Introduction

Since the proposition of the “one gene - one enzyme hypothesis” by George W. Beadle and Edward L. Tatum in 1941 (Beadle & Tatum, 1941), it has become increasingly evident that the complexity of the proteome is more intricate than initially anticipated. It is firmly established that alternative splicing can give rise to multiple mRNA isoforms (Berget et al, 1977), and in certain instances, these isoforms can be translated into different protein variants. Alongside alternative splicing events, such as exon-skipping, which can potentially result in protein isoforms with distinct structures, the alternative use of transcription start sites can also lead to protein variants lacking specific domains that exhibit contrasting functions compared to the full-length variant (Svoboda et al, 1988).

Recent endeavors in uncovering the small peptidome, employing riboproteomics, a combination of Ribo-seq and proteomics, have unveiled thousands of novel protein-coding genes in humans (Mudge et al, 2022). Some of these short open reading frames (sORFs) originate from their corresponding loci in an overlapping manner and are frequently encoded in an alternate reading frame. Certain upstream sORFs can possess a regulatory role in controlling the larger downstream ORF at the transcriptional or translational level (Luo et al, 1995; Orbach et al, 1990). Furthermore, long non-coding RNAs (lncRNAs), which arise from pervasive transcription of the seemingly non-coding regions of the genome, represent another abundant source of small and potentially emerging protein-coding genes. It has been suggested that translated lncRNAs may even serve as a source for de novo gene birth (Ruiz-Orera et al, 2020).

Small proteins are abundant in both prokaryotes and eukaryotes, and many of them exhibit dynamic expression patterns. Notably, sORFs have been demonstrated to possess significant regulatory roles in plant development (Hanada et al, 2013) as well as in animals (Kondo et al, 2010; Zhang et al, 2020). In the field of human biology, certain sORFs have been identified to exhibit disease-specific expression patterns, suggesting their potential utility as diagnostic markers or future therapeutic targets (Chong et al, 2020).

When compared to animals, plants display a greater degree of developmental plasticity and exhibit substantial changes in gene expression in response to environmental fluctuations. Due to their sessile nature, plants constantly monitor their surroundings, particularly in relation to light, which is their main source of energy. Plants possess various photoreceptor systems, such as phytochromes and cryptochromes, which enable them to continuously assess the quality of light and determine their growth conditions in competition with other plants (Halliday et al, 1994; Pedmale et al, 2016). Phytochrome photoreceptors respond to both red and far-red light in a toggle switch manner where red light activates PHYB and converts it to the far-red light-responsive Pfr form, while far-red light inactivates PHYB and converts it to the inactive Pr form (Quail, 2002). Shade-sensitive plants like Arabidopsis promote elongation growth in response to a low red-to-far-red ratio. This growth response is mediated by the stabilization of PHYTOCHROME INTERACTING FACTOR (PIF) transcription factors (Lorrain et al, 2008). Under conditions of high red light (no shade), PHYB exists in the active Pfr form, leading to the degradation of PIFs. When present, PIFs directly activate genes encoding YUCCA enzymes, resulting in increased production of the plant hormone auxin in response to shade (Müller-Moulé et al, 2016; Won et al, 2011).

In addition to the pivotal role played by PIFs in regulating growth in response to shade, the class II homeodomain leucine zipper (HD-ZIPII) family of transcription factors has long been recognized as being light-regulated (Carabelli et al, 1993). Among the 10 genes that encode HD-ZIPII transcription factors, for HAT1, HAT2, HAT3, ATHB2, and ATHB4, it has been demonstrated that their expression is transcriptionally upregulated in response to shade (Carabelli et al, 2013; Ciarbelli et al, 2008). The presence of an EAR-domain in all of these HD-ZIPII genes suggests their function as transcriptional repressors. Initial experiments showed that ATHB2 is capable of binding to a cis-element within its own promoter to suppress its own expression (Ohgishi et al, 2001; Steindler et al, 1999). In addition to their role in self-regulation, HD-ZIPIIs have been found to influence the expression of genes encoding enzymes and signaling components involved in hormone pathways such as auxin, brassinosteroids, and gibberellic acid, all of which are known to promote growth (Sorin et al, 2009). Furthermore, HD-ZIPIIs have been shown to play a crucial role in leaf and meristem development, as loss-of-function mutants exhibit severe defects in maintaining meristems and developing leaves (Bou-Torrent et al, 2012; Turchi et al, 2015). Interestingly, REVOLUTA (REV), a member of the class III homeodomain leucine zipper (HD-ZIPIII) family, acts upstream of HD-ZIPII factors and also participates in the regulation of shade-induced growth by transcriptionally upregulating HAT2, HAT3, ATHB2, and ATHB4 (Brandt et al, 2012). Furthermore, it has been discovered that a dimer composed of HD-ZIPII and HD-ZIPIII can bind to a conserved cis-element in the microRNA genes MIR165/166, leading to the repression of their expression specifically in the adaxial domain of developing leaves (Merelo et al, 2016). Studies have also revealed that shade influences the rate of leaf proliferation, which is controlled by ATHB2 and ATHB4 (Carabelli et al, 2018). Additionally, in the hypocotyl, REV has been found to regulate xylem proliferation in a shade-dependent manner, likely providing structural support to the elongating hypocotyl while increasing water flow (Botterweg-Paredes et al, 2020).

HD-ZIPIII factors act as key regulators of meristem maintenance and leaf development, and therefore their regulation occurs at multiple levels, including post-transcriptional and post-translational. At the post-transcriptional level, microRNA165/166 modulates HD-ZIPIII expression, limiting it to the adaxial domain of developing leaves (Mallory et al, 2004). At the post-translational level, HD-ZIPIII factors are subject to regulation by LITTLE ZIPPER (ZPR) microProteins (Kim et al, 2008; Wenkel et al, 2007). MicroProteins are short regulatory proteins consisting of approximately 100 amino acids that share structural similarities with the proteins they regulate (Bhati et al, 2021; Eguen et al, 2015). The ZPR microProteins specifically possess a leucine zipper domain similar to that found in HD-ZIPIII factors, and they are directly and positively regulated by HD-ZIPIII transcription factors (Brandt et al, 2013). Evolutionarily, ZPRs are related to HD-ZIPIII factors, and it has been demonstrated that ZPRs originated from a duplication of an ancient HD-ZIPIII paralog in the euphyllophyte clade, followed by evolutionary trimming through the accumulation of degenerative mutations, resulting in the preservation of only the leucine zipper domain (Floyd et al, 2014).

Here, we investigate how shade affects the transcription landscape in Arabidopsis. We find that in response to shading alternative transcription start sites are more often employed, potentially, increasing proteome diversity. One of the alternative transcription start sites is located in the third exon of ATHB2 and the alternative transcript encodes a microProtein resembling the ZPRs that is devoid of the homeodomain DNA binding moiety but contains the leucine zipper part, necessary to interact with HD-ZIPII proteins. Ectopic expression of ATHB2miP, the microProtein isoform, inactivates ATHB2 by engaging it in a non-productive heterodimer and thereby affects growth of the hypocotyl and the root. By using ATHB2miP as a tool, we can effectively inactivate ATHB2 functions. CRISPR-induced mutations that remove the leucine zipper domain from the ATHB2 gene result in defects in responding to shade and additional defects in root development. The finding that ATHB2 localizes to nuclear photobodies, mediated by the leucine zipper domain, supports a fundamental role for this domain in the formation of higher-order protein aggregates and signaling. Transgenic plants that ectopically express ATHB2miP exhibit overlapping transcriptome changes with athb2 mutant plants. Gene ontology and cluster analysis reveal significant changes in genes encoding components of auxin synthesis and signaling, as well as genes controlling root development and nutrient uptake, specifically iron. After discovering that genes related to iron uptake were upregulated in athb2 mutants, we measured the internal iron levels in both shoots and roots. Our findings showed that the levels were increased in both athb2 mutants and transgenic plants that ectopically expressed ATHB2miP. Experiments on shade growth conducted on low and high iron media revealed that the availability of iron affects shoot and root growth, which is dependent on the activity of ATHB2. In summary, these findings demonstrate that ATHB2 and ATHB2miP are key mediators between shoot and root growth in response to environmental fluctuations and the nutritional status of the plant.

