TurboID-based proximity labeling of meiotic chromosome axes in Arabidopsis thaliana

Abstract

During meiotic prophase I sister chromatids are arranged in a loop-base array along a proteinaceous structure, called meiotic chromosome axis. This structure is essential for synapsis and meiotic recombination progression and hence formation of genetically diverse gametes. Proteomic studies in plants aiming to unravel the composition and regulation of meiotic axes are constrained by limited meiotic cells embedded in floral organs. We report TurboID (TbID)-based proximity labeling (PL) in meiotic cells of Arabidopsis thaliana. TbID fusion to the two meiotic chromosome axis proteins ASY1 and ASY3 enabled the identification of their proximate “interactomes” based on affinity-purification coupled with mass spectrometry. We identified 39 ASY1 and/or ASY3 proximate candidates covering most known chromosome axis-related proteins. Functional studies of selected candidates demonstrate that not only known meiotic candidates but also new meiotic proteins were uncovered. Hence, TbID-based PL in meiotic cells enables the identification of chromosome axis proximate proteins in A. thaliana.

Introduction

Reshuffling of genetic information between homologous chromosomes creates variable gametes during meiosis. Genetic variation is assured by the repair of programmed DNA double strand breaks (DSBs) either as crossover (CO) or non-crossover (NCO) during meiotic recombination¹.

Meiotic recombination occurs in the context of the meiotic chromosome axis^{2, 3, 4} a proteinaceous structure along which sister chromatids are arranged in a loop-base array during leptotene. In A. thaliana, the chromosome axis is composed of the HORMA domain protein ASY1 (Hop1 in S. cerevisiae)^{5, 6}, the coiled-coil proteins ASY3 (Red1 in S. cerevisiae) and ASY4^{7, 8} as well as cohesin/cohesion-associated factors such as REC8 (also known as SYN1 or DIF1)⁹. In budding yeast and A. thaliana data suggests that DSBs are formed in loops tethered to the axis and most recombination proteins are found axis-associated implying meiotic chromosome axis as a scaffold for meiotic recombination^{4, 10, 11}. The axis-associated yeast proteins Hop1 and Red1 are critical for meiotic DSB formation¹², whereas in A. thaliana only ASY3 and not ASY1 is required for this process^{5, 7}. In plant meiotic axis mutants, the formation of COs is impaired highlighting the role of the chromosome axis for meiotic recombination^{13, 14, 15, 16}.

By zygotene with the installation of the transverse filament protein ZYP1 (ZYP1a/b in A. thaliana; Zip1 in S. cerevisiae)^{17, 18, 19}, aligned homologous chromosomes start to get physically connected along their length culminating during pachytene in the formation of the synaptonemal complex (SC). Concurrently with ZYP1 installation, ASY1 gets progressively depleted from synapsed regions. The conserved AAA+ ATPase PCH2/TRIP13 is a key player during the remodeling process^{20, 21, 22} with COMET likely acting as an adaptor between PCH2 and ASY1²³. In addition, similar to other species^{24, 25, 26} also post-translational modifications (PTMs) of chromosome axis components such as phosphorylation of ASY1 at 142T²⁷ are critical for chromosome axis assembly in Arabidopsis. Phosphorylation of axis proteins seems conserved in plants as in Brassica oleracea multiple phosphorylated residues in ASY1 and ASY3 are found²⁸ and in rice PAIR2 (homolog of ASY1) is phosphorylated²⁹. Interestingly, ASY1 acts as a dosage-dependent suppressor for telomere-led COs³⁰ and in autotetraploid Arabidopsis distinct ASY1 and ASY3 alleles contribute to the stabilization of newly formed autotetraploids^{31, 32}. Due to the key role of axis proteins such as ASY1 and ASY3 for meiotic fidelity including frequency and distribution of COs, further insights into the composition and regulation of plant meiotic chromosome axes are of interest. However, proteomic studies using A. thaliana meiocytes are challenging due to the limited number of meiotic cells embedded within non-meiotic flower tissues. To circumvent some of the proteomic limitations in A. thaliana, ASY1-affinity proteomics were performed in B. oleracea that identified 589 proteins co-precipitating with ASY1 and multiple phosphorylated residues within ASY1 and ASY3²⁸. However, given the large number of identified candidates, assessing their significance based on functional studies is challenging.

More recently proximity-dependent biotin identification (BioID) enabled the identification of proximate protein “interactomes”³³. BioID is based on the fusion of a protein of interest (POI) to a promiscuous biotin ligase that catalyzes biotin into biotinoyl-5’-AMP with ATP as second substrate. In a distance-dependent manner (radius ^~10 nm³⁴) diffusing biotinoyl-5’-AMP can react with lysine residues of proximate proteins leaving the proteins labeled with biotin³⁵. Biotinylated proteins can be affinity purified under harsh lysis conditions reducing possible false-positive protein interactions post cell lysis. After the initial biotin ligase BirA* for in vivo biotin labeling³⁶ improved versions were reported^{37, 38}. More recently, two engineered biotin ligases, TurboID (TbID) and miniTurbo (mTb), increased in vivo biotinylation efficiency even further with biotin labeling times decreased to as few as 10 min³⁹, suggesting their potential to identify also transient or weak (proximate) protein interactions. So far, TbID has been employed in several species including plants in distinct tissues, cell types, or cellular compartments^{39, 40, 41, 42} but not to identify the proximate “interactome” of a meiotic POI.

Here we report the application of TbID-based proximity labeling for the two meiotic chromosome axis-associated proteins ASY1 and ASY3 to identify their proximate “interactomes” in A. thaliana. Both ASY1-eYFP-TbID and ASY3-eYFP-TbID trigger efficient biotinylation in meiocytes and complement their respective male meiosis mutant phenotypes. 39 ASY1 and/or ASY3 proximate candidates were identified using control samples and stringent data filtering. Besides known meiotic chromosome axis-related proteins, proteins so far not described to play a role during meiosis were found. While most candidates did not directly interact with axis proteins in yeast two-hybrid (Y2H), T-DNA insertion mutants of selected candidates showed a clear meiotic phenotype including the loss of the obligate CO. Finally, meiotic in planta expression of two candidates was found.

Results

ASY1-/ASY3-eYFP-TurboID complement male meiosis in asy1/asy3

To test whether TbID-based proximity labelling of meiotic chromosome axes is feasible in A. thaliana, the meiotic chromosome-axis proteins ASY1 and ASY3 were selected as POIs^{5, 7}. A plant codon-optimized version of TbID was synthesized showing similar activity as the mammalian version in Arabidopsis protoplasts (Extended Data Fig. 1a, b). The codon-optimized TbID fused with eYFP was inserted in-frame of the genomic ASY1 and ASY3 sequences including their native promoter and terminator sequences. Resulting expression cassettes pASY1::ASY1-eYFP-TbID::tASY1 (ASY1-eYFP-TbID) and pASY3::eYFP-TbID-ASY3::tASY3 (ASY3-eYFP-TbID) were transformed into Arabidopsis asy1 and asy3 plants, respectively. Three further lines pASY1::ASY1-eYFP::tASY1 in asy1 (ASY1-eYFP), pASY3::eYFP-ASY3::tASY3 in asy3 (ASY3-eYFP) and pUBQ10::_NLSeYFP-TbID::tNos in WT (_UBQeYFP-TbID, constitutive nuclear expression of eYFP-TbID) were generated as controls.

To identify whether the presence of TbID interferes with the function of ASY1, ASY3 and/or meiotic progression, we phenotypically studied and compared ASY1-eYFP-TbID, ASY1-eYFP, ASY3-eYFP-TbID and _UBQeYFP-TbID with asy1, asy3 and WT plants. No differences in vegetative growth and development were found in any of the transgenic lines including plants constitutively expressing eYFP-TbID (Extended Data Fig. 2a). In terms of overall fertility (silique length and seeds per silique), all transgenic lines except ASY1-eYFP-TbID were similar to WT (Fig. 1a, b). In ASY1-eYFP-TbID seed numbers were ^~70% of WT levels compared with only ^~20% in asy1, suggesting partial complementation. Interestingly, in case of pollen viability assessed by Alexander staining, all transgenic lines including ASY1-eYFP-TbID were similar to WT (Extended Data Fig. 2b), suggesting WT-like male meiosis in all transgenic lines. To confirm this, male meiotic chromosome spread analysis was performed in all transgenic lines.

Fig. 1. Phenotype of ASY1-eYFP-TbID and ASY3-eYFP-TbID plants.

(a) Siliques of WT (Col-0), ASY1-eYFP (asy1), _UBQeYFP-TbID (Col-0), ASY1-eYFP-TbID (asy1), ASY3-eYFP-TbID (asy3), asy1 and asy3. Scale bar = 1 cm. (b) Quantification of seeds per silique, data are presented as mean ± SEM (ASY1-eYFP (53.35 ± 4.96) vs. Col-0 (51.63 ± 3.91), p=0.0921; _UBQeYFP-TbID (50.40 ± 6.98) vs. Col-0, p=0.3428; ASY1-eYFP-TbID (37.55 ± 5.73) vs. Col-0, p=1.28 x 10^-20; ASY3-eYFP-TbID (48.65 ± 6.10) vs. Col-0, p=0.0126; asy1 (8.68 ± 1.98) vs. Col-0, p=3.13 x 10^-54; asy3 (13.00 ± 2.19) vs. Col-0, p=2.58 x 10^-53; two-sided Student’s t test; n=40), (c) male meiotic chromosome behavior (scale bar = 10 μm; DNA counterstained with DAPI in gray), and (d) minimum chiasmata number (MCN) (Col-0, n=28; ASY1-eYFP, n=25; _UBQeYFP-TbID, n=36; ASY1-eYFP-TbID, n=34; ASY3-eYFP-TbID, n=27; p=0.4262; one-way ANOVA) in plants indicated in (a). Distinct plants from a single transgenic line (or WT) were used for seed counting or chromosome spread analysis. ***, P<0.0001. N.S., not significant.

