Identification and Analysis of the Expression of microRNAs during the Low-Temperature Dormancy Release of Tulipa thianschanica Seeds

Abstract

With versatile biological functions, microRNAs (miRNAs) participate in the regulation of post-transcriptional gene expression in plants. Tulipa thianschanica Regel is a key wild tulip resource of the Liliaceae; however, it reproduces poorly under natural conditions, and the spreading and expansion of its population rely on the release of its seeds. In this study, T. thianschanica seeds were subjected to stratification at 4 °C for three different durations: 1 d (physiological dormancy), 20 d (dormancy release), and 40 d (non-dormancy). Nine samples were selected, and miRNA databases were established and annotated. This research revealed information on 34 miRNA families, including 14 newly discovered families and 20 families that were previously known. TpmiRNA167 and TpmiRNA395 were observed to be upregulated. Conversely, TpmiRNA166, which targets the key enzymes involved in antioxidant reactions, was downregulated. Notably, PC-5p-84014 and TpmiRNA159 were found to target TpPYL and TpDELLA, respectively, which influenced the abscisic acid/gibberellin (ABA/GA) ratio. Additionally, TpmiRNA160 and TpmiRNA164 were found to participate in the indole-3-acetic acid signal transduction pathway and interact with ABA, thereby contributing to the release of dormancy. In conclusion, this study provides a comprehensive understanding of the mechanism used to release dormancy in T. thianschanica seeds at the molecular level.

1. Introduction

Tulipa thianschanica Regel is distributed in western Xinjiang and central China and showcases its striking colors, unique flower types, and strong resistance. T. thianschanica reproduces through bulbs or seeds in the wild [1,2]. Bulb propagation is advantageous because it expedites the time of cultivation, which ensures earlier returns on investment. However, this method has the following drawbacks: (1) it diminishes the population of bulbs in natural habitats and harms the environment and (2) the coefficient of bulb propagation is low because a 4 to 5 year old bulb yields only approximately 1–3 cormels. Consequently, seed propagation has become the preferable choice to artificially cultivate T. thianschanica on a large scale [3]. Unfortunately, T. thianschanica seeds display low rates of germination under natural conditions. Seeds released from mature capsules require an extended period to fully break morphological and physiological dormancy before germination [3]. Investigating the mechanism of germination of tulip seeds holds significant implications to accelerate the germination and breeding of wild tulip seeds.

MicroRNAs (miRNAs), a class of biosynthesized small RNAs that are involved in degrading target gene mRNAs or inhibiting mRNA translation [4,5], are widely studied and relatively conserved in plants. Seed development is a complex biological process that encompasses the development of embryos, endosperms, and seed coats and the accumulation of storage substances [6]. This process undergoes intricate genetic regulation at multiple levels with miRNAs playing a crucial role as a mode of genetic regulation. miRNAs govern the development, size, vitality, and activity of seeds, while also influencing their physicochemical properties and nutritional composition [7,8]. Studies have demonstrated the regulatory roles of various miRNAs, including miR160, miR159, miR417, miR395, miR402, miR165/166, miR164, miR167, miR156, miR172, and miR158, in the germination and dormancy of seeds [9,10,11]. An increase in the levels of miR156 in mature embryos leads to a decrease in the levels of SQUAMOSA Promoter-Binding Protein-Like (SPLs) and miR172, which effectively reduces the speed of the seed developmental process and maintains dormancy [9]. During seed imbibition, 12 miRNA families, namely, miR156, miR159, miR164, miR166, miR167, miR168, miR169, miR172, miR319, miR393, miR394, and miR397, have been reported to be down regulated to varying degrees, while four miRNA families (miR398, miR408, miR528, and miR529) are upregulated during seed germination [12]. miR156 and miR172 have also been associated with the transition from the vegetative to the reproductive period [13], which highlights their diverse functions. The identification of two abscisic acid (ABA)-sensitive mutants, absg1 and absg2, revealed that there is complex regulation between indole-3-acetic acid (IAA) and the small RNAs, with absg1 and absg2 identified as alleles of dcl1 and hen1, respectively; these mutants positively regulate the expression of ABA-responsive genes [14,15]. miR159 plays a crucial role in the regulation of seed germination by modulating gibberellic acid (GA) and ABA signals [11]. miR159 regulates GAMYB mRNAs during flower development, fertility, and seed germination [5]. The GAMYB protein acts as a positive regulator, while DELLA serves as a negative regulator in the GA signaling cascade [16,17]. Additionally, miR159 also regulates the transcription factors (TFs) MYB33 and MYB101, which act as positive regulators of ABA signals during seed germination and dormancy [5,11]. miR159 was upregulated in the rdr2, dcl2, dcl3, and dcl4 triple mutants. RDR2, DCL2, DCL3, and DCL4 are critical factors that are involved in the generation of small interfering RNAs (siRNAs), particularly during the biosynthetic pathway of heterochromatin siRNAs [18,19]. This indicates that in addition to miRNAs, different types of small RNAs also play a significant role in seed germination and dormancy.

