1 Genotype effect on lifespan following vitellogenin knockdown 2 3 Kate E. Ihle1,2*, M. Kim Fondrk1,3, Robert E. Page1 and Gro V. Amdam1,4 4 5 1 Arizona State University, School of Life Sciences, Tempe, AZ 85287, USA 6 2 Smithsonian Tropical Research Institute, Panama City, Panama Apartado Postal 0843-03092, Republic of Panama 7 8 3 Shields Avenue, Davis, CA 95616-5270, USA 9 10 11 University of California, Davis, Department of Entomology and Nematology, 4 Norwegian University of Life Sciences, Department of Chemistry, Biotechnology and Food Science, Aas, 1432, Norway 12 13 14 Author email addresses: kateihle@gmail.com, mkfondrk@ucdavis.edu, Robert.Page@asu.edu, Gro.Amdam@asu.edu 15 16 *Author for correspondence: Kate Ihle, address: Smithsonian Tropical Research 17 Institute, Panama City, Panama Apartado Postal 0843-03092, Republic of 18 Panama, email: kateihle@gmail.com, telephone: +1 612-321-8833 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Abstract Honey bee workers display remarkable flexibility in the aging process. 34 This plasticity is closely tied to behavioral maturation. Workers who initiate 35 foraging behavior at earlier ages have shorter lifespans, and much of the 36 variation in total lifespan can be explained by differences in pre-foraging lifespan. 37 Vitellogenin (Vg), a yolk precursor protein, influences worker lifespan both as a 38 regulator of behavioral maturation and through anti-oxidant and immune 39 functions. Experimental reduction of Vg mRNA, and thus Vg protein levels, in 40 wild-type bees results in precocious foraging behavior, decreased lifespan, and 41 increased susceptibility to oxidative damage. We sought to separate the effects 42 of Vg on lifespan due to behavioral maturation from those due to immune and 43 antioxidant function using two selected strains of honey bees that differ in their 44 phenotypic responsiveness to Vg gene knockdown. Surprisingly, we found that 45 lifespans lengthen in the strain described as behaviorally and hormonally 46 insensitive to Vg reduction. We then performed targeted gene expression 47 analyses on genes hypothesized to mediate aging and lifespan: the insulin-like 48 peptides (Ilp1 and 2) and manganese superoxide dismutase (mnSOD). The two 49 honey bee Ilps are the most upstream components in the insulin-signaling 50 pathway, which influences lifespan in Drosophila melanogaster and other 51 organisms., while manganese superoxide dismutase encodes an enzyme with 52 antioxidant functions in animals. We found expression differences in the llps in fat 53 body related to behavior (llp1 and 2) and genetic background (Ilp2), but did not 54 find strain by treatment effects. Expression of mnSOD was also affected by 55 behavior and genetic background. Additionally, we observed a differential 56 response to Vg knockdown in fat body expression of mnSOD, suggesting that 57 antioxidant pathways may partially explain the strain-specific lifespan responses 58 to Vg knockdown. 59 60 61 Keywords: vitellogenin, RNA-interference, Apis mellifera, foraging, longevity 62 Abbreviations: vitellogenin (Vg), insulin/insulin-like signaling (IIS), insulin-like 63 peptide 1 (Ilp1), insulin-like peptide 2 (Ilp2), manganese superoxide dismutase 64 (mnSOD), RNA-interference (RNAi), juvenile hormone (JH), double-stranded 65 RNA (dsRNA), injected control treatment (injC), non-injected control treatment 66 (noREF), vitellogenin knockdown treatment (vgRNAi) 67 68 1. Introduction The extraordinarily plastic nature of honey bee life history and aging has 69 70 recently made this eusocial insect a model organism for the study of senescence 71 (Amdam et al., 2005; Remolina et al., 2007; Corona et al., 2007). From a single 72 genome, female honey bees can develop along three very different aging 73 trajectories triggered by environmental conditions: queens, which typically live 74 between 1-2 years (Page and Peng, 2001); spring and summer workers, which 75 live between 15-60 days (Fukuda and Sekiguchi, 1966); and diutinus workers 76 (winter bees), which emerge in the late summer and early autumn in temperate 77 regions and can live 6-8 months (Maurizio, 1950). Additionally, worker aging is 78 not entirely chronological. Instead, senescence in workers appears to also be 79 closely tied to behavioral maturation (Neukirch, 1982; Seehuus et al., 2006a; 80 Behrends et al., 2007; Rueppell et al., 2007; Baker et al., 2012). 81 Honey bee workers exhibit an age-associated division of labor in which 82 young (nurse) bees perform in-hive tasks such as brood care and food storage 83 while older bees leave the nest to forage (Winston, 1987). The age at which a 84 worker initiates foraging is strongly correlated with lifespan: the earlier a worker 85 leaves the nest to forage, the shorter her lifespan is likely to be (Neukirch, 1982; 86 Robinson et al., 1992a: Rueppell et al., 2007). This age-related behavioral 87 progression can be slowed, accelerated, or even reversed based on internal, 88 social, and environment factors (Robinson et al., 1992b; Huang et al., 1998; 89 Pankiw and Page, 2001). The dynamic nature of the aging process in honey 90 bees has enabled targeted studies on the physiological and molecular pathways 91 involved in their aging and senescence (Münch and Amdam, 2010; Amdam, 92 2011). 93 Aging and lifespan in honey bees are affected in several ways by 94 Vitellogenin (Vg), a yolk precursor protein. Vg is the most abundant protein in the 95 hemolymph of nurse bees, comprising 30-50% of total protein (Engels and 96 Fahrenhorst, 1974), and impacts both anti-oxidant and immune function (Amdam 97 et al., 2004; Seehuus et al., 2006b; Corona et al., 2007). In honey bees, Vg 98 appears to be the primary zinc carrier, and hemolymph zinc levels are closely 99 tied to fluctuations in Vg titer (Amdam et al., 2004). In foragers, a behavioral 100 stage with low Vg titers, zinc levels fall so low that apoptosis is induced in 101 hemocytes, cells that function in the innate immune response of insects (Amdam 102 et al., 2005). Additionally, Vg itself has anti-oxidant properties and is 103 preferentially carbonylated in response to oxidative damage induced via paraquot 104 injection (Seehuus et al., 2006b). When faced with an oxidative challenge, 105 caged workers with experimentally reduced Vg expression have higher mortality 106 than control workers (Seehuus et al., 2006b). 