Pathogenicity of Beauveria bassiana PfBb and Immune Responses of a Non-Target Host, Spodoptera frugiperda (Lepidoptera: Noctuidae)

Simple Summary In this study, we tested the pathogenicity of Beauveria bassiana PfBb on the important agricultural pest Spodoptera frugiperda (Lepidoptera: Noctuidae) by determining the relative activities of protective enzymes and detoxifying enzymes in different larval instars. Our results show that the B. bassiana PfBb strain could infect all six larval instars of S. frugiperda, and its virulence to S. frugiperda larvae gradually increased with an increase in spore concentration. Furthermore, the activities of protective enzymes (i.e., catalase, peroxidase, and superoxide dismutase) and detoxifying enzymes (i.e., glutathione S-transferases, carboxylesterase, and cytochrome P450) in S. frugiperda larvae of the first three instars infected with B. bassiana PfBb changed significantly with infection time, but such variations were not obvious in the fifth and sixth instars. These findings laid the foundation for further research on the mechanism by which B. bassiana controls S. frugiperda. Abstract Exploring the pathogenicity of a new fungus strain to non-target host pests can provide essential information on a large scale for potential application in pest control. In this study, we tested the pathogenicity of Beauveria bassiana PfBb on the important agricultural pest Spodoptera frugiperda (Lepidoptera: Noctuidae) by determining the relative activities of protective enzymes and detoxifying enzymes in different larval instars. Our results show that the B. bassiana PfBb strain could infect all six larval instars of S. frugiperda, and its virulence to S. frugiperda larvae gradually increased with an increase in spore concentration. Seven days after inoculation, the LC50 of B. bassiana PfBb was 7.7 × 105, 5.5 × 106, 2.2 × 107, 3.1 × 108, 9.6 × 108, and 2.5 × 1011 spores/mL for first to sixth instars of S. frugiperda, respectively, and the LC50 and LC90 of B. bassiana PfBb for each S. frugiperda instar decreased with infection time, indicating a significant dose effect. Furthermore, the virulence of B. bassiana PfBb to S. frugiperda larvae gradually decreased with an increase in larval instar. The activities of protective enzymes (i.e., catalase, peroxidase, and superoxide dismutase) and detoxifying enzymes (i.e., glutathione S-transferases, carboxylesterase, and cytochrome P450) in S. frugiperda larvae of the first three instars infected with B. bassiana PfBb changed significantly with infection time, but such variations were not obvious in the fifth and sixth instars. Additionally, after being infected with B. bassiana PfBb, the activities of protective enzymes and detoxification enzymes in S. frugiperda larvae usually lasted from 12 to 48 h, which was significantly longer than the control. These results indicate that the pathogenicity of B. bassiana PfBb on the non-target host S. frugiperda was significant but depended on the instar stage. Therefore, the findings of this study suggest that B. bassiana PfBb can be used as a bio-insecticide to control young larvae of S. frugiperda in an integrated pest management program.