Drawing on the characteristic stem-ring structure of pre-miRNAs, many researchers have developed multiple software programs to predict pre-miRNAs and mature miRNAs in plants. These studies primarily utilize genomic data combined with the sequencing information for small RNAs. The comprehensive collection of miRNA data is available in miRbase V23 [20], which is a database that encompasses miRNAs from both plant and animal species. Recently, a plant-specific miRNA database called the Plant MicroRNA Encyclopedia (PmiREN) has been introduced, which incorporates information on the miRNA of 88 plant species and more than 16,000 predicted miRNA loci (http://www.pmiren.com/, accessed on 16 December 2022). While high-throughput sequencing has been used to identify numerous miRNAs from various plant species that have been cataloged in miRBase22.1 (http://www.mirbase.org/, accessed on 14 March 2021), there are no reported miRNAs specific to T. thianschanica.

In this study, we used high-throughput sequencing to analyze nine small RNA sequencing libraries. Our objective was to classify miRNAs and their corresponding target genes during the release of T. thianschanica seeds from dormancy under low-temperature conditions. The aim was to predict the regulatory role of miRNAs in seed dormancy and advance our understanding of the mechanisms that underlie the seed dormancy of T. thianschanica. Furthermore, molecular methods were utilized to modulate germination and accelerate the breeding of tulip seeds.

2. Materials and Methods

2.1. Collection and Pre-Treatment of Seeds

The materials used in this experiment were obtained from the National Tulip Germplasm Bank, Liaoning Academy of Agricultural Sciences (Shenyang, China) (N 41°49′41.40″, E 123°32′39.66″). Mature T. thianschanica seeds were dried and stored at room temperature. T. thianschanica seeds (Figure 1) were subjected to stratification (according to Zhang, 2020) at 4 °C for 1 day (physiological dormancy [PD]), 20 days (dormancy release [DR]), and 40 days (non-dormancy [ND]) and then stored at −80 °C.

2.2. Total RNA Extraction and Establishment of the miRNA Library

The total RNA was extracted from T. thianschanica seeds using the CTAB method. The degree of RNA degradation and contamination was assessed using 1.0% agarose gel electrophoresis. The concentration and purity of RNA (A_260/280 and A_260/230, respectively) were determined using a Nanodrop 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). MiRNA sequencing libraries were established for nine samples, including PD-1, PD-2, PD-3, DR-1, DR-2, DR-3, ND-1, ND-2, and ND-3, using TruSeq Small RNA Sample Prep Kits (Illumina, San Diego, CA, USA). Subsequently, sequencing was performed on an Illumina HiSeq 2500 platform with a 50 bp/read length. The resulting high-quality data were compared with the GtRNAdb, Silva, Rfam, and Repbase databases. Non-coding RNAs, such as rRNA, tRNA, snRNA, and snoRNA, as well as repeat sequences, were filtered out. miRDP2-v1.1.4 software (San Diego, CA, USA) was utilized to compare the sequenced reads with the reference genome sequences of closely related species to obtain mapped reads. PD was the control group, and DR and ND were the treatment groups.

2.3. Identification and Functional Annotation of the Target Genes of miRNAs

BLAST was utilized to compare the predicted miRNA target genes with the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases to obtain functional annotations. Low-quality and spliced tags were removed to obtain high-quality clean tags and the corresponding clustering data. These data were then compared with the Rfam database to annotate non-coding RNAs, and degradation sites were identified using a Cleveland analysis with a significance threshold set at p < 0.05.

2.4. Analysis of the Differential Expression of miRNA

To identify differentially expressed miRNAs, the levels of expression of miRNA were normalized as tags per million (TPM), and differential expression was detected using IDEG6 with the criteria of |log2(FC)| ≥ 1; FDR ≤ 0.01. The fold change (FC) represents the ratio of the expression between two samples. The p value of the original hypothesis represents the probability of expressing indifference, which was corrected by Benjamini–Hochberg, and FDR was eventually selected as the key indicator to screen differentially expressed genes.