Results

The ATHB2 gene harbors an alternative transcription start site in the third exon

To mine the transcriptome of Arabidopsis thaliana for novel microProteins regulating shade avoidance, we used 5’ PEAT sequencing, a method that enriches capped RNA transcripts and results in reads that correspond to transcription start sites (Ni et al, 2010). Plants were grown for ten days in a white light regime (WL; R:FR = 7.7) and then exposed to additional far-red light, to simulate shade (WL+FR; R:FR = 0.13; deep shade). After shade treatment for 45- and 90-minutes samples, along with an untreated control (0 minutes shade) were collected. Overall, we identified 12,398 transcription start sites (TSSs). The majority of these corresponded to putative TSSs from TAIR10, but 40% (4,982) corresponded to sites that had not previously been annotated (Fig. 1A). Most of the reads mapped to the 5’UTR of already annotated genes, but a subset was found to map to regions within gene bodies. As we were particularly interested in identifying novel transcripts that might encode microProteins (i.e. overlapping gene products), we focused on the subset of genes encoding such alternative microProteins. We therefore established a set of criteria in order to identify sites in our dataset that might correspond to an alternative microProtein mRNA transcript. Firstly, we limited our search to open reading frames (ORFs) that would encode a protein within the size range of a typical microProtein of 5-15 kDa but would differ significantly in size to the overlapping full-length gene product. Secondly, we selected for ORFs that contained a single protein-protein interaction domain allowing the alternative proteins to interfere with the full-length isoforms by forming heterodimers. Finally, we considered the biological relevance of each alternative transcript: the number of reads must exceed a certain threshold to ensure robust expression of the transcript isoform. By applying these criteria, we identified 377 potential microProtein candidates, corresponding to 3% of the total number of TSSs identified in our dataset (Fig.1A). One of the alternative microProtein candidates we identified overlaps with the end of ATHB2 (Fig. 1B), a gene encoding a class II HD-ZIP transcription factor that plays a central role in the shade avoidance response (Ciarbelli et al., 2008; Schena et al, 1993). Like the full-length transcript, the alternative transcript appears to be shade-sensitive, with 5’PEAT-seq showing that expression peaks at 45 minutes. The ATHB2 alternative transcript was also identified by 5’RACE (Fig. S1A) and by sequencing the PCR product, we found that the alternative transcript is spliced at the same site as the full-length transcript and starts around position 653 of the full-length transcript (Fig. S1B). In Arabidopsis, the average length of a 5’UTR is 155 bp (Srivastava et al, 2018), therefore, we searched for alternative start codons no more than 300 bp downstream of the alternative TSS. We identified two near-cognate start codons with strong Kozak sequence similarity (≥ 0.7 and < 0.8) using the online tool TISpredictor (Gleason et al, 2022). The two start codons are TTG and CTG, both in-frame with the full-length ATHB2 protein sequence and producing ORFs that are 126 and 102 amino acids long, respectively, and consisting of the leucine zipper domain, and, in the case of the former, parts of the homeodomain (Fig. S1B). It is unclear whether the alternative transcript we have identified is the result of an independent transcription event or due to post-transcriptional processing of the full-length ATHB2 transcript involving downstream recapping (Ni et al., 2010). To confirm that ATHB2miP is the result of an independent transcription event, we sequenced full-length cDNAs using PacBio isoseq. This analysis revealed isoforms that start in the third exon of ATHB2, supporting an independent transcription event (Fig. 1C). Furthermore, we also observed antisense transcripts covering the introns. These changes occurred for both isoseq and the 5’PEAT-seq only at the 45-minute timepoint, indicating major chromatin reorganization in response to short-term shading. At 90 minutes, the situation stabilized, and ATHB2 was expressed at higher levels (Fig. 1C). To determine whether there are sequences in the gene body of ATHB2 that can promote transcription and translation, we fused a β-glucuronidase uidA (GUS) reporter gene to different sequences from the ATHB2 gene body (Fig. 1D). First, we fused GUS to the sequence 1 kb upstream of the ATHB2 full-length start codon ATG (pATHB2::GUS). As a control, we fused GUS to the sequence immediately upstream of an ATG located in the middle of intron two (pATHB2control::GUS). Finally, we fused GUS to the sequence immediately upstream of the first alternative, in-frame codon that we had identified, TTG (pATHB2miP::GUS). Whilst pATHB2control::GUS did not display any GUS staining upon treatment with the substrate 5-bromo-4-chloro-3-indolyl ß-d-glucuronide (X-Gluc; Fig. 1F), both pATHB2::GUS and pATHB2miP::GUS showed staining in the shoot apical meristem (SAM) and the hypocotyl, though to a lesser extent in the latter in pATHB2miP::GUS (Fig. 1E, G). This suggests that there are sequences within the second intron and third exon, that are excluded from pATHB2control::GUS, that can promote both transcription and translation. This might also lend support to our hypothesis that the alternative transcript arises through an independent transcription event to the full-length ATHB2 transcript.

An alternative transcription start site in the third exon of ATHB2 gives rise to a shade sensitive microProtein.

(A) 5’PEAT-seq reads map to 12,398 unique TSSs, 43% representing novel TSSs and 1.5% being candidates for alternative microProteins. (B) 5’PEAT-seq reads mapping to ATHB2. (C) Iso-seq reads mapping to ATHB2. (D) Design of constructs with GUS fused to either the promoter of ATHB2 (pATHB2::GUS), the sequence upstream of an ATG in the second intron of ATHB2 (pATHB2control::GUS; negative control), or the sequence upstream of a near-cognate start codon TTG in the third exon of ATHB2 (pATHB2miP::GUS). (E) GUS staining in pATHB2::GUS accumulates in the hypocotyl, SAM and cotyledon vasculature. (F) GUS staining in pATHB2control::GUS does not produce any GUS signal. (G) GUS staining in pATHB2miP::GUS plants accumulates in the SAM and hypocotyl. Scale bars = 200 µM.