Introduction
Insect epidemics caused by entomopathogenic fungi can naturally suppress pest populations, reduce pesticide use, and maintain ecological balance. A wide range of entomopathogenic fungi, particularly those belonging to the Beauveria, Isaria, Lecanicillium, and Metarhizium genera, have been widely studied [1][2][3][4]. However, an improved understanding of the pathogenicity of entomogenous fungi is a prerequisite for evaluating pest control efficiency. To date, some entomopathogenic fungi have been commercialized and widely used for pest management in greenhouses and fields, especially Beauveria bassiana [5][6][7]. Many studies have shown that B. bassiana has significant lethal effects on a variety of insect pests, such as Ostrinia furnacalis [8], Cylas formicarius [9], Bemisia tabaci [10], Lycorma delicatula [5], Frankliniella occidentalis [6], and Leptinotarsa decemlineata [4]. Although B. bassiana has strong host specificity, the virulence of different strains to non-target host pests varies greatly [11][12][13]. Therefore, clarifying the virulence of a new B. bassiana strain to non-target host pests can provide essential information for potential application in pest control on a large scale.
Insects are usually exposed to xenobiotics in the environment, in which their successful adaptation to these environmental risks requires an efficient system to detoxify and eliminate these substances from their bodies. For example, B. bassiana affects the normal physiological and metabolic activities in infected insects [14]; it causes damage to the free amino acids in the hemolymph, and interferes with a variety of important metabolic enzymes, such as glutathione S-transferases (GST) [15], carboxylesterase (CarE) [16], and cytochrome P450 (CYP450) [17], which play an important role in the detoxification of penetrated xenobiotics [18,19]. In addition, the protective enzyme systems in insects, mainly including catalase (CAT), peroxidase (POD), and superoxide dismutase (SOD), could reduce oxidative damage by degrading H 2 O 2 [20]. The toxic reactive oxygen species (ROS) are generated and accumulated by the innate immune cells of organisms when insects are under stressful conditions, such as the applications of chemical pesticides and entomopathogenic fungi [21][22][23]. In many studies, to maintain normal cellular function and reduce oxidative damage, different types of ROS are scavenged by protective enzymes [24,25]. Therefore, understanding the activity of detoxifying enzymes and protective enzyme systems is helpful for exploring the resistance mechanism of insects during fungal infection, which is of great significance for pest control.
The fall armyworm (FAW), Spodoptera frugiperda (J.E. Smith) (Lepidoptera: Noctuidae), is a notorious migratory pest originating from the Americas [26]. It invaded Africa in 2016 [27], and then Southeast Asian countries such as India [28], Myanmar, and Vietnam in 2018 [29]. Spodoptera frugiperda was first recorded in Yunnan, China, in 2019 [30]. It has become a major agricultural pest across borders and continents due to its wide host range, high reproductive capacity, and fast dispersal ability [31]. A variety of natural enemies attack Spodoptera frugiperda, but they have rarely been used in the biological control of this pest in China [32]. Thus far, the control of S. frugiperda has mainly relied on chemical insecticides [33,34]. However, the long-term application of insecticides could result in the development of insecticide resistance and damage to the environment. Alternatively, a large number of studies have shown that B. bassiana is pathogenic to S. frugiperda [35,36], and that B. bassiana could be compatible with natural enemies to improve the control effect [37]. Therefore, entomopathogenic fungi could be prime substitutes for insecticides [38,39]. However, most entomopathogenic isolates show high mortality only to young larvae but low pathogenicity to old larvae [40]. The potential causes are still unknown.

Insects
The larvae of S. frugiperda were collected from corn fields at Jiaoyi township (119 • 28 E, 22 • 83 N), Hengzhou City, Guangxi Zhuang Autonomous Region, China. The larvae were individually reared with corn leaves in plastic Petri dishes (9 cm diameter × 1.5 cm height) at 26 ± 1 • C, with 75 ± 10% RH, and a photoperiod of 14:10 h (L:D) until pupation in the artificial climate chambers (LRH-250, Changzhou Putian Instrument Manufacturing Co., Ltd., Changzhou, China). Newly emerged moths were paired in plastic cups (11.5 cm diameter × 15.5 cm height) and fed with 10% honey solution supplied in a small cotton wick. The cups were replaced with a new one after the eggs laid on the cup wall were observed. The cups with eggs were maintained in plastic boxes (24.5 cm diameter × 12.0 cm height) until the hatching of neonate larvae. S. frugiperda larvae were reared on leaves of corn seedlings (Meiyu No. 3; Hainan Lvchuan Seed Co., Ltd., Hainan, China) planted in the laboratory, and newly hatched larvae of the fourth generation were used for the experiments.

Virulence of B. bassiana PfBb on S. frugiperda Larvae
Four conidial concentrations (i.e., treatments: 1 × 10 5 , 1 × 10 6 , 1 × 10 7 , and 1 × 10 8 spores/mL) with Tween-80 (0.05%) as the control (CK) were set up to test the virulence of B. bassiana PfBb on six S. frugiperda larval instars. There were three replicates for each treatment, with 30 individuals per treatment. For each treatment, the S. frugiperda larvae were immersed individually in the solution for 10 s and placed in a Petri dish. The larvae were reared at 26 ± 1 • C, 75 ± 10% RH, and a photoperiod of 14:10 h (L:D) and checked twice (09:00 and 21:00) a day for seven consecutive days. A larva was regarded as dead if it did not respond when touched with a brush. The number of dead larvae was recorded, and the cumulative mortality was calculated as [(treatment group mortality-control group mortality)/(1-control group mortality) × 100%]. The median lethal concentration (LC 50 ) of S. frugiperda larvae was calculated based on the cumulative mortality of each treatment during the first seven days. The dead larvae were removed into a culture dish sterilized with 75% alcohol to moisten the culture. If mycelia were observed on the larval integument, they were considered cadavers. The number of insect cadavers was recorded to calculate the cadaver rate (number of dead larvae growing mycelium/total number of dead larvae × 100%) [43].  50 and LC 90 were calculated using the model to simulate time and dose-response parameters. The establishment of the TDM model requires a bioassay, including i dose (the dose is i), day after treatment is j, the cumulative mortality probability p ij (p ij = 1-exp[-exp(τ j + βlog 10 (d j ))]) caused by any dose d i (i = 1, 2, . . . , i) at any time t j (j = 1, 2, . . . , j), where β is the slope of dose effect and τ j is the cumulative time effect parameter from t 1 to t j . Note that p ij is a time-dependent variable. However, the above equation cannot directly fit the bioassay data because the binomial variable, p ij , for modeling does not satisfy the requirement for independence of time. To guarantee that the observed mortality probability was independent of time, the true mortality that occurred at d j at the interval [t j−1 , t j ] was considered. That is, the conditional death probability q ij , which can be expressed as q ij = 1-exp[-exp(γ j + βlog 10 (d j ))], where, γ j is the time effect parameter to be estimated in the time interval [t j−1 , t j ] [44].

