Ectopic expression of miR160 results in auxin hypersensitivity, cytokinin hyposensitivity, and inhibition of symbiotic nodule development in soybean

Symbiotic root nodules in leguminous plants result from interaction between the plant and nitrogen-fixing rhizobia bacteria. There are two major types of legume nodules, determinate and indeterminate. Determinate nodules do not have a persistent meristem while indeterminate nodules have a persistent meristem. Auxin is thought to play a role in the development of both these types of nodules. However, inhibition of rootward auxin transport at the site of nodule initiation is crucial for the development of indeterminate nodules, but not determinate nodules. Using the synthetic auxin-responsive DR5 promoter in soybean, we show that there is relatively low auxin activity during determinate nodule initiation and that it is restricted to the nodule periphery subsequently during development. To examine if and what role auxin plays in determinate nodule development, we generated soybean composite plants with altered sensitivity to auxin. We over-expressed microRNA393 to silence the auxin receptor gene family and these roots were hyposensitive to auxin. These roots nodulated normally suggesting that only minimal/reduced auxin signaling is required for determinate nodule development. We overexpressed microRNA160 to silence a set of repressor ARF transcription factors and these roots were hypersensitive to auxin. These roots were not impaired in epidermal responses to rhizobia, but had significantly reduced nodule primordium formation suggesting that auxin hypersensitivity inhibits nodule development. These roots were also hyposensitive to cytokinin, and had attenuated expression of key nodulation-associated transcription factors known to be regulated by cytokinin. We propose a regulatory feedback loop involving auxin and cytokinin during nodulation.


Introduction
Auxin plays a crucial role in the initiation and development of a number of plant organs (reviewed in (Vanneste and Friml, 2009)). Auxin is perceived by a set of TIR1-like F-box proteins that form part of an SCF-type ubiquitin ligase complex (Ruegger et al., 1998;Dharmasiri et al., 2005;Kepinski and Leyser, 2005). Molecular and biochemical characterization of TIR1 in arabidopsis revealed that it directly binds auxin and aids in the degradation of Aux/IAA repressor proteins (Dharmasiri et al., 2005;Kepinski and Leyser, 2005). Aux/IAAs are short-lived nuclear proteins encoded in general by primary/early auxin response genes (Ulmasov et al., 1997). Aux/IAA proteins form protein complexes with auxin response factor (ARF) transcriptional activators and repress the expression of auxin-responsive genes (Reed, 2001;Tiwari et al., 2001;Tiwari et al., 2004). Dissociation of ARF-Aux/IAA protein complexes by TIR1 in an auxin-dependent manner results in auxin-responsive gene expression (Gray et al., 1999;Gray et al., 2001). The ARF gene family consists of both transcriptional activators and repressors that bind with specificity (at auxin response elements or auxREs) to promoters of primary/early auxin response genes. Repressor ARFs are capable of repressing gene expression with or without auxin (Ulmasov et al., 1999;Tiwari et al., 2003). It has been hypothesized recently that they are more likely to act independent of Aux/IAA proteins (Guilfoyle and Hagen, colonization of root hairs by rhizobia cells leads to root hair deformation and curling to entrap rhizobia. Subsequent invagination of the root hair cells forms specialized structures termed infection threads. Simultaneously, cell divisions occur in cortex and pericycle cell layers in preparation to form nodule primordia. Infection threads transport rhizobia in to the developing nodules (formed by cortex cell divisions) where they differentiate into bacteroids and fix nitrogen. Ultimately, the two cell division foci merge and symplastic and vascular connections are formed enabling transport of nutrients to and from the mature nodules. There are two major types of nodules formed in legume roots: indeterminate and determinate (reviewed in (Hirsch, 1992;Sprent, 2007)). Indeterminate nodules are oblong and characterized by the presence of a persistent meristem. Examples of plants that form indeterminate nodules include temperate legumes viz. pea, Medicago truncatula and clover. In contrast, determinate nodules are spherical and lack a persistent meristem. There is no sustained cell division during determinate nodule development; cell expansion rather than cell division results in nodule growth. Examples of plants producing determinate nodules include tropical legumes viz. soybean, common bean, and Lotus japonicus. Additionally, indeterminate nodules arise from inner cortical cell layers whereas determinate nodules arise from outer/mid cortical cell layers. A number of different plant hormones including auxin and cytokinin are hypothesized or shown to play a role in the initiation and development of both these types of nodules (Hirsch and Fang, 1994;Ferguson and Mathesius, 2003;Ding and Oldroyd, 2009;Suzaki et al., 2013).
