Members of the Arabidopsis auxin receptor gene family are essential early in embryogenesis and have broadly overlapping functions

The TIR1/AFB proteins are auxin co-receptors that mediate diverse responses to the plant hormone auxin. The genome of Arabidopsis encodes six TIR1/AFB proteins representing three of the four clades that were established prior to angiosperm radiation. To determine the role of these proteins in plant growth and development we have performed an extensive genetic analysis involving the generation and characterization of all combinations of multiply mutant lines. We find that loss of all six proteins results in defects in embryogenesis as early as the first division of the apical cell. Mutant embryos progress, but exhibit frequent errors in cell division and proliferation of the suspensor. Despite this dramatic phenotype, a single wild-type allele of TIR1 or AFB2 is sufficient to support growth of the plant throughout vegetative


INTRODUCTION
The phytohormone auxin regulates diverse processes throughout the entire plant life cycle. Auxin acts as a signal to promote cell differentiation during morphogenetic events such as embryogenesis, root development, and shoot organ formation. Auxin also mediates responses to environmental cues such as light, gravity, water availability, and pathogens. Auxin regulation of transcription involves three families of proteins; AUXIN RESPONSE FACTOR (ARF) transcription factors, Aux/IAA transcriptional repressors, and TRANSPORT INHIBITOR RESPONSE1 (TIR1)/AUXIN-SIGNALING F-BOX (AFB) proteins. Auxins, of which indole-3-acetic acid (IAA) is the predominant natural form, are perceived by a co-receptor complex consisting of TIR1/AFB and Aux/IAA proteins. Formation of the co-receptor complex leads to degradation of the Aux/IAA protein and activation of ARF-dependent transcription (Reviewed in (Lavy and Estelle 2016). In addition to this established pathway, recent studies demonstrate that the TIR1/AFB proteins are required for very rapid auxin responses that are independent of transcription (Dindas et al. 2018;Fendrych et al. 2018). The details of TIR1/AFB function in these rapid responses are currently unknown.
Members of the TIR1/AFB protein family are encoded by three pairs of ohnologs in the Arabidopsis genome. Each protein contains an amino-terminal F-Box followed by eighteen leucine-rich repeats (LRRs). Only tir1, afb2, and afb5 mutants have been identified in forward-genetic screens (Ruegger et al. 1997;Ruegger et al. 1998;Alonso et al. 2003;Walsh et al. 2006;Parry et al. 2009), but reverse-genetic analyses revealed functional redundancies between TIR1, AFB2, and AFB3 as well as between AFB4 and

AFB5.
Gene duplication events provide the primary source material for the evolution of biological innovation. In plants, whole genome duplication (WGD) events have been especially important since events have preceded the radiation of several key plant lineages including seed plants, flowering plants, and core eudicots (Reviewed in . Following duplication, the paralogs are often redundant, allowing one copy to degenerate into a pseudogene. In Arabidopsis, the average half-life of a duplicate gene has been estimated at 17.3 million years (Lynch and Conery 2003). In many cases, however, both duplicates are retained for one or a combination of reasons (reviewed in (Panchy et al. 2016). Occasionally, one of the paralogs evolves a novel function (neofunctionalization), but often the two paralogs fulfill different aspects (enzymatically, temporally, or spatially) of the role of the ancestral gene (subfunctionalization). Following subfunctionalization, there may be changes in selective pressure allowing each paralog to evolve specialized functions without affecting functions carried out by the other paralog. This mechanism likely played a prominent role in the evolution of plant gene families and, in turn, in the radiation and diversification of land plants.
The TIR1/AFB, Aux/IAA, and ARF gene families expanded during land plant evolution after the divergence of bryophytes and vascular plants (Remington et al. 2004;Rensing et al. 2008). Because auxin has a central role in many important adaptations that occurred during land plant evolution, such as vascular development, lateral root formation, and organ polarity; it seems likely that the acquisition of new roles for auxin was enabled by the duplication and diversification of these three gene families. Here we present the comprehensive genetic analysis of the TIR1/AFB gene family of Arabidopsis which revealed extensive functional overlap between even distantly related members as well as an essential role for the TIR1/AFB pathway in early embryos.

