Full-Length Transcriptome Profiling of the Complete Mitochondrial Genome of Sericothrips houjii (Thysanoptera: Thripidae: Sericothripinae) Featuring Extensive Gene Rearrangement and Duplicated Control Regions

Simple Summary

A total of 37 mitochondrial genes appears in a certain arrangement in most insects, and the transcription of this conserved mitochondrial genome (mitogenome) has been well-studied in species like Drosophila melanogaster, Erthesina fullo and Coridius chinensis. However, transcription of the mitogenome with extensive gene rearrangement has rarely been studied. In this research, we sequenced the mitogenome and mitochondrial transcriptome of Sericothrips houjii (Thysanoptera: Thripidae: Sericothripinae). The mitogenome of S. houjii exhibited extensive gene rearrangement and contained two control regions (CRs) with sequence repeats. Compared to the insect mitogenome with a typical gene order, the mitogenomic feature, relative gene expression level, transcriptional model and post-transcriptional cleavage of S. houjii were quite different. Unlike other insects where ribosomal RNA (rRNA) is typically highly expressed, ND4/ND4L in S. houjii exhibits the highest expression. Both strands of this mitogenome were entirely transcribed and the bicistronic messenger RNA (mRNA) COI/ND3 was reported for the first time in insects. Our study provides new insights into the transcriptional and post-transcriptional regulation processes in the insect mitogenome with extensive gene rearrangement and duplicated CRs.

Abstract

The mitochondrial genome (mitogenome) of Thysanoptera has extensive gene rearrangement, and some species have repeatable control regions. To investigate the characteristics of the gene expression, transcription and post-transcriptional processes in such extensively gene-rearranged mitogenomes, we sequenced the mitogenome and mitochondrial transcriptome of Sericothrips houjii to analyze. The mitogenome was 14,965 bp in length and included two CRs contains 140 bp repeats between COIII-trnN (CR1) and trnT-trnP (CR2). Unlike the putative ancestral arrangement of insects, S. houjii exhibited only six conserved gene blocks encompassing 14 genes (trnL2-COII, trnD-trnK, ND2-trnW, ATP8-ATP6, ND5-trnH-ND4-ND4L and trnV-lrRNA). A quantitative transcription map showed the gene with the highest relative expression in the mitogenome was ND4-ND4L. Based on analyses of polycistronic transcripts, non-coding RNAs (ncRNAs) and antisense transcripts, we proposed a transcriptional model of this mitogenome. Both CRs contained the transcription initiation sites (TISs) and transcription termination sites (TTSs) of both strands, and an additional TIS for the majority strand (J-strand) was found within antisense lrRNA. The post-transcriptional cleavage processes followed the “tRNA punctuation” model. After the cleavage of transfer RNAs (tRNAs), COI and ND3 matured as bicistronic mRNA COI/ND3 due to the translocation of intervening tRNAs, and the 3′ untranslated region (UTR) remained in the mRNAs for COII, COIII, CYTB and ND5. Additionally, isoform RNAs of ND2, srRNA and lrRNA were identified. In summary, the relative mitochondrial gene expression levels, transcriptional model and post-transcriptional cleavage process of S. houjii are notably different from those insects with typical mitochondrial gene arrangements. In addition, the phylogenetic tree of Thripidae including S. houjii was reconstructed. Our study provides insights into the phylogenetic status of Sericothripinae and the transcriptional and post-transcriptional regulation processes of extensively gene-rearranged insect mitogenomes.

1. Introduction

The insect mitochondrial genome (mitogenome) is a double-strand circular DNA, which typically includes 13 protein-coding genes (PCGs), 22 transfer RNA genes (tRNAs), 2 ribosomal RNA genes (rRNAs) and a control region (CR) that plays a key role in regulating replication and transcription [1,2]. In many insect groups, including those with earlier origins, these components are arranged in a compact manner with a typical order, which is inferred as the ancestral gene arrangement of insect mitogenomes (hereinafter referred to as “ancestral arrangement”) [3,4,5]. With the rapid advancement of DNA sequencing technology, the duplication of CRs and gene rearrangements in various insect taxa have been increasingly characterized as more insect mitogenomes are sequenced [6,7,8]. To date, studies on insect mitogenomes have primarily focused on the aspects such as base composition, codon usage, gene arrangement and phylogenetic evolution [4,9,10], while studies on their transcriptional model and post-transcriptional processing have been limited to few species. Five transcriptional cassettes in Drosophila have been inferred by Berthier et al. [11] and confirmed by Stewart et al. through 5′ and 3′ RACE (rapid amplification of cDNA ends), circularization and RT-PCR (reverse transcription polymerase chain reaction) methods [12]. Roberti et al. [13] discovered the Drosophila mitochondrial transcription termination factor (DmTTF) binding to two transcription termination sites (TTSs) and proposed two models of mitogenomic transcription. The difference among these models was whether the gene blocks of astrnS2-asCYTB-asND6 and astrnF-asND5-astrnH-asND4-asND4L have been transcribed. The post-transcriptional cleavage process was thought to follow the reverse cleavage of the “tRNA punctuation” model, which suggested tRNAs were cleaved from 3′ end to 5′ end in the primary transcript to release the messenger RNA (mRNA) and rRNA [12,14].