The roles of auxin in nodule initiation and development were hypothesized initially based on the effect of exogenous application of auxin or auxin transport inhibitors (Allen et al., 1953;Hirsch et al., 1989), subsequently based on the expression of reporter gene constructs (e.g. GH3:GUS (Mathesius et al., 1998;Boot et al., 1999;Pacios-Bras et al., 2003;Takanashi et al., 2011)) and 7 1998)) nodule-producing legumes. However, the expression pattern was clearly distinct in the roots and other nodule tissues between the two types of nodules. For example, GH3:GUS expression was reduced in the vasculature rootward to the inoculation site, but increased in cortex cells around the site of rhizobial inoculation in white clover, an indeterminate noduleforming legume (Mathesius et al., 1998). Subsequently, GH3:GUS expression was detected in the dividing primordium cells, but moved to the peripheral cells (presumably those forming the nodule vasculature) during differentiation. In contrast, GH3:GUS expression in the root vasculature was not reduced in response to rhizobial inoculation in L. japonicus, a determinate nodule-forming legume. Nevertheless, GUS expression was detected initially in the dividing cells, later in peripheral cells in nodule primordia, and in the nodule vasculature in mature nodules (Pacios-Bras et al., 2003;Takanashi et al., 2011). GH3:GUS expression pattern in root vasculature and measurement of auxin transport suggested that inhibition of rootward auxin transport might precede the initiation of indeterminate nodules, but not determinate nodules (Mathesius et al., 1998;Boot et al., 1999). We and others obtained clear genetic evidence for the role of auxin transport inhibition in nodule development by silencing the biosynthesis of flavonoids, a group of phenolic compounds that act as endogenous auxin transport inhibitors (Subramanian et al., 2006;Wasson et al., 2006;Zhang et al., 2009). These data conclusively demonstrated that inhibition of auxin transport in the roots at the site of nodule initiation is crucial for the development of indeterminate nodules, but not for determinate nodules. Other studies showed that components of auxin transport machinery (PIN2 and LAX) might play a role in the development of indeterminate nodules as well (de Billy et al., 2001;Huo et al., 2006). We concluded that the requirements of auxin distribution during primordium initiation/development are different between these two types of nodules (Subramanian et al., 2007).
Another line of evidence pointing towards a role for auxin in nodule development is the coordinate expression of microRNAs (miRNAs) regulating auxin signaling during this process in 8 receptor TIR1/AFB gene family (Parry et al., 2009;Vidal et al., 2010;Si-Ammour et al., 2011), miR160 regulates members of the ARF10/16/17 repressor ARF family (Mallory et al., 2005;Wang et al., 2005), miR167 regulates ARF8 family (Wu et al., 2006) and miR390 regulates ARF3 family (Fahlgren et al., 2006;Marin et al., 2010). In soybean roots, the expression of miR393 that regulates the TIR1/AFB auxin receptor gene family was transiently up-regulated where as miR160 that regulates a group of repressor ARFs was down-regulated upon Bradyrhizobium japonicum inoculation. More recently, detailed analysis of the specific and sensitive auxin-inducible marker, DR5:GFP-NLS marker in L. japonicus showed that auxin induction during nodule primordium development might occur downstream of cytokinin perception (Suzaki et al., 2012). While all of these data strongly suggested that auxin signaling is regulated during nodule development, it was not known if and what role auxin plays during the development of symbiotic nodules, specifically determinate nodules. We sought to address this long-standing question by (i) monitoring the activity of the auxin-specific synthetic promoter DR5 during soybean nodule development and (ii) examining nodule development in soybean roots with altered auxin sensitivity.