Major lineages of auxin receptors diverged prior to fern-seed plant split
To better understand the timeframe during which the auxin receptor family diversified, we built upon previous phylogenetic analyses (Parry et al. 2009;Mutte et al. 2018) with more taxon sampling at key nodes. As shown earlier (Hori et al. 2014;Mutte et al. 2018), the TIR1/AFB genes likely evolved from a gene encoding an F-Box/LRR protein similar to those present in the genomes of extant streptophyte algae. These algal proteins form a sister clade to three distinct land plant F-Box families, the TIR1/AFB auxin receptors, the COI1 jasmonate-Ile (or dinor-OPDA) receptors, and the 'XFB' clade of unknown function conserved in the genomes of mosses and some lycophytes but not in other land plants (Bowman et al. 2018). While the last common  (Schranz and Mitchell-Olds 2006).
One noteworthy aspect of the phylogenetic tree is the pronounced branch-length asymmetry within the TIR1+AFB1 clade (Figure 1-figure supplement 1A and 1B).
Since the last common ancestor of Arabidopsis (Brassicaceae) and Tarenaya (Cleomaceae), the AFB1 gene has accumulated over three times as many nonsynonymous changes as TIR1 (Figure 1-figure supplement 1B) despite being under selection based on the ratio of non-synonymous and synonymous substitutions (Delker et al. 2010;Wright et al. 2017). AFB1 also differs from the other TIR1/AFBs in that it contains two of three substitutions in the first α-helix of the F-Box that were previously shown to weaken TIR1's interaction with CUL1 (Yu et al. 2015). The substitution with the largest effect, Glu8Lys (equivalent to Glu12Lys in TIR1), appeared shortly after the At-β WGD that produced AFB1, and the Phe14Leu substitution appeared prior to the crown group of the Brassicaceae family ( Figure 1-figure supplement 1C). Interestingly, AFB1 orthologs from members of the Camelina genus-C. sativa (all three homeologs), C. laxa, C. hispida, and C. rumelica-additionally contain the third substitution ( Figure   1-figure supplement 1C).

Genetic Analysis of the Arabidopsis TIR1/AFB Gene Family
Previous studies have assessed the functional overlap between the TIR1, AFB1, AFB2 and AFB3 genes (Dharmasiri et al. 2005;Parry et al. 2009) and separately between the AFB4 and AFB5 genes (Prigge et al. 2016). To study the genetic interactions between all members of the family, and to determine the effects of the complete absence of the TIR1/AFB-mediated auxin signaling, lines with strong loss-offunction mutations in the six TIR1/AFB genes were intercrossed to generate all sixtythree mutant combinations. We used the following alleles tir1-1, afb1-3, afb2-3, afb3-4, afb4-8, and afb5-5 (Figure 1-figure supplement 2A;(Ruegger et al. 1998;Parry et al. 2009;Prigge et al. 2016). The tir1-1 allele, which causes an amino acid substitution within the leucine-rich repeat domain of the protein, has been reported to act as a dominant-negative allele (Dezfulian et al. 2016;Wright et al. 2017). However, we found that the root elongation phenotype of plants heterozygous for the tir1-1, tir1-10, and tir1-9 alleles were not significantly different from each other and each displays a semidominant phenotype (Figure 1-figure supplement 2B). These results argue against a dominant negative effect for tir1-1 since neither tir1-9 or tir1-10 produce significant levels of transcript (Ruegger et al. 1998;Parry et al. 2009). Nevertheless, because it is possible that a dominant-negative effect might be revealed in higher-order mutants and because the afb2-3 allele may not be a complete null allele (Parry et al. 2009), we generated selected mutant combinations using the tir1-10 (Parry et al. 2009) and the afb2-1 (Dharmasiri et al. 2005) T-DNA insertion alleles. The afb2-1 allele was introgressed from the Ws-2 background into the Col-0 background through at least eight crosses. For brevity, mutant line names will be simplified such that "tir1afb25" corresponds to the tir1-1 afb2-3 afb5-5 triple mutant line, for example, unless other allele numbers are specified.
The sixty-three mutant combinations displayed a wide range of phenotypes from indistinguishable from wild type to early-embryo lethality (Figure 1-figure supplement 3). The non-lethal higher-order mutant combinations displayed a cohort of phenotypes associated with mutants defective in auxin signaling: smaller rosettes, reduced inflorescence height, reduced apical dominance, fewer lateral roots, and partially or wholly valveless gynoecia ( Figure 1; Figure 1-figure supplements 3 and 4). The three viable quintuple mutants-tir1afb1245, tir1afb1345, and afb12345-had rosettes approximately half the diameter and inflorescences less than half the height of WT Col-0. Despite being smaller, these lines produced approximately twice as many branches as WT ( Figure 1A; Figure 1-figure supplement 4). Remarkably, lines retaining only one copy of TIR1 (tir1/+ afb12345) or one copy of AFB2 (afb2/+ tir1afb1345) were viable.
The rosettes of these two lines were much smaller than those of WT plants with the tir1/+ afb12345's rosette phenotype being slightly more severe ( Figure 1B). In contrast, afb2/+ tir1afb1345 plants developed shorter primary inflorescences and appeared to completely lack apical dominance as all axillary meristems became active upon flowering. The afb2/+ tir1afb1345 and especially tir1/+ afb12345 plants produced very few seeds, an average 0.6 and 14 seeds per plant, respectively (n=10 and 9). Lines containing the alternate alleles-afb2-1/+ tir1-10 afb1345 and tir1-10/+ afb2-1 afb1345-displayed indistinguishable phenotypes ( Figure 1B), although they tended to produce slightly more seed, 1.1 and 36 seeds per plant, n=18 and 8, respectively, so we cannot rule out a slight difference in allele behavior.
Auxin plays an important role in many aspects of root development. To begin to assess the role of the TIR1/AFBs during root growth, we measured the effect of exogenous IAA on primary root growth in the mutant lines. The response ranged from indistinguishable from WT to nearly insensitive to 0.5 µM IAA (Figure 1-figure supplement 4). As seen previously (Dharmasiri et al. 2005;Parry et al. 2009), the roots of lines containing the tir1 and afb2 mutations displayed strong IAA resistance. In addition, we found that the afb3 and afb5 mutations had substantial effects on auxin response, while the afb4 mutation had a more modest effect. The mutant lines also responded similarly to exogenous auxin with respect to lateral root production. The lines with the strongest resistance to IAA in root elongation produced very few if any lateral roots ( Figure 1-figure supplement 4).

