Deciphering miRNAs involved in crosstalk between auxin and cold stress in Arabidopsis roots

Background: The phytohormone auxin and microRNA-mediated regulation of gene expressions are key regulators for plant growth and development at both optimal and under low-temperature stress conditions. However, the mechanistic link between microRNA and auxin in regulating plant cold stress response remains elusive. Results: To better understand the role of microRNA in the crosstalk between auxin and cold stress responses, we took advantage of the mutants of Arabidopsis thaliana with altered response to auxin transport and signal. Screening of the mutants for root growth recovery after cold stress at 4°C revealed that the auxin signaling mutant, solitary root 1 ( slr1; mutation in Aux/IAA14), shows a hypersensitive response to cold stress. Genome-wide expression analysis of miRNA in wild-type and slr1 mutant roots using next-generation sequencing revealed 180 known and 71 novel cold-responsive microRNAs. Cold stress also increased the abundance of 26 nt-31 nt small RNA population in slr1 compared with wild-type. Comparative analysis of microRNA expression shows signicant differential expression of 13 known and 7 novel miRNAs in slr1 at 4°C compared with wild-type. Target gene expression analysis of the members from one potential candidate miRNAs, miR169 revealed the possible involvement of miR169- NF-YA module in the auxin-mediated cold stress response. Conclusions: Taken together, these results indicate that SLR/IAA14, a transcriptional repressor of auxin signaling, plays a crucial role in integrating miRNA in auxin and cold responses. The present study provides a basic platform to explore the genetic and cellular mechanisms of auxin and miRNAs regulating the cold stress response. Identication of handful number of target miRNAs, including miR169 from the comparative RNA seq analyses between wild-type and auxin signaling mutant slr1 indicates that auxin regulated miRNAs play essential roles in maintaining the cellular signaling system to ensure an optimal cold stress response. Future overexpression or loss of function studies with the specic miRNAs that are altered in slr1 mutant and their targets will further clarify how the auxin and miRNA mediated pathways contribute in regulating the cold stress response.

processes. Long term cold stress results in coordinated down and upregulations of Aux/IAA and ARF protein family respectively [12], while short term cold stress affects the polar transport of auxin through modulating GNOM-mediated intracellular cycling of PIN2, a detrimental process for the functionality of PIN proteins [9,13].
It was also demonstrated that cold stress alters the auxin homeostasis and considerably increases the auxin level in the root meristem resulting in root growth inhibition. Restoring the auxin level through increased transport could restore the root growth even under cold stress [9,13]. Recently, the involvement of auxin maximum in the quiescent center (QC) has been shown to be an essential factor in preserving the root stem cells at a quiescent status under chilling stress [14]. The authors also demonstrated that this chilling stress-speci c sacri ce-for survival mechanism not only protects the stem cell niche from chilling stress but also improves the root's ability to withstand the accompanying environmental pressures and to recover when ambient temperatures rise to an optimal level [14]. In rice, 4°C cold stress resulted in 1.2-1.6 fold increase in IAA level [15]. Differential expression of three auxin e ux carrier genes, 9 ARF genes and 10 Aux/IAA genes were observed under cold stress [16]. The genome-wide analysis of the early auxin-responsive gene families in rice under cold stress revealed both up and downregulation of several genes of GH3, Aux⁄IAA, SAUR and ARF families [17]. OsGH3-2 overexpression-induced increase in cold tolerance was attributed to the combined effects of reduced free IAA content, alleviated oxidative damage, and decreased membrane permeability [18].
Additionally, it has also been shown that several known components of the cold signaling pathway are linked to auxin. For instance, SIZ1, a central regulatory component of cold response pathway, that stabilizes ICE1, is directly linked to auxin-mediated root architecture patterning [19,20]. Another downstream component of cold signaling pathway, AtNUP160, which plays a critical role in the nucleocytoplasmic transport of mRNAs under cold stress [21], has also been shown to play an essential role in auxin signaling [22]. Collectively, these results demonstrate the importance of auxin in regulating plants cold response. However, what remains obscure are the molecular components that integrate auxin and cold stress response. . For instance, overexpression of miR408, which targets cuproproteins belonging to the phytocyanin family and laccase, results in cold tolerance [33,34]. Consistently, miR408 knockout lines show a hypersensitive response to cold stress [34]. Overexpression of miR397, which targets laccases and a casein kinase beta subunit 3, also results in increased freezing tolerance after cold acclimation [35]. Overexpression of miR394a and LCR have demonstrated the positive role of this miRNA-target pair in response to low-temperature stress [36]. In rice, miR319 overexpression lines show an increased survival rate under cold stress [37,38]. In trifoliate orange, overexpression of the precursor of ptr-miR396b results in enhanced cold tolerance [39]. Collectively, these results suggest that miRNAs are potential regulators of cold stress response pathway across the plant species.
Reports on miRNAs have also demonstrated their strong potentials in modulating auxin signal transduction and several genes in auxin signaling have been reported as targets of miRNAs. For example, miR393 targets four closely related F-box genes, including the auxin receptor TIR1 [40,41]. miR393 also targets, a basic helix-loop-helix transcription factor from Arabidopsis that is homologous to GBOF-1 from tulip and annotated as an auxin-inducible gene [42]. Interestingly, Liu et al. (2017) showed that the heterologous expression of rice miR393a results in enhanced cold tolerance in switch grass (Panicum virgatum L.) [43]. Some auxin response factors (ARFs) have also been reported as a target of miRNAs [44,45]. ARF10, ARF16 and ARF17 are regulated by miR160 [46,47], while miR167 negatively regulates the expression of ARF2, ARF3, ARF4, ARF6 and ARF8 [48]. Few reports show that miR160 and the target ARFs are conserved between dicots and monocots [49,50]. Liu et al. (2012) suggested that miR167 is essential for the appropriate expression of at least four OsARFs that contribute to the normal growth and development of rice [51]. Additionally, Lv et al. (2010) reported the involvement of miR167 during cold stress in rice, which has also been shown to be involved in regulating auxin signaling through modulating auxin response factors in several plant species [32, [51][52][53][54]. Moreover, miR164 ne-tunes the auxin signals by targeting the NAC domain transcription factors [55]. Taken together, these ndings suggest that miRNA could be the potential link in integrating the auxin and cold stress responses.
In the present work, we tried to decipher the miRNAs that may regulate both auxin and cold stress responses by identifying auxin mutant that shows altered response to root growth after cold stress, followed by comparative analyses of genome-wide miRNAs in wild type and cold stress-sensitive auxin mutant by deep sequencing. Our results revealed that Aux/IAA14 mutant slr1 shows a hypersensitive response to cold-induced root growth inhibition. Comparative microRNA expression analysis displayed signi cant differential expression of 13 known and 7 novel miRNAs in slr1 during cold stress compared with wild-type. Interestingly, majority of the differentially expressed miRNAs were downregulated in slr1 in comparison to Col-0. The alteration of a signi cant number of miRNAs at 4°C in slr1 background suggests that SLR/IAA14 plays a crucial role in integrating cold and miRNA responses. Further, expression analysis of the target genes of one of the potential candidates, miR169 revealed that miR169-NF-YA module may play an important role in integrating auxin signaling, miRNA and cold stress.