In WT, homologous chromosomes are synapsed at pachytene, five bivalents are inevitably found at metaphase I and faithful meiotic chromosome segregation during meiosis I and II results in balanced dyads and tetrads, respectively. In asy1 and asy3, no normal pachytene nuclei and univalents (homologous chromosome pairs failed to form chiasmata) at metaphase I were found leading to unequal chromosome segregation and unbalanced dyads and tetrads (Fig. 1c). In all transgenic lines, male meiotic chromosome behavior and the minimum chiasmata number (MCN) were similar to WT (Fig. 1c, d). Hence, ASY1 and ASY3 fusion proteins independent of TbID presence functionally complement their corresponding male meiotic mutant phenotypes. Moreover, the presence of TbID in male meiocytes (even constitutive) does not interfere with meiotic progression and CO formation (Fig. 1c, d). However, partial fertility recovery in ASY1-eYFP-TbID with roughly 30% of ovule abortion (Extended Data Fig. 2c), suggests that fusion of eYFP-TbID but not eYFP to ASY1 may either not fully rescue asy1 female meiosis, despite expression of fusion proteins in female meiotic cells (Extended Data Fig. 2d), and/or may impact ovule development. To address this, we performed female meiotic chromosome spread analysis in ASY1-eYFP-TbID and ASY3-eYFP-TbID. While in ASY3-eYFP-TbID female meiosis appeared normal, i.e. five bivalents were found at diakinesis/metaphase I (n = 8), in case of ASY1-eYFP-TbID we frequently (^~40 % of cells, n = 18) found the occurrence of one or two pairs of univalents (Extended Data Fig. 2e) likely causing the reduced seed setting. Given that in Arabidopsis female meiosis occurs in older flower buds than male meiosis⁴³ and given that ASY1-eYFP-TbID fully rescues male meiosis in asy1, we collected inflorescence samples (the inner part of the inflorescence was isolated) with young flower buds containing only male meiocytes for all following experiments to assure in planta biotin labeling of WT-like meiotic chromosome axes.

We also tested mTb^{39, 41} for proximity labeling in meiocytes by generating pASY1::ASY1-eYFP-mTb::tASY1 in asy1 (ASY1-eYFP-mTb). Similar to ASY1-eYFP-TbID, ASY1-eYFP-mTb functionally complements asy1 male meiosis but it does not fully complement asy1 fertility (Extended Data Fig. 3a-e). Moreover, ASY1-eYFP-TbID outperformed ASY1-eYFP-mTb in terms of in planta biotinylation efficiency in meiocytes (Extended Data Fig. 3f, g). Therefore, we chose TbID for all further studies.

ASY1-/ASY3-eYFP-TurboID biotinylate meiotic chromosome axes

To determine whether ASY1 and ASY3 fusion proteins with or without TbID show similar dynamics during male meiosis, we performed meiotic live-cell imaging within intact anthers using light sheet fluorescent microscopy (LSFM)⁴⁴. Both ASY1- and ASY3-eYFP-TbID fusion proteins were found in prophase I nuclei for a duration of around 30 hrs (from nuclear appearance until nuclear envelope breakdown), respectively, similar to ASY1- and ASY3-eYFP (Extended Data Fig. 4a, b), suggesting that the presence of TbID does not alter temporal dynamics of ASY1 or ASY3. Constitutive nuclear expression of eYFP-TbID in meiotic and somatic nuclei was found while no eYFP signals above background in a WT were found (Extended Data Fig. 4a).

All ASY1 and ASY3 fusion proteins with and without TbID localized to the meiotic chromosome axis during prophase I (Fig. 2a). Moreover, the localization of ASY1 and ZYP1 was similar in ASY1-eYFP-TbID, ASY3-eYFP-TbID and WT (Extended Data Fig. 4c, d) with ASY1 getting depleted from ZYP1-positive synapsed chromosome regions and complete SC formation, suggesting WT-like meiotic chromosome axis remodeling and SC formation in both ASY1-eYFP-TbID and ASY3-eYFP-TbID plants. Notably, we also found the presence of ASY1 and ZYP1 in WT tetrad nuclei (Extended Data Fig. 4c).

Fig. 2. Biotinylation of meiotic chromosome axes in ASY1-eYFP-TbID and ASY3-eYFP-TbID plants.

(a) Immunolocalization using fluorophore-conjugated streptavidin (biotin detection, magenta) and anti-GFP antibody (eYFP detection, green) reveals meiotic chromosome axis localization of ASY1-eYFP, ASY1-eYFP-TbID, ASY3-eYFP and ASY3-eYFP-TbID fusion proteins while only in ASY1-eYFP-TbID and ASY3-eYFP-TbID meiocytes biotinylation of meiotic chromosome axes after treatment with 0.5 mM biotin. Scale bar = 5 μm. (b) SIM (Super-resolution microscopy) of ASY3-eYFP-TbID pachytene nucleus after immunolocalization as in (a). The experiments in a-b were repeated three times independently with similar results. Arrowheads indicate biotin signal covering ASY3-eYFP-TbID weak or negative region. Scale bar = 2 μm.

Next, we asked whether meiotic chromosome axes were biotinylated in ASY1- and ASY3-eYFP-TbID. Immunolocalization using fluorochrome-conjugated streptavidin for biotin detection showed that in ASY1- and ASY3-eYFP-TbID abundant biotinylation was found on chromosome axes while neither in ASY1-nor ASY3-eYFP (Fig. 2a). Spatially structured illumination microscopy (3D-SIM) revealed that at pachytene, two nearly linear biotin signals overlapped with axis protein ASY3(-eYFP-TbID) on the two chromosome axes, supporting that TbID trigger biotinylation in a distance-dependent manner. Notably, the biotin signal also covered ASY3 weak or negative chromatin regions suggesting persistent biotinylation upon ASY3 depletion (Fig. 2b).

TurboID-based proximity labeling in meiocytes

Exogenous biotin treatment enhances in planta TbID-mediated biotinylation⁴¹. Hence, we adopted a stem-cut method used for BrdU-labelling in Arabidopsis meiocytes⁴⁵ to test whether exogenous biotin delivery can improve TbID-mediated biotinylation in meiocytes. Based on a series of biotin concentrations (0, 0.1, 0.5 and 1.0 mM) tested for ASY1-eYFP-TbID, by immunoblots, the level of in planta biotinylation was rather similar among samples due to the vast amount of endogenous TbID-independent biotinylated proteins (Fig. 3a). However, when compared with 0 mM or 0.1 mM biotin treatment, 0.5 mM and 1 mM exogenous biotin treatment seems to increase in planta biotinylation, in particular the level of biotinylation of ASY1-eYFP-TbID (asterisk; Fig. 3a). To more directly assess the impact of different biotin concentrations on the level of biotinylation of meiotic chromosomes, we performed immunolocalization of ASY1 and biotin in ASY1-eYFP-TbID prophase nuclei after treatment with different biotin concentrations. By immunolocalization, 0.5 mM biotin treatment significantly increases biotinylation of meiotic chromosome axes in ASY1-eYFP-TbID meiocytes compared with 0 mM or 0.1 mM biotin treatment (Fig. 3b, c). Interestingly, no further increase in biotinylation of meiotic chromosome axes was found after treatment with 1.0 mM biotin compared with 0.5 mM (Fig. 3b, c). Notably, even without exogenous biotin treatment, meiocytes in ASY1-eYFP-TbID plants were found biotinylated albeit weaker (Fig. 3b). Moreover, in WT anthers also comparatively strong biotin signals were found in organelles of mitotic cells (Extended Data Fig. 5a), which is consistent with the presence of free biotin in plant cell cytosol and bound biotin in organelles⁴⁶. Based on the above, we used for all further experiments 0.5 mM of exogenous biotin treatments.

Fig. 3. Exogenous biotin treatment enhances biotinylation in ASY1-eYFP-TbID plants.

(a) Immunoblot after streptavidin affinity purification from 30 dissected inner inflorescences samples with flower buds containing cells undergoing male meiosis of ASY1-eYFP-TbID plants treated with 0, 0.1, 0.5 or 1.0 mM of exogenous biotin for 6 hours using the stem-cut method. Coomassie brilliant blue (CBB) stained protein gel as loading control. Asterisk indicates biotinylated ASY1-eYFP-TbID. The experiment was repeated three times independently with similar results. (b) Immunolocalization using fluorophore-conjugated streptavidin (biotin detection, magenta) and anti-ASY1 antibody (green) in male meiocytes of ASY1-eYFP-TbID plants treated with 0, 0.1, 0.5 or 1.0 mM of exogenous biotin (Scale bar = 10 μm). (c) Quantification of biotin signal intensity (normalized with ASY1) in immuno-stained cells as in (b) using Fiji (ASY1-eYFP-TbID, 0.1 mM (0.44 ± 0.15, n=18) vs. 0 mM (0.33 ± 0.15, n=21), p=0.033; 0.5 mM (1.25 ± 0.41, n=23) vs. 0/0.1 mM, p=9.19 x 10^-11/1.24 x 10^-9; 1.0 mM (1.19 ± 0.59, n=38) vs. 0/0.1 mM, p=7.07 x 10^-11/2.6 x 10^-9; 1.0 mM vs. 0.5 mM, p=0.6759; two-sided Student’s t test). For each treatment, samples from distinct plants were used for analysis. ***, P<0.0001. N.S., not significant.

In ASY1-eYFP-TbID and ASY3-eYFP-TbID, biotinylation of meiotic chromatin was found from leptotene to tetrads (Extended Data Fig. 5b), while ASY1 and ASY3 appear associated with chromatin during prophase I, suggesting that TbID-mediated biotinylation in meiocytes is a rather stable modification. Hence at least some of the ASY1-eYFP-TbID- and ASY3-eYFP-TbID-mediated biotinylated proteins during prophase I are associated with the chromosomes throughout meiosis.

