Hedgehog signaling regulates regenerative patterning and growth in Harmonia axyridis leg

Abstract

Appendage regeneration has been widely studied in many species. Compared to other animal models, Harmonia axyridis has the advantage of a short life cycle, is easily reared, has strong regeneration capacity and contains systemic RNAi, making it a model organism for research on appendage regeneration. Here, we performed transcriptome analysis, followed by gene functional assays to reveal the molecular mechanism of H. axyridis leg regenerative growth process. Signaling pathways including Decapentaplegic (Dpp), Wingless (Wg), Ds/Ft/Hippo, Notch, Egfr, and Hedgehog (Hh) were all upregulated during the leg regenerative patterning and growth. Among these, Hh and its auxiliary receptor Lrp2 were required for the proper patterning and growth of the regenerative leg. The targets of canonical Hh signaling were required for the regenerative growth which contributes to the leg length, but were not essential for the pattern formation of the regenerative leg. dpp, wg and leg developmental-related genes including rn, dac and Dll were all regulated by hh and lrp2 and may play an essential role in the regenerative patterning of the leg.

Electronic supplementary material

The online version of this article (10.1007/s00018-020-03631-7) contains supplementary material, which is available to authorized users.

Introduction

Some organisms have regeneration capacity to restore the lost body parts by remodeling and growth from the remaining tissues [1]. From vertebrates to invertebrates, many model organisms have been found to possess such ability. Drosophila melanogaster has been intensively studied to reveal the genetic and molecular mechanism due to its convenient genetic manipulation. Meanwhile, the wing and leg imaginal discs of D. melanogaster have less regenerative ability because they are hard to regenerate to normal size [2–6]. Two-spotted crickets and axolotls have been used as the main leg regeneration models during the larval stage in invertebrate and vertebrates, respectively [7–9]. In addition, as an appendage, fin regeneration in zebrafish was also well studied [10–12].

At the beginning of Drosophila imaginal discs regeneration, JNK signaling is activated in response to the reactive oxygen species (ROS) and promote wound healing [4, 13, 14]. For a formed appendage in two-spotted cricket Gryllus bimaculatus, the blastema consisting of proliferative cells is formed at the wound [15]. Signaling pathways are essential for blastema formation during wound healing, including the Hedgehog (Hh), Dpp/TGFβ, Wg/Wnt, FGF, EGF, IGF, MAPK and JAK/STAT signal pathways [5, 15–20]. In the leg regeneration process of two-spotted cricket G. bimaculatus, JAK/STAT signaling, negatively regulated by Socs, plays an important role in promoting blastema formation [15]. Cell proliferation during the regeneration process of D. melanogaster imaginal disc is also coordinated by JAK/STAT signaling with Dilp8-mediated developmental delay [2]. Activation of Wg/Wnt and Dpp signaling in blastema is necessary for leg regeneration. When the signaling activity of Wg was blocked, no significant regeneration of the amputated legs was observed in G. bimaculatus [21]. Inhibition of Dpp signaling pathway directly by double-strand RNAs against dpp and mothers against dpp (Mad) also can cause leg regeneration defects in the American cockroach [22].

The next stage to blastema formation is regenerative growth and patterning. Dachsous (Ds)/Fat (Ft) signaling pathway can regulate the leg size and shape of regenerating legs in G. bimaculatus. Loss of ds and ft expression induces shorter and thicker regenerative legs. In contrast, expanded/mer can inhibit the over-proliferation along the PD axis during regeneration [23]. An extended steepness model has been proposed to interpret how regenerating tissues are guided [24, 25]. The Ds/Ft signal provides a gradient information in leg segment and growth would stop when the gradient information slope dropped below a certain threshold level [25]. The segment patterning during leg regeneration in the two-spotted cricket can be regulated by methylation of histone H3K27. This phenomenon is achieved by epigenetically regulating the expression of developmental-related genes dac and Egfr [26].

