature of the plant-cultivar × bacterial strain relationship.Each strain could possess or express different levels of adhesion factors or resistance-related genes when associating with lettuce that allow stronger attachment to the leaf surface or to better overcome lettuce defense mechanisms based on the physicochemical properties and defense response of the given cultivar.In our study, assessing an outbreak related strain-cultivar pair, strain 2705C did not exhibit traits that indicated any exceptional advantage in the lettuce niche.On the other hand, strain EDL933, isolated from hamburger meat several decades ago (33), may not represent typical behavior for an environmental or leafy green associated strain.Due to the extremely high variability of enteric pathogen-plant associations, using relevant strains is crucial for accurately building our understanding of these systems.
The higher bacterial recovery from the abaxial than the adaxial side of leaves corroborates previous studies, where the recovery of E. coli O157:H7 at 2-3 days after inoculation was significantly higher when lettuce was inoculated on the abaxial side of leaves as opposed to the adaxial side (29,34,35).The leaf surface properties that may promote better bacterial adhesion or persistence on the abaxial leaf side possibly include increased stomatal size and density (19,36) and higher physicochemical heterogeneity (35).In our study, leaf side affected recovery only for strain 2705C and only when cultivar was differentiated, suggesting a weak effect of leaf side in the pathogen-cultivar systems used.On the other hand, incubation time did interact with leaf side for strain 2705C, whereby higher recovery was obtained from abaxially inoculated leaves after 24 hours compared to adaxially inoculated leaves 48 hpi.This discrepancy suggests that this strain's dieoff rate was accelerated on the adaxial leaf side.
In this study, the choice of sample processing method (manual homogenizing vs stomaching) did not appear to significantly affect the transfer of E. coli O157:H7 from Romaine lettuce to the wash solution, resulting in equivalent bacterial recovery from lettuce.Recovery of Salmonella from leafy vegetables using homogenizing or stomaching has also been reported to yield equivalent results (37,38).Liberation of microorganisms from plant matrices is a critical step in the quantitative assessments of microbial populations on fresh produce (39).In our study, we opted for stomaching as a frequently used and recommended method for bacterial retrieval from leafy greens (37,39,40), and leaf tissue homogenizing for its theoretical capacity to capture both epiphytically and endophytically associated bacteria (38,39,41).We had hypothesized that the recovery method would yield different bacterial counts based on their effect on plant tissue.While maceration of plant tissue with the aid of a mortar and pestle or via blending could release both epiphytic and endophytic bacteria associating with plants, the tissue homogenizing method leads to leakage of plant intracellular content, such as polysaccharides and phenolic compounds, which could serve as microbial growth or PCR inhibitors and affect the efficiency in bacterial detection (42).In addi tion, tissue injury may lead to the activation of a membrane-bound NADPH oxidase and generate reactive oxygen species as one of the earliest responses at wound sites (also known as the "oxidative burst"), which could also induce antimicrobial stress in bacterial cells (43).Studies also suggested that Romaine lettuce could respond to various bacteria and to injury by producing a diverse range of secondary metabolites, including benzoic acid, lettucenins (sesquiterpene phytoalexins), antimicrobial peptides, or essential oil compounds, which provide protection against bacterial invasion (24,44,45).The resultant stress from antimicrobial compounds in the wash solution could underestimate bacterial quantification by the homogenizing method.On the other hand, the stomaching method asserts mechanical force to dislodge cells from plant tissues by operating motorized reciprocating paddles to apply pressure against the sample bag and could also recover E. coli O157:H7 cells that may be attached to or embedded within the lettuce surface (39,46,47).However, stomaching may yield incomplete bacterial recovery and the surface roughness and hydrophobic properties of the Romaine lettuce leaves may require careful optimization of stomaching duration to dislodge the maximal number of bacterial cells from the plants (38,48).In this study, we indeed noticed that the stomaching method resulted in significantly less leaf tissue damage and visible leakage of plant cellular material in the wash solution, compared to homogenizing, but no difference in bacterial recovery was detected between the two methods.As a result, while neither of the processing methods is likely completely "accurate" in recovering 100% of the E. coli O157:H7 cells from Romaine lettuce, the techniques yielded equiva lent outcomes.The choice between these methods may be left to practical considera tions, such as operator skill, equipment availability, and the specific characteristics of the lettuce samples being tested.

