Strategy for Insulator Selection
Insulator isolation was initially based on a 3-step process in which vectors were designed for the selection of candidate insulator (In) sequences, the testing of re-isolated sequences in a second vector to confirm insulator activity and final cloning into a third vector to determine that the sequence was not a strong repressor and that expression from a tissue-specific seed coat promoter was observed.
pC1 (pCAM1300-35S46-codA) is a negative selection vector (Fig. 1A) containing cytosine deaminase gene (codA) (Perera et al., 1993) expressed from the core CaMV35S promoter (35S46) (Covey et al., 1981). Expression is boosted by CaMV35S enhancer (“35S double enhancer”) responsible for driving expression of hygromycin phosphotransferase II (hptII) used for plant selection. Cytosine deaminase converts the non-toxic, cytosine analogue 5-Fluorocytosine (5-FC) into a highly toxic metabolite 5-Fluorouracil (5-FU) (1989; Perera et al., 1993; Stougaard, 1993) that in planta irreversibly inhibits thymidylate synthase thus depriving cells of dTTP and leading to cell death (Mullen et al., 1992) The pathway is illustrated in Fig, S1. The presence of a DNA fragment with a functional insulator between the CaMV35S enhancer and 35S46 will block enhancer influence on codA and plants will survive. Loss of codA expression for one of many possible reasons such as cloning or an insulator or a strong repressor or mutation within codA that could allow plant survival. Hence the need to further testing.
pB31 (pCAMBIA1300-35S46-GUS) is a screening vector with in which codA from pC1 is replaced by the β-glucuronidase (GUS) reporter gene (Fig. 1B). As in pC1, the presence of an insulator between the CaMV35S enhancer and the 35S46 core promoter will block the expression of the transgene resulting in no blue stain in any of the tissues tested -flowers, leaves, siliques- while the absence of a functional insulator will result in constitutive GUS throughout the plant. This does not distinguish between an insulator and a strong repressor.
The final vector, pL1 (pCAMBIA 1391 napin-GUS) uses a similar strategy to pB31 except that GUS is now expressed from the napin seed-specific promoter (Ericson et al., 1991) in lieu of the 35S46 core promoter (Fig. 1C). In pL1 the napin promoter under the influence of the CaMV35S enhancer drives non-specific GUS expression, showing blue staining in all tissues tested-flowers, leaves, siliques. Cloning of a strong repressor would block all gene expression, resulting in no staining in transgenic tissues (Gudynaite-Savitch et al. 2007). An insulator cloned between the two elements blocks non-specific expression, as a result GUS expression is found only in seeds and at the same time verifies that the insulator functions with two different promoters.
Key: hptII: hygromycin phosphotransferase; codA: cytosine deaminase; GUS: marker gene β-glucuronidase; 35S: CaMV 35S promoter S46: CaMV 35S core promoter Napin: Brassica napus seed-specific promoter
Selection of insulator sequences
124 bp random DNA oligonucleotide library plus flanking sequences was cloned into pC1 for transformation into A. thaliana. Plantlets containing potential insulator sequences were selected on agar plants containing 30µg/ml hygromycin-B and 250µg/ml timentin supplemented with 2000 µg /ml 5-Fluorocytosine. Initial screening identified 100 plantlets that survived selection on agar plates; subsequent transfer to soil resulted in 60 candidate plants with putative insulators. Seed was collected from these plants for further propagation and DNA was isolated from leaf tissue for PCR and sequence analyses. They were classified into four groups (I, II, III, and IV) based upon the selection trial.
Twenty-one were chosen based on plant growth and health on selective media and in soil for subsequent cloning of the insert (potential insulator) into the vector pB31. Upon further analysis, nine had the phenotype expected for a fragment without insulator activity (Table 1A), eight had the phenotype expected for insulator activity (Table 1B) and four were in the category “too few transgenic plants to infer the category”.
These results suggest that the 5-FC can be used to select insulator sequences from a library of random sequences. However, a strong repressor activity could be expected to mimic an insulator in this assay. To differentiate between these two possibilities inserts from clones in Table 1B were cloned into the vector pL1 in which an insulator positioned between the 35S double-enhancers and the napin promoter will block the action of the enhancers but not the tissue-specific napin promoter expressing GUS (Gudynaite-Savitch et al. 2009). Clone InIII-80 may also be a good candidate for an insulator but at this stage too few transgenic plants were analysed (Table 1B). As indicated in Table 1C, 5 candidates, InI-3, InII-10, InII-12, InIII-50, and InIII-78Δ5’ (a spontaneous deletion of InIII-78) demonstrate the expected insulator GUS phenotype. Based upon this small survey, we conclude that insulator activities can be isolated de novo from a random library. Actually, the source of the DNA should be immaterial.
Defining Insulator Sequences
The sizes of the inserts presented in Table 1, and in the other clones not yet completely analysed, are not just the expected 154 bp for the insert plus primer binding sites but range from ~ 150 bp up to 446 bp. Sequencing revealed that the ~ 450 bp inserts were ligation products of three different ~ 154 bp sequences to give a trimeric insert (Gandorah, 2012). The minor differences between them could result from mutations/deletions generated during PCR and prior to ligation or by deletions due to recombination during plant transformation.
