Changes in Aphid Host Plant Diet Influence the Small-RNA Expression Profiles of Its Obligate Nutritional Symbiont, Buchnera

In general, the genomes of intracellular bacterial symbionts are reduced compared to those of free-living relatives and lack many key regulatory genes. Many of these reduced genomes belong to obligate mutualists of insects that feed on a diet that is deficient in essential nutrients, such as essential amino acids. It is unclear if these symbionts respond with their host to changes in insect diet, because of their reduced regulatory capacity. Emerging evidence suggests that these symbionts express small RNAs (sRNAs) that regulate gene expression at the posttranscriptional level. Therefore, in this study, we sought to determine if the reduced genome of the nutritional symbiont Buchnera in the pea aphid responds to changes in the aphid’s host plant diet. This study demonstrates for the first time that Buchnera sRNAs, some conserved in two or more Buchnera lineages, are differentially expressed when aphids feed on different plant species and potentially target genes within essential amino acid biosynthesis pathways.

H erbivorous insects are faced with the challenge of using a food resource that contains defensive compounds and varies in nutritional profiles both spatially and temporally (1). Consequently, insect herbivores have evolved a diversity of mechanisms that facilitate host plant use (1,2). One mechanism that has aided various phytophagous insects in using plants as nutrient resources is the acquisition of microbial symbionts (3)(4)(5)(6). Within the insect order Hemiptera, many examples of such bacterial symbioses are widely found (7). In the hemipteran suborders Auchenorrhyncha and Sternorrhyncha, many insects have obligate nutritional symbionts that are housed intracellularly in specialized insect cells (bacteriocytes) within the insect's body (3,7). These obligate intracellular symbionts have highly eroded genomes that lack many key regulatory genes (8,9). Consequently, it is unclear if intracellular symbionts can respond to changes in insect diet to facilitate host plant use, especially in comparison to extracellular symbionts that often directly interface with the plant material and have larger, more dynamic genomes.
Within Hemiptera, the partnership between Acyrthosiphon pisum (pea aphid) and the bacterium Buchnera is one of the best-characterized nutritional, intracellular symbioses within insects. In this symbiosis, each partner depends on the other for the production of amino acids (10)(11)(12). The symbiont Buchnera possesses the majority of genes within pathways for essential amino acid biosynthesis (13), which are not possessed by the aphid (14). The aphid possesses genes that are predicted to be vital in the regulation of this microbial symbiosis. These aphid genes include those that are involved in amino acid transport, ammonia recycling, synthesis of amino donors, and metabolite intermediates, as well as the terminal steps for essential amino acid biosynthesis that are missing from the Buchnera genome (11,12,(14)(15)(16). Homologs of these aphid genes are also upregulated in bacteriocytes from other hemipteran taxa, suggesting that host control of these intracellular symbioses is widespread among hemipterans (17)(18)(19)(20). These insect genes potentially can respond to host-plant diet for example; Kim et al. (21) observed the upregulation of several of these collaborating aphid genes in bacteriocytes when aphids fed on their suboptimal, specialized host plant, alfalfa (Medicago sativa), compared to their universal host plant, fava bean (Vicia faba).
Currently, there is limited evidence that supports the role of intracellular symbionts responding to changes in insect diet to facilitate host plant use. For example, Buchnera displays negligible gene expression responses at the mRNA level when aphids feed on an artificial diet or plant material that has been spiked with different concentrations of amino acids (22,23). However, Viñuelas and colleagues (24) demonstrated that Buchnera's pLeu plasmid can respond to changes in the aphid's diet by increasing plasmid copy number and upregulating the genes on the plasmid (leuABCD, repA1, repA2, and yqhA) in response to limited concentrations of leucine in an artificial diet. These results suggest that in the face of genome reduction, Buchnera displays limited transcriptional control on its chromosome in response to diet variation; however, other regulatory strategies may still be maintained via plasmids. Currently, it is unknown if posttranscriptional regulation is important in Buchnera in response to host plant diet.