107 Vg also impacts honey bee lifespan and aging by mediating the pacing of 108 behavioral maturation in concert with juvenile hormone (JH). In insects JH is 109 typically a gonadotropin associated with high titers of Vg (Flanagan and 110 Hagedorn, 1977; Chen et al., 1979; Flatt et al., 2005). In honey bees, however, 111 Vg and JH function in a mutually repressive feedback loop (Ramamurty and 112 Engles, 1977; Pinto et al., 2000; Guidugli et al., 2005). Nurse bees have high Vg 113 titers and low JH titers, while foragers have low Vg titers and high JH levels (Rutz 114 and Lüscher, 1974; Rutz et al., 1976; Engels and Fahrenhorst, 1974; Robinson 115 et al., 1991). Vg knockdown in the fat body induces early onset of foraging 116 behavior and decreased JH titers (Guidugli et al., 2005; Nelson et al., 2007; 117 Marco Antonio et al., 2008). Likewise, treatment with JH and its analogues 118 induces early foraging and decreased Vg titers (Ramamurty and Engles, 1977; 119 Robinson, 1987; Robinson et al., 1992a). In free-flying workers, Vg knockdown in 120 the fat body decreases lifespan, likely through a combination effects on behavior, 121 immunity and oxidative stress resistance (Nelson et al., 2007). 122 Studies using two selected lines of honey bees demonstrated that the 123 measured physiological and behavioral responses to Vg knockdown are 124 genotype dependent (Amdam et al., 2007; Ihle et al., 2010). The high and low 125 pollen hoarding strains were bidirectionally selected for colony levels of pollen 126 stores, but also exhibit a variety of differences at the level of behavior, 127 physiology, and gene expression (reviewed in Page and Fondrk, 1995; Page et 128 al., 2012; Page 2013). The sensitivity of the Vg/JH feedback loop is one such 129 trait that differs between the strains (Amdam et al., 2007; Ihle et al., 2010). High 130 strain bees have higher peak titers of Vg that decline faster at the onset of 131 foraging, relative to low strain bees (Amdam et al., 2007). Experimental 132 manipulations suggest that this could be due to a stronger coupling of the Vg/JH 133 feedback relationship in the high strain: in response to Vg knockdown, high strain 134 bees exhibit an increase in JH titers while the low strain does not (Amdam et al., 135 2007). High strain workers, like wild-type, forage earlier in response to Vg 136 knockdown in the fat body, while behavioral maturation was not affected by Vg 137 knockdown in the low strain (Ihle et al., 2010). 138 We investigated how Vg influences lifespan in the genetic background of 139 the high and low strains to separate its effects on lifespan due to behavioral 140 maturation from those due to antioxidant and immunological functions. In the Vg- 141 responsive high strain, we predicted that Vg knockdown would decrease lifespan 142 consistent with results in wild-type bees. In the low strain, where Vg knockdown 143 does not appear to affect behavioral maturation and JH dynamics, we predicted 144 either no response or a decreased lifespan reflecting reduced protection from 145 oxidative and immune challenges. 146 Lifespan of high strain workers was not significantly affected by Vg 147 knockdown. Contrary to our prediction, Vg knockdown increased lifespan in the 148 low strain. We hypothesized that the observed genotype-specific differences in 149 lifespan after Vg knockdown could be traced to differentially effective 150 compensatory mechanisms between the strains. We then performed targeted 151 expression analysis on same age nurses and foragers of both strains to 152 investigate pathways associated with lifespan and oxidative damage. 153 The conserved insulin/insulin-like signaling (IIS) pathway is known to 154 influence many important biological processes including aging, reproduction, and 155 nutrition in both insects and vertebrates, (reviewed in Wu and Brown, 2006). The 156 functions of the IIS pathway in honey bees are less well understood, but it has 157 been shown to be associated with behavior (Ament et al., 2008; Wang et al., 158 2010), nutritional status (Ament et al., 2011; Nilsen et al., 2011; Ihle et al., 2014), 159 and lifespan (Corona et al., 2007). Interestingly, the IIS pathway appears to be a 160 central regulator of the divergent phenotypes of the high and low strains (Page et 161 162 al., 2012). The most upstream components of the IIS pathway in honey bees, insulin- 163 like peptides (Ilp1 and 2), in particular have been liked to Vg and JH levels 164 (Corona et al., 2007; Nilsen et al., 2011). Peripheral Ilp1 expression is positively 165 correlated with Vg expression (Nilsen et al., 2011), while expression of Ilp1 in the 166 head is increased by JH-analogue treatment (Corona et al., 2007). Ilp1 167 expression increases in response to sugar and amino acid supplementation 168 (Ament et al., 2011; Nilsen et al., 2011; Ihle et al., 2014) and has been implicated 169 in the extended lifespan phenotype of queens (Corona et al., 2007). In contrast, 170 Ilp2 expression is correlated with JH titer (Nilsen et al., 2011), and does not 171 respond to nutrient manipulation (Ament et al., 2011; Nilsen et al., 2011; Ihle et 172 al., 2014). Expression of both peptides appears to be sensitive to behavioral 173 maturation or age (Corona et al., 2007; Ament et al., 2011). Thus, we 174 hypothesized that Ilp1 and Ilp2 could mediate the divergent lifespan responses to 175 Vg knockdown observed in the high and low strains. 176 As Vg may affect lifespan by mitigating oxidative damage, we further 177 reasoned that low strain workers may have compensatory mechanisms that 178 shield them from some of the potential deleterious effects of naturally low Vg 179 expression, which may be particularly sensitive to Vg knockdown. We targeted 180 manganese superoxide dismutase (mnSOD), because as a mitochondrial 181 antioxidant important during aerobic respiration, its activity may be especially key 182 during the aerobically challenging flight of honey bee foragers (Fridovich, 1995). 183 Expression dynamics of mnSOD differ between queens and workers and can be 184 associated with aging (Corona et al., 2005). 185 Bidirectional selection has demonstrated that some honey bees can 186 respond to reduced Vg levels with increased lifespan, perhaps via alternative 187 mechanisms of self-maintenance. This low strain phenotype is associated with 188 severely reduced Vg sensitivity in behavioral and hormonal regulation, and may 189 be uncommon in wild type that relies on intact Vg functions. 190 191 192 2. Materials and Methods 193 194 2.1. Bees 195 Worker progeny were derived from queens in the 33rd generation of selection. 196 High pollen strain colonies stored nearly 7 times the amount of low pollen strain 197 colonies. Mean area of pollen for highs (n=16) was 173.2 square inches and the 198 mean for low colonies (n=17) was 24.8 square inches (Mann-Whitney: U=2, 199 Z=4.74, p>0.0001). Queens from two source high and low strain colonies were 200 caged over night to allow easy collection of same-aged bees for both the lifespan 201 and gene expression trails. After 20 days, frames were removed from the 202 colonies and worker bees were emerged in an incubator set at 34°C with 203 approximately 70% relative humidity. Newly emerged bees were randomly 204 assigned to one of three groups: vgRNAi, the experimental dsRNA injected 205 group, the double control groups of noREF, a non-handled reference group; and 206 injC, a control group injected with vehicle, according to established protocols, 207 e.g. (Guidugli et al., 2005; Nelson et al., 2007). Bees to be used in the longevity 208 experiment (below) were individually tagged (BeeWorks, Orillia, Ontario, 209 Canada) while bees used in the gene expression study were marked with paint 210 (Testors Enamel; Testor Corporation, Rockford, Illinois, United States) to indicate 211 treatment group. 