2.5. Verification of the Target Genes of miRNAs by qRT-PCR

To explore the correlation between miRNAs and their predicted target genes, six pairs of miRNA–mRNA were randomly selected from the sequencing database for a real-time quantitative reverse transcription PCR (qRT-PCR) analysis of the target genes using tulip Actin as the internal reference. Reference and candidate genes were measured simultaneously with each sample repeated three times, and gene expression was calculated using the 2^−ΔΔCt method. Specific primer sequences are shown in Table S1. The method of extracting the total RNA was the same as that described above. The RNA from T. thianschanica seeds was extracted after stratification for 1 day, 20 days, and 40 days to quantitatively measure the gene. The qRT-PCR reaction system, program, and reactants are shown in Tables S2 and S3.

2.6. Statistical Analysis

SPSS 24.0 (IBM, Inc., Armonk, NY, USA) was utilized for statistical analyses. The data were expressed as the mean ± standard deviation, and Duncan’s method was used to assess the significance and perform inter-group comparisons. A difference was considered significant when p < 0.05. GraphPad Prism 8.0.1 (GraphPad Software, La Jolla, CA, USA) was used to visualize the data.

3. Results and Analysis

3.1. miRNA Sequencing of Seeds at Different Stages of Low-Temperature Stratification

To investigate the regulatory mechanism of miRNAs during the germination of T. thianschanica seeds at different stages of stratification, the total RNA was collected and subjected to sequencing. Illumina sequencing technology was utilized, which resulted in the generation of 46,145,598 raw reads across nine samples (three for each of PD, DR, and ND). After the adapter sequences were removed along with low-quality reads and sequences outside the range of 18–30 nt, more than 60% of the original reads remained, which yielded 73,534,151 valid reads from the nine samples tested (

Table 1. Results of miRNA sequencing of the Tulipa thianschanica seed libraries.

Sample	Total Reads (%)	3ADT and Length Filter (%)	Junk Reads (%)	Rfam (%)	Repeats (%)	Valid Reads (%)
PD-1	17,035,681 (100)	3,287,565 (19.30)	53,503 (0.31)	1,425,279 (8.37)	8550 (0.05)	12,266,027 (72.00)
PD-2	13,530,443 (100)	3,063,073 (22.64)	38,381 (0.28)	1,346,965 (9.96)	7442 (0.06)	9,079,240 (67.10)
PD-3	15,579,474 (100)	3,112,538 (19.98)	49,475 (0.32)	1,286,262 (8.26)	8658 (0.06)	11,127,570 (71.42)
DR-1	14,669,850 (100)	5,914,655 (40.32)	18,357 (0.13)	2,270,576 (15.48)	45,245 (0.31)	6,456,516 (44.01)
DR-2	11,288,606 (100)	3,695,920 (32.74)	27,623 (0.24)	1,439,851 (12.75)	16,100 (0.14)	6,120,230 (54.22)
DR-3	13,881,039 (100)	4,534,585 (32.67)	36,418 (0.26)	1,700,172 (12.25)	15,765 (0.11)	7,604,768 (54.79)
ND-1	14,222,034 (100)	4,408,012 (30.99)	27,732 (0.19)	1,718,145 (12.08)	21,017 (0.15)	8,060,083 (56.67)
ND-2	12,884,793 (100)	4,592,940 (35.65)	18,327 (0.14)	1,563,487 (12.13)	26,889 (0.21)	6,701,697 (52.01)
ND-3	12,143,228 (100)	4,863,545 (40.05)	25,445 (0.21)	1,131,736 (9.32)	11,006 (0.09)	6,118,020 (50.38)

Note: Rfam, number of non-miRNA sequences annotated in the Rfam database; repeats, the number of non-miRNA sequences annotated in the Repbase database.

).

As shown in Figure 2, small RNA sequences that ranged from 18 to 25 nt in length were subjected to a statistical analysis. The most abundant small RNA sequence was 24 nt, followed by 20 nt and 22 nt. The frequency share of 24 nt decreased from PD to DR and then to ND, and the frequency share of 19 nt, 21 nt, and 22 nt increased from PD to ND. There was a lower proportion of 18 nt sequences than those for the other small RNA sequences, and their summed frequency share accounted for <40%. These patterns of length distribution closely resembled the typical distribution of small RNAs in angiosperms.

The tested miRNAs were simultaneously compared with ribosomal RNAs (rRNAs), transfer RNAs (tRNAs), small nuclear RNAs (snRNAs), and nucleolar RNAs (snoRNAs). The results revealed that only a cumulative small fraction of 8.5–16.0% of the miRNAs detected could be identified. During the release of T. thianschanica seeds from dormancy, more than 80% of miRNAs with unknown functions were not annotated, as shown in

Table 2. Numbers of reads classified according to the types of miRNAs identified.