Enzymatic Sample Preparation
The S. frugiperda larvae of different instars were treated by the immersion method with 1 × 10 8 spores/mL of B. bassiana PfBb and Tween-80 (0.05%) as the treatment group and control group, respectively. The treated S. frugiperda larvae were reared the same way as above. During culturing, the treated larvae were sampled randomly at 12, 24, 36, 48, 60, and 72 h after treatment with B. bassiana PfBb or Tween-80 (0.05%). The number of samples for each test group at every time point was 50 (1st and 2nd instars), 10 (3rd and 4th instars), and 5 larvae (5th and 6th instars), respectively. The insect samples were homogenized with 0.1 mol/L PBS (sample weight: PBS = 1 g:9 mL) using an electric homogenizer (JXFSTPRP-24L, Shanghai Jingxin Industrial Development Co., Ltd., Shanghai, China). Next, the tissue homogenate was centrifuged (Centrifuge 5424R, Eppendorf, Hamburg, Germany) at 12,000 rpm and 4 • C for 15 min. The supernatant, as the enzyme source, was transferred to new tubes and stored at −20 • C [45]. Each test was repeated three times, and the enzymatic activities associated with the observed larvae were determined.
Activity assays of each enzyme were carried out by following the manufacturer's instructions. The NH 3 + group of the protein molecule can be combined with the anion of Coomassie brilliant blue to make the solution turn blue, and the protein content can be calculated by measuring the absorbance at 595 nm wavelength after 10 min of reaction. CAT activity determination is defined as the amount of 1 µmol H 2 O 2 decomposed per milligram of tissue protein per second as 1 unit (U). Briefly, H 2 O 2 was utilized as a substrate, and the absorbance of the reaction was measured at 405 nm wavelength for 1 min. The determination of POD activity was defined as 1 U per mg of tissue protein per minute to catalyze 1 µg of substrate per minute at 37 • C. In short, the absorbance after 30 min of tissue protein reaction was measured at a wavelength of 420 nm. For SOD activity determination, the amount of SOD corresponding to the SOD inhibition rate of 50% per mg of tissue protein in 1 mL of reaction solution was taken as 1 U. In summary, the absorbance of the enzyme solution after reacting at 37 • C for 40 min was measured at a wavelength of 550 nm. GST activity determination. The non-enzymatic reaction was subtracted from the reaction per mg of tissue protein at 37 • C for 1 min, and the GSH concentration in the reaction system was reduced by 1 µmol/L as 1 U. That is, after the enzyme solution was reacted at 37 • C Insects 2022, 13, 914 5 of 21 for 10 min, the absorbance was measured at a wavelength of 412 nm. The CarE activity was determined as 1 U per mg of tissue protein at 37 • C with an increase of 1 in catalytic absorbance per minute. Briefly, the difference in absorbance was measured at a wavelength of 450 nm after the enzyme solution reacted for 10 and 190 s [45]. To determine CYP450 activity, a linear regression equation was made with the standard product. The absorbance of the enzyme solution at 450 nm after reacting at 37 • C for 60 min was measured, and the concentration value and enzyme activity of each sample were calculated according to the equation. CAT, POD, SOD, and GST enzyme activities in S. frugiperda larvae were calculated based on their protein content ( Figure S1).