212 213 2.2. dsRNA preparation and injection 214 Double stranded RNA (dsRNA) toward Vg was prepared according to the RNA 215 interference (RNAi) protocol of Amdam et al. (Amdam et al., 2003; Amdam et al., 216 2006). The cDNA clone AP4a5 was used as template (GenBank accession #: 217 AJ517411), and primers were fused to T7 promoter sequence (underlined): Fw: 218 5’-TAATACGACTCACTATAGGGCGAACGACTCGACCAACGACTT-3’, Re: 5’- 219 TAATACGACTCACTATAGGGCGAAACGAAAGGAACGGTCAATTCC-3’. PCR 220 product was purified using the QIAquick PCR purification kit (Qiagen, Valencia, 221 California, United States), and RNA was prepared with the Promega RiboMax T7 222 system (Promega, Madison, Wisconsin, United States). RNA was extracted by 223 TRIzol LS reagent (GIBCO-BRL, San Diego, California, United States), 224 resuspended in nuclease-free water, heated at 96°C for 2 min, and left to cool at 225 room temperature for 20 min. dsRNA products were brought to a final 226 concentration of 5 !g/!l in nuclease-free water (Qiagen) (Amdam et al., 2003). 227 Before injection bees were cold anaesthetized, and secured to wax covered 228 plates by crossing pins between the thorax and abdomen. Injections were 229 performed between the fifth and sixth tergites using Hamilton syringes with G30 230 disposable needles (BD, Palo Alto, California, United States). Injection volume 231 was 2 !l for both vgRNAi and injC groups. 232 233 2.3. Lifespan 234 Injections took place over three days for each colony. Treated bees (n = 200 235 bees per treatment, per genotype, per colony) were individually tagged, and 236 placed into each of two, three-framed, glass-walled observation colonies with a 237 wild-type background population. Colonies were surveyed in the evening after all 238 foragers had returned for the night. Each frame was scanned twice per side, and 239 all ID tags were recorded to determine surviving individuals. Age of death was 240 considered to be the day after the last sighting of an individual (Nelson et al., 241 2007). We determined which bees had initiated foraging during our study window 242 243 by monitoring glass-topped runways into the colonies during peak foraging 244 windows. The ID tags of all returning foragers were recorded. Individuals 245 observed returning from foraging trips more than once were considered 246 confirmed foragers in our analyses. We determined total lifespan for the 247 experimental population as a whole as well as for confirmed foragers. 248 Additionally, we divided total lifespan into pre- and post-foraging initiation 249 components. Pre-foraging lifespan was determined by age of foraging initiation or 250 age of death if death occurred before the onset of foraging behaviour. Post- 251 foraging lifespan was considered to be the span between foraging initiation and 252 death. 253 254 2.4. Gene expression 255 Injections took place on a single day for each colony. Treated bees (n = 100 bees 256 per treatment, per genotype, per colony) were paint marked to indicate treatment 257 group (VgRNAi or injC) and introduced into one of two five-frame nuclear 258 colonies with a wild-type background population. Treated bees were allowed to 259 mature inside the host colonies for 10 days. 10-day old bees were collected from 260 the colonies at 8:00am to avoid the circadian fluctuations of antioxidant gene 261 expression observed previously (Williams et al., 2008; Elekonich, 2009) and to 262 enable collection of both nurses and foragers. Bees identified as foragers were 263 collected as they alighted on the colony entrance after returning from a flight. 264 Nurses were identified as bees that had been observed to put their heads into 265 cells containing young brood. Fat body and brain were dissected as described 266 before (Nelson et al., 2007; Nunes et al., 2013). Briefly, fat body tissue was 267 harvested by removing the gut tract, leaving fat body tissue clinging to the 268 abdominal carcase. 269 RNA was extracted from fat body tissue using a combined Trizol 270 (Invitrogen) and RNeasy Kit (Qiagen) method according (Amdam et al., 2003). 271 Brain RNA was obtained through phenol:chloroform extraction. We measured 272 RNA quality and concentration using a NanoDrop ND-1000 (NanoDrop 273 Technologies, Wilmington, DE, USA) and diluted all samples to 25 µg/µL. 274 Relative gene expression levels were determined by one-step reverse 275 transcription- polymerase chain reaction (RT-qPCR) using QuantiTect SYBR 276 Green RT-PCR Master Mix kit (Qiagen) and ABI Prism 7500 (Applied 277 Biosystems, Foster City, CA, USA). All samples were run in triplicate. Negative 278 controls, samples run without the addition of the RT enzyme, confirmed the 279 absence of genomic DNA contamination. Knockdown of Vg in fat body was 280 verified relative to injected controls as previous work has demonstrated that 281 injected controls do not differ from the non-handled reference group (Amdam et 282 al., 2007; Nelson et al., 2007). Transcript levels of Vg (accession # AJ517411), 283 Ilp1 (GB17332-PA), Ilp2 (accession # GB10174_PA) and mnSOD (accession # 284 AY329356) were quantified relative to !-actin (accession # AB023025) 285 expression using real-time RT PCR as before e.g. (Amdam et al., 2004; 286 Lourenco et al., 2008). "-actin is stably expressed in several tested honey bee 287 tissues and has been demonstrated to be an effective control gene when 288 measuring gene expression in adult honey bee fat body (Lourenco et al., 2008; 289 Scharlaken et al., 2008). As such, "-actin is a commonly used reference in 290 studies of honey bee gene expression (Chen et al., 2005). Primers are listed in 291 Supplementary Table 1. Control reactions without reverse transcriptase were 292 preformed for each sample to ensure reactions were not contaminated with 293 genomic DNA. 294 295 2.5. Statistical analysis 296 Lifespan data were analyzed using Kaplan-Meier survival analysis (Amdam et al. 297 2007). Planned pair-wise comparisons were made between the vgRNAi and injC 298 groups as well as between the injC and noREF groups with the Cox-Mantel test 299 (Nelson et al., 2007; Ihle et al., 2010). Gene expression data were log 300 transformed to approximate a normal distribution, and analyzed by factorial 301 ANOVA (Rieu and Powers, 2009; Wang et al., 2010). Post-hoc analysis was 302 performed with Fisher’s LSD test. All analyses were conducted using Statistica 303 6.0 (StatSoft, Inc. Tulsa, Oklahoma, United States). 