Hh signaling is highly expressed in blastema that can regulate the limb regeneration process. Downregulation of hh at the initial stage can cause the formation of a supernumerary leg [21, 27, 28]. In salamanders, at the early time of limb regeneration, Hh signaling is critical for blastema formation. Cell proliferation can be inhibited at the initial stages in blastema due to the absence of hh and results in pattern defect including no-regeneration or digit loss by using cyclopamine (a sonic hedgehog inhibitor) [29]. The expression feedback loop of Fgf8 and hh from anterior to posterior was sufficient for limb regeneration process, and Grem1 is indispensable for this loop [9]. Expression level of hh decreases with development due to the hypermethylation of hh enhancer [30]. It might be the reason why regeneration defects occur in later development stages. Lrp2 is a newly discovered receptor with the initial binding site for Hh and controls cellular uptake and intracellular trafficking of sonic hedgehog (Shh) during forebrain development in mice, but there is no evidence to show that Lrp2 may be involved in leg regeneration process [31]. It is a type-1 transmembrane receptor of the LDL receptor family expressed in the vertebrate and invertebrate species [32].

H. axyridis is an important predator and widely used to keep the balance of the ecological environment [33]. Meanwhile, its excellent adaptability to the environment facilitated its emergence as an invasive species in Europe and North America [34, 35]. Appendage regeneration ability could be a major contributing factor behind its successful invasion [36, 37]. In our study, compared to cricket or axolotls, its shorter life cycle, easy to rear, stronger regeneration capacity and systemic RNAi makes it an ideal model organism for appendage regeneration research. To uncover the molecular basis during regenerative growth in H. axyridis, we performed comparative transcriptome analysis and then focused on Hh signaling. Canonical Hh signaling regulates the regenerative leg growth which contributes to leg length. Hh and Lrp2 pattern the regenerative leg, possibly downstream Dpp and Wg signaling, and may represent leg patterning genes.

Materials and methods

Source of insects and leg amputation

All larvae and adults of H. axyridis were maintained at 25 ± 1 ℃ with 70 ± 5% humidity. The population was established under laboratory conditions and raised for more than ten generations. Larvae were raised in rectangular plastic boxes (20 × 10 × 8 cm) at densities of 30 individuals per cage and fed with bean aphid (Acyrthosiphon pisum) after amputation or treatment with dsRNA. Following treatment, larvae were moved to a smaller box (12 × 8 × 6 cm) and continued to feed until the adult stage.

H. axyridis larvae take 10–11 days for the larval stage development and are divided into four larval instars. The size of primary hatched larvae is too small (full body length of the 1st instar larva is less than 2 mm) to be manipulated surgically, so the second instar and later stage larvae were selected for leg amputation to observe the regeneration process. The third instar larvae were selected for gene functional research. The mesothoracic legs were amputated with capsulotomy Vannas scissors to remove the distal part of the femur as schematically illustrated in Fig. 1b. To distinguish target genes from other species, we use Ha (short for Harmonia axyridis) plus gene name to present our targets, which is an expression habit for normal insect research.

Fig. 1.

H. axyridis leg was completely regenerated after amputation. a Dorsal view of H. axyridis at second instar. Scale bar represents 1 mm. b Schematic of H. axyridis mesothoracic leg and experimental design, the distal femur was removed and the vertical line marked the amputation position. c The length ratios of regenerated legs which were amputated at the second, third, the first day and the last day of fourth instars, respectively. d One regenerative case of H. axyridis leg (right column) when amputated at the second instar compared with the contralateral leg (left column) during the regenerative growth. Scale bar represents 200 μm

Sample collection and RNA extraction

The regenerative leg (RL) segments were collected at 48 h post-amputation (n = 30), and the same segments of contralateral normal legs (NL) were collected as control group (n = 30) at the third instar larval stage. All treatments have three biological replications. Total RNA samples were extracted by TRIzol Regent, Invitrogen^™, Code No. 12183555. Quality of RNA was verified by AGE and Nanodrop 2000. Then RNA was treated with DNase for 30 min at 37 °C before sequencing.

Transcriptome analysis

The transcript libraries were constructed via Illumina HiSeq 2000 sequencing platform. Raw reads were filtered to remove low-quality reads less than Q20 and the sequence reads contained adapters and poly-A/T tails. The resulting clean reads were assembled using Trinity [38] and unigenes were annotated using BLASTX against various protein databases such as Nr, Swiss-Prot, GO, and KEGG [39]. The expression level of each transcript is presented by FPKM value [40]. Each sample was repeated three times and we used DESeq to analyze the different expression genes (DEGs), and fold change ≥ 2, FDR < 0.01 are screen conditions [41].