TurboID-based proteomic chromosome axis profiling

250 inner inflorescences (from each inflorescence the inner part with young flower buds containing only male meiocytes was dissected) each of ASY1-eYFP and ASY1-eYFP-TbID with and without biotin treatment as well as of ASY3-eYFP-TbID and _UBQeYFP-TbID with biotin treatment (in total six independent samples in triplicate) were used for protein affinity purification and mass spectrometry (MS) analysis. By immunoblots, successful biotin labeling was confirmed in all TbID lines compared to ASY1-eYFP (Fig. 4a). The large amount of biotin signal observed in _UBQeYFP-TbID is consistent with the ubiquitous expression of TbID (Fig. 4a). Note, we collected ASY3-eYFP-TbID and _UBQeYFP-TbID only with biotin treatment, given an increase in biotinylation upon exogenous biotin treatment (Fig. 3; see below). We also used ASY1-eYFP instead of Col-0 as “WT control” across all experiments. We reasoned that the eYFP marker would enable collecting flower buds at similar stages between TbID samples and controls given also that ASY1-eYFP fully rescues asy1. Given the existence of endogenous plant biotinylated proteins⁴¹, we used this control sample to identify and filter for TbID-independent in planta biotinylated proteins. To decipher whether an exogenous biotin treatment is beneficial for MS-based identification of meiotic proximate candidates, we compared the proximate “interactomes” of ASY1-eYFP-TbID with and without biotin treatment. Upon exogenous biotin treatment, around 78.1% of proteins showed a positive fold change, compared with samples without prior biotin treatment. Among them, 127 and 23 candidates were identified with significant positive and negative fold change, respectively (Extended Data Fig. 5c). Notably, the known axes protein ASY4 was among the candidates with positive fold change, suggesting that biotin treatment is beneficial for TbID-based proximity labeling in meiotic cells in agreement with immunostaining and WB data (Fig. 3).

Fig. 4. Identification of meiotic chromosome axis proximate candidates in ASY1-eYFP-TbID and ASY3-eYFP-TbID plants.

(a) Biotinylation in 30 dissected inflorescences (inner parts) of ASY1-eYFP, ASY1-eYFP-TbID, ASY3-eYFP-TbID and _UBQeYFP-TbID plants with (+) or without (-) exogenous biotin treatment assessed by streptavidin-affinity purification and protein blot using streptavidin (SA) and anti-GFP antibody. Coomassie brilliant blue (CBB) stained protein gel as loading control. Asterisks mark the size of TbID fusion proteins. The experiment was repeated four times independently with similar results. (b-e) Volcano plots depicting candidates and their enrichments identified by MS by comparing (b) ASY1-eYFP-TbID or (c) ASY3-eYFP-TbID versus control plants (ASY1-eYFP) and (d) recovered candidates from (b) or (e) recovered candidates from (c) versus _UBQeYFP-TbID. Statistical significance of differentially expressed proteins was determined using limma. Arrows indicate known axis proteins. (f) Venn diagram depicting all axis proximate candidates identified by statistical and PSM filtering specific for ASY1-eYFP-TbID and/or ASY3-eYFP-TbID.

TbID-based proximity labeling combined with affinity purification and MS analysis was performed as biological triplicates for each of the six independent samples. Data analysis was performed across all datasets leading to the identification of 4807 proteins. To filter the data for ASY1 and/or ASY3 proximate candidates, stringent filtering conditions were applied. First, all proteins not significantly enriched in ASY1-eYFP-TbID or ASY3-eYFP-TbID samples versus the control ASY1-eYFP were removed to exclude the endogenous biotinylated proteins and candidates TbID-independently purified (e.g. proteins bound unspecific to the streptavidin-conjugated beads) leaving 405 and 667 proteins for ASY1-eYFP-TbID and ASY3-eYFP-TbID, respectively (Fig. 4b, c and Supplemental Data 1). Second, candidates not significantly enriched in ASY1-eYFP-TbID or ASY3-eYFP-TbID samples versus the control _UBQeYFP-TbID were further removed to subtract TbID-mediated but POI-independent biotinylated proteins. After this step, 12 and 15 proteins were left for ASY1-eYFP-TbID and ASY3-eYFP-TbID, respectively (Fig. 4d, e). We noticed that some known meiotic chromosome axis-related proteins such as ZYP1a/b, PCH2 or PRD3 were excluded during our stringent statistical filtering. However, considering numbers of peptide spectrum matches (PSMs; see material and methods for details), these proteins were specifically identified in ASY1-eYFP-TbID or ASY3-eYFP-TbID samples while not in any of our control samples. Hence, their low enrichments (likely due to their overall low amount in total protein extracts from inner inflorescence samples with limited numbers of meiotic cells) hindered their identification upon our stringent statistical filtering (see material and methods for details). To account for these excluded low abundant candidates, we included PSM filtering. Here proteins uniquely identified with three or more PSMs in ASY1-eYFP-TbID or ASY3-eYFP-TbID while absent in both controls were considered as additional candidates. After merging both sets of candidates from our statistical and PSM filtering, in total 39 meiotic chromosomes axis proximate candidate proteins were identified. 25 and 20 of these proteins were identified in ASY1-eYFP-TbID and ASY3-eYFP-TbID samples, respectively, with six shared between both samples (Fig. 4f, Table 1 and Supplemental Data 2).

Table 1. The list of meiotic chromosome axis proximate candidates.

Full list of candidates identified by MS (filtered by statistical analysis and PSM numbers) for both ASY1-eYFP-TbID and ASY3-eYFP-TbID including gene identifier, gene name, subcellular localization and description. The candidates were filtered using stringent cut-offs (see methods) and listed according to the fold change (first ASY1-eYFP-TbID then ASY3-eYFP-TbID versus _UBQeYFP-TbID) or PSMs number in descending order (for details see Supplemental Data 2). POI: protein of interest. *: Information according to The Arabidopsis Information Resource (TAIR, https://www.arabidopsis.org/index.jsp)

Can. Name	Accessions	Genes	Subcellular localization*	Description (prediction)*	POI	Filtered by
ATC1	AT1G67370	ASY1	chromosome; nucleus	Meiotic axis component	ASY1, ASY3	Statistical analysis
ATC2	AT2G4698O	ASY3	chromosome; nucleus	Meiotic axis component	ASY1, ASY3	Statistical analysis, PSMs
ATC3	AT5G17160	AT5G17160	nucleus	Aspartic/glutamic acid-rich protein	ASY1	Statistical analysis
ATC4	AT2G33793	ASY4	chromosome; nucleus	Meiotic axis component	ASY1, ASY3	Statistical analysis
ATC5	AT2G32235	AT2G32235	nucleus	Unknown protein	ASY1	Statistical analysis, PSMs
ATC6	AT1G31190	IMPL1	chloroplast	Myo-Inositol monophosphatase like 1	ASY1	Statistical analysis
ATC7	AT1G66670	CLPP3	chloroplast	Nuclear-encoded caseinolytic protease	ASY1	Statistical analysis
ATC8	AT2G36660	PAB7	cytoplasm; nucleus	Polyadenylate-binding protein	ASY1	Statistical analysis
ATC9	AT4G32820	AT4G32820	nucleus	Tetratricopeptide repeat (TPR)- like superfamily protein	ASY1	Statistical analysis
ATC10	AT3G12510	AT3G12510	nucleus	MADS-box family protein	ASY1	Statistical analysis, PSMs
ATC11	AT3G23670	KINESIN-12B	cytoplasm; cytoskeleton	Microtubule motor kinesin	ASY1	Statistical analysis
ATC12	AT3G61430	PIP1A	membrane; mitochondrion; vacuole	Subfamily PIP1	ASY1	Statistical analysis
ATC13	AT1G22275	ZYPIb	nucleus	Meiotic synaptonemal complex (SC)component	ASY1, ASY3	PSMs
ATC14	AT1G22260	ZYPIa	nucleus	Meiotic synaptonemal complex (SC) component	ASY1, ASY3	PSMs
ATC15	AT5G05490	SYN1	chromosome; nucleus	Meiotic axis component	ASY1, ASY3	PSMs
ATC16	AT1G77600	PDS5B	chromosome; nucleus	Cohesion component	ASY1	PSMs
ATC17	AT4G24710	PCH2	chromosome; membrane	Meiotic chromosome remodeling	ASY1	PSMs
ATC18	AT1G56050	ENGD-2	cytoplasm	GTP-binding protein-like protein	ASY1	PSMs
ATC19	AT1G20575	DPMS1	endoplasmic reticulum; membrane	Component of dolichol phosphate mannase synthase (DPMS) complex	ASY1	PSMs
ATC20	AT5G15510	TPXL5	nucleus	TPX2 (targeting protein for Xklp2) protein family	ASY1	PSMs
ATC21	AT4G18490	AT4G18490	nucleus	Unknown protein	ASY1	PSMs
ATC22	AT3G09430	AT3G09430	nucleus	Peptide transporter family protein	ASY1	PSMs
ATC23	AT4G22960	AT4G22960	cytoplasm	FAM63A-like protein	ASY1	PSMs
ATC24	AT1G01690	PRD3	nucleus	Meiotic DNA double strand break formation	ASY1	PSMs
ATC25	AT2G20900	DGK5	N.D.	Diacylglycerol kinase 5	ASY1	PSMs
ATC26	AT5G56710	AT5G56710	cytoplasm	Ribosomal protein L31e family protein	ASY3	Statistical analysis, PSMs
ATC27	AT4G11220	BTI2	nu cleus; endoplasmic reticulum	VIRB2-interacting protein 2	ASY3	Statistical analysis
ATC28	AT1G20720	AT1G20720	mitochondrion; nucleus	RAD3-like DNA-binding helicase protein	ASY3	Statistical analysis
ATC29	AT2G21600	RER1B	endoplasmic reticulum; Golgi; membrane	Key player of retrieval of ER membrane proteins	ASY3	Statistical analysis
ATC30	AT5G10780	AT5G10780	endoplasmic reticulum; nucleus	ER membrane protein complex subunit-like protein	ASY3	Statistical analysis
ATC31	AT5G53530	VPS26A	membrane	Homolog of yeast retromer subunit VPS26	ASY3	Statistical analysis
ATC32	AT4G25050	ACP4	chloroplast; cytosol	Acyl carrier protein	ASY3	Statistical analysis
ATC33	AT4G3434O	TAF8	nucleus	Member of SAGA complex	ASY3	Statistical analysis
ATC34	AT3G53520	UXS1	endosome; Golgi; membrane	Isoform of UDP-glucuronic acid decarboxylase	ASY3	Statistical analysis
ATC35	AT5G16870	AT5G16870	mitochondrion	Peptidyl-tRNA hydrolase II (PTH2) family protein	ASY3	Statistical analysis
ATC36	AT4G19350	EMB3006	kinetochore; nucleus	Component of MIS12 complex	ASY3	Statistical analysis
ATC37	AT3G48210	SPC25	nucleus	Kinetochore protein	ASY3	PSMs
ATC38	AT4G34850	LAP5	cytoplasm, endoplasmic reticulum	Chalcone and stilbene synthase family protein	ASY3	PSMs
ATC39	AT1G02050	LAP6	cytoplasm, endoplasmic reticulum	Chalcone and stilbene synthase family protein	ASY3	PSMs