Posttranscriptional strategies of gene regulation, such as small RNAs (sRNAs), have been widely observed throughout all domains of life (25)(26)(27)(28)(29), and emerging evidence supports the role of regulatory sRNAs in intracellular bacterial symbiont gene regulation (30). For example, sRNAs expressed from the tick symbionts Rickettsia prowazekii and Rickettsia conorii are hypothesized to be important in facilitating host niche adaptation (31,32). In Buchnera, sRNAs are hypothesized to be important in regulating genes at the posttranscriptional level when Buchnera transitions between different life stages (33,34). In turn, there is potential for Buchnera sRNAs to aid in regulating essential amino acid and vitamin biosynthesis pathways in response to changes in nutrient demand. Changes in nutrient demand may occur when aphids feed on host plants that vary in free amino acid content as well as defensive compounds such as saponins and flavonols, which may inhibit nutrient uptake by the aphid, especially on alfalfa (M. sativa) (35)(36)(37)(38)(39)(40).
In this study, we sought to determine if sRNAs expressed by the intracellular symbiont Buchnera respond to aphid host plant diet. Specifically, we investigated if Buchnera sRNAs are differentially regulated when A. pisum feeds on fava bean (V. faba) compared to alfalfa (M. sativa). These host plants were chosen because they vary in amino acid profiles and host plant defenses (35)(36)(37)(38)(39)(40). Moreover, the A. pisum (LSR1) sublines used in this study display higher fitness when they feed on their "universal" host plant, fava bean (FB), than on their specialized host plant alfalfa, (ALF), and several aphid genes involved in the nutritional symbioses are differentially expressed between host plant diets (21). In this study, we use transcriptome sequencing (RNA-seq) to characterize Buchnera sRNA expression between the aphid's universal (FB) and specialized (ALF) host plant diets for two different Buchnera life stages: (i) aphid ovarioles (referred to here as embryos [EMB]), in which early in aphid development Buchnera is in its extracellular proliferating state, and (ii) maternal bacteriocytes (BAC), in which Buchnera is in an intracellular state (41). These two Buchnera life stages were collected separately because it was previously shown that Buchnera displays differential sRNA (34) and protein expression profiles (33) between these life stages.

RESULTS
Buchnera sRNA expression profiles are influenced by aphid host plant diet and Buchnera life stage. Across all 12 RNA-seq samples, ALF-BAC1 to -3, ALF-EMB1 to -3, FB-BAC1 to -3, and FB-EMB1 to -3, an average of 96% of all high-quality reads mapped to either the Buchnera or the aphid genome (see Table S1 in the supplemental material). Each sample was sequenced to a similar read depth, with an average in ALF samples of 2.67 ϫ 10 7 reads and 3.12 ϫ 10 7 reads in FB samples mapping to the Buchnera genome (Table S1).
After identifying expressed sRNAs using Rockhopper, principal-component analysis (PCA) and permutational multivariate analysis of variance (PERMANOVA) were run to determine how similar the Buchnera sRNA expression profiles were across treatments. Using PERMANOVA, the sRNA expression profiles of the groups ALF-BAC, ALF-EMB, FB-BAC, and FB-EMB were found to be significantly different from each other (P Յ 0.01; F ϭ 1.696). When grouped by Buchnera life stage, the sRNA expression profiles were found to be significantly different from each other (P ϭ 0.001; F ϭ 8.2119). When grouped by host plant treatment, the sRNA profiles were also found to be significantly different (P ϭ 0.001; F ϭ 5.0376). This segregation of sRNA expression profiles by host plant treatment and life stage was supported by the PCA ordination results (Fig. 1). The first three PCA axes explained 73% of the variation in the data set, with the first PCA axis (PC1) accounting for 35% (Ϯ 13.21 standard deviations [SD] of the variation), the second axis (PC2) accounting for 22% (Ϯ 10.43 SD), and the third axis (PC3) accounting for 16% (Ϯ 8.90 SD). Combined, these results suggest that both aphid host plant and Buchnera life stage have significant effects on sRNA expression profiles. Among the samples from aphids feeding on alfalfa, there was a more heterogenous response in the sRNA expression profiles (Fig. 1) than for samples from aphids feeding on fava bean. Despite this heterogenous response among ALF samples, it is important to note that there was still a significant difference in sRNA expression profiles observed between host plant treatments for both life stages. Moreover, across samples no batch effect was observed (see Text S1 [supplemental results]).