Recently, the full-length transcriptome has been utilized to profile mitochondrial gene expression in insects. In Erthesina fullo, five primary transcripts were mapped into the five transcriptional cassettes of Drosophila, two novel long non-coding RNAs (lncRNAs) of CR have been detected and both the reverse cleavage and forward cleavage model (tRNAs were cleaved from 5′ end to 3′ end) have been reported [15,16]. Further studies investigated the mitochondrial transcriptional models in Coridius chinensis and Aphidius gifuensis suggest that the mitogenome is transcribed continuously from CR, producing an almost complete primary polycistronic transcript from both strands [17,18]. The only exception is that the antisense genes downstream of trnS2 are not transcribed. Each strand has been transcribed into one primary polycistronic transcript except the majority strand (J-strand) in the A. gifuensis mitogenome, which forms at least five primary polycistronic transcripts as there were five TTSs recognized. In A. gifuensis, the gene clusters of CR-trnI-trnM-trnQ and trnW-trnC-trnY have been rearranged as trnI-CR1-trnM-CR2-trnQ and trnW-trnY-trnC, but there is no repeated sequence between the two CRs [18]. These rearranged gene clusters seem to have little effect on the multiple TTSs of the J-strand, and the TISs of two strands were distributed in different CRs. It remains unclear whether more extensive gene rearrangement and duplicated CRs affects transcription, post-transcriptional cleavage and the number of TIS and TTS.

Compare to the ancestral arrangement, the thysanopteran mitogenome is well-known for its extensive gene rearrangement, with only 10–18 relative conserved genes encompassed by 3–6 gene blocks [19]. Therefore, the mitogenome of Thysanoptera serves as an excellent model for studying the processes of mitogenomic transcription in an extensive gene rearrangement. In this study, we sequenced the mitogenome and the full-length mitochondrial transcriptome of Sericothrips houjii. Firstly, we precisely annotated the mitogenome of S. houjii by using the information of mitochondrial transcripts and illustrate the characteristics of mitogenome. Next, we constructed a quantitative transcription map using high-quality full-length mitochondrial transcripts which were generated from single-molecule, real-time (SMRT) sequencing. Then, the polycistronic RNAs, mature RNAs and isoform RNAs were counted in detail. Finally, the models of mitochondrial transcription and post-transcriptional cleavage in S. houjii were proposed and compared with other insects. These results revealed the unique features of mitochondrial transcripts in S. houjii and offered valuable insight into the patterns of gene transcription and post-transcriptional regulation in a mitogenome with extensive gene rearrangement.

Currently, four subfamilies, Panchaetothripinae, Dendrothripinae, Sericothripinae and Thripinae, are recognized in Thripidae [20]. The phylogenetic analyses based on morphology and several molecular loci support Panchaetothripinae, Dendrothripinae and Sericothripinae nested within Thripinae [21,22]. Meanwhile, the phylogenetic analyses based on mitochondrial genes support only Sericothripinae nested within Thripinae [19,23,24], indicating that the mitochondrial genes are useful to recover the monophyly of the subfamilies. However, only one mitogenome of a species, Neohydatothrips samayunkur, from Sericothripinae has been published to date. The monophyly and phylogenetic status of Sericothripinae remains unclear. Here, we sequenced the mitogenome from Sericothrips, which is the type genus of Sericothripinae. To further understand the phylogenetic status of Sericothripinae, the phylogeny of Thripidae was analyzed using Aeolothripidae as the outgroup based on the published mitogenomes. Our study provides insights into the taxonomic decision and phylogeny of Thripidae.

2. Materials and Methods

2.1. Sample Collection

Specimens of S. houjii were collected from alfalfa (Medicago sativa) in Hongsibao, Wuzhong, Ningxia, China, in August 2020. Living thrips were frozen by drikold (−80 °C) for transcriptome extraction or soaked in absolute ethyl alcohol stored at −20 °C for total genomic DNA extraction. Voucher specimens were preserved at the Entomological Museum of China Agricultural University, Beijing, China.

2.2. Acquisition and Annotation of Mitogenome

The total genomic DNA were extracted from seven adults using the DNeasy Blood and Tissue kit (Qiagen, Dusseldorf, Germany) according to the manufacturer’s protocol. An Illumina TruSeq library was prepared with an average insert size of 350 bp and sequenced on the Illumina Hiseq 6000 platform (Berry Genomics, Beijing, China) with 150 bp paired-end reads. A total of 6 Gb raw data was obtained and adapters were trimmed from raw reads using Trimmomatic [25]. The short and low-quality reads were removed by using Prinseq v0.20.4 [26] with the parameter poly-Ns > 15 bp, <75 bp in length and quality score < 3. The remaining clean data were used to de novo assemble by IDBA-UD [27], with overlapping similarity > 98%, and minimum and maximum k values of 45 bp and 145 bp, respectively. A 541 bp fragment of the COI gene (GenBank accession number is HQ605969) was used to identify the mitochondrial scaffold using BLASTN [28] searches with at least 98% similarity. To confirm the accuracy of the assembly, clean reads were mapped to the obtained mitogenome using Geneious v10.1.3 (http://www.geneious.com/, accessed on 15 September 2019), with mismatches of up to 2%, a maximum gap size of 3 bp and a minimum overlap of 40 bp. Finally, the complete mitogenome of S. houjii was obtained with a 35,545× average sequencing depth.

The mitogenome of S. houjii was preliminarily annotated by MitoZ [29] with settings “genetic_code 5” and “clade Arthropoda”, and further corrected by alignment with homologous mitochondrial genes of Neohydatothrips samayunkur (Thysanoptera: Thripidae: Sericothripinae. GenBank accession number is MF991901).