Sample	rRNA	tRNA	snoRNA	snRNA	Rfam RNA
PD-1	813,906 (4.78%)	519,783 (3.05%)	11,783 (0.07%)	33,001 (0.19%)	46,806 (0.27%)
PD-2	765,727 (5.66%)	476,440 (3.52%)	12,430 (0.09%)	46,725 (0.35%)	45,643 (0.34%)
PD-3	683,463 (4.39%)	504,865 (3.24%)	12,422 (0.08%)	40,735 (0.26%)	44,777 (0.29%)
DR-1	1,267,257 (8.64%)	865,419 (5.90%)	18,345 (0.13%)	29,577 (0.20%)	89,978 (0.61%)
DR-2	887,866 (7.87%)	468,740 (4.15%)	6274 (0.06%)	28,979 (0.26%)	47,992 (0.43%)
DR-3	1,091,435 (7.86%)	501,606 (3.61%)	7150 (0.05%)	39,114 (0.28%)	60,867 (0.44%)
ND-1	879,050 (6.82%)	317,343 (2.61%)	4989 (0.04%)	26,628 (0.22%)	48,241 (0.40%)
ND-2	734,535 (6.05%)	571,427 (4.43%)	9242 (0.07%)	43,361 (0.34%)	60,407 (0.47%)
ND-3	1,061,289 (7.46%)	541,407 (3.81%)	8030 (0.06%)	41,243 (0.29%)	66,176 (0.47%)

Note: DR, dormancy release; ND, non-dormancy; PD, physiological dormancy; rRNA, ribosomal RNA; tRNA, transfer RNA; snRNA, small nuclear RNA; snoRNA, nucleolar RNA.

.

3.2. Identification of the Conserved miRNAs

Numerous miRNAs are relatively conserved among plants. As illustrated in Figure 3, a consensus-based approach identified 54 conserved miRNAs, which belonged to 18 families. The most abundant family was miRNA167, which consisted of seven members, followed by miRNA159, miRNA166, and miRNA171. Furthermore, most families, such as miRNA156, miR169, miRNA160, miRNA164, and miRNA403, only contained one or two members.

3.3. Identification of Novel miRNAs

To identify novel miRNAs from unlabeled small RNA labels, miREvo and miRDeep2 were used to predict the secondary structure and Dicer cutting sites and to measure the minimum free energy. As shown in

Table 3. Novel miRNAs identified in the Tulipa thianschanica seeds.

miRNA	miRNA Sequence	Length (nt)	CG %	dG	Level of Expression
PC-5p-12886_384	TGGACCAATCACAGCAAGAAT	21	35.40	−101.90	high
PC-3p-2448_3032	TATTCTTGCTGTGATTGGTCC	21	35.40	−101.90	high
PC-5p-29011_123	ACGGAACGCACCTCCGAACGCACC	24	61.00	−104.10	high
PC-3p-190777_6	TGCGTTTCGGAGGTGCGTTCC	21	61.00	−104.10	low
PC-5p-26456_140	ACCCCAAATCCCCCGAGAAGC	21	54.00	−48.10	high
PC-3p-212_24745	TTTTCGGGTGATTTGAGGTGG	21	54.00	−48.10	high
PC-5p-20023_207	GGAATGGTGTGATCGGTAAAT	21	44.00	−53.30	high
PC-5p-84014_23	GTTCCCTCCGGCACTTCACC	20	54.40	−51.70	high
PC-5p-80803_25	AAGCAAAGTGAACTAGAGC	19	46.30	−68.90	high
PC-3p-73564_29	TTGTATCTCACTTGTAGTCCT	21	41.90	−79.60	high
PC-3p-104391_16	CTCAATTGTAGACTTGACC	19	37.80	−28.70	high
PC-3p-95380_19	TTGTAGACCTAACCTTGTAGA	21	43.60	−101.30	high
PC-5p-54971_47	TGATAGTGATTGATGAAGCTC	21	34.30	−31.50	high
PC-5p-46859_61	ACCTCAAATCCCCCGAAAAGC	21	55.10	−66.30	high

Note: miRNA, microRNA.

, 14 novel miRNAs were predicted from nine samples. The length of new mature miRNAs ranged from 18 to 24 nt, and approximately 64.3% of the miRNAs were 21 nt. The rest were 19 nt, 20 nt, and 24 nt. There was >35% CG in all the novel miRNAs. Except for PC-3p-190777_6, the expression abundance of other miRNAs was relatively high. There were fewer newly discovered miRNAs (14) than conserved miRNAs (67), and these novel miRNAs could represent novel miRNA families or correspond to unknown members of known miRNA families.

