Orthogonal RNA-regulated destabilization domains for three-color RNA imaging with minimal RNA perturbation

Abstract

RNA is one of the key molecules responsible for controlling gene expression regulation, and visualizing individual RNA molecules in living cells offers unique insights into the dynamics of this process. Recently, RNA-regulated destabilization domain was developed for live-cell imaging of single RNA. However, this method is limited to single-color RNA imaging and its long RNA tag induces destabilization of the tagged RNA. Here we describe two orthogonal RNA-regulated destabilization domains (mDeg and pDeg) that enable three-color mRNA imaging in living mammalian cells. We show that these destabilization domains can image mRNA tethered to the endoplasmic reticulum membrane, the inner surface of the plasma membrane, and in the cytosol. In addition, we show that mDeg can detect mRNA more effectively than the previously reported tDeg system. Moreover, mDeg can be combined with a short RNA tag (9XMS2) for single-molecule RNA imaging without perturbation of RNA stability.

INTRODUCTION

RNA is one of the most fundamental biomolecules for gene expression regulation. Small nuclear RNAs can form spliceosomes to generate mature messenger RNAs (mRNAs), mRNAs carry genetic information for protein synthesis and microRNAs can fine-tune protein expression levels^1,2. Therefore, robust methods for illuminating the spatiotemporal dynamics of different RNAs and their interactions in living cells are critical for understanding their roles in cellular function.

Conventional RNA imaging methods based on RNA hairpin tethering have high background fluorescence. In these methods, 24 repeats of an RNA hairpin, such as MS2, are fused to the 3’ untranslated region (3’UTR) of an RNA of interest³. Each of the MS2 RNA hairpins can bind and recruit fluorescent proteins (FPs) to the RNA of interest for imaging by MS2-MCP (MS2 coat protein) interactions³. However, the excess unbound FP-MCP creates background fluorescence. To decrease background fluorescence, FP-MCP often needs to be knocked-in to host cells to ensure low expression levels. Although orthogonal RNA hairpin tethering methods have been reported⁴, knocking-in multiple FPs into the host cell genome with proper expression levels can be highly laborious.

An alternative method to image RNA relies on fluorogenic RNA aptamers^5-11. These fluorogenic RNA aptamers can bind and turn on the fluorescence signals of otherwise nonfluorescent small-molecule dyes^5-11. When fused to the 3’UTR of an RNA of interest, these fluorogenic RNA aptamers can confer fluorescence signals to this RNA for imaging. Despite recent developments, many of the dyes for fluorogenic RNA aptamers have limited cell permeability and suboptimal photostability¹². This limits the application of fluorogenic RNA aptamers for imaging cellular RNA with single-molecule resolution.

Recently, another RNA imaging method was developed based on an RNA-regulated destabilization domain¹³. In this method, an FP is fused to an RNA-regulated destabilization domain, termed tDeg. Because tDeg contains a degron sequence, this FP-tDeg fusion protein can be degraded rapidly in cells. However, FP-tDeg will only be stabilized when tDeg specifically binds to the Pepper RNA. By fusing multiple Peppers to the 3’UTR of an RNA of interest, this method enables imaging cellular RNA with single-molecule resolution¹³. Unlike conventional RNA hairpin tethering method, FP-tDeg can be introduced to host cells by transiently transfecting its encoding DNA plasmids, making this method more accessible.

Although the Pepper-tDeg method allows imaging cellular RNA with single-molecule resolution, it still has two major limitations: (1) it only allows imaging of one RNA at a time in living cells due to a lack of orthogonal systems; and (2) it requires a long RNA tag (1,812 nucleotides), which could destabilize the stability of the tagged RNA through nonsense-mediated mRNA decay^13,14. The second problem is also shared by the other RNA imaging tags, such as the 24XBroccoli tag¹⁵ and the 24XMS2 tag¹⁴. Although tethering the poly(A) binding protein, PABPC1, or the polypeptide chain release factor, eRF3, to the 3’UTR of the tagged RNA can alleviate this problem¹⁴, matching the stoichiometry between the tethered protein to the tagged RNA, while avoiding overexpression of the tethered protein, could be challenging.

To address these two major limitations, we describe the development of two orthogonal systems to Pepper-tDeg for live-cell RNA imaging with minimal perturbation to RNA stability. These systems are based on the most widely used RNA-RNA binding protein pairs, MS2-MCP³ and PP7-PCP (PP7 coat protein)⁴, respectively. We converted the single-chain tandem MCP (stdMCP) and single-chain tandem PCP (stdPCP) to two orthogonal RNA-regulated destabilization domains, termed “mDeg” and “pDeg,” respectively. We showed that mDeg and pDeg can image mRNA tethered to the endoplasmic reticulum (ER) membrane, the inner surface of the plasma membrane and in the cytosol. MS2-mDeg can image single mRNA much more effectively than the previously reported Pepper-tDeg system, representing a major improvement for RNA imaging. We also showed that MS2-mDeg, PP7-pDeg, and Pepper-tDeg are fully orthogonal to each other and they enable three-color simultaneous RNA imaging in living cells. Last, we developed a short MS2 tag (9XMS2) that can be used for mRNA imaging with minimal perturbation of RNA stability while retaining single-RNA sensitivity.