Aphid host plant diet significantly affects Buchnera sRNA expression for genes involved in essential amino acid biosynthesis. To determine how sRNA expression profiles (including tRNAs) differ between host plant treatments for each life stage, Rockhopper was used. Within the comparison of ALF-BAC to FB-BAC, 103 sRNAs were upregulated in the ALF-BAC treatment group and 28 sRNAs were upregulated in FB-BAC treatment group. Within the comparison of ALF-EMB to FB-EMB, 96 sRNAs were upregulated in the ALF-EMB treatment group and 74 sRNAs were upregulated in the FB-EMB treatment group (FDR-adjusted P value, q Յ 0.05; 1.5-fold change [ Table 1]).  Of the 279 sRNAs (not including tRNAs) differentially expressed between the two aphid host plant treatments for both life stages (BAC: ALF-BAC versus FB-BAC) and (EMB: ALF-EMB versus FB-EMB), 74 sRNAs were found to be differentially expressed in both BAC and EMB ( Fig. 2 and Tables S2A and B, S3A and B, and S4A and B). Largely due to this observed overlap, the same 11 PANTHER GO pathways associated with 11 putative coding DNA sequence (CDS) targets of sRNAs were identified for both life stage comparisons between the two aphid host plant treatments (BAC and EMB) ( Table 2). Five of these GO pathways are associated with essential amino acid biosynthesis, specifically the arginine, chorismate, isoleucine, threonine, and valine biosynthesis pathways ( Table 2). The vitamin B 6 metabolism GO pathway was also found to  a Pathways in bold are related to essential amino acid biosynthesis. "n.s." indicates that the specific sRNA was not differentially expressed in one of the host plant comparisons. Note that some sRNAs are present in more than one pathway.
be targeted by a differentially expressed sRNA ( Table 2). For the comparison of ALF-BAC to FB-BAC, there are three sRNAs (aroA_1, ilvI_1, and thrC_3) that were significantly expressed in the bacteriocyte comparison that were not significantly expressed in the embryo comparison of ALF-EMB to FB-EMB. Similarly, in the comparison of ALF-EMB to FB-EMB, there are three sRNAs (ilvI_2, thrB_2, and thrB_3) that were significantly expressed in the embryo comparison that were not significantly expressed in the bacteriocyte comparison of ALF-BAC to FB-BAC (Table 2).
In the comparison of ALF-BAC to FB-BAC, 7 of the 11 sRNAs associated with the PANTHER GO pathways were upregulated in FB-BAC ( Table 2). The antisense sRNAs aroA_1, ilvC_2, murD_1, and murD_2 were upregulated in ALF-BAC. In the comparison of ALF-EMB to FB-EMB, 8 of the 11 sRNAs predicted to target CDSs were upregulated in the FB-EMB samples, with the antisense sRNAs ilvC_2, murD_1, and murD_3 being upregulated in the ALF-EMB samples ( Table 2). Of the 14 differentially expressed sRNAs found within both comparisons, only the antisense sRNA ilvD was found to be conserved among the Buchnera lineages of A. pisum, Uroleucon ambrosiae, and Schizaphis graminum (33) (Table S2A). Eight out of 14 sRNAs are unique to the A. pisum Buchnera lineage and were observed previously in the strains 5A and LSR1 (33) ( Table S2A). The antisense sRNAs aroC, ilvI_2, prsA, thrB_2, and thrB_3 were identified for the first time in this study for the A. pisum Buchnera strain LSR1 (33) (Table S2A).
Antisense tRNA expression has been previously observed within Buchnera (34,42). Within this study, more than 60% of the differentially expressed tRNAs were expressed in the antisense direction (Table 1 and Table S6A and B). All antisense tRNAs were upregulated in the FB host plant treatments for both comparisons of ALF-BAC to FB-BAC and ALF-EMB to FB-EMB, and of these antisense sRNAs, four (28%) (antisense asparagine, histidine, glutamate, and methionine tRNAs) are conserved in multiple Buchnera lineages (42) (Table S6A and B).
