3.Results
3.1. Data distribution
We compared gene expression patterns for both switch experiments using principal component (PC) analysis. Overall, individuals did not cluster by pre- or post-switch host at either time point. (Supplementary Figures 1 and 2). Rather, individuals that experienced specific transitions between hosts stood out from the remaining larvae. For example, in Switch A, individuals that were switched from Ulmus toUrtica tended to cluster separately from other samples. This pattern could be seen in the first two PC axes after 2 hours, and for the second and third PC axes after 17 hours. In Switch B, individuals that originally fed on Salix clustered separately from the other larvae after 2 hours, but this pattern was lost after17 hrs. In contrast, individuals moved to Urtica clustered only after 17 hours (PC axes 1 & 2, Supplementary Figure 2e). While individuals did not cluster by family on any of the first three PC axes in Switch A, family clusters were clearly discernible in the Switch B PCA. This was especially true for PC axis 3. Due to this, family structure was accounted for in the downstream analyses.
3.2. Global assessment of gene expression in response to host switches
We assessed the influence of the two hosts involved in a host switch in an ANOVA-like framework, testing for an overall effect of Host 1 on expression, an overall effect of Host 2 on expression, or an interacting effect of Host 1 and Host 2 at each timepoint. To do this, we combined all of the different results across individual hosts, in order to get a global assessment of host switching effects in each of our two experiments. Overall, the total number of significantly differentially expressed genes (FDR < 0.05) differed between the two switch experiments (Figure 2). In Switch A, a total of 4.2% (n = 550) and 1.9% (n = 249) of genes were significantly differentially expressed after 2 and 17 hours of feeding, respectively. At both time points, the highest number of differentially expressed genes was associated with the host they were feeding on before the switch (i.e. Host 1). This differed considerably from Switch B, in which most differences in expression between larvae that could be attributed to Host 1 and Host 2 shifted over time. After 2 hours 32.9% of genes (n = 4148) were significantly differentially expressed, and the majority of these were associated with the pre-switch plant (Host 1). After 17 hours 32.7% (n = 4216) of genes showed significantly different expression. Here, the highest number of differentially expressed genes was driven by the new host (Host 2). This suggests that the actual plants that are involved in a switch experiment play an important role in the time course and extent of expression differences.
3.3. Differentially expressed genes between switches
To further understand the transcriptional effects of switching between two particular host plants, we compared differential gene expression between caterpillars that were exclusively reared on the same host (i.e. Control, e.g. UdUd ) with individuals that were moved to an alternative host (i.e. Switch, e.g. UdSc ). In both switch experiments, the host pairs that showed the highest number of differentially expressed genes after 2 hours clearly differed from those at the 17 hours after the switch (Figure 3).
In Switch A, the highest number of significantly differentially expressed genes was generally found between comparisons in which larvae were switched onto Urtica (aka Ud ; Figure 3a). After 2 hours, most differentially expressed genes were found in the switch fromUlmus to Urtica (UgUg - UgUd ). In the other comparisons, the number of significantly differentially expressed genes was 6 or fewer. After 17 hours, the number of differentially expressed genes was significantly lower. Here, a maximum of 3 significantly differentially expressed genes were found in the comparison between theUrtica and Salix -control line (UdUd - ScSc ).
Many more genes were found to be differentially expressed in Switch B than in Switch A. Moreover, a more dynamic pattern was found when comparing the two time points. After 2 hours, the highest number of differentially expressed genes was measured in comparisons betweenUrtica and Salix and between Salix andRibes . After 17 hours, overall higher numbers of differentially expressed genes were found in switches that involved Ribes. By this time, most transcriptional differences were detected in the pairwise comparison of the Urtica control line with the switchUrtica to Ribes (i.e. UdUd - UdRu ). This further indicates that the same or similar processes underlie the use of plants with similar chemical properties. In contrast, the utilization of hosts that represent a chemically or structurally more challenging environment requires more specific mechanisms and, thus, a higher number of transcriptional changes.
Due to the low number of transcriptional differences after 17 hours in Switch A, we focus subsequent analyses on Switch B. However, the corresponding figures for Switch A can be found in the
supplementary material (Supplementary Figure 3, Supplementary Table 1).
3.4. Transcriptional adjustments to the host switch
Given that larvae in switch B showed distinct shifts in transcriptional profiles between 2 and 17 hours after switching to a novel host, we wanted to explore whether these larvae were recovering the expression patterns of larvae reared continuously on the novel host (i.e, the control lines). This would demonstrate extensive adaptive transcriptional plasticity with an adjustment of a core set of genes specific to a particular host plant. Alternatively, gene expression in switched larvae may reflect stress responses caused by switching to any novel host as well as transient adjustments to a specific host plant (which would not be seen in control lines). To test these alternative hypotheses, we evaluated the transcriptional differences between switches and control lines against expectations from pairwise comparisons with larvae exclusively feeding on one host (e.g,UdUd - UdRu compared to UdUd - RuRu , i.e. Control). This framework allowed for transcriptional profiles to be classified into gene expression patterns that either matched or contradicted their control line comparisons (Figure 4).