Relative Gene Effects
Each of the tir1/afb mutations, except for afb1, affected the phenotypes that we scored but to various extents. To appraise the effects of each mutation on the phenotypes, we plotted the mean values for each phenotype minus that of the corresponding line without that mutation. Larger effects are indicated by greater deviations from zero. For both the root elongation assay and the induction of lateral root primordia, the tir1 allele had the largest effect with the afb2, afb5, afb3, and afb4 mutations having smaller median effects (Figures 2A and 2B). The afb1 mutation had little or no effect on root elongation but surprisingly had an opposite effect on lateral root formation.
Only tir1 and, to a lesser degree, afb2 affect rosette diameter in most contexts with the median effects for afb3, afb4, and afb5 being very close to zero ( Figure 2C). However, they have huge effects together: the afb2345 quadruple mutant is over 6 cm smaller than each of the four triple mutants (arrowheads; Figures 2C). Consistent with previous reports that AFB5 plays a key role in regulating inflorescence branching and height (Prigge et al. 2016;Ligerot et al. 2017), the afb5 mutation has the largest effect on these phenotypes, although each mutation, except for afb1, had some effect ( Figures 2D and 2E).
While the afb1 mutation had minimal effect on most aspects of plant growth, it suppressed the lateral root phenotype of some mutant lines both with and without auxin treatment ( Figure 2B; below). We found that the afb1 mutation suppressed the phenotype of both the afb234 (2.15±0.13 versus 1.75±0.10 lateral roots/cm) and afb345 triple mutants (3.10±0.13 versus 1.96±0.14 lateral roots/cm) measured after 12 days on media not supplemented with IAA. This behavior was not observed in an otherwise WT background (2.76±0.11 for afb1 versus 3.23±0.15 lateral roots/cm for Col-0) nor in a tir1-1 background (1.75±0.08 versus 2.34±0.09 lateral roots/cm for tir1). Each of the pairs were significantly different (two-tailed t-test, P < 0.03).