Buchnera sRNAs are differentially expressed between life stages when aphids feed on either host plant. To determine if Buchnera sRNA expression is different between life stages, we analyzed the following sRNA expression profiles for each host plant treatment: (i) ALF-BAC compared to ALF-EMB and (ii) FB-BAC compared to FB-EMB. For the ALF host plant treatment, one sRNA was upregulated in the ALF-BAC life stage and the remainder, 29 sRNAs, were upregulated in the ALF-EMB life stage (q Յ 0.05; 1.5-fold change [ Table 3]). This pattern of more sRNAs being upregulated in the EMB than in the BAC life stage was also observed in the FB host plant treatment, where 222 sRNAs were upregulated in the FB-EMB life stage and 15 were upregulated in the FB-BAC life stage (q Յ 0.05; 1.5-fold change [ Table 3]. Notably, there was a high level of overlap for the sRNAs that were differentially expressed between life stages for both host plant comparisons. Of the 30 sRNAs differentially expressed in the comparison of ALF-BAC to ALF-EMB, 83% were also differentially expressed, in the same direction for the comparison of FB-BAC to FB-EMB ( Fig. 2 and Tables S2C and D, S3C and D, S4C and D, and S5C and D).
For the comparison of ALF-BAC to ALF-EMB, the antisense sRNAs ilvI_1 and ilvI_5 were predicted to target CDSs within the PANTHER GO pathways for the isoleucine and valine biosynthesis pathways. In the comparison of FB-BAC to FB-EMB, 15 antisense sRNAs, which were all upregulated in the FB-EMB samples, were predicted to target CDSs within 16 PANTHER GO pathways ( Table 4). The antisense sRNAs aroA_1, aroC, carB_1, ilvD, ilvI_1, and ilvI_5 were predicted to target CDSs within five PANTHER GO pathways that are associated with essential amino acids (arginine, chorismate, isoleucine, threonine, and valine) ( Table 4). The antisense sRNA serC_2 was predicted to target three GO pathways, including the vitamin B 6 pathway. Of these 15 antisense sRNAs, 3 (carB-1, ilvD, and pta) are conserved across two or more Buchnera lineages (33) ( Table S2D). The remaining 12 antisense sRNAs are specific to the A. pisum Buchnera lineages, with antisense the sRNAs aroA_1, ilvI_1, ilvI_5, and murD_3 being conserved in both the Buchnera A. pisum strains 5A and LSR1 (33) (Table S2D).
Antisense tRNAs made up the majority of the differentially expressed tRNAs identified between life stage comparisons for both aphid host plant treatments. Within the comparison of ALF-BAC to ALF-EMB, three differentially expressed antisense tRNAs were upregulated in the ALF-EMB samples (Table S6C). One of these antisense tRNAs (glutamate) was also identified in the comparisons of ALF-BAC to FB-BAC and ALF-EMB to FB-EMB and is conserved among multiple Buchnera lineages (42). In the comparison of FB-BAC to FB-EMB, 14 antisense tRNAs were differentially expressed and upregulated in the FB-EMB samples (Table S6D). Four of these antisense tRNAs (asparagine, histidine, glutamate, and methionine tRNAs), which were also identified in the comparisons of ALF-BAC to FB-BAC and ALF-EMB to FB-EB, are conserved among Buchnera lineages (42) (Table S6C and D).

DISCUSSION
This study demonstrated for the first time that Buchnera sRNAs are differentially expressed between host plant species and potentially target genes within essential Antisense serc_2 EMB a Pathways in bold are related to essential amino acid biosynthesis. Note that some sRNAs are present in more than one pathway. b Antisense sRNAs dut_1, ilvI_1, and ilvI_5 were also detected in the comparison of ALF-BAC to ALF-EMB and were upregulated in the EMB life stage group. None of the other sRNAs targeting CDS in the PANTHER GO pathways listed were differentially expressed in the comparison of ALF-BAC to ALF-EMB, and therefore a column was not included for this comparison.
