Genome-wide identification and expression profile under abiotic stress of the barley non-specific lipid transfer protein gene family and its Qingke Orthologues

Plant non-specific lipid transfer proteins (nsLTPs), a group of small, basic ubiquitous proteins to participate in lipid transfer, cuticle formation and stress response, are involved in the regulation of plant growth and development. To date, although the nsLTP gene family of barley (Hordeum vulgare L.) has been preliminarily identified, it is still unclear in the recently completed genome database of barley and Qingke, and its transcriptional profiling under abiotic stress has not been elucidated as well. We identified 40 barley nsLTP (HvLTP) genes through a strict screening strategy based on the latest barley genome and 35 Qingke nsLTP (HtLTP) orthologues using blastp, and these LTP genes were divided into four types (1, 2, D and G). At the same time, a comprehensive analysis of the physical and chemical characteristics, homology alignment, conserved motifs, gene structure and evolution of HvLTPs and HtLTPs further supported their similar nsLTP characteristics and classification. The genomic location of HvLTPs and HtLTPs showed that these genes were unevenly distributed, and obvious HvLTP and HtLTP gene clusters were found on the 7 chromosomes including six pairs of tandem repeats and one pair of segment repeats in the barley genome, indicating that these genes may be co-evolutionary and co-regulated. A spatial expression analysis showed that most HvLTPs and HtLTPs had different tissue-specific expression patterns. Moreover, the upstream cis-element analysis of HvLTPs and HtLTPs showed that there were many different stress-related transcriptional regulatory elements, and the expression pattern of HvLTPs and HtLTPs under abiotic stress also indicated that numerous HvLTP and HtLTP genes were related to the abiotic stress response. Taken together, these results may be due to the differences in promoters rather than by genes themselves resulting in different expression patterns under abiotic stress. Due to a stringent screening and comprehensive analysis of the nsLTP gene family in barley and Qingke and its expression profile under abiotic stress, this study can be considered a useful source for the future studies of nsLTP genes in either barley or Qingke or for comparisons of different plant species.


Key message
• 40 barley nsLTP (HvLTP) and 35 Qingke Orthologues nsLTP (HtLTP) genes were identified through stringent screening using the new barley and Qingke genomes and divided into four types (1, 2, D and G) by comprehensive analysis.
• The HvLTPs and HtLTPs were unevenly distributed in barley and Qingke chromosomes and had obvious gene clusters.
• Combined with the analysis of upstream ciselements of HvLTPs and HtLTPs and their expression pattern under abiotic stress, it was found that numerous HvLTP and HtLTP genes may change their regulatory modes due to different upstream cis-elements and cause different abiotic stress responses.