304 305 3. Results 306 307 3.1. Knockdown verification 308 We first confirmed Vg knockdown in fat body (F (1, 79) =67.20, p < 0.0001). Vg 309 expression was significantly reduced by Vg dsRNA injection across strains and 310 task group (LSD: high strain nurses p < 0.0001, high strain foragers p < 0.0001, 311 low strain nurses p = 0.0020, low strain foragers p=0.0008; df = 80; Fig. 3A-D). 312 313 3.2. Lifespan effects of Vg knockdown 314 We tested the effects of Vg knockdown on the lifespan of all experimental bees 315 as well as on the lifespan of confirmed foragers. When we considered the entire 316 experimental population, we found that in contrast to results from wild type 317 (unselected) bees (Nelson et al., 2007), Vg knockdown had no effect on the 318 lifespan of the high strain bees (Kaplan-Meier !2 = 3.517, df = 2, p = 0.1723; n= 319 noREF:189, injC:227, and VgRNAi:211; Fig. 1A). There was a significant effect of 320 treatment in the low strain (Kaplan-Meier !2 = 30.576, df = 2, p < 0.0001; n= 321 noREF:233, injC:210, and VgRNAi:212; Fig. 1B). Intriguingly, Vg knockdown in 322 the low strain increased lifespan relative to injected controls (Cox-Mantel U = 323 25.4983, p = 0.0067). 324 Within the confirmed forager population, there was an overall effect of 325 treatment in the high strain (Kaplan-Meier !2 = 7.2342, df = 2, p = 0.0268; n = 326 noREF: 113, injC:173, and VgRNAi:169). The differences between treatment 327 groups to be compared were not significant (Cox-Mantel: inC vs. VgRNAi U = 328 11.066, p = 0.173; injC vs noREF U = -7.259, p = 0.310), but we observed a non- 329 significant pattern mirroring that seen in wild-type bees with vgRNAi bees dying 330 earlier than controls (Nelson et al., 2007). There was also an effect of treatment 331 in low strain confirmed foragers (Kaplan-Meier !2 = 17.7264, df = 2, p < 0.0001; 332 n= noREF:128, injC:91, and VgRNAi:107). Vg knockdown resulted in increased 333 lifespan in confirmed low strain foragers relative to the control (Cox-Mantel: injC 334 vs. VgRNAi U = -16.92, p = . 0.0052). Additionally, there was a significant 335 negative effect of handling on lifespan (Cox-Mantel: injC vs noREF U = -23.02, p 336 < 0.0001). 337 While Vg knockdown did not affect total lifespan in the high strain, it did 338 affect both the pre-foraging (Kaplan-Meier !2 = 13.898, df = 2, p < 0.001) and 339 post-foraging initiation (Kaplan-Meier !2 = 7.323, df = 2, p = 0.0270) lifespans, 340 albeit in different directions. In the high strain, Vg knockdowns had significantly 341 shorter pre-foraging lifespans than did injected controls (Cox-Mantel: injC vs. 342 vgRNAi U = -28.596, p < 0.01; Fig 2A) consistent with their early foraging 343 initiation (Ihle et al., 2010). Likewise, high strain vgRNAi bees had significantly 344 longer post-foraging onset lifespans than the injC group (Cox-Mantel: inC vs. 345 VgRNAi U = 13.899, p = 0.0482; Fig 2B), displaying an established negative 346 correlation between pre- and post-foraging initiation components of lifespan 347 (Neukirch, 1982; Rueppell et al., 2007). In the low strain, there was no effect of 348 treatment on either pre-foraging lifespan (Kaplan-Meier !2 = 22.878, df = 2, p < 349 0.0001; Fig. 2C) or post-foraging initiation lifespan (Kaplan-Meier !2 = 0.424, df = 350 2, p = 0.8090; Fig 2D), suggesting that the increase in overall lifespan is not an 351 artifact of the age of foraging, a trait unaffected by Vg reduction in the low strain. 352 353 3.2. Gene expression 354 Vg expression in fat body was significantly affected by strain (F(1, 79) =11.99, p = 355 0.0009), treatment (F(1, 79) =67.20, p < 0.0001) and task (F(1, 79) = 53.39, p < 356 0.0001), but not colony (F(1, 79) = 2.08, p > 0.152). Fat body Vg expression was 357 not different between high and low strain injected control (LSD: p = 0.8670, df = 358 78) or injected control foragers (LSD: p = 0.0886, df = 78). A significant strain x 359 treatment effect (F(1, 79) = 5.85, p = 0.0178) indicates that the knockdown was 360 stronger in high strain bees. Vg expression in brain was upregulated in response 361 to knockdown in abdominal fat body (Fig 3E-H). Expression of Vg in brain was 362 affected by treatment (F(1, 73) = 16.2737, p < 0.0001) and task (F(1, 73) = 8.1764, p 363 = 0.0055), but not by strain (F(1, 73) = 1.3415, p = 0.25) or colony (F(1, 73) = 3.5260, 364 p = 0.0643). Expression of Vg in brain is higher in foragers than it is in nurses 365 (LSD: p = 0.0006, df = 74), and higher in the vgRNAi group than in injected 366 controls (LSD: p < 0.0001, df = 74). 367 Ilp1 expression in the abdominal fat body was influenced by task (Factorial 368 ANOVA: F(1, 79) = 28.93, p < 0.0001), but not by strain (F(1, 79) = 0.126, p = 369 0.7240), treatment (F(1, 79) = 3.03, P > 0.0854), or colony (F(1, 79) = 1.75, P > 370 0.189). When the strains were considered separately, the effect of task was 371 independently significant with nurses having higher expression than foragers in 372 both high and low strains (LSD: high strain p < 0.0001; low strain p < 0.0013). 373 There was no influence of strain (F(1, 69) = 0.62, p = 0.4354), treatment (F(1, 69) = 374 1.28, p = 0.2620), task (F(1, 69) = 0.63, p = 0.4290), or colony (F(1, 69) = 0.49, p = 375 0.4840) on Ilp1 expression in brain. 376 377 As reported before, abdominal fat body expression of Ilp2 was not affected by Vg knockdown (Factorial ANOVA: F(1, 67)=0.22, p = 0.7730) (Nilsen et al. 378 2011). However, strain (F(1, 67)= 18.75, p < 0.0001) and task (F(1, 67)=11.82, p < 379 0.001) did influence Ilp2 expression. The low strain has higher Ilp2 expression 380 than the high strain, (LSD: p < 0.0001, df = 68) and nurses have higher 381 expression than foragers (p < 0.0009, df = 68). Brain Ilp2 expression was not 382 influenced by strain (Factorial ANOVA: F(1, 78) = 0.076, p = 0.783), treatment (F(1, 383 78) 384 = 0.86, p = 0.358), or task (F(1, 78) = 2.47, p = 0.120). mnSOD expression in abdominal fat body was affected by strain (Factorial 385 ANOVA: F(1, 79) = 6.81, p = 0.011) and task (F(1, 79) = 16.85, p < 0.0001), but not 386 by colony (F(1, 79)=1.23, p = 0.271) or treatment (F(1, 79) = 0.07 , p = 0.7832). Low 387 strain bees had higher mnSOD expression than high strain bees (LSD: p < 388 0.0034, df = 80), and nurses had higher expression than foragers (p < 0.0001, df 389 = 80). Additionally, there was a significant interaction effect between strain and 390 treatment (F(1, 79) = 7.77, p = 0.0066), revealing that the strains responded 391 differently to Vg knockdown (Fig 6A-D). While the effects of treatment were not 392 independently significant for either strain (LSD: high strain p = 0.1150, df = 80; 393 low strain p = 0.1470, df = 80), there was a non-significant trend in the high strain 394 for Vg knockdown to decrease mnSOD expression in abdominal fat body and a 395 non-significant increase in the low strain. In brain tissue, mnSOD expression is 396 influenced by task (Factorial ANOVA: F(1, 78) = 8.29, p = 0.0051), but not strain 397 (F(1, 78) = 1.48, p = 0.226) or treatment (F(1, 78) = 1.20, p = 0.276). In contrast to 398 expression patterns in abdomen, foragers had higher expression mnSOD 399 expression in head tissue than did nurses. However, when these results are 400 broken down by strain, the effect was only significant in the low strain (LSD: low 401 strain p = 0.0122; high strain p = 0.607, df = 80; Fig 6E-H). 