Stars and symbiosis: microRNA- and microRNA*-mediated transcript cleavage involved in arbuscular mycorrhizal symbiosis.

The majority of plants are able to form the arbuscular mycorrhizal (AM) symbiosis in association with AM fungi. During symbiosis development, plant cells undergo a complex reprogramming resulting in profound morphological and physiological changes. MicroRNAs (miRNAs) are important components of the regulatory network of plant cells. To unravel the impact of miRNAs and miRNA-mediated mRNA cleavage on root cell reprogramming during AM symbiosis, we carried out high-throughput (Illumina) sequencing of small RNAs and degradome tags of Medicago truncatula roots. This led to the annotation of 243 novel miRNAs. An increased accumulation of several novel and conserved miRNAs in mycorrhizal roots suggest a role of these miRNAs during AM symbiosis. The degradome analysis led to the identification of 185 root transcripts as mature miRNA and also miRNA*-mediated mRNA cleavage targets. Several of the identified miRNA targets are known to be involved in root symbioses. In summary, the increased accumulation of specific miRNAs and the miRNA-mediated cleavage of symbiosis-relevant genes indicate that miRNAs are an important part of the regulatory network leading to symbiosis development.


Introduction
Small silencing RNAs are a complex group of short RNAs with sizes in the range of ~20-25 nucleotides (nt) in length which can regulate gene expression on transcriptional and post-transcriptional levels (Bartel, 2004;He and Hannon, 2004).
MiRNAs in plants are mostly transcribed from intergenic located MIR genes through RNA polymerase II activity, resulting in a 5' capped and 3' polyA tailed primary transcript (Cai et al., 2004;Lee et al., 2004). A region within the primary transcript folds into an imperfect stem-loop structure, which is cut by DCL resulting in a miRNA precursor. This precursor is subsequently processed by the same DCL protein to yield the miRNA/miRNA* duplex with a 2 nt overhang at the 3' ends (Papp et al., 2003;Vazquez et al., 2004;Xie et al., 2004). In contrast to animal miRNAs, both duplex strands of plant miRNAs are 2'-O-methylated at their 3' ends by HEN1 to prevent them against degradation (Yu et al., 2005;Yang et al., 2006). The duplex is then exported to the cytosol presumably by HASTY (Park et al., 2005). An AGO protein then incorporates one of the duplex strands, which is then referred to as mature miRNA (Liu et al., 2004;Meister et al., 2004). Its counterpart from the duplex is called miRNA* and will often be degraded after release of the mature strand (Khvorova et al., 2003;Schwarz et al., 2003). The miRNA within the AGO protein then guides the RISC to the target mRNA (Bartel, 2009). This is achieved by pairing of the miRNA with a specific binding site within the target transcript (Wang et al., 2008b;Wang et al., 2008a). The RISC then suppresses expression of the transcript by inhibiting translation or transcript cleavage of the miRNA/target duplex, normally

Results
Deep sequencing of M. truncatula small RNAs and degradome tags The goal of this study is to identify miRNAs and miRNA targets which are related to transcriptional reprogramming of plant cells during arbuscular mycorrhizal symbiosis.
For this purpose, we used high throughput sequencing with Illumina technology to detect and profile miRNAs and miRNA targets of M. truncatula. We inoculated M. truncatula seedlings with G. intraradices and harvested the mycorrhizal plants and non-inoculated non-mycorrhizal plants at 3 weeks after inoculation and extracted root RNA. Two small RNA (sr) and degradome (deg) libraries were prepared from either mycorrhizal (myc) or non-mycorrhizal (nm) roots of M. truncatula. In total we obtained more than 30 million reads (table 1) and more than 97% of all reads were of the appropriate sizes, i.e. 15-31 nucleotides (nt) for the small RNA libraries and 20-21 nt for the degradome libraries. Within the small RNA libraries, the 21 nt and 24 nt molecules were the most abundant forms and the 24 nt small RNAs were more diverse than the 21 nt class (figure 1). Within the degradome libraries, we obtained slightly more 20 nt than 21 nt sequences (figure 1).