2.3. Acquisition of Mitochondrial Transcriptome

A total of 50 adults of S. houjii were pooled for total RNA extraction, using TRIzol Universal (Tiangen, Beijing, China). The concentration and purity of total RNA was detected by NanoDrop 2000 spectrophotometry (Thermo Fisher, Waltham, MA, USA), and the integrity was detected by the Agilent 2100 system (Agilent Technologies, Santa Clara, CA, USA). Total RNA was reverse transcribed into cDNA using the Clontech SMARTer PCR cDNA Synthesis Kit (Clontech Laboratories, Inc., Mountain View, CA, USA) with 3′ SMART CDS Primer II A (5′-AAGCAGTGGTATCAACGCAGAGTAC-T₍₃₀₎-3′) and 5′ SMARTer II A Oligonucleotide (5′-AAGCAGTGGTATCAACGCAGAGTACATGGG-3′). The cDNA was amplified and constructed a library with an insert size between 1 and 10 kb was constructed according to the PacBio IsoSeq protocol after size selection. Finally, one SMRT cell was performed on the PacBio Sequel platform with circular consensus sequencing (CCS) mode at Berry Genomics Company (Beijing, China).

The PacBio SMART Analysis v10.2 (http://www.pacb.com/devnet/, accessed on 17 December 2021) was performed on the raw data. Of which, CCS v5.0.0 with parameters (Minimum Full Passes = 1, Minimum Predicted Accuracy = 0.9) was used to process the sequenced reads into the high-quality circular consensus sequencing (CCS) reads, lima v2.0.0 was used to produce full-length non-chimera (FLNC) reads also known as draft transcripts by removing the full-length non-chimera, primers of cDNA and polyA sequences in 3′ end, and Samtools v1.11 (http://www.htslib.org/, accessed on 18 December 2021) was used to read the sequences in the binary file (BAM file).

2.4. Mitochondrial Transcript Identification

The mitochondrial transcripts were mapped to the circular topology mitogenome using Geneious v10.1.3, with maximum gap per read = 10%, maximum gap size = 8, minimum overlap identity = 90%, maximum mismatches per read = 5%, and maximum ambiguity = 5. The coverage depth of each base in the reference mitogenome were calculated to the coverage histogram, namely, quantitative transcription map. The 5′ and 3′ ends of mature transcripts, polycistronic transcripts and antisense transcripts were identified and classified after precise modification of the mitogenome according to the full-length transcripts. The representative transcripts from each of the same type were plotted on the quantitative transcription map. The mature and precursor transcripts were identified by aligning the boundaries of genes or gene blocks. The transcripts with 5′ and/or 3′ ends truncated (not mapped to the gene boundaries of the gene or gene block) were considered as intermediate degradation products. Isoform RNA can be identified by multiple transcripts of a single gene with identical 5′ and 3′ ends, but the 5′ and/or 3′ ends are not aligned to the gene boundary.

2.5. Phylogenetic Analysis

The amino acid of each PCG and nucleotide of each rRNA were individually aligned using MAFFT [30]. The poorly aligned sites of rRNA were removed using trimAL [31] with the parameter “-automated1”, while additional parameters “-ignorestopcodon -backtrans” was used to remove the stop codon and back translation of the amino acid sequence. The PCG alignments excluded the third codon position in MEGA 7.0.21 [32]. Alignment concatenation was performed in Geneious v10.1.3 (http://www.geneious.com/, accessed on 15 September 2019) to generate the datasets PCG_rRNA (13 PCGs and two rRNAs) and PCG12_rRNA (13 PCGs with the third codon position excluded and two rRNAs) for phylogenetic analyses. A site-heterogeneous mixture model (CAT + GTR) implemented in PhyloBayes avoids the false grouping of unrelated taxa with similar base composition and accelerated evolutionary rate in the reconstruction of the phylogeny in Thysanoptera [19]. Thus, the datasets were analyzed under a CAT+GTR model using Phylobayes MPI 1.8c [33] on the CIPRES Science Gateway [34]. In each analysis, two independent chains starting from a random tree were run for 30,000 cycles, with trees being sampled every cycle until 30,000 trees were sampled. The initial 7500 trees of each MCMC run were discarded as burn-in. A consensus tree was generated from the remaining trees combined from two runs.

3. Results

3.1. The Mitogenome Feature of Sericothrips houjii

We annotated the mitogenome of S. houjii using MitoZ and homologous alignment, and further modified the annotations precisely according to the full-length transcripts. The annotation with MitoZ and homologous alignment are based on the nucleotide alignment, the mitochondrial gene code of PCG and the cloverleaf structure of tRNA [29]. MitoZ was precise in the gene arrangement and homologous alignment corrected the translation frame, but did not predict the gene boundaries well. Compared to the annotation results of MitoZ and homologous alignment, the boundary of seven PCGs (ND3, COIII, CYTB, ND2, ND1, ATP6 and ND6), two tRNAs (trnR and trnT) and two rRNAs (sRNA and lrRNA) was modified according to the full-length transcripts (

Table 1. Comparison of the annotation of the Sericothrips houjii mitogenome based on DNA sequences and full-length transcripts.