Auxin-inducible gene expression during root nodule initiation in soybean
We examined auxin-inducible gene expression during nodule development in soybean transgenic composite plants (Collier et al., 2005) using a marker gene construct where the synthetic auxinresponsive DR5 promoter (Ulmasov et al., 1997) drove the expression of the fluorescent protein tandem dimer Tomato (tdT; (Campbell et al., 2002)). To distinguish auto-fluorescence observed on the soybean root surface when using confocal microscopy, we examined DR5:tdT together with GFP driven by the constitutive super Ubiquitin promoter. The yellowish red tdT + GFP signal can be clearly distinguished from the bright red auto-fluorescence on the root surface. DR5:tdT expression was detected in the root meristem and columella cells of the root cap ( Figure 1A) and lateral root (LR) primordia ( Figure 1B). Treatment of the roots with exogenous auxin (1μM 2,4-D for 24h) resulted in an increased tdT expression domain at the root tips accompanied by a noticeable "swelling" of the root tips ( Figure S1A & B). The expression pattern and auxin-responsiveness was consistent with what has been reported in arabidopsis and other species (e.g. (Ni et al., 2001;Ilina et al., 2012)) indicating that the construct was suitable for monitoring auxin-responsive gene expression in soybean composite plant roots.
We inoculated these plants with B. japonicum cells and examined DR5:tdT expression at 3, 7 and 14 days post inoculation (dpi). Surprisingly, we observed a much lower expression of DR5:tdT in the majority of nodule primordia and emerging nodules ( Figure 1C). DR5:tdT expression was localized primarily to the nodule "apex". The level of DR5:tdT expression in nodule primordia was so low that it could not be detected reliably by regular fluorescent microscopy ( Figure S1E), but only by confocal microscopy. In contrast, DR5:tdT expression in the root tips and LR primordia was readily detectable by regular fluorescent microscopy ( Figure   S1 C & D). In transverse sections of mature nodules, DR5:tdT expression was detected closer to the nodule periphery primarily along the vasculature and was clearly absent from the central infection zone ( Figure 1D). We also examined DR5:GUS expression during nodule development, due to the relatively longer half-life and higher sensitivity of GUS. Consistent with results from confocal microscopy, we detected the expression of DR5:GUS in the majority of nodule primordia ( Figure 1E) with intense staining closer to the nodule apex. In transverse-sections of mature nodules, DR5:GUS expression was limited to the nodule periphery and not detected in the infection zone ( Figure   1F). Together, these observations suggested that (i) there is auxin activity during nodule initiation and development in soybean; but the level of auxin activity in nodule primordia is much lower compared to that in LR primordia, (ii) there is no or minimal auxin activity in the nodule infection zone during subsequent nodule development, and (iii) the majority of the auxin activity is localized around the vasculature in mature nodules.

Manipulating auxin sensitivity using miR393 and miR160
To examine the role of auxin in symbiotic nodule initiation in soybean, we sought to manipulate auxin signaling in transgenic composite plant roots by over-expressing miRNAs against the auxin receptor and a set of repressor ARF genes. We reasoned that this would help overcome functional redundancy usually associated with components of auxin signaling in classical genetic mutants (see Discussion). Secondly, we have previously shown that the levels of these miRNAs are influenced by B. japonicum inoculation in soybean (Subramanian et al., 2008). We overexpressed miR393 to silence the TIR1/AFB auxin receptor family and to obtain auxin hyposensitive plants. Over expression of miR393 precursor using the Cassava Vein Mosaic Virus CVP2 promoter (CsVMV, a kind gift from Dr Claude Fauquet, Donald Danforth Plant Science Center, St Louis, MO) resulted in increased levels of mature miR393 ( Figure S2A) and a corresponding decrease in two of the three validated targets of this miRNA ( Figure S2B). We over-expressed miR160 to silence a set of repressor ARFs belonging to ARF10/16/17 family and to obtain auxin hypersensitive plants. Over expression of miR160 precursor also resulted in increased levels of the corresponding mature miRNA ( Figure S2C) and a decrease in the levels of all nine validated targets ( Figure S2D).
We assayed physiological responses to auxin in the roots of these plants by treating them with 0, of LRs after one week. In the absence of exogenous auxin, we observed no statistically significant difference between the vector control and miRNA over-expressing lines in either the rate of root growth (Table S1A) or LR density (Table S1B). When treated with auxin, the elongation of vector control roots was significantly inhibited in response to both levels of auxin treatment (Table S1A) in a dose-dependent manner ( Figure 2A). However, inhibition of root growth in miR393 over-expressing (miR393ox) plants was significantly lower than that of vector control at both levels of auxin treatment ( Figure 2A; Table S1A) suggesting that these roots were hyposensitive to auxin. In contrast, there was significantly enhanced inhibition of root growth in miR160 over-expressing (miR160ox) plants relative to vector control plants at both levels of auxin treatment (Figure 2A; Table S1A). This suggested that miR160ox roots were hypersensitive to auxin. We also examined increase in LR density (both primordia and emerged LRs) in response to auxin treatment in miR393ox and miR160ox roots. Vector control roots had a significant increase in LR density in a dose-dependent manner in response to auxin ( Figure 2B; Table S1B). Similar to the response observed for inhibition of root elongation, miR393ox roots displayed auxin hyposensitivity (reduced LR density vs vector control; Figure 2B; Table S1B) where as miR160ox roots displayed auxin hypersensitivity (significantly higher LR density vs vector control; Figure 2B; Table S1B).