Mapping of small RNAs and degradome tags to the M. truncatula genome
The small RNA reads from both libraries were 3' trimmed and reads containing low complexity regions, e.g. poly(A) stretches, and low quality reads (containing N's) were removed. Reads were filtered for length between 15 to 31 nt. The reads of both libraries (sr-myc and sr-nm) were mapped to the M. truncatula genome (versions 2.0 and 3.0), annotated cDNA sequences, annotated non-coding RNA as well as to available Glomus intraradices sequences (The Glomus consortium, unpublished) using 100% (no mismatch) as identity threshold. 69% of redundant sr-myc and 72% of the sr-nm could be mapped with 100% identity to the M. truncatula genome. Of the sr-myc reads, 1.6% could be mapped with 100% identity to G. intraradices sequences. Of the 0.05 % reads of the nm libraries, which match G. intraradices sequences, 95% are rRNA sequences. 39% and 38 % of non-redundant nm and myc reads respectively were assigned to so far uncharacterized regions of the mt3.0 genome, supporting the assumption that most small RNAs originate from intergenic regions. A summary of read annotation of redundant (total) and non-redundant (unique) sr reads is shown in figure 2.
www.plantphysiol.org on September 2, 2017 -Published by 1 0 Mapping the degradome tags with 100% identity to M. truncatula and G. intraradices sequences, revealed a match of 86% and 76% of the total nm-and the myc-deg sequences to the mt3.0 genome, respectively. Of the myc-deg sequences, 7.7% mapped to G. intraradices sequences. The higher proportion of degradome tags mapping to G. intraradices as compared to the small RNAs is due to the fact that most degradome tags are of mRNA origin and available G. intraradices sequences contain a high proportion of mRNA sequencing data (The Glomus consortium, personal communication). An summary of read annotation of redundant and nonredundant degradome reads is shown in figure 3.
The degradome sequences are correlated to transcriptome data of mycorrhizal and non-mycorrhizal roots Recent degradome analyses in Rice and Arabidopsis showed that only a small fraction of these sequence tags represent miRNA-mediated transcript cleavage (Li et al., 2010;Zhou et al., 2010), the vast majority of these sequence tags represent non-miRNA mediated, unspecific RNA degradation. This means that degradome sequence tag data of two tissues can be used to compare expression levels, when neglecting the proportion of miRNA-mediated mRNA cleavage in the dataset. Hence, a correlation of transcript abundance represented by read counts for each transcript between both tissues in the degradome libraries and previous transcriptome studies is expected. We analyzed transcript regulation by calculating read counts in both degradome libraries and compared the log2-fold changes to transcriptome data obtained by microarray hybridization of mycorrhizal versus non-mycorrhizal M. truncatula roots (Gomez et al., 2009). To compensate ratio compression by saturation effects on microarrays, strongly specifically expressed genes (no sequence tags in one condition and at least ten tags in the second condition), we used the half minimum normalized abundance for those transcripts. The comparison between log2-fold changes values of both data sets revealed a significant correlation (Pearson correlation coefficient: 0.51, p value <0.00001) ( Figure S1). Hence, the degradome sequencing data can be used to detect miRNA-mediated mRNA cleavage and in parallel, they can also be used to investigate transcriptional changes between the two tissues.
Analysis of annotated miRNAs www.plantphysiol.org on September 2, 2017 -Published by 1 1 In order to investigate the expression of known (miRBase 16 annotated) miRNAs in mycorrhizal and non-mycorrhizal roots, we searched for known miRNAs in our small RNA libraries. 375 M. truncatula miRNAs are currently annotated at the miRBase (release 16) database (Griffiths-Jones et al., 2008). 162 mature miRNAs could be identified through 100% matches of reads to the mature sequence in one or both of our libraries (table S1), representing 43.2% of the currently known M. truncatula miRNAs. Next, we analyzed if we find reads matching the annotated miRNA* or precursor sequences. Surprisingly, in some cases we found reads matching the miRBase-annotated precursor sequence, but reads matching the annotated mature miRNA sequences with 100% identity were absent. This indicates that some miRNAs can occur as different isoforms or siblings (Vazquez et al., 2008;Zhang et al., 2010), i.e. miRNAs originating from the same primary transcript with distinct mature forms.
MiRNA isoforms (isomiRs) often contain a small sequence shift or additional nucleotides (Ebhardt et al. 2010), whereas miRNA siblings are non-overlapping in comparison to the originally annotated mature or star miRNAs (Zhang et al., 2010).
We detected 17 isoforms of miRNAs whose mature sequence was missing (miR169e/h/I/n/o, miR393b, miR2594a/b, miR2611, miR2674, miR2678, miR2679c miR2680a-e, respectively). In addition, for two miRNAs we found reads of the annotated star sequence but no reads matching the mature sequence (miR171f, miR398a). Taken this into account we found a total of 181 known miRNAs in mycorrhizal and non-mycorrhizal M. truncatula roots.
To investigate the specific abundance of annotated mature miRNAs and the corresponding miRNA* and found isomiRs, we calculated the read counts of the corresponding sequences in our small RNA libraries. Interestingly, for 11 miRNAs, the isoforms of the mature strand were found to have a higher abundance than the sequence annotated as mature miRNA sequence at miRBase (miR156, miR166b/c/f, miR1509a, miR1519b, miR2597, miR2610a/b, miR2620, miR2645). In addition, we found a high abundance for miRNA* sequences of several legume specific miRNAs like miR1507*, miR2086*, miR2089* and miR2118* (Szittya et al., 2008;Jagadeeswaran et al., 2009). Additionally, the star strand for 8 miRNAs (miR160b/e, miR169d/l/m, miR369a, miR2089, miR2118) and a different isoform of 19 miRNA star strands (miR171c, miR2088a, miR2592a-g/i-j/o-s, miR2595, miR2612a, b) showed a higher abundance than the corresponding mature strand. 4 miRNAs showed a similar expression of mature and star strand (miR172, miR399c/j/k). Hence, we could www.plantphysiol.org on September 2, 2017 -Published by 1 2 identify 48% of all annotated M. truncatula miRNAs in our two small RNA libraries.
Notably, a high proportion of these miRNA seemed to occur in different isoforms than the annotated mature miRNA sequence.
Prediction of novel M. truncatula miRNAs In order to find novel M. truncatula miRNAs, we subjected the 15-31 nt reads of the small RNA libraries to a miRNA prediction pipeline. In order to calculate the discovery rate of the adapted miRDeep algorithm, we analyzed the ability of the algorithm to recover previously annotated miRNAs from a given dataset. To simulate the prediction pipeline as described in the Methods section, where we excluded reads that matched known miRNAs, we used here for evaluation only reads that were sequenced at least 20 times, that were not matching more than 25 times on the M. truncatula genome mt3, and that matched with 100% in sense direction onto annotated miRNAs precursor sequences.
Only 139 of the 340 non-redundant miRBase precursor had at least one read with an abundance of greater than 20, which is the minimum threshold we used for prediction before. Only 96 of these 139 known miRNA precursor had a mature miRNA read abundance of at least 20 reads. Since the excising of the correct precursor sequence depends on the position and hybridization of the (most likely) mature miRNA, we assume here that using the described parameters the new method should be able to predict those 96 known miRNAs from the given dataset. Using exactly the same prediction pipeline as for the new miRNAs (see Methods section) we could predict 87 (=90,26%) of the known miRNAs (data not shown). Applying the plant miRDeep program to our small RNA datasets, revealed the prediction of 515 novel miRNA candidates of 20-25 nt length.
To further validate these 515 miRNA candidates, we included criteria for plant miRNA annotation (Meyers et al. 2008). All 515 predicted miRNA candidates are produced from a single-stranded stem-loop precursor, which is required for miRNA prediction within the miRDeep pipeline. For 69 candidate miRNAs the corresponding miRNA* sequence was present in the deep sequencing data, hence these candidates fulfill the primary criterion for miRNA annotation. Additional 369 out of 515 miRNA candidates satisfy at least one of the ancillary criteria including conservation amongst one or more plant species, clustering into miRNA families and confirmed target by degradome analysis. However, criteria for miRNA annotation from deep sequencing 1 3 data have been updated recently, and include now the requirement that high confident miRNAs have multiple cloned nucleotide reads with relatively fixed 5´ends at both arms of the precursor sequence and the possibility to identify the miRNA/miRNA* duplex pair (Berezikov et al. 2010). Taking these criteria into account, we annotated 243 of our miRNA candidates as miRNAs at miRBase (Griffiths-Jones et al., 2008). A list of these miRNAs is shown in