Quantitative PCR (q-PCR)

RNA extracted from NL (n = 30) and RL (n = 30) segments was used to validate the RNA-seq result. Following dsRNA treatment, all RNA samples were extracted from the same segments (n = 30). For RNA-seq validation, each treatment has 3 biological replications. cDNA was synthesized with TaKaRa PrimeScript^™RT reagent Kit with gDNA Eraser Code No. RR047A, and q-PCR was achieved with PerfectStart^™ Green qPCR SuperMix (+ Dye II) in QuantStudio 6 Flex platform. RPS18 was the housekeeping gene for q-PCR. For primers, see supplementary material (Table S2 and S3).

Cloning and RNAi for H. axyridis

All targets were cloned with TaKaRa pMD^™19-T Vector Code No. 6013. Double-strand RNA fragments were synthesized with PromegaTM ^T7 RiboMAX^™ Express RNAi System (Code No. P1700); see details on instruction of the kit. For RNAi experiments, we used dsGFP as negative control. dsRNA was injected into the larvae in the gap between the third and fourth abdomen segments after 48 h post-amputation (hpa). Before injection, all the larvae will be narcotized with CO₂ for 3 min. After RNAi treatment, the larvae were moved into a 5 × 5 × 5 cm box, and each box contained three individuals fed with aphids. For target primers, see supplementary materials (Table S3).

Fluorescence RNA in situ hybridization

Leg segments were collected 48 h post-injection (96 h post-amputation) and fixed with paraformaldehyde (PFA) in phosphate-buffered saline with 0.1% Tween 20 (PBT) for 30 min at room temperature. Then leg samples were dehydrated with 30%, 50%, 70% and absolute ethyl alcohol. After slicing and rehydration, samples were refixed in 4% PFA/PBT, and washed in PBT.

Antisense and sense probes were synthesized with DIG RNA Labeling Kit (SP6/T7) REF11175025910. The RNA hybridization was performed as in a previous report [42].

Results

Regeneration capacity of H. axyridis

To investigate the regeneration ability of H. axyridis, different staged larvae were selected to observe the leg regenerating process. The second instar (Fig. 1a) and later stage larvae (Fig. S1a–g) were subjected to leg amputation. The distal part of the femur was removed on the first day of the second instar (Fig. 1b), then the lost tissue was gradually restored along the P/D axis during regeneration (Fig. 1d). The amputated leg can be fully recovered after eclosion (Fig. 1c, d). From the second instar to the early stage of the fourth instar, amputated segments can be regenerated almost completely (Fig. 1d and S1a–g). Regenerated leg length showed no apparent difference between the legs amputated at the second instar and the ones amputated at the third instar (Fig. 1c). The leg regenerative growth declined when the amputation was performed in the late fourth instar, but all segments can still be observed (Fig. S1g). These results suggest that H. axyridis larvae have excellent capacity to regenerate the lost parts of their severed appendages.

In the two-spotted cricket, blastema is formed within 2 days post-amputation (dpa) during the leg regeneration process [27, 28]. Consistently, the blastema structure can be also observed at 48 h post-amputation (hpa) in the H. axyridis leg (Fig. S1h). Ha’Egfr and Ha’CyclinE were highly upregulated to more than 300% (Fig. S1i, j), which means cell proliferation is accelerated within 48 hpa.

Sequencing and assembly of regenerating leg transcriptome of H. axyridis

After the formation of blastema, there should be many signals activated to promote the regenerative growth. To uncover these factors, we performed a transcriptome analysis of the regenerating legs following the formation of blastema. The regenerating legs (RLs) with the distal femur removed and the contralateral normal legs (NLs) were both collected at 48 hpa (Fig. 2a). The RNA samples extracted from amputated and unamputated insects were respectively sequenced using the Illumina Hiseq platform. All sequence reads were assembled into 206,121 transcripts (Table S1) using transcriptome assembly software Trinity [43]. The average length of the assembled transcripts was 503 (Table S1) and each read was mapped to the assembled transcripts to calculate its FPKM value.

Fig. 2.