402 403 4. Discussion 404 Using two divergently selected strains of honey bees that differ in the strength of 405 their Vg/JH feedback relationships, we were able to uncover lifespan responses 406 to Vg knockdown not observed in a previous study on wild-type bees (Nelson et 407 al., 2007). In the total population of treated bees, lifespan was not significantly 408 different between high strain Vg knockdowns and controls. Intriguingly, in the low 409 strain, in which Vg knockdown does not affect behavioral maturation, Vg 410 knockdown resulted in increased lifespan relative to injected controls. We 411 predicted that the lifespan effects of Vg knockdown would be more pronounced 412 in the sub-population of confirmed foragers. Foraging honey bees have among 413 the highest metabolic rates measured (Harrison et al., 1996). Therefore, foraging 414 represents a major metabolic and oxidative challenge for worker bees. While 415 foragers have lower hemolymph titers of Vg than nurses, RNAi against Vg further 416 reduces expression levels in the fat bodies of both high and low strains, 417 potentially exposing Vg knockdowns to increased oxidative damage (Seehuus et 418 al. 2006). There was no significant effect of Vg knockdown in confirmed high 419 strain foragers, but they did display a non-significant trend toward decreased 420 lifespan in response to Vg knockdown, similar to results from wild-type 421 (unselected) honey bees (Nelson et al., 2007). In the low strain, we again 422 observed an increase in total lifespan in the vgRNAi group. 423 We initially hypothesized that the Vg knockdown-induced increase in 424 lifespan observed in low strain workers could be the result of delayed foraging 425 onset, as age of foraging initiation has been shown to be strongly correlated with 426 lifespan (Neukirch, 1982; Robinson et al., 1992a; Rueppell et al., 2007). 427 However, in the low strain we found no effect of Vg knockdown on either pre- 428 foraging or post-foraging initiation lifespan. This suggests that the overall lifespan 429 increases observed in low strain Vg knockdowns occur throughout both general 430 life stages, but were not statistically significant within either component of total 431 lifespan. In the high strain, Vg knockdowns initiated foraging earlier in life (Ihle et 432 al., 2010) and had longer foraging lifespans than did controls, consistent with 433 previous work demonstrating a strong negative relationship between pre-foraging 434 and post-foraging onset lifespans (Rueppell et al., 2007). 435 As the increased lifespan in the low strain Vg knockdown group could not 436 be explained by behavioral ontogeny, we hypothesized that it may instead be 437 due to differential Vg response to knockdown or compensatory signaling from 438 pathways that intersect Vg action and which are associated with aging and 439 senescence. While RNAi-induced Vg knockdown in the fat body is effective in 440 both strains, the high strain experiences a greater reduction in Vg expression in 441 the fat body than the low strain. This reflects the Vg dynamics in untreated low 442 strain bees that naturally have slower declines in Vg titer than do high strain bees 443 (Amdam et al., 2007). The slower Vg decline observed in the low strain is likely 444 due to a weak Vg/JH relationship in this strain. We did not find the previously 445 observed strain differences in Vg expression in the fat body (Amdam et al., 446 2007). However, at 10 days of age, the bees in this study were likely past peak 447 expression, after which expression of Vg decreases faster in the high strain. 448 In brain, we found that Vg expression is upregulated in response to both 449 natural (injC foragers relative to nurses) and experimental (vgRNAi nurses and 450 foragers relative to injC nurses and foragers respectively) declines in fat body- 451 produced Vg. Our results are in contrast to those of a recent study using wild- 452 type bees which found that brain Vg expression decreased in response to 453 injection of dsRNA against Vg into the hemolymph (Nunes et al. 2013). However, 454 these bees were collected at 15 days of age while our sample came from 10 day 455 old bees. It is possible that we observed a short-lived phenotype present only 456 immediately after foraging initiation or that, despite care, different dissectors 457 produced samples containing different tissue types. The mechanism by which Vg 458 affects behavior is not yet clear, nor are the relative contributions of fat body and 459 brain expression levels. Further study is needed to determine which cell types 460 can express Vg in honey bees, and how fat body and brain expression of Vg are 461 related. 462 We found no evidence that Ilp1 or Ilp2 expression mediates the strain- 463 specific, Vg knockdown induced differences in lifespan. There were no strain or 464 strain-by-treatment effects on Ilp1 expression. In same-age nurses and foragers, 465 we found that nurse bees have higher fat body Ilp1 expression than do foragers. 466 In caged bees, fat body Ilp1 expression increases in response to available amino 467 acids and carbohydrates demonstrating a nutrient signaling function for Ilp1 468 independent of behavioral phenotype (Nilsen et al., 2011; Ihle et al., 2014) . Our 469 results support these findings and extend them to free-flying bees as nurses 470 have high nutritional stores relative to foragers (Crailsheim, 1986; Toth and 471 Robinson, 2005). We did not find previously reported differences in brain Ilp1 472 expression correlated with task (Ament et al., 2008). However, as chronological 473 age was not controlled in that study (Ament et al., 2008), the forager group was 474 likely older than the nurse group. Thus, the increased brain expression of Ilp1 475 observed in foragers may be due to a positive correlation between Ilp1 476 expression and age (Corona et al., 2007). 