Gene ontology (GO) analysis revealed that 23 of the 39 proteins show nuclear localization (Table 1). The two POIs ASY1 and ASY3 showed the highest enrichment and peptide coverage in the corresponding ASY1-eYFP-TbID and ASY3-eYFP-TbID candidate lists, respectively. Combined peptide coverage was up to 51.5% and 44.0% (per sample up to 49.7% and 40.5%) for ASY1 and ASY3, respectively (Extended Data Fig. 6). Strikingly, in addition to ASY1 and ASY3 also further proteins known to play a role in meiosis were found, these include axis-associated ASY4, cohesin subunit proteins REC8 and PDS5b, SC component proteins ZYP1a and ZYP1b, axis remodeling proteins PCH2 and DSB-machinery protein PRD3, suggesting high confidence of our candidates identified by TbID-based PL. Notably, some of the meiotic proteins were only identified by PSM filtering. Hence, PSM-filtering seems beneficial to identify candidates with low enrichment due to low abundance in a sample (rare cell type). Additionally, except the nine candidates known to be meiotic axis-related, 21 out of the 30 remaining candidates are expressed in young flower buds⁴⁷.

We noted that some potential meiotic chromosome axes proximate proteins (in particular with expression also in mitotic cells) were removed during the second stringent filtering step versus _UBQeYFP-TbID. Among those, 10 were reported with meiotic function including e.g. TAF4B, ATM, MRE11, SMC1 or ICU2 (Extended Data Fig. 7a, b and Supplemental Data 1). Possibly, their _UBQeYFP-TbID-mediated biotinylation occurs either in meiocytes and/or likely primarily in mitotic cells as meiosis-specific candidates such as REC8, ZYP1a/b or PCH2 were not removed. Hence, _UBQeYFP-TbID seems particularly suitable to identify a limited number of high confidence candidates specifically expressed in meiotic cells being in proximity of ASY1-eYFP-TbID and/or ASY3-eYFP-TbID.

However, stringent _UBQeYFP-TbID-mediated filtering comes with the possible downside of “losing” some “true” candidates. To gain more insight into the quality of datasets generated by TbID-based PL in meiotic cells, the candidate proteins of the ASY1- and ASY3-eYFP-TbID samples after the first filtering step versus ASY1-eYFP (no TbID) were further analyzed (Supplemental Data 1; statistical filtering). 73.5% and 69.8% of candidate proteins from ASY1-eYFP-TbID and ASY3-eYFP-TbID showed nuclear localization based on GO cellular component analysis, indicating preferential TbID-mediated biotinylation in the nucleus (Extended Data Fig. 7c, d). The top three molecular functions, i.e. protein binding, catalytic activity and DNA binding, were similar for both ASY1-eYFP-TbID and ASY3-eYFP-TbID (Extended Data Fig. 7e, f).

High peptide coverage enabled us to search for post-translational modifications of POIs ASY1 and ASY3. Indeed, certain number of phosphorylation sites was detected (Extended Data Fig. 6). In ASY1, threonine 365 (T365), serine 382 (S382) and threonine 295 (T295) were found phosphorylated. T365 and S382 are predicted to be potential targets of CDK kinase and T295 is a potential target of ATM/ATR kinase (SQ/TQ as target site) likely homologous to T318 of yeast Hop1, being critical for inter-homolog recombination²⁴. We asked whether these sites play a role for ASY1 function during meiosis. To answer that, we generated transgenic Arabidopsis lines expressing ASY1 with putative phosphorylation sites mutated to alanine, namely ASY1^hexa-A and ASY1^365A+382A. Due to a possible functional redundancy of SQ/TQ target sites in vivo, ASY1^hexa-A has six SQ/TQ sites mutated (representing two clusters, SCD1 [SQ/TQ Cluster Domain 1] and SCD2). ASY1^365A+382A has the putative in planta phosphorylated CDK sites mutated (Extended Data Fig. 8a). Full complementation of asy1 plants transformed with either ASY1^hexa-A or ASY1^365A+382A was found based on fertility and male meiotic chromosome spread analysis (Extended Data Fig. 8b-d). Hence, in planta phosphorylation of these sites seems not critical for ASY1 function in Arabidopsis. Whether subtle differences are found between ASY1 and ASY1^hexa-A or ASY1^365A+382A needs further studies.

ASY1 and/or ASY3 proximate candidates: new meiotic players

To refine the list of possible candidates for functional studies, we initially selected 12 new Axis TbID Candidates (ATC) with strong expression in flower buds⁴⁷ and with predicted nuclear (10 candidates) or cytoplasmic localization (2 candidates), as well as the 3 identified candidates ASY4 (ATC4), ZYP1b (ATC13) and PRD3 (ATC24) known to be meiotic chromosome axes related (Fig. 5a and Table 1). We tested these 12 newly identified and 3 known candidates (meiotic chromosome axes-related) for a direct interaction with axis proteins ASY1, ASY3 and REC8 (identified as ATC1, ATC2 and ATC15 in the candidate list) via Y2H. Note, interactions among axis proteins ASY1, ASY3 and REC8 were also tested and in addition to the full-length version of ASY1 also a truncated version (termed ASY1T) with the last C terminal 26 aa deleted (reported to interfere with the interaction between ASY1 and other axis protein, i.e. ASY3²⁷) was employed. We detected for ASY1 (ASY1T) and for REC8 one interactor (self-interaction not considered), ASY3 and ATC8 (PAB7), respectively, while for ASY3 four interactors (ASY4, PRD3, ZYP1b and ATC8) (Fig. 5a). The Y2H interactions between ASY1 and ASY3 as well as ASY3 and ASY4 have been reported previously^{8, 27}. In fact, in our MS data, ASY3/ASY4 and ASY1/ASY4 are highly enriched among ASY1-eYFP-TbID and ASY3-eYFP-TbID proximate candidates, respectively. Interestingly, for most of the tested candidates, no Y2H interaction with known axes candidates was detected. This lack of interaction could imply that the new candidates are only in proximity of ASY1 and/or ASY3 and hence cannot directly interact with axis components in yeast. To address this hypothesis, 7 out of the 12 new candidates tested by Y2H were selected (ATC3, ATC5, ATC8, ATC21, ATC22, ATC23 and ATC28) to investigate whether they play a critical role in meiosis. Homozygous T-DNA insertion mutant alleles for each of them were identified, the absence of transcripts (CDS) was confirmed (Extended Data Fig. 9a), and male meiotic chromosome spread analysis was performed. In case of both candidates, ATC3 and ATC21, univalents (suggesting a failure to form chiasmata including the “obligate” CO, which resulted in a lack of bivalent formation) and unequal chromosome segregation during the first meiotic division were found (Fig. 5b), with a frequency of 10% and 45% for atc3-1 and atc21-1, respectively (Fig. 5b, c). To confirm the observed meiotic phenotype in atc3-1 and atc21-1, a second independent allele with the absence of full-length CDS was analyzed for ATC3 (atc3-2) and ATC21 (atc21-2), showing similar meiotic defects (Fig. 5b, c). No obvious meiotic defects were found in male meiocytes of the remaining candidates (Fig. 5b, c). However, in the case of atc5 we noted an obvious reduction in fertility (reduced number of seeds per silique), despite male meiotic chromosome behavior being WT-like (Fig. 5b and Extended Data Fig. 9b). Next, we expressed the three candidates ATC3, ATC5 and ATC21 as HA-mRuby2 fusion proteins in WT plants, given their mutants showing either meiotic defects (ATC3 and ATC21) or fertility reduction (ATC5). While for ATC3 no specific fusion protein signal was detected, expression of ATC5 was found in tetrad nuclei (Fig. 5d and Extended Data Fig. 9c) and, strikingly, expression of ATC21 was detected in meiotic prophase I nuclei (Fig. 5e). Together, among functionally tested candidates at least three are critical for meiosis and/or fertility as well as two of these are specifically expressed in meiotic nuclei (including prophase I) while the expression/localization of the third candidate could not be discerned. Hence, among our stringent candidates identified by TbID-based PL of meiotic chromosome axes not only candidates known to play a role in meiosis but also new bona fide meiotic (chromosome axis) proteins are found.

Fig. 5. Functional dissection of selected candidates.

(a) Interactions between meiotic chromosome axis protein ASY1, ASY3 or REC8 and Axis TbID Candidate (ATC) proteins detected by Y2H. The interaction status is indicated by the color code, the color key is on the right side, note that TDO (SD/-LTH) is a less stringent medium, QDO (SD/-LTHA) is the most stringent medium for selection. (b) Male meiotic chromosome spreads of T-DNA insertion mutants of the selected candidates ATC3, ATC5, ATC8, ATC21, ATC22, ATC23 and ATC28. Col-0 as WT. Note, in the case of both, atc3 and atc21, presence of univalents at metaphase I and unbalanced meiotic chromosome segregation during the first meiotic division. DAPI-stained DNA in grey. Scale bar = 10 μm. (c) Quantification of cells (n = number of cells analyzed per T-DNA insertion lines) with normal or abnormal meiosis. Cells with univalents or unbalanced chromosome segregation were categorized as abnormal. (d, e) Immunolocalization using anti-ZYP1 or anti-ASY1 (green) and anti-HA (magenta) antibodies in plants expressing ATC5- or ATC21-HA-mRuby2 fusion proteins in meiotic cells. DNA counterstained with DAPI in blue. Scale bar = 10 μm. The experiments were repeated three times independently with similar results.