Comparative transcriptome analysis of RLs and NLs. a Schematic illustration of RNA-seq sample preparation. Legs were amputated at the third instar. The newly formed blastemas of regenerative legs (RLs) were collected for RNA extraction 48 hpa. RNA samples of the contralateral unamputated leg segments (NLs) were set up as the control groups. b–d Analysis of different expressed genes (DEGs) with volcano plot and GO annotation. e–k Heatmap of Hh, MAPK/Egfr, Notch, Hippo, Dpp/TGFβ, Wg/Wnt pathways and development and cell cycle-related genes

Annotation of transcripts in both RLs and NLs

All transcripts were analyzed using the BLASTX program in NR and Gene Ontology (GO) terms databases. In total, 67,995 transcripts of H. axyridis had significant hits with known genes annotated in NR database and showed high homologies to the orthologous genes of Tribolium castaneum (44.34%) and Dendroctonus ponderosae (10.59%) (Fig. S2).

Genome or EST data are currently insufficient for H. axyridis; only 17.8% (36,740 in 206,121) of the transcripts were annotated in GO terms database. To obtain a general overview of the function, these unigenes were divided into three main categories (cellular component, molecular function, biological process) and 55 functional groups (Fig. 2d). In each of the three main categories of the GO terms classification, the ‘metabolic and cellular’, ‘membrane, cell and cell part’, ‘catalytic activity and binding’ terms were the most dominant in both RLs and NLs. There were no unigenes that enriched in ‘rhythmic process and channel regulator activity’ groups in NLs and RLs (Fig. 2d).

Analysis of differentially expressed genes (DEGs)

Compared to the data of NLs, expression of 2765 genes significantly changed in RLs. A total of 1012 genes were upregulated and 1753 were downregulated (Fig. 2b, c), and 87.65% (887 in 1,1012) of upregulated genes were enriched in the section of log2 (fold change) = [1, 3] presented in the volcano plot (Fig. 2c). To validate the results of the comparative transcriptome analysis, we checked the mRNA level of 18 genes that were randomly selected from all DEGs. All selected targets showed similar expression pattern to DEGs analysis (Fig. S3).

To verify whether or not the regenerating process of H. axyridis legs share similar regulating signals with previous models, we performed heatmaps of signaling pathway related to the regenerative process. From the result of comparative analysis, DEGs were enriched in various signaling pathways important for patterning and length maintenance after blastema formation, including Hh, Ds/Ft/Hippo, Egfr, Notch, Dpp/TGFβ, Wg/Wnt pathways [1, 23, 24, 27, 44, 45] and cell cycle-related factors (Fig. 2e–k).

Functional analysis of Hh during the RL growth

In Drosophila, Hh regulates the morphogenetic growth of imaginal discs. We first checked the expression pattern of Haʹhh and Haʹlrp2 during the regenerative growth. Haʹhh was upregulated to 424.97% at 48 hpa and maintained a high expression level until the pupa stage (Fig. 3c and S5b), and from DEGs and q-PCR analysis, expression pattern of Haʹlrp2 was consistent with Ha’hh (Fig. 3c and S5d). To confirm the function of Haʹhh during regenerative growth, we injected the dsRNA at 48 hpa at which point the blastema had already been formed (Fig. 4a). The contralateral legs were set up as endogenous controls to rule out the potential influence of dsHaʹhh on larval leg development (Fig. 4b–i). After eclosion, the phenotypes of adult can be categorized into four classes. In class I (35.71%, n = 10/28), no regeneration occurred but the wound healed well at the distal femur (Fig. 4c). In class II (7.14%, n = 2/28), regenerated legs lost the tarsal and claw segments (Fig. 4e). In class III (50.00%, n = 14/28), regenerated legs had a much smaller tarsus and claw (Fig. 4g). In class IV (7.14%, n = 2/28), there are no apparent regeneration defects (Fig. 4i). These results suggest that Hh is required for the regenerative patterning and growth. This conclusion is not consistent with the result that lacking Hh cannot induce pattern defect after the blastema was formed [29]. Thus, we prepared two sequences to knockdown Hh to rule out the possibility of off-target effect (Fig. 3a, b). Both dsHaʹhh_N and dsHaʹhh_C could decrease the mRNA level of Haʹhh, indicating efficient RNAi (Fig. 4k and S4h). dsHaʹhh_C phenotype also showed pattern defects (Fig S4 d) with similar ratio of each phenotype class to that of dsHaʹhh_N (Fig. 4j and S4g). These results indicated that Hh can regulate leg pattern and cell proliferation along P/D axis during regenerative growth after blastema formed.