amino acid biosynthesis pathways. The results from this study also suggest that even though sRNAs are differentially expressed between Buchnera life stages (34; this study), sRNA expression profiles are heavily influenced by what host plant species the aphid feeds upon. This study also found that 21% and 26% of Buchnera sRNAs that are differentially expressed between aphid host plant species and between life stages, respectively, are conserved in two or more Buchnera taxa. It is unclear at this time if Buchnera displays differential protein expression when aphids feed on different host plants, and if putative differential protein expression is linked to the candidate CDS targets for sRNAs identified in this study. It will be of interest for future studies to address whether these sRNA candidates are important in facilitating posttranscriptional regulation in Buchnera when aphids feed on different host plant diets. Buchnera sRNA expression is more heterogenous among biological replicates in both life stages when A. pisum feeds on its specialized host plant alfalfa than when it feeds on its universal host plant, fava bean (Fig. 1). It is uncertain if this heterogenous response is driven by one of the aphid sublines or is a variable response to its specialized host plant, alfalfa. Despite this diffuse response in sRNA expression profiles, there is a significant difference in sRNA expression observed between host plant treatments for both life stages, and no batch effects were observed. We hypothesize that the potential variation of metabolites (which include amino acids and plant defense compounds) in alfalfa may contribute to the heterogeneity observed in Buchnera sRNA expression profiles when aphids feed on alfalfa. The two A. pisum host plants used in this study are known to vary significantly in essential amino acid profiles (35) and metabolite profiles, including host plant defense compounds (38,39). Aphid stylectomy experiments have also shown that fava bean and alfalfa have different concentrations of arginine, isoleucine, leucine, lysine, phenylalanine, tryptophan, tyrosine, valine, and histidine, with alfalfa generally having lower concentrations of essential amino acids than fava bean (35). Regardless of Buchnera life stage, sRNAs predicted to target genes within the arginine, chorismate (phenylalanine, tryptophan, and tyrosine), isoleucine, threonine, and valine biosynthesis pathways are differentially expressed between aphid host plant treatments, which corresponds to the difference in host plant amino acid content ( Table 2, Fig. 3, and Fig. S1). Previous studies have shown that A. pisum aphids that feed on alfalfa display lower mass than that of aphids that feed on fava bean, which is suggestive that they obtain fewer nutrients and/or expend more energy when they feed on alfalfa (21,43,44). This variation in aphid mass could also be due to the variation in metabolites that act as feeding deterrents in alfalfa. When comparing the metabolite profiles among the A. pisum host plants alfalfa, fava bean, pea, and red clover, Sanchez-Arcos et al. (38) found that alfalfa plants contain the highest number of unique metabolites, especially compared to that of fava bean plants. Notably, Sanchez-Arcos et al. (38) found that saponins, a known A. pisum feeding deterrent (43,44), were specific to alfalfa and the most abundant plant chemical compound within this host plant. An alternative hypothesis to explain this host plant-specific pattern is that Buchnera sRNAs may respond sporadically, in a nondirected manner, if Buchnera is stressed when the aphid feeds on a metabolically complex diet such as alfalfa (43,44). Of the 322 differentially expressed sRNAs identified in this study, 18% are predicted to target genes that are associated with stress responses in Escherichia coli (45). Future work is needed to determine if the observed aphid host plant-specific response of Buchnera sRNAs results in the differential expression of predicted Buchnera protein targets, which potentially can impact this symbiosis in either an adaptive nutritional manner or a nonadaptive random stress response.