Gene	Coding Strand	Annotation Using DNA Sequencing				Annotations According to Transcripts
		Position	Length (bp)	Start Codon	Stop Codon	Position	Length (bp)	Strat Codon	Stop Codon
COI	J	1–1539	1539	AUA	UAA	1–1539	1539	AUA	UAA
ND3	J	1552–1892	341	AUU	UA	1552–1891	340	AUU	U
trnL2	J	1892–1955	64			1892–1955	64
COII	J	1956–2618	663	AUU	UAA	1956–2618	663	AUU	UAA
trnR	J	2665–2718	54			2664–2718	55
trnG	J	2719–2779	61			2719–2779	61
trnD	J	2798–2858	61			2798–2858	61
trnK	J	2897–2957	61			2897–2957	61
COIII	J	3070–3867	798	AUA	UAA	2959–3867	909	AUA	UAA
CR1		3868–4052	185			3868–4052	185
trnN	J	4053–4117	65			4053–4117	65
trnE	J	4115–4179	65			4115–4179	65
trnQ	J	4183–4251	69			4183–4251	69
trnA	J	4291–4353	63			4291–4353	63
trnS1	J	4354–4407	54			4354–4407	54
trnI	J	4409–4477	69			4409–4477	69
CYTB	J	4482–5588	1107	AUA	UAA	4479–5588	1110	AUA	UAA
trnY	N	5650–5586	65			5650–5586	65
ND2	J	5678–6656	979	AUA	U	5678–6659	982	AUA	U
trnW	J	6660–6722	63			6660–6722	63
ND1	J	6729–7641	913	AUU	U	6723–7641	919	AUA	U
trnM	J	7642–7702	61			7642–7702	61
trnF	J	7712–7777	66			7712–7777	66
srRNA	J	7771–8504	734			7778–8500	723
ATP8	N	8501–8669	169	AUU	U	8501–8669	169	AUU	U
ATP6	J	8636–9346	711	AUA	UAA	8696–9346	651	AUU	UAA
trnL2	J	9347–9410	64			9347–9410	64
trnT	J	9418–9483	66			9411–9477	66
CR2		9484–9736	253			9478–9736	259
trnP	N	9801–9737	65			9801–9737	65
ND5	N	11,532–9856	1677	AUA	UAG	11,532–9856	1677	AUA	UAG
trnH	N	11,592–11,533	60			11,592–11,533	60
ND4	N	12,896–11,593	1304	AUU	UAA	12,896–11,593	1304	AUU	UAA
ND4L	N	13,168–12,890	279	UAG	UAG	13,168–12,890	279	UAG	UAG
trnC	J	13,207–13,267	61			13,207–13,267	61
ND6	J	13,298–13,762	465	AUU	UAA	13,283–137,62	480	AUA	UAA
trnV	J	13,764–13,819	56			13,764–13,819	56
lrRNA	J	13,819–14,907	1087			13,820–14,902	1083
trnS1	J	14,903–14,965	63			14,903–14,965	63

Note: The modification of the annotation was bold to highlight.

). After modification, the overlapping regions between ND3 and trnL2, trnF and srRNA, srRNA and ATP8, trnV and lrRNA, and lrRNA and trnS1 were eliminated, and the intergenic spacers between COII and trnR, trnK and COIII, trnI and CYTB, trnW and ND1, ND2 and trnW, trnL2 and trnT and trnC and ND6 were reduced or disappeared, which recovered the compactness of this mitogenome.

The mitogenome of S. houjii is a single circular molecule 14,965 bp in length (GenBank accession number is PP697967), which contains 37 encoding genes (13 PCGs, 22 tRNAs and 2 rRNAs) and two putative control regions (CR1 and CR2). The minority strand (N-strand) encodes 3 protein-coding genes (ND4, ND4L and ND5) and three tRNAs (trnY, trnP and trnH), whereas the J-strand encodes the remaining 31 genes. This mitogenome contains a repeat region of 230 bp in length, which covered 144 bp of CR1 and 220 bp of CR2. The 76 bp repeats before CR1 are a part of the complete COIII genes but were joined to CR2 as a pseudogene (Figure 1). The motifs of the control region located at the repeat region include G(A)nT, polyT/polyA, A(T)n, stem-loop and (TA)n.

Compared with the ancestral arrangement, the gene order of the S. houjii mitogenome underwent extensive rearrangement (Figure A1). At least ten rearrangement events were proposed in the evolution of the S. houjii mitogenome, with eight genes (trnQ, ND1, trnF, srRNA, trnL1, trnC, trnV and lrRNA) inverted and only six relative conserved gene blocks (trnL2-COII, trnD-trnK, ND2-trnW, ATP8-ATP6, ND5-trnH-ND4-ND4 and trnV-lrRNA) remaining (Figure A1). Gene blocks COI-ND3, trnR-trnG, COIII-CR1-trnN-trnE-trnQ-trnA-trnS1-trnI-CYTB-trnY, ND1-trnM-trnF-srRNA, trnL1-trnT and trnC-ND6 were newly generated in the processes of mitochondrial gene rearrangement.

3.2. The Quantitative Transcription Map of Sericothrips houjii Mitogenome

We obtained a total of 3.32 Gb high-quality CCS (accuracy ≥ 0.9) with 1,144,294 transcripts from the raw data of the full-length transcriptome sequencing, and 1,005,191 draft transcripts totaling 2.18 Gb. In these draft transcripts, 37,040 transcripts (GenBank accession number is PRJNA1130360) were mapped to the complete mitogenome of S. houjii with a mean coverage of 4796.7× in depth. The quantitative transcription map showed that the transcripts of the N-strand (19,126) were more abundant than that of the J-strand (17,914) (Figure 2). We found that the relative expression levels of mitochondrial genes, ranking from highest to lowest, were ND4, ND4L, ND3, COI, srRNA, COIII, lrRNA, CYTB, ND1, ND5, ND2, ATP6, ATP8, COII and ND6.