3.4. Analysis of the Differentially Expressed miRNAs during the Release of Seed Dormancy

Based on the findings shown in Figure 4, 60 differentially expressed miRNAs (p < 0.05) were determined from nine samples throughout three sequencing periods. A total of 21 miRNAs were significantly differentially expressed, while 71 miRNAs reached a differential level of expression of 0.1. A screening criterion of 0.05 was used to observe that during the ND vs. PD process, the highest number of differentially expressed miRNAs was 24 with 14 upregulated and 10 downregulated. During the early stages of low-temperature stratification (PD-DR), 13 miRNAs were downregulated, whereas eight miRNAs were upregulated. As the dormancy gradually dissipated at the later stages of stratification, the number of differentially expressed miRNAs decreased. Notably, there were five downregulated miRNAs, which was precisely half of the number of 10 upregulated miRNAs.

Considering the characteristics of the expression of miRNAs, 21 miRNAs that exhibited different patterns of co-expression were selected for cluster analysis. As shown in Figure 5, the 21 differentially expressed miRNAs could be categorized into two groups. The first group was highly expressed in the PD stage, and their levels of expression gradually decreased with the progression of stratification stages. In this group, eight miRNAs showed no significant changes from DR to ND but increased in PD, and three miRNAs changed from higher, lower, to higher. The second group exhibited an increasing pattern of expression as the time of stratification advanced. There were three miRNAs that also changed from higher, lower, to higher, and seven miRNAs showed a trend of continuous increasing. These phased changes in expression signified the critical regulatory role of miRNAs during the process of seed germination.

3.5. Prediction and Functional Annotation of the Target Genes of the miRNAs

To further understand the important role of miRNAs in the release of T. thianschanica seeds from dormancy, the known sequenced miRNAs and their corresponding all target genes were analyzed by GO enrichment. As shown in Figure 6, 337 target genes were noted to be involved in biological processes (150, total 44.51%), cell components (101, total 29.97%), and molecular functional modules (86, total 25.52%) with significant differences in expression (p < 0.05).

As shown in

Table 4. GO functional annotation of the target genes of differentially expressed miRNAs.

GO Term	GO ID	Annotated	Significant	%
Biological Process
Primary shoot apical meristem specification	GO:0010072	5	3	60
Regulation of transcription, DNA-templated	GO:0006355	15	12	80
Response to deep water	GO:0030912	2	2	100
Stomatal complex morphogenesis	GO:0010103	2	2	100
Leaf development	GO:0048366	3	3	100
Polarity specification of adaxial/abaxial axis	GO:0009944	4	2	50
Lateral root development	GO:0048527	2	2	100
Transcription, DNA-templated	GO:0006351	9	7	77.8
Regulation of lignin biosynthetic process	GO:1901141	1	1	100
Meristem initiation	GO:0010014	3	1	33.3
Regulation of RNA biosynthetic process	GO:2001141	1	1	100
Determination of bilateral symmetry	GO:0009855	3	1	33.3
Cellular response to freezing	GO:0071497	1	1	100
Integument development	GO:0080060	3	1	33.3
Response to gibberellin	GO:0009739	2	2	100
Auxin-activated signaling pathway	GO:0009734	2	2	100
Cellular component
Histone deacetylase complex	GO:0000118	1	1	100
Mitochondrion	GO:0005739	9	5	55.6
Nuclear chromosome	GO:0000228	1	1	100
Nucleus	GO:0005634	20	8	40
Integral component of mitochondrial outer membrane	GO:0031307	1	1	100
Peroxisome	GO:0005739	8	2	25
Molecular function
Double-stranded DNA binding	GO:0003690	4	4	100
Sequence-specific DNA binding transcription factor activity	GO:0003700	9	7	77.8
Sigma factor activity	GO:0016987	1	1	100
Plastid sigma factor activity	GO:0001053	1	1	100
Carboxypeptidase activity	GO:0004180	1	1	100
Acetyl-CoA C-acetyltransferase activity	GO:0003985	1	1	100
Chorismate mutase activity	GO:0004106	2	1	50
L-ascorbate peroxidase activity	GO:0016688	2	1	50
Methyltransferase activity	GO:0008168	2	2	100
Calcium-transporting ATPase activity	GO:0005388	1	1	100
Caffeoyl-CoA O-methyltransferase activity	GO:0042409	1	1	100
DNA binding	GO:0003677	8	6	75
Signal transducer activity	GO:0004871	2	1	50
ATP binding	GO:0005524	4	2	50
Oxidoreductase activity	GO:0016491	1	1	100

Note: GO, Gene Ontology; miRNA, microRNA.

