Our expression data suggest a complex interaction between the comma butterfly and its host plants. On the one hand, each plant species appears to require a very specific subset of genes to be regulated in the midgut upon feeding. This involves digestion proteins, immunity related genes, general metabolism genes and ribosomal protein as well as translational and transport genes. In the restbody we also find cell signaling related domains, hormone synthesis genes and genes involved in silk production, suggesting a gene regulation, tuned specifically for each plant species. On the other hand, counting of upregulation of genes in the midgut shows a suggestive pattern (Figure ). Both species of the Urticales family (Ulmus
), as well as both trees, have a higher agreement of gene regulation than do the stinging nettle and the great sallow. Especially the digestion and ribosomal genes show clear differences here (Additional File 1
; Table S1). This suggests that phylogenetic and/or growth form relatedness demand more similar expression profiles in the midgut of the caterpillars.
However, this pattern only holds true for the midgut of the comma butterfly; the restbody of the caterpillars show a completely different picture. Here the stinging nettle and great sallow diet in contrast share the highest similarity in gene expression, while the wych elm appears to require less gene upregulation in general (Additional File 1
; Table S2, Figure ). The midgut is the place of contact with the food bolus, and the location where the first detoxifying and digestive actions will take place, whereas the restbody is only indirectly involved in this process by receiving the solubilized products and only sometimes toxic or toxin degradation products. Here the actual nutrient content is of more importance. This might explain the converse picture we observe in those two tissues. While the defense compounds of the phylogenetically or growth form-related plants might be more similar, our data suggests that the nutritional value does not follow this line.
Previous data on larval performance show that caterpillars of P. c-album
perform differently on the plant species used in this study, suggesting different levels of adaptation and/or different nutritional values (e.g. [34
]). The general pattern is that the plants in Urticales (U. dioica
and U. glabra
) support the highest growth rates, while larvae grow slower on S. caprea
(but pupal weights are typically larger). These differences are also reflected in the preference hierarchy of the ovipositing females, which means that the preference-performance correlation is good on a population level [33
] (Figure ).
Figure 3 Correlation between preference and performance in Polygonia c-album. Correlation between relative oviposition preference (degree of preference in simultaneous choice trials) and larval performance (mean growth rate) for the plants used in the present (more ...)
Oviposition preference and larval performance on U. glabra are more variable between years than the other plants, so that in some years it is considerably less preferred than U. dioica, with a correspondingly lower performance (Nylin, S. and Janz, N., unpublished data). This could be caused by higher between-year variability in U. glabra compared to the other two hosts. Such unpredictable variation would make it more difficult to adapt to the host. Indeed, the lower amount of uniquely upregulated genes on Ulmus found in this study would suggest such insufficient adaptation to this host.
In the midgut we could not detect any differentially expressed known detoxifying genes, such as cytochromes P450s or glutathione-S-transferases (GST). However, those enzymes are with high probability active in the gut of the comma butterfly larvae when encountering plant material. The GeneFishing method identifies solely differentially expressed genes. Hence, P. c-album might very well express such Phase I and Phase II detoxifying genes, but in a constant manner independent of the diet. Being a highly polyphagous insect species, the comma butterfly might be endued with very broad acting detoxifying enzymes that can cope with a wide variety of the compounds and hence are expressed continuously when feeding independent of the diet. The lack of differentially expressed detoxifying enzymes could also be due to overlaps of toxic chemistry on the three host plants.
With the exception of cytochrome P450s and GSTs not much is known about detoxifying enzymes applied by generalist insects. Hence such expressed enzymes could not be identified in databases. In the midgut we identified 4 unknown differentially expressed genes, which could very well be unknown genes involved in detoxification. We could, however, identify a differentially expressed gene homologous to a cytochrome P450 of P. xylostella in the restbody of P. c-album larvae. Cytochrome P450's are also involved in the metabolism of many endogenous compounds, hence this P450 expressed in the restbody is possibly not involved in the detoxification of plant defense compounds, but in metabolism.