477 We found that fat body Ilp2 expression is higher in nurses than in foragers, 478 a result consistent with the hypothesis that Ilp2 is a broad indicator of nutrient 479 availability (Nilsen et al., 2011). Additionally, we found that fat body Ilp2 480 expression is higher in low strain than in high strain workers. We have previously 481 hypothesized that Ilp2 may be an antagonist of the honey bee insulin receptors 482 and may suppress the metabolic changes that accompany the transition to 483 foraging, but this has yet to be tested experimentally (Nilsen et al., 2011). Here, 484 high expression of Ilp2 in the fat body would act remotely to suppress 485 transduction of the IIS pathway in brain tissue and so inhibit synthesis of JH, a 486 downstream target. It is possible that the high expression of Ilp2 in low strain 487 bees plays a role in the reduced sensitivity of their Vg/JH relationship, and may 488 slow the transition to forager physiology. While these findings do not support a 489 role of Ilp2 expression in the strain-specific lifespan response to Vg reduction, 490 they do suggest that the low strain bees may be a valuable tool in determining 491 the mechanisms that underlie not only the behavioral transition from nursing to 492 foraging behavior but also the metabolic changes that accompany it. 493 Our data indicate that one potential mechanism underlying the longer 494 lifespans in the low strain following Vg knockdown is an up regulation of genes 495 associated with defense against oxidative damage. Low strain workers have 496 higher fat body expression of mnSOD, which encodes an enzyme active in the 497 degradation of superoxide radicals in mitochondria (Fridovich, 1995). Expression 498 of mnSOD between the strains is differentially affected by Vg knockdown: 499 expression is decreased slightly in the high strain and slightly increased in the 500 low strain following Vg suppression. In low strain workers, brain expression of 501 mnSOD is higher in foragers than in nurses. The generally higher mnSOD 502 expression in low strain bees may mean that they invest more heavily in 503 alternative pathways to combat oxidative damage when Vg expression is 504 generally low. As such, in response to Vg knockdown, they may be exposed to 505 less oxidative damage than is the high strain. Perhaps in low strain workers, 506 expression patterns of mnSOD, and potentially other genes that mitigate 507 oxidative damage, combined with increased local expression of Vg in the brain 508 buffer the effects of senescence generally observed in the brains of foragers 509 enough to significantly lengthen lifespan in response to Vg knockdown. However, 510 while oxidative damage has been shown to correlate with aging in many 511 organisms (Bokov et al., 2004), more recent work has called into question a 512 causal role for oxidative stress in aging (Perez et al., 2009; Salmon et al., 2010). 513 Future work is needed to quantify how Vg titer impacts oxidative damage in the 514 brains and peripheral tissues of free-flying honey bees, and whether such 515 damage translates to functional declines. 516 In this study, we found evidence to suggest that bidirectional selection on 517 behavior has altered systems of self-maintenance: In the relative absence of Vg, 518 alternative mechanisms extend worker life in the low but not the high pollen 519 hoarding strain. This response occurred in the physiological context of severely 520 reduced Vg sensitivity and generally low Vg levels that characterize low strain 521 workers throughout life (Amdam et al. 2004a, 2007). This phenotype is likely 522 uncommon in wild type bees, which rely on intact Vg functions in regulation of 523 health and behavior. However, the low strain developed in an artificial selective 524 context that favored Vg deficiency because it reduced pollen hoarding. Vg 525 deficiency has negative consequences for bee health (Amdam et al. 2004b; 526 Seehuus et al. 2006b), but for each generation only the healthiest colonies were 527 used in breeding of the high and low strains. The low strain colonies that were 528 used in breeding may, therefore, have represented genotypes that were able to 529 recruit alternative self-maintenance systems to replace Vg. 530 531 We were unable to demonstrate a clear molecular explanation for the extended lifespans in the low strain after Vg knockdown. However, we feel this 532 paper provide strong data to support the use of bidirectional selection in the 533 study of honey bee health and aging. 534 535 536 5. Acknowledgements 537 We thank Adam Dolezal, Ying Wang, and Nicholas Baker for assistance with the 538 experimental setups and M. Teague O’Mara for helpful comments on the 539 manuscript. KEI was supported by The Research Council of Norway 540 (216776/F11) and a postdoctoral fellowship from Arizona State University and 541 the Smithsonian Tropical Research Institute. 542 6. References 543 Amdam, G. V. (2011). Social context, stress, and plasticity of aging. Aging Cell 544 10, 18-27. 545 Amdam, G. V. and Page Jr, R. E. (2010). The developmental genetics and 546 physiology of honeybee societies. Animal Behaviour 79, 973-980. 547 Amdam, G. V., Aase, A. L., Seehuus, S. C., Kim Fondrk, M., Norberg, K. and 548 Hartfelder, K. (2005). Social reversal of immunosenescence in honey bee 549 workers. Experimental Gerontology 40, 939-947. 550 Amdam, G. V., Nilsen, K. A., Norberg, K., Fondrk, M. K. and Hartfelder, K. 551 (2007). Variation in endocrine signaling underlies variation in social life history. 552 American Naturalist 170, 37-46. 553 Amdam, G. V., Norberg, K., Page, R. E. J., Erber, J. and Scheiner, R. (2006). 554 Downregulation of vitellogenin gene activity increases the gustatory 555 responsiveness of honey bee workers (Apis mellifera). Behavior Brain Research 556 169, 201-205. 557 Amdam, G. V., Simoes, Z. L., Guidugli, K. R., Norberg, K. and Omholt, S. W. 558 (2003). Disruption of vitellogenin gene function in adult honeybees by intra- 559 abdominal injection of double-stranded RNA. BMC Biotechnology 3, 1. 560 Amdam, G. V., Simoes, Z. L., Hagen, A., Norberg, K., Schroder, K., 561 Mikkelsen, O., Kirkwood, T. B. and Omholt, S. W. (2004). Hormonal control of 562 the yolk precursor vitellogenin regulates immune function and longevity in 563 honeybees. Experimental Gerontology 39, 767-773. 564 Ament, S. A., Corona, M., Pollock, H. S. and Robinson, G. E. (2008). Insulin 565 signaling is involved in the regulation of worker division of labor in honey bee 566 colonies. Proceedings of the National Academy of the United States of America 567 105, 4226-4231. 568 Baker, N., Wolschin, F. and Amdam, G. V. (2012). Age-related learning deficits 569 can be reversible in honeybees Apis mellifera. Experimental Gerontology 570 Behrends, A., Scheiner, R., Baker, N. and Amdam, G. V. (2007). Cognitive 571 aging is linked to social role in honey bees (Apis mellifera). Experimental 572 Gerontology 42, 1146-1153. 573 Bokov, A., Chaudhuri, A. and Richardson, A. (2004). The role of oxidative 574 damage and stress in aging. Mech Ageing Dev 125, 811-826. 575 Chen, T. T., Couble, P., Abu-Hakima, R. and Wyatt, G. R. (1979). Juvenile 576 hormone-controlled vitellogenin synthesis in Locusta migratoria fat body. 577 Hormonal induction in vivo. Developmental Biology 69, 59-72. 578 Chen, Y. P., Higgins, J. A. and Feldlaufer, M. F. (2005). Quantitative real-time 579 reverse transcription-PCR analysis of deformed wing virus infection in the 580 honeybee (Apis mellifera L.). Applied and Environmental Microbiology 71, 436- 581 441. 