In the life stage comparisons of ALF-BAC to ALF-EMB and FB-BAC to FB-EMB, there was notable overlap in the differentially expressed antisense sRNAs identified for both host plant treatments ( Fig. 2 and Table S2C 6 (ansA_1, alaS_2, leuS, ilvI_1, ilvI_5, and truB_1) had significant thermodynamically stable secondary-structure prediction. The presence of thermodynamically stable secondary structures in part suggests that these sRNAs may potentially have a functional role (46). Other than the antisense sRNAs ilvI_1 and ilvI_5, which were predicted to target CDS within the isoleucine and valine biosynthesis PANTHER GO pathways, the remaining four antisense sRNAs were predicted to target genes in eight PANTHER GO biological processes, which include mRNA pseudouridine synthesis and pseudouridine synthesis (Table S5). Buchnera is unable to produce its own uracil, a precursor for pseudouridine synthesis, and therefore, it depends on the host for uracil biosynthesis (13) (Fig. 3). Pseudouridine is a widespread and functionally important posttranscriptional modification of RNAs within eukaryotes (47). Within E. coli, not only can pseudouridine modification of mRNAs influence translation, reducing protein expression (48), but also the presence of pseudouridine in a stop codon can result in readthrough (49). It is therefore possible that pseudouridine modification is important in Buchnera to facilitate different translation rates between life stages. In the comparisons of ALF-BAC to ALF-EMB and FB-BAC to FB-EMB, there was also a trend in which more sRNAs were upregulated in the EMB life stage than in the BAC life stage (Table 3). A similar trend was also observed in previous experiments comparing Buchnera sRNA expression with the EMB and BAC aphid life stages (34). One key difference between these two aphid life stages is that within the EMB life stage, Buchnera transitions from an extracellular proliferating state within the aphid embryos to an intracellular nonproliferating one in the aphid bacteriocytes (41). The difference in sRNA expression between the two life stages suggests that dynamic gene regulation is required during the extracellular transitionary period of the aphid embryo in contrast to the intracellular state within the homeostatic environment of the aphid bacteriocyte.
Insects, like many other animals, cannot synthesize B vitamins de novo, and so they obtain these necessary micronutrients from their diet or their association with various Buchnera sRNA Response to Shifts in Host Diet ® microorganisms (50). Within aphids, Buchnera possesses several genes for the biotin (B 7 ), folic acid (B 9 ), pantothenate (B 5 ), pyridoxine (B 6 ) riboflavin (B 2 ), and thiamine (B 1 ) biosynthesis pathways (13,51). Vitamin B 6 is an important bacterial and eukaryotic cofactor, especially in amino acid metabolism (52,53), and neither Buchnera nor A. pisum has all the genes needed for its biosynthesis. Previous work looking at aphid gene expression has shown that within bacteriocytes of A. pisum that have fed on alfalfa, the vitamin B 6 pathway was significantly enriched (21). In this study, it was found that in the comparison of ALF-BAC to FB-BAC, the antisense sRNA predicted to target thrC, one of the two Buchnera genes vital for vitamin B 6 biosynthesis, was upregulated in the ALF-BAC samples (Table S2A). In addition, a second antisense sRNA, which was upregulated in the FB-EMB samples compared to the FB-BAC samples, was predicted to target serC, the other Buchnera gene vital for vitamin B 6 biosynthesis. Further studies that directly manipulate vitamin B nutrition, in conjunction with proteomics, will help illuminate whether the antisense sRNAs thrC and serC differently regulate their predicted CDSs.
One of the most important tasks in the field of bacterial sRNA research is determining if a particular putative sRNA has a functional role. Working in nonmodel, nonculturable systems, such as Buchnera, increases this challenge because many of the techniques used to validate and determine functionality are not easily implemented or possible (30). Nevertheless, the results from this study are consistent with previous Buchnera sRNA experiments, which provide support for the potential function of sRNA candidates. For example, Thairu et al. (34) observed Buchnera sRNA expression between bacteriocytes and embryos of aphids that fed on fava bean. Thairu and colleagues (34) identified 90 differentially expressed sRNAs, 27 of which were also identified in the comparison of FB-BAC to FB-EMB conducted in this study (34) (Tables S2D, S3D, S4D,  and S6D). Both studies also identified sRNAs predicted to target CDSs within similar PANTHER GO pathways, specifically, the acetate utilization, arginine biosynthesis, de novo purine biosynthesis, de novo pyrimidine ribonucleotide biosynthesis, isoleucine biosynthesis, peptidoglycan biosynthesis, and valine biosynthesis pathways. Though the two studies used the same aphid lines, methodological differences, such as different insect rearing locations and differences in sequencing platforms, library preparation, and centers, could account for some the differences observed.
