Together, this evidence suggests the presence of a larger cis-regulatory module that may include up to five other transcription factors besides WUS. Slightly further downstream, another conspicuous feature of the 3’ enhancer is the presence of two large and perfectly conserved sequence blocks, spanning 42 and 32 bp, respectively . Both the size and the degree of sequence conservation in these two regions were exceptional in that they exceeded those found in the coding region of AtCLV3. The two regions also overlapped with the DNA family transposable element At2TE50665, suggesting that these may represent coding sequences of the transposon, rather than AtCLV3 regulators. When examined with the 5bp sliding window, the two conserved blocks were found to be surrounded on both sides by a strikingly periodic arrangement of three cis-motifs, spaced 11bp apart, strongly reminiscent of the pattern associated with the WUS-binding region. Superficially, the region around the two conserved blocks resembled an inverted repeat , though the underlying sequence in each half showed little or no sequence similarity. The repeated occurrence of this pattern suggests that the cis-motifs are organized around a higher-ordered protein structure, each of which may bridge up to 5-7 unique transcription factors. Such clustersof cis-motifs can be described as cis-regulatory modules, which in recognition of their similar structure, are provisionally identified here as CRM 1, 2, and 3 . Of the three modules, evidence from the WUS binding sites and a promoter deletion analysis, suggests that only CRM1 has a direct role in AtCLV3 activation. Although CRM2 and CRM3 might belong to a transposable element,drainage gutter the overall sequence of At2TE50665 is poorly conserved. It is also an old transposon, which likely shared a common ancestor with all five orthologs more than 20mya.
Suspecting functional diversification, the two large sequence blocks were subjected to additional BLAST searches in order to similar motifs in the A. thaliana genome. The genes located next to the motifs were identified, and there expression patterns of a select subset of identified genes were compared using microarray data using the eFP browser. In all, 32 genes were identified, most of which contained only one of the two conserved blocks.Unexpectedly, this analysis revealed that many genes had similar expression patterns in the lateral root cap, the columella, and root procambium tissue . This root expression pattern was also found to be shared between At2G27240 and AtCLV3, though the root expression of CLV3 was much reduced compared to its levels in the SAM. Further alignment of the oligomers returned by the blast searches identified three potential cismotifs, each of which was found to correlate with these expression patterns. Motifs were clearly related to root expression patterns, whereas the third motif was less tissue specific, but associated with several disparate structures in the shoot. However, the predicted transcription factor binding sites found within CRM2 and CRM3 did not strongly support any of these putative functions. The AAATCTAT motif overlapped with a predicted cytokinin response element , and an AGAMOUS binding site. Cytokinin responses occur in both root and shoot tissues, though AGAMOUS expression is clearly confined to the flowers, indicating a shoot-related function. The root-related function of the TGGCGATTTCG motif was at least partially supported by a predicted “right part of the root hair cis-element” motif, but the associated transcription factor is unknown. Finally, the ATTATCCTTAAT motif overlapped with two predicted targets, one for AtHB-2/HAT4, which has root and shoot expression patterns, and the other was a computer identified “sucrose response element” originally identified from lateral buds.
To test the function of the identified regulatory regions, a series of deletions were performed in the pCLV3m:H2B-YFP reporter construct . This construct containes three previously described mutated WUS binding sites, which enhanced expression of the reporter by 120% . Five large deletions, each about 500bp long were initially performed in the regulatory regions, three of which occurred in the upstream 1.5kb region and two in the downstream 1.2kb . Based on initial findings, two of the initial deletions were subdivided into smaller segments, identified here as 3.1, 5.1, and 5.2 . Most deletions produced a binary on-off response in the SAM, though a faint signal remained in the flower meristems of deletions 4 and 5, which occurred in 25% of all independent alleles. Only deletion 5.1 produced an intermediate fluorescent signal, though the pattern indicates reflection off the surface of the SAM, rather than actual fluorescence. Overall the deletions revealed that the AtCLV3 promoter is located in a small region between -154 and +25 bp TSS, while the 3’ enhancer required sequences between +584 and +1389 bp TSS . These regions closely correspond to the previously identified conserved footprints, and contained 8 out of 10 footprints in the 5’ promoter, all four footprints in the 3’UTR, and 14 out of 18 in the 3’enhancer region.Previous deletions of AtCLV3 found the 5’ promoter region was less than 812 bp long, and detected a large 3’ enhancer approximately 950bp long. These findings are broadly consistent with the results of the present study, where deletions 1, 2, 3, and 3.1 further narrowed down the 5’ promoter to just 154bp, and deletions 4, 5, 5.1 and 5.2 support the existence of a regulatory region just 805bp long in the 3’ enhancer. However, the two studies disagree in the functional annotation of these regulatory regions, as the present analysis found only positive regulatory regions , while both positive and negative regulatory functions were found in a previous deletion analysis. Disregarding the 5’ end of deletion V1 which had no significant footprint in the present analysis, a comparison of the two studies revealed that the previously identified negative regulatory region, corresponds to a 334 bp sequence containing CRM2 and CRM3.
