Comparative analysis identifies micro‐RNA associated with nutrient homeostasis, development and stress response in Arabidopsis thaliana upon high Zn and metal hyperaccumulator Arabidopsis halleri

Abstract miRNAs have been found to be key players in mineral homeostasis, both in the control of nutrient balance and in the response to toxic trace elements. However, the effect of Zn excess on miRNAs has not been elucidated; moreover, no data are present regarding miRNAs in hyperaccumulator species, where metal homeostasis is tightly regulated. Therefore, expression levels of mature miRNAs were measured by RNA‐Seq in Zn‐sensitive Arabidopsis thaliana grown in control conditions and upon high Zn, in soil and in Zn‐hyperaccumulator Arabidopsis halleri grown in control conditions. Differential expression of notable miRNAs and their targets was confirmed by real‐time RT‐PCR. The comparison in A. thaliana revealed a small subset modulated upon Zn treatment that is associated with stress response and nutrient homeostasis. On the other hand, a more consistent group of miRNAs was differentially expressed in A. halleri compared with A. thaliana, reflecting inherent differences in nutritional requirements and response to stresses and plant growth and development. Overall, these results confirm the involvement of miRNAs in Zn homeostasis and support the hypothesis of distinct regulatory pathways in hyperaccumulator species.

responses. In particular, they are key players in determining phenotypic plasticity in response to environmental stimuli, such as light, temperature and nutrient availability. Regarding stresses, they are involved in plant immunity toward a variety of pathogens, as well as in response to different abiotic stresses (Song et al. 2019). Indeed, modulation of large sets of miRNAs during abiotic stress response has been demonstrated for extreme temperatures (Chen et al. 2012;Yu et al. 2012;Zeng et al. 2018;Zhang et al. 2014aZhang et al. , 2014b, drought (Bertolini et al. 2013;Zhang et al. 2014aZhang et al. , 2014b, salinity (Carnavale Bottino et al. 2013;Sun et al. 2015), and wounding (Tang et al. 2012;Wang et al. 2014). In addition to miRNA transcriptional control, the whole biogenetic machinery responds to stress conditions through transcriptional, posttranslational and proteolytic regulation on several elements of the biogenetic complexes (Manavella et al. 2019).
Among the environmental conditions triggering miRNA modulation, metal content in soil is a significant one. Indeed, plant homeostasis of trace elements is tightly regulated, as plants must both ensure adequate uptake and metabolism of essential elements and avoid toxicity due to excess of micronutrients as well as the presence of nonessential elements. In this context, miRNAs are key players in the finetuning of metal homeostasis. For example, analyses in Cu-deficient conditions revealed the up-regulation of a highly conserved set of miRNAs including miR397, miR398, miR408, and miR857, leading to repression of dispensable Cu-containing proteins and redistribution of Cu reserves toward essential processes such as photosynthesis (Abdel-Ghany & Pilon 2008;Lu et al. 2011). On the other hand, Cu excess, an increasingly common situation due to agricultural practices, promotes down-regulation of previously cited Cu-responsive miRNAs, as well as modulation of others involved in stress response and metal transport (Fu et al. 2019;Jiu et al. 2019). Several toxic metals and metalloids have also been found to induce significant modulation of miRNAs. The involvement of the latter in stress due to Cd, Hg, As, Al, Pb, Cr, and Mn has been extensively reviewed (Ding et al. 2020;Noman et al. 2019;Noman & Aqeel 2017). In these cases, miRNA activity generally leads to reorganization of plant development and modulation of antioxidants and stress responses, although the modulated sets of miRNAs are element-specific (Ding et al. 2020;Noman & Aqeel 2017).