Background
Plant lipid transfer protein (LTPs), named for their function that transfer phospholipids and fatty acids between cell membranes in vitro, and they are also known as non-specific LTPs (nsLTPs) because of the characteristic of non-specific binding to different lipids [1,2]. Plant nsLTPs are usually 6.5 to 10.5 kDa, although some nsLTPs are approximately 15 kDa, and their isoelectric points (pIs) are usually between 8.5 and 12 (occasionally less than 5). Therefore, nsLTPs are a group of small, basic and ubiquitous proteins, that have eight cysteine residue motifs (8CM, ECM) in a highly conserved backbone sequence, C-Xn-C-Xn-CC-Xn-CXCXn-C-Xn-C, and a high content of αhelices with a central hydrophobic cavity to bind lipids. Almost all nsLTPs carry an N-terminal signal peptide in their nascent polypeptides [3][4][5]. Plant nsLTPs are involved in multiple physiological functions, such as cuticular lipid transport, cutin synthesis, cell wall extension, pollen development, pollen tube growth and guidance, stigma and pollen adhesion, plant signalling, and seed maturation [1,[6][7][8]. In addition, some plant nsLTPs have been identified as related allergens in plant food and pollen [9][10][11]. The expression of some nsLTPs can be induced by biotic and abiotic stresses, including low or high temperature, drought, heavy metal exposure and disease [12][13][14][15][16][17][18]. In particular, many studies have shown that nsLTP genes are closely related to abiotic stress resistance in plant [9,12,13,19,20].
Barley (Hordeum vulgare L.) is the fourth largest crop in the world after wheat, rice, and maize, and it is one of the oldest food and feed crop in the world (http:// faostat.fao.org). It is an important crop in China that has high economic and food value, especially on the Qinghai-Tibet Plateau. Tibetan hulless barley (highland barley), called "Qingke" in Chinese and "nas" in Tibetan, is the principal cereal cultivated on the Qinghai-Tibetan Plateau for at least 3500 years. Due to its high soluble dietary β-glucan and arabinoxylan contents, it is beneficial to human health and has attracted considerable interest [21][22][23]. Karen Skriver cloned and analysed the gene structure of the LTP1 gene in barley [24]. The crystal structure of this gene, specifically in ligand binding preferences, was also studied [25]. Then, some other barley genes (LTP2, LTP3, and LTP4) have been cloned in succession [26]. Interestingly, the barley nsLTP gene (blt4) was also found in the cold stress response [27,28]. The latest physical, genetic and functional sequence assemblies of the barley and Qingke genomes were completed in 2012 and 2016 [29,30] and 2015, 2018, and 2020 [22,31,32], respectively, and they provided an important reference for future crop breeding, improvement, gene function and evolution research.
Barley has strong drought resistance, low-temperature tolerance, and salt and alkali capacity, and its environmental adaptation mechanism is complex [19,20]. Qingke is mainly cultivated on the Tibetan Plateau (>4000 m above sea level), which has high UV-B radiation, low temperatures and low barometric pressure, and thus may therefore have greater extreme environmental adaptability than cultivated barley from other regions [21,31]. Currently, the nsLTP genes and their multigene families have been studied in many plants, including Arabidopsis, carrot, rape (Brassica napus), broccoli (Brassica oleracea), tobacco (Nicotiana tabacum), sugar beet (Beta vulgaris), sesame (Sesamum indicum), tomato (Solanum lycopersicum), maize (Zea mays), sorghum (Sorghum vulgare), cotton (Gossypium hirsutum), rice (Oryza sativa) and wheat (Triticum aestivum) [1,4,9,13,17,[33][34][35][36][37][38][39]. For barley, although 70 barley nsLTP genes have been identified in 2012 barley genome [40], there are still many deficiencies that need further comprehensive research by using the genome database of barley recently completed in 2016. At the same time, the aim is to further understand the molecular mechanism underlying the stronger ability of Qingke to adapt to extreme environments than barley. Therefore, in this study, nsLTP gene family members were identified and a bioinformatics analysis of barley and Qingke was performed through strict screening based on the latest barley and Qingke genomes. Using qRT-PCR technology, the expression profile of this gene family under abiotic stress was also discussed. The results of this study lay a foundation for further studies on the biological and molecular functions of nsLTP genes in barley and Qingke.