Expression profiling of miRNAs by read count analysis
In order to investigate a putative role of novel and conserved miRNAs in the mycorrhizal symbiosis, we searched for miRNAs with induced or decreased expression in mycorrhizal roots when compared to non-mycorrhizal roots. To identify these differentially expressed miRNAs, the abundances of miRNAs were calculated by comparing the normalized numbers of reads in both libraries. Using a log2-fold cutoff of 0.7, 20 distinct miRNA sequences showed a differential abundance between mycorrhizal and non-mycorrhizal roots (table 2). MiR5229a/b showed the strongest elevated abundance levels (log2-fold change 9.7) in mycorrhizal roots as compared to non-mycorrhizal roots. Interestingly, we also found miRNA star sequences to be regulated in mycorrhizal roots. MiR169d*/l*/m*/e.2* was more abundant as the annotated mature miRNA sequence, and in addition 4-fold higher abundant in mycorrhizal roots. This is of particular interest because the miR169d*/l*/m*/e.2* target identified by degradome analyses (see below) is MtBcp1, which itself is specifically transcribed in arbuscule containing cells. In addition miR160f* was not detected in non-mycorrhizal roots but was clearly abundant in mycorrhizal roots. In summary, we could identify 20 miRNA with significantly (χ 2 test; P 1 4