Defects in Embryonic Root Formation
The tir1afb23 and tir1afb123 lines were previously shown to display a variably penetrant phenotype in which seedlings lack roots, lack both roots and hypocotyls, or fail to germinate (Dharmasiri et al. 2005;Parry et al. 2009). All lines homozygous for both tir1 and afb2 plus either afb3, afb4, or afb5 display these defects to some degree ranging from 1% in tir1afb24 to 99% in tir1afb1234 ( We had noticed a sizeable difference in the proportion of seedlings lacking roots from different batches of seeds. To test whether the temperature at which the seeds mature affects the penetrance of the rootless seedling phenotype, we grew tir1afb23, tir1afb123, tir1afb245 and WT in parallel at 17°C, 20°C, and 23°C and scored the progeny seedling phenotypes (Figure 1-figure supplement 5). The penetrance of the phenotype for all three lines was significantly lower at 20°C than at either 17°C or 23°C for all with the exception that the difference with tir1afb245 at 17°C was not significant using the Fisher's exact test. This suggests that aspects of the auxin regulatory system are sensitive to temperature.

Early-Embryo Defects of the tir1afb235 and tir1afb12345 mutant lines
Because tir1afb235 seedlings were not identified among the progeny of tir1/+ afb235 or afb2/+ tir1afb35 plants, we examined developing embryos in the siliques from these lines. Roughly one-quarter of the embryos from each line lacked cotyledons and had over-proliferated suspensors while the rest had a WT phenotype ( Figure 3A-3B).
Identical defective embryos were seen for lines additionally homozygous for the afb1 mutation ( Figure 1-figure supplement 3). Because all mutant combinations that we expected to produce one-quarter tir1afb2345 or tir1afb12345 progeny were either seedling lethal or infertile, we looked at the progeny of plants that were heterozygous at two loci: tir1/+ afb2/+ afb345, tir1/+ afb2/+ afb1345, and tir1/+ afb5/+ afb1234. We observed embryos indistinguishable from those seen for the tir1afb235 quadruple mutant at a rate close to the expected 1/16 ratio.
To facilitate a more detailed analysis of the sextuple mutant, we assembled the complementing genomic fragments encoding TIR1, AFB2, and AFB5, each carboxy-terminally fused with the coding sequences for different monomeric fluorescent proteins (mOrange2, mCitrine, and mCherry, respectively) into a single locus (Figure 3-figure supplement 1). This construct was transformed into afb1234 plants, crossed into a sextuple mutant background, and two TO/5Ch/2Ci lines were identified that complemented the sextuple mutant phenotype when hemizygous and segregated as a single locus. Using this approach, one-quarter of the progeny of plants hemizygous for these transgenes display embryo defects, while the complemented siblings are easily identified because they expressed fluorescent TIR1, AFB2, and AFB5 fusion proteins.
The first discernable difference between sextuple mutants and the complemented siblings is that the initial division of the embryo proper was displaced from vertical in nearly half of the sextuple 2-cell embryos (6 of 15) compared to only one slightly skewed division out of 18 sibling embryos ( Figure

Marker gene expression in tir1afb235 quadruple mutant embryos
To learn more about the early embryo defects, we introgressed marker genes into lines segregating the tir1afb235 quadruple mutant. Expression of the auxinresponsive marker DR5rev:3×Venus-N7 was undetectable in embryos displaying the mutant phenotype ( Figure (Crawford et al. 2015). In mutants, NTT-YPet appears normally in the suspensor cells, but not in the hypophysis, and is progressively lost in the distal suspensor cells before the abnormal lateral cell divisions occur ( Figure 4V-X) . PIN7-Venus is also normally expressed in the suspensor, hypophysis, and hypophysis-derived cells ( Figure 4K-L). In mutant embryos, PIN7-Venus is faintly detectable in these cells in early-globular embryos. Unexpectedly, the signal is stronger in the lower tier of the embryo proper, starting at the late-globular stage ( Figure 4O-P). The same pattern was observed with the PIN7-GFP marker ( Figure 4U). The PIN1-Venus marker initially is expressed in a reciprocal pattern to PIN7-Venus, in all the cells above the hypophysis and is later refined to strips from the provascular cells out to the cotyledon tips ( Figure 4I-J). In the mutants, PIN1-Venus appears faintly in early embryos and never resolves to specific cell types in later embryos ( Figure 4M-N). The WOX5:GFP marker is first expressed in the hypophysis prior to its asymmetric division then persists in the quiescent center cells ( Figure 4D).
No WOX5:GFP signal was detected in the mutant embryos ( Figure 4H).