In addition to the above physiological responses, we also assayed molecular responses to auxin in miRox roots by examining the expression of DR5:GUS, the synthetic auxin-responsive marker. In untreated vector control roots, expression of DR5:GUS ("auxin maximum") was observed primarily in the root cap columella and the root meristem ( Figure  to vector control roots. This suggested that these roots were hypersensitive to auxin.

2
We also examined the expression of seven endogenous auxin-responsive marker genes in miRox roots by RT-qPCR (Table 1). Of these five were clearly induced by auxin in vector control roots.
For example, a 40-fold induction of GH3 expression was observed in response to treatment with 1μM 2,4-D for 6h in vector control roots (Table 1). Auxin induction of GH3 was significantly attenuated in miR393ox roots (Table 1). On the other hand, in miR160ox roots, there was a significantly higher induction of GH3 in response to auxin (Table 1). Similarly, four of the six Aux/IAA genes examined were strongly induced by auxin treatment (GmIAA1, IAA8, IAA13 and IAA20). Similar to GH3, auxin induction of all the four genes was significantly attenuated in miR393ox roots, compared to vector control (Table 1). Auxin induction of three of these genes was significantly enhanced in miR160ox roots (the exception being GmIAA13) compared to vector control (Table 1). In summary, at least four of the five auxin-inducible marker genes had an attenuated response in miR393ox roots and enhanced response in miR160ox roots. All the above results strongly indicate that miR393ox resulted in auxin hyposensitivity while miR160ox resulted in auxin hypersensitivity. We used these roots to examine the effect of altered auxin sensitivity on symbiotic nodule development.

Auxin hypersensitivity in the roots or nodule primordia inhibits nodulation
miR393ox and miR160ox composite plants were inoculated with B. japonicum cells and the extent of nodulation examined 14 dpi. Vector control roots had an average of 20.2 ± 1.8 nodules per root. The number of nodules on miR393ox roots (16.3 ± 2.2) did not significantly differ from that of vector control roots (Poisson distribution analysis) suggesting that auxin hyposensitivity did not significantly influence nodulation in soybean ( Figure 3A). In contrast, miR160ox roots had significantly fewer nodules (19.7 ± 1.7 in vector control vs 9.6 ± 1.4 in miR160ox roots) suggesting that auxin hypersensitivity inhibited nodulation in soybean ( Figure 3B). However, this experiment did not answer if this inhibition is a direct effect of auxin sensitivity on nodule tissues or a pleiotropic effect of altered auxin sensitivity in the root system. To address this question, we "mis-expressed" these miRNAs using the nodule-specific soybean ENOD40 promoter (Yang et al., 1993). When examined in composite plant roots using a ENOD40:tdT construct, soybean ENOD40 was not expressed or expressed at very low levels in uninoculated roots (including root tips and LRs), but highly expressed in nodule primordia, emerging nodules and mature nodules in soybean ( Figure S3). Therefore, altered auxin sensitivity in roots misexpressing these miRNAs would be limited to these target tissues. Composite plants misexpressing miR393 (miR393mx) nodulated as efficiently as vector control plants (14.9 ± 2.2 vs. 17.4 ± 2.6 nodules per root respectively; Figure 3C) consistent with results from miR393ox plants. Interestingly, miR160mx plants had significantly less number of nodules compared to vector control roots (15.8 ± 2.4 vs 7.7 ± 1.0; Figure 3D) as observed in miR160ox plants. This suggested that auxin hypersensitivity during and at the sites of nodule primordia formation was sufficient to inhibit nodulation in soybean.