582 Corona, M., Velarde, R. A., Remolina, S., Moran-Lauter, A., Wang, Y., 583 Hughes, K. A. and Robinson, G. E. (2007). Vitellogenin, juvenile hormone, 584 insulin signaling, and queen honey bee longevity. Proceedings of the National 585 Academy of Sciences of the United States of America 104, 7128-7133. 586 Crailsheim, K. (1986). Dependence of protein metabolism on age and season in 587 the honeybee (< i> Apis mellifica carnica Pollm). Journal of Insect Physiology 32, 588 629-634. 589 Elekonich, M., M. (2009). Extreme thermotolerance and behavioral induction of 590 70-kDa heat shock proteins and their encoding genes in honey bees. Cell Stress 591 & Chaperones 14, 219-226. 592 Engels, W. and Fahrenhorst, H. (1974). Alters-und kastenspezifische 593 Veränderungen der Haemolymph-Protein-Spektren beiApis mellificia. 594 Development Genes and Evolution 174, 285-296. 595 Flanagan, T. R. and Hagedorn, H. H. (1977). Vitellogenin synthesis in the 596 mosquito: the role of juvenile hormone in the development of responsiveness to 597 ecdysone. Physiological Entomology 2, 173-178. 598 Flatt, T., Tu, M. P. and Tatar, M. (2005). Hormonal pleiotropy and the juvenile 599 hormone regulation of Drosophila development and life history. Bioessays 27, 600 999-1010. 601 Fridovich, I. (1995). Superoxide radical and superoxide dismutases. Annual 602 Review of Biochemistry 64, 97-112. 603 Fukuda, H. and Sekiguchi, K. (1966). Seasonal change of the honeybee worker 604 longevity in Sapporo, North Japan, with notes on some factors affecting the life- 605 span. Japanese Journal of Ecology 16, 206-212. 606 Guidugli, K. R., Nascimento, A. M., Amdam, G. V., Barchuk, A. R., Omholt, 607 S., Simoes, Z. L. and Hartfelder, K. (2005). Vitellogenin regulates hormonal 608 dynamics in the worker caste of a eusocial insect. FEBS Letters 579, 4961-4965. 609 Harrison, J. F., Fewell, J. H., Roberts, S. P. and Hall, H. G. (1996). 610 Achievement of thermal stability by varying metabolic heat production in flying 611 honeybees. Science 274, 88-90. 612 Huang, Z. Y., Plettner, E. and Robinson, G. E. (1998). Effects of social 613 environment and worker mandibular glands on endocrine-mediated behavioral 614 development in honey bees. Journal of Comparative Physiology [A] 183, 143- 615 152. 616 Ihle, K., E., Page, R., E., Jr., Frederick, K., Fondrk, M. K. and Amdam, G., V. 617 (2010). Genotype effect on regulation of behaviour by vitellogenin supports 618 reproductive origin of honeybee foraging bias. Animal Behaviour 79, 1001-1006. 619 Ihle, K. E., Baker, N. A. and Amdam, G. V. (2014). Insulin-like peptide response 620 to nutritional input in honey bee workers. Journal of Insect Physiology 69, 49-55. 621 Lourenco, A. P., Mackert, A., Cristino, A. D. and Simoes, Z. L. P. (2008). 622 Validation of reference genes for gene expression studies in the honey bee, Apis 623 mellifera, by quantitative real-time RT-PCR. Apidologie 39, 372-U33. 624 Marco Antonio, D. S., Guidugli-Lazzarini, K. R., do Nascimento, A. M., 625 Simoes, Z. L. and Hartfelder, K. (2008). RNAi-mediated silencing of vitellogenin 626 gene function turns honeybee (Apis mellifera) workers into extremely precocious 627 foragers. Naturwissenschaften 95, 953-961. 628 Maurizio, A. (1950). The influence of pollen feeding and brood rearing on the 629 length of life and physiological condition of the honeybee. Bee World 31, 9-12. 630 Münch, D. and Amdam, G. V. (2010). The curious case of aging plasticity in 631 honey bees. FEBS letters 584, 2496-2503. 632 Mutti, N. S., Dolezal, A. G. and Wolschin., F. (2011). IRS and TOR nutrient- 633 signaling pathways act via juvenile hormone to influence honey bee caste fate. 634 The Journal of Experimental Biology 214, 3977-3984 635 Nelson, C. M., Ihle, K. E., Fondrk, M. K., Page, R. E. and Amdam, G. V. 636 (2007). The gene vitellogenin has multiple coordinating effects on social 637 organization. PLoS Biology 5, e62. 638 Neukirch, A. (1982). Dependence of the life span of the honeybee (Apis 639 mellifica) upon flight performance and energy consumption. Journal of 640 Comparative Physiology B: Biochemical, Systemic, and Environmental 641 Physiology 146, 35-40. 642 Nilsen, K. A., Ihle, K. E., Frederick, K., Fondrk, M. K., Smedal, B., Hartfelder, 643 K. and Amdam, G. V. (2011). Insulin-like peptide genes in honey bee fat body 644 respond differently to manipulation of social behavioral physiology. The Journal 645 of Experimental Biology 214, 1488. 646 Nunes, F. M. F., Ihle, K. E., Mutti, N. S., Simões, Z. L. P. and Amdam, G. V. 647 (2013). The gene vitellogenin affects microRNA regulation in honey bee (Apis 648 mellifera) fat body and brain. The Journal of Experimental Biology 216, 3724- 649 3732. 650 Page, R. E. (2013). The Spirit of the Hive. Harvard University Press. Cambridge, 651 MA. 652 Page, R. E. and Peng, C. Y. S. (2001). Aging and development in social insects 653 with emphasis on the honey bee, Apis mellifera L. Experimental Gerontology 36, 654 695-711. 655 Page, R. E., Fondrk, M. K. and Rueppell, O. (2012). Complex pleiotropy 656 characterizes the pollen hoarding syndrome in honey bees (Apis mellifera L.). 657 Behavioral Ecology and Sociobiology 1-8. 658 Page, R. E. and Fondrk, M. K. (1995). The Effects of Colony Level Selection of 659 the Social-Organization of Honey-Bee (Apis melifera L) Colonies - Colony Level 660 Components of Pollen Hoarding. Behavioral Ecology and Sociobiology 36, 135- 661 144. 662 Pankiw, T. and Page, R. E. (2001). Genotype and colony environment affect 663 honeybee (Apis mellifera L.) development and foraging behavior. Behavioral 664 Ecology and Sociobiology 51, 87-94. 665 Pérez, V. I., Bokov, A., Remmen, H. V., Mele, J., Ran, Q., Ikeno, Y. and 666 Richardson, A. (2009). Is the oxidative stress theory of aging dead? Biochim 667 Biophys Acta 1790, 1005-1014. 668 Pinto, L. Z., Bitondi, M. M. and Simoes, Z. L. (2000). Inhibition of vitellogenin 669 synthesis in Apis mellifera workers by a juvenile hormone analogue, 670 pyriproxyfen. Journal of Insect Physiology 46, 153-160. 671 Ramamurty, P. S. and Engles, W. (1977). Allatectomy and Juvenile-Hormone 672 Effects on Synthesis and Incorporation of Vitellogenin in Honeybee-Queen (Apis- 673 mellifica). Zool Jahrb Allg Zool 81, 165-176. 674 Remolina, S. C., Hafez, D. M., Robinson, G. E. and Hughes, K. A. (2007). 675 Senescence in the worker honey bee Apis Mellifera. Journal of Insect Physiology 676 53, 1027-1033. 677 Rieu, I. and Powers, S., J. (2009). Real-Time Quantitative RT-PCR: Design, 678 Calculations, and Statistics. Plant Cell 21, 1031-1033. 679 Robinson, G. E. (1987). Regulation of honey-bee age polyethism by juvenile- 680 hormone. Behavioral Ecology and Sociobiology 20, 329-338. 681 Robinson, G. E., Page, R. E., Strambi, C. and Strambi, A. (1992a). Colony 682 integration in honey-bees- Mechanisms of behavioral reversion. Ethology 90, 683 336-348. 684 Robinson, G. E., Strambi, C., Strambi, A. and Feldlaufer, M. F. (1991). 685 Comparison of juvenile-hormone and ecdysteroid hemolhmph titers in adult 686 worker and queen honey-bees (Apis-mellifera). Journal of Insect Physiology 37, 687 929-935. 