Viability of sextuple mutant gametophytes
Because the maternal supply of auxin and the endosperm both play important roles in embryo development, it is possible that female gametophytes lacking auxin receptors would not be viable. That we were able to identify sextuple mutant embryos shows that the gametophytically-expressed auxin receptors are not completely required, but it is possible that they are required for robust transmission. To test the transmission through sextuple mutant megagametophytes and pollen, we carried out reciprocal crosses between wild type (Col-0) and lines homozygous for four loci and heterozygous for the other two (tir1/+ afb2/+ afb1345, tir1/+ afb5/+ afb1234, afb2/+ afb5/+ tir1afb134, and afb2/+ afb4/+ tir1afb135) and then determined the genotypes of all the progeny ( Figure 3-figure supplement 2). When the tir1/afb mutant lines are used as the pollen donor, approximately one-quarter of the progeny were heterozygous at all six loci indicating that transmission of sextuple mutant pollen is not discernably different from non-sextuple pollen. When the tir1/afb lines were used as the female parent, though, we detected fewer than the expected number of progeny that are heterozygous at all six loci, although the deviation was not highly significant ( 2 = 3.74, P=0.053, for the four populations combined). Thus, gametophytically expressed TIR1/AFBs may contribute to megagametophyte viability but are not essential.

TIR1/AFB protein expression
To see if differences in expression patterns can account for the differences in