Identification of miRNA cleavage targets by degradome analysis
The small RNA sequences presented here revealed the identification of 243 novel miRNAs and additional miRNA candidates. However, these data offer limited biological information without reliable target identifications in order to get information about the cellular processes where these miRNAs might be involved. For this purpose, we carried out a degradome analysis, which is a large-scale experimental method for miRNA target identification. Sequencing degradome libraries of Medicago non-mycorrhizal and mycorrhizal roots revealed a total number of 5.6 million reads and 7.8 million reads, respectively. In order to find miRNA cleavage sites in the M.  (table 3). In total, we identified 185 mRNAs (mt3) to be miRNA, miRNA* or miRNA candidate cleavage targets, of which 157 are targets of the novel miRBase annotated miRNAs. We next wanted to estimate the frequency of miRNA mediated mRNA cleavage in the M. truncatula root transcriptome. The degradome sequence tags could be mapped to 27,729 mt3 genes. Hence, we detected miRNA mediated transcript cleavage for 0.67% of the represented root transcriptome. For a functional characterization of all miRNA targets identified, we carried out a MapMan analysis (Usadel et al., 2005). Interestingly, disease resistance proteins (27 genes) and transcription factors (33 genes) were clearly overrepresented within the identified target genes (Figure 4). This indicates that particularly the transcript regulation and the defense responses are controlled by miRNA mediated mRNA cleavage in roots. miRNA and miRNA star mediated mRNA cleavage events were found for genes involved in root endosymbioses We wanted to investigate if transcripts, which are highly regulated in mycorrhizal roots, also represent miRNA targets. For this purpose, we used a read count analysis 1 5 of our degradome sequence tags to identify transcripts, which are strongly over-or underrepresented in mycorrhizal roots as compared to non-mycorrhizal roots. The log2-fold values were calculated for each of the 27,729 genes detected in the degradome sequences. Using a log2-fold cutoff of +/-1, we could identify 111 miRNA targets with differential transcription in mycorrhizal versus non-mycorrhizal roots. Table 4 shows all detected miRNA targets with a log2-fold cutoff of +/-1.5. The most strongly regulated miRNA target (Medtr8g109760.1) encodes a putative GRAS transcription factor, which is specifically transcribed in arbuscule containing cells (Gaude et al. submitted). Interestingly, transcripts encoding a putative phosphate transporter (Medtr5g076920.1) strongly accumulate in mycorrhizal roots and are a target of the miR399 family. In addition, we identified previously described mycorrhizal symbiosis induced genes to be miRNA targets, such as MtBcp1 (Medtr7g102930.1) (Pumplin and Harrison, 2009;Paradi et al., 2010), which is cleaved by miR169d*/l*/m*/e.2* ( Figure 5E). Additionally, this miRNA itself is induced in mycorrhizal roots (table 2). Interestingly, we found that one mycorrhizal symbiosisspecific transcript MtGst1 (Wulf et al. 2003) seems to be a target of miR5282 and in addition of the predicted miRNA candidate new_miRc_275 (figure 5B). It is further worth mentioning that MtNsp2 transcripts, which are a target miR171h (figure 5A), show elevated levels (log2-fold 1.6) in mycorrhizal roots. Figure 5 shows additional target plots indicating the mRNA cleavage mediated by miRNAs induced in mycorrhizal roots, namely miR160c and miR167; both are cleaving mRNAs encoding auxin response factors (figure 5C and 5D) and miR169 targeting a CCAAT-binding transcription factor in mycorrhizal roots (figure 5F). Notably, within the population of transcripts with decreased levels in mycorrhizal roots, disease resistance genes are highly overrepresented. We found 16 miRNA cleavage sites within disease resistance genes encoding transcripts with decreased levels in mycorrhizal roots.