Results
Auxin signaling mutant slr1 is susceptible to low temperature To decipher microRNA mediated auxin signaling during plant cold stress response, we rst focused on the identi cation of the cold-responsive auxin mutants. For the selection of cold-responsive auxin mutants, we used the root growth recovery assay developed in our lab earlier [9]. 12 h cold treatment at 4°C typically results in approximately 40-50% root elongation inhibition during the recovery phase at 23°C [9]. Hence, we selected 12 h incubation at 4°C as an optimal treatment for analyzing the effect of cold stress. Both auxin signaling mutants slr1, tir1, axr1-3, and afb2-1, and auxin transport mutants aux1-7, eir1-1, pin3-3 and pin4-3 were subjected to cold stress screening [56][57][58][59][60][61][62][63]. Percent root elongation recovery, compared to control (grown at 23°C) of each genotype was used as a measure of screening. 5-day-old seedlings grown at 23°C were transferred to new plates and kept at 4°C for 12 h and brought back at 23°C for recovery. The root elongation recovery was analyzed after 6 h and 24 h of recovery ( Fig. 1) [9].
Consistent with previous results, we also found that cold stress inhibits root growth recovery by approximately 50% in the wild-type after 6 h (Fig. 1A). Through root growth recovery screening, we could identify slr1, an Aux/IAA14 mutant as a potential candidate as slr1 showed slower root elongation recovery at both 6 h and 24 h time point (Fig. 1, Supplementary Fig. S1). The other mutants did not show any signi cant difference compared with wild-type for root elongation recovery except auxin transport mutant pin4-3, which shows a slight but statistically signi cant slower root recovery response at 24 h ( Fig. 1). To further con rm whether slr1 response to cold stress persists for a longer time, we measured the root growth recovery till 24 h, and found that slr1 shows slower root growth recovery at all-time points we tested ( Fig. 2A and 2B). Because the slr1 response to cold stress was consistent at all-time points, we selected slr1 for miRNA study.
High-throughput sequencing of small RNA libraries To identify the microRNAs that are responsive to auxin-mediated cold stress response, root tissues from 5 days old seedling grown on modi ed Hoagland media followed by treatment at 4°C for 12 h, and control grown at 23°C were used to construct small RNA libraries. We made eight small RNA libraries; four from wild-type (two each from 23°C and 4°C) and four from auxin mutant slr1 (two each from 23°C and 4°C different tags from slr1-23°C and 554,119 different tags from slr1-4°C respectively. The nucleotide length distribution in all the libraries showed that the majority of sequences ranged from 19 to 29 nt in size. We found 21 nt long small RNAs were the most abundant in all four libraries, followed by 24 nt. The total abundance of 2 types of small RNA population, i.e. 21 nt and 24 nt were 46.26% in Col-0 23°C, 56.44% in Col-0 4°C, 42.66% slr1-23°C and 30.43% in slr1-4°C respectively ( Table 2). The signi cant change in small RNA abundance in the cold stressed slr1 compared with wild-type suggests the importance of auxin signaling in miRNA modulated cold stress response. The populations of 26 nt-31 nt small RNAs were drastically affected in slr1 backgrounds by cold stress as an increase in 26 nt-31 nt were observed ( Table  2). We also found a signi cant decrease in the 21 nt small RNA population in slr1 under cold stress, while in Col-0 this was increased. These results further reinforce the idea that auxin directly regulates miRNA expression (Fig. 3, Table 2). We did not observe any signi cant changes in 24 nt reads for treatments and genotypes (Table 2, Fig. 3). The signi cant change in the major population of small RNAs in the slr1 mutant under cold stress con rms the involvement of auxin response in regulating miRNA function linked to cold stress response pathway, and also provides a possible explanation for the hypersensitive response of slr1 to cold stress.