DISCUSSION
The TIR1/AFB protein family has expanded through a series of gene duplication events that began before fern-seed-plant divergence. Despite the fact that three major subclades were established approximately 400 MYA (Morris et al. 2018), our genetic studies reveal that the TIR/AFB proteins retain largely overlapping functions, at least during standard growth conditions. In general, TIR1 is most important for normal growth and development, but AFB5 and AFB2, and to a lesser extent AFB3 and AFB4, also play significant roles. Spatial differences are also apparent; TIR1 has a major role in the root while AFB5 is relatively more important in inflorescence development.
Although all six genes are broadly expressed, it seems likely that differences in the relative importance of individual TIR1/AFB proteins in various organs are at least partly related to differences in expression. For example, AFB5 is more broadly expressed than the other genes in the inflorescence while in the root, TIR1 and AFB2 are most highly expressed. The AFB4 gene is expressed at a lower level in all tissues consistent with its relatively minor role. Additional differences in patterns of expression are also apparent, particularly in the inflorescence. Further studies will be required to determine if these differences are important.
Our studies demonstrate that the levels of the TIR1/AFB proteins are not uniform throughout the plant. This is true for individual members of the family and for total TIR1/AFB levels across different tissues and cell types. Earlier experiments also showed that TIR1/AFB levels can be dynamic in a changing environment (Wang et al. 2016). These observations may have important implications for use of DII-Venus-based auxin sensors to estimate relative auxin levels, since levels of the sensor protein are dependent on both auxin and the TIR1/AFBs (Brunoud et al. 2012;Liao et al. 2015).
It is important to emphasize that individual members of the family may have specialized functions in particular environmental conditions. For example, the microRNA miR393 is known to target TIR1, AFB2, and AFB3 but not other members of the family (Jones-Rhoades and Bartel 2004; Navarro et al. 2006). Regulation of miR393 abundance modulates the levels of these three TIR1/AFBs to facilitate various growth processes, such as lateral root formation and hypocotyl elongation in response to environmental signals (Vidal et al. 2010;Pucciariello et al. 2018).
Previous in vitro studies have documented some differences in the biochemical activity of members of the TIR1/AFB family (Calderón Villalobos et al. 2012;Lee et al. 2014). Similarly, an auxin-induced degradation assay in yeast reveals differences in the behavior of TIR1 and AFB2 (Wright et al. 2017). In contrast, our results do not reveal any biochemical specificity, except for AFB1 (see below). Thus, a single TIR1 or AFB2 allele is sufficient to support viability throughout the plant life cycle albeit with dramatically reduced fertility. This contrasts to functional diversification seen in other well-studied gene families that diverged in a similar time frame such as the phytochrome photoreceptors and Class III HD-Zip transcriptional regulators (Prigge et al. 2005;Franklin and Quail 2010;Strasser et al. 2010). It is possible that the retention of overlapping functions reflects stricter constraints on TIR1/AFB protein function. One possibility is that the different TIR1/AFB paralogs have been maintained because they contribute to the robustness of the auxin signaling system. Of course, specific functions may be revealed in future studies.
The AFB1 protein is unique among the auxin co-receptors and may have undergone neofunctionalization during the early diversification of the Brasssicales order.
Although AFB1 can interact with Aux/IAA proteins in an auxin-dependent manner, it is not able to efficiently assemble into an SCF complex and is not primarily localized in the may negatively regulate the other members of the family through its ability to sequester auxin, Aux/IAAs, ASK1 and/or other TIR1/AFB-interacting partners. For example, at high auxin concentrations, AFB1 may act to reduce the level of auxin available to the other active co-receptors. It is noteworthy that unlike the other TIR1/AFB genes that are broadly expressed in most cells, AFB1 is expressed very highly in some tissues (root epidermis and vascular tissue) and not at all in others (meristematic pericycle). Based on our genetic studies, AFB1 appears to have a negative effect on lateral root initiation in the afb234 and afb345 lines despite the fact that AFB1 is not expressed in the pericycle. Perhaps AFB1 sequesters auxin that is flowing through the epidermis or stele and limits the amount of auxin that reaches the pericycle.
Most of TIR1 and AFB2 through AFB5 are localized to the nucleus as expected, but we also observe some protein present in the cytoplasm. The significance of this localization is unknown. Perhaps cellular Aux/IAA pools are regulated in part through degradation in the cytoplasm. One intriguing possibility is that cytoplasmic co-receptors are important for the recently described very rapid auxin responses in the root (Dindas et al. 2018;Fendrych et al. 2018).
The importance of auxin in patterning of the developing embryo is well established (Palovaara et al. 2016). Auxin signaling, as evidenced by activity of the DR5 reporter, is first apparent in the apical cell of the embryo (Friml et al. 2003). The essential role of auxin in the apical cell and later in the hypophysis is clearly demonstrated by the defects in the division of these cells in the tir1afb235 quadruple and tir1afb12345 sextuple mutant (Figure 3). Similar defects are observed in a number of other auxin mutants including those affecting response (monopteros and bodenlos), auxin synthesis (yuc1 yuc4 yuc10 yuc11 and taa1 tar1 tar2) and transport (pin1 pin3 pin4 pin7 and aux1 lax1 lax2) (Berleth and Jürgens 1993;Hardtke and Berleth 1998;Hamann et al. 1999;Hamann et al. 2002;Friml et al. 2003;Cheng et al. 2007;Stepanova et al. 2008;Robert et al. 2015). However, none of these lines exhibit the fully penetrant embryo-lethal phenotype observed for the tir1afb235 quadruple and tir1afb12345 sextuple mutants. In the other mutants, a significant fraction of embryos escape embryo lethality and germinate, albeit often as rootless seedlings.
The expression of key embryonic markers in the mutants also reveals profound defects in embryonic patterning by the dermatogen stage. Although tir1afb235 embryos form a morphologically normal hypophysis cell, this cell never expresses NTT-YPet or WOX5:GFP. The proliferation of suspensor cells in the mutant is associated with reduced expression of the suspensor marker PIN7-Venus and to a lesser extent NTT-YPet suggesting that the TIR1/AFB pathway is required to maintain the suspensor cell fate, consistent with an earlier study (Rademacher et al. 2012). PIN1-Venus is expressed in the globular embryo distal from the hypophysis in tir1afb235 mutants as in WT, but none of the dynamic changes related to cotyledon and provascular tissue specification occur. It was surprising to observe that PIN7-Venus exhibits ectopic expression in the basal half of the embryo proper. The reason for this is obscure but one possibility is that PIN7 expression is normally repressed in the embryo by a TIR1/AFB-dependent pathway.