, further attention was paid to subclasses that are related to seed dormancy, including regulation of transcription (DNA-templated) in the BP components, transcription (DNA-templated), primary shoot apical meristem specification, meristem initiation, regulation of RNA biosynthetic processes, cellular response to freezing, response to gibberellin, and auxin-activated signaling pathway. The CC components primarily included the histone deacetylase complex, mitochondrion, nuclear chromosome, and peroxisome, which are involved in transport and aerobic metabolism. Of the MF components, the activities and transport functions of the enzymes related to energy metabolism have been largely annotated, including double-stranded DNA binding, ATP binding, carboxypeptidase activity, and acetyl-CoA C-acetyltransferase activity, among others, were annotated. There were some differences in the number of commented target genes in different subclasses with the proportion of significantly commented target genes ranging from 33.3 to 100%.

3.6. Analysis of the Expression of miRNAs Related to Dormancy and Their Target Genes

To further investigate the role of miRNAs in the seed dormancy of T. thianschanica, a comprehensive screening was conducted to identify 26 candidate miRNAs associated with the release of seed dormancy considering the annotation information of target genes (

Table 5. Target genes of the miRNAs that are related to the release of dormancy in Tulipa thianschanica seeds.

miRNA Family	Target Genes	Annotation
zma-miR156j-3p_L-1_2ss14	TRINITY_DN30618_c0	Uncharacterized protein LOC108987072
osa-miR156a	TRINITY_DN30823_c0	Gibberellin 20 oxidase 1-D-like
cas-miR159b-3p_R+1	TRINITY_DN42042_c1	Chorismate mutase 3, chloroplastic
aly-miR159c-3p_R+1_1ss20CT	TRINITY_DN42042_c1	Chorismate mutase 3, chloroplastic
aof- miR159f	TRINITY_DN46102_c0	DELLA protein DWARF8-like
bdi-miR162	TRINITY_DN41525_c2	Plant UBX domain-containing protein 7
aly-miR164a-5p	TRINITY_DN32775_c0	NAC domain-containing protein 21/22
	TRINITY_DN31804_c0	NAC domain-containing protein 21/22-like
	TRINITY_DN35987_c1	NAC domain-containing protein 100-like
zma-miR166m-5p	TRINITY_DN42896_c0	Uncharacterized protein LOC103700952
	TRINITY_DN42749_c0	Probable calcium-binding protein CML21
lus-miR166e	TRINITY_DN45178_c0	Homeobox-leucine zipper protein HOX32
	TRINITY_DN38326_c0	NDR1/HIN1-like protein 12
ata-miR166a-3p	TRINITY_DN45178_c0	Homeobox-leucine zipper protein HOX32
vca-miR167b-3p	TRINITY_DN27282_c0	Non-specific lipid-transfer protein-like
vca-miR168a-5p	TRINITY_DN45301_c1	Hypothetical protein POPTR_0012s03410g
osa-miR171d-5p	TRINITY_DN33637_c1	--
vvi-miR390	TRINITY_DN43406_c0	Serine/threonine-protein kinase BAM3
	TRINITY_DN39224_c0	LRR receptor-like serine/threonine-protein kinase
	TRINITY_DN46340_c0	-
	TRINITY_DN37038_c0	LRR receptor-like serine/threonine-protein kinase
osa-miR395b_L-1R+1	TRINITY_DN43730_c0	ATP sulfurylase 1, chloroplastic-like
	TRINITY_DN16935_c0	-
	TRINITY_DN40931_c0	Transcriptional corepressor LEUNIG isoform X4
aof-MIR395b-p5_2ss11AG20	TRINITY_DN46806_c1	Calcium-transporting ATPase 8
cme-miR396e_R+1	TRINITY_DN36128_c0	Growth-regulating factor 4-like
	TRINITY_DN34323_c1	Growth-regulating factor 1
	TRINITY_DN45792_c0	Growth-regulating factor 10-like isoform X2
	TRINITY_DN44678_c0	CSC1-like protein ERD4
	TRINITY_DN34323_c1	Growth-regulating factor 1-like
zma-miR166l-3p	TRINITY_DN45178_c0	Homeobox-leucine zipper protein HOX32
	TRINITY_DN38326_c0	NDR1/HIN1-like protein 12
gma-miR4995	TRINITY_DN37500_c0	Abscisic acid 8’-hydroxylase 1-like
osa-miR5083_R-1	TRINITY_DN41267_c1	Uncharacterized protein LOC105045431 isoform
	TRINITY_DN39322_c1	--
	TRINITY_DN44382_c0	--
	TRINITY_DN42545_c0	Uncharacterized protein LOC103973876 isoform
Novel miRNA
PC-3p-212_24745	TRINITY_DN36128_c0	Growth-regulating factor 4-like
PC-3p-2448_3032	TRINITY_DN41146_c2	--
	TRINITY_DN37385_c0	Uncharacterized protein LOC103721212
	TRINITY_DN22436_c0	--
	TRINITY_DN31474_c0	Transcription factor vib-1
PC-5p-20023_207	TRINITY_DN34607_c0	Uncharacterized protein LOC103979330 isoform
PC-5p-12886_384	TRINITY_DN41146_c2	--
PC-5p-29011_123	TRINITY_DN39092_c1	Ethylene-responsive transcription factor 4
PC-5p-84014_23	TRINITY_DN39500_c3	Abscisic acid receptor PYR1-like