3.3. Polycistronic RNA and Precursor RNA Cleavage

In addition to ATP8/ATP6 and ND4/ND4L maturing as bicistronic mRNA like other insects [12,15,17,18], another bicistronic mRNA COI/ND3 was recognized in the S. houjii mitogenome, which was firstly identified in the insect mitogenome. Excluding degraded transcripts, there were 584 sense polycistronic transcripts (except COI/ND3, ATP8/ATP6 and ND4/ND4L) mapped to the mitogenome of S. houjii, of which six transcripts were transcribed from J-strand can be divided into four types and 578 were transcribed from N-strand can be divided into 25 types (Figure 2). These polycistronic transcripts covered 10 PCGs (ATP8, ATP6, COI, COII, COIII, CYTB, ND1, ND2, ND3 and ND6), 2 rRNAs (srRNA and lrRNA) and 17 tRNAs (trnL2, trnR, trnD, trnK, trnN, trnE, trnQ, trnA, trnS1, trnI, trnW, trnM, trnF, trnT, trnC, trnV and trnS2). We found the polycistronic transcripts trnS2/COI/ND3, trnL2/COII, trnI/CYTB/astrnY/ND2, trnW/ND1, trnC/ND6 and trnV/lrRNA lack their downstream tRNAs in 3′ end, indicating that their primary transcript had the ability to remove tRNAs from 3′ end to 5′ end. However, the polycistronic transcripts COI/ND3/trnL2/COII, trnD/trnK/COIII/CR* (* means the 3′ end or 5′ end of the transcript was not aligned to the gene boundary) and trnK/COIII/CR* indicating the trnS2 and trnD were removed from 5′ end to 3′ end. By comparing the related mature mRNAs and rRNAs of the polycistronic transcripts, it was found that the cleavage site of polycistronic transcripts was the gene boundaries of tRNAs. After cleave the RNAs from polycistronic transcripts, the untranslated regions (UTRs) transcribed from the non-coding regions (NCRs) at the 5′ ends of COIII, CYTB, ND2, ND4L/ND4 and ND6 were removed to expose the start codon. However, the 3′ ends of the mRNAs COII, CYTB and ND5 were connected to the 5′ end of their downstream tRNA, wherein the 3′ UTR was retained. Beyond the situation, the 3′ UTRs of COIII transcripts were variable in length from 5 to 120 bp, and the polyadenylate sites were optional from nucleotide position 3872 to 3987.

3.4. Isoform RNA

We identified in total 153 mature transcripts of ND2 (ChrM: 5678–6659). Among them, 18 transcripts ended at position 6582 and 135 transcripts ended at position 6659, resulting in two isoforms of ND2 transcripts. The isoform1 of ND2 mRNA is 982 bp in length (ChrM: 5678–6659) and isoform2 is 899 bp (ChrM: 5678–6582). Excluding the degraded and polycistronic transcripts, 4023 transcripts were mapped to srRNA (ChrM: 7778–8500). The 5′ ends of these transcripts were mapped to position 7778 but the 3′ ends were mapped to 17 different positions (Figure 3). Considering the nucleotide error of PacBio Sequel sequencing and the inconsistency of polyA length at the 3′ end of the transcripts, we classified these transcripts into eight isoform types (isoform1-8) according to the nearby 3′ end. Each isoform type of srRNA was supported by multiple transcripts (Figure 3). For lrRNA (ChrM: 13,820–14,902), 397 mature transcripts were detected. The 5′ ends of these transcripts were mapped to position 13,820 or 13,942, but the 3′ ends were mapped to positions 14,902, 14,620 and 14,243, which is supported by 22 transcripts (isoform1, ChrM: 13,820–14,243), 62 transcripts (isoform2, ChrM: 13,820–14,620), 116 transcripts (isoform3, ChrM: 13,820–14,902), 64 transcripts (isoform4, ChrM: 13,942–14,620) and 133 transcripts (isoform5, ChrM: 13,942–14,902), respectively. In addition, the 3′ ends of three polycistronic transcripts trnW/ND1/trnM/trnF/srRNA* (see nos. 145884403, 159975201 and 19530688) were identical with isoform1 and isoform4 of srRNA and 32 bicistronic transcripts trnV/lrRNA* were identical with the isoforms of lrRNA. These isoforms are easy to distinguish from the degraded transcripts as multiple transcripts were identical in 5′ and 3′ end.