Fig. 3.

Expression pattern of HaʹHh and Haʹlrp2 during the leg regeneration process. a, b Protein structure and mRNA sequences of Ha’hh and Ha’Lrp2. Double-sided arrows presented RNAi fragments against mRNA and short, red lines mark the qPCR sequences sites. HH_signal: hedgehog amino-terminal signaling domain, YncE: DNA-binding beta-propeller fold protein YncE, cEGF: complement Clr-like EGF-like. c Expression pattern of Haʹhh and Haʹlrp2 related to normal leg before amputation. The mRNA levels of Haʹhh and Haʹlrp2 increased to their peak from 24 to 48 hpa, then decreased at the pupal stage. The error bar represents the SD of the average from three independent experiments

Fig. 4.

Interference of Ha’Hh leads to pattern defect and shortened regenerative leg. a Schematic of RNAi treatment. Distal femur parts were removed at the third instar stage and dsRNA was injected into the larvae at the fourth instar, 48 hpa. Scale bar represents 0.5 mm. b–i Phenotypes of dsHa’hh_N treated RLs. NLs were set up as controls. The phenotypes can be divided into four classes. In class I, no regeneration occurred distal to the femur; however, the wounded place healed well. In class II, partial regeneration occurred. Here, the tibia of amputated legs regenerated well, but no tarsus and claw segment was found. In class III, shape defect occurred. All segments regenerated including tibia, tarsus, and claw, but the size of each segment was smaller than that of control groups. In class IV, no regeneration defect occurred and the shape and length of regenerated legs were similar to the control. The black arrowhead points to claw structure. j The ratios of defective legs after treatment with dsHaʹhh_N. Phenotype class I: 35.71%; II: 7.14%; III: 50.00%; IV: 7.14%, n = 28. k Expression level of Haʹhh in dsHaʹhh_N treated group (n = 30) went down to 47.97%, relative to the dsGFP group

Hh regulated the RL patterning with dosage effect

In the above RNAi experiments, hh was suppressed ~ 50%, and the phenotypes were various. Therefore, we wondered if the different classes of dsHaʹhh phenotypes were caused by the relative low efficiency of RNAi. To test this hypothesis, we injected a mixture of dsHaʹhh_N and dsHaʹhh_C into the larvae. As expected, the relative expression level of Haʹhh decreased from 47.97% to 22.35% (Figs. 4k, 5f). Severe pattern defects could be found in dsHaʹhh_N + C phenotypes (Fig. 5b), and the ratio of pattern defective legs (I and II classes) also increased from 42.85% (n = 12/28) to 73.33% (n = 22/30) (Figs. 4j, 5e).

Fig. 5.

Severe regenerative defects were dependent on the efficient interference of Haʹhh and Haʹlrp2. a, b Phenotypes of treatments with dsHaʹhh_N and dsHaʹhh_C mixture. Severe pattern defects were induced. Scale bar represents 0.5 mm. c, d Phenotype of dsHaʹlrp2 treatment. The regeneration process was totally blocked. e The ratios of defective legs after RNAi treatments. In dsHaʹhh_N + C group, phenotype class I: 33.33%; II: 40.00%; III: 20.00%; IV: 8.33%, n = 30. In dsHaʹlrp2 group, I: 91.67%; IV: 8.33%, n = 24. f, g Relative to dsGFP group, the mRNA levels of Haʹhh and Haʹlrp2 decreased to 22.35% and 17.91% in dsHaʹhh_N + C and dsHaʹlrp2_N groups, respectively

Lrp2 is an auxiliary receptor for Hh and controls cellular uptake and intracellular trafficking of sonic hedgehog (Shh) during forebrain development in mice [31]. To further confirm the above conclusion from Haʹhh^RNAi, we downregulated the expression of Haʹlrp2 by injection of dsHaʹlrp2_N (Fig. 5d) into the insects with leg amputation. A total of 91.67% (n = 22/24) of adults failed to regenerate amputated legs (Fig. 5e). The ratios of class I phenotype was higher in the dsHaʹlrp2_N group compared to that in the dsHaʹlrp2_C group, which is also reflected by the higher gene suppressing efficiency in the dsHaʹlrp2_N group (Fig. 5e, g) compared to that in the dsHaʹlrp2_C group (Fig. S4f, g), indicating that Hh receptor activity is also required for the regenerative patterning.