Zn is an extremely interesting element in this context: indeed, as a micronutrient, Zn is essential for all living organisms. This metal is a fundamental co-factor, with both structural and catalytic functions, in a large variety of proteins. It has been estimated that, on average, about 9% of the whole eukaryotic proteome is composed of Znbinding proteins, of the latter, the majority is predicted to be either enzymes (47%) or transcription factors (44%; Andreini et al. 2009). In Arabidopsis thaliana, more than 2000 proteins have been proposed by bioinformatic analysis to bind or transport Zn (Andreini et al. 2009), involved in a variety of extremely important processes, including DNA synthesis, transcription and translation, photosynthesis and proteolytic control of protein activity (Hänsch & Mendel 2009). On the other hand, Zn excess can prove detrimental for plants: toxicity can derive from competition with other metallic co-factors and indirect formation of reactive oxygen species (ROS), resulting in impairment of photosynthesis, cell death and generally stunted growth (DalCorso 2012). In light of this, Zn homeostasis in plants needs to be under tight control. Although precise determination of Zn requirements is difficult to achieve, it has been estimated that the internal level of free Zn is below the nanomolar range in eukaryotic cells. In plants, optimal Zn concentrations, including both free and chelated or compartmentalized metal, are generally between 15 and 50 mg kg À1 dry biomass (Hänsch & Mendel 2009;Sinclair & Krämer 2012). However, this range is extremely variable, in line with the vast natural diversity of plants adapted to different environments and edaphic conditions.
Interestingly, a class of plants has been identified, called hyperaccumulators, able to accumulate extremely high metal concentrations in their above-ground tissues and to tolerate them without showing toxicity symptoms (Baker & Whiting 2002;Krämer 2010). Among them, the facultative metallophyte Arabidopsis halleri is particularly interesting due to its constitutive ability to hyper accumulate Zn and its close phylogenetic proximity with model species A. thaliana (Krämer 2010). Transcriptomic analyses comparing nonaccumulator A. thaliana with hyperaccumulator A. halleri highlighted differential modulation of several genes involved in nutrient homeostasis and stress responses, many of which are constitutively expressed at high levels in A. halleri (Becher et al. 2004;Talke et al. 2006;Weber et al. 2004).
Despite the amount of data available for protein-coding transcriptome, very little work has been produced regarding the Zn effect on small regulatory RNAs. Zn-deficient conditions were analyzed in Brassica juncea and Sorghum bicolor, revealing a comparatively small set of modulated miRNAs, mostly involved in plant development and stress response Shi et al. 2013 Meyer et al. 2015) were used for this study. Seeds were stratified for 3 days at 4 C to break seed dormancy, then sown in garden soil and grown in a growth chamber on a short-day regime (8 hr light/16 hr dark, illumination 100-120 μmol m À2 s À1 , day/night temperature 22/18 C). Four-week-old plants were watered with either water (A. thaliana control and A. halleri) or with 500 μM ZnSO 4 (A. thaliana + Zn) for 1 week. Total Zn content in treated and untreated soil was measured at the end of the experiment by inductively coupled plasma atomic emission spectrometry (ICP-AES), as described by Fasani et al. (2019). Total Zn in soil was 52.6 ± 1.1 mg kg À1 DW in control conditions and 165.0 ± 5.9 mg kg À1 DW in Zn-treated conditions; these results are comparable with the mean values observed worldwide for Zn content in soils and with a moderately Zn-rich soil, respectively (Alloway 2008;Kabata-Pendias 1995).
CaCl 2 -extractable Zn, corresponding to soluble and exchangeable metal, was obtained by incubating air-dried soil in 10 mM CaCl 2 in a 1:2.5 proportion for 16 hr; the resulting solution was filtered and analyzed by ICP-AES. CaCl 2 -extractable Zn was 0.23 ± 0.01 mg kg À1 DW in control conditions and 0.75 ± 0.02 mg kg À1 DW in Zn-treated conditions; these values fall in the range of CaCl 2 -extractable Zn obtained in previous studies (Esnaola et al. 2000;Pueyo et al. 2004).
Rosettes of untreated and treated plants were collected and frozen in liquid nitrogen for further analyses. Three pools of five plants each were collected for each condition and genotype and considered as three different biological replicates.