Identi cation of known and novel miRNAs
To identify the number of known miRNAs in small RNA libraries, we used CLC genomics workbench v12.0 and aligned unique reads to known Arabidopsis miRNA sequences in miRBase v22.1, allowing up to 0 mismatches. A total of 180 known miRNAs, representing 70 families, were identi ed in eight small RNA libraries made from root tissue (Fig. 4, Fig. 5, Supplementary Table S1). In all the eight libraries, miR166 family was the most abundant, followed by miR165 and miR168 families. During lowtemperature stress, we found differential expression patterns of miRNAs in wild type and slr1 (Fig. 4, Supplementary Table S1). The comparative analysis between cold treated Col-0 and slr1 revealed that the expression of 13 miRNAs signi cantly changed in slr1 during cold stress (Table 3). We also observed altered expression of 10 miRNAs in slr1 background between control and cold treatment (Table 3). Interestingly, except for one, all the miRNAs expression was downregulated in slr1 mutant in comparison to Col-0.
For the identi cation of novel microRNAs, mirDeep-P pipeline was used. A total of 71 sequences were predicted to be potentially novel microRNAs from unannotated small RNAs (Supplementary Table S2 and S3, Supplementary Fig. S2). The abundance of novel miRNAs was lower compared to conserved miRNAs and their length varied from 19 nt to 30 nt. The comparative analysis of novel miRNAs between Col-0 and slr1 revealed that 7 of them signi cantly changed in slr1 under cold stress (Table 4, Supplementary Fig.  S2).

Validation of miRNA expression patterns
The consistency in microRNA expression identi ed by deep sequencing was validated using quantitative realtime-PCR. We selected 13 differentially expressed miRNAs that were regulated in response to cold in the root. The validation was performed using 10 known miRNAs (miR156, miR164b-3p, miR169a-5p, miR171-5p, miR390-5p, miR5642a, miR408-5p, miR398a-5p, miR472-3p and miR774a-5p) and 3 novel miRNAs (miR_Pred7, miR_Pred27 and miR_Pred37). In general, qRT-PCR validation results of miRNAs expression patterns were in agreement with the deep sequencing data, con rming the accuracy of the sequencing data (Fig. 6). Collectively, the qRT-PCR data con rm that the observed differences in miRNA expression in NGS are consistent and reproducible.

miRNA Target Prediction
The degree of sequence complementarity between miRNA and its binding site within the target determines the mode of action of miRNA. High sequence complementarity results in cleavage of targets [47, 64,65], while low sequence complementarity results in translational inhibition [66,67]. Several online resources such as psRNATarget use the same strategy to identify the plant miRNA targets. We carried out target prediction to understand the function of identi ed miRNAs by using psRNATarget server with preset values. The predicted targets for these miRNAs were from different classes of proteins associated with development, transport, auxin regulation, signaling and stress response (Fig. 7, Supplementary Table S1 and S2). For instance, cold stress response and signaling related proteins were targeted by miRNAs such as miR396b-3p, which targets MYB like transcription factors and miR156, which targets SQUAMOSA-PROMOTER BINDING-LIKE (SPL) transcription factor (Table 5). miRNAs like miR390a/b-5p targets the TASI-ARF, which is involved in auxin signaling. The possible roles of the target proteins in integrating auxin and cold stress responses have been discussed in detail in the discussion section. In general, we predict that the regulation of essential proteins contributing to cold stress tolerance in slr1 is possibly linked to the cold susceptible phenotype of the mutant.