Auxin hypersensitivity did not affect root hair responses to rhizobia, but affected nodule primordium formation
We closely examined nodule development in miR160ox and miR160mx roots to identify which stage(s) of nodule development are influenced by auxin hypersensitivity. We used B. japonicum . miR160ox roots had a significantly lower number of infections (2 ± 1) that showed this response (P=0.002; Figure 4B) but there was no significant difference between vector control and miR160mx (4 ± 1) roots (P=0.1; Figure 4B). This observation suggested that auxin hypersensitivity in the entire root system (resulting from miR160ox) inhibited infection thread growth in the cortex while auxin hypersensitivity in nodule-associated cells (resulting from miR160mx) did not affect this process. We also counted the number of nodule primordia in these roots 8 dpi. Vector control roots had 3.9 ± 0.8 primordia per root ( Figure 4C). Both miR160ox (1.3 ± 0.3) and miR160mx (2.2 ± 0.4) plants had significantly reduced number of nodule primordia ( Figure 4C), but with different levels of severity indicated by statistical significance (P=0.0002 and 0.02 respectively). We conclude that auxin hypersensitivity in the entire root system (resulting from miR160ox) caused a severe inhibition of nodule primordium formation while auxin hypersensitivity in noduleassociated cells (resulting from miR160mx) was sufficient to cause a moderate inhibition of nodule primordium formation.
Next, we examined nodule development and maturation in miR160ox and miR160mx plants by classifying the number of nodules at 14 dpi into emerging and mature nodules (see Methods).
Consistent with a reduction in the number of infection threads reaching the cortex and a severe reduction in the number of nodule primordia, miR160ox roots had a severe reduction in the number of emerging nodules compared to the vector control roots (12.4 ± 1.1 vs. 5.9 ± 0.9; Figure 5A). Interestingly, miR160mx roots also had a severe reduction in the number of emerging nodules (13.9 ± 2.0 vs. 7.2 ± 1.0; Figure 5C) even though these roots were unaffected in the number of infections reaching the cortex and had only a moderate reduction in the number of nodule primordia. This observation suggested that auxin hypersensitivity in nodule tissues might inhibit the development of nodule primordia into emerging nodules. Consistent with the severe reduction in emerging nodules, both miR160ox and miR160mx roots had a severe reduction in the number of mature nodules as well ( Figure 5B & D). Together, our results suggest that auxin hypersensitivity resulting from ectopic expression of miR160 inhibits not only the formation of nodule primordia but also their subsequent development in soybean.

Roots over-expressing miR160 are hyposensitive to cytokinin
Our results suggested that hypersensitivity to auxin inhibits nodule primordium formation and it is known that cytokinin promotes this process (see Discussion). Since auxin and cytokinin are known to act antagonistically during a number of different plant developmental processes, we examined cytokinin sensitivity in roots over-expressing miR160. First we examined the expression of AtARR5:GUS which has been used as a marker for cytokinin activity in legumes (Lohar et al., 2004). In untreated vector control roots, AtARR5:GUS expression was detected primarily in the root cap ( Figure 6A), vasculature ( Figure 6B) and at the base of LR primordia of mature root regions (not shown). There was no obvious difference in AtARR5:GUS expression between untreated vector control and miR160ox roots ( Figure 6C & D). When roots were treated with 5 μ M BAP for 6 hours, there was a clear induction of AtARR5:GUS expression in the root tip and vasculature of vector control roots ( Figure 6E & F). In contrast, a very weak induction was observed in 160ox roots ( Figure 6G & H). This suggested that these roots were indeed hyposensitive to cytokinin.
Next, we examined the expression of soybean orthologs of arabidopsis ARR5 and ARR9 in response to cytokinin. Some of these type-A RRs were shown to regulate nodule development in M. truncatula (Op den Camp et al., 2011). Based on BLASTx searches and previously published information, we identified 4 potential soybean orthologs for each of AtARR5 and AtARR9. All eight genes were significantly induced by 5 μ M BAP as early as 30 min in both vector control roots and miR160ox roots (Table 2). However, the level of cytokinin induction was significantly attenuated in 160ox roots compared to vector control for 6 of these RRs (Table 2). Together, these observations suggested that miR160ox roots were hyposensitive to cytokinin.