In the report by Thairu et al. (34), the authors provided evidence for the functionality of the antisense sRNA carB (carB_1 in this study) by heterologously expressing the sRNA in E. coli. This study not only detected this sRNA but also found that it had the same expression pattern as described by Thairu et al. (34) between different Buchnera life stages. There was also an overlap between this study and the proteomic study conducted by Hansen and Degnan (33), who characterized the differentially expressed Buchnera proteins between Buchnera life stages when aphids feed on fava bean. Specifically, 27 of the 54 differentially expressed proteins identified by Hansen and Degnan (33) were either directly or indirectly (i.e., within an operon) associated with a differentially expressed sRNAs identified in this study (Table S7). This study also found 72 (53 antisense sRNAs, 15 UTR sRNAs, and 4 intergenic sRNAs) differentially expressed conserved sRNAs, which were also characterized by Hansen and Degnan (33) (Tables S2,  S3, and S4). The conservation of sRNAs across four Buchnera lineages (Buchnera A. pisum, Acyrthosiphon kondoi, Uroleucon ambrosiae, and Schizaphis graminum) that diverged over 65 million years ago strongly suggests that some sRNAs may be important in Buchnera's gene regulation. Recent genomic studies further support the hypothesis that sRNA regulation is likely conserved across Buchnera strains. Specifically, Chong et al. (54) recently compared the genomes of 39 Buchnera strains and found that across these taxa, 29 genes are under strong positive selection within the aphid subfamily Aphidinae. Interestingly, of these 29 genes, there is evidence that 24 are predicted targets of conserved antisense or UTR sRNAs (33). Twenty of the 29 Buchnera genes under positive selection (54) were predicted to be either direct or indirect targets of one or more differentially expressed antisense sRNAs identified in this study. Four of these genes (asps, mtlA, rnr, and serC) were also found to be differentially expressed as proteins in Buchnera embryos and maternal bacteriocytes, which corresponds to differential expression patterns of antisense sRNA identified in this and other studies (33,34).
In recent years, increasing numbers of studies have begun to find sRNAs being expressed in highly reduced genomes, including organelles (30,55). Though the roles of sRNAs within these reduced genomes are still poorly understood, within mitochondria and chloroplasts they have been described to respond to stress (56)(57)(58)(59) and host development and/or tissue type (60)(61)(62)(63). Based on the trends observed in other reduced genomes and the patterns that we observed within the Buchnera system, we hypothesize that sRNAs can be maintained and evolve in reduced genomes potentially to help compensate for the loss of regulatory proteins. Host-restricted bacterial symbionts with highly eroded genomes, such as Buchnera, face the challenge of having to compensate for the loss of key genomic material, and until recently, many of the compensatory methods that have been described have been only been host meditated. Moving forward, based on our data and other studies, especially since positive selection can occur in these reduced genomes (54), we predict that more examples of symbiont mediated control will emerge. For each subline treatment, two life stage samples were cocollected from the same 4th-instar aphid nymph individual via dissection, similar to the method of Thairu et al. (34). These two samples represent two different Buchnera life stages: (i) aphid ovarioles (EMB) and (ii) maternal bacteriocytes (BAC). For each sample, approximately 200 aphids from each subline were dissected for each life stage and pooled. In total, 12 samples were collected: three bacteriocyte biological replicate samples from aphids feeding on alfalfa (ALF-BAC), three bacteriocyte biological replicate samples from aphids feeding on fava bean (FB-BAC), three embryo biological replicate samples from aphids feeding on alfalfa (ALF-EMB), and three embryo biological replicate samples from aphids feeding on fava bean (FB-EMB). All tissues were immediately stored in RNAprotect bacterial reagent (Qiagen, Germantown, MD) and stored at Ϫ80°C.

Small
For each sample (n ϭ 12), RNA was extracted using an miRNAeasy kit (Qiagen, Germantown, MD). Library preparation and sequencing were then performed on the small-RNA-enriched fraction (Յ250 nucleotides [nt]) using the Illumina (San Diego, CA) mRNA directional sequencing protocol by the University of California, San Diego, Institute for Genomic Medicine Genomics Center (UCSD IGM Genomics Center). Each library was then sequenced as 75-nt single-end reads on an Illumina Hi-seq 4000 at the UCSD IGM Genomics Center.