Canonical Hh signaling affected the length of RL

We next examined whether canonical Hh signaling activity is required for the regenerative patterning and growth. To block canonical Hh signaling, we prepared dsRNA of Haʹsmo, Haʹci and HaʹCk1 which also were upregulated in DEGs analysis (Fig. 2e), then injected them, respectively, into the larvae after blastema formation. In the phenotypes of dsHaʹsmo, no pattern defect was found. Amputated legs restored all segments with tibia, tarsus and claw along the P/D axis, but with shorter leg length compared to unamputated legs (Fig. 6a, b). mRNA level of Ha’smo decreased to 31.99% (Fig. 6g). Similarly, all legs recovered with all segments intact except for leg length in groups treated with dsHaʹCk1 and dsHaʹci (Fig. 6d, f), indicating that canonical Hh signaling promotes the regenerative growth but maybe not involved in the regenerative patterning.

Fig. 6.

Canonical Hh signaling regulated the regenerative leg length. a–f The phenotypes of dsHaʹsmo, dsHaʹCk1 and dsHaʹci. Only length defect was present in these groups. g–i The mRNA levels of Haʹsmo, HaʹCk1, and Haʹci after RNAi treatments decreased to 31.99%, 21.78%, and 38.83%, respectively. j The expression levels of leg development-related genes were downregulated in the regenerative leg when Hh signaling was suppressed. The asterisks indicate significant differences (*p < 0.05; **p < 0.01; ***p < 0.001) compared to the relevant control with a two-tailed t test. Plotted are means ± SEM

To figure out the potential mediators of the roles Hh and Lrp2 play on the regenerative patterning, we examined the expression of Haʹdpp, Ha’wg and leg pattern genes including rotund (rn), dachshund (dac) and distal-less (Dll) [46–49] that were upregulated in DEGs analysis (Fig. 2f, i, k). From the qPCR results, when Haʹhh and Haʹlrp2 were suppressed during regenerative growth, mRNA levels of Haʹwg, Haʹ dpp, Haʹrn, Haʹdac, HaʹDll and HaʹCyclin E all decreased (Fig. 6j and S5e–x). These preliminary results imply that the patterning role of Hh and Lrp2 may be dependent on Dpp, Wg signaling and leg pattern factors. Further RNAi experiments would need to be done to clarify the relationships between these factor and successful appendage regeneration.

Discussion

H. axyridis has remarkable appendage regeneration ability

In our experiments, H. axyridis showed remarkable leg regeneration capacity. Amputated legs could be fully recovered when the distal femur is removed before the late larval stage. Regenerated leg parts are close to normal legs in regard to either segment structure or in shape and length. Even in legs amputated 1 day before pupation had regenerated smaller legs that were also well patterned with all segments (Fig. S1f, g).

It takes about 25 days for H. axyridis to develop from egg to mature adult under the experimental condition [50]. Only 10 days were needed for complete leg regeneration (Fig. 1d). However, leg regeneration needs 18 days in two-spot cricket and 60 days in red-spot newt, which are often coupled with smaller size if the distal femur was amputated.

Systemic RNA silencing in Coleoptera beetles [51–53] also provide us an excellent tool for studying H. axyridis gene function during the leg regeneration process, as verified in our experiments; therefore H. axyridis may become a model animal for appendage regeneration studies.

Conserved signaling pathways in regulating appendage regeneration

Amputated legs formed blastema within 48 hpa (Fig. S1h). After that, regenerative growth process was accelerated to restore the lost part with good pattern formation and leg length. Comparative transcriptome analysis showed that 1012 genes were upregulated at this time, including Ds/Ft/Hippo, Egfr, Notch, Dpp/TGFβ, Wg/Wnt pathways and developmental related genes, Dll, rn, and dac etc. (Fig. 2a, e–k).