After 2 hours, gene expression profiles in larvae switched to a new host mainly corresponded to those of their pre-switch host (Figure 5), consistent with category 1 in Figure 4. This supported our previous results (Figure 2), and suggests that after 2 hours larvae had not yet adjusted their transcriptional response to the new plant. Overall, this pattern was found for all combinations of host plants, and the direction of a switch did not seem to play a crucial role. However, a slight deviation from this broader pattern was found in the switch fromSalix to Urtica ; after 2 hours, six genes showed early transcriptional adjustments to the new host (category 2, Figure 4). These genes were all involved in the “regulation of peptidoglycan recognition protein signaling pathway” (Supplementary Table 2). In the reciprocal switch from Urtica to Salix , these differences were not found. In addition, one gene that matched category 2 expectations was found 2 hours after individuals were moved fromSalix to Ribes . This gene showed a significant enrichment for “regulation of transcription involved in G1/S transition of mitotic cell cycle ” and “biomineralization ”. Again, the switch into the other direction (Ribes to Salix ) did not show these differences.
After 17 hours, the differences in the gene expression indicated host-specific transcriptional plasticity, depending strongly on the plants involved and the direction of the switch. Though few genes were differentially expressed between Urtica - Salix switches and the respective control lines after 17 hours (e.g., UdUd -UdSc and ScSc - ScUd ; Figure 5), a large proportion (51.4% (n=38), 28% (n=14) respectively) of these genes were not also differentially expressed between the control lines (consistent with category 3, Figure 4). This suggests these genes are not typically involved in the more stable patterns of gene expression promoting the ability to eat these different hosts (category 1), but instead reflect a general stress response due to the switch between hosts, or a transient response in order to regain homeostasis on a specific host. The strongest transcriptional discrepancy was found in the switches betweenUrtica and Ribes . In both directions a large proportion of the genes showed an adjustment to the new host (UdUd -UdRu : 9.5% (n=123); RuRu - RuUd : 17.9% (n=90)). Especially in the switch to Ribes, most of these genes were associated with functions directly involved in metabolic processes (Supplementary Table 2). This pattern was also found in the switch fromSalix to Ribes (ScSc - ScRu ) but not in the other direction, which further supports a crucial role of these genes for the utilization of Ribes . The majority of the genes were, however, uniquely differentially expressed between switch but not the control lines (UdUd - UdRU : 84.6% (n=1096), RuRu -RuUd : 60.4% (n=304), ScSc - ScRu : 73% (n=580)). Our hypothesis that these genes are associated with a response to the experience of eating a novel host is supported by an enrichment of genes that were involved in metabolic processes and genes that could be associated with a stress response. These patterns are in line with our expectations as they confirm that a switch to an alternative host can not only represent a stressful situation but also requires a specific regulation of molecular and physiological processes to regain cellular homeostasis in the new environment.
3.5. Congruence with previous studies
Switches between Urtica and Ribes are hypothesized to be the most challenging for P. c-album larvae (Celorio-Manceraet al. 2023). Our results support this hypothesis, finding the largest expression differences associated with switching between these two hosts (Figures 2, 5), and a large proportion of differentially expressed genes that are unique to switches (i.e. consistent with category 3, Figure 4). We thus aimed to evaluate the extent to which our results overlap with these earlier results, in which larvae were not switched between hosts but instead reared on the same host until they were sampled (Celorio-Mancera et al. 2023). First, it can be noted that the results of the two studies are highly congruent. In the present study, there were 199 genes differing significantly in gene expression between Urtica and Ribes controls. Of 113 genes upregulated on Ribes , 91 could be matched to genes in the earlier database, and 29 of these genes were also significantly upregulated onRibes in the earlier study (Celorio-Mancera et al. 2023). 85 genes were upregulated on Urtica in our study, out of which 75 could be matched, and 20 of these genes were also upregulated onUrtica in the previous study. Only one gene was found to be expressed in the opposite direction between the two studies (upregulated on Urtica in our study, downregulated in the previous). Moreover, genes differing most strongly (highest fold-change) in both studies were most likely to overlap. For instance, 7 of the genes with highest fold-change in the present study were found among the 20 genes differing most strongly in the earlier study (Celorio-Manceraet al . 2023). In both studies, a synaptic vesicle protein and an antibacterial peptide were strongly upregulated on Urtica . Similarly, a strong upregulation of an aldo-keto reductase, a proline dehydrogenase, a spermine oxidase-like and one non-annotated gene was found on Ribes in both studies. We further used an extensive functional annotation from Celorio-Manceraet al . (2023) to identify the putative function of some of the genes that were highly up- or downregulated after a switch fromUrtica to Ribes or vice versa at 17 hours after the switch. These genes are candidates for representing a direct response to the new host plant rather than a downstream effect. In this study, we could confirm the involvement of genes that had previously been indicated by Celorio-Mancera et al. (2023). Moreover, we were also able to identify new candidates that appear to play an important role in the direct response and, thus, the utilization of Urticaand Ribes (Table 1).
3.6. Larval performance
In order to test whether switching to a new host also has consequences at a phenotypic level, the performance of larvae was measured until they reached the pupal stage. Despite the transcriptional responses found here, the host switches had no significant effect (χ2= 13.372, df= 8, p=0.0997) on the larval growth rate (Figure 6). Even after excluding pre-switch days from the analysis, no significant effect (χ2= 3.1934, df= 8, p=0.9216) on growth rate could be measured (Supplementary Figure 4). This suggests that the caterpillars of Polygonia c-album can adjust to a new host without major expenses for their development.