| Physiological analysis of plants
Chlorophylls were extracted in 80% aqueous acetone buffered with NaCO 3 ; total chlorophyll content was measured as described by Porra et al. (1989).
In situ O 2 À accumulation in above-ground tissues was detected by nitroblue tetrazolium (NBT) staining, as reported in Rossetti and Bonatti (2001). Superoxide dismutase (SOD) activity was evaluated by native polyacrylamide gel electrophoresis (PAGE) and in-gel NBT staining, as in Chu et al. (2005); equal protein loading (30 μg) was demonstrated by Coomassie staining of a replica gel in SDS-PAGE.
Quantification of global SOD activity was achieved by scanning the gels and determining band intensity with Quantity OneR software v4.4.1 (Bio-Rad).
Zn concentration in leaves was determined by ICP-AES as previously indicated (Fasani et al. 2019). Each of the analyses here reported was performed in triplicate. Reads were processed using the miRPlant tool . At first, reads were trimmed for the adapter sequence, and the reads with a length lower than 18 and higher than 22 were discarded. The reads were aligned against the A. thaliana miRNAs in the miRBase database (http://mirbase.org/; Kozomara et al. 2019), with the Javacoded bowtie algorithm implemented in miRPlant software not allowing any mismatch. For miRNA quantification, only reads completely covering a mature miRNA were considered. The differential miRNA expression analysis between the samples was performed using the edgeR package (Robinson et al. 2010

| Northern blot analysis
Small RNAs were purified from the previously collected samples using the mirPremier microRNA Isolation Kit (Sigma-Aldrich). About 1.5 μg small RNAs was separated on a denaturing 15% polyacrylamide gel containing 7 M urea and then transferred on a Hybond-N+ nylon membrane (GE Healthcare) in a semi-dry electroblotter. DNA oligo probes were labeled with [γ-32 P]ATP using the mirVana Probe & Marker Kit (Thermo Fisher Scientific); probe sequences are reported in Table S1.
Blots were pre-hybridized for 30 min in the ULTRAHyb-Oligo Hybridization Buffer (Thermo Fisher Scientific). Hybridization was performed at 38 C overnight, according to the manufacturer's instructions. Signals were detected on an autoradiography film. Blots were then stripped and rehybridized with a probe complementary to U6 (Table S1) as a loading control to overcome unintended differences in RNA loading.
Band intensity was measured using the Image Lab software (Bio-Rad).
Accumulation of mature miRNAs was evaluated as the ratio between miRNA probe intensity and U6 probe intensity.

| Real-time RT-PCR
Total RNAs were purified from the previously collected samples using the TRIzol reagent (Thermo Fisher Scientific), according to the manufacturer's instructions. After DNase treatment, first-strand cDNA was synthesized from 2 μg of total RNA using the Superscript III Reverse Transcriptase Kit (Thermo Fisher Scientific). Real-time RT-PCR was performed using the Platinum SYBR Green qPCR SuperMix-UDG kit (Thermo Fisher Scientific) and a StepOnePlus Real-Time PCR System (Applied Biosystems). Each reaction (40 amplification cycles) was carried out in triplicate; melting curve analysis was applied to confirm amplification specificity. Primers for miRNAs and miRNA targets are listed in Table S2. Endogenous reference genes for data normalization were β-ACTIN (At5g09810) and UBIQUITIN 10 (At4g05320). Relative expression was evaluated using the 2 ÀΔΔCT method (Livak & Schmittgen 2001). this is associated with higher global SOD activity, as highlighted by ingel SOD analysis ( Figure 1D). Zn accumulation was moderately, although not significantly, higher in Zn-treated A. thaliana plants compared to untreated ones; Zn concentration in A. halleri leaves was double than in A. thaliana ( Figure 1E).