Discussion

One major benefit of TbID-based proximity labeling is that TbID-mediated biotinylation occurs in vivo. To achieve specific in vivo labeling, it is essential to ensure the functionality of employed POI-TbID fusion proteins³⁵. In our case, male meiosis and male fertility were restored in both ASY1-eYFP-TbID and ASY3-eYFP-TbID. However, while female meiosis and overall fertility were recovered in ASY3-eYFP-TbID, they were neither fully recovered in ASY1-eYFP-TbID nor ASY1-eYFP-mTb despite full recovery in ASY1-eYFP. Possibly fusion of TbID (or mTb) to ASY1 to some extent hampers the functionality of ASY1 during female meiosis and/or ASY1-eYFP-TbID (or -mTb)-dependent biotinylation to some degree impairs female meiosis. This difference between male and female meiosis in ASY1-eYFP-TbID might be due to sexual dimorphic function of ASY1 and/or differential expression levels/spatiotemporal localization of ASY1 between male and female meiocytes. In line with this, sexual dimorphism was reported in Arabidopsis with regards to differential axis/SC length or CO patterning⁴⁸. In any case, by employing dissected inner inflorescences samples with young flower buds that only contain male meiotic cells expressing ASY1-eYFP-TbID, female meiocytes were excluded. Recently, a new biotin ligase version (ultraID) was reported with comparable efficiency as TbID while being smaller in size⁴⁹. Hence, it will be tempting to test the smaller ultraID with POI ASY1 in terms of female fertility.

Most proteins identified by MS are shared among all samples, highlighting the importance of using stringent controls to filter out interesting candidates. We used ASY1-eYFP (WT) to filter out TbID-independent candidates (endogenous biotinylated proteins and/or “unspecific” purified proteins), leaving 405 proteins for ASY1-eYFP-TbID and 667 for ASY3-eYFP-TbID. To further exclude TbID-mediated but POI-independent candidates, a second control is necessary. Given that ASY1- and ASY3-eYFP-TbID are specifically expressed in meiotic cells, ideal controls could theoretically be plants with meiotic TbID expression levels/dynamics similar to ASY1 or ASY3. We tested the ASY1 promoter but no eYFP-TbID expression was detected in male meiotic cells as reported²⁰. Accordingly, we used as our second control _UBQeYFP-TbID plants, in which eYFP-TbID was expressed in mitotic and meiotic nuclei under control of the UBQ10 promoter. In fact, stringent filtering using _UBQeYFP-TbID as control left us with a rather low number of high confidence candidates, including various candidates known to play a role in meiosis and some candidates not yet described to play a role during meiosis. Hence, we consider _UBQeYFP-TbID as a suitable control to identify high confidence meiotic proximate candidates. Notably, due to our stringent filtering using _UBQeYFP-TbID some axis-related candidates that are expressed in both somatic and meiotic cells (e.g. ATM or MRE11) were removed. Whether control lines with meiosis-specific TbID expression could help to recover these candidates is unclear.

The presented TbID-based approach is highly efficient because even the proteome of a rare cell type like the meiotic cell embedded within flower tissues can be analyzed. For each sample, affinity purification followed by MS was performed using only 250 dissected inner inflorescences parts with young flower buds. The enrichment of some known meiotic axis-related proteins (e.g. PCH2, PRD3 or ZYP1a/b) was low and hence they were only identified by filtering based on PSMs. Whether increased starting material amounts may improve candidate coverage or numbers could be explored. Moreover, a meiocyte enrichment step before affinity-purification, e.g. by employing dissected anthers or isolated/enriched meiotic cells^{50, 51}, may improve data quality.

ASY1 antibody-based affinity proteomics in B. oleracea using isolated anthers and meiocytes identified a list of 492 candidates including 12 proteins with a reported role in meiosis²⁸. Except DMC1, all 12 proteins were also recovered in our ASY1-eYFP-TbID MS dataset using dissected inner inflorescences samples of Arabidopsis. The lack of DMC1 in our dataset could be due to e.g. DMC1 being not in sufficient proximity to ASY1 or no lysine residue (receptor) in DMC1 being available for ASY1-eYFP-TbID-mediated biotinylation. This could also be the reason for the absence of COMET, which was recently reported to interact with ASY1²³. By two-step stringent filtering 25 proximate candidate proteins for ASY1-eYFP-TbID were left with high confidence (including most known axis proteins) compared to 492 candidates identified by ASY1-affinity proteomics. REC8, PDS5b and ZYP1b were not among the 492 candidates found by ASY1-affinity proteomics in B. oleracea but among our 25 candidates. Moreover, considering one-step filtering data of ASY1-eYFP-TbID versus ASY1-eYFP, further candidates with a reported meiotic role are found in our list (e.g. MRE11, RAD50, NBS1, TAF4B or ATM) but not identified by ASY1-affinity proteomics, while candidates such as SMC1 and ICU2 were recovered by both approaches. Notably, MRE11, RAD50 and NBS1 form a complex crucial for DSB repair⁵², TAF4B is a CO-modifier and ATM is a key meiosis regulator in plants^{53, 54}. In the case of ASY3, no comparison is possible given the lack of available affinity proteomics data. However, some meiotic proteins have been reported to interact with ASY3 in Y2H, including ASY1, ASY4, ZYP1, PRD3, MTOPVIB and PRD2^{8, 15, 55, 56}. Four of them are found among the candidate list of ASY3-eYFP-TbID. Together, our data suggest that TbID-based PL is an efficient tool for the identification of meiotic candidate proteins.

In total 25 and 20 axis proximate candidates for ASY1-eYFP-TbID and ASY3-eYFP-TbID (including ASY1 and ASY3), respectively, were found. Interestingly, only six candidates are shared between ASY1-eYFP-TbID and ASY3-eYFP-TbID, all are meiotic chromosome axis-associated. This may suggest, in addition to differences in spatiotemporal dynamics of the POIs, also for instance a different sub-localization of ASY1 and ASY3 and/or different TbID labelling sub-areas due to the position of TbID relative to the POI. Recently expansion stochastic optical reconstruction microscopy (STORM) revealed a different sub-localization of HORMAD1 (homologue of ASY1) and SYCP2 (homologue of ASY3) in mouse⁵⁷. Whether a similar situation is found in Arabidopsis is unclear. Moreover, TbID was fused to the C terminus of ASY1 while to the N terminus of ASY3. Given that the N terminus of ASY3 (closure motif) binds to the N terminal HORMA domain of ASY1⁵⁸, resulting differences in TbID labeling sub-areas between ASY1-eYFP-TbID and ASY3-eYFP-TbID may in part explain why most of the candidates including some known meiotic candidates (e.g. PCH2 or PRD3) are not shared between ASY1-eYFP-TbID and ASY3-eYFP-TbID. In future, a comparison of candidates identified from both N- and C-terminal POI-TbID fusions may give further insights.

A limited number of the tested candidates showed a direct interaction with axis proteins ASY1, ASY3 or REC8 by Y2H. Likely, most candidates are rather in proximity of ASY1 and/or ASY3 or require modifications/further proteins for interaction in yeast. Hence, TbID-based PL may enable to identify proximate candidates that are not recovered using antibody-based affinity-proteomics or that are not identified when screening for direct interactions using e.g. meiotic cDNA Y2H libraries. In addition to the recovery of various known meiotic chromosome axis-associated proteins, initial data suggest new candidates with high confidence. By chromosome spread analysis, obvious male meiotic defects including univalent formation and unbalanced meiotic chromosome segregation were found in T-DNA insertion mutant lines of two (atc3 (AT5G17160) and atc21 (AT4G18490)) out of seven selected candidates, which strongly suggest that these two candidates are critical for meiosis. In the case of atc5 (AT2G32235), despite WT-like male meiosis, a reduction in fertility was found. Among the remaining four candidates that do not show obvious meiotic defects by male meiotic chromosome spread analysis three (atc8 (PAB7), atc23 (AT4G22960) and atc28 (AT1G20720)) have one or more paralogue(s) in the Arabidopsis genome, notably, ATC8 was found to interact with ASY3 and REC8, and ATC28 is predicted to be involved in meiotic DSB repair. Hence, to address whether they play an essential meiotic role, the generation of double or multiple mutants might be necessary. In addition to the candidates studied by meiotic chromosome spread analysis of respective T-DNA insertion mutants, ATC11 (Kinesin-12B) is reported to be involved in phragmoplast assembly during male gametogenesis⁵⁹ while ATC9 (AT4G32820) and ATC33 (TAF8) in chromatin organization or histone modifications, possibly suggesting a role for meiotic chromatin. ATC36 (EMB3006) and ATC37 (SPC25) are two kinetochore proteins that might be found at meiotic centromeres. Moreover, candidates predicted to be localized to the endoplasmatic reticulum, vacuole, endosome, membrane or Golgi (e.g. ATC19 [AT1G20575], ATC25 [DGK5], ATC30 [AT5G10780] or ATC34 [UXS1]) could be specifically biotinylated ASY1-eYFP-TbID- and/or ASY3-eYFP-TbID-dependent during translation and/or transport of ASY1-eYFP-TbID and/or ASY3-eYFP-TbID.

By in planta HA-mRuby2 fusion protein expression, ATC5 was expressed in meiotic tetrad nuclei. Given ASY1 expression found in WT tetrad nuclei, likely ATC5 was identified based on ASY1-eYFP-TbID-mediated biotinylation in tetrad nuclei. Most strikingly, expression of an ATC21 fusion protein was found during prophase I. Notably, no Y2H interaction of ATC21 was found with tested axis proteins and ATC21 (as well as ATC3 and ATC5) was not found in ASY1-affinity proteomics in B. oleracea, highlighting the power of TbID-based proximity labeling to identify new proximate candidates when compared to Y2H or traditional antibody-based affinity approaches. Future studies are required and are currently in progress to decipher the exact meiotic role of ATC3 and ATC21 but also of further identified axis proximate candidates.

In a nutshell, we have demonstrated that TbID-based proximity labeling coupled with affinity-purification and MS analysis can be used to identify proximate “interactomes” of POIs and their modifications in rare cell types such as meiotic cells in Arabidopsis. In the future, we envision that TbID-mediated biotinylation could be used for the direct identification of yet unknown plant meiotic proteins or modifications involved for instance in DSB/CO formation and processing as well as of components of the SC, the meiotic kinetochore, cytoskeleton or nuclear envelope.