ATHB2miP can heterodimerize with itself and the full-length ATHB2 isoform

We hypothesized that the ATHB2miP might inhibit the full-length ATHB2 in a manner similar to that of the LITTLE ZIPPERs (Kim et al., 2008; Wenkel et al., 2007). To test this hypothesis, we posed the following questions. Firstly, does the microProtein localize to the same cellular compartment as the full-length protein? It was previously reported that ATHB4, another member of the HD ZIPII family, contains a nuclear localization signal (NLS) in its homeodomain that is highly conserved in ATHB2 (Gallemi et al, 2017). However, our own analysis using LOCALIZER (Sperschneider et al, 2017), which has been trained specifically on plant proteins, suggests that there is an NLS in the LZ domain that is conserved in other HD-ZIPIIs, including ATHB4. Interestingly, LOCALISER also predicts a chloroplast transit peptide (cTP) at residues 1-46 of ATHB2 with a probability of 0.8. The cTP does not appear to be conserved, suggesting that, if real, it is specific to ATHB2. To determine the subcellular localization of ATHB2 and ATHB2miP, we transiently expressed eGFP fusions of the proteins in Nicotiana benthamiana leaves (Fig. 2A; eGFP-ATHB2, eGFP-ATHB2miP). eGFP-ATHB2miP encompasses the entirety of exon 3 and 4 of ATHB2. To further test whether the homeodomain of ATHB2 could target the nucleus, as shown previously for ATHB4 (Gallemi et al., 2017), we also fused a variant of ATHB2 lacking the leucine zipper to eGFP (eGFP-ATHB2HD). We observed that all variants of ATHB2 localized exclusively to the nucleus, unlike the negative control (eGFP) which was dispersed along the cell membrane and in the nucleus (Fig. 2A). We also observed that the distribution of eGFP-ATHB2 was not uniform in the nucleus, but instead localized to distinct nuclear speckles or, in some case, to the nucleolus specifically. Nuclear speckles were mostly absent in leaves expressing eGFP-ATHB2miP, but entirely so in leaves expressing eGFP-ATHB2HD, suggesting that interactions mediated by the leucine zipper domain and in part by the N-terminal domain are required for the speckles to form. We next asked whether we could find evidence that ATHB2 and ATHB2miP can physically interact with each other. It has been demonstrated that ATHB2 homodimerizes in order to bind to its target DNA sequence and that this interaction is directed by the leucine zipper domain (Sessa et al, 1993). To test whether ATHB2 and ATHB2miP can interact, we performed a yeast-2-hybrid (Y2H) study. As in the localization assay, ATHB2 or ATHB2miP were fused to either the activation or the DNA binding domain of GAL4. Constructs were transformed into yeast carrying the HIS3 gene fused to the target DNA sequence of GAL4. Growth of transformed cells on media lacking histidine revealed that ATHB2 and ATHB2miP can interact and form heterodimeric complexes as well as homodimers (Fig. 2B). To further confirm this interaction in planta we carried out a FRET-FLIM assay. Here again, ATHB2 and ATHB2miP were fused to either eGFP or mCherry and then transiently co-expressed in tobacco leaves. Using a confocal microscope, fluorescence lifetime of eGFP-ATHB2 was measured. We observed that in cells that co-expressed fluorescently tagged versions of ATHB2 or ATHB2miP there was a significant decrease in fluorescence lifetime of eGFP-ATHB2 when compared to cells that expressed eGFP-ATHB2 alone or with a control (mCherry; Fig. 2C), supporting that ATHB2 and ATHB2miP can physically interact.

ATHB2 and ATHB2miP localise to the nucleus and interact in vitro and in planta.

(A) Graphic representation of eGFP fused to ATHB2 variants: ATHB2, ATHB2miP, and ATHB2HD; panels show subcellular localisation of ATHB2 variants when fused to eGFP in tobacco leaves, with nuclear localisation shown in square and enlarged in lower panel. Arrows point to nuclear speckles that form when eGFP-ATHB2miP is expressed. Scale bars: upper panel = 50 µM; lower panel = 5 µM. (B) Yeast-2-hybrid with growth on histidine drop-out media when ATHB2 or ATHB2miP are fused to the activation domain and binding domain of GAL4. (C) FRET-FLIM assay. Co-expression of eGFP-ATHB2 with mCherry-ATHB2 or mCherry-ATHB2miP in tobacco leaves shows co-localization in the nucleus and causes significant reduction in fluorescent lifetime of eGFP-ATHB2. Scale bar = 5 µM. Significance levels: * = p<0.05, ** = p<0.01, *** = p<0.001, ns = not significant.

ATHB2miP perturbs the ability of ATHB2 to regulate transcription

To determine the functional consequences of the interaction between ATB2miP and ATHB2, we decided to study the transcriptional changes elicited by ectopic expression of ATHB2miP. For this, we grew transgenic plants overexpressing ATHB2miP, an athb2 mutant, and wild-type plants in white light conditions. Prior to RNA extraction, plants were exposed to 0, 45 and 90 minutes of additional far-red light. RNA-seq was then used to compare the transcriptomes of respective plants under white light and shade conditions. We hypothesized that overexpressing ATBH2miP would result in the inactivation of ATHB2, similar to the inactivation of HD-ZIPIII transcription factors by ZPR microProteins (Kim et al., 2008; Wenkel et al., 2007). Differential gene expression analysis, adjusting for genotype, environment and genotype by environment interaction, identified approximately 5,900 differentially expressed genes (DEGs) with FDR ≤0.05 and significant up/down-regulation (logFC>1; logFC<1). As it is known that HD-ZIPIIs establish an autoregulatory feedback loop by binding to their own promoter and repressing their own expression (Ciarbelli et al., 2008), we examined the ATHB2 gene in all three genotypes. No reads mapping to the ATHB2 locus were detected in the t-athb2 mutant (Figure S2), indicating a true null allele. The transgenic plants overexpressing ATHB2miP showed an increase in the number of reads covering exons 1 and 2 compared to wild type plants, suggesting an increase in the endogenous expression of ATHB2 (Figure S2). These findings indicate that ectopic expression of ATHB2miP sequesters the endogenous ATHB2 protein and inactivates the negative auto-repressor loop.

In total, we identified 95 DEGs whose expression depended on genotype, environment, and genotype x environment (Fig. 3A). The GO analysis of these 95 genes revealed an overrepresentation of genes involved in the shade avoidance response, growth, and hormone/auxin signaling (Fig. S3A). These findings demonstrate that both t-athb2 mutant plants and the transgenic plants overexpressing ATBH2miP can detect shade and respond to it at the molecular level. In order to compare the genotypes and environmental conditions, we carried out an unbiased gene cluster analysis using mFuzz (Kumar & Futschik, 2007). Our analysis revealed 17 individual clusters with distinct expression patterns (Fig. S3B). Cluster 1 (Fig. 3B) contains 200 genes that are upregulated only in 35S::miP, but not in either of the other two genotypes. We found an overrepresentation of genes involved in root development in this cluster. Cluster 2 contains 382 genes with lower expression in both 35S::miP and t-athb2. This cluster is also enriched in genes encoding hormonal regulators and regulators of root development. Finally, cluster 14 contains 152 genes that are up-regulated in both 35S::miP and t-athb2 plants (Fig. 3B). Genes in this cluster have known roles in the shade avoidance response. We also found an overrepresentation of genes involved in iron uptake and signalling, which are strongly upregulated in 35S::miP and t-athb2 plants. The latter finding suggests that ATHB2 may play a role in mediating between nutritional status and growth responses.

Gene expression profiling of the t-athb2 and 35S::miP lines.

(A) Heatmap showing expression changes in 95 genes whose expression depends on both the genotype and the shade treatment. Lowly-expressed genes are in blue, highly-expressed genes are in yellow. (B) 3 of the 17 clusters generated with MFuzz. Expression changes of the genes included in each cluster are on the y-axis. The 3 genotypes and shade conditions are on the x-axis. (C, D, E) Venn diagrams showing the number of genes that are upregulated and downregulated in each of the three genotypes included in the RNA-Seq and the overlap between them in WL, and 45 and 90 mins of WL+FR.