miR169-NF-YA module is altered in slr1 under low-temperature stress
To understand the biological signi cance of the RNAseq results, we selected one of the potential miRNA candidates, miR169, which is evolutionarily conserved, reported to be present in various plant species including monocots, dicots, ferns and gymnosperms [68][69][70], and has been shown to be a central regulator of various abiotic stresses, including drought, salt, cold, heat, oxidative and hypoxia [71]. The miR169 family of Arabidopsis has 14 members, that matures into four types of different isoforms, differing only 1 or 2 nucleotides [71]. Phylogenetic analysis of miR169 revealed that apart from miR169a, b, c, and h, there are three obvious clades: clade I (mir169d, e, f, g), clade II (miR169i, k, m) and clade III (miR169j, l, n) [71]. The miR169 family members show distinct temporal and differential expression patterns and thus regulate diverse target genes [71,72]. One of the major targets of miR169 for abiotic stress response is NUCLEAR FACTOR Y (NF-YA), a heterotrimeric transcription factor composed of NF-YA, NF-YB and NF-YC proteins [73]. The link of the NF-Y family members in regulating plant developmental and stress response pathways has been demonstrated in several studies [73][74][75][76][77][78][79][80][81]. Earlier, a direct effect of temperature on miR169h and NF-YA was demonstrated [82]. miR169h abundance is directly in uenced by temperature; while at high temperature the abundance was high, at low temperature the abundance was considerably low. As expected, miR169 target gene NF-YA expression was reciprocal to the abundance of the miR169 expression [82]. They further demonstrated that NF-Y complex regulates the temperaturedependent owering and petiole length through directly binding to the promoters of owering regulator FT and the auxin biosynthesis gene YUC2. These results make an elegant model linking miR169, NF-Y and auxin. We tested whether a similar module works for cold stress response in slr1. For better clari cation of the role of the miR169 family, we selected at least one member from each clade, and miR169a, b and h. Beside miR169m, in cold stressed slr, all the tested miR169 members showed either a signi cant reduction in expression (miR169a, miR169d, miR169e, mir169h) or no changes in expression (miR169b, miR169g) (Fig. 8). The expression data suggest that among miR169 family members, miR169a, miR169d and miR169h function as major regulators to link auxin response and cold.
Next, we investigated whether cold stress-induced change in miR169 affects the NF-YA abundance reciprocally. In Col-0, under cold stress, we observed a decrease in NF-YAs abundance (8,15 and 22 fold decrease respectively for NF-YA3, NF-YA5 and NF-YA8 compared with 23°C, while in slr1, there was a higher accumulation of NF-YA transcripts compared with Col-0 (3.94, 4.3 and 4.4 fold increase for NF-YA3, NF-YA5 and NF-YA8 respectively, Fig. 8). Based on the above ndings, we speculate that the altered miR169-NF-YA module in slr1 is possibly contributing to its susceptible phenotype to cold stress. Further, miR169-NF-YA module might be playing a pivotal role during cold stress recovery, and miR169 regulates the expression of this module in an SLR dependent manner.