Methods

Plant growth conditions and materials

A. thaliana plants were grown at constant 22°C under short day conditions (8/16 h light/dark) for 4 weeks and then under long day conditions (16/8 h light/dark) until maturity.

Col-0 was used as WT. The following T-DNA insertion lines provided by the Nottingham Arabidopsis Stock Centre (NASC) were used⁶⁰: asy1-4 (AT1G67370; SALK_046272)²¹, asy3-1 (AT2G46980; SALK_143676)⁷, atc3-1 (AT5G17160; SALKseq_105562), atc3-2 (AT5G17160; FLAG_172D11), atc5-1 (AT2G32235; SAIL_221_A10), atc8-1 (AT2G36660; SALKseq_056996), atc21-1 (AT4G18490, SAILseq_210_G05), atc21-2 (AT4G18490; SALK_023936), atc22-1 (AT3G09430; SAILseq_718_D10), atc23-1 (AT4G22960; SALKseq_047548), and atc28-1 (AT1G20720; SAILseq_423_C04). Details of primers used for genotyping of T-DNA lines are found in Supplemental Data 3.

Plant transformation was performed according to the flower dip method⁶¹. pASY1::ASY1-eYFP-TbID::tASY1 (ASY1-eYFP-TbID), pASY1::ASY1-eYFP-mTb::tASY1 (ASY1-eYFP-mTb) and pASY1::ASY1-eYFP::tASY1 (ASY1-eYFP) were transformed into asy1-3 +/-, pASY3::eYFP-TbID-ASY3::tASY3 (ASY3-eYFP-TbID) and pASY3::eYFP-ASY3::tASY3 (ASY3-eYFP) into asy3-1 -/-, and pUBQ10::_NLSeYFP-TbID::tNos (_UBQeYFP-TbID) into the WT. Primary transformants were selected on Murashige and Skoog medium⁶² containing the relevant antibiotic. Among at least 10 T₁ transgenic lines per transformation based on an initial fertility and microscopic expression assessment of the respective fusion protein, for each construct a representative line for further analysis was selected. T₂ or T₃ transgenic lines were used for analysis.

Plasmids construction

For details of all primers used for construct generation see Supplemental Data 3.

Constructs for protoplast transfection

To generate p35S::3xHA-TbID::tNos and p35S::eYFP::tNos, 3xHA-TbID was either amplified from 3xHA-TurboID-NLS_pCDNA3 (Addgene #107171; mammalian TbID)³⁹ and/or from custom synthesized TbID (Eurofins genomics; plant codon-optimized TbID) using 35SBAM_Turbo_F/R, 35SBAM_HA_F/R or 35SBAM_Turbo_F2/R2, eYFP was amplified using 35SBAM_eYFP_F/R. The amplicons were then cloned into p35S-BAM (dna-cloning-service.com) via BamHI/HindIII.

Constructs for TbID-based PL of meiotic chromosome axes

ASY1-eYFP was generated as described⁴⁴. To generate pASY1::ASY1-eYFP-TbID::tASY1 (ASY1-eYFP-TbID), plant codon-optimized TbID was amplified using AtASY1-TbID_F/R and cloned into ASY1-eYFP via AscI in-fusion cloning (Takara). pASY1::ASY1-eYFP-mTb::tASY1 (ASY1-eYFP-mTb) was generated as above using primers AtASY1-mTb_F/R, the mTb sequence was amplified by mutagenesis PCR with TbID (codon-optimized) as template.

To generate pASY3::eYFP-ASY3::tASY3 (ASY3-eYFP), the ASY3 genomic sequence from A. thaliana accession Col-0 was PCR amplified in two parts using ASY3_clo_F2/ASY3_clo_R4, cloned via AfeI/XbaI in the modified (BseYI site deleted) pBM vector (dna-cloning-service.com) generating N-ASY3, as well as using ASY3_clo_F3/ASY3_clo_R2, cloned via XbaI/SalI in the modified pBM vector generating ASY3-C. AscI/FseI were introduced in ASY3-C by Q5 site-directed Mutagenesis (NEB) using primers ASY3_N_MCS_F/ASY3_N_MCS_R and eYFP was inserted from ASY1-eYFP via AscI/FseI generating ASY3-C-eYFP. ASY3-C-eYFP was inserted in N-ASY3 via XbaI/SalI and resulting expression cassette was subcloned into the binary vector pLH7000 (dna-cloning-service.com) via SfiI generating ASY3-eYFP. To generate pASY3::eYFP-TbID-ASY3::tASY3 (ASY3-eYFP-TbID), TbID was PCR amplified using primers AtASY3-TbID_F/R and cloned into ASY3-eYFP via AscI by in-fusion cloning.

To generate pUBQ10::_NLSeYFP-TbID::tNos (_UBQeYFP-TbID), eYFP-TbID was PCR amplified from ASY1-eYFP-TbID with eYFP-TbID_F/R and cloned into the vector p35S-BAM via NotI/HindIII generating pBAM-eYFP-TbID. The promoter of AtUBQ10 (Col-0) was PCR amplified with UBQ10_F/R and cloned into pBAM-eYFP-TbID via SnaBI by Gibson Assembly (NEB). The resulting expression cassette was finally subcloned into pLH7000 via SfiI generating _UBQeYFP-TbID.

Constructs for ASY1 phosphor-null mutant lines

To generate ASY1 phosphor-null constructs ASY1^hexa-A (267A, 269A, 295A, 566A, 569A and 593A) and ASY1^365-382A (365A and 382A), a fragment was amplified from ASY1-eYFP and cloned into p35S-BAM. The mutations for ASY1^365-382A (ASY1^365-382A_F/R) were introduced via Q5 site-directed mutagenesis. Mutations for ASY1^hexa-A were introduced either via Q5 site-directed mutagenesis (295A: A295_2mutF/A295_2mutR; 566A+569A: TSS_2mutF/TSS_WtR) or via insertion of custom-synthesized Genestrands (Eurofins Genomics) carrying desired mutation(s) (267A+269A via SfoI/EcoO109I; 593A via AscI/PmlI). Clones with desired mutations were cloned via BamHI/ScaI back into ASY1-eYFP with the original sequence substituted.

Constructs for Y2H assays

The coding sequences of axis proteins ASY1, ASY1T (with the last 26 C terminal aa deleted), ASY3, and REC8 as well as selected ATCs (see Fig. 5a) were PCR amplified from Col-0 flower bud cDNA and cloned into pGBKT7 and pGADT7 vectors (Takara) by Gibson Assembly (NEB), respectively.

Constructs for HA-mRuby2 fusions of ATC3, ATC5 and ATC21

The genomic sequences of ATC3, ATC5 and ATC21 were amplified from Col-0 genomic DNA using ATC3-F/R, ATC5-F/R and ATC21-F/R, respectively. Amplicons were cloned (Gibson Assembly) into p35S-BAM via NotI/StuI generating pBAM-ATC3, pBAM-ATC5 and pBAM-ATC21. A 3xHA tag as well as an AfeI site were introduced into them via Q5 site-directed mutagenesis using ATC3-Q5-F/R, ATC5-Q5-F/R and ATC21-Q5-F/R, followed by insertion of the mRuby2 CDS (amplified using ATC3-mRuby2-F/R, ATC5-mRuby2-F/R and ATC21-mRuby2-F/R from pcDNA3-mRuby2 (Addgene #40260)⁶³) via AfeI by Gibson Assembly making pBAM-ATC3-HA-mRuby2, pBAM-ATC5-HA-mRuby2 and pBAM-ATC21-HA-mRuby2. Finally, these expression cassettes were subcloned via SfiI into pLH7000. Details of primer sequences are found in Supplemental Data 3.

Arabidopsis protoplasts transfection

Isolation of Arabidopsis mesophyll protoplasts was performed according to⁶⁴. Isolated protoplasts were washed two times with W5 buffer (154 mmol/L NaCl, 5 mmol/L KCl, 125 mmol/L CaCl₂, 2 mmol/L MES; pH 5.7) and resuspended in Mmg buffer (4 mmol/L MES, 0.4 mol/L Mannitol, 15 mmol/L MgCl₂; pH 5.7) at a concentration of ^~10⁶ cells/mL. For each transformation, 190 μL of protoplast suspension were mixed with 10 μg of plasmid DNA, followed by adding 200 μL of 40% PEG solution (40% PEG4000, 100 mmol/L CaCl₂, 0.6 mol/L mannitol), mixing by gently pipetting and incubation for 18 min at RT. The protoplasts were then washed two times with W5, resuspended in W5 and incubated under light (long day) at 20°C for 24 hrs.

TurboID-based PL in meiotic cells

Biotin treatment

Exogenous biotin treatment was performed as described ⁴⁵. Briefly, 5-6 cm long inflorescence stems with young flower buds were cut under water, immediately transferred into biotin solution (0 - 1.0 mM) and incubated under light for 6 hrs. After the treatment, the inner part of the inflorescences containing young flower buds was dissected, frozen in liquid nitrogen and stored at -80°C.

Protein Affinity-purification

Affinity-purification was performed according to^{40, 41} with some modifications. For each sample, 250 frozen dissected inner inflorescences samples were ground into a fine powder in a mortar using a pestle, transferred into 2 ml reaction tubes, and 0.6 mL of RIPA lysis buffer (50 mM Tris-HCl [pH 7.5], 500 mM NaCl, 1 mM EDTA, 1% NP40 [v/v], 0.1% SDS [w/v], 0.5% sodium deoxycholate *w/v+, 1 mM DTT, 1 tablet of cOmplete™ Protease Inhibitor Cocktail [Roche], 1 mM PMSF) was added for cell lysis. After vortex mixing, 5 μl of DNaseI (NEB), 1 μl of RNase (Thermo Scientific) and 6 μL of 100XHalt Phosphotase+protease inhibitors (Thermo Scientific) were added and samples were incubated on ice for 30 min. After four times sonication (Sonics & Materials, VC 60) on ice (20 sec sonication + 40 sec break), samples were centrifuged at 16,200 x g for 15 min at 4 °C. The upper solution was run through a Zeba™ Spin Desalting Column (Thermo Fisher Scientific) to remove the excess of biotin in protein extracts. Protein concentrations were determined using the Bradford assay (Bio-Rad). To enrich for biotinylated proteins from the protein extracts, 15 μL of streptavidin-coated magnetic beads (MERCK) were washed twice with RIPA lysis buffer and the desalted protein extracts were then incubated with the equilibrated beads on a rotator overnight at 4 °C. The next day, beads were separated from the protein extract with a magnetic rack and washed as described³⁹. For sequential beads washing, 2x cold extraction buffer (beads were transferred into a new tube the first time), 1x cold 1 M KCl, 1x cold 100 mM Na₂CO₃, 1x 2M Urea in 10 mM Tris pH 8 at RT and 2x cold extraction buffer without Protease Inhibitor tablet and PMSF, were used. After washing, 10% of the beads were boiled at 95°C for 10 mins in 50 μl of 4x Laemmli buffer supplemented with 20 mM DTT and 2 mM biotin for protein blots. After removing buffers, the remaining beads were stored at -80°C until further use.