Q-PCR expression analysis
To confirm the regulation of selected miRNAs, we carried out quantitative real time (q RT) PCR analysis. In addition to mycorrhizal roots, we measured mature miRNA abundance in nodulated roots fertilized with either 20 µM or 1 mM phosphate and in roots grown under full nutrition (1 mM phosphate and 5 mM nitrate). This allows first insight into the specificity of the miRNA regulation with regard to nodule symbiosis www.plantphysiol.org on September 2, 2017 -Published by 1 6 and nutrient availability. The mature miRNA abundance was measured by stem-loop qRT-PCR (Chen et al., 2005) ( Figure 6).
For expression analysis, we selected miR5229a/b and miR5204 because of their strongly elevated levels in mycorrhizal roots. In addition the miR160f* was chosen for expression analysis because this miRNA was undetectable in non-mycorrhizal roots.
Also miR160c was selected due to the elevated levels of mature miRNAs in mycorrhizal roots and additionally because it belongs to the same family as the mycorrhizal root specific new miR160f*. As expected, the qRT-PCR confirmed the induction in mycorrhizal roots as compared to non-mycorrhizal roots. Two of the measured miRNA candidates (miR160f* and miR5229a/b) were only detectable in mycorrhizal roots. MiR160c showed an increased abundance in mycorrhizal roots as compared to all other treatments measured. MiR167 was induced in mycorrhizal roots as compared to non-mycorrhizal roots under low phosphate, but showed increased expression in nodulated roots and under full nutrition. MiR169 and miR169d*/e.2*/l*/m* showed a clear positive response to increased phosphate nutrition. miR171h exhibited strongest expression in nodulated roots. This is of particular interest since we identified MtNsp2, a transcription factor necessary for nodulation symbiosis signaling (Kalo et al., 2005), to be a target of this miRNA (table   S4 and table 3). The miR5204 was increased in mycorrhizal roots according to the read count analysis, which could be confirmed by qRT-PCR. However, we found further increased expression under full nutrition, indicating phosphate to be a positive regulator of miR5204. Comparable low expression levels of this miRNA in nodulated roots grown at high phosphate might be due to the root nodules as strong phosphate sinks. In summary, the expression profiling showed miRNAs specifically accumulating in mycorrhizal roots as well as miRNAs induced in mycorrhizal roots but also regulated by other stimuli.

Localization of mature and star miRNAs by in situ hybridization
Mature miRNAs and star sequences, which were proven to show elevated levels in mycorrhizal roots were analyzed by in situ hybridization in order to get information about their spatial accumulation in mycorrhizal roots. As negative control we used a DIG-labelled scramble probe, which is a random 21 nt LNA enhanced DNA The criteria for the annotation of novel miRNA from high throughput sequencing data have been re-evaluated recently, since re-analysis of miRNA datasets of invertebrates and plants demonstrated a considerably high fraction of erroneously annotated miRNA (Rajogopalan et al. 2006, Ruby et al. 2006, Ruby et al. 2007). With the availability of deep sequencing data is now possible and essential to confidently determine the precise 5´ end of a mature miRNA (Berezikov et al. 2011;Chiang et al. 2010). In addition, reads matching the miRNA* should be present and have the potential to pair to the mature miRNA candidate with approximately 2 nt 3´ overhangs (Chiang et al. 2010). Taken these two latter criteria into account, we identified 243 novel miRNAs of M. truncatula, which are now deposited at miRBase.