Having established that ectopic expression of ATHB2miP perturbs ATHB2 function and alleviates the negative auto-repressive feedback loop we assessed DEGs in the three different environmental conditions (0-, 45-, and 90-minutes additional FR-light). Focusing on the genes upregulated in both t-athb2 and 35S::miP plants, we found 22 genes in white light (Fig. 3C), 157 genes in 45-minute shade (Fig. 3D), and 20 genes in 90-minute shade (Fig. 3E). As previously observed, we identified upregulated genes related to auxin synthesis and signaling, as well as genes involved in root development. Genes related to auxin-biology and hypocotyl elongation (Table 1) include YUCCA3 and YUCC9, both of which have known functions in auxin biosynthesis in response to shade. (Goyal et al, 2016). Genes associated with root development (Table 2) include several cell wall modifying enzymes, most notably expansins, as well as transcription factors. Taken together, these findings support the function of ATHB2 as a suppressor of auxin signaling and root development. Analysis of genes down-regulated in both t-athb2 and 35S::miP plants showed that genes encoding defence regulators were enriched in all conditions (0, 45 and 90 min additional FR light). In addition, we found that genes involved in the biosynthesis of tryptophan were down-regulated in the shaded environments. Since auxin is mainly synthesised from tryptophan (Tao et al, 2008; Won et al., 2011), the latter finding may explain why athb2 mutant plants have defects in shade-induced growth promotion (Ciarbelli et al., 2008; Schena et al., 1993).

ATHB2 controls elongation growth and root development in response to shade

In shade-avoidant plants, such as A. thaliana, shade caused by neighboring plants elicits a set of growth responses including elongation of stem-like structures, hyponasty and leaf extension, and acceleration of flowering (Martinez-Garcia & Rodriguez-Concepcion, 2023). It has been demonstrated that ATHB2 plays a central role in the shade avoidance response, with transgenic Arabidopsis plants where ATHB2 is affected displaying varying degrees of shade response defects, the most obvious being in the elongation of the hypocotyl (Steindler et al., 1999). Early studies found that when knocked down or overexpressed, ATHB2 was associated with subsequent shade responses where there was less or more hypocotyl elongation compared to the WT, respectively. For this reason, and because we show that the two proteins interact, we expected that transgenic plants where ATHB2miP (35S::miP) is overexpressed would show a phenotype indicative of lost ATHB2 function, i.e. shade insensitivity. To better understand the role of the region corresponding to ATHB2miP, we also created mutant plants by CRISPR-Cas9 gene editing to compare with the overexpression and KO line already described (Fig. 4A). The first, athb21LZ, presents a loss-of-function allele of ATHB2 caused by a 25-bp deletion in the third exon, causing a frameshift and nonsense mutation, and resulting in a protein product lacking the leucine zipper and part of helix III of the homeodomain. Secondly, athb21, is a complete loss-of-function of ATHB2, caused by a frameshift mutation in the beginning of the first exon, resulting in a truncated protein that shares only its first twelve amino acids with ATHB2. Based on RT-qPCR results where we analyzed expression of ATHB2 after 45 minutes in shade, we see that transcription is not affected in any of the CRISPR lines we have generated, unlike in t-athb2 where ATHB2 is clearly downregulated (Fig. 4B), as also shown in the RNA-seq analysis (Fig. S2). To measure responsiveness to shade, we grew these mutant and transgenic plants alongside the Col-0 wild-type (WT) under three different shade regimes (Fig. 4C). We grew plants in deep shade (13 µmol m-2 s-1 PAR in WL and WL+FR; R:FR = 0.13 in WL+FR) with reduced PAR and a very low R/FR-ratio. In addition, we used canopy shade (45 µmol m-2 s-1 PAR in WL and WL+FR; R:FR = 0.15 in WL+FR) and proximity shade (45 µmol m-2 s-1 PAR in WL and WL+FR; R:FR = 0.06 in WL+FR) with higher PAR but different F/FR-ratios. In deep shade, we observed that athb2 mutants were hypersensitive, with elongated hypocotyls in white light conditions and, as in the case of 35S::miP, with very long hypocotyls in shade. Interestingly, the CRISPR mutants athb21 and athb21LZ exhibited responses similar to those of wild-type plants. It is worth noting that ATHB2 is still expressed in these mutants (Fig. 4B), which suggests that ATHBmiP and other transcript or protein isoforms may still be produced. Under canopy shade, the athb2 mutants exhibited a growth behaviour that was less sensitive to shade, resulting in reduced hypocotyl elongation in shaded conditions. However, under proximity shade, there were no significant differences in hypocotyl elongation between the wild type and athb2 mutant plants (Fig. 4C). The results indicate that ATHB2 has a dual function. In deep shade, it acts as a growth repressor, whereas in canopy and proximity shade, it acts as a growth promoter.

ATHB2 and ATHB2mip mediate hypocotyl elongation and root growth.

(A) Graphic representation of the gene structure of ATHB2. Location of T-DNA insertion (SALK_106790) is shown and below that the predicted protein structure of WT, transgenic and mutant isoforms of ATHB2. Blue and red triangles indicate Cas9 targeted sites. (B) RT-qPCR of ATHB2 in mutant and transgenic plants with and without WL+FR treatment. Coloured regions correspond to those in part A. (C) Plots of hypocotyl length in plants after 5 days growth in three different shade regimes: deep shade, canopy shade, and proximity shade. (D) Plot of the number of lateral roots in plants after 11 days of growth in WL or WL+FR (proximity shade). Letters in all signify results of pairwise comparisons by two-way ANOVA.

Class II HD-ZIPs have been suggested to have some redundant function in the shade avoidance response, so it is possible that ATHB2miP also interacts with these other proteins in the absence of full-length ATHB2. To determine whether the effect of 35S::miP depends on the presence of the endogenous copy of ATHB2, we complemented the t-athb2 knock-out with 35S::miP. In deep shade, we observed that both 35S::miP and 35S::miP t-athb2 plants developed longer hypocotyls under white light, indicating that ATHB2miP may have inhibited other transcription factors. Additionally, 35S::miP plants had significantly longer hypocotyls than 35S::miP t-athb2 plants in deep shade, suggesting that the absence of full-length ATHB2 may contribute to this difference.

The shade avoidance response is also associated with changes in root architecture. For example, one study found that lateral root density was decreased in plants that were exposed to longer periods of shade (van Gelderen et al, 2018). Mutations affecting ATHB2 have been reported to affect root development, disrupting both root length and lateral root formation (Schena et al., 1993), and we also identified a number of genes related to root morphogenesis in our RNA-seq analysis. Therefore, we questioned whether ATHB2miP could also have a role in root development.

Following the shade regimes described earlier, plants were grown in either WL+FR or WL for a period of 11 days and primary root length was measured. Our results indicate that there were no significant differences in all conditions tested, except for the t-athb2 mutant plants which had slightly shorter roots in white light and even shorter roots in the proximity shade regime (Suppl. Fig. S4). We also assessed the ability of plants to form lateral roots under light and shade conditions. While no difference in the number of lateral roots respective to geneotypes and environmenal conditions was observed in deep shade and canopy shade, we detected significant differences in proximity shade conditions (Fig. 4D). In white light, t-athb2 mutants, CRISPR mutants (athb21 and athb21LZ), and 35S::miP transgenic plants all all had significantly lower numbers of lateral roots, suggesting that ATHB2 promotes lateral root emergence broadly. In response to shade, a decrease in the number of lateral roots was observed in wild type plants. However, in t-abhb2, athb21 and athb21LZ mutant plants, the number of lateral roots did not change in shade. This indicates that ATHB2 is required to repress lateral root emergence in shade. Ectopic expression of ATHB2miP severely affected lateral root development in shade, suggesting that both ATHB2miP and ATHB2 act to repress lateral root emergence in shade. The 35S::miP transgenic plants exhibited a strong suppression phenotype, which was reversed upon loss of ATHB2 in 35S::miP transgenic plants (t-athb2 35S::miP), indicating that a functional ATHB2 is required and that ATHB2miP and ATHB2 work together in a repressive complex.