Discussion
Several studies indicate that auxin and microRNAs play essential roles in plant cold stress response [27,30,83]. Auxin plays a pivotal role in regulating temperature stress response, and high-temperature stress response directly affects the auxin biosynthesis through altering phytochrome interacting factors (PIFs) expression [84]. More recently, phytochromes have been shown to function as thermosensors in Arabidopsis [85,86]. Additionally, auxin transport and auxin signaling have also been shown to be altered in response to high temperature [10,87,88]. Under cold stress, polar and lateral auxin transports are altered, resulting in slower growth [1,9,89]. As for microRNAs, a large number of miRNA has been implicated in regulating the low-temperature response in different plant species [30, [90][91][92][93]. Although these studies represent a link 1) between auxin and cold stress, and 2) between miRNA and cold stress, the mechanistic link between microRNA and auxin in regulating plant cold stress response remains elusive. Here we report that 1) downstream auxin signaling response is crucial for miRNA mediated cold stress response in Arabidopsis root, 2) loss of auxin response resulted in altered expression of speci c miRNAs under cold stress, 3) the hypersensitive response of slr1 to cold stress is possibly linked to the differential expression of cold regulated miRNAs, and 4) miR169-NF-YA module possibly plays a major role in integrating auxin signaling, miRNA and cold stress.
Intracellular auxin response and miRNA-mediated gene expression tightly regulate the root growth developmental process in plants. For instance, root elongation is inhibited by the accumulation of auxin in the cell elongation zone [94]. Similarly, the timely regulation of expression of several genes by miRNA is indispensable for proper root growth [95,96]. Earlier it has been reported that the cold-induced inhibition of root growth is linked to the accumulation of auxin in the root meristematic zone, which results from the GNOM-regulated dysfunction of auxin e ux carrier, PIN2 [9]. In rice, genes linked to auxin signaling such as Aux/IAA and ARF were found to be altered by cold stress [16,17]. Consistently, in our current screening, we found slr1 showing hypersensitive response to cold treatment during root elongation recovery process. SLR encodes IAA14 protein, which functions as a repressor for auxin-induced gene expression, and a downstream auxin signaling component [56]. Mutation in slr1 results in stable IAA14 protein, which does not degrade in response to IAA, resulting in auxin insensitivity [97]. The nding that slr1 shows hypersensitive response to cold stress indicates the importance of downstream auxin signaling pathway in the process. miRNA modulates the plant development, plant response to various environmental challenges and auxin response by regulating the gene expression post-transcriptionally. From the beginning of plant miRNA study, its role in low temperature and other stresses has been reported [25, [98][99][100]. Consistently, in the present work, we also identi ed a genome-wide change in small RNA expression. Cold stress resulted in a de nite shift in the expression of small RNAs with longer nucleotide length as well as altered expression in several miRNAs (Fig. 3, Table 2). Several proteins help to regulate the multiple steps of miRNA biogenesis in optimal and stress conditions. The biogenesis of miRNA is precisely regulated during stress to regulate gene expression using the gene silencing machinery [101]. During the biogenesis, the pri-miRNAs sequentially get processed in miRNA-miRNA* duplexes by protein complex of DICER-LIKE 1 (DCL1), HYPONASTIC LEAVES1 (HYL1) and SERRATE (SE). These miRNA-miRNA* duplexes are then methylated, and ultimately the mature miRNAs incorporate with AGO1 and loaded to miRNA-induced silencing complexes (miRISCs) [102]. In Arabidopsis, one of the four Dicer-like (DCL1-4) proteins process the double stranded RNA into distinct size (18-21 nt long by DCL1, whereas DCL2, DCL3 and DCL4 produce 22 nt, 24 nt and 21 nt long, respectively) of small RNA duplexes [24].The abundance of 21 nt small RNA in all the libraries suggest that during the Arabidopsis miRNA biogenesis, the pri-and pre-miRNAs are mainly processed by DCL1 (Table 2). Till date, majority of the reported miRNAs are of 21 nt long in size followed by 24 nt (http://www.mirbase.org). It is also reported that even a single nt change in miRNA length affects its e ciency of target slicing [103]. The signi cant change in 21 nt small RNA abundance and an increase in nonspeci c 26nt-31nt small RNAs in the cold stressed slr1 compared with wild-type suggests that SLR might be playing a pivotal role in the biogenesis of miRNAs during cold stress (Fig. 3, Table 2). The SLR mediated auxin response could be regulating the 21 nt miRNA biogenesis under cold stress.
The ndings that at least 13 known miRNAs among 180 identi ed miRNAs were signi cantly regulated in slr1 under cold stress indicate the possible involvement of miRNAs in low-temperature stress response and recovery (Fig. 4, Table 3). Among the altered miRNAs, miR398, miR171, miR169, and miR396 have already been shown to be regulated by cold stress [34,39,104]. Few miRNAs that were found to be regulated by cold are also involved in auxin response viz. miR169, miR852 and miR390 [105,106].
Previous studies suggest that during stress conditions, an increase in miRNA expression deregulates the negative regulators of stress. In contrast, a decrease in miRNAs expression leads to the accumulation of positive regulators of stress [107,108]. Signi cant reduction in the expression of several miRNAs in the slr1 mutant compared to Col-0 suggests that there could be an increased activity of negative regulators in mutant leading to a susceptible phenotype ( Table 3). The observed differences in post-transcriptional regulation of NF-YA transcripts by miR169 in Col-0 and slr1 under cold stress supports this notion (Fig. 8). The miR169-NF-YA module, which is one of the stress-regulated miRNA target modules, showed altered expression pattern in slr1 under cold stress. Our results demonstrate that miR169a, miR169d and miR169h are the primary targets of cold in slr1 mutant (Fig. 8). This is consistent with the idea that the temporal and differential expressions of the members within the same microRNA family widely varies depending on the growth stages and acquired stresses [109]. miR169m showed a complete opposite expression pattern in slr1 under cold stress compared to other members. This is interesting, but at present the signi cance of this result is unknown. The reduced expression of miR169a, miR169d and miR169h directly in uenced the expression of NF-YA transcripts, resulting in higher accumulation in slr1 compared with Col-0 under cold stress (Fig. 8). miR169-NF-YA module has already been shown to be linked in the regulation of the temperature response in previous studies [82,110,111]. In addition, NF-YA has been shown to modulate auxin response by regulating its biosynthesis [82]. Taken together, these ndings suggest that the miR169-NF-YA module, which functions in a SLR/IAA14 dependent manner could function as a major regulator to integrate auxin, miRNA and cold stress response.
Beside miR169, several other microRNAs may be involved in regulating auxin-mediated cold stress response. Low-temperature stress also signi cantly decreased the expression of miR390 in slr1 mutant compared to Col-0. miR390 has been suggested to direct the production of tasiRNAs from Trans-acting siRNA3 (TAS3) transcripts, which regulate the ARF genes essential for auxin signaling [105,112]. Recently it has been demonstrated that the miR390/TAS3/ARFs module plays a key role in regulating lateral root development in salt-stressed poplar through modulating the auxin pathway [113]. Moreover, miR390b-3p targets AtVps11, an essential component for endosome organization, intracellular protein transport, vacuole biogenesis and pollen tube growth [114]. It also targets clathrin heavy chains which mediate endocytosis, intracellular transport, and are required for proper polar distribution of PINs [115]. Differentially regulated miR390 in wild-type and slr1 under low-temperature stress may also function in linking auxin and miRNAs during low-temperature stress response.
The signi cantly altered miRNAs were mostly downregulated in the mutant (Table 3, Supplementary  Table S1). The differential expression of microRNA under cold stress in wild-type and auxin signaling mutant slr1 indicates that the cold susceptible phenotype of the mutant is possibly linked to the miRNA targets. For instance, the members of miR156 family, which is highly induced by heat stress, were signi cantly downregulated in slr1 mutant background under cold stress, indicating their possible involvement in temperature responses [116]. miR156 also targets SPL transcription factor genes that play an essential function in Arabidopsis growth and development, including vegetative phase change, lateral root development, different plant organ and response to stress [117][118][119]. Another downregulated miRNA in slr1 mutant was miR396b-3p, whose expression was 2-3 fold more in the wild-type compared to the mutant. miR396b-3p targets the putative MYB transcription factors, which have been shown to regulate various developmental processes, including auxin homeostasis, biotic and abiotic stress responses [120][121][122]. Additionally, miR396a-5p targets growth-regulating factors that are involved in plant growth and development [123,124].
Another important miRNA that could be contributing to the susceptible phenotype of slr1 mutant is miR398, that positively regulates the heat tolerance in Arabidopsis [125]. Moreover, enhancement of freezing tolerance in Chrysanthemum dichrum by overexpression of ICE1 was also linked to the downregulation of miR398 [126]. Besides, miR398a-5p targets SETH2 protein involved in GPI anchor protein (GAP) biosynthetic pathway and pollen germination [127]. GAPs are found abundantly in the plasma membrane of plants and display similarities with plasma membrane receptors, peptides, and lipid transfer-like proteins [128,129]. Also, miR399, known to participate in phosphate homeostasis, is regulated by changes in ambient temperature [130,131]. Additionally, miR399b,c-3p targets the wallassociated kinase involved in cell surface receptor signaling, could also be involved in low-temperature stress signaling. Undoubtedly, during low-temperature stress, the membrane is a primary site of injury, and consistently the freezing stressed plants show higher electrolyte leakage [132]. To combat lowtemperature stress, cold-hardy plants modify their membrane as a necessary physiological process [133]. The possible downregulation of proteins required for membrane components by miRs in mutant may contribute as a signi cant factor for its susceptibility to cold.
The essential amino acid tryptophan (Trp) is well known for its requirement in auxin biosynthesis. The plant needs Trp for the synthesis of various proteins and many metabolites. Interference in the Trp biosynthesis leads to various developmental defects in plants [134]. The targeting of tryptophan synthetase by miRNA5642 during low temperature suggests a possible crosstalk of low temperature and auxin biosynthesis during the cold stress response. Moreover, the miR774a, which targets several F-box proteins, could also be attenuating the auxin signaling as several F-box proteins play indispensable roles in auxin signaling and response [135]. Taken together, these results con rm a complex triangular relation of miRNA, cold and auxin response.
Intriguingly, the differentially regulated miRNAs target a wide range of proteins involved in response to the stimulus, developmental processes, cellular component organization and biogenesis, biological regulation, cellular processes, metabolic processes and stress signaling (Supplementary Table S1, Fig 7). Although many of the proteins targeted by speci c miRNAs are still functionally not characterized, few of the target proteins have already been reported for their role in regulating various processes, including abiotic stresses such as cold [30,108]. Characterization of the other target proteins in this list will help to reveal the functional signi cance of these altered miRNAs in integrating auxin response and cold stress.
In agreement with the notion that often the miRNAs are expressed at a lower level, majority of the novel miRNAs displayed a lower expression compared to known miRNAs (Supplementary Table S2). As per miRBase 22.1, these miRNAs have not been described previously in Arabidopsis, which could be due to their low non-detectable expression level. In the present study, we possibly discovered most of the miRNAs in Arabidopsis during cold stress. The novel predicted miRNAs as well as known miRs viz. miR5656, miR774a and miR8181 that show altered expression patterns in wild-type and slr1 target the transposable elements (TE) and transposons (Table 5 and Supplementary Table S2). TEs can have a myriad of effects when they insert into new locations [136][137][138]. These effects vary depending on the sequence of the TEs and where precisely the TEs are inserted. TEs are also responsive and susceptible to environmental changes. Stress-activated TEs might generate the raw diversity that species require over evolutionary time to survive stressful situations [139]. From bacteria to mammals, TE-induced mutations are associated with environmental adaptations (For review, see E Casacuberta and J Gonzalez [139]. In the plants, TE-induced mutations result in adaptation to high altitude in soybean, adaptation to changing light environment in Arabidopsis and adaptation to a wide range of environments in wheat [104,[140][141][142][143]. Interestingly, TEs have also been shown to contribute to duplication of Aux/IAA genes in soybean [144]. The ndings that the cold stress stimulates miRNAs that potentially target TEs in slr1 suggest a possible involvement of TEs in integrating auxin and cold stress responses and need further studies.