Statistical Analyses
The normality and homoscedasticity of all data were tested prior to analysis using the Kolmogorov-Smirnov and Levene's tests, respectively. Data on cadaver rates and enzyme activities were analyzed using a one-way analysis of variance (ANOVA), followed by Tukey's honestly significant difference (HSD) multiple tests. The regression equation, LT 50 , LT 90 and their 95% confidence limits were calculated by Probit regression analysis. The effects of B. bassiana PfBb on enzyme activities in infected S. frugiperda larvae were analyzed by an independent samples t-test. The statistical significance level was determined to be alpha ≤0.05. A probit regression was applied to analyze the virulence of B. bassiana PfBb to six instars of S. frugiperda larvae, and calculate the LC 50 , LC 90 , and their 95% confidence limits. Survival curves were subjected to a log-rank test and graphed using GraphPad Prism 8.0.2 (GraphPad Software Inc., San Diego, CA, USA). The Hosmer-Lemeshow test was performed on the time-dose-mortality model using DPS 7.05, and a t-test was used to analyze the significance of the dose and time-effect parameters of different instar larvae. Data analyses were performed using SPSS 25.0 (IBM Corp., Armonk, NY, USA).

Effect of B. bassiana PfBb on S. frugiperda Larvae
The S. frugiperda larvae of different instars can be infected by B. bassiana PfBb. After the dead larvae were wet-cultured, white hyphae grew on both sides of the integument, and then the hyphae wrapped the whole body ( Figure 1).
Insects 2022, 13, x FOR PEER REVIEW 5 of 21 reaction system was reduced by 1 μmol/L as 1 U. That is, after the enzyme solution was reacted at 37 °C for 10 min, the absorbance was measured at a wavelength of 412 nm. The CarE activity was determined as 1 U per mg of tissue protein at 37 °C with an increase of 1 in catalytic absorbance per minute. Briefly, the difference in absorbance was measured at a wavelength of 450 nm after the enzyme solution reacted for 10 and 190 s [45]. To determine CYP450 activity, a linear regression equation was made with the standard product. The absorbance of the enzyme solution at 450 nm after reacting at 37 °C for 60 min was measured, and the concentration value and enzyme activity of each sample were calculated according to the equation. CAT, POD, SOD, and GST enzyme activities in S. frugiperda larvae were calculated based on their protein content ( Figure S1).

Statistical Analyses
The normality and homoscedasticity of all data were tested prior to analysis using the Kolmogorov-Smirnov and Levene's tests, respectively. Data on cadaver rates and enzyme activities were analyzed using a one-way analysis of variance (ANOVA), followed by Tukey's honestly significant difference (HSD) multiple tests. The regression equation, LT50, LT90 and their 95% confidence limits were calculated by Probit regression analysis. The effects of B. bassiana PfBb on enzyme activities in infected S. frugiperda larvae were analyzed by an independent samples t-test. The statistical significance level was determined to be alpha ≤0.05. A probit regression was applied to analyze the virulence of B. bassiana PfBb to six instars of S. frugiperda larvae, and calculate the LC50, LC90, and their 95% confidence limits. Survival curves were subjected to a log-rank test and graphed using GraphPad Prism 8.0.2 (GraphPad Software Inc., San Diego, CA, USA). The Hosmer-Lemeshow test was performed on the time-dose-mortality model using DPS 7.05, and a ttest was used to analyze the significance of the dose and time-effect parameters of different instar larvae. Data analyses were performed using SPSS 25.0 (IBM Corp., Armonk, NY, USA).

Effect of B. bassiana PfBb on S. frugiperda Larvae
The S. frugiperda larvae of different instars can be infected by B. bassiana PfBb. After the dead larvae were wet-cultured, white hyphae grew on both sides of the integument, and then the hyphae wrapped the whole body ( Figure 1).