Finally, we sought to examine the influence of miR160ox on nodulation pathway genes dependent on cytokinin perception/activity. In Lotus japonicus and soybean, the expression of NIN, NSP1, HAP2-1 and HAP2-2 is induced during nodule development (Heckmann et al., 2011;Hayashi et al., 2012). NIN and HAP2 are directly induced by cytokinin treatment and all these genes act downstream of cytokinin perception during nodule development in L. japonicus. We examined the expression of soybean orthologs of these genes along a time course of B. japonicum inoculation in vector control and miR160ox roots. The expression of NIN and NSP1 increased moderately at 5 dpi and very highly at 8 dpi in response to B. japonicum inoculation ( Figure 6E & F). In miR160ox roots the expression of NIN and NSP1 also increased at moderate levels at 5 dpi, but their expression was much lower than in vector control roots at 8 dpi ( Figure   6E & F). Between the two HAP2 genes we examined, HAP2-2 was induced at much higher levels compared to HAP2-1 in response to B. japonicum inoculation in vector control roots  Figure 6G & H). The expression of these genes was lower in miR160ox roots compared to vector control roots along the entire time course (Figure 6G & H). However, the induction of HAP2-2 was affected to a larger extent in miR160ox roots compared to HAP2-1. We also examined the expression of two nodulation-inducible marker genes, ENOD40 (Yang et al., 1993) and FWL1 (Libault et al., 2010), at 0, 5 and 8 dpi in vector control and miR160ox roots. The expression of both ENOD40 and FWL1 was very low or undetectable at 0 dpi, but significantly increased at 5 and 8 dpi in vector control roots ( Figure 6M & N). Consistent with reduced expression of NIN and NSP1 and reduced nodule primordia initiation, miR160ox roots had significantly lower expression of these genes in response to B. japonicum inoculation ( Figure 6M & N).
Our data clearly show that over expression of miR160 resulted in hypersensitivity to auxin and hyposensitivity to cytokinin. Consistently, the expression of cytokinin-dependent nodulation genes was also reduced in miR160ox roots in response to B. japonicum inoculation. These results suggest that hypersensitivity to auxin and hyposensitivity to cytokinin results in impaired nodule development in soybean.

There is relatively low auxin activity during determinate nodule initiation
Using the auxin-inducible marker gene constructs, DR5:tdT and DR5:GUS, we identified that there is low or transient auxin activity during soybean nodule initiation and development. the rhizobium-colonized zone consistent with our results. Therefore, the use of markers with enhanced detectability (e.g. GFP-NLS) or stability (e.g. GUS) detected auxin-inducible gene expression during nodule initiation suggesting that it was low or transient. The observation that DR5:tdT was readily detectable in LR primordia and emerging LRs, but not in nodule primordia or nodules suggested that the level of auxin activity during determinate nodule initiation and development is much lower than that during LR initiation. In addition, it appears that an initial auxin maximum occurs during nodule primordium initiation and subsequently diminishes from the primordium in to the nodule periphery during nodule maturation (our results and those of (Suzaki et al., 2012). We conclude that the requirement of auxin activity changes in a spatiotemporal manner during determinate nodule development.
We over-expressed or mis-expressed specific miRNAs to modulate auxin sensitivity in soybean roots. We reasoned that the use of miRNAs would overcome pitfalls of functional redundancy often associated with genetic mutants especially in auxin signaling. Unfortunately, soybean mutants with genetic lesions in TIR/AFB genes are not available for such a comparison to miR393ox roots. Nevertheless, reduced sensitivity to auxin in miR393ox soybean roots resulted in reduced LR initiation, but did not affect nodule formation. Due to the expression of DR5:TdT along the vasculature in mature nodules, we also examined miR393ox nodules under light microscope for any defects in vascular development. We observed no obvious defects in vascular development in miR393ox nodules. This suggested that only minimal TIR1/AFB activity is required for proper nodule formation and development in soybean (and likely other determinate nodule-forming legumes as well). This is consistent with the reduced level of auxin-responsive gene expression observed during determinate nodule formation.

Auxin hypersensitivity inhibits determinate nodule development
Over expression of miR160 resulted in a clear auxin hypersensitivity in soybean roots. Similar observations have been made in arabidopsis as well. For example, miR160ox arabidopsis plants had increased LR density (Wang et al., 2005) and proteins encoded by ARF10, 16 and 17 (targets of miR160) have been proposed to encode repressor ARF proteins (Mallory et al., 2005).