Conclusions
The present study provides a basic platform to explore the genetic and cellular mechanisms of auxin and miRNAs regulating the cold stress response. Identi cation of handful number of target miRNAs, including miR169 from the comparative RNA seq analyses between wild-type and auxin signaling mutant slr1 indicates that auxin regulated miRNAs play essential roles in maintaining the cellular signaling system to ensure an optimal cold stress response. Future overexpression or loss of function studies with the speci c miRNAs that are altered in slr1 mutant and their targets will further clarify how the auxin and miRNA mediated pathways contribute in regulating the cold stress response.
Surface sterilized seeds were placed in round, 9-cm petri plates on modi ed Hoagland medium [1,9,145] containing 1 % (w/v) sucrose and 1 % (w/v) agar. The plates were kept at 4°C in the dark for 2 days for seed strati cation. After strati cation, the plates were transferred to the growth chamber (LPH-220S, NK System) at 23°C under continuous white uorescent light at an intensity of 100 μmol m -2 sec -1 and seedlings were grown vertically for 5 days.
Cold stress treatment and analysis of root growth recovery Cold stress treatment and growth recovery were performed as described earlier [9]. Brie y, 5-day-old seedlings were transferred to new plates containing the modi ed Hoagland medium and kept at 4°C for 12 h in the growth chamber (NK System; LH-1-120.S). After cold stress, plates were put back at 23°C for recovery, whereas the control plates were kept continuously at 23°C. Primary roots were analyzed after 6 h and 24 h of recovery. The experiments were repeated at least three times, with eight seedlings per treatment. To measure cold stress recovery, seedlings were photographed by a digital camera (Canon Power Shot A640) and root growth recovery was analyzed by ImageJ software (http://rsbweb.nih.gov/ij/).