| Several miRNAs are differentially expressed under Zn treatment and between A. thaliana and A. halleri
The miRNA-Seq analysis, considering untreated and Zn-treated A. thaliana and untreated A. halleri, identified 129 expressed miRNAs for a total number of reads ranging between about 61 and 81 million/ sample. The most represented family was miR166, which was also the most expressed in all genotypes and conditions (for 3p strands, 94% of all reads in A. halleri, about 88% in A. thaliana); also abundant were miR398, miR396, and miR165 isoforms ( Figure 2A, Data S2). On the contrary, miR169 3p isoforms and miR395 were not expressed in A. thaliana. Moreover, no reads were detected for several miRNAs in A. halleri; of them, miR172e-5p, miR391-3p, and -5p showed complete sequence identity between A. thaliana and A. halleri, verified by sequence alignment with the A. halleri spp. gemmifera genome (Briskine et al. 2017), and therefore confirmed as not expressed. As for the other miRNAs not identified in A. halleri, they were either absent in the genome sequence or not conserved regarding sequence identity and were therefore not considered for further analysis.
Given these premises, the comparative analysis revealed 81 differentially expressed miRNAs belonging to 33 already described families ( The function of a wide proportion (37%) of the modulated miRNAs is unknown: of these, the majority consists of the complementary sequences referenced as "passenger strands" and still poorly characterized. Of the remaining, the most represented functional class (41%) is associated with plant development; several miRNAs are also involved in nutrient homeostasis and response to biotic stresses (12% and 5%, respectively, Figure 2D).
Validation of RNA-Seq results was performed on a set of remarkable miRNAs involved either in development (miR157, miR159, miR319, and miR390) or nutrient homeostasis (miR395, miR398, and miR408). Northern blot on the mature miRNA or real-time RT-PCR on its precursor was applied according to their expression levels, esti-   LACCASE13 (LAC13, AT5G07130) for miR408 (Figure 4). Transcription factor SPL3 was expressed at higher levels in A. halleri, in line with the overall lower levels of its regulators miR156-5p and miR157-5p. Analogously, the precursor of trans-activated siRNA3, TAS3, the transcription factor TCP4 has an expression profile that is consistent with its in view of no significant modulation of miR398 (Figure 4). However, scarce notice has been given to miRNAs in Zn homeostasis, except for some works considering Zn deficiency Shi et al. 2013). Due to its dual condition as an essential micronutrient and a toxic trace element when in excess (Andreini et al. 2009;DalCorso 2012), Zn uptake and distribution must be kept under strict control. This makes the analysis of plant response to Zn very interesting, although challenging: indeed, plant behavior toward this metal is extremely variable and associated with the adaptation to a wide range of different edaphic conditions. In particular, hyperaccumulator species show hypertolerance as well as tightly regulated uptake, translocation and compartmentalization of metals. Yet, despite the precise control of ionic balances displayed, no data are available on miRNAs in hyperaccumulators.