Ds/Ft/Hippo pathway has already been mentioned in regulating length and shape during the regeneration process in two-spotted cricket G. bimaculatus [24]. Egfr signaling is required for the regeneration of the distal leg structures, the tarsus and claws, with the activation of Dll, dac and al [54]. Notch signaling can coordinate proliferation with differentiation during regenerative growth by maintaining blastema cells in a proliferative state [12]. Normal development segmentation of D. melanogaster legs also can be mediated by Notch signaling [55–57]. Morphogens including Dpp and Wg play important roles in regulating cell fate determination, differentiation and proliferation in D. melanogaster imaginal discs [58–63]. During leg regeneration process, Dpp and Wg also participate in the regulation of blastema formation, repatterning and proliferation [21, 22, 27].

Hh presents multiple functions in regulating appendage regeneration

Amputated legs were unable to recover the correct pattern when lacking Hh activity during the blastema formation, which is the initial stage of regeneration. In red-spotted newts, lack of Hh activity leads to bone structure defect with less fingers or no regeneration [29]. In two-spotted crickets, lack of Hh activity induces shorter supernumerary leg at the distal position [28]. In mouse neural tube development process, Shh presents a gradient mechanism at long range in regulating cell fate determination [64]. From the results of smo^RNAi, ci^RNAi and Ck1^RNAi, canonical Hh signaling only can regulate the cell proliferation process along the P/D axis (Fig. 6a–f). Lrp2 also known as megalin, which can shuttle extracellularly applied Hh, TGFβ, but it does not participate in regulating the expression of these morphogens directly [31, 65].

We examined the function of Hh after blastema formation and found that Hh also can regulate cell proliferation along the P/D axis and patterning of the regenerating legs during regenerative growth (Fig. 4b–i). The ratios of pattern defects can be increased by intensive inhibition of Haʹhh and HaʹLrp2 (Fig. 5 and S4), which means phenotypes caused by dsHaʹhh may be dependent on the concentration of Hh. Haʹdpp, Haʹwg, and that other leg developmental genes were downregulated when inhibiting Haʹhh and Haʹlrp2, indicating a role of non-canonical Hh signaling in leg patterning during regenerative growth. In our experiments, we set up the contralateral legs as endogenous control which showed no developmental defect after Haʹhh or Haʹlrp2 downregulation (Figs. 4, 5). The transcriptome analysis and RNA in situ hybridization both revealed very low expression level of Haʹhh in larval legs (Fig. S5). Therefore, Hh signaling was not required for the normal leg development since larval stage in H. axyridis. Similarly, Dachsous/Fat signaling is also required for crickets leg regeneration, but not for larval leg development. This phenomenon is considered as regeneration dependent RNAi [23]. But we cannot rule out the possibility that Hh regulates leg development during embryonic and very early larval stages. T. castaneum larval leg regeneration also requires Hh signaling, but Hh signaling is also involved in metamorphic transformation of the legs. Lack of Hh caused T. castaneum pupa death with size and pattern defective legs [66].

From previous studies, the process of regeneration from blastema to a well-patterned leg shares similarity with a leg bud or imaginal disc development [7, 28]. Therefore, the pattern defect caused by dsHaʹhh and dsHaʹlrp2 during regenerative growth could be owing to downregulation of leg development-related transcription factors including Haʹrn, Haʹdac and HaʹDll.

In conclusion, this study demonstrates that H. axyridis larva has remarkable appendage regeneration capacity and it shares similar regulating pathway including Dpp, Wg, Hippo, Notch, Egfr, and Hh during blastema formation. In addition, we find that Hh presented multiple functions in regulating leg regenerative growth process. Canonical Hh signaling is required for regenerative growth along P/D axis. While non-canonical Hh signaling might be mediated by its auxiliary receptor Lrp2, and downstream targets genes dpp, wg, and other leg development-related genes regulate the regenerative patterning.

Electronic supplementary material

Below is the link to the electronic supplementary material. The primer sequences and the supplementary figures are all listed in detail in the Supplementary Materials.

Author contributions

HZ, ZM, and ZW performed experiments. HZ, JS, SY, and DW designed the experiments, analyzed and interpreted the data, and wrote the manuscript.

Compliance with ethical standards

Conflict of interest

The authors declare no competing interests.