Taken together, the results indicate that ATHB2 regulates hypocotyl elongation in response to shade in a shade-dependent manner. It is hypothesized that ATHB2 homodimers exist in balance with heterodimers of ATHB2 and ATHB2miP. The role of ATHB2miP is to sequester ATHB2 and prevent excessive inhibition of hypocotyl growth in early shade. Furthermore, it was observed that ectopic expression of ATHB2miP resulted in heightened sensitivity to deep shade in comparison to the other LOF alleles. This observation may suggest that ATHB2miP has the potential to interfere with other HD-ZIPs involved in shade avoidance, although its primary target is ATHB2. This effect is also evident in the root, where ectopic expression of ATHB2miP affects lateral root emergence in an ATHB2-dependent manner.

The shade avoidance response is influenced by nutritional status regulated by ATHB2

ATHB2 regulates the expression of genes involved in shade avoidance and root development (Fig. 3) and mutations in ATHB2 result in defects in these processes (Fig. 4). After observing an increase in the expression of genes involved in iron transport and metabolism (Supplementary dataset 2), we investigated whether iron content was altered in athb2 mutants and whether iron affected shade-induced hypocotyl elongation and root development.

Iron homeostasis is under the regulation of a transcriptional cascade involving genes encoding iron transporters and a family of small peptides involved in iron deficiency signalling (Spielmann et al, 2023). The basic helix-loop-helix family of transcription factors, specifically bHLH38, bHLH39, bHLH100, and bHLH101, have been identified as intermediaries in iron deficiency signaling for the iron transporter IRT1. They achieve this by dimerizing with FIT1, FER-LIKE IRON DEFICIENCY INDUCED TRANSCRIPTION FACTOR1 (Colangelo & Guerinot, 2004).

Transcriptome profiling of wild type, t-athb2 and 35S::miP seedlings revealed genes that were upregulated in both t-athb2 and 35S::miP plants (Fig. 3B), indicating that these genes are repressed by ATHB2. The iron deficiency cascade, which includes bHLH38, bHLH39, bHLH100, bHLH101, FIT1, the iron transporter IRT1, and the IRONMAN genes (IMA1, IMA2, and IMA3), was among the ATHB2 repressed genes and found to be upregulated in 35S::miP and t-abth2 seedlings (Fig. 5A,B). In white light conditions, a smaller set of genes, including bHLH38, bHLH101, IMA1, and IMA2, are already highly expressed (Fig. 5A). The remaining genes are upregulated after 45 minutes of additional far-red light (Fig. 5B), suggesting negative regulation by ATHB2 in response to shading. This regulation is uncoupled in t-athb2 and 35::miP plants.

ATHB2 and ATHB2miP have a role in iron transport regulation and homeostasis.

(A) Heatmap showing the expression levels in white light of genes upregulated in both the t-athb2 mutant and the 35S::miP lines that are involved in iron transport, iron homeostasis and iron deficiency response. Lowly expressed genes are coloured blue, highly expressed genes are coloured red. Additionally, the expression level for all three replicates per genotype is displayed. (B) Heatmap showing the gene expression levels in shade conditions. (C) Concentration of Fe (mg/g) measured in the roots (grey) and the shoots of seedlings of the three genotypes (green). (D) Plots of hypocotyl length (red) and root length (blue) of plants grown on low or high iron supplemented ½ MS media in WL and WL+FR. Hypocotyls were measured after five days in either condition whilst roots were measured after nine days.

To determine whether the observed increases in transcription of genes encoding iron signaling and uptake components are related to iron levels, we measured the iron content of wild type, 35S::miP, and t-athb2 mutant seedlings. Iron is taken up by the roots and then distributed throughout the plant, and we therefore decided to measure iron levels in both roots and shoots. As anticipated, higher concentrations of iron were found in the roots than in the shoots (Fig. 5C). Both t-athb2 and 35S::miP transgenic plants exhibited significantly higher levels of iron in both roots and shoots. In comparison, 35S::miP plants had the highest internal iron concentration, which can be explained by the ability of ATHB2miP to affect class II HD-ZIP transcription factors beyond ATHB2. These findings collectively support the role of ATHB2 as a negative regulator of iron uptake.

To test whether nutrient status affects root development and shade avoidance responses, we performed shade avoidance assays with seedlings grown on low iron (10μM) and high iron (100μM) media. Overall, we observed a significant hypocotyl shortening and reduced hypocotyl response when seedlings were grown on high iron media compared to low iron media. The responses to deep shade on low-iron medium were similar to those observed earlier (Fig. 4C), except that t-athb2 mutants did not elongate hypocotyls under white light conditions. On high iron media, we observed a dampening of the shade avoidance response, but t-athb2, athb21 and athb21LZ mutants showed longer hypocotyls than wild type under shade conditions. Transgenic plants ectopically expressing ATHB2miP (35S::miP and t-athb2 35S::miP) showed slightly elongated hypocotyls in white light and almost no elongation in shade. These results show that nutrient status has a strong influence on the shade avoidance response and that the function of ATHB2 is dependent on iron homeostasis. Under low iron conditions, ATHB2 acts as an activator of shade growth, whereas under high iron conditions it suppresses shade growth. With regard to root growth, we measured the length of the primary root on low and high iron media in the absence and presence of shade. Overall, we did not observe strong changes in root elongation between white light and shade grown seedlings, supporting what we had observed previously (Fig. S4). However, we observed that all athb2 mutant plants, with the exception of athb21 mutants, developed significantly shorter roots when grown on low-iron medium. This finding may indicate that these mutants, in which iron uptake is strongly up-regulated, do not need to grow longer roots to increase their iron uptake. The latter hypothesis is supported by the finding that plants ectopically expressing ATHB2miP have the shortest roots but show the greatest hypocotyl elongation in shade. On high-iron medium, we did not observe major changes in root length, except for athb21 mutants, which had significantly shorter roots in both white light and shade conditions. In conclusion, these results show that nutrient status affects both root length and hypocotyl elongation in response to shade. ATHB2 mediates between root growth and hypocotyl elongation and can either activate or repress hypocotyl elongation in response to different iron concentrations.

Discussion

In this study, we assessed the impact of shading on transcript isoform production and their potential to encode microProteins. It was found that almost half of the Arabidopsis genes that had reads at the transcription start site also had alternative transcription start sites in the gene body. Only a fraction of these genes had the potential to encode microProteins. It is furthermore fair to assume that many more potential microProteins exist, but these are produced in a tissue-specific and/or inducible fashion. We filtered the dataset to prioritize alternative transcripts with robust expression levels, excluding those unsupported by multiple reads, which again could be candidates that have a narrower pattern of expression. Our approach focused on enriching microProtein candidates likely to have biological significance. To enhance the likelihood of identifying microProteins involved in the shade avoidance response, we considered alternative methods for studying transcript isoforms, including parallel analysis of RNA 5′ends from low-input RNA (nanoPARE), and ribosome profiling (Ingolia, 2014; Schon et al, 2018). Ribosome profiling, which employs deep sequencing to monitor in vivo translation, can provide additional insights into protein formation. Such future approaches can both confirm the purely sequencing-based identification of candidate microProteins but also aid to identify novel candidates.

The list of microProtein candidates generated through the 5’PEAT-seq approach contained few shade avoidance regulators. Notably, microProteins may serve distinct functions compared to their full-length counterparts, albeit often targeting evolutionarily related proteins. Adjusting filtering criteria could identify more transcript isoforms related to shade responses, although they may not classify as microProteins. The ATHB2miP microProtein, chosen due to its similarity to the known LITTLE ZIPPER microProteins (Kim et al., 2008; Wenkel et al., 2007), was expected to interact with ATHB2 and regulate its activity, akin to ZPRs and HD-ZIPIIIs. Yeast-Two-Hybrid tests and pull-down assays confirmed physical interactions between ATHB2 and ATHB2miP. Notably, ATHB2miP can also homodimerize, a feature attributed to the leucine zipper domain’s role as a protein-protein interaction motif in HD-ZIP proteins.