Effect of B. bassiana PfBb on Mortality of S. frugiperda Larvae
Mortality of S. frugiperda larvae on each instar increased gradually with the increase in the spore concentration of B.  Figure 2). For example, the mortality rate of the first instar larvae increased from 25.00% at day 1 to 82.33% at day 5 after being treated at 1.0 × 10 8 spores/mL (Figure 2A). A similar mortality pattern was also detected for the second to fifth instars ( Figure 2B-E). After 7 days of treatment, the cumulative mortality of the sixth instar larvae treated with 1.0 × 10 7 and 1.0 × 10 8 spores/mL suspensions was 15.35% and 17.02%, respectively, which are significantly higher than those treated with CK (3.33%), 1.0 × 10 5 (3.42%), and 1.0 × 10 6 spores/mL (6.67%) ( Figure 2F). Obviously, the cumulative mortality gradually reduced with the increase in the larval instar (Figure 2A-F). After 7 days of treatment, the LC 50 of spore concentrations increased with the instar stage of S. frugiperda (Table 1).  Figure 2). For example, the mortality rate of the first instar larvae increased from 25.00% at day 1 to 82.33% at day 5 after being treated at 1.0 × 10 8 spores/mL (Figure 2A). A similar mortality pattern was also detected for the second to fifth instars ( Figure 2B-E). After 7 days of treatment, the cumulative mortality of the sixth instar larvae treated with 1.0 × 10 7 and 1.0 × 10 8 spores/mL suspensions was 15.35% and 17.02%, respectively, which are significantly higher than those treated with CK (3.33%), 1.0 × 10 5 (3.42%), and 1.0 × 10 6 spores/mL (6.67%) ( Figure 2F). Obviously, the cumulative mortality gradually reduced with the increase in the larval instar (Figure 2A-F). After 7 days of treatment, the LC50 of spore concentrations increased with the instar stage of S. frugiperda (Table 1).     (Figure 3). For example, the cadaver rate of the treatment of 1.0 × 10 8 spores/mL (76.67 ± 5.77%) was significantly higher than those other treatment concentrations at the first instar larvae (Figure 3). The cadaver rate of the 4th (36.67 ± 7.64%) and 5th (33.33 ± 7.64%) instar larvae under the treatment of 1.0 × 10 8 spores/mL was also significantly higher than other treatment concentrations ( Figure 3). However, there was no significant difference in the cadaver rate of the sixth instar larvae between different spore concentrations (  .49 × 10 31 y is the expected mortality and x is the logarithm of concentration. All data were calculated using the mortality rates on the seventh day after treatment.

Effect of B. bassiana PfBb on the Cadaver Rate of Infected S. frugiperda Larvae
For a given larval instar (i.e., first to fifth instars), the cadaver rate of instar larvae significantly increased with the increase of spore concentration (first instar: F3, 8 (Figure 3). For example, the cadaver rate of the treatment of 1.0 × 10 8 spores/mL (76.67 ± 5.77%) was significantly higher than those other treatment concentrations at the first instar larvae (Figure 3). The cadaver rate of the 4th (36.67 ± 7.64%) and 5th (33.33 ± 7.64%) instar larvae under the treatment of 1.0 × 10 8 spores/mL was also significantly higher than other treatment concentrations ( Figure  3). However, there was no significant difference in the cadaver rate of the sixth instar larvae between different spore concentrations (F3,8 = 3.42, p = 0.073) (Figure 3).

Time-Dose-Mortality (TDM) Model of B. bassiana PfBb on S. frugiperda Larvae
The cumulative mortality data of 1-5 days for the first to second instar larvae, 1-6 days for the third to fifth instar larvae, and 2-7 days for the sixth instar larvae were applied to the TDM model for analysis according to the mortality of S. frugiperda larvae before and after infection with spores of B. bassiana PfBb. The Hosmer-Lemeshow test for the heterogeneity of the goodness of fit for the binomial variable pij was not significant (first instar:  Table 2). The cumulative effect parameter τj increased with the increase in time, indicating that the dose effect, time effect, and their interaction effect significantly influenced the mortality of S. frugiperda larvae. The slopes (β) of dose response of first to sixth instar larvae to B. bassiana PfBb were

