(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Identification of exceptionally potent adenosine deaminases RNA editors from high body temperature organisms [1]

['Adi Avram-Shperling', 'The Nano Center', 'The Mina', 'Everard Goodman Faculty Of Life Sciences', 'Bar-Ilan University', 'Ramat Gan', 'The Institute Of Nanotechnology', 'Advanced Materials', 'Eli Kopel', 'Itamar Twersky']

Date: 2023-03

(A, B) (i) Logarithmic cells carrying hADAR2, sqADAR2, owADAR2, hbADAR2, and mdADAR1, were grown in SC-URA+Raf. Cells were then normalized to an optical density at a wavelength of 600nm (O.D 600nm ) of 0.1 and transferred to SC-URA+Gal (A) and SC-URA+Glu (B) (ADAR expression on and off respectively). Growth was assessed in 96 wells plate using a TECAN microplate reader, by measuring the O.D 600nm in 30 minutes intervals for 22 hours. Cells harboring an empty URA3 marked plasmid were used as a control. Error bars show the standard deviation between three independent experiments. (ii) The growth in A and B was quantitated by calculating the area under the curve (AUC). Whiskers represent the minimum and maximum values measured from three independent experiments. Stars above the boxplots indicate a statistically significant difference between the means of two samples (ns: p > 0.05; *: p < = 0.05; **: p < = 0.01; ***: p < = 0.001; ****: p < = 0.0001).

To further confirm and quantitate the growth defect of these strains, we measured the optical density of liquid cultures over 22 hours (in 30-minutes intervals) for these three affected strains, as well as two reference strains carrying the squid ADAR2 (sqADAR2) and Human ADAR2 (hADAR2) which have shown a weaker effect ( Fig 2A ). The growth curves confirmed that the inducible expression of all five ADAR enzymes resulted in differential growth impairment, with a stronger effect for owADAR2, hbADAR2, and the most discernible effect in cells overexpressing mdADAR1. No effect was detected when the same strains were grown in a glucose-containing media, where no ADAR protein expression is detected ( Fig 2B ).

Exogenous expression of heterologous ADARs in yeast differently affects their growth rate (A) Schematic representation of the yeast-based expression system. The coding sequence of heterologous ADAR enzymes N-terminally fused to a FLAG-tag were cloned into a URA3 marked plasmid under a galactose inducible promoter (GAL1p). ADARs were originated from Mallard duck (Anas platyrhynchos; mdADAR1/2); Hummingbird (Calypte anna; hbADAR1/2); Orca whale (Orcinus orca; owADAR1/2); Squid (Loligo opalescens; sqADAR1/2); and Human (Homo sapiens; hADAR1/2). (B) Multiple alignment of the heterologous ADARs used in this study. The ADAR enzymes contain a catalytic deaminase domain (DD, red), two or three double stranded RNA binding domains (dsRBD, blue), and, in ADAR1 only, one or two Z-DNA binding domains (Z-DBD, green). Domains were identified using PROSITE database. All schemes are to scale. (C) Western blot analysis confirms the expression of the ADAR proteins in yeast. Logarithmically growing cells harboring the plasmids described in A were grown in minimal media (SC-URA) supplemented with raffinose (Raf) (SC-URA+Raf) (ADAR expression is off). 2% Galactose (Gal) was then added for 6 hours, to induce the ADAR genes expression. Cell lysates from the SC-URA+Gal media were separated by SDS-PAGE and immunoblotted with anti-FLAG (α-FLAG) antibody. Cells harboring an empty URA3 marked plasmid were used as a negative control (URA3). Ponceau staining was used as a loading control. (D) The inducible expression of specific heterologous ADAR enzymes impairs cells growth. Left: 10-fold serial dilutions of the indicated strains were spotted on Glucose (top, Glu, ADAR expression is off) and Galactose containing media (bottom, Gal, ADAR expression is on). Plates were incubated at 30°C for 40h. Cells carrying an empty URA3 plasmid were used as a control. Right: Quantitation of growth impairment. Growth was assessed using a digital image of a drop (second dilution) to generate an estimate of the growth on SD-Gal relative to SD-Glu based on pixel density. Whiskers represent the minimum and maximum values measured from three independent experiments. Stars above the boxplots indicate a statistically significant difference between the means of two samples (ns: p > 0.05; *: p < = 0.05; **: p < = 0.01; ***: p < = 0.001; ****: p < = 0.0001).