The presence of P and S-rich middle regions suggested that soybean ARF10/16/17 family members might act as repressors as well ( Figure S5). Indeed, a mutation in miR160 leading to increased expression of ARF10, 16 and 17 resulted in auxin-resistant phenotypes in arabidopsis (Liu et al., 2010). In addition, miR160ox resulted in consumption of the root cap in arabidopsis www.plantphysiol.org on August 19, 2017 -Published by Downloaded from Copyright © 2013 American Society of Plant Biologists. All rights reserved. (Wang et al., 2005). Surprisingly, we observed intact root caps in soybean miR160ox roots when examined normally (e.g. Fig. 2) or through columella starch staining (not shown). It is possible that this difference is due to the different promoters used to drive over-expression of miR160 in arabidopsis and soybean.
Results from miR160mx roots suggested that auxin hypersensitivity in nodule tissues causes only a moderate inhibition of primordium formation, but a severe inhibition of subsequent nodule development in soybean. What role does auxin play in nodule development? Signaling events during nodule development occur in two distinct phases, the first one in the epidermis in response to nod factor perception, and the second one in cortex cells following the activation of cytokinin signaling (Oldroyd et al., 2011). Auxin hypersensitivity does not appear to affect epidermal responses ( Figure 4A), but clearly affects the second phase of events in the cortex influencing infection thread growth, primordium formation and subsequent development ( Figure   4B & C). Does auxin regulate infection thread growth? We think this might be an indirect effect.
Infection thread growth in the cortex is thought to be determined by cells dividing to form nodule primordia (Murray et al., 2007;Oldroyd and Downie, 2008). Therefore, it is likely that the lack of nodule primordia caused the arrest of infection thread growth in the cortex of miR160ox soybean roots. Consistent with this observation, a moderate reduction in primordium initiation did not affect infection thread growth in the cortex of miR160mx roots. We cannot exclude the possibility that auxin directly influenced rhizobial infection process similar to its effect on infection by bacterial pathogens in arabidopsis (e.g. (Navarro et al., 2006;Cui et al., 2013)).
However, while auxin sensitivity promotes susceptibility to bacterial pathogens, we observed inhibition of rhizobial infection threads suggesting that these mechanisms are not likely to be conserved.
What effect does auxin have on nodule initiation in the cortex? There is very low auxinresponsive gene expression during nodule initiation and it appears specifically in the apex region of nodule primordia ( Figure 1C). We examined nodule initials (cell division foci) in vector control and miR160ox roots using roots expressing the DR5:GUS marker ( Figure S6). The number of nodule initials per root was not significantly different between vector control and miR160ox roots at 3 dpi (2.7 ± 0.9 vs 2.0 ± 0.6 respectively; Poisson distribution analysis P=0.2). We conclude that auxin hypersensitivity did not affect nodule initial cell divisions.
During post-initiation stages of nodule development, the majority of auxin activity gets restricted to the nodule periphery and there is no or minimal auxin activity in the infection zone, at least in determinate nodule-forming legumes ( Figure 1D; (Suzaki et al., 2012)). Interestingly, ENOD40 and DR5 have minimal overlap between their expression domains ( Figure S3F vs Figure 1D). Therefore, it is likely that there was sustained auxin activity in the nodule initials / infection zone of miR160ox and miR160mx roots and this might have inhibited nodule development. We opposite roles in nodule initiation as well? It is known that cytokinin promotes nodule primordium formation and development. For example, the cytokinin-responsive marker, AtARR5:GUS, is specifically induced in cortex cells dividing to form nodule primordia in L.
japonicus (Lohar et al., 2004). Similarly, cytokinin insensitivity caused an inability to initiate nodule primordia in both L. japonicus (a determinate nodule forming legume; (Murray et al., 2007)) and M. truncatula (an indeterminate nodule forming legume; (Plet et al., 2011)). In clear agreement with these observations, gain of function mutations in L. japonicus LHK1 resulted in spontaneous nodule formation even in the absence of rhizobia (Tirichine et al., 2007). We show that miR160ox resulted in hypersensitivity to auxin, hyposensitivity to cytokinin and a reduction in nodule primordium formation. Interestingly, loss of cytokinin sensitivity in L. japonicus hit1-1/lhk1 mutant plants resulted in nodulation phenotypes similar to that of soybean miR160ox roots. While infection threads formed normally, nodule primordium initiation was significantly reduced in these mutants (Murray et al., 2007). Similarly, both miR160ox roots as well as lhk1 mutants had an attenuated expression of NIN and ENOD40 in response to rhizobial inoculation.