Identification, sequence analysis and classification of HvLTPs and HtLTPs
To identify the entire collection of putative nonredundant nsLTP genes in the barley genome, an accurate search workflow of nsLTP identification and data mining was performed (Fig. 1). Initially, three search methods were used to identify barley nsLTP family members. First, 109 amino acid sequences were obtained after conducting a BLASTP analysis of the IPK Barley BLAST Server using previously reported nsLTP protein sequences of Arabidopsis (79), maize (63), cabbage (63) and rice (77) as queries (Table S1), and the redundancy was checked. Second, a total of 265 proteins with the conserved Tryp alpha amyl domain (Pfam domain PF00234) were retrieved by the HMM search method. Third, the 39 relevant barley nsLTP genes were downloaded by keyword searches from the NCBI, IPK and Phytozome databases. Then, the above results were merged, and the redundant sequences were examined and removed. Finally, all the barley deduced nsLTP protein sequences were downloaded from the NCBI and IPK databases and the presence of LTP domain cl07890 was verified by BatchWeb CD-Search and Pfam validation using the domain PF00234. After that initial identification step, a total of 160 putative nsLTP protein sequences were identified. Then, each of the deduced protein sequences was manually assessed through the analysis of the cysteine residue motifs (8CM), and 107 proteins lacking the Cys residues were omitted from the remaining set. In addition, 11 proteins lacking Nterminal signal sequences (NSS) were also excluded by NSS prediction, and 8 proteins with C-terminal glycosylphosphatidylinositol (GPI) anchors remained by GPI anchor signal prediction. Subsequently, 2 proline-rich proteins were also excluded, and no proteins similar to At2S1-At2S4 and RATI were found. Finally, we identified 40 nsLTPs in the whole barley genome and named them HvLTPs, and 35 nsLTPs were also identified in Qingke genome by local BLAST analysis with HvLTPs and named them HtLTPs (Table 1). Furthermore, to better understand the characteristics of the HvLTP and HtLTP proteins, we analysed the theoretical isoelectric point (pI) and molecular weight (MW) for all putative HvLTP and HtLTP proteins and summarized them in Table 1 13,238 Da and average theoretical pI of HtLTPs is 7.94, which demonstrates its small basic properties with subtle differences. We also analysed the instability index, aliphatic index and grand average of hydropathicity (GRAVY) of the HvLTPs and  HtLTPs. The results showed that most instability index of HvLTPs and HtLTPs were greater than 40, the most aliphatic indices were greater than 75.5 and the GRAVY value was greater than zero, which indicated that most HvLTPs and HtLTPs were unstable, aliphatic and hydrophobic proteins.
Based on the Mw of the mature proteins, plant nsLTPs can be classified into two main types: nsLTP1 (9 kDa) and nsLTP2 (7 kDa). Then, according to sequence similarity, Boutrot divided the nsLTPs into nine types (I, II, III, IV, V, VI, VII, VIII and IX) [4]. Recently, plant nsLTPs have been categorized into four major and several minor types (1, 2, C, D, E, F, G, H, J, K, X) by intron position, sequence identity and spacing between the cysteine residues in the 8CM, as well as post-translational modifications [41,42]. Compared with the classification proposed by Edstam,40 HvLTPs and 35 HtLTPs could be divided into four types, including 16/13(type 1), 5/5(type 2), 11/10(type D) and 8/7(type G) nsLTP genes ( Table  1). As shown in Table 1, the molecular weights of HvLTPs and HtLTPs are usually between 9206.81 Da and 19,981.15 Da, and the molecular weight of Type 2 is smaller than that of the other types, with an average of 9 kDa. The molecular weight of the type G nsLTPs is relatively large and mostly between 16 and 19 kDa, with an average of 18 kDa. In the type G nsLTPs, the transcripts encode not only a C-terminal signal sequence, but also an N-terminal sequence, thus leading to a post-translational modification which a glycosylphosphatidylinositol (GPI)-anchor is added to the protein to attach the protein to the exterior side of the plasma membrane [43], leading to a much higher molecular weight of the type G nsLTPs than other types.
The main characteristic of plant nsLTPs is the presence of 8CM motifs. To establish a specific 8CM consensus for each nsLTP type obtained, we conducted a multiple sequence alignment using the 8CMs from 40 HvLTPs and 35 HtLTPs. The amino acid sequence alignment of the 8CMs of HvLTPs and HtLTPs revealed a variable number of inter-cysteine amino acid residues. We also found that the alignment results are consistent with the classification results. For the CXC motif, most of the residues at the X position in type 1 nsLTP are only hydrophilic, while in type 2, D and G nsLTPs, the X position is usually occupied by hydrophobic residues (Fig. 2). These conserved hydrophobic or hydrophilic Phylogenetic analysis, conserved motifs, and gene structure of the HvLTP and HtLTP families To analyse the evolutionary relationship, a phylogenetic tree of 293 nsLTPs from four species, including the HvLTPs and HtLTPs with maize, rice and Arabidopsis, was constructed. Comparing the previous classification data with phylogenetic analysis, it was found that the previous classification and phylogenetic analysis have the same type of nsLTPs; that is, the four groups in the HvLTP and HtLTP classification were consistent with the 1, 2, D and G types of the nsLTPs of the other three species, except for sporadic interlaces in types D and G (Fig. 3). The members of types 1 and 2 formed specific clades, indicating that these genes share a common ancestor in major nsLTP types. Based on the distribution of predicted motifs, 40 HvLTPs and 35 HtLTPs were categorized into four distinct subfamilies, which was consistent with the classification from the phylogenetic analysis (Fig. 4) As a type of evolutionary relic, the intron-exon arrangement carries the imprint of gene family evolution. The gene structures of the HvLTPs and HtLTPs were also investigated (Fig. 5). Investigation of the HvLTP and HtLTP gene structures revealed a low diversity distribution of intronic regions amid the exonic sequences. 40 HvLTP and 35 HtLTP genes were predicted to be interrupted by 0-2 introns positioned −9 to 104 bp downstream of the codon encoding the eighth cysteine in 8CM (Table 1). Additionally, it was interesting to find a similar exon/intron pattern in each group. For instance, the HvLTP and HtLTP genes in type 2 lack introns while the type G contains 2 introns. Except for HvLTPd1, HvLTPd7/HtLTPd7, and HvLTPd9/HtLTPd9, no introns were detected in the coding regions of type D genes.