We assessed potential alternative ATHB2miP promoters using the GUS protein assay, identifying one active promoter, pATHB2miP::GUS, which did not possess a typical ATG start codon. This observation is consistent with prior findings of alternative start sites not commencing with an ATG. Differences in GUS expression patterns between ATHB2 and ATHB2miP promoters were observed, with ATHB2miP primarily present in the shoot apical meristem (SAM) and early leaf primordia. The absence of ATHB2miP in other tissues suggested that it does not directly regulate ATHB2 in those areas.

Analysis of the subcellular localization patterns of ATHB2miP and ATHB2 protein-fusions to GFP revealed that ATHB2 localizes mostly to larger subnuclear speckles, a phenomenon also observed for phytochromes and other light signaling components and therefore often referred to as photobodies (Kim et al, 2023; Kircher et al, 2002). Interestingly, ATHB2miP protein showed a more diffuse nuclear localization implying discrete functions of both isoforms. However, when co-expressed both isoforms appear in speckles, but these are more dispersed, indicating a direct physical interaction where the full-length isoform is partially removed from the speckles. The full-length ATBH2 isoform has an intrinsically disordered amino terminus and could undergo phase separation and engage in more flexible protein-protein interactions within the photobodies. As ATHB2miP lacks the disordered part, it may prevent the phase separation activity of ATHB2.

To investigate the regulatory role of ATHB2miP, we compared ATHB2miP overexpression plants to athb2 mutants. Surprisingly, in deep shade conditions all athb2 knock-down and knock-out mutants as well as ATHB2miP overexpression plants showed elongated hypocotyls in white light conditions compared to Col-0 wild type plants while the hypocotyl length in shade grown seedlings showed no significant differences when compared to wild type. Except 35S::miP transgenic plants that showed excessive hypocotyl elongation in both white light and far-red enriched conditions (Fig. 4C). This indicates that ATHB2 functions as a growth repressor in deep shade and ATHB2miP has the capacity to inactivate ATHB2 which results in hypocotyl elongation in shade. The finding that the growth differences can only be observed in deep shade conditions indicates that ATHB2 activity is dependent on other components that are only present in this type of shade. In canopy and proximity shade ATHB2 functions as a promoter of elongation growth, hence the shorter hypocotyls in these types of shade.

ATHB2 has also been reported to be involved in controlling root growth and development (Schena et al., 1993) prompting us to investigate the role of ATHB2 and ATHB2miP in this process. The analysis of root development in different types of shade did not reveal significant differences in root length. However, there were notable differences in the ability of athb2 mutants to produce lateral roots. Additionally, all tested athb2 mutants appeared to be insensitive to shade, as the number of lateral roots did not change in response to proximity shade, unlike in the wild type. In summary, these results suggest that ATHB2 acts as a suppressor of lateral root growth in response to proximity shade. The overexpression of ATHB2miP had a significant impact on lateral root development in both white light and shade. Notably, in the t-athb2 mutant background (t-athb2 35S::miP), the strong shade suppression was not observed, indicating that ATHB2 is required for strong shade suppression. These findings suggest that ATHB2miP has a role beyond ATHB2 inhibition. The discovery that ATHB2 has diverging functions depending on the type of shade experienced by the plant raises questions about the upstream signalling processes that control ATHB2 activity and the type of protein complexes in which ATHB2 participates. Additionally, the localization of ATHB2 to photobodies hints at a role in phytochrome B signalling. In deep shade, phyA may have additional roles that cause ATHB2 to switch from being a growth activator to being a growth inhibitor. To better understand ATHB2’s molecular functions in different light conditions, it is important to study the protein complexes it interacts with.

To elucidate the molecular events resulting from the absence of ATHB2 or the inhibition of ATHB2 by overexpression of ATHB2miP, we conducted an RNA-sequencing analysis comparing Col-0 wild type with t-athb2 and transgenic plants overexpressing ATHB2miP (35S::miP), all grown under both white light and deep shade conditions. In line with the exaggerated growth phenotype of the hypocotyl in deep shade, we found auxin- and shade-related genes to be expressed at elevated levels. The fact that the expression of these genes follows the same trend in both the t-athb-2 and the 35S::miP line also supports the negative feedback mechanism we hypothesised. In addition, we also found genes involved in root growth and development to be over-represented in white light grown seedlings, which corresponds to what we observe in our lateral root phenotyping results. Genes involved in the formation of lateral roots were found to be upregulated in both the t-athb2 and the 35S::miP seedlings, even in shade. Using gene clustering analysis, especially cluster 2 (Fig. 3B) comprises genes that are down-regulated in both the t-athb2 and the 35S::miP lines and several of these genes are involved in root growth and development. Surprisingly, genes that were upregulated in t-athb2 and ATHB2miP overexpression plants encode iron transporters. The finding that iron concentrations were elevated in these plants implies that these roots might have stopped growing because of a better nutrient supply. Our findings show that iron availability influences the shade avoidance response and primary root length. Furthermore, it appears that the nutritional status influences the activity of ATHB2, suggesting that ATHB2 is integrating different environmental inputs (shade and nutrient availability) to formulate an appropriate response that can either activate or repress growth.

Shade not only promotes elongation growth, but it also affects the plant’s hormone homeostasis, weakening its ability to defend against pathogens (de Wit et al, 2013). The recent discovery that regulation of iron uptake is influenced by internal IRONMAN peptides, which are sensitive to external flagellin perception (Cao et al, 2024), raises the question of whether ATHB2 may play a central role in mediating between nutritional immunity and shade-induced growth promotion or suppression.

In summary, our work uncovered transcript isoforms that are responsive to shading, including ATHB2miP, a microProtein derived from an alternative transcription start site in the ATHB2 gene. This microProtein appears to intercede between root and shoot growth in response to shade. In summary, our data support a model for ATHB2/ATHB2miP as integrator of nutrient status and elongation growth of both the shoot and the root.

Materials and methods

Plant materials and growth conditions

All plants were grown in growth chambers on ½ Murashige and Skooge (MS) 1% agar under long day (LD) conditions at 22°C during daytime and 20°C during night. For the shade treatment for the 5’PEAT-seq and RNA-seq experiments, seedlings were transferred after 10 days from a white light only compartment (PAR of 13 μmol/m2/s, R:FR = 7.7) to a compartment with additional far-red lights (PAR of 13 μmol/m2/s, R:FR = 0.2) for treatments of 0, 45 and 90 minutes. Whole seedlings frozen in liquid N2 before grinding with a mortar and pestle and RNA extraction. For the shade treatments related to measuring hypocotyl length and root growth, half of the plants were transferred after two days from a white light only compartment (PAR of 45 μmol/m2/s, R:FR = 5.5) to a compartment in the same chamber with additional far-red lights (PAR of 45 μmol/m2/s, R:FR = 0.15). The other half remained in the white light only compartment. After five and eleven days, seedlings were photographed and hypocotyls and roots measured using IMAGEJ.

For the growth and phenotyping of seedlings on iron-enriched and iron-depleted media, two different growth media were prepared. Both media contained the same vitamins, macro and micro elements of standard MS salt (M0222, Duchefa) and in the same concentrations, with the exception of FeNAEDTA, used in a concentration of 10uM in the iron-depleted medium and 100uM in the iron-enriched medium. Seedlings were grown on 1% agar iron-depleted and iron-enriched media under the same growth conditions of the other growth experiments. Half of the seedlings on plates were transferred after two days from a white light only compartment (PAR of 13 μmol/m2 /s, R:FR = 7.7) to a compartment in the same chamber with additional far-red lights (PAR of 13 μmol/m2 /s, R:FR = 0.2). The other half remained in the white light only compartment. After five and eleven days, seedlings were photographed and hypocotyls and roots measured using IMAGEJ.