We cloned each of these ten heterologous enzymes into a URA3 marked plasmid that enables their expression under the galactose inducible promoter (GAL1p) ( Fig 1A and 1B ). Next, we immunoblotted protein lysates samples from yeast cells carrying these plasmids and were able to validate the expression of most enzymes, except hbADAR1, owADAR1 and sqADAR1, which were most probably eliminated by the proteasome ubiquitin system or autophagy ( Fig 1C ). These strains were then serially diluted and spotted on galactose- and glucose-containing media (ADAR expression is on and off, respectively) and incubated at 30°C. As shown in Fig 1D , the strongest growth impairment effect was seen upon inducible expression of Orca whale ADAR2 (owADAR2), Hummingbird ADAR2 (hbADAR2), and Mallard duck ADAR1(mdADAR1).

Endothermic vertebrates maintain a fixed core body temperature. This body temperature is between 35 and 42°C, which is 10–15°C higher than the typical body temperature of ectothermic vertebrates [ 35 ]. We have therefore looked at ADAR1 and ADAR2 orthologs from five species–two mammals and two birds, and one ectotherm invertebrate, that inhabit different environmental niches: (1) Human (Homo sapiens), whose ADARs are commonly used in base editing research [ 32 , 36 , 37 ]; (2) Squid (Loligo opalescens), an ectotherm invertebrate living in diverse temperatures, which was shown to have extraordinarily high levels of A-to-I editing in coding sequences [ 38 – 40 ]; (3) The marine mammal, Orca whale (Orcinus orca), with core body temperature of ~38°C [ 41 ]; (4, 5) Hummingbird (Calypte anna) and Mallard duck (Anas platyrhynchos), two birds with relatively high core body temperatures of ~40°C and ~42°C respectively, the warm end of the endothermic spectrum [ 42 , 43 ].

ADARs contain dsRNA Binding Domains (RBDs) that recognize RNA secondary structures surrounding the target adenosine. The stability of these structures is temperature-dependent [ 33 , 34 ], and weaker (less paired) structures are generally expected at higher temperatures. We have therefore speculated that species evolved to live with higher core body temperatures may have developed ADAR enzymes that target weaker dsRNA structures.

Yeast cells do not encode the ADAR proteins, and thus are not capable of A-to-I RNA editing and are not adapted to it. Accordingly, imposing editing on these cells by introduction of exogenous ADAR enzymes results in impairment of their growth rate [ 32 ]. We thus proposed to screen different ADAR enzymes using growth impairment as a probe for editing activity.

The impaired growth rate of yeast cells expressing heterologous ADARs is the result of editing within yeast transcripts

To verify that the exogenously induced ADAR enzymes are catalytically active in the yeast cell, we sequenced RNA samples from the strains described in Fig 2 and looked for evidence of A-to-I editing. In addition, we looked at three negative controls, cells harboring an empty URA3-marked plasmid, and cells expressing catalytically inactive mdADAR1 and hADAR2, (mdADAR1-E619A and hADAR2- E396A) in which a critical glutamic acid was converted to alanine. A similar mutation was shown to inhibit catalytic activity in hADAR1 [44].