The 5′ ends of the degraded transcripts from ND2 were variable within genes, while the 3′ ends were identical in two isoforms, indicating that the ND2 was degraded from the 5′ end to 3′ end. However, the tricistronic transcripts of CYTB/astrnY/ND2* indicated the degradation from the 3′ end to 5′ end. In this study, we detected the 5′ ends of srRNA transcripts were variable in positions between 7778 (5′ end of srRNA) and 8186, and the 3′ ends of srRNA transcripts were variable in positions between 8013 (3′ end of isoform1) and 8500 (3′ end of isoform8), which indicates the srRNA were degraded from both ends. The lrRNA isoforms had two 5′ end positions (13,820 and 13,942) and three 3′ end positions (14,243, 14,620 and 14,902). The trapezoid formed by the lrRNA transcripts in the quantitative transcription map indicates a significantly exonucleolytic degradation from 5′ end to 3′ end (Figure 2). A small number of lrRNA transcripts mapped their 5′ end to position 13,820 or 13,942 but were variable in the 3′ end, which can be regarded as the exonucleolytic degradation from 3′ end to 5′ end. In addition, polycistronic transcripts of trnI/CYTB/astrnP/ND2*, CYTB/astrnP/ND2* and trnW/ND1/srRNA* and bicistronic transcripts of trnV/lrRNA* indicate that the cleavage of polycistronic transcripts and exonucleolytic degradation can proceed simultaneously.

3.5. The Proposed Model of Mitochondrial Transcription

The ncRNA transcribed from CR was considered the precursor of the transcription initial RNAs (tiRNAs) [35] and the polyA/polyT motifs in CR are the signal to initiate the transcription of mitogenome [36]. Three ncRNAs mapped to the J-strand of CRs and one ncRNA mapped to the N-strand of CR in S. houjii mitogenome contained the motifs, which was considered as regulating the transcription of mitogenome (Figure 1). So, we proposed both CRs contained the TISs of both strands. As the 5′ end of ncRNAs was located within the COIII when the 3′ end was mapped within CR1, whether the TIS was function in the transcription of N-strand remains unclear. The relative expressions from asATP6 to astrnN and from asCOIII to astrnC were significantly lower than their upstream region (from trnP to ND4L), indicating CRs may weakly function in the transcription initiation but function well in the termination of the J-strand. In addition, the 5′ end of polycistronic transcript aslrRNA*/astrnV/asND6/astrnC/ND4L/ND4 (see no. 29428453) and three antisense transcripts aslrRNA*/astrnV/asND6/astrnC (see nos. 108333364, 108333364 and 156437306) have the same 5′ end position, ChrM-13,864, which may be another TIS of the J-strand.

The 3′ end of 2175 transcripts of COIII, and the 3′ end of 10 tricistronic transcripts trnD/trnK/COIII and bicistronic transcripts trnK/COIII were mapped to the repeat region in CR1 (position 3887 to 3968, and position 3987), indicating CR1 contained a TTS of the N-strand. The 3′ end of the transcripts trnT/CR2* (see nos. 138282373, 28902972 and 59901598) were mapped to the repeat region in CR2 (position 9601, 9607, 9651 and 9673), indicating CR2 contained a TTS of the N-strand. Therefore, both CRs were functional in the termination of N-strand transcription. Similarly, all antisense transcripts of the gene block CYTB-trnI-trnS1-trnA-trnQ-trnR-trnN-CR1 were ended within the repeat region of CR1 and the transcript ND5*/trnP/CR2* ended within the repeat region of CR2, indicating both CRs were functional in the termination of J-strand transcription.

Based on the above analyses, we proposed the transcriptional model of the S. houjii mitogenome (Figure 4). Both the J-strand and N-strand contain at least two TISs and TTSs in the CRs, respectively. The J-strand has an additional TIS within aslrRNA to enhance the transcription of the gene block ND5-trnH-ND4-ND4L. The mitogenome of S. houjii was transcribed into a complete primary polycistronic transcript as CR1 was fully transcribed from N-strand (see no. 83820671) and CR2 was fully transcribed from both strands (see nos. 70518759 and 114557833). The antisense transcripts of the gene blocks trnR-trnG-trnD-trnK-COIII and srRNA-ATP8-ATP6 were not detected, which may be degraded quickly due to their non-functionality. Further research was needed to determine the presence and exact location of TISs and TTSs, such as 5′ and 3′ RACE, RT-PCR and so on.

3.6. Phylogenetic Analysis and Mitochondrial Gene Rearrangement of Sericothripinae

A total of 46 mitogenomes were used for phylogeny analyses, including 37 species from Thripidae and 9 outgroups from Aeolothripidae (

Table A1. List of samples included in phylogenetic analysis.

Family	Subfamily	Species	GenBank Accession Number
Aeolothripidae		Aeolothrips fasciatus	ON210959
		Aeolothrips indicus	MW899051
		Aeolothrips melaleucus	ON210970
		Aeolothrips sp.	ON210960
		Aeolothrips xinjiangensis	NC_063848
		Desmothrips sp.	KY751031
		Franklinothrips megalops	ON210963
		Franklinothrips stasseni	ON210964
		Franklinothrips vespiformis	MN072395
Thripidae	Panchaetothripinae	Opimothrips tubulatus	NC_069977
		Phibalothrips peringueyi	MW603839
		Rhipiphorothrips cruentatus	MN072396
		Selenothrips rubrocinctus	MT872374
	Dendrothripinae	Dendrothrips minowai	NC_037839
		Pseudodendrothrips mori	NC_050743
	Sericothripinae	Neohydatothrips samayunkur	MF991901
		Sericothrips houjii	PP697967
	Thripinae	Anaphothrips obscurus	NC_035510
		Anaphothrips sudanensis	ON210961
		Aptinothrips stylifer	OQ559124
		Arorathrips mexicanus	OP913452
		Bregmatothrips sinensis	OP913450
		Ctenothrips transeolineae	OP913440
		Echinothrips americanus	ON210962
		Ernothrips longitudinalis	OP913449
		Frankliniella intonsa	JQ917403
		Frankliniella occidentalis	NC_018370
		Frankliniella panamensi	NC_081011
		Frankliniella schultzei	MT872372
		Lefroyothrips lefroyi	NC_083274
		Megalurothrips distalis	OP913453
		Mycterothrips gongshanensis	NC_067744
		Mycterothrips nilgiriensis	MT872373
		Odontothrips loti	ON210965
		Odontothrips pentatrichopus	OP913451
		Odontothrips phaseoli	NC_084197
		Scirtothrips dorsalis	NC_025241
		Scirtothrips hansoni	OR044712
		Stenchaetothrips bicolor	OP913448
		Stenchaetothrips biformis	ON653412
		Stenchaetothrips minutus	OP913447
		Taeniothrips eucharii	OP913454
		Taeniothrips sp.	ON210966
		Taeniothrips tigris	MW751816
		Thrips palmi	NC_039437
		Thrips imaginis	AF335993

