Dissection of TALE-dependent gene activation reveals that they induce transcription cooperatively and in both orientations

Abstract

Plant-pathogenic Xanthomonas bacteria inject transcription activator-like effector proteins (TALEs) into host cells to specifically induce transcription of plant genes and enhance susceptibility. Although the DNA-binding mode is well-understood it is still ambiguous how TALEs initiate transcription and whether additional promoter elements are needed to support this. To systematically dissect prerequisites for transcriptional initiation the activity of one TALE was compared on different synthetic Bs4 promoter fragments. In addition, a large collection of artificial TALEs spanning the OsSWEET14 promoter was compared. We show that the presence of a TALE alone is not sufficient to initiate transcription suggesting the requirement of additional supporting promoter elements. At the OsSWEET14 promoter TALEs can initiate transcription from various positions, in a synergistic manner of multiple TALEs binding in parallel to the promoter, and even by binding in reverse orientation. TALEs are known to shift the transcriptional start site, but our data show that this shift depends on the individual position of a TALE within a promoter context. Our results implicate that TALEs function like classical enhancer-binding proteins and initiate transcription in both orientations which has consequences for in planta target gene prediction and design of artificial activators.

Fig 1. TALE mediated gene activation is dependent on surrounding DNA sequences.

(A) Model for the DNA-binding mode of TALEs by using the example of Hax3 aligned to the Hax3-box. TALEs contain a central repeat region (red), two nuclear localization signals (NLS) and an acidic activation domain in the C-terminal part. The amino acid sequence of a Hax3 repeat is shown in single letter code. The repeat variable diresidue (RVD) is shaded in grey. Each RVD specifies one nucleotide in the DNA-target sequence. (B) Sequence overview of the analysed 75 bp long DNA fragments that originate either from the Bs4 promoter (pBs4, region from -278 bp to +25 bp) or from the Bs4 open reading frame (oBs4). Potential promoter elements that were predicted [37] are labelled in grey (TATA-box) blue (W-box), black (CAAT-box) or italic and underlined (5' untranslated region, UTR). (C) TALE-dependent activation of reporter constructs. Each promoter fragment was placed downstream of the Hax3-box and inserted in front of a promoterless uidA reporter gene. Agrobacterium strains delivering hax3 under control of the 35S promoter were co-inoculated into N. benthamiana leaves along with an Agrobacterium strain delivering the respective reporter construct. As negative control, all reporter constructs were inoculated with an empty vector construct (-). The quantitative β-glucuronidase measurement was performed 2 dpi, error bars were calculated on the basis of three independent replicates and represent the standard deviation. (4-MU, 4-methyl-umbelliferone).

Fig 2. Bs4 promoter swaps to identify regions that support TALE activity.

(A) The two 75 bp fragments (A, positive, green and B, negative, red) were subdivided into smaller fragments (A1-4, B1-4). Putative promoter elements that were predicted [37] are labelled in grey (TATA-box), blue (W-box), black (CAAT-box) or italic and overlined (5´UTR). (B) Overview of the analysed promoter swaps. The origin of the fused fragments is marked with green and A or red and B, respectively. The fragments were placed downstream of the Hax3-box (H3-B) and upstream of a promoterless uidA reporter gene. The dashed line indicates the location of the attB1 site preceding the uidA reporter gene. TALE-dependent activation of the reporter constructs was determined by β-glucuronidase-measurement. Agrobacterium strains delivering hax3 under control of the 35S promoter were co-inoculated into N. benthamiana leaves along with an Agrobacterium strain delivering the respective reporter construct. The ß-glucuronidase measurement was performed two dpi and calculated as relative activity based on the activity of the reference fragment A (100%). Error bars represent the standard deviation of three independent replicates (4-MU, 4-methyl-umbelliferone).

Fig 3. TALEs enhance transcription from diverse positions in the rice OsSWEET14 promoter in N. benthamiana in a partially TATA-box dependent manner.

(A) Overview of the reporter construct containing the OsSWEET14 wildtype (WT) promoter fragment (1kb upstream of the ATG) and the OsSWEET14-11 promoter mutant (Δ 33 bp; 967 bp upstream of the ATG; [41]) fused to the promoterless uidA reporter gene. The dashed line indicates the location of the attB1 site preceding the uidA reporter gene. The binding sites and binding orientations of artificial TALEs (coloured) and the natural TALEs TalC and AvrXa7 (white) are marked with arrows, the arrowhead indicates the orientation of the activation domain relative to the uidA gene. Reverse binding TALEs are labeled with "-" in front of the number. RVD sequences and target sites are listed in S1 Table. (B)-(E) Activity of the artificial and natural TALEs. Agrobacterium strains delivering TALE constructs under control of the 35S promoter were co-inoculated into N. benthamiana leaves along with an Agrobacterium strain delivering the reporter construct. The TALE Hax3 that does not bind to the OsSWEET14 promoter was used as negative control to exclude background promoter activity. The ß-glucuronidase measurement was performed 2 dpi. Error bars represent the standard deviation of three independent replicates (4-MU, 4-methyl-umbelliferone). (B) Activity of forward TALEs in combination with the WT reporter. (C) Activity of reverse TALEs in combination with the WT reporter. (D) Activity of forward TALEs in combination with the mutant reporter. (E) Activity of reverse TALEs in combination with the mutant reporter. (D, E) TALEs whose binding sites overlap with the deletion in the mutant promoter are indicated with Δ33 bp. Arrows highlight a strong reduction of TALE activity in comparison to the WT reporter.