688 Robinson, G. E., Strambi, C., Strambi, A. and Huang, Z. Y. (1992b). 689 Reproduction in worker honey bees is associated with low juvenile hormone titers 690 and rates of biosynthesis. General and Comparative Endocrinology 87, 471-480. 691 Rueppell, O., Bachelier, C., Fondrk, M. K. and Page, R. E. (2007). Regulation 692 of life history determines lifespan of worker honey bees (Apis mellifera L.). 693 Experimental Gerontology 42, 1020-1032. 694 Rutz, W., Gerig, L., Wille, H. and Lüscher, M. (1976). The function of juvenile 695 hormone in adult worker honeybees, Apis mellifera. Journal of Insect Physiology 696 Rutz, W. and Lüscher, M. (1974). The occurrence of vitellogenin in workers and 697 queens of Apis mellifica and the possibility of its transmission to the queen. 698 Journal of Insect Physiology 20, 897. 699 Salmon, A. B., Richardson, A. and Pérez, V. I. (2010). Update on the oxidative 700 stress theory of aging: does oxidative stress play a role in aging or healthy 701 aging? Free Radical Biology and Medicine 48, 642-655. 702 Scharlaken, B., de, G., DC, Goossens, K., Brunain, M., Peelman, L. J. and 703 Jacobs, F. J. (2008). Reference gene selection for insect expression studies 704 using quantitative real-time PCR: The head of the honeybee, Apis mellifera, after 705 a bacterial challenge. Journal of Insect Science 8, ARTN 33. 706 Seehuus, S. C., Krekling, T. and Amdam, G. V. (2006a). Cellular senescence 707 in honey bee brain is largely independent of chronological age. Experimental 708 Gerontology 41, 1117-1125. 709 Seehuus, S. C., Norberg, K., Gimsa, U., Krekling, T. and Amdam, G. V. 710 (2006b). Reproductive protein protects functionally sterile honey bee workers 711 from oxidative stress. Proceedings of the National Academy of Sciences of the 712 United States of America 103, 962-967. 713 Sunderman, F. W. (1995). The influence of zinc on apoptosis. Annals of Clinical 714 & Laboratory Science 25, 134-142. 715 Toth, A. L. and Robinson, G. E. (2005). Worker nutrition and division of labour 716 in honeybees. Animal Behaviour 69, 427-435. 717 Wang, Y., Azevedo, S. V., Hartfelder, K. and Amdam, G. (2013). Insulin-like 718 peptides (AmIlp1 and AmIlp2) differentially affect female caste development in 719 the honey bee (Apis mellifera). The Journal of Experimental Biology doi:10.1242. 720 Wang, Y., Mutti, N., S., Ihle, K., E., Siegel, A., Dolezal, A., G., Kaftanoglu, O. 721 and Amdam, G., V. (2010). Down-Regulation of Honey Bee IRS Gene Biases 722 Behavior toward Food Rich in Protein. PLoS Genetics 6, e1000896. 723 Williams, J. B., Roberts, S. P. and Elekonich, M. M. (2008). Age and natural 724 metabolically-intensive behavior affect oxidative stress and antioxidant 725 mechanisms. Experimental Gerontology 43, 538-549. 726 Winston, M. L. (1987). The Biology of the Honey Bee. Harvard University Press. 727 Boston, MA. 728 Wu, Q. and Brown, M. R. (2006). Signaling and function of insulin-like peptides 729 in insects. Annual Review of Entomology 51, 1-24. 730 731 7. Figure Captions 732 733 Fig 1. Vg expression in (a-d) fat body and brain (e-h). Relative expression by 734 strain, task and treatment group. (a-d) Fat body RNAi against Vg (vgRNAi) was 735 highly effective in both strains and across behavioral groups, but the knockdown 736 was stronger in the high strain. (e-h) In brain, Vg expression is higher in foragers 737 than in nurses and higher in the vgRNAi group than in the injC group. Asterisks 738 indicate p-values < 0.05. 739 740 Fig 2. Effect of vitellogenin knockdown on total lifespan in (a) high and (b) low 741 strain bees. Panels display the cumulative proportion of treated bees still alive at 742 a given point in time. There was no effect of treatment in high strain workers. In 743 low strain workers the vgRNAi group, bees in which Vg was experimentally 744 reduced from emergence, lived significantly longer than did the injC group, bees 745 that received a control injection of the vehicle. Injection stress also impacted 746 lifespan in the low strain with the noREF group, non-injected control bees, lived 747 significantly longer than the injC treatment group. 748 749 Fig 3. Total lifespan divided into pre- and post-foraging initiation components for 750 both high and low strains. (a) high strain vgRNAi bees have significantly shorter 751 pre-foraging lives than do the injC bees consistent with their earlier foraging 752 onset (Ihle et al., 2010). (b) In contrast, the post-foraging onset lifespans of high 753 strain vgRNAi bees are significantly longer than those of injC bees. (c) In the low 754 strain, while there was an overall effect of treatment, there was no difference in 755 pre-foraging lifespan between the vgRNAi and injC groups. (d) There was no 756 effect of treatment on the post-foraging initiation lifespan in the low strain. 757 758 Fig 4. Ilp1 expression in (a-d) fat body and brain (e-h). Relative expression by 759 strain, task and treatment group. (a-d) Fat body expression of Ilp1 is higher in 760 nurses than in foragers, but was not influenced by strain or treatment. (e-h) 761 Expression of Ilp1 in brain was not affected by any of the factors included in this 762 study. 763 764 Fig 5. Ilp2 expression in (a-d) fat body and brain (e-h). Relative expression by 765 strain, task and treatment group. (a-d) Fat body expression of Ilp2 is higher in 766 nurses than in foragers and higher in low strain workers than in high strain 767 workers. Vg knockdown did not affect fat body expression of Ilp2. (e-h) 768 Expression of Ilp2 in brain was not affected by any of the factors included in this 769 study. 770 771 Fig 6. mnSOD expression in (a-d) fat body and brain (e-h). Relative expression 772 by strain, task and treatment group. (a-d) Fat body expression of mnSOD is 773 higher in nurses than in foragers and higher in low strain workers than in high 774 strain workers. mnSOD expression was not affected by Vg knockdown. However, 775 there was a significant strain by treatment response revealing that the low strain 776 vgRNAi group had increased mnSOD expression relative to the high strain 777 vgRNAi group. (e-h). Brain expression of mnSOD is higher in foragers than in 778 nurses, but this effect is driven by differences in the low strain. 779 Figure(s)           Figure(s)    Figure(s)          Figure(s)          Figure(s)          Figure(s)          Supplemental Data SUPPLEMENTAL INFORMATION Table S1. List of primers and accession numbers. Gene Primer Sequences Accession Number ȕ-actin )¶-TGCCAACACTGTCCTTTCTG- ¶ 5¶-AGAATTGACCCACCAATCCA- ¶ AB023025 vitellogenin )¶- GTTGGAGAGCAACATGCAGA - ¶ 5¶- TCGATCCATTCCTTGATGGT - ¶ AJ517411 insulin-like peptide 1 )¶-CGATAGTCCTGGTCGGTTTG- ¶ 5¶-CAAGCTGAGCATAGCTGCAC- ¶ GB17332-PA insulin-like peptide 2 )¶- TTCCAGAAATGGAGATGGATG- ¶ 5¶-TAGGAGCGCAACTCCTCTGT- ¶ GB10174-PA manganese superoxide dismutase F: 5'- GGTGGTGGTCATTTGAATCATTC-3' R: 5'- AAGAAGTGCAGCGTCTGGTTTAC-3' AY329356