Chromosomal localization and gene duplication of HvLTPs and HtLTPs
After determining the genomic location information of 40 HvLTP and 35 HtLTP genes, the genomic location The tree divided the HvLTP and HtLTP proteins into 4 types represented within the tree. The left/right phylogenetic tree was constructed from the nsLTP protein sequence of barley/Qingke, maize, rice, and Arabidopsis thaliana. The phylogenetic tree was derived using the neighbourjoining (NJ) method with 1000 bootstrap replications in MEGA-X information belonging to the 2012 IPK genes was not found for one of them (MLOC_57612.1). The Chromosomal localization results showed that 39 HvLTP and 35 HtLTP genes were unevenly distributed on 7 chromosomes of barley and Qingke, and there were obvious HvLTP and HtLTP gene clusters. The maximum number of HvLTP and HtLTP genes was contained on chromosome 2 (11), and the minimum number (1) was shown on chromosome 6 ( Fig. 6).
Gene duplication is generally considered to be a major driving force in evolutionary innovation and leads to genomic complexity. In this study, six tandem repeats were identified in the barley genome (Table S2), which was consistent with obvious HvLTP gene clusters in the barley chromosome. Two significant clusters were found on chromosomes 2 and 3. In addition, one sister pair appeared to be generated from segmental duplication (Table S2, Figure S1A). Furthermore, we analysed the collinearity of HvLTPs with nsLTP genes in rice and wheat. Eleven out of 39 HvLTP genes had collinear genes with rice, while 22 HvLTP genes had syntenic members between barley and wheat ( Figure S1B).