). The two phylogenetic trees based on two datasets (PCG_rRNA and PCG12_rRNA) and inferred from the PhyloBayes analysis showed the congruent relationships. As for the results, S. houjii is sister to N. samayunkur, and the monophyly of Sericothripinae was supported (Figure 5A). Sericothripinae is nested within Thripinae and sister to the group of (Echinothrips + Scirtothrips).

Then, we compared the gene arrangement between S. houjii and its related species, N. samayunkur (Figure 5B). Gene blocks of trnI-CYTB-trnY-ND2-trnW-ND1-trnM-trnF-srRNA-ATP8-ATP6-trnL1-trnT and ND5-trnH-ND4-ND4L were conserved in two mitogenomes, while the gene blocks of COI-ND3-trnL2-COII, trnD-trnR-trnG-trnK-COIII-trnN-trnE and ND6-lrRNA-trnS2 in N. samayunkur were divided by tRNA translocation in S. houjii. The rearrangement degree between these two mitogenomes was calculated in CREx [37], and the results showed that there were six tRNAs (trnQ, trnA, trnV, trnC, trnL2 and trnS1) translocated with a breakpoint distance of 15. Mitochondrial gene arrangement in Sericothripinae is diverse like in other thrips taxa.

4. Discussion

Mitochondrial gene rearrangement has been deemed to generate the duplication of genes or CR duplication, change in relative gene order, redundancy of pseudogenes and modification of intergenic spacers [38,39,40]. The mitogenome of Thysanoptera have been known for extensive gene rearrangement, a duplicatable control region and highly variable NCRs [19,41]. Due to the rapid evolution, the insertions and deletions of nucleotides may change the length of genes and intergenic spacers. As a result, transcript, which serve as information carriers of gene expression, can annotate the mitogenome more accurately than gene structure predication and homologous sequence alignment. Compare with the annotation according to MitoZ and homologous alignment, the annotations according to transcripts modified the overlaps and intergenic spacers of insect mitogenomes and recovered their compactness. The relative mitochondrial gene expression level was similar in the mitogenomes of E. fullo, C. chinensis and A. gifuensis. These species were found to have the highest expression level in lrRNA, and cytochrome c subunits constantly had higher expression levels than NADH-dehydrogenase [15,17,18]. However, the relative expression level was significantly different in S. houjii mitogenome. ND4-ND4L expression was highest, COI-ND3 and COIII expression was higher than the remaining NADH-dehydrogenase and rRNAs and the expression of COII was only higher than ND6. Extensive gene rearrangement and duplicated CR changed the relative distance between the TISs and genes, which may affect the relative efficiency of post-transcriptional cleavage that affects the relative expression level of genes. In addition, the duplicated sequences produced during gene rearrangement, such as gene duplication and random lose (TDRL), may remain as NCR between genes [19,42]. The NCRs of the S. houjii mitogenome have been transcribed as the UTRs of the transcripts, the 5′ UTRs that were removed may be related to the translation need an initiation codon, while the 3′ UTRs remain in the mature mRNA of COII, COIII, CYTB and ND5. It was suggested that the 3′ UTR of mRNA is related to the stability of mRNA by forming a secondary structure with a polyadenylate tail [43,44], which maybe another reason for the relative gene expression level change in the mitogenome with extensive gene rearrangement.

There are two theories regarding to the mechanism of isoform formation. One is exonucleolytic degradation, because the isoforms of the same gene have identical 5′ or 3′ ends, which has been thought to fit the exonucleolytic degradation with two orientations [17]. The other is transcription termination, because the corresponding polycistronic transcript have been detected, and there was no transcript mapped to the downstream of the 3′ end of the isoforms [18]. The rRNA isoforms of S. houjii were observed to degrade from both ends, and the endings of degraded transcripts were observed to be variable and located between the endings of the isoforms. Thus, the isoforms of the RNAs in the S. houjii mitogenome were more likely generated by exonucleolytic degradation from 3′ end to 5′ end or/and 5′ end eto 3′ end instead of the termination of transcription. Neither the 5′ end nor the 3′ end of the degraded mRNA of ND2 were located between the 3′ ends of the two isoforms. It seems that the isoform2 of ND2 was generated by transcription termination. But the degradation from 3′ end to 5′ end was observed in the tricistronic transcript of CYTB/astrnY/ND2. Although isoform2 of ND2 was probably generated by exonucleolytic degradation from 3′ end to 5′ end, but the exonucleolytic degradation occurred before the cleavage of CYTB/astrnY/ND2 and the 3′ ends of two isoforms are so close in distance that the downstream transcript is too short to sequence or degrade quickly.