RESULTS

Development of mDeg, an MS2-regulated destabilization domain

To develop an orthogonal RNA-destabilization domain pair to Pepper-tDeg, we sought to convert stdMCP to an MS2-regulated destabilization domain. To achieve this, we sought to fuse a previously described C-terminal degron (Arg-Arg-Arg-Gly)^16,17 close to the MS2 binding site in a circularly permuted stdMCP, such that this degron can be recognized by the E3 ubiquitin ligase complex to induce protein degradation in the absence of MS2. However, MS2 binding would block this degron from being recognized by the E3 ubiquitin ligase complex and thus inhibiting protein degradation. After close examination of the MS2-MCP crystal structure¹⁸ (Protein Data Bank 2BU1), we chose Arg49 and Arg83 of stdMCP for circular permutation because (1) both Arg are the same as the first amino acid of the Arg-Arg-Arg-Gly degron and we only needed to append Arg-Arg-Gly to extend this Arg to the full degron; (2) both Arg are close to the MS2 binding site so that MS2 binding may block the Arg-Arg-Arg-Gly degron from recruitment of the proteasomal machinery needed for proteolysis. We therefore termed these two circularly permuted stdMCP with appended Arg-Arg-Gly as stdMCP-CP49-RRRG and stdMCP-CP83-RRRG, respectively (Extended Data Fig. 1).

We first asked whether fusing stdMCP-CP49-RRRG and stdMCP-CP83-RRRG, respectively, to a protein confers protein instability. To test this, we fused stdMCP-CP49-RRRG and stdMCP-CP83-RRRG to the C terminus of enhanced yellow FP (EYFP), and expressed this fusion protein in human embryonic kidney 293T (HEK293T) cells. We found that cells expressing EYFP-stdMCP-CP49-RRRG and stdMCP-CP83-RRRG showed yellow fluorescence that is 50% and 2% of cells expressing EYFP, respectively (Supplementary Fig. 1). Since stdMCP-CP49-RRRG only showed limited destabilization effect to EYFP, we further explored whether the destabilization effect from stdMCP-CP83-RRRG can be inhibited by binding to MS2. Our results showed that coexpression of EYFP-stdMCP-CP83-RRRG and circular MS2¹⁹ led to only a very limited increase in yellow fluorescence (Supplementary Fig. 2). Together, these results indicate that even though stdMCP-CP83-RRRG can confer protein instability to EYFP, this effect cannot be regulated by MS2.

Since stdMCP-CP83-RRRG’s stability cannot be regulated by MS2, we sought to further engineer stdMCP-CP49-RRRG to increase its destabilization effects. We reasoned that the decreased destabilization effects in stdMCP-CP49-RRRG might be due to the decreased accessibility of the Arg-Arg-Arg-Gly degron by the proteasomal machinery. We therefore asked whether adding a flexible linker between stdMCP-CP49 and the Arg-Arg-Arg-Gly degron can increase the accessibility of the degron by the proteasomal machinery and thus result in an increase in destabilization effect. To test this, we constructed DNA plasmids encoding EYFP-stdMCP-CP49-RRRG variants containing a flexible linker from one amino acid to six amino acids, respectively, between stdMCP-CP49 and the Arg-Arg-Arg-Gly degron (Fig. 1). Each EYFP-stdMCP-CP49-RRRG variant (termed a mDeg variant) was expressed in HEK293T cells without the MS2 RNA. Compared to stdMCP-CP49-RRRG (mDeg variant0), we observed a gradual yellow fluorescence decrease from 28% to 92% as the linker length increased from mDeg variant1 to variant6, respectively (Fig. 1). Together, these results indicate that adding a flexible linker between stdMCP-CP49 and the Arg-Arg-Arg-Gly degron increases the degradation efficiency of the resulting destabilization domains.