Time-Dose-Mortality (TDM) Model of B. bassiana PfBb on S. frugiperda Larvae
The cumulative mortality data of 1-5 days for the first to second instar larvae, 1-6 days for the third to fifth instar larvae, and 2-7 days for the sixth instar larvae were applied to the TDM model for analysis according to the mortality of S. frugiperda larvae before and after infection with spores of B. bassiana PfBb. The Hosmer-Lemeshow test for the heterogeneity of the goodness of fit for the binomial variable pij was not significant (first instar: X 2 7 = 0.45, p = 0.999; second instar: X 2 7 = 2.63, p = 0.917; third instar: X 2 7 = 1.72, p = 0.974; fourth instar: X 2 8 = 1.93, p = 0.983; fifth instar: X 2 8 = 2.13, p = 0.977; sixth instar: X 2 9 = 4.09, p = 0.905), indicating that the data fit the model well ( Table 2). The cumulative effect parameter τ j increased with the increase in time, indicating that the dose effect, time effect, and their interaction effect significantly influenced the mortality of S. frugiperda larvae. The slopes (β) of dose response of first to sixth instar larvae to B. bassiana PfBb were 0.4068, 0.5825, 0.5602, 0.6144, 0.6952, and 0.4701, respectively, suggesting that the fifth instar larvae were the most sensitive to the increase of B. bassiana PfBb spore concentration. The time effect parameters of first and second instar larvae to B. bassiana PfBb reached the maximum on the third day after infection (γ 3 ), and that of third instar larvae on the second day (γ 2 ), fourth and fifth instar larvae on the fourth day (γ 4 ), and sixth instar larvae on the fifth day after infection (γ 5 ). The subscript number of parameter symbol indicates the days after being treated.

Effect of B. bassiana PfBb Concentration on the Lethality of S. frugiperda Larvae
According to the TDM model, the estimated dose effect of S. frugiperda larvae infected with B. bassiana PfBb is shown in Figure 4. With the extension of time, the lethal concentration decreased, and the dose effect increased (Figure 4). With the increase in lethal concentration, the time effect increased (Figure 4). However, the actual maximum cumulative mortality of the first to sixth instar larvae is not up to 90%. Therefore, the estimated LC 90 lethal concentrations corresponding to the first to sixth instar larvae were rather large ( Figure 4). with B. bassiana PfBb is shown in Figure 4. With the extension of time, the lethal concen-tration decreased, and the dose effect increased (Figure 4). With the increase in lethal concentration, the time effect increased (Figure 4). However, the actual maximum cumulative mortality of the first to sixth instar larvae is not up to 90%. Therefore, the estimated LC90 lethal concentrations corresponding to the first to sixth instar larvae were rather large ( Figure 4).

Effect of B. bassiana PfBb on SOD Enzyme Activity in S. frugiperda Larvae
The SOD activity in S. frugiperda larvae did not significantly vary over the infection time in the first and fifth instars (F 5,12 = 0.92 and 2.20, respectively; p > 0.05) but significantly changed over time in the second, third, fourth, and sixth instars with a peak detected at 48, 24, 24, and 48 h, respectively (F 5,12 = 11.28, 5.16, 4.46, and 4.49, respectively; p < 0.05) (Figure 7). Compared to that in the control, SOD activity was significantly higher at 12 h (t 4

Effect of B. bassiana PfBb on GST Enzyme Activity in S. frugiperda Larvae
The GST activity in the first to fifth instars infected by B. bassiana PfBb significantly changed over infection time, with a peak detected at 36,12,12,36

Effect of B. bassiana PfBb on CYP450 Enzyme Activity in S. frugiperda Larvae
The CYP450 activity in the first five instars significantly changed over infection time (F5,12 = 6.87, 4.24, 19.00, 6.91 and 7.83, respectively; p < 0.05), while in the sixth instar it did not significantly vary over infection time (F5,12 = 1.05, p = 0.435) ( Figure 10). Compared to that in the control, CYP450 activity was significantly lower at 24 h (t4 =

Discussion
Beauveria bassiana is a promising biological control agent for insect pests. In this study, we show that B. bassiana PfBb, a strain collected from P. flammans larvae, had strong pathogenicity to the young larvae of S. frugiperda, and that the protective enzymes and detoxifying enzymes in S. frugiperda larvae play important roles in resisting the infection of B. bassiana PfBb, especially on young larvae. Our results show that the pathogenicity of B. bassiana PfBb on non-target host S. frugiperda, as well as the protective enzymes and detoxifying enzymes in immune resistance, are instar-stage dependent.
Virulence is a major measure of pathogenicity in entomopathogenic fungi. The degree of virulence is directly related to the ability of entomopathogenic fungi to reduce insect mortality despite the existence of host resistance. The results of the current study show that B. bassiana PfBb, a strain collected from P. flammans larvae, had strong virulence to first to third instar larvae of S. frugiperda, though its pathogenicity gradually decreased with the increase of the larval instars (Figures 2 and 3; Tables 1 and 2). Similar results have also been reported when pathogenic fungi are collected from other host species [35] or when other pest insects are infected with B. bassiana [46,47]. For example, the application of B. bassiana has a higher effect on the young larvae of Indarbela dea [48], suggesting younger larvae are more susceptible to B. bassiana than older larvae [49]. This may be attributed to four reasons. First, the insect epidermis is the first barrier to pathogenic fungal Figure 10. Effect of Beauveria bassiana PfBb on CYP450 activity in Spodoptera frugiperda. (A-F) represents the first to sixth instar larvae, respectively. Means (± SD), followed by different lowercase letters in the control or treatment line, are significantly different (Tukey's HSD test: p < 0.05). Asterisks indicate a significant difference between control and treatment at a given time (t-test: p < 0.05).