). Among these candidates, 20 were conserved miRNAs, whereas 6 were newly annotated miRNAs, which corresponded to a total of 37 and 9 target genes, respectively. Remarkably, it was observed that certain miRNAs can simultaneously regulate multiple target genes. For example, aly-miR164a-5p was found to target three genes, while cme-miR396e_R+1 regulated five target genes. Interestingly, more than half of the miRNAs targeted a single gene. Additionally, miRNAs from the same family exhibited diverse functions by regulating different target genes. For example, cas-miR159b-3p_R+1 and aly-miR159c-3p_R+1_1ss20CT targeted CM1 (Chorismate mutase 1), and aof-miR159f targeted DELLA. Similarly, some miRNAs targeted the same gene, such as lus-miR166e and zma-miR166l-3p, suggesting that miRNAs from different families may share similar regulatory patterns or possibly represent functionally equivalent miRNAs. Similar findings were observed for the newly annotated miRNAs in this study, such as PC-3p-212_24745 and cme-miR396e_R+1, which both targeted ARF4.

To establish the correlation between the dynamic expression of miRNAs and their target genes, fluorescence quantitative experiments were conducted on six miRNAs that were related to hormone metabolism and their corresponding target genes. Figure 7 illustrates the patterns of expression of these miRNAs, which could be classified into two categories. The first category included Tp-mi156 and PC-5p-84014, which were continuously upregulated or downregulated. The second category comprised the remaining four miRNAs, which exhibited phased changes in their expression. Notably, Tp-mi156, Tp-mi159, Tp-mi164, Tp-mi4995, and PC-5p-84014 negatively correlated with their respective target genes TpGA20ox-like, TpDELLA8-like, TpNAC22-like, TpCYP707A1-like, and TpPYL1-like. This observation was consistent with the expected function of miRNA-mediated cleavage. However, PC-3p-212-24745 and its corresponding target gene TpARF4-like displayed similar patterns of expression, indicating a positive correlation. These findings suggest the possibility of a feedback regulatory mechanism between the miRNA and its target gene (Figure 8).

4. Discussion

4.1. miRNA Dynamic Patterns of Expression during the Dormancy Release of T. thianschanica Seeds

In most eukaryotic cells, miRNAs play a crucial role as a class of non-coding small RNAs involved in gene regulation. Extensive research has demonstrated their widespread involvement in various stress responses, such as drought, cold, and salt [21,22]. Over the past decade, significant progress has been made in understanding miRNAs in numerous plant species. However, the identification and mining of miRNAs using traditional methods, such as Sanger sequencing and bioinformatics predictions, are time-consuming, labor-intensive, and often unable to detect new miRNAs or those that are present at low levels. Advancements in high-throughput sequencing have revolutionized the field and enabled the exploration of novel or rare miRNAs and the analysis of small RNA clusters at different developmental stages under diverse natural environmental stress conditions [23,24]. High-throughput sequencing was first applied to the model species Arabidopsis thaliana, and it has since played a pivotal role in the discovery of miRNAs in various plant species, including Magnolia sieboldii, narrow-leafed lupin (Lupinus angustifolius), Ginkgo biloba L., and Chlorophytum borivilianum [25,26,27,28]. To date, 7474 hairpin sequences and 9168 mature miRNA sequences have been identified from 74 angiosperms. Despite the incorporation of these advances into miRBase 22.1, the sequence information for tulip miRNAs remains completely absent from the database.