Promoter and stress expression analysis of HvLTPs and HtLTPs
Plant nsLTPs display a complex tissue-specific and developmental expression patternand are mainly expressed in the tapetum, pericarp and epidermal cells of embryos, stems, leaves and roots [42][43][44]. According to on the RPKM values of each HvLTP gene published on the IPK website, an expression heatmap of 40 HvLTP genes in the 16 different tissues were mapped (Fig. 7A). The results showed that the HvLTPs had distinctly different expression patterns during different developmental stages as well as in different plant tissues. All HvLTPs were not expressed in INF1, and the expression level of HvLTPs in INF2 was also extremely low. Among the different HvLTP types, the relative expression levels of type 1 and type 2 genes were higher and more specific than those of type D and type G genes, except that the expression  HvLTP genes showed obvious tissue-specific expression patterns. For example, HvLTP1.2 is mainly expressed in LEA. HvLTP1.13 is specifically expressed in CAR15, and HvLTP2.4 and HvLTP2.5 are specifically highly expressed in CAR5 and CAR15, which indicates that these three genes might be involved in grain development. Given the high expression level of barley type 1 HvLTPs in the above transcriptome data, we simultaneously detected the tissue expression patterns of 10 type 1 barley and Qingke nsLTP genes in five different tissues by qRT-PCR (Fig. 7B). The results showed that the expression levels of type 1 LTP genes in barley and Qingke were very low in the roots, and the expression pattern of barley type 1 HvLTPs was basically consistent with that of transcriptome data. However, the expression patterns of 10 nsLTP genes were different in the barley and Qingke tissues. For example, the expression level of barley LTP1.4 was higher in the leaves, while that of Qingke LTP1.4 was lower in the leaves. The expression level of barley LTP1.11 was higher only in the seeds, and that of Qingke LTP1.11 was higher in the stems and flowers. An in silico analysis of the 1.5 kb upstream region (starting from the translation initiation site) of 35 HvLTP and HtLTP genes revealed various regulatory elements associated with development and abiotic or biotic stress signalling (Fig. 8.). In addition to the TATA-box and CAAT-box, the A-box is the most common type of ciselements in the HvLTP and HtLTP genes. In this study, we found that the cis-elements of the HvLTP and HtLTP genes included stress response elements (ARE, MBS, MYB, LTR, TC-rich motif, and DRE), hormone-related elements (ABRE, TGACG-element, CGTCA-motif, TGA-element, TCA-element, GARE-motif, P-box, TATC-box, AuxRR-core, ERE, and Wbox), indicating that HvLTP and HtLTP are involved in stress response and hormone signalling. Additionally, the regulatory element involved in light responsiveness also appears to be enriched in the HvLTP and HtLTP genes, including G-Box, ACE, Box 4, GT1-motif, Sp1, TCT-motif,   for MYB transcription factor to modulate plant tolerance to freezing and drought stress [45], suggesting that these HvLTPs and HtLTPs containing MBS or MYB elements may participate in abiotic stress signalling of MYB. The difference is that the number of hormone response elements and light responsive elements in Qingke was higher than in barley, and ABRE and G-box are the most important and different hormone and light-responsive regulatory elements in HvLTP and HtLTP genes, which may mean that Qingke may rely on the abscisic acid hormone pathway and G-box related light-response pathway to adapt to more severe plateau environment [39].
qRT-PCR was performed further to explore the expression patterns of 16 HvLTP and 7 HtLTP genes in the root and leaf tissue under abiotic stress, which means that barley seedlings were treated for 2 days with drought, cold and salt stress, and then recovered for 2 days (Fig. 9, Figure S2). Similar to spatiotemporal tissue expression of HvLTPs, the expression level of Type 1 genes is higher than that of other types in HvLTPs under abiotic stress; therefore, the expression of type 1 genes of HtLTP genes were also assessed by qPCR assay. The seven Type 1 HvLTP and HtLTP genes had significant responses in the leaves under cold, drought and salt stress, but no or low responses in the roots, which was consistent with the low expression in the root in the IPK database (Fig. 9). The expression level of all barley Type 1 HvLTP genes was decreased under cold stress, while four HtLTP genes were upregulated in Qingke. The 4 Type 1 HvLTP and HtLTP genes were upregulated in barley and Qingke under drought stress, but two of them were different. For example, the expression of HvLTP 1.7 was downregulated in barley, but HtLTP 1.7 was upregulated in Qingke; and the expression of HvLTP 1.3 was upregulated in barley, but HtLTP 1.3 was downregulated in Qingke. Only one Type 1 HvLTP gene was upregulated in barley under salt stress, while six Type 1 HtLTP genes were upregulated in Qingke. In general, the expression patterns of seven Type 1 HvLTP and HtLTP genes in barley and Qingke under abiotic stress were different. After abiotic stress was removed, the change trend of the type 1 gene was also different Unlike type 1, two genes (HvLTP2.1/HvLTP2.2) of type 2 responded significantly to drought, cold and salt stresses in the roots and HvLTP2.1 responded significantly to salt stress in the leaves. The two genes showed the same change trends in the roots under the removal of abiotic stresses basically. For example, the two genes were upregulated continuously under cold stress and after the removal of cold stress, but they were downregulated under drought stress and upregulated after the removal of drought stress; moreover, slightly different were observed under salt stress. HvLTP2.1 was upregulated continuously with and without salt stress, while the HvLTP2.2 gene changed in opposite directions, i.e., it was downregulated first and then upregulated. Type D genes were downregulated under cold and drought stress, while most type D genes were upregulated under salt stress in the roots. The HvLTPd2 and HvLTPd5 genes also showed significant effects and were upregulated in the leaves under drought and salt stress. After the removal of abiotic stress, the overall fluctuation of type D genes was the opposite, but the HvLTPd7 gene showed the same trend. In type G, the response of HvLTPg1 in the roots was upregulated significantly under cold stress, while the response in the leaves was upregulated significantly under drought and salt stress. HvLTPg2 was downregulated significantly in roots and leaves under the three abiotic stresses. Only HvLTPg5 was upregulated significantly in leaves under cold stress. In general, the type G genes were not affected due to low expression after removal of abiotic stress, and only some genes had the opposite changing trends, including the HvLTPg1 genes in roots and HvLTPg2 genes in roots and leaves under drought stress; and the HvLTPg5 gene in leaves under the three abiotic stresses ( Figure S2).