Discussion
Beauveria bassiana is a promising biological control agent for insect pests. In this study, we show that B. bassiana PfBb, a strain collected from P. flammans larvae, had strong pathogenicity to the young larvae of S. frugiperda, and that the protective enzymes and detoxifying enzymes in S. frugiperda larvae play important roles in resisting the infection of B. bassiana PfBb, especially on young larvae. Our results show that the pathogenicity of B. bassiana PfBb on non-target host S. frugiperda, as well as the protective enzymes and detoxifying enzymes in immune resistance, are instar-stage dependent.
Virulence is a major measure of pathogenicity in entomopathogenic fungi. The degree of virulence is directly related to the ability of entomopathogenic fungi to reduce insect mortality despite the existence of host resistance. The results of the current study show that B. bassiana PfBb, a strain collected from P. flammans larvae, had strong virulence to first to third instar larvae of S. frugiperda, though its pathogenicity gradually decreased with the increase of the larval instars (Figures 2 and 3; Tables 1 and 2). Similar results have also been reported when pathogenic fungi are collected from other host species [35] or when other pest insects are infected with B. bassiana [46,47]. For example, the application of B. bassiana has a higher effect on the young larvae of Indarbela dea [48], suggesting younger larvae are more susceptible to B. bassiana than older larvae [49]. This may be attributed to four reasons. First, the insect epidermis is the first barrier to pathogenic fungal infection [50]. The pathogenicity of entomogenous fungi to pests often decreases with the increase of larval instars due to the higher content of melanin in the insect epidermis and midgut of older instars, hindering fungal budding [51]. Second, the virulence of entomogenous fungi to pests is also closely related to the structure of the larval body wall [52], because the body wall of young larvae is relatively thin, while the waxy layer of the body wall gradually thickens with the increase of the instar stage, which prevents the invasion of B. bassiana [53]. Third, the developmental period of the first to sixth instar larvae of S. frugiperda is 3.0, 2.0, 2.0, 1.1, 1.5, and 4.3 d, respectively, at 25 • C [54]. We speculate that the time period of spores attaching to the body surface of older instars is shorter due to the shortening molting process. Fourth, insects can overcome different poisons by regulating changes in protective and detoxifying enzymes in the body [55,56]. The reduced pathogenicity of B. bassiana PfBb to older S. frugiperda larvae in this study may be due to the enzyme systems in the non-target host (see discussion below).
After the pathogen successfully invades a host, it activates the host's defense system, resulting in changes in the host's protective enzymes, especially SOD, POD, and CAT. It is well known that a large quantity of active oxygen will accumulate in the insect bodies, which stimulates an antioxidant enzyme response when insects suffer from stress caused by adverse factors [19,57]. Earlier studies have shown that SOD and CAT activities are inversely correlated with larval instars [58]. In this study, our results illustrate that the activity of protective enzymes and reaction times in different larval instars varied differently after being infected with B. bassiana PfBb. These findings indicate that young larvae are more sensitive faster response time to B. bassiana PfBb. The protective enzyme activity in the first and second larval instars significantly increased first and then decreased. Similar results have been reported in Bemisia tabaci, Mythimna separata, and Xylotrechus rusticus [19,59,60]. For example, B. bassiana leads to an increase in reactive oxygen species in X. rusticus larvae in the early stage of infection and a quick activation of the antioxidant enzyme system [19]. However, the infection of pathogenic fungi leads to the inhabitation of antioxidant enzyme activity in the late instar stages of S. frugiperda (Figures 5-7), so the ability of infected larvae to scavenging free radicals in the body is weakened. Previous studies have shown that the oxygen balance in the body is easily disrupted when insects are infected with B. bassiana [21,61] [62,63]. Moreover, the activity of protective enzymes determines the intensity of external stimuli [23], insecticide resistance, and stress resistance of insects [58,64]. In this study, we revealed that SOD and CAT activity in the fifth and sixth instars were not significantly different compared with the control. We speculate that B. bassiana PfBb has weaker stimulation to the older stage of non-target hosts and lower H 2 O 2 concentration after infection, resulting in insignificant changes in the activity of protective enzymes. However, we only measured the activity of protective enzymes in susceptible larvae of S. frugiperda, and the changes in oxygen free radicals and gene expression in susceptible larvae need further research.