Conclusion
In recovering E. coli O157:H7 from live Romaine lettuce, comparison of tested factors found that cultivar was the dominant influencing factor, followed by incubation time, leaf side inoculated, and bacterial strain.Sample processing method, on the other hand, had no significant effect on the bacterial recovery rate.Green Romaine cultivar "Rio Bravo" yielded higher levels of E. coli O157:H7 than the red variety "Outredgeous" even when all the other factors were differentiated individually or as interacting factors.This emphasized the stronger effect of cultivar over other factors driving E. coli O157:H7 persistence on lettuce, and pointed to the importance of choosing appropriate cultivars in experimental design.Inoculation and incubation procedures were also relevant variables to standardize.Abaxial inoculation and an incubation time of 24 hours were optimal, although the latter factor interacted with lettuce cultivar, bacterial strain, and leaf side inoculated.The overall bacterial recovery from lettuce plants also varied by bacterial strain, with higher counts obtained for strain EDL933 compared to strain 2705C.This highlighted the implication of preferentially selecting suitable outbreak strains such as 2705C over model strains such as EDL933 in human pathogen on plant studies, for a more truthful representation of the pathogen-host interaction.Indeed, this study pointed out the significance of selecting relevant pathogen-cultivar pairs.While outbreak strain information is accessible to researchers, outbreak-implicated plant cultivar information is not usually divulged.The strong effect of cultivar in driving E. coli O157:H7-lettuce associations revealed in this study supports the need to start collecting cultivar information during foodborne illness outbreak investigations, to help design the most appropriate pathogen-plant interaction studies, and identify plant characteristics that may increase food safety risk.The observed influence of the different experimental factors on the recovery of E. coli O157:H7 on live Romaine lettuce could contribute to the development of standardized methods for the inoculation, retrieval, and enumeration of enterobacteria on live plants, to enhance the recovery efficiency and promote compara bility in studies evaluating lettuce-bacteria interactions.

vs 4 .
5 ± 0.1 log CFU/leaf at 48 hpi; P < 0.001) (Table incubation times was observed for EDL933.In addition, recovery from the abaxial side of the leaves at 24 hpi (5.3 ± 0.1 log CFU/leaf ) was higher than at the 48-hour incubation (4.8 ± 0.1 log CFU/leaf ) (P < 0.01).No difference between 24 and 48 hpi counts was observed on plants inoculated on the adaxial side.

FIG 1
FIG 1 Mean recovered levels of E. coli O157:H7 strains EDL933 and 2705C in log CFU/leaf from abaxial or adaxial sides of lettuce leaves of cv."Rio bravo" or "outredgeous" following 24-or 48-hour incubation periods on live plants.Recovery was via manual homogenizing or stomaching.Error bars indicated SE.Different letters indicate statistically significant difference (P < 0.05).Average ± SE.

TABLE 1
Comparison of the least square means (LS mean) of the recovered bacterial counts (log CFU/leaf ) for the tested independent factors a a Tested factors include bacterial strains (EDL933 and 2705C), lettuce cultivar ("Rio Bravo" [RB] and "Outredgeous" [OT]), leaf side inoculated (abaxial [Abx] and adaxial [Adx]), on-plant incubation period (24 and 48 hours), and sample processing method (homogenizing [H] and stomaching [ST]).Results of the effects test for the independent factors included in the model are displayed on the first row for each factor listed.SE is the standard error.Different letters indicate statistically significant differences within each factor (P < 0.05).NS stands for not significant.

TABLE 2
Comparison of the least square means (LS mean) of the recovered bacterial counts (log CFU/leaf ) for the interactions of the tested independent factors a