Distinct miRNAs are upregulated in mycorrhizal roots
The read count analysis of our miRNA candidates showed that members of 20 miRNA are regulated in mycorrhizal roots as compared to non-mycorrhiza roots. For 8 miRNAs of the most strongly regulated candidates (log2-fold >1), a significant differential expression was proven by real time PCR analysis. The highest upregulation was observed for miR5229a/b (log2-fold 9.7), which is remarkably high taking into account that cellular differences in miRNA expression are hardly detectable, since whole root samples were used for library construction. We found miR5229a/b exclusively expressed in mycorrhizal roots and highly abundant in arbuscule-containing cells, suggesting a specific role during arbuscule development.
So far, no target for this novel miRNA could be detected, but in silico predictions suggest a transcript encoding a haem peroxidase to be a target of this miRNA. 2 0 accumulation in the phloem and around fungal hyphae give evidence that the miR169 family is also involved in the AM symbiosis. Moreover, miRNA* sequences of the miR169 family were also strongly induced in mycorrhizal roots and were shown to have a similar tissue specific accumulation as mature miR169. Remarkably, we were able to identify MtBcp1 to be a target of this miRNA*. MtBcp1 encodes a protein specifically accumulating in the periarbuscular membrane (Pumplin and Harrison, 2009) and it might be speculated that the miR169* sequences accumulating in mycorrihzal roots are involved in restricting MtBcp1 expression to arbuscule containing cells. A further miRNA, which was exclusively detectable in mycorrhizal roots was miR160f*, which was primarily detected in the phloem. However, the function of this miRNA* is yet elusive, because no targets were found by degradome sequencing. The same situation is true for the miR5204. Nevertheless, several targets could be in silico predicted including a heavy metal transport protein and a zinc finger transcription factor. Notably, miR5204 appears to be a phosphate responsive miRNA and is located around individual arbuscules. We hypothesize that Future investigation will unravel the physiological relevance of the discussed miRNAs during AM symbiosis.

Regulation of symbiosis relevant transcripts by miRNAs
The here applied transcriptome wide degradome analysis gives the unique opportunity to identify mRNAs, which undergo miRNA-mediated cleavage. Besides

2
This study provides evidence for miRNA*-mediated mRNA cleavage in M. truncatula roots. Moreover, we often found a notably high accumulation of miRNA* and for several miRNA* molecules even a higher abundance as compared to their mature miRNA. It is currently anticipated that one arm of the RNA duplex preferentially accumulates and is then referred as mature miRNA whereas the opposite strand is called miRNA*. Previous models suggested that the choice of the dominant miRNA arm encoding the mature miRNA depends on thermodynamic and structural properties of the duplex RNA molecule (Khvorova et al. 2003, Schwarz et al. 2003.
Nevertheless it was recognized that many miRNA* species also accumulate to substantial levels and are able to downregulate target mRNAs (   which were distinct from transcripts targeted by the corresponding mature strand. One example is miR169, which targets a CCAAT transcription factor important for root nodule development (Combier et al. 2006), whereas the star strand, which is more abundant than the mature miR169 in mycorrhizal roots, mediates the cleavage of the arbuscule-specific protein MtBcp1. Hence, it might be anticipated that the dominant form of this miRNA switches during the two endosymbioses and restricts the expression of different targets to distinct cell types. In addition, we also found miRNAs, of which targets were identified so far only for the miRNA*, particularly for miR5204*, which targets a GRAS TF specifically expressed in mycorrhizal roots. In summary, our findings propose that the establishment of a mycorrhizal symbiosis leads to a reprogramming of the miRNA-target network in roots including miRNA strand preferences. we used all reads from the small RNA libraries that could be mapped in sense direction without any mismatch onto the in miRBase16 annotated miRNA precursor sequences having a minimum abundance of 20 reads, but not representing annotated miRNA mature and star sequences. We defined isomiRs as reads that are not shifted more than 5 positions from their original mature or star 5' position, miRNAsiblings were then all other reads that could be mapped to the miRNA precursor.

MiRNAs modulate the defense response upon mycorrhizal infection
Small RNA deep sequencing and sequence processing The small RNA reads were 3' trimmed using the NOVOALIGN software       Normalization was carried out against a reference gene index (MtPdf2; MtEf1).
Primers were designed to measure specific miRNAs, if possible. If primers bind to more than one mature or star miRNA sequence, it is indicated in the figure.

6
Tables:      Devers et al.     Primers were designed to measure specific miRNAs, if possible. If primers bind to more than one mature or star miRNA sequence, it is indicated in the figure.