For further bioinformatics approaches the longer sequences were subdivided into 3 units (A,B,C) in silico based upon the positions of the internal BamHI and EcoRI sites. Further editing removed the primer binding sites used for PCR and the terminal restriction sites used for cloning from each unit. Treated in this way resulted in a total of 11 sequences were recovered for the insulators InI-3, InII-10, InII-12, InIII-50, and InIII-78Δ5’. None were identical, suggesting we had identified at least 5 unique insulator sequences.
Several insulators from plant or other sources have been defined (Pérez-González and Caro 2019; She et al. 2010; Tran and Johnson 2020). Our initially focus was on the group UASrpg, BEAD1c, gypsy to determine whether our experimentally defined insulators contained sequences that were similar to them. Both orientations were investigated. Finally, a set of insulators with known sequences (scs, CoreScs, GAGA, MCP, HS4, HS185) was tested. No 100% matches were observed but near identical matches were identified (Fig. S3); InI-3 with, MCPRC (7/9) and CoreScsRC (8/10); InII-10 with UASrpg1RC (8/10); InII-12 with UASrpg1 (7/10) and UASrpg1RC (7/10); InIII-50 with scsRC (7/10). No matches were observed for InIII-78_5’.
1A. GUS staining results for candidate sequences cloned into pB31 with no apparent insulator activity. The expected phenotype of insulator activity is no GUS expression in all tissues.
| | | GUS positive staining | | | GUS |
expression |
| | No. plants tested | | | | | |
Construct | Size* | Flowers | Leaves | Siliques | Stained | Not stained | |
Name | (bp) | | | | | | |
pB31 | | 62 | 11 | 39 | 38 | 74.2% | 25.8% | |
InII-3 | 126 | 17 | 9 | 6 | 9 | 76.5% | 23.5% | |
InII-7 | 128 | 30 | 11 | 14 | 8 | 63.3% | 36.7% | |
InIII-4 | 414 | 26 | 16 | 4 | 3 | 65.4% | 34.6% | |
InIII-17 | 417 | 30 | 25 | 15 | 2 | 86.7% | 13.3% | |
InIII-22 | 399 | 36 | 26 | 11 | 3 | 80.6% | 19.4% | |
InIII-27 | 102 | 13 | 13 | 4 | 2 | 100.0% | 0.0% | |
InIII-53 | 126 | 15 | 13 | 14 | 14 | 93.3% | 6.6% | |
InIII-55 | 406 | 23 | 22 | 5 | 2 | 95.7% | 4.3% | |
InIII-57 | 128 | 16 | 2 | 8 | 9 | 68.8% | 31.2% | |
3B. GUS staining results for candidate sequences cloned into pB31 with apparent insulator activity. The phenotype of insulator activity is no GUS expression in all tissues. |
| | | GUS positive staining | | GUS | |
| expression | |
| | No. plants tested | | | | | | |
Construct | Size* | Flowers | Leaves | Siliques | Stained | Not stained | |
Name | (bp) | | | | | | |
InI-3 | 412 | 51 | 0 | 0 | 0 | 0.0% | 100.0% | |
InI-6 | 436 | 24 | 0 | 0 | 0 | 0.0% | 100.0% | |
InII-10 | 128 | 40 | 0 | 0 | 0 | 0.0% | 100.0% | |
InII-12 | 401 | 37 | 0 | 0 | 3 | 8.1% | 91.9% | |
InIII-50 | 417 | 11 | 0 | 0 | 0 | 0.0% | 100.0% | |
InIII-52 | 420 | 15 | 3 | 1 | 0 | 20.0% | 80.0% | |
InIII-78 | 407 | 20 | 0 | 0 | 0 | 0.0% | 100.0% | |
InIII-80 | 425 | 9 | 1 | 0 | 0 | 11.1% | 88.9% | |
1C. GUS staining results for candidate sequences re-cloned into pL1 with apparent insulator activity. The phenotype of insulator activity is GUS expression only in seeds.
Table 1
The β-glucoronidase (GUS) staining results for potential Arabidopsis thaliana insulator sequences. The columns indicate the sequence number, construct name, size, number of total transformations, the number of individual transformants tested for flower, leaf and silique staining, and a calculation of the number of stained and unstained samples. The data sets for plant staining and potential insulator sequences can be found in Supplementary Information Table S2.1 (Staining of pB31 transgenic plants), Table S3.2 (Staining of pL1 transgenic plants) and Fig. S2 (Verification of insulator sequences in transgenic pL1)
| | | GUS positive staining | | | Seed-specific | |
GUS expression |
| | No. plants tested | | | | | | | |
Construct | Size* | Flowers | Leaves | Siliques | Seeds | Specific | Nonspecific | None |
Name | (bp) | | | | | | | |
pL1 | | 53 | 41 | 49 | 42 | 41 | 3.9% | 96.1% | 2 |
InI-3 | 412 | 20 | 0 | 0 | 2 | 16 | 89.8% | 11.1% | 2 |
InII-10 | 128 | 27 | 0 | 1 | 0 | 22 | 95.7% | 4.4% | 4 |
InII-12 | 401 | 14 | 0 | 0 | 0 | 10 | 100.0% | 0.0% | 4 |
InIII-50 | 417 | 13 | 0 | 0 | 0 | 12 | 100.0% | 0.0% | 0 |
InIII-78Δ5’ | 116 | 16 | 0 | 0 | 1 | 14 | 93.3% | 6.7% | 1 |
* The size of the insert cloned into either pB31 or pL1 after removal of the primer sequences.