While the detoxifying mechanisms appear not to be differentially expressed we see genes belonging to other classes differentially expressed. Short-chain dehydrogenases (SCDH) for example are upregulated in the midguts of on Urtica
feeding larvae. SCDH form a large protein family with highly different enzymes, which only share 15-30% identity among each other [35
]. They are present in all the life forms studied so far, have a very wide substrate spectrum and are generally involved in cellular differentiation and signaling [36
Not surprisingly, also many digestive enzymes are differentially expressed in the midgut (Additional File 1
; Table S1). For example an alpha-amylase was down-regulated in both tree diets in comparison to the stinging nettle. Amylases are enzymes participating in carbohydrate digestion. It is known that insects possess different amylases for starches [37
]. It is possible that the differential expression we observe is due to different starch contents in the stinging nettle. We also found several serine proteases being differentially expressed upon feeding on the host plants, namely trypsins and chymotrypsins. Plants possess proteinase inhibitors (PIs), which are insect inducible peptidases that can suppress insect proteinases and by that reduce the digestibility [38
]. It has been shown that lepidopteran larvae adapt their proteinases expression profile to the PI content of their food plant, upregulating proteinases that are insensitive to the plant PIs [37
]. Our expression patterns suggest different proteinase requirements for the three host plants. We also excised four bands that were identified as homologous to Bmlipase-1; three of them were highly expressed in larvae feeding on Urtica
and one on Salix
. Insect midgut lipases have been studied in few insects so far and little to nothing is known on different expression of lipases in insects [37
]. They form in insects a gene family that underwent many duplication events [40
] with resulting diverse and overlapping function. Bmlipase-1 from Bombyx mori
shows a high antiviral activity against Bombyx mori
nucleopolyhedrovirus, although it is not inducible by viral infection. Its main function is probably as a digestive enzyme, as it is exclusively expressed in the gut tissue and has lipase activity [41
We also found many ribosomal proteins to be differentially regulated upon feeding on different host plants, namely seven in the midgut and twelve in the restbody of P. c-album
. Ribosomal genes are considered to be stably expressed and have been suggested and used as housekeeping genes in expression analysis. However, there are a number of ribosomal proteins that have been found to be differentially expressed upon different treatment and in different tissues [42
], suggesting that there are major differences in expression patterns between different ribosomal proteins.
The complex pattern found in this study should perhaps not be surprising, considering that secondary metabolites, leaf texture and nutrient contents vary greatly between different plant species, and also depending on the plant age [43
]. The three chosen P. c-album
host plants have different leaf architecture, with the stinging nettle possessing thin hairy leaves, and both trees waxy leaves, that differ in shape. Besides the prominent stinging trichomes that only provide an effective defense against vertebrate herbivores [44
], stinging nettle leaves also contain the phenolic compound caffeic acid, tannins, nicotine in measurable amounts and flavonoid glucosides [45
leaves has low levels of toxins but more condensed tannins [47
], which is similar to the situation in elm [45
Phenolic compounds, nicotine, tannins and flavonoids have been shown to affect insects feeding on them. Tannins for example are astringent, bitter plant polyphenols that either bind and precipitate or shrink proteins. Studies on the effect of tannins on the gut pH and redox potential of larvae found that many specialist and generalist insects have developed adaptations to cope with them [48
]. Tannin metabolites are for example oxidized during the gut passage in the aquatic caterpillar Acentria ephemerella
. Resistance against polyphenols appears to be correlated in general with better repair mechanisms in the gut tissue that enables to cope with free radical stress occurring during oxidation better [50
]. Nicotine acts as agonist of the postsynaptic nicotinic acetylcholine receptors of the insect central nervous system. Metabolism of nicotine has been attributed to the action of cytochrome P450 [51
]. Flavonoids in contrast can be sequestered from the diet and used for protection and pigmentation. In B. mori
for example three flavonoid glycosides could be isolated from the cocoon shell [53
]. The role of these secondary plant compounds in the interaction between P. c-album
and its host plant and the adaptations P. c-album
developed to them however, is not known.
Different plants also possess different microfloras [54
], which affect immune response-related gene expression in the midgut of lepidopteran larvae [55
]. We detected cobatoxin and gloverin homologous genes that are known to be inducible by bacterial challenge to be differentially expressed in the midgut tissue [56
]. We see gloverin expression only in Urtica
fed larvae and cobatoxin in Salix
fed larvae, suggesting different microbial environments on those two species. In addition to differences in plant secondary metabolites P c.-album
must therefore also face different bacterial quantities and qualities on its various host plants.