On bead digest

Beads were resuspended in 40 ul of 100 mM ammonium bicarbonate (ABC), supplemented with 400 ng of lysyl endopeptidase (Lys-C, Fujifilm Wako Pure Chemical Corporation) and incubated for 4 hrs on a Thermo-shaker with 1200 rpm at 37°C. The supernatant was transferred to a fresh tube and reduced with 0.5 mM Tris 2-carboxyethyl phosphine hydrochloride (TCEP, Sigma) for 30 mins at 60°C and alkylated in 4 mM methyl methanethiosulfonate (MMTS, Fluka) for 30 min at RT. Subsequently, the sample was digested with 400 ng of trypsin (Trypsin Gold, Promega) at 37°C overnight. The digest was acidified by addition of trifluoroacetic acid (TFA, Pierce) to 1%. A similar aliquot of each sample (30%) was analyzed by LC-MS/MS

LC-MS/MS analysis

The system used was an UltiMate 3000 RSLC nano system coupled to a Q Exactive HF-X mass spectrometer, equipped with an Easyspray nanospray source (all from Thermo Fisher Scientific). Peptides were loaded onto a trap column (Thermo Fisher Scientific, PepMap C18, 5 mm × 300 μm ID, 5 μm particles, 100 Å pore size) at a flow rate of 25 μL/min using 0.1% TFA as mobile phase. After 10 min, the trap column was switched in line with the analytical column (Thermo Fisher Scientific, PepMap C18, 500 mm × 75 μm ID, 2 μm, 100 Å). Peptides were eluted using a flow rate of 230 nl/min, starting with the mobile phases 98% A (0.1% formic acid in water) and 2% B (80% acetonitrile, 0.1% formic acid) and linearly increasing to 35% B over the next 180 min, followed by a gradient to 90% B in 5 min, staying there for 5 min and decreasing in 2 min back to the gradient 98% A and 2% B for equilibration at 30°C. The Q Exactive HF-X mass spectrometer was operated in data-dependent mode, using a full scan (m/z range 380–1500, resolution of 60 000, target value 1E6) followed by MS/MS scans (resolution of 30000, target value 1E5, maximum injection time 105 ms) of the 10 most abundant ions. MS/MS spectra were acquired using normalized collision energy of 28 % and an isolation width of 1 m/z. Precursor ions selected for fragmentation were put on a dynamic exclusion list for 60 s. The intensity threshold was set to 4.8E4.

Proteomics data analysis

Raw MS data were loaded into Proteome Discoverer (PD, version 2.5.0.400, Thermo Scientific). All MS/MS spectra were searched using MS Amanda v2.0.0.16129⁶⁵. Peptide and protein identification was performed in two steps. An initial search was performed against the database TAIR10_pep_20101214.fasta (32,785 sequences; 14,482,855 residues) with common contaminants appended. Trypsin was specified as proteolytic enzyme without proline restriction and allowing for up to 2 missed cleavages. Mass tolerances were set to ±5 ppm at the precursor and ±8 ppm at the fragment mass level. Beta-methylthiolation of cysteine was searched as fixed modification, oxidation of methionine as variable modification. The minimum peptide length was set to 7. The result was filtered to 1 % FDR on protein level and a sub-database of proteins identified in this search was generated for further processing. For the second step, the RAW-files were searched against the created sub-database using the same settings as above plus considering additional variable modifications: Phosphorylation on serine, threonine and tyrosine, deamidation on asparagine and glutamine, biotinylation on lysine, ubiquitination on lysine, acetylation on protein N-Terminus, glutamine to pyroglutamate conversion at peptide N-termini. Again, proteins were filtered to 1% FDR. Additionally, a minimum MS Amanda score of 150 was applied on PSM level and proteins had to be identified by a minimum of 2 PSMs. The localization of the modification sites within the peptides was performed with the tool ptmRS, which is based on phosphoRS⁶⁶. Peptide areas were quantified using the in-house-developed tool apQuant⁶⁷. Proteins were quantified by summing unique and razor peptides and applying intensity-based absolute quantification (iBAQ)⁶⁸. Protein-abundances-normalization was done using sum normalization. The statistical significance of differentially expressed proteins was determined using limma⁶⁹. For statistical candidates filtering, a cut-off of P value<0.05 and FD (fold change)>2 was used for ASY1-eYFP-TbID or ASY3-eYFP-TbID versus controls (ASY1-eYFP, _UBQeYFP-TbID), the scatter plots were created using Microsoft Excel 2016 (Microsoft Corporation, USA). For filtering by PSM numbers, a cut-off of a minimum three of PSMs (identified in ASY1-eYFP-TbID or ASY3-eYFP-TbID; the sum of PSMs from three replicates) and 0 PSMs (identified in controls, ASY1-eYFP and _UBQeYFP-TbID) applied. Note, for one peptide sequence of a given protein, more than one MS/MS spectrum (PSM) can be recorded (i.e. highly abundant peptides). Hence, the PSM number can be used for label-free semi-quantitative quantification, as it reflects the abundance of corresponding peptides (or protein).

Western blot

Protein extracts were separated using 10% SDS-PAGE gels and proteins were transferred onto PVDF membrane (Millipore) for blotting. Antibodies used for blotting: anti-HA-tag (1:2000, Santa Cruz Biotechnology, sc-7392), anti-GFP (1:1000, Chromotek, 3H9), IRDye 800CW goat anti-Mouse (1:2000, LI-COR, 926-32210), IRDye 800CW goat anti-Rat (1:2000, LI-COR, 926-32219), and IRDye 800CW Streptavidin (1:2000, LI-COR, 926-32230). Membrane blotting and washing were performed according to the manufacturer’s instructions (LI-COR). After blotting, the membranes were visualized using the LI-COR Odyssey 9120 imaging system.

Cytological procedures and microscopy

Alexander staining⁷⁰ was performed to assess pollen viability in Arabidopsis anthers. Anthers were dissected, carefully squashed in Alexander staining solution and stained overnight.

Immunostaining was performed according to^{6, 16, 71}. Both the primary antibodies, anti-ASY1⁶, anti-ZYP1¹⁹, anti-GFP (Chromotek, PABG1) and Streptavidin-Cy3 (Thermo Fisher Scientific, 434315), as well as the secondary antibodies, anti-rabbit texas Red (Abcam, Ab6719), anti-Guinea pig Alexa 488 (Jackson Immuno Research, 706-545-148), anti-rabbit Alexa 488 (Jackson Immuno Research, 711-545-152), were diluted 1:500.

Male meiotic chromosome spread analysis and minimum chiasmata counting was performed according to⁷².

Epifluorescence images were acquired using either a Nikon Eclipse Ni-E equipped with a Nikon DS-Qi2 camera and NIS-Elements-AR version 4.60 software (Nikon, Tokyo, Japan) or an Olympus BX61 microscope equipped with an ORCA-ER CCD camera (Hamamatsu, Japan). Images were processed with Adobe Photoshop CS5 (Adobe Inc., USA). To detect the ultrastructural chromatin organization of meiocytes at a resolution of ^~120 nm (super-resolution achieved with a 488 nm laser excitation) spatial structured illumination microscopy (3D-SIM) was performed with a 63×/1.4 Oil Plan-Apochromat objective of an ElyraPS.1 microscope system and the software ZENBlack (Carl Zeiss GmbH). Images were captured separately for each fluorochrome using the 561, 488, and 405 nm laser lines for excitation and appropriate emission filters^{73, 74}.

The quantification of signal intensity (maximum intensity projection of all Z-stacks) was performed with Fiji ImageJ v. 2.9.0 (open source). The relative biotin intensity was obtained by normalizing biotin signal intensity against ASY1 signal intensity.

Light sheet fluorescence microscopy (LSFM) was performed using a Lightsheet 7 (Carl Zeiss GmbH) equipped with two pco.edge 4.2 sCMOS cameras (PCO AG) according to previous reports^{44, 75} with minor modifications. To monitor the duration of ASY1- and ASY3-eYFP (with and without TbID) within prophase I nuclei (ten nuclei per sample) from initial nuclear appearance until nuclear envelope breakdown as well as to evaluate the expression of TbID in meiotic and somatic nuclei of _UBQeYFP-TbID compared with the WT, ^~0.3 mm or ^~0.4 mm flower buds were chosen, respectively. Outer sepals were carefully removed and flower buds were inserted into glass capillaries containing 1% low melting agarose (Sigma-Aldrich, A9045) in Murashige and Skoog (MS) medium (Duchefa Biochemie, M0221) supplemented with 1% sucrose (pH 5.7). For short-term imaging, samples were mounted in ^~1.5 mm (inner diameter) glass capillaries (size 3, green mark; Carl Zeiss GmbH, 701908) and, once fitted in the microscope, the agarose cylinder containing the sample was extruded into the imaging chamber filled with MS medium with 1% sucrose. For long-term imaging (from 24 h up to several days), ^~2.15 mm (inner diameter) glass capillaries (size 4, blue mark; Carl Zeiss GmbH, 701910) were coupled with 2.8 mm (inner diameter) fluorinated ethylene propylene (FEP) tubes (Wolf-Technik). After mounting the sample, the bottom of the tube was sealed with a glue gun and imaging was done through the FEP tube. To match the refractive index of the FEP tubes with that of the medium, the imaging chamber was filled with 6% glycerol which was manually replaced roughly every 12 h. Temperature in the chamber was set to 21°C and no artificial source of light was supplemented. Imaging was done with 10x (W Plan-Apochromat 10x/0.5) detection and 5x (LSFM 5x/0.1 foc) illumination objectives and 2.5x zoom. Excitation was done with 1% 488 nm laser line. Dual-side illumination was chosen, and light sheet pivot was on to reduce shadowing. Z-stacks covering the whole anther were taken every 1, 5 or 30 min. Images were acquired with ZEN Black 3.1 and processed (3D Maximum Intensity projections, sample drift correction, partial movies fusion and image export) with ZEN Blue 3.4 (both from Carl Zeiss GmbH).