| DISCUSSION
In light of the evidence, this study has focused on miRNA Under the tested conditions that were chosen in order not to induce excessive stress in nonaccumulator A. thaliana, Zn-treated plants did not show apparent toxicity symptoms, apart from an increase in O 2 À accumulation and in global SOD activity. Zn accumulation in leaves was slightly, but not significantly, higher, consistently with the characterization of A. thaliana as an excluder species (Arrivault et al. 2006). On the other hand, Zn accumulation in untreated A. halleri was significantly higher than in A. thaliana, although lower than what was observed in previous studies in native metallicolous soil and upon hydroponic conditions (Corso et al. 2021;Schvartzman et al. 2018). However, in this study, plants were grown in unpolluted soil, having low Zn content and bioavailability. This considered, it is possible that A. halleri was under moderate Zn deficiency, but the condition was not so substantial as to produce an appreciable phenotype of Zn deprivation, as evidenced by the physiological characterization.
The most striking result emerging from miRNA-Seq analysis is that a consistent number of miRNAs is differentially expressed when comparing the two species considered, whereas a significantly smaller subset is modulated upon Zn treatment in A. thaliana. Analogously, Zn deficiency in S. bicolor produced only a small set of modulated miRNAs in leaves, despite a significant reduction in plant growth . Moreover, it must be remembered that the Zn treatment applied in this study produced a condition of moderately Zn-rich soil in order to not induce excessive stress in Very low levels of miR163 expression had been previously found also in Arabidopsis arenosa (Ng et al. 2011), a pseudo-metallophyte evolved independently from A. halleri, coherently with the hypothesis of partially convergent adaptive processes in the behavior toward metals (Preite et al. 2019). Interestingly, some other miRNAs involved in defense against biotic stresses were found as constitutively expressed at different levels in A. halleri in comparison to A. thaliana (e.g., miR400, miR472, miR825, and miR846). This evidence is coherent with the evolution of different defense mechanisms in A. halleri by changes in copy number and expression levels of biotic stress-related genes (Becher et al. 2004;Suryawanshi et al. 2016). Indeed, this adaptive strategy is the result of both convergence between response strategies against biotic/abiotic stresses and metal hyperaccumulation, providing some form of F I G U R E 4 Expression analysis of miRNA targets, evaluated by real-time RT-PCR. Statistically significant variations (P < 0.05), evaluated by one-way ANOVA followed by a post hoc Tukey's test, are marked with letters, the same letter corresponding to nonstatistically significant differences elemental defense against pathogens (Shahzad et al. 2013;Stolpe et al. 2017).
In addition to the stress-related miRNAs reported above, a sub-  (Sunkar et al. 2006), as well as the associated Cu chaperone CCS1 (Beauclair et al. 2010). In this view, miR398 links mineral homeostasis with the control of oxidative stress. Indeed, the lower expression of miR398-3p in A. halleri is consistent with the constitutively higher expression of the targets CSD1, CSD2 and CCS1 observed in this study and the lower accumulation of reactive oxygen species already described in the pseudo-metallophyte (Baliardini et al. 2015;Chiang et al. 2006). On the other hand, miR398-3p isoforms were not modulated upon Zn treatment in A. thaliana. This is apparently in contrasts with the higher O 2 À accumulation and SOD levels observed in Zn-treated A. thaliana in this study and with miR398 down-regulation upon oxidative stress and excess of redoxactive metals (Sunkar et al. 2006). However, Zn is not a directly redox-active metal (Cuypers et al. 1999), and it has been proposed to alter redox homeostasis indirectly, with no significant effect on expression levels of miR398b and c (Remans et al. 2012). As for their targets, CSD1 expression was down-regulated in the same conditions, whereas CSD2 and CCS1 were moderately but significantly induced, and global SOD levels are almost double in Zn-treated plants than in control conditions. Cu/Zn SOD modulation was contrary to what was observed by Remans et al. (2012); however, it should be considered that the treatment imposed in this study is different in both the growth substrate and the duration, thus resulting in a milder stress.
Interestingly, by the miRNA-Seq analysis, only miR398a-5p was upregulated upon Zn treatment in A. thaliana. This isoform belongs to the poorly characterized passenger strands; the targets predicted in this study do not allow a clear definition of its possible role in the plant, although the drought-inducible transcription factor ERF053 has been proposed as a putative target . Overall, Zn treatment upon sub-toxic conditions determines a moderate alteration of redox status in A. thaliana, that correlates with the modulation of SOD genes but not with that of miR398. On the other hand, the expression of the whole regulatory hub associated with miR398 is markedly different in A. halleri, leading to constitutively activated strategies for defense against oxidative stress as a part of the adaptive background of metal hypertolerance.
In conclusion, in A. thaliana, high Zn in soil induces the modulation of a small set of miRNAs involved in stress response, nutrition and plant development that are constitutively down-regulated in the facultative metallophyte A. halleri. In addition to these, several other miRNAs have substantially different transcript levels in A. halleri than A. thaliana, coherently with native differences in development and nutrient homeostasis, as well as with constitutively activated strategies for stress response, in particular for oxidative stress. Overall, these results support the hypothesis that adaptation to metalliferous soils implicates the reorganization of plant growth, allocation of resources and global mineral nutrition.

ACKNOWLEDGMENTS
The authors would like to acknowledge colleagues of the IGA Technology Services (Udine, Italy) for the sequencing of small RNAs.

DATA AVAILABILITY STATEMENT
All data supporting the findings of this study are available within the paper and within its supplementary materials published online.