Chemicals
Difco Bacto Agar TM was purchased from BD Biosciences, Japan. Other chemicals were from Wako Pure Chemical Industries, Japan.

Small RNA isolation and sequencing
The total RNA was isolated from the roots of cold stress treated and control samples by using the RNeasy Kit (Qiagen, USA) according to the manufacturer's instructions. Eppendorf BioPhotometer plus (Germany) was used to detect the quality and the concentration of RNA. Construction of the sRNA libraries and deep sequencing were carried out by BGI (Beijing, China). Brie y, RNA with lengths of  nt was separated and puri ed using denaturing polyacrylamide gel electrophoresis, followed by sequential 3' and 5' RNA adaptor ligation to the small RNAs using T4 RNA Ligase. The adaptor-ligated samples were then reverse transcribed and ampli ed by PCR to construct the nal libraries. Then, the prepared libraries were sequenced using an Illumina HiSeq 4000 platform (Illumina, USA).

Data Deposition Information
The sequencing data that support the ndings of this study have been deposited in NCBI SRA database with the SRA accession code: PRJNA579274. The SRA record is accessible with the following link: https://www.ncbi.nlm.nih.gov/sra/PRJNA579274 Bioinformatic analysis of the sRNAs sequencing data After sequencing, the raw reads were ltered and adapter sequences were removed along with contamination and low-quality reads from raw reads. The remaining unique sequences (clean reads) were then processed to identify known and novel microRNAs.
Known miRNAs from Arabidopsis root after cold stress were identi ed using CLC Genomics Workbench v12.0 (CLC Bio, Denmark). Brie y, clean reads processed from raw sequencing reads after trimming adaptor sequences and removing low-quality reads were further analyzed by CLC workbench to extract and group sRNA. Sequences shorter than 16 nt and larger than 36 nt along with non-coding RNA such as rRNA, tRNA, and snRNA were excluded from further analysis. The remaining reads were and then annotated to identify the known Arabidopsis miRNAs. To identify known miRNAs, small RNA sequences were annotated against miRBase 22.1 (http://www.mirbase.org/index.shtml) using CLC Genomics Workbench 12.0 based on their sequence homology. Finally, the mapped miRNAs were obtained, which then was normalized using the reads per million reads (RPM) method. Normalized reads were then used to determine the fold change between the control and stressed samples.
Novel miRNAs were identi ed by using mirDeep-P, a plant-speci c miRNA identi cation pipeline [146].
Brie y, for each sequenced small RNA library, reads were ltered by length and only those between 16 nt and 36 nt were retained. FASTA-formatted reads were then analyzed by miRDeep-P using the Arabidopsis genome as a reference.