Seed germination is initiated by the absorption of water, which causes the seeds to expand [29]. This results in the mobilization of nutrients within the seeds and subsequent cell elongation until the radicle breaks through the seed coat. The mobilization of stored substances involves a series of metabolic responses that can be modulated by alterations in gene expression and the utilization of different metabolic or signal transduction pathways under biological and abiotic stress conditions [30,31]. T. thianschanica seeds are physiologically dormant, and their germination requires exposure to low temperatures. This distinctive regulatory mechanism necessitates breaking dormancy under cold conditions before germination can be completed. In this study, nine small RNA libraries were constructed and analyzed to identify and predict the miRNAs and their target genes that are involved in the release of T. thianschanica seeds from dormancy under low-temperature conditions. A total of 67 conserved miRNAs were identified, which represented 34 families that encompassed the majority of miRNAs that have been reported in previous studies on other plant species. Additionally, 14 novel miRNAs were discovered. The predominant identified miRNA family was miRNA167, which was composed of seven members, followed by miRNA159, miRNA166, and miRNA171 with six, five, and five members, respectively. Several other families, including miRNA156, miRNA160, and miRNA164, were also identified. These families had mostly one or two validated members.

4.2. Regulatory Mechanism of the miRNA of T. thianschanica Seeds

The characteristics of T. thianschanica seeds, particularly the presence of dormancy, limit the successful establishment of seedlings, which thereby impedes the utilization of this wild species. Numerous studies have confirmed the significant regulatory role of miRNAs in the processes of seed dormancy and germination. To obtain a better understanding of the miRNA–mRNA regulatory relationship, we performed an miRNA–mRNA correlation analysis using the expression data of small RNAs obtained in this study and previously acquired transcriptome data by predicting sequencing data. In this experiment, six pairs of miRNA–target gene interactions were validated, and five of them exhibited a negative correlation in their patterns of expression, which was consistent with the silencing function of miRNAs. Only one pair showed a positive correlation, which could be attributed to the presence of a feedback regulatory mechanism between the target gene and miRNA or the combined effect of multiple miRNAs and target genes. Zhou [32] also reported a similar positive correlation between miRNAs and patterns of target gene expression in her study on the dormancy characteristics of a lily with fine leaves. Therefore, further research is required to confirm the intricacies of this regulatory mechanism.

The target genes ARF and miRNA396, which were identified in this study, have been extensively reported to be involved in auxin transduction [32]. Furthermore, the newly annotated miRNA PC-3p-212_24745 may have a similar functional role to that of miRNA396; furthermore, we will next verify this bold conjecture. Additionally, CYP707A1, a key gene that encodes ABA 8′-hydroxylase that is involved in the catabolism of ABA, could potentially be the target gene for miRNA4995 [33]. In A. thaliana, CYP707A that encodes ABA 8′-hydroxylase plays critical roles in various physiological processes [33]. Previous studies have demonstrated the importance of ABA 8′-hydroxylase genes, including CYP707A1 and CYP707A2 in A. thaliana, in regulating the endogenous levels of ABA during seed dormancy and germination [34,35]. In our study, seeds in the early stratification stage exhibited dormancy. Genes for the synthesis of ABA were upregulated along with miRNA4995, while CYP707A1 was downregulated. However, as dormancy was released owing to the continuous catabolism of ABA, the level of expression of miRNA4995 decreased, while that of CYP707A1 increased. Moreover, the prediction of this study identified PYL as the target gene for the newly predicted miRNA PC-5p-84014_23, which directly participates in ABA metabolism. Throughout the process, from dormancy to germination (PD to ND), PC-5p-84014_23 was consistently downregulated, whereas the corresponding target gene PYL was continuously upregulated. Previous studies have suggested that miRNA159 regulates the TFs that encode GA-MYB, which interacts with the GA response elements and influences the duration of flowering and anther development [36]. In this study, the expression of miRNA159 was inhibited in the absence of GA, indicating that the target gene DELLA directly interacts with miRNA159. Additionally, since DELLA is also targeted by GA-induced proteolysis, miRNA159 is closely associated with GA.

5. Conclusions

The miRNA databases for the PD, DR, and ND stages were constructed, and the corresponding raw reads were 46,145,598, 36,839,495, and 39,250,055, respectively. A total of 73,534,151 valid reads were obtained from the nine samples with 24 nt and 21 nt miRNAs identified as the predominant types. Furthermore, 34 miRNA families were identified and annotated. They encompassed functions that were related to seed germination, auxin signaling, and various aspects of signal transduction and endogenous metabolism. Among the miRNAs identified, TpmiRNA159, TpmiRNA156, TpmiRNA4995, and PC-5p-84014 target TpDELLA, TpGA20ox, TpCYP707A1, and TpPYL, respectively, thereby influencing the GA/ABA ratio. In particular, TpmiRNA4995 and TpCYP707A1 exhibit direct negative regulation with the level of expression of TpCYP707A1 increasing and the level of ABA decreasing during the late lamination period. Consequently, these molecular interactions contribute to the process of germination of T. thianschanica seeds.