Discussion
Plant nsLTPs are a large transporter family composed of 79 members in Arabidopsis, 77 in rice, 63 in maize, 58 in sorghum, 63 in cabbage and 156 in wheat, all of which are classified as different types [4,17,38,39,46]. In 2018, the wheat nsLTP family was comprehensively analysed again, and 461 putative TaLTPs were identified from the whole wheat genome [47]. Edstam provided comprehensive information about the categorization of the nsLTP gene family based on phylogenetic clustering and facilitated further functional analysis [41]. In this study, 40 and 35 nsLTPs were identified in the barley and Qingke genomes and classified into 4 subfamilies (Type 1, 2, D and G) following Edstam's classification (Table 1, Fig. 2). In addition, the number and classification of HvLTPs were also compared with those in Arabidopsis, rice, maize and cabbage ( Table 2). The results showed that the total number of nsLTP gene families in barley and Qingke was less than that in other species. Moreover, type E is unique to dicotyledons, while HvLTPs and HtLTPs are also deficient in type C and type X (Table 2). Meanwhile, the proportion of nsLTPs in each subfamily indicated that type G seemed to have contracted in barley, but no expansion was observed. Therefore, the decrease in nsLTP gene number in barley may be due to the loss of individual types and the reduction of individual populations. In addition, this number may be due to the inhibition of gene expansion. Zhang (2019) showed that 70 barley nsLTP genes were identified by keyword searching in Phytozome v12.1.6 and divided into five types (1, 2, C, D, and G) [40], which indicates that previous research results of this barley nsLTP gene family expansion are different from ours. Compared with Zhang's research, our research collected more comprehensive data through a BLASTP search, HMM search and keyword search of the barley genomes, including in the IPK, Ensembl, NCBI and Phytozome databases, and obtained 160 potential HvLTP genes. Referring to previous screening methods [17], the screening method was also more stringent and finally identified 40 barley HvLTP genes; therefore, our results were more accurate, comprehensive and reliable.
The role of gene duplication in the origin of evolutionary novelty and complexity has long been recognized [48]. In our study, we found one segmental duplication and six tandem duplication events in the barley nsLTP gene family (Table S2; Figure S1). These results suggest that gene expansion is inhibited overall and the evolution of the barley nsLTP gene family not only involves gene retention but also gene loss and mutation. The retention and loss of genes may be associated with related functions during plant evolution [49]. In addition, these paralogous duplicated genes may retain some essential functions in subsequent evolution. For example, five pairs of tandem duplication genes (HvLTP1.1/ HvLTP1.2/HvLTP1.3, HvLTP1.5/HvLTP1.7/HvLTP1.8, HvLTP1.6/HvLTP1.9, HvLTP2.1/HvLTP2.2, HvLTPd2/ HvLTPd5/HvLTPd6) shared similar expression profiles. However, one pair of duplicated genes (HvLTP2.4/ HvLTP2.5) showed significant divergence in expression. At the same time, another pair of segmental duplication genes (HvLTP1.12/HvLTP1.13) also showed different expression patterns. The differential expression patterns of these duplicated genes in barley indicated that these genes may be functionalized after duplication events during the evolutionary process, leading to significant variation in gene regulation [50,51].
Sufficient evidences has been obtained to show that nsLTPs are involved in various types of stress resistance, including resistance to phytopathogens, freezing, drought and salt [42,52]. In our study, the stressdependent cis-elements in the promoter regions of the HvLTP and HtLTP genes were analysed. The results showed that the promoters of the HvLTP and HtLTP genes contained stress response elements (STRE, DRE, MBS, MYB, TC-rich repeats) and hormone-related elements (ARE, LTR, ABRE, ERE, TCA-element, TGAelement, TGACG-element, CGTCA-motif, and W box), indicating that the HvLTPs and HtLTPs are involved in the stress response. Meanwhile, the expression patterns of 17 HvLTP genes in response to abiotic stresses also demonstrated their correlation with abiotic stresses. In general, the response of these genes in the root was weaker than that in the leaf. Under different abiotic stresses, the response levels of different type members were inconsistent. For example, the response of HvLTP type 1 was significantly higher than that of the other subfamilies. The blt4 gene of barley is a low-temperature response gene with different responses to drought, pathogen attack, and abscisic acid (ABA) [28]. Previous studies have shown that blt4 belongs to the barley nsLTPs located on chromosome 3 and can act as a regulatory protein to stabilize plasma membrane activity and resist low-temperature injury [12]. Interestingly, blt4 genes belong to the barley nsLTP type 1, and some of them appear in tandem duplication events, such as blt4.3 (HvLTP1.8)/blt4.9 (HvLTP1.7) and blt4.1 (HvLTP1.6) ( Table S2, Fig. 6). The expression patterns of these tandem duplication genes were similar, indicating that these genes may retain the same function and coordinate the regulation of the stress response through the tandem repeat of gene clusters.
Different promoter cis-elements and their epigenetic changes have been reported to affect gene regulation, thereby resulting in different gene expression levels and further affecting adaptation to the environment, including altitude changes [53][54][55]. We also compared the nsLTP gene promoter and stress expression pattern between barley and Qingke and found that the nsLTP gene sequences between barley and Qingke were basically the same, but the promoter and stress expression pattern were different. Moreover, the expression of nsLTP gene was directly related to its promoter, which indicated that the stress expression pattern of nsLTP gene was changed by its promoter. Combined with the analysis of upstream cis-elements of HvLTPs and HtLTPs and their expression patterns under abiotic stress, it was found that a large number of HvLTP and HtLTP genes may change their regulatory modes due to different upstream ciselements and cause different abiotic stress responses. Generally, there is little difference between them in terms of the HvLTP and HtLTP genes themselves, including the gene structure, conserved motifs, phylogenetic analysis and classification. The main difference between them lies in its promoter regulatory elements, which may lead to the difference of expression patterns between HvLTP and HtLTP genes, thus leading to the adaptability of Qingke to extreme plateau climate.