Together, these results suggest that reduced nodulation in auxin hypersensitive miR160ox roots is due to the suppression of cytokinin activity. Is a balance between auxin and cytokinin (biosynthesis and/or signaling) crucial for proper nodule primordium development? Indeed, such a hypothesis was proposed recently by Oldroyd & colleagues (Oldroyd et al., 2011). However, such interactions are likely to be cell type-and developmental stage-specific. For example, it was discovered very early that addition of both auxin and cytokinin was necessary to initiate cell divisions opposite xylem poles in isolated pea cortex cell explants (Libbenga et al., 1973).
Vectors were electroporated into Agrobacterium rhizogenes K599 cells and transgenic composite plants for gene expression analyses and miRNA over/mis-expression were generated as previously described (Collier et al., 2005) using two week-old soybean (Glycine max cv Williams82) seedlings. Transgenic roots of interest were identified using GFP epifluorescence. in root length and number of total LRs (0 vs. 7 day measurements) were calculated for each root.
Student's t-test was used to compare different treatments and genotypes using Microsoft Excel.
For DR5:GUS and marker gene expression assays (auxin response), transgenic plants with tagged roots (see above) were transferred to sterile de-water with or without 0.2 or 1.0 μ M 2, 4-D and incubated for 12 h (DR5:GUS) or 6 h (qPCR assays) at 25°C in the dark. For AtARR5:GUS and marker gene expression assays (cytokinin response), transgenic plants were treated with sterile de-water with or without 5.0 μ M BAP for 1 h at 25°C in the dark. Histochemical localization of GUS was performed as described before (Jefferson et al., 1987). For gene expression assays, whole roots were harvested after auxin/cytokinin treatment, blot dried and immediately frozen in liquid N 2 .

B. japonicum inoculation and nodulation assays:
For nodulation assays, transgenic composite plants (3 weeks post transformation with the respective miRox or miRmx construct) were transferred to 4" pots containing a mixture of vermiculite-perlite (1:3), allowed to grow for one week (16h light; 25°C; 50% relative humidity) and inoculated with a suspension of B. japonicum USDA110 (OD 600 =0.08). Two weeks after inoculation, roots were harvested, GFP positive transgenic roots separated and observed under a dissection microscope for nodulation. Nodules appearing as "bumps" were classified as emerging nodules, and those that were round and pink were classified as mature nodules. Mock-inoculated composite plants were used as inoculation control in each experiment and no nodules were observed on these plants. Nodule numbers between different ox and mx roots and the respective controls were examined for statistically significant difference if any using zero-inflated Poisson distribution analysis in the statistical analysis package R. To examine rhizobial colonization and infection thread development, composite plants were inoculated as above with B. japonicum transformed with a nptII:GUS construct (a kind gift from Dr. Gary Stacey, University of Missouri, Columbia, MO). Tissue fixation and GUS staining to visualize rhizobial colonization and infection thread formation were performed as described previously (Loh et al., 2002). Nodule primordia were visualized by clearing roots in 10% bleach for 10 min. Since nodule primordium cell division occurs in the outer cortex in soybean, it is easy to distinguish nodule primordia from LR primordia. For gene expression assays, whole roots were harvested at appropriate time points, rinsed briefly in sterile Microscopy: To examine DR5:tdT expression, mock-or B. japonicum-inoculated transgenic composites plant roots were observed under a laser confocal microscope (Olympus FV300) or a fluorescent compound microscope (Olympus AX70) at 3, 7, 10 and 14 dpi. Images shown in Figure 1A-D were obtained using the confocal microscope with the following settings (Channel 1: 488ex/515em for GFP; Channel 2: 568ex/635em for TdT; 1.5% gain; Kalman acquisition). To examine DR5:GUS expression, mock-and B. japonicum inoculated roots were stained for GUS activity (see above).
Gene expression assays: Total RNAs were isolated from transgenic roots using Trizol reagent and cDNAs prepared from 2 μ g total RNA using oligo-dT and M-MuLV reverse transcriptase (NEB, Ipswich, MA). qPCR assays for gene expression were performed using a Stratagene MX3000P equipment (Stratagene, La Jolla, CA) and SYBR premix (Clontech, Mountain View, CA). Gene expression levels were normalized to that of GmActin using the ddCt method (Livak and Schmittgen, 2001) and further confirmed using two additional house-keeping genes, GmCONS7 and GmCONS15 (data not shown, (Libault et al., 2008)