5’-PEAT library preparation and analysis

Arabidopsis Col-wild type seedlings were grown for 10 days on 1/2 MS agar plates as described before. Whole seedlings were harvested directly into liquid nitrogen after 0, 45 or 90 minutes of shade treatment. Total RNA was extracted with Spectrum Plant Total RNA Kit (Sigma). PEAT library preparation was done exactly as described earlier (Ni et al., 2010), but all steps beginning with the circularization of the PCR product were omitted and the fragments were inserted into NEBNext Ultra™ II DNA Library Prep K it for Illumina and quantified with NEBNext Library Quant Kit for Illumina. The finished libraries had a fragment size of 500 bp. BGI tech solutions (Hongkong) performed sequencing on an Illumina HiSeq (150bp paired end, 40 M read pairs per sample).

Quality control was done with FastQC (v0.11.5), available online at: http://www.bioinformatics.babraham.ac.uk/projects/fastqc) and generic Illumina adapters were removed with Cutadapt (Martin, 2011). Additionally, Cutadapt removed PEAT-adapters and marked read direction for later identification of the RNA start. HISAT2 (v2.0.5) (Kim et al., 2015) aligned the reads with more than 90% success to the Arabidopsis thaliana TAIR9 genome assembly (Lamesch et al, 2012). Samtools (v0.1.19) (Li et al, 2009) and bedtools (v2.17.0) (Quinlan & Hall, 2010) filtered for properly paired reads with high aligment score (-q 10) and performed file conversions. Around 30 million read pairs per sample were further analysed. Using previously marked read directions, RNA starts were identified and recorded in bedgraph format. Only genes with a transcription start site (TSS) in the 5’UTR or upstream region and at least one alternative TSS in coding sequence or intron region were further analysed. The difference between the TSS and the alternative start site was set to 10%. To test whether the regulation of the 5’UTR-TSS and the alternative TSSs are different, the number of TSS-reads in the 5’UTR-TSS and one alternative TSS were compared in two samples with Fisher’s exact test and Benjamini-Hochberg corrected false discovery rate in R software (v3.2.2). The protein size was set between 7 and 20 kDa. Pfam-A domains (v28, ftp://ftp.ebi.ac.uk/pub/databases/Pfam/) (Finn et al, 2016) were searched in protein sequences with hmmer3 (http://hmmer.org/) and cutoffs of e-value <= 0.1 and c-value <= 0.05. Interaction domains were identified using iPfam (v1.0, http://www.ipfam.org/) (Finn et al, 2014).

Generation of transgenic and CRISPR-Cas9 edited plants

To generate the overexpression of ATHB2miP (35S::ATHB2miP) in Arabidopsis oligos were designed to amplify exons 3 and 4 of ATHB2 and the resulting sequence then cloned into the vector pJAN33 by gateway cloning. Plants were transformed by floral dipping and transformants selected for by BASTA treatment.

For CRISPR mutagenesis, sgRNAS were cloned into the CRISPR/Cas9 vector pKI1.1R. The cloning was performed as described (Tsutsui and Higashiyama, 2016). Athb2-cr2 was made by Cas9 gene editing targeted using the following sgRNA combination: 5’-TACAGTCTCAAGCTCTACA-3’ and 5’-GTTTCAGAACAGACGAGCA-3’. Athb2-cr3 was made by Cas9 gene editing targeted using the single sgRNA: 5’-ACGATCTGGGTCTAAGCTT-3’. Plants were transformed by floral dipping and then RFP-positive seeds selected for in the T1 generation. Genomic DNA of 4-week old plants was isolated using the Edwards method and the plants were genotyped by PCR. The PCR products were gel purified using the E.Z.N.A® Cycle-Pure Kit (Omega Bio-tek) and sequenced at Macrogen (Amsterdam).

Expression in transgenic and mutant plants was measured by RT-qPCR. First, RNA was extracted from 4-week old plants using the Spectrum Plant Total RNA Kit (Sigma). Purified RNA (1 μg) was used for reverse transcription using the iScript™ cDNA Synthesis Kit (BIO-RAD). RT-qPCR was performed using SYBR green (ThermoScientific) on a Biorad CFX384, using oligos against ATHB2: Region A FWD 5’-GATTGCAAAATCTCTCTCTCTCTC-3’ with REV 5’-GACCCAGATCGTCTTTCTCG-3’; Region B FWD 5’-TGTTCGAGAAAGACGATCTGG-3’ with REV 5’-GTCGATTCCTCGGATGAAAG-3’; Region C FWD 5’-CAGTGACGATGAAGATGGTGA-3’ with REV 5’-GAAACCAAACTTCCACTTGTCT-3’; Region D FWD 5’-GCTGAAGCAAACGGAGGTAG-3’ with REV 5’-GGACCTAGGACGAAGAGCGT-3’. Gene expression levels were standardized to the housekeeping gene GAPDH (FWD 5’-AAAGTGTTGCCATCCCTCAA-3’ with REV 5’-TCGGTAGACACAACATCATCCT-3’) using the Pfaffl equation.

Yeast-2-Hybrid

For the Y2H assay, the coding sequences of ATHB2 and ATHB2miP (exons 3 and 4 of ATHB2, as above) were recombined into vectors pGBKT7-GW and pGADT7-GW by gateway cloning. The constructs were transformed to yeast strains PJ694α and YM4271A using lithium acetate and heat shock for 15 min (42°C). Yeast colonies that grew after two days at 30°C on SD dropout medium plates lacking either tryptophan (−W) or leucine (−L) were selected for, mixed with each other and spread on SD dropout medium lacking tryptophan and leucine (−LW). Colonies were finally plated on dropout −LW medium that additionally lacked histidine (−LWH), with additional 10 mM 3-aminotriazole to test for protein interactions.

GUS localization assay

Oligos were designed to amplify the 1 kb sequence upstream of ATHB2, and the sequences upstream of the alternative start codon TTG in exon 3 and upstream of a control start codon ATG in intron 2, up to and including the putative start codon ATG of ATHB2. Sequences were cloned into pBGWFS7 by gateway cloning and the resulting vectors transformed into plants by floral dipping. Arabidopsis thaliana (ecotype Col-0) seeds were surface-sterilized and sown on Murashige and Skoog (MS) agar plates. The plates were incubated at 4°C for 2 days for stratification and subsequently transferred to a growth chamber under long-day conditions. Seven day old seedlings were immersed in GUS staining solution (XGLUC; 14-mM NaH2PO4, 36-mM Na2HPO4, 2-mM K4[Fe(CN)6], 2-M K3[Fe(CN)6], 10% Triton X-100). Seedlings were vacuum infiltrated for 5 minutes, incubated overnight at 37 °C in the dark, and rinsed with PBS to remove excess staining solution and dehydrated through an ethanol series (30%, 50%, 70%, 90%, and 100%) to remove chlorophyll and enhance contrast.. Pictures of seedlings were taken with a canon EOS 450D camera on a Nikon Eclipse 80i Fluorescence microscope. White balance and scalebar were added via ImageJ.

Subcellular localization and FRET-FLIM analysis

As above, the coding sequences of ATHB2, ATHB2miP and ATHB2HD (from exon 1 to the beginning of exon 4) were cloned into the vectors pK7WGF2 that contains an N-terminal eGFP driven by a CaMV35S promoter and a modified vector of pEarlyGate104 containing an N-terminal mCherry driven by a CaMV35S promoter (provided by Sabine Müller/Dorothee Stöckle, ZMBP Tübingen). Tobacco leaves were transiently transformed by agrobacterium infiltration (three individual plants per infiltration) and nuclei of the epidermal cells of leaves were imaged 2-3 days post-infiltration using a Leica Stellaris 8 confocal laser-scanning microscope. All vectors were co-transformed with p19 and for the FRET-FLIM assay, eGFP-ATHB2 was additionally co-transformed with either mCherry, mCherry-ATHB2, or mCherry-ATHB2miP. Fluorescence lifetime measurements were collected using the TauInteraction tool in the LAS X software; samples were excited with a 470 nm pulsed laser (10 MHz) and emission from 500 nm to 560 nm was recorded.