Fig 4. TALEs can synergize for gene induction.

Agrobacterium strains delivering TALE constructs and the OsSWEET14::uidA reporter, respectively were mixed at a ratio of 3:1 to a final OD of 1.6. If less than three different TALEs targeting the OsSWEET14 promoter were used, the remainder was substituted with a Hax3-delivering strain not targeting this reporter to keep the total ratio of TALEs constant. The highly active TAL8 was used as positive control. The bacterial mixture was inoculated into N. benthamiana leaves and GUS measurements were taken 2 dpi, error bars represent the standard deviation of three independent replicates (4-MU, 4-methyl-umbelliferone). Please note that the GUS values of individual low-active TALEs was consistently very low here in comparison to Fig 3, probably due to the higher dilution factor of the reporter construct in this experimental design.

Fig 5. The use of an alternative Activation Domain (AD) or DNA-binding domain (dCas9 activator) does not change the activation pattern of TALEs at the OsSWEET14 promoter in N. benthamiana.

(A) Overview of the binding sites and binding orientation of a selected number of artificial TALEs that were fused to the VP64 AD and the binding sites and binding orientation of the analyzed sgRNAs. The dashed line indicates the location of the attB1 site preceding the uidA reporter gene (B) Activity of artificial TALEs (filled bars) in comparison to the TALE-VP64 derivatives (framed bars). Agrobacterium strains delivering the TALE constructs under control of the 35S promoter were co-inoculated into N. benthamiana leaves along with an Agrobacterium strain delivering the reporter construct. TAL1ΔAD was used as an internal control. The ß-glucuronidase measurement was performed 2 dpi. Error bars represent the standard deviation of three independent replicates (4-MU, 4-methyl-umbelliferone). (C) The nucleolytically inactive dead Cas9 (dCas9) variant was fused to the C-terminus of Hax3 to generate a dCas9 activator. Agrobacterium strains delivering dCas9 activator constructs and the reporter construct that contains the 1 kb promoter fragment of OsSWEET14 fused to a promoterless uidA gene were co-inoculated into N. benthamiana leaves. GUS measurement was performed 2 dpi, error bars represent the standard deviation of three independent replicates (4-MU, 4-methyl-umbelliferone). Please notice the different scales on the TALE and dCas9 activator graphs indicating that TALEs are more potent activators in this example.

Fig 6. Artificial TALEs do not always shift the natural Transcriptional Start Site (TSS) at the OsSWEET14 promoter in N. benthamiana.

Overview of analyzed artificial and natural TALEs that bind to different positions in the OsSWEET14 promoter. The natural OsSWEET14 TSS in rice is marked with a red "A", and the AvrXa7-induced TSS is marked with a grey "G" [20]. Agrobacterium strains delivering the 35S-controlled TALE constructs were co-inoculated into N. benthamiana leaves along with an Agrobacterium strain delivering the reporter construct. 1 dpi leaf discs were harvested and total RNA was extracted. cDNA was produced and used for 5' RACE. The first nucleotide of each identified TSS is labeled with a capital letter. The distance of the TSS to the 3´end of forward-orientated TALE target sequences and to the 5´end of reverse-orientated TALE target sequences as well as the number of analyzed clones ("x") is indicated to the right. TSSs that overlap with the natural OsSWEET14 TSS in rice are marked in red.

Fig 7. Reverse binding artificial TALEs can activate OsSWEET14 expression in a natural Xoo-rice infection.

(A) Overview of the binding sites and binding orientation of analyzed artificial TALEs (coloured arrows) and of the natural TALE TalC (white arrow). (B) Reverse- and forward-oriented TALEs can both support disease development of Xoo in rice. Leaves of Oryza sativa cv. Nipponbare were inoculated with the Xoo strains BAI3, BAI3ΔtalC or BAI3ΔtalC carrying an empty vector plasmid (ev), talC or an artificial TALE on a plasmid. Inoculation of rice leaves with MgCl₂ served as negative control. Pictures of phenotypes were taken 5 dpi. Water soaking symptoms are marked with a white arrow. (C) OsSWEET14 expression levels following Xoo infection. Leaves of Oryza sativa cv. Nipponbare were inoculated with the same Xoo strains as in (B). 1 dpi leaf material was harvested and the transcript levels of OsSWEET14 were determined by quantitative reverse transcription polymerase chain reaction (qRT-PCR). The error bars indicate the standard deviation of three biological replicates. The fold-change of OsSWEET14 expression was calculated based on the negative control BAI3ΔtalC + ev. An asterisk (*) indicates a significant increase in OsSWEET14 expression calculated with the students t-Test.