Inosines are read by the reverse-transcriptase and the following sequencing protocol as guanosines. Thus, editing detection tools look for A-to-G mismatches between the reference genome and the RNA-seq reads. As a control, we looked at the number of mismatches of different types. Since our RNA-sequencing libraries were not strand-specific, there are only six possible types of mismatches (e.g., T-to-C mismatches are indistinguishable from A-to-G ones). First, we used the RES-scanner pipeline [45] followed by multiple testing correction to identify mismatch sites de-novo (see Methods). We found that the number of non-A-to-G substitutions were similar for all samples, while the number of A-to-G mismatches (presumably induced by A-to-I editing) increased substantially upon ADAR induction. For samples containing active ADARs 74–99% of the identified sites were A-to-G mismatches, and the number of these mismatches was at least 25-fold higher than that observed in control samples (Fig 3A). The number of editing sites per sample varied from 597 for hADAR2 (~25-fold increase from the baseline A-to-G mismatch level shown in the control samples) to 113,672 for mdADAR1 (~5000-fold increase from the baseline), which has the highest body temperature of the species screened. The detected A-to-G sites exhibit the familiar ADAR motif [46, 47] (Fig 3B). We thus attributed these mismatches to A-to-I RNA-editing events. The majority of these editing sites resides within protein coding sequences, and the distribution of editing levels is skewed towards lower values, as expected (see S1 Fig).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Editing activity in ADAR-expressing yeast strains is strongest for mallard duck ADAR1. (A) Res-scanner detection of editing sites (Methods) for five yeast strains expressing different active ADAR enzymes results in varying amounts of sites, spanning two to three orders of magnitude. Note that due to the large numbers for mdADAR1 (113,672 sites, in a 12Mbp genome) and hbADAR2 (10,443), the results for the other strains are invisible in the main graph, and thus these results are plotted again in the inset. The (three rightmost) control strains show no excess of A-to-G mismatches, attesting for no editing activity as expected. (B) Distribution of the neighboring nucleotides to the predicted editing sites reveals the familiar editing motif (mostly a depletion of G upstream) for all strains expressing active ADARs, but not for the control ones. (C) Thermodynamic stability of the predicted secondary structures surrounding the detected sites (Methods). Lower (more negative) ΔG indicates a more stable structure. Note that no structure was found for 113, 379, 939, 2526 and 32197 sites, for the five organisms, respectively. Their ΔG was set to zero. (D) Genome-wide editing index (Methods) for the five active ADAR strains and three control ones. The dashed line represents the base-line noise level. (E-F) ADAR expression levels (TPM) (E) and protein levels (F) for the different strains do not account for the wide differences in the number of sites detected. Logarithmically growing cells carrying the indicated plasmids were grown in SC-URA+Raf (ADAR expression is off). 2% Gal was then added for 4 hours, to induce the ADAR genes expression. Cell lysates from the SC-URA+Raf and SC-URA+Gal (t-0 and t-4 respectively) were separated by SDS-PAGE and immunoblotted with anti-FLAG (α-FLAG. Anti-3-phosphoglycerate kinase 1 (α-PGK1) was used as a loading control. Relative intensity was measured as pixel density for each sample with consideration of the loading control. Protein levels were quantitated relative to the loading control in three independent experiments. Whiskers represent the minimum and maximum values measured from three independent experiments. Stars above the boxplots indicate a statistically significant difference between the means of two samples (ns: p > 0.05; *: p < = 0.05; **: p < = 0.01; ***: p < = 0.001; ****: p < = 0.0001). https://doi.org/10.1371/journal.pgen.1010661.g003

We wanted to further test our hypothesis that the increased number of editing sites using ADARs originating from species with higher core body temperatures is due to targeting of weaker dsRNA structures by these enzymes. We thus looked at the thermodynamic stability of the secondary structures surrounding the editing sites detected per strain (Methods). As expected, enzymes editing more sites have also targeted sites with a lower stability (Fig 3C). In particular, the structures edited by mdADAR1 are significantly less stable that those edited by hADAR2, in agreement with our proposed model.

In addition, in order to obtain a quantitative comparison between the different enzymes we used the editing index approach (see Methods), which is less sensitive but more robust [48]. Briefly, the editing index represents the fraction of nucleotides expressed from genomically-encoded adenosines that has been edited. Here, we have calculated the index across the full yeast genome. In the URA3 empty plasmid sample and the two inactive ADAR samples, the measured editing index was 0.023–0.025%, representing the baseline noise level (mainly due to genomic polymorphism sites). The samples expressing active ADARs exhibited elevated index values, as expected, and the increase correlated to the number of de-novo sites reported above (Fig 3D). We also quantified the RNA and protein expression levels of the ADAR enzyme in each sample (Fig 3E and 3F). While there are variations between the samples, the high editing seen for mdADAR1 is not explained by higher RNA or protein levels, and likely represents the intrinsic potency of this enzyme. Notably, for all ADAR enzymes tested, the majority of detected sites are also detected with mdADAR1 (S2 Fig).