The functionality of the truncated stop codon T or TA of PCGs are probably recovered by the post-transcriptional polyadenylation [45]. Although the 3′ end of the transcripts polyadenylated, the stop codon in isoform2 of ND2 is missing, indicating that the isoform2 of ND2 may not be translated accurately. The evolution rate of the mitochondrial gene is much higher than that of nuclear gene [46] and the mitogenome with extensive gene rearrangement shows a higher rate of nonsynonymous substitution [6,19], but the mitochondrial genes were still homologous and have the relatively conserved domains. Isoform was ubiquitous in the rRNA transcripts of the insect mitochondrial transcriptome, but the isoform type and the orientation of exonucleolytic degradation were different, and the 3′ ends of isoforms were mapped to the nonhomologous regions of genes and the length of the truncated RNAs was variable [15,17,18]. Therefore, the truncated RNAs may not be able to perform normal physiological functions.

Compared with the transcriptional models of the insect mitogenome represented by Drosophila, in which the both strands were almost entirely transcribed except the NCR downstream trnS2 on the N-strand [13,15,18], both strands of the S. houjii mitogenome were entirely transcribed, even the control regions. Both CRs were observed to have the initiation potential of mitogenome transcription, but the CR1 was doubtful as its initiative function needs to combine a part sequence of COIII. Similarly to the Drosophila mitochondrial genome, in which the TIS located at the ND6 of the Drosophila increases the expression of gene block trnP-trnT-ND4L-ND4-trnH-ND5-trnF [11,13], the J-strand TIS located within aslrRNA (position 13,864) of S. houjii increased the expression of the gene block ND4L-ND4-trnH-ND5-trnP. This indicates that the J-strand TIS is located upstream of the major template region, regardless of whether the flanking genes were rearranged. Although the A. gifuensis mitogenome possessed two CRs, both strands were proposed to have only one TIS. However, both strands of the S. houjii mitogenome were proposed to have at least two TISs. The TIS of the mitogenome seems primarily influenced by the duplication of the control region and the location of a major template region. Both CRs of S. houjii mitogenome contained the TTS of the J-strand and N-strand, suggesting that the CRs contain the mitochondrial transcription termination factor (mTTF) binding sites. But the mTTF bound to CRs would only serve as an attenuator of the transcription instead of a strict terminator because both CRs were transcribed entirely.

The mitochondrial post-transcriptional cleavage process followed the “tRNA punctuation” model [14,45] in S. houjii, which was similar to the patterns observed in other insects [12,15,17,18]. The cleavage processes of six transcripts, trnS2/COI, trnL2/COII, trnI/CYTB/astrnY/ND2, trnW/ND1, trnC/ND6 and trnV/lrRNA, followed the “reverse cleavage” model proposed in Drosophila [12]. While the cleavage processes of the transcripts COI/ND3/trnL2/COII, trnD/trnK/COIII/CR* and trnK/COIII/CR* followed the “forward cleavage” model, which was raised in E. fullo [15]. However, the cleavage patterns of mRNAs were slightly different. As the tRNA insertion caused by gene rearrangement, the mature bicistronic mRNA ND6/CYTB found in Bemisia tabaci and Maruca vitrata, and the tricistronic mRNA ATP8/ATP6/COIII found in M. vitrata [47,48] were cleaved to ND6, CYTB, ATP8/ATP6 and COIII in S. houjii. The monocistronic mRNAs COI and ND3 were matured as the bicistronic mRNA COI/ND3 in S. houjii since there was no tRNA insertion between them after gene rearrangement. Post-transcriptional cleavage processes of PCGs were affected by the tRNA inversion occurred downstream of the PCGs and insertion or removal between PCGs, rather than rearrangement of PCGs.

Sericothripinae includes three valid genera, Sericothrips, Hydatothrips and Neohydotothrips, of which a synapomorphic characteristic is the microtrichia rows dense in the abdominal tergite and sternite. In the phylogenetic tree, Sericothripinae is related to Scricothrips and Echinothrips [19,22], of which the microtrichia are also rows dense on abdominal tergites and sternites [49,50]. Based on this character, Sericothripini was recognized as a tribe together with Sericohripina and Scirtothripina (now placed in the Scricothrips genus-group) [49,51,52]. However, this character is also identified in other genra of Thripinae, such as the Thrips, Mycterothrips and Projectothrips, which are unrelated to Sericothrips, Hydatothrips and Neohydotothrips [19,21,23,52]. Microtrichia rows dense in abdominal tergite and sternite may not be the key characteristic for the tribe of Sericothripini or the subfamily of Sericothripinae. Moreover, limited mitogenomic resources and small sampling has hampered widely comparative studies for the phylogenetic status and taxonomic decision within Thripidae.

5. Conclusions

In summary, an insect mitogenome with extensive gene rearrangement is different in the transcriptional model, relative gene expression level and post-transcriptional cleavage process. The transcription and cleavage of the polycistronic transcript were not influenced when the location was interchanged among tRNAs or PCG, but the replicates of CR sequences will influence the transcription of the mitogenome, and the tRNA insertion or removal between PCGs and NCR in the 3′ end of PCG will influence the cleavage of polycistronic transcript. Our findings provide new insights into mitochondrial gene transcription, RNA processing and RNA degradation in insect mitogenomes with gene rearrangements and CR duplication.