Fig. 1 ∣. Development of mDeg, an MS2 RNA-regulated protein destabilization domain.

a, Schematic drawing of an MS2 RNA-regulated protein destabilization domain, mDeg. b, MS2 RNA stabilizes EYFP fused to mDeg (variant2) in HEK293T cells. Widefield fluorescence images of HEK293T cells expressing each EYFP-mDeg variant with and without circular MS2. We termed variant2 as mDeg. Shown are representative images from three independent cell cultures. All cells were stained with Hoechst. Scale bar: 100 μm. c, Summary data of normalized fluorescence of each EYFP-mDeg variant with and without circular MS2 as in b. Normalized average cellular yellow fluorescence of individual cells from three independent cell cultures is plotted. (For each bar from left to right, n = 382, 433, 421, 435, 419, 434, 435, 411, 469, 432, 436, 394, 429, and 433 cells, respectively). Values are means ± s.d. ****P_{variant 0} = 6.5 x 10⁻¹¹⁸; ****P_{variant 1} = 2.9 x 10⁻¹¹⁴; ****P_{variant 2} = 2.1 x 10⁻¹²²; ****P_{variant 3} = 1.5 x 10⁻¹⁰⁹; ****P_{variant 4} = 2.9x10⁻⁹⁵; ****P_{variant 5} = 1.2 x 10⁻⁶²; ****P_{variant 6} = 1.5 x 10⁻⁴⁶ using an unpaired, two-tailed, Student’s t-test.

Given flexible linkers between stdMCP-CP49 and the Arg-Arg-Arg-Gly degron increase protein degradation efficiency, we wondered whether this effect could be inhibited by MS2 binding. To test this, we coexpressed each EYFP-mDeg variant with and without circular MS2 in HEK293T cells, respectively. We found that all EYFP-mDeg variants can be regulated by MS2 but to different extents (Fig. 1). Among these, mDeg variant2 containing a ‘Gly-Arg’ linker showed the highest yellow fluorescence increase of 55-fold on binding to circular MS2 (Fig. 1). Biochemical characterization showed that mDeg variant2 binds MS2 with a high affinity (dissociation constant K_d = 35 ± 6 nM) (Supplementary Fig. 3). In addition, MS2-mDeg variant2 can regulate the stability of EYFP in different cell types (Supplementary Fig. 4). Thus, we chose this variant for further characterization and termed this MS2-regulated destabilization domain, ‘mDeg’.

Development of pDeg, a PP7-regulated destabilization domain

Since stdMCP and stdPCP share high structural homology⁴, we therefore asked whether the design of mDeg can be applied to stdPCP for making a PP7-regulated destabilization domain. To test this, we constructed DNA plasmids encoding circularly permuted stdPCP variants at position 45 (equivalent to position 49 in stdMCP) with an appended linker followed by the Arg-Arg-Arg-Gly degron to the C terminus (Extended Data Fig. 2). Similar to the strategy for designing mDeg, we titrated linker length from one to six amino acids between stdPCP-CP45 and the Arg-Arg-Arg-Gly degron (Extended Data Fig. 3). Each of these variants (termed pDeg variants) was coexpressed with and without circular PP7 in HEK293T cells. Among these variants, we observed that pDeg variant2 containing a ‘Gly-Arg’ linker showed the highest fold change in yellow fluorescence of 22-fold (Extended Data Fig. 3) when coexpressed with circular PP7. We also observed yellow-fluorescent puncta in all pDeg variants (Supplementary Fig. 5). Together, these results suggest that pDeg variant2 can serve as a destabilization domain and its destabilization effect can be inhibited when bound by PP7.

Even though pDeg variant2 can be regulated by the PP7 RNA, the yellow-fluorescent puncta suggest aggregation of the EYFP-pDeg variant2-PP7 complex, which could generate artifacts when used for RNA imaging. We therefore sought to eliminate this aggregation phenomenon of the EYFP-pDeg variant2-PP7 complex.

One possibility of the yellow-fluorescent puncta observed is that they result from aggregation of the circular PP7 RNA. To test this, we transiently expressed EYFP-pDeg variant2 lacking the Arg-Arg-Arg-Gly degron (termed pDeg variant7) in HEK293T in the absence of circular PP7 RNA. We found that EYFP-pDeg variant7 showed similar yellow-fluorescent puncta as in cells expressing the EYFP-pDeg variant2-PP7 complex (Supplementary Fig. 5 and 6). As a control, we expressed non-circularly permuted stdPCP fused to EYFP in HEK293T cells. We found that cells expressing EYFP-stdPCP showed the same yellow-fluorescent puncta as in cells expressing the EYFP-pDeg variant2-PP7 complex (Supplementary Fig. 5 and 6). Together, these results indicate that the aggregation of the EYFP-pDeg variant2-PP7 complex does not come from the circular PP7 RNA.

Wild-type PCP can oligomerize to form capsids for the Pseudomonas bacteriophage via interactions between a loop region, called the FG loop²¹. Although we used a capsid assembly-deficient PCP variant with part of its FG loop (Cys67 to Phe74) deleted when constructing the pDeg variants, we nevertheless asked whether the remaining FG loop residues could be the cause of