To inhibit further damage by pathogenic fungi, detoxification enzymes, such as CarE, GST, and CYP450, in insects can efficiently metabolize exogenous toxic compounds [17]. Theoretically, insects resist the poisoning of different exogenous substances by enhancing the activity of detoxification enzymes and promoting the expression of detoxification enzyme genes, thereby improving their resistance to pathogenic fungi. In the present study, the detoxification enzyme activity and reaction time in different larval instars varied differently after being infected with B. bassiana PfBb. The detoxification enzyme activity in the first to fourth larval instars changed significantly compared with the control (Figures 8-10). Similar findings have also been found in X. rusticus and B. tabaci [19,59]. Our results revealed that young larvae are more susceptible and have a faster reaction time to B. bassiana. GST is an important detoxification enzyme in the metabolism of endogenous and exogenous substances in insects, and the increase in GST activity can be used as a sensitive indicator of tissue damage [65]. Our results illustrate that GST in S. frugiperda larvae was activated after B. bassiana PfBb was inoculated, but because the tissue was destroyed by poison, the GST synthesis capability was reduced with infection time, resulting in the GST increasing first and then decreasing (Figure 8). CarE is a specific catalyzing ester bond hydrolase that not only participates in lipid metabolism but also acts as a detoxification enzyme to metabolize exogenous toxins [16]. Our results show that CarE activity increased first and then decreased with increasing infection time (Figure 9), which is consistent with a previous report [19]. In contrast, the activity of CarE first decreased and then increased after being infected by B. bassiana, which is also reported in P. flammans [66]. The opposed CarE activity in target and non-target hosts could be attributed to their different resistances to the B. bassiana PfBb strain. Empirical evidence demonstrates that numerous responsive genes are differentially expressed in various hosts during infection by B. bassiana. For example, the increased or decreased expression of some special genes [i.e., the adhesion (Mad1), protease (Pr2), and secretory lipase] is an important factor for the difference in pathogenicity of B. bassiana to different hosts [13]. The differential expression of relative genes in target and non-target hosts infected by the B. bassiana PfBb strain needs further study. CYP450 is also a key enzyme involved in detoxification metabolism in insects. In this study, the CYP450 enzyme activity of the first instar larvae in the treatment was significantly lower than that of the control larvae ( Figure 10). Cytochrome P450 enzymes are known to activate ecdysone [67]. We suggest that CYP450 enzyme activity decreases after the first instar larvae are infected, and that the synthesis of ecdysone is inhibited, eventually leading to death. Furthermore, the CYP450 enzyme activity in the sixth instar larvae was not significantly different from that in the control (Figure 9), probably due to the fact that the B. bassiana PfBb is not activate in the fatty acid degradation pathway, because the differences in cytochrome P450 gene expression are caused by differential degradation of thrips fatty acids of strains with similar virulence [68]. In addition, a study found that the detoxification enzyme activity of Culex quinquefasciatus increased with increases in B. bassiana concentrations [69]. Therefore, we assume that the pathogenicity of B. bassiana PfBb is related to the activity of detoxification enzymes.

Conclusions
Our findings confirm that the B. bassiana PfBb strain can infect all larval instars of S. frugiperda and result in a higher cadaver rate, especially for younger larvae. Furthermore, the virulence of B. bassiana PfBb to S. frugiperda larvae gradually increased with the increase in spore concentration, and the LC 50 and LC 90 of B. bassiana PfBb for each S. frugiperda instar decreased with infection time, indicating a significant dose effect. Although protective enzymes and detoxification enzymes in S. frugiperda larvae are usually activated between 12 and 48 h after treatment, the activity of these enzymes in the first three larval instars changed significantly over infection time, while such change in the fifth and sixth larval instars is not obvious. Our results indicate that the pathogenicity of B. bassiana PfBb on a non-target host, S. frugiperda, is significant but instar-stage dependent. In summary, our results suggest that B. bassiana PfBb can be used as a bio-insecticide to control young larvae of S. frugiperda in the integrated pest management program.