Conclusions
In summary, 40 HvLTPs and 35 Qingke nsLTPs were identified in barley and Qingke in this study. A comprehensive study of HvLTPs and HtLTPs will reveal important features of the nsLTP gene family, such as gene structure, evolution, chromosome distribution, conserved motifs, segmental and tandem duplication, upstream cis-elements, and stress expression patterns. The study results could be considered a useful source for future nsLTP gene research in either barley or Qingke or for comparisons of different plant species. This study will help to provide a foundation for future research on the molecular mechanisms of barley and Qingke stress adaption.

Plant materials, growth conditions and abiotic stress treatment
The barley variety 'Morex' and Qingke variety 'Dulihuang' were used as the plant material in this study. The Then, the deduced protein sequences of candidate nsLTPs were manually examined to determine whether they harboured the 8CMs (C … C … CC … CXC … C … C), and proteins lacking the essential cysteine residues were removed. Subsequently, the proteins without NSSs (N-terminal signal sequence prediction, http://www.cbs.dtu.dk/ services/SignalP, checked by the PROTTER http://wlab. ethz.ch/protter/start/) were also removed, and the remaining C-terminal glycosylphosphatidylinositol (GPI) anchors (GPI anchor signal prediction, http://mendel. imp.ac.at/gpi/plant_server.html and http://psort.hgc.jp/ form.html) remained. Subsequently, putative prolinerich or hybrid proline-rich proteins, which are characterized by a high proportion of proline, histidine and glycine residues in the sequence located between the NSS and the 8CM, were excluded from further analyses. The protein sequences of At2S1-At2S4 and RAT1 were then BLAST-searched against the rest of the candidate nsLTP proteins to exclude possible inhibitors and cereal storage proteins. Proteins with more than 120 amino acids at maturity were also discarded, and the final remaining amino acid was identified as the target amino acid sequence, named HvLTPs (the nsLTPs in barley). The entire screening process was undertaken strictly according to Fig. 1. The Qingke (Tibetan hulless barlay) genome was downloaded from the NCBI (https://ftp.ncbi.nlm.nih. gov/genomes/genbank/plant/Hordeum_vulgare/all_ assembly_versions/GCA_004114815.1_Hulless_Barley_ ass.V2/), and putative Qingke nsLTPs Orthologues were identified using HvLTP sequences by local BLAST and named HtLTPs (nsLTPs in Qingke).