Meanwhile, the positive regulatory region corresponded to a 290 bp sequence containing CRM1. The source of the discrepancy is obscure, but might potentially be related to the different techniques used to observe CLV3 expression. The previous deletion analysis relied on a GUS reporter system that used whole plant extracts, which might have missed ectopic expression patterns, while the present study did not attempt to observe any other tissue outside of the SAM. Thus it would be of interest to examine similar deletion constructs in a future study, to see if ectopic expression patterns actually occur following the loss of negative regulatory region. The slight increase in the reporter expression in the mutant control compared to a wild-type pCLV3:H2B-YFP reporter is consistent with the repressive function of WUS transcription factors, but suggests that WUS alone is insufficient to repress CLV3 expression. The failure to detect strong reporter activity in deletions 5 and 5.1 might also indicate an interaction between their regulatory modules, as the presence of CRM1 alone could not activate the reporter in the absence of CRM2 and CRM3. The reverse is also true, as CRM2/3 were not able to activate the reporter in the absence of CRM1 . It is not clear how the modules might interact with each other, as CRM1 is separated from CRM2/3 by 280-335bp, indicating that they are located on non-adjacent nucleosomes. There are however, four conserved regions located between CRM1 and CRM2 , which if recognized by additional DNA binding proteins, might help bridge the nucleosome gap. The AtCLV3 expression pattern occurs in an inverted cone-shaped domain in the apex of the SAM,large square pots and displays layer-specific patterns. The L1 is often strongest, while the signal intensity fades with tissue depth. In tangential sections, L2 cells often have noticeably weaker expression levels . However, the L2 “gap” often disappears in perfect longitudinal sections, suggesting that AtCLV3 is actually expressed in two closely spaced domains: a broad L1 sheet in the CZ, and a smaller, but roughly spherical domain directly underneath. Ideally, it would be possible to predict these two patterns using the conserved cis-motifs identified in this study, but unfortunately the function of individual regulatory regions often cannot be completely determined with the available data. Deletion 3.1 for example , suggests that Motif #2 and the redundantly predicted MYB site have no apparent function in the 5’ promoter, yet this contrasts with their unique and strongly conserved footprints. There is also predicted cytokinin-response element located at -102bp, but it is poorly conserved among the four orthologous sequences. A predicted AGL15 binding site might produce the L1 pattern by partially suppressing L2-L4 expression, as AGL15 is known to interact with transcriptional repressors such as TOPLESS and SAP18. However, an examination microarray data with the eFP browser suggests that AGL15 is only transcribed in a small subset of root tissues, where it is unlikely to affect the SAM.
However, considering that the AGAMOUS-LIKE gene family contains more than 100 members , it is possible that one or more of them might function redundantly to suppress AtCLV3 in a subset of SAM cells. If these four cis-elements are removed from consideration, the remaining portion of the AtCLV3 5’ promoter is surprisingly small, and is potentially less than 66bp long. Within this small region are three predicted cis-motifs, in addition to the previously noted initiator for a TATA-less promoter, and several conserved regions around the transcriptional start site. Two of the cismotifs, GT-1 and AGAMOUS, partially overlap with each other and are clear transcriptional activators. The presence of the AG site might also explain the weak expression pattern found in the flower meristems of deletions 4 and 5. The role of GT-1 is harder to explain though, as it is homogenously expressed in most plant tissues, and presumably would lead to widespread ectopic expression of AtCLV3. However, the absence of such ectopic expression patterns might be explained in terms of a nearby auxin response element, which is recognized by the AUXIN RESPONSE FACTOR1. Based on the pDR5rev:3xVENUS-N7 reporter, no auxin responses occured in the SAM itself, but they can be readily detected in the apices of lateral anlagen. Although it is not immediately clear how this might relate to AtCLV3 expression patterns in the CZ, a review of the ARF gene family finds that it includes 5 activators and 17 repressors. The cone-shaped expression domain of AtCLV3 is thus most consistent with repressive auxin responses in the peripheral zone, and might even suggest that the “cone” shape is actually pyramidal based on the auxin response foci observed with the pDR5rev:3xVENUS-N7 reporter. This model is also consistent with the peripheral expression of ARF3 and ARF9, both of which have been demonstrated to be transcriptional repressors. However, this interpretation is at odds with the transcriptional activator ARF5/MONOPTEROS, which is also largely expressed in the peripheral zone, and trace amounts extend into the CZ. As AtCLV3 itself is known to be up regulated by auxin responses at least within the narrow confines of the CZ, the potential functional significance of the GT-1 site might be questioned by an alternative regulatory hypothesis. It is equally probable for example, that AtCLV3 is activated in the CZ primarily by ARF5 or other ARF paralogs, and repressed though an unrelated molecule produced in the peripheral zone. So long as this occurs at levels below the detection threshold of the pDR5rev:3xVENUS-N7 reporter, this model would be indistinguishable from the GT-1 activation/peripheral auxin repression model. Currently, the only evidence that might be able to discriminate between these two hypothesis is rather indirect, and relies on the enhancement of cytokinin responses through the alcohol inducible RNAi system to silence ARR7/15 expression . Interestingly, the CLV3 expression level was reduced in this system, which is consistent with auxin-based activation. In contrast with the promoter, the 3’ enhancer region of AtCLV3 is quite large, with conserved regions spanning a minimum of 460 bp. This region quite likely contains three cis-regulatory modules, each of which may contain 5-7 unique cis-motifs, and together they might support upwards of 20 different transcription factors from multiple gene families. Clearly, regulation of AtCLV3 from the 3’ enhancer is likely to be complicated. One hint about how this might occur lies in the regular spacing of cis-motifs 11-15 bp apart, which is suggestive of helical phasing. The motifs themselves show little or no sequence similarity between modules, implying that transcription factors themselves are interchangeable, while their spacing pattern is governed by a higher-order protein complex that bridges all 5-7 cis-motifs simultaneously.