Differential Gene Expression Analysis and Clustering

Total RNA was extracted from Arabidopsis thaliana Col-0 wild-type (WT), t-athb2, and 35S::ATHB2miP genotypes under three different light conditions as described: White Light (WL), 45 minutes of Far-Red (FR) enrichment, and 90 minutes of FR enrichment. Three biological replicates per genotype-condition combination were used. RNA quality and concentration were assessed using a NanoDrop spectrophotometer and Agilent Bioanalyzer. RNA was sequenced on an Illumina HiSeq platform.

Raw sequencing reads were quality-checked using FastQC to ensure data integrity. Trimming of reads to remove adapter sequences and low-quality bases was performed using Trimmomatic. After preprocessing, high-quality reads were retained for downstream analysis. Read alignment to the Arabidopsis reference genome (TAIR10) was conducted using the STAR aligner within the R environment. This generated BAM files for each sample, which were subsequently used for quantification of gene expression. Differential gene expression analysis was carried out using the edgeR package in R. Raw counts were normalized using the TMM (Trimmed Mean of M-values) method. Genes with low expression were filtered out, and statistical analysis was performed to identify differentially expressed genes (DEGs) between genotypes and conditions. P-values were adjusted for multiple testing using the false discovery rate (FDR) method. DEGs were subjected to gene ontology (GO) enrichment analysis to gain insights into biological processes and pathways associated with the observed expression changes. The topGO package in R was used for this purpose.

The raw counts obtained from the RNA-seq analysis were used as input for the MFuzz clustering. These count data represented the expression levels of genes across the three genotypes and three light conditions.

For the clustering, the count data were first normalized (TMM method) to account for differences in sequencing depth and library size. MFuzz, an R package specifically designed for model-based clustering of high-dimensional gene expression data, was utilized for clustering analysis. The resulting 17 gene clusters were visually inspected to find the ones with the wished expression patterns across the samples. Genes belongind to clusters of interest were extracted and subjected to gene ontology (GO) enrichment analysis.

The reads were visualised on IGV (Integrative Genome Viewer) from the BAM mapping files generated.

Preparation of samples and iron concentration measurements

WT, t-athb2 and 35S::ATHB2miP seedlings (about 200 per genotype) were grown on standard MS medium for 12 days. They were carefully removed from the medium to avoid damage and the aerial part of each seedling was excised from the roots. In all steps, no iron nor glass tools were used. The roots and shoots were rinsed in a 2 mM CaCl2 solution to remove iron accumulated on the surface. The material was placed in an Eppendorf tube and dried in a stove at 70°C for 3 days. Each of the dry samples was weighed (dry root weights ranged from 2 to 6 mg, seedlings from 15 to 25 mg). Samples were digested in 2.5 mL of 70% HNO3 + 1 mL of H2O2. Digestion conditions: maximum temperature = 240°C; maximum pressure = 200 bar; cycle duration: 1-2 hours. Following digestion, the samples were diluted in ultrapure Milli-Q water to achieve a [HNO3] = 3.5%. The liquid samples were then analyzed using an ICP-QQQ-MS 8800 Agilent instrument. NIST certified reference materials were employed to verify the accuracy of the data.

Acknowledgements

We thank Jaume Martinez-Garcia for comments on the manuscript.

Additional information

Funding

We acknowledge funding through NovoCrops Centre (Novo Nordisk Foundation project number 2019OC53580 to S. W.), the Independent Research Fund Denmark (0136-00015B and 0135-00014B to S. W.) and the Novo Nordisk Foundation (NNF18OC0034226 and NNF20OC0061440 to S. W.).

Author contribution

Conceptualization: Stephan Wenkel.

Data curation: Daniel Straub, Maurizio J. Chiurazzi.

Formal analysis: Ashleigh Edwards, Maurizio Junior Chiurazzi, Anko Blaakmeer, Ylenia Vittozzi, Ashish Sharma, Sanne Matton, Valdeko Kruusvee, Daniel Straub, Giovanna Sessa, Monica Carabelli, Giorgio Morelli, Stephan Wenkel

Funding acquisition: Stephan Wenkel.

Investigation: Ashleigh Edwards, Maurizio Junior Chiurazzi, Anko Blaakmeer, Ylenia Vittozzi, Ashish Sharma, Sanne Matton, Valdeko Kruusvee, Daniel Straub, Giovanna Sessa, Monica Carabelli

Methodology: Daniel Straub, Maurizio Junior Chiurazzi, Valdeko Kruusvee

Supervision: Stephan Wenkel.

Validation: Daniel Straub, Valdeko Kruusvee, Giovanna Sessa, Monica Carabelli, Giorgio Morelli

Writing – original draft: Ashleigh Edwards, Maurizio Junior Chiurazzi, Anko Blaakmeer

Writing – review & editing: Giovanna Sessa, Monica Carabelli, Giorgio Morelli, Stephan Wenkel.

Additional files

Supplementary dataset 1. Alternative transcripts encoding potential microProteins discovered by 5’PEAT seq.

Supplementary dataset 2. Differentially expressed genes in wild type, t-athb2 and 35S::miP seedlings exposed to 45 and 90 minutes of shade.

Supplementary dataset 3. Gene clusters identified by RNA-seq.

Data availability

All sequencing data have been deposited in NCBI’s Gene Expression Omnibus under GEO Series accession no. GSE241976, comprising RNA-seq (GSE241974) and 5’PEAT-seq (GSE241975).

Note

This reviewed preprint has been updated to use the correct text; previously, due to a productione error, content from an unrelated submission was presented here.

Supplementary figures

Identification of an alternative ORF in the coding sequence of ATHB2.

(A) 5’RACE carried out on 10 day old seedlings that were exposed to 0, 45 and 90 mins of WL+FR light. Red asterisk signifies the small transcript arising from ATHB2 that when purified and sequenced corresponds to the underlined portion of ATHB2 in part B. (B) Genomic sequence of ATHB2. The exonic sequence with protein sequence below is shown in bold. Double-sided blue arrows denote sites corresponding to 5’PEAT-seq read clusters. Blue highlighted sequence denotes the leucine zipper domain. Red circles indicate two alternative non-canonical start codons identified by TIS predictor. Underlined sequence corresponds to the transcript purified from 5’RACE in part A.

RNA-seq reads corresponding to ATHB2 at 0, 45 and 90 mins of WL+FR.

RNA-seq reads peak in the WT at 45 mins of WL+FR and then decrease (see y axis for maximum reads) at 90 mins. No reads are detected in t-athb2 mutants whereas 35S::miP mutant plants are highly overexpressing exon 3 and 4 of ATHB2.

(A) GO enrichment analysis performed on the 95 DEGs whose expression is shown in Figure 3 yielded several significant GO terms (FDR < 0.05). These GO terms are related to shade avoidance, hormone/auxin responses, and growth. (B) The 17 clusters generated by MFuzz. The expression changes of the genes included in each cluster are on the y-axis. The three genotypes and shade conditions are on the x-axis.

ATH2B and ATHB2miP slightly effect root elongation.

Main root length of 11-day old plants grown in either WL or WL+FR in three different shade regimes: deep shade, canopy shade and proximity shade.