Multiple sequence alignment and classification
The amino acid sequences of the putative 40 HvLTPs and 35 HtLTPs were downloaded and multiple alignment of the 8CM part of these sequences was then conducted and manually edited using the DNAMAN program. Additionally, the amino acid sequences were submited to the online site Compute pI/Mw tool (http://web.expasy.org/compute_pi/) to calculate the isoelectric point and molecular weight. Sub-cellular localization of these genes was predicted by the Plant-mPLoc online service (http://www.csbio.sjtu.edu. cn/bioinf/plant-multi/). The three-dimensional structures of all putative HvLTPs and HtLTPs were also predicted by SWISS-MODEL (https://swissmodel. expasy.org).
The HvLTPs and HtLTPs can be divided into four major and several minor types according to sequence identity, spacing between the Cys residues and intron position in the 8CM, and the post-translational modifications based on the presence of a GPI modification site. In the second round of classification, the HvLTPs and HtLTPs were sorted based on the identity matrix calculated from the multiple sequence alignments [41].

Protein motif and gene structure analysis
The predicted barley and Qingke nsLTP protein sequences were submitted to the online MEME (http:// meme-suite.org/tools/meme) to identify 20 distinct conserved motifs in the nsLTPs. The following parameters were used: repetitions are arbitrary, maximum number of bases is 10, and optimal base widths are limited to between 6 and 50 residues.
The prediction analysis of gene structure was carried out using GSDS (http://gsds.cbi.pku.edu.cn/) using the DNA and cDNA sequences of each predictive HvLTP and HtLTP gene from the barley and Qingke genomes.

Chromosomal mapping and gene duplications
The chromosome location information of the nsLTPs was searched in the barley and Qingke genome databases and MapInspect software was used to generate chromosomal distribution images.
Gene duplication was investigated following the method described by Kong et al [56]. The MCScanx and Circos programs were used to retrieve and map the collinearity between different plant genomes.

Promoter analysis of HvLTPs and HtLTPs
To investigate the cis-elements in promoter sequences of barley and Qingke nsLTP genes, the upstream sequences (~1500 bp) of each identified HvLTP and HtLTP were retrieved from the barley and Qingke genomes using a Perl script. The PlantCARE website (http:// bioinformatics.psb.ugent.be/webtools/plantcare/ html/) was used to identify cis-elements in the promoters.

Tissue expression profile analysis of HvLTPs
The publicly available Barley RNA-Seq datasets were downloaded from the IPK database, and the value of fragments per kilobase of transcript per million fragments mapped (FPKM) of these genes was used visualize the heat map using HEMI software and then used to analyse the barley tissue expression profiles of the identified HvLTP genes.

RNA isolation and quantitative real-time PCR
Total RNA was extracted using the Plant RNA isolation kit (Takara, Shiga-ken, Japan) following the manufacturer's instructions. The RNA quality was checked using a 1.0% (w/v) agarose gel stained with ethidium bromide (EB), and the RNA samples were inspected for quality and quantity using a NanoDrop® spectrophotometer and gel imager analysis. First-strand cDNA was synthesized from DNase-treated RNA with a PrimerScript 1st Strand cDNA synthesis kit (TIANGEN, Beijing, China). HvLTP and HtLTP gene-specific primers were designed based on their coding sequences (CDSs) using an online tool in NCBI (https: //www.ncbi.nlm.nih.gov/tools/primerblast) and then synthesized commercially (Shenggong, Shanghai, China) (Table S3). qRT-PCR was performed with SYBR GREENI and the CFX96 Real-time System (Bio-Rad, France) by strictly following the manufacturer's instructions. The thermal cycling conditions were as follows: 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s. The relative transcription levels were calculated using the 2 −ΔΔ CT method, and three technical replicates were performed for each sample. Availability of data and materials All data generated or analysed during this study are included in this article and its additional files. The genome data of Hordeum vulgare was downloaded from https://webblast.ipk-gatersleben.de/downloads/barley_ pangenome/Morex/ and http://plants.ensembl.org/Hordeum_vulgare/Info/ Index [29,30], the genome data of Qingke were downloaded from https:// ftp.ncbi.nlm.nih.gov/genomes/genbank/plant/Hordeum_vulgare/all_ assembly_versions/GCA_004114815.1_Hulless_Barley_ass.V2/ and http:// www.ibgs.zju.edu.cn/ZJU_barleygenome.htm [22,31], All the RNA-Seq data are available at IPK from https://apex.ipk-gatersleben.de/apex/f?p=284:57.

Declarations
Ethics approval and consent to participate The plant materials (barley and Qingke) used in this article were obtained from the Academy of Agriculture and Forestry of Qinghai University. All plant materials were provided free of charge and maintained in accordance with international guidelines. This article did not contain any studies with human participants or animals and did not involve any endangered or protected species.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.