Phylogeny
The sources for the amino-acid sequences ( Taxa were selected based on availability, quality, and diverse sampling at key nodes. A reduced set were included for COI1 homologs. The AFB1 genes from Camelina hispida, C. laxa, and C. rumelica were amplified from genomic DNA using Phusion Polymerase (New England Biolabs or ThermoFisher) and primers to regions of the 5′ and 3′ UTRs conserved in all three C. sativa AFB1 genes. The PCR products were subcloned, and three C. hispida and C. laxa clones and a single C. rumelica clone were sequenced.
The CamhiAFB1 and CamlaAFB1 sequences included in analysis appeared in two of the three clones (GenBank accession numbers MK423960-MK423962).
To build the alignment of F-Box-LRR protein sequences, sequences from distinct subclades were aligned using T-COFFEE (Notredame et al. 2000) to identify and trim unique unalignable regions from individual sequences before aligning the whole set.
Ambiguous regions of the full alignments were removed in Mesquite (Maddison and Maddison 2018). The raw alignment of nucleotide CDS sequences of Brassicales TIR1/AFB1 genes were adjusted so that gaps fell between adjacent codons.
Phylogenetic trees were inferred using MrBayes (Ronquist et al. 2012). For the TIR1/AFB/XFB/COI1 phylogeny, a total of six runs of four chains were split between two Apple iMac computers using the parameters aamodelpr=mixed, nst=6, and rates=invgamma. Only four of the six runs had converged after 16 million generations, so the analysis was restarted with three runs each starting with the best tree from one of the initial runs and with more heating (temp=0.5) for 10 million generations. The

Fluorescently tagged TIR1/AFB lines
Genomic regions containing each of the TIR1/AFB genes were amplified using Phusion polymerase (New England Biolabs or ThermoFisher) from corresponding genomic clones (JAtY51F08, JAtY62P14, JAtY53F15, JAtY61O12, and JAtY52F19) except for AFB3 which was amplified from Col-0 genomic DNA. See Supplementary File 2 for primers used. The PCR products were cloned into pMiniT (New England Biolabs), and the stop codon was altered to create a NheI site using site-directed mutagenesis.

Microscopy
For confocal microscopy of the root meristem, five-to seven-day-old seedlings were stained in a 10 µg/ml aqueous solution of propidium iodide for one minute, rinsed in water, mounted with water, and viewed with either a Zeiss LSM 880 inverted microscope or a Zeiss LSM 710 inverted microscope. Embryos were fixed and stained with SCRI Renaissance 2200 (SR2200; Renaissance Chemicals, UK; (Crawford et al. 2015). Briefly, using fine forceps and a 27-gauge needle as a scalpel, developing seeds were dissected from siliques and immediately immerged in fix solution (1×PBS, 4% formaldehyde (Electron Microscopy Sciences, 15713), and 0.4% dimethyl sulfoxide) in a six-well plate with 100µ-mesh strainers. A vacuum was pulled and held three times for 12 minutes each time, before rinsing twice with 1×PBS for 5 minutes. The embryos were transferred to SR2200 stain [3% sucrose, 4% diethylene glycol, 4% dimethyl sulfoxide and 1% SR2200 and stained overnight with vacuum pulled and released 3-4 times. Seeds were mounted (20% glycerol, 0.1×PBS, 0.1% dimethyl sulfoxide, 0.1% SR2200, and 0.01% Triton X100) and the embryos were liberated by pressing on the coverslip. To detect mCitrine in the shoot apices, we removed stage 5 and older floral buds using fine forceps, fixed and rinsed (as with the embryos), soaked in ClearSee (Kurihara et al. 2015) for seven to ten days changing the solution every two to three days, and then stained with basic fuchsin (not shown) and Fluorescent Brightener 28 (Calcofluor White M2R) as described (Ursache et al. 2018). Confocal image channels were merged using ImageJ or FIJI (Schindelin et al. 2012;Schneider et al. 2012).
Cleared embryos were viewed by mounting dissected ovules in a solution containing 2.5 g chloral hydrate dissolved in 1 ml 30% glycerol and viewing with a Nikon E600 microscope.