Identification and categorization of Buchnera sRNAs. Reads for each sample were independently quality screened using Trimmomatic v.0.33 (64). Adapters were then removed using Cutadapt v.2.1 (65). To remove aphid reads, sequences were aligned to the aphid genome (GCA_000142985.2) using Bowtie2 v.2.2.9 (66). Once aphid reads were removed, Bowtie2 was used to map the remaining reads to the Buchnera genome (GCA_000174075.1). For all Bowtie2 runs the default settings were used (66). Rockhopper v.2.0.3 (67) was then used to identify putative Buchnera sRNAs in each sample. The Rockhopper parameters optimized for identifying Buchnera sRNAs by Hansen and Degnan (33) were used to identify sRNAs in this study. Reads were normalized by the upper-quartile method in Rockhopper. sRNA boundaries were manually determined by inspecting directional coverage curves in Artemis v.16 (68). sRNAs were then binned into three different categories, similar to those of Hansen and Degnan (33) and Thairu et al. (34): (i) sRNAs expressed antisense to the gene (antisense sRNAs), (ii) sRNAs expressed within the untranslated regions of genes (UTR sRNAs), and (iii) sRNAs identified within the intergenic spacer regions (intergenic sRNAs). As in the works of Hansen and Degnan (33) and Thairu et al. (34), antisense and UTR sRNAs are named after their predicted target coding sequence (CDS) based on direct basepairing interactions. For example, the antisense sRNA aroC, which is expressed antisense to the aroC coding sequence, is predicted to target the CDS aroC. If multiple sRNAs are predicted to target different regions of the same CDS, then a number follows (e.g., antisense sRNAs ilvI_1 and ilvI_2). For sRNAs expressed within the intergenic spacer region, the sRNA name contains both the upstream and downstream CDS names (e.g., intergenic sRNA argH-yibN).
Though target binding and functionality of the putative cis-acting Buchnera sRNAs identified in this study have not been determined, we predict that they would function similarly to cis-acting antisense sRNAs in free-living bacterial systems (see reviews in references 69 to 71). Previous work with Buchnera has validated the functionality of an expressed antisense sRNA in vitro using heterologous expression (34). Further, in the work of Hansen and Degnan (33), 52 protein targets showed evidence of posttranscriptional regulation, which may be mediated by these types of cis-RNA interactions.
Using the read counts for each of the identified sRNAs from Rockhopper, a principal-component analysis (PCA) was conducted in R v.3.5.2 (72) using the package DEbrowser v.1.10.6 (73) to compare how similar Buchnera sRNA expression profiles were across all treatments. The reads were filtered using the default DEbrowser settings and were normalized by the upper-quartile method, the same method used by Rockhopper (67). The following four groups were compared: ALF-BAC, ALF-EMB, FB-BAC, and FB-EMB. Using R, a permutational multivariate analysis of variance (PERMANOVA) was used to determine if sRNA profiles were significantly different between treatments (74).
Rockhopper was also used to determine if sRNAs were differentially expressed between host plant treatments and life stage categories. sRNA expression comparisons include (i) bacteriocytes from aphids feeding on alfalfa (ALF-BAC) compared to bacteriocytes from aphids feeding on fava (FB-BAC), (ii) embryos from aphids feeding on alfalfa (ALF-EMB) compared to embryos from aphids feeding on fava (FB-EMB), (iii) bacteriocytes compared to embryos from aphids feeding on alfalfa (ALF-BAC and ALF-EMB, respectively), and (iv) bacteriocytes compared to embryos from aphids feeding on fava (FB-BAC and FB-EMB, respectively). Significance criteria of a Ն1.5-fold change between samples and a q of Ͻ0.05 were used to determine if an sRNA was differentially expressed. The putative secondary structures were predicted for all differentially expressed sRNAs using RNAalifold v.2.1 (75,76), following the same methods as Hansen and Degnan (33).
Using PANTHER GO functional gene list analysis (77), the GO pathways and GO biological processes were identified for the predicted CDS targets of sRNAs that displayed both significant differential expression and predicted thermodynamic stability. Only putative cis-acting sRNAs (e.g., antisense and UTR sRNAs, which have hypothetical direct base-pairing interactions with their CDS target) were included in this analysis, because it is uncertain whether putative intergenic sRNAs target one or both neighboring genes or act in trans and target other genes located distantly in the genome.
Accession number(s). The sequence data for RNA-seq reads from all samples were submitted to the NCBI database under BioProject accession number PRJNA549470.