RNA extraction and reverse transcription-PCR

Total RNA was isolated from Arabidopsis seedlings (Col-0 and mutant lines) using TRIzol Reagent (Invitrogen) according to manufacturer’s instructions. After DNA removal with DNaseI (NEB, M0303S), 1 μg of total RNA per sample was used for first strand cDNA synthesis (Thermo Scientific, K1632). The cDNA samples were then used as PCR templates. Primer details are found in Supplemental Data 3.

Yeast two-hybrid assays

Y2H experiments were performed following the manufacturer’s instructions (Takara). Bait and prey plasmids (with empty plasmids as controls) were co-transformed into yeast strain Y2HGold (Takara, 630489). Selection assays were performed using Minimal SD Base medium supplement with DO Supplement – Leu/–Trp (DDO, 630417) or DO Supplement –His/–Leu/–Trp (TDO, 630419) or DO Supplement –Ade/–His/–Leu/–Trp (QDO, 630428).

Extended Data

Extended Data Fig. 1. Activity of mammalian and plant codon-optimized TbID in Arabidopsis protoplasts.

(a) Immunoblots (IB) of protein extracts from protoplasts transformed with mammalian or plant codon-optimized TbID and treated with 0.05 mM biotin for 2 hrs. Anti-HA antibody and Streptavidin (SA) used for blotting. Protein extract from non-transformed protoplasts used as control. The experiment was repeated three times independently with similar results. (b) DNA sequence of custom synthesized TbID (plant codon-optimized).

Extended Data Fig. 2. Phenotype of ASY1-eYFP-TbID and ASY3-eYFP-TbID transgenic lines.

(a) Flowering plants of Col-0, ASY1-eYFP (asy1), _UBQeYFP-TbID (Col-0), ASY1-eYFP-TbID (asy1), ASY3-eYFP-TbID (asy3), asy1 and asy3. Scale bar = 5 cm. (b) Pollen viability in plant lines as indicated in (a) revealed by Alexander red staining. Scale bar = 50 μm. (c) Seed setting in fruit pods of Col-0, ASY1-eYFP, ASY1-eYFP-TbID and asy1. Scale bar = 0.5 cm. (d) Localization of ASY1-eYFP and ASY1-eYFP-TbID fusion proteins in female meiotic cells. Pistils containing cells undergoing female meiosis squashed in H₂O. Scale bar = 50 μm. (e) Female meiotic chromosome spread analysis in ASY1-eYFP-TbID and ASY3-eYFP-TbID. Scale bar = 5 μm. The experiments in b-e were repeated at least three times with similar results.

Extended Data Fig. 3. ASY1-eYFP-mTb plants: Phenotype and biotinylation.

(a) Siliques (Scale bar = 1 cm), (b) seeds per silique (ASY1-eYFP-mTb (48.65 ± 3.80) vs. Col-0 (51.63 ± 3.91), p=1.04 x 10^-3; two-sided Student’s t test; n=40; *, P<0.01), (c) pollen viability assessed by Alexander staining (Scale bar = 50 μm), (d) male meiotic chromosomes from pachytene to tetrads (Scale bar = 10 μm; DNA stained with DAPI in gray), and (e) minimum chiasmata number (MCN) in WT (n=28) and ASY1-eYFP-mTb (n=27); two-sided Student’s t test; N.S., not significant.Biotinylation in ASY1-eYFP, ASY1-eYFP-mTb and ASY1-eYFP-TbID plants treated with 0.5 mM of exogenous biotin based on (f) indirect immunolocalization of fluorophore-conjugated streptavidin in male meiocytes (Scale bar = 5 μm) and (g) immunoblot analysis. Streptavidin (SA) used for blotting, Coomassie brilliant blue (CBB) stained protein gel as loading control. The experiments in f-g were repeated three times independently with similar results.

Extended Data Fig. 4. Meiosis-specific ASY1-eYFP-TbID and ASY3-eYFP-TbID fusion protein expression.

(a) Fusion protein expression in flower buds of _UBQeYFP-TbID, ASY1-eYFP, ASY1-eYFP-TbID, ASY3-eYFP and ASY3-eYFP-TbID plants revealed using light sheet fluorescence microscopy (LSFM). Scale bar = 50 μm. (b) ASY1-eYFP-TbID and ASY3-eYFP-TbID fusion proteins show similar dynamics in meiotic nuclei when compared with ASY1- and ASY3-eYFP, respectively. ASY1-eYFP duration in meiotic nucleus as example (left) from initial signal appearance until nuclear envelop break down (star) and average time duration determined (right) for ASY1-eYFP-TbID (30.21 ± 1.52)/ASY3-eYFP-TbID (29.55 ± 1.64) and ASY1-eYFP (29.85 ± 1.56)/ASY3-eYFP (30.50 ± 1.18) fusion proteins (n=10 nuclei). N.S., not significant (p=0.4262; one-way ANOVA). Immunolocalization of ASY1 (magenta) and ZYP1 (green) during (c) meiosis in WT (Scale bar = 5 μm) as well as (d) early zygotene and pachytene in ASY1-eYFP-TbID, ASY3-eYFP-TbID and WT plants (Scale bar = 10 μm). Note, ASY1 and ZYP1 presence in WT tetrad nuclei in (c). The experiments in a, c and d were repeated at least three times with similar results.

Extended Data Fig. 5. Assessment of endogenous biotin(ylation), turnover of biotinylation (on meiotic chromatin) and impact of exogenous biotin treatment.

(a) In WT anthers, immunolocalization of biotin (magenta) in organelles of mitotic cells. Scale bar = 5 μm. (b) Male meiotic chromosome spreads from leptotene to tetrad stage from flower buds of Col-0, ASY1-eYFP-TbID and ASY3-eYFP-TbID plants treated with 0.5 mM of exogenous biotin for 20 hrs and immunostained for biotinylation (magenta). DNA counterstained with DAPI in blue. Scale bar = 5 μm. (c) Scatter plots depicting increased protein intensities after biotin treatment in ASY1-eYFP-TbID. Statistical significance of differentially expressed proteins was determined using limma. The experiments in a-b were repeated at least three times with similar results.

Extended Data Fig. 6. Peptide coverage of ASY1 and ASY3 and in planta phosphorylated residues.

(a, b) Blue highlighted are combined peptide coverage for ASY1 and ASY3 by MS analysis. Residues (bold and underlined) identified being phosphorylated by MS analysis within (a) ASY1 (in total six samples, triplicate of both ASY1-eYFP-TbID with and without biotin treatment) and (b) ASY3 (three samples, triplicate of ASY3-eYFP-TbID with biotin treatment), respectively. Indicated above each site in how many samples a given residue was covered (denominator) and in how many of these the residue was found being phosphorylated (numerator) by MS analysis.

Extended Data Fig. 7. ASY1-eYFP-TbID and ASY3-eYFP-TbID proximate candidates after the first filtering step versus ASY1-eYFP.

(a) Venn diagram shows candidates identified for ASY1-eYFP-TbID and/or ASY3-eYFP-TbID (versus ASY1-eYFP). See also Supplemental Table 1. (b) Proteins with reported meiotic function. (c-f) Cellular component and Molecular function of candidates identified from ASY1-eYFP-TbID (c, e) or ASY3-eYFP-TbID (d, f) by gene ontology classification. GO analysis performed by PANTHER (https://www.arabidopsis.org/tools/go_term_enrichment.jsp).

Extended Data Fig. 8. Functional dissection of putative ASY1 phosphoresidues in planta.

(a) Phosphorylation sites within ASY1: six predicted ATM/ATR (blue) and two predicted CDK sites (red). (b) Seeds per silique (Col-0 (48.67 ± 4.96, n=24), ASY1^hexa-A (46.68 ± 5.29, n=40) and ASY1^365A+382A (47.13 ± 1.55, n=40)), (c), male meiotic chromosomes and (d) MCN (minimum chiasmata number) in ASY1^hexa-A (six predicted ATM/ATR sites (S/T) modified to A, n=22), ASY1^365A+382A (two predicted CDK sites modified to A, n=29) and WT (Col-0, n=28) plants. N.S., not significant (p=0.4512 in b, p=0.0711 in d; one-way ANOVA). Scale bar = 10 μm.

Extended Data Fig. 9. Expression analysis of selected candidate T-DNA mutants by reverse transcription-PCR and functional dissection of ATC5.

(a) Full-length transcripts (CDS) of candidate genes (ATC3, ATC5, ATC8, ATC21, ATC22, ATC23 and ATC28) are present in Col-0, while absent in their respective homozygous T-DNA insertion mutants. For ATC3 and ATC21, two T-DNA insertion alleles are tested. ACTIN used as a positive control. (b) Seed setting in fruit pods of Col-0 and atc5. Scale bar = 0.2 cm. (c) ATC5-HA-mRuby2 fluorescence (RFP, magenta) in squashed anther (at tetrad stage) compared with Col-0 as control. Scale bar = 50 μm. The experiments in a, c were repeated three times independently with similar results.

Data availability

All data supporting the findings of this research are presented in the main text, figures and supplementary information. Generated materials are available from the corresponding author upon reasonable request. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE⁷⁶ partner repository with the dataset identifier PXD034241. The gene/protein sequences and accession-codes of genes used in this study are found in the following databases: TAIR (https://www.arabidopsis.org/), Ensembl Plants (http://plants.ensembl.org/index.html). Source data are provided with this paper.