Phenotype comparisons
The viable tir1afb lines were divided based on whether they contained the tir1 mutation, and the two batches were grown sequentially. The afb123 line included in the initial batch was noticed to display a long-hypocotyl phenotype that may have been picked up after an earlier cross to the afb4-2 mutant, so a third batch was made up of alternative isolates for five lines whose pedigrees included a cross to afb4-2. Each batch included Col-0 and tir1-1. Seeds were surfaced sterilized, stratified in water for five days, spotted onto ½ MS medium containing 1% sucrose, and incubated in a light chamber (22°C). Twelve five-day-old seedlings for each genotype were transferred to 120 mm square plates containing the same medium containing either 0, 20, or 100 nM IAA (batch a), 0, 100, or 500 nM IAA (batch b), or 0, 20, 100, or 500 nM IAA (batch c).
Each plate received six seedlings from six genotypes spread out over two rows.
Seedlings for each genotype were present on the top row of one plate and the lower row on a second plate placed in a different part of the growth chamber after marking the position of the root tips with a marker and scanning with Epson V600 flatbed scanners.
The plates were scanned again after 72 hours, and the growth during the 72 hours was measured using imageJ. The plates containing 100 nM IAA were grown for a fourth day before the numbers of lateral roots protruding through the epidermis were counted using a dissecting microscope. Five seedlings from the no-IAA control plates were transferred to soil in 6cm pots and grown an additional 34 days. The genotypes for two plants per line were confirmed by PCR. For each 42-day-old plant, the height from the rosette to the tip of the longest inflorescence and the maximum rosette diameter were measured, and the numbers of branches of at least 1 cm were counted. The IAA effects on root elongation data is presented as the percent relative to the growth without IAA ± the relative standard error of the ratio. For the gene effect analyses, the averages from each batch were normalized using measurements for Col-0 and tir1-1 plants that were included with each batch.

ACKNOWLEDGMENTS
We thank Yingluo Wang, Diane Le, and Whitnie Szutu for technical assistance; Yi Zhang and Brian Crawford for help with microscopy; and the Arabidopsis Biological Resource Center and the US National Plant Germplasm System for seeds. This work was supported by a grant from the NIH (GM43644 to ME). NK was supported in part through a UC San Diego Biological Sciences Eureka Summer Research Scholarship.   (Figure 1-figure supplement 4), the normalized mean for each genotype with the given mutation was subtracted from the normalized mean for the corresponding genotype lacking that mutation and plotted (circles). The red bars indicate the median difference attributable to the given mutation. (A) Effects of each mutation on IAA-inhibition of root elongation. For each genotype, twelve five-day-old seedlings were transferred and grown for three days on media containing 100 nM IAA, and their average growth was divided by that of twelve seedlings grown on media lacking added auxin. (B) Effects of each mutation on auxin-induced lateral root production. Twelve five-day-old seedings for each genotype were grown for four days on media containing 100 nM IAA and the numbers of emerged lateral roots were     in the first helix of the F-Box that were shown to interfere with SCF assembly. The Salvadora AFB1 transcript assembly lacked the sequence encoding this helix so that ancestor's sequence could not be predicted.   (Figure 1-figure supplement 4), the ranges for each of the phenotypes were divided into five bins, from "-" to "++++" in increasing severity. NA, Not applicable (due to embryo or seedling lethality); ND, not determined.     Figure 5. The same microscope settings for mCitrine detection in panels A-B, E-G, J-K (asterisks). The settings used for panels C and L were less sensitive (-) and those for panel I were more sensitive (+).

FIGURES AND TABLES
Scale bars equal 25 µm. M, Sensitivities of AFB4-expressing transgenic lines to picloram. Root elongation was measured for seedlings grown on media containing 20 µM picloram, expressed as a percentage of elongation on media lacking picloram. Lines AFB4-mCitrine#3 and AFB4-tdTomato#16 are more sensitive to picloram than WT indicating that the transgene is expressed at higher levels than the endogenous AFB4 locus.  genotype with the given mutation was subtracted from the normalized mean for the corresponding genotype lacking that mutation and plotted (circles).

Supplementary
The red bars indicate the median difference attributable to the given mutation. (A) Effects of each mutation on IAA-inhibition of root elongation. For each genotype, twelve five-day-old seedings were transferred and grown for three days on media containing 100 nM IAA, and their average growth was divided by that of twelve seedlings grown on media lacking added auxin. (B) Effects of each mutation on auxin-induced lateral root production. Twelve five-day-old seedings for each genotype were grown for four days on media containing 100 nM IAA and the numbers of emerged lateral roots were