To further validate these results, we sequenced again strains including empty vectors, inactive mdADAR1-E619A and active mdADAR1, five replicates each. The number of A-G/T-C mismatch sites detected by RES scanner was 6±2 and 18±7 for the empty plasmid control strain and the mutated mdADAR1, respectively, and increased to 240,600±44,700 upon induction of the active mallard duck enzyme. Similarly, editing index was similar for the empty plasmid control strain and the mutated mdADAR1 (0.0154±0.0006% and 0.0161±0.0008%, respectively), but increased substantially for the active enzyme– 0.313±0.050% (S3 Fig).

Altogether, we found 471,784 distinct adenosine positions that were identified as edited in at least one (out of five) mdADAR1 samples (S3 Table - https://docs.google.com/spreadsheets/d/1wXGKUPB39t4P-4AiXq9ZGwfxsR_NrHcw/edit?usp=share_link&ouid=102710005033051837999&rtpof=true&sd=true). That is, we observe editing in about 1 in 16 A:T sites in the whole yeast genome (regardless of the expression level of the locus and strand in which the adenosine resides). Of these, 85,726 sites (~18%) reoccur in all five samples (S4 Fig). Most of the sites were weakly edited, but 79,687 sites (~17%) have shown editing levels of 10% or more in at least one of the samples. Overall, the index value of ~0.3% means that ~1/300 of the adenosines expressed in the transcriptome is converted to inosine. These findings imply that in the absence of long-time adaptation in the ADAR-naïve yeast genome, a sizable amount of adenosine sites are located within preferable substrates for mdADAR1 activity. These results, in agreement with the observations from cell growth assay, demonstrate the potency of mdADAR1.

mdADAR1 and hbADAR2 carry a normal number of dsRBDs (Fig 1B). Their sequences exhibit multiple differences with respect to ADAR enzymes from other taxa, and it is not easy to pinpoint the source for their exceptional editing potency. To identify the domains that contribute to the unique editing capability of mdADAR1 we carried out domain deletion experiments, expressing under GAL1p constructs that include mdADAR1 with its (a) first, (b) first and second, or (c) all three dsRBDs deleted (Fig 4A). In addition, we carried out domain swapping experiments, testing the growth of strains carrying the following hybrids: (i) mdADAR1 combined with the human-dsRBDs1,2,3 (mdADAR-DD-hRBD1-3), and (ii) human ADAR1-DD combined with the md-dsRBDs1,2,3 (hADAR1-DD-mdRBDs1-3) (Fig 4B). The results show that the growth rate of strains where some or all of the dsRBMs were deleted was similar to the wt control, implying that the combination of the DD and all three dsRBDs is important for mdADAR1 editing. Furthermore, the growth of the hADAR1-DD-mdRBDs1-3 strain was also comparable to the wt. In contrast, growth was partially impaired in the mdADAR-DD-hRBD1-3 strain, even though not as much as for the full mdADAR1 strain (Fig 4C). These results imply that the mdRBDs are partially replicable by the hRBD2-3, and that the mdADAR1 catalytic domain has a larger contribution to its potency in yeast.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. mdADAR1 potency is mainly due to its catalytic domain. (A, B) Schematic representation of the full length mdADAR1, the shorter deletion (A), and the domain swapping constructs tested (B). C. 10-fold serial dilutions of the indicated strains were spotted, and growth was quantitated as described in 1D. Plates were incubated at 30°C for 40h. Cells carrying an empty URA3 plasmid were used as a control. https://doi.org/10.1371/journal.pgen.1010661.g004

To conclude, introduction of active ADAR proteins into the ADAR-naïve yeast cells resulted in editing of thousands of sites, which can be readily probed by growth impairment. Of the enzymes tested, mdADAR1 is by far the most potent one, resulting in hundreds of thousands of editing sites and a severe growth impairment.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1010661

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.