Prediction of miRNA targets
The targets of identi ed miRNAs were predicted using psRNATarget (the plant small RNA target server, 2011 Release; http://plantgrn.noble.org/psRNATarget/) by aligning with Arabidopsis transcripts and default parameters which included a threshold cut-off of 3.0, a complementarity scoring length of 20 bp, and the energy required for target accessibility equal to 25 kcal/mole.

Quantitative RT-PCR validation of selected deferentially expressed miRNAs
Analysis of miRNAs expression was performed using the poly(T) adaptor RT-PCR method by Mir-X miRNA First-Strand Synthesis kit (Clonetech, Takara Bio, USA) as per manufacturers instruction. Brie y, for polyadenylation and cDNA synthesis 1 μg of DNaseI treated total RNA was incubated at 37°C for 60 min in a 10 μl reaction volume containing mRQ enzyme then the reaction was terminated at 85°C for 5 min to inactivate the enzymes. Quantitative RT-PCR (qRT-PCR), was run on a TaKaRa Dice Real Time apparatus (Takara, Japan) with the SYBR Green I Master kit (Bio-Rad, USA). The reaction conditions for qRT-PCR included following steps: 10 sec at 95°C followed by 40 cycles of denaturation for 10 s at 95°C and annealing for 20 s at 60°C, and extension for 15 s at 72°C. miRNAs were quanti ed using speci c primer pairs with the translation initiation factor elongation factor 1-α (EF1α) as the normalization controls.
Relative transcript abundance was calculated using 2 -ΔΔC T method [147]. All experiments were performed using three biological replicates and three technical replicates. The primers used in the study are listed in the supplementary Table S4.

Statistical Analysis
Results are expressed as the means ± SE from the appropriate number of experiments. A two-tailed Student's t-test was used to analyze statistical signi cance. Availability of data and materials All the data and materials that are required to reproduce these ndings can be shared by contacting the corresponding author. The sequencing data can be found in NCBI SRA database with the SRA accession code: PRJNA579274.
Five-day old Arabidopsis seedlings were transferred to new agar plates and kept at 4°C for 12 h. Plates were brought back to 23°C for recovery after cold stress. Percent root recovery compared to control was analyzed from primary root growth. Vertical bars represent mean ±S.E. Data are from at least three independent experiments (n=3 or more) with 8-10 seedlings per treatment. Asterisks represent the statistical signi cance between control and treatment as judged by the Student's t-test (* P < 0.05 and *** P < 0.001).

Figure 2
Auxin mutant slr1 shows a hypersensitive response to cold stress A) Five-day old Arabidopsis seedlings were transferred to new agar plates and kept at 4°C for 12 h. Plates were brought back to 23°C for recovery after cold stress. Root growth was analyzed from primary root growth elongation 6 h, 12 h and 24 h of recovery. Vertical bars represent mean ± SE. Data are from at least three independent experiments (n=3 or more) with 8-10 seedlings per treatment. slr1 root growth recovery at 4°C was statistically signi cant at all-time points (*** P < 0.001) as judged by the Student's t-test. B) Root phenotype of Col-0 and slr1 during 24 h recovery period after cold stress at 4°C for 12 h. Tick marks indicate the starting point of the recovery at 23°C. Scale bar = 10mm.

Figure 3
Nucleotide length distribution in Col-0 and slr1 Nucleotide length distribution of small RNAs libraries made from ve-day old control (23°C) and 12 h cold treated (4°C) Col-0 and slr1 roots. The results are obtained from two independent biological replicates. Vertical bars represent mean ± SD. Asterisks represent the statistical signi cance between the treatments as judged by the Student's t-test (* P < 0.05, ** P < 0.01 and *** P < 0.001).   Validation of miRNAs expression in response to cold stress A) RT-qPCR validation of selected miRNAs from cold stressed NGS library. RNA was isolated from the roots of 5-day-old 12 h cold (4°C) stressed seedlings. microRNA expression is expressed in fold change of expression against ef1α calculated by 2−∆∆CT. Vertical bars represent the mean ± SE of three biological replicates. Asterisks represent the statistical signi cance between control and treatment as judged by the Student's t-test (* P < 0.05, ** P < 0.01 and *** P < 0.001).  seedlings. The fold change of expression is calculated by 2−∆∆CT against 23°C using ef1α as internal control. Vertical bars represent the mean ± SE of two biological replicates. Asterisks denote the statistical signi cance between the Col-0 and slr1 at 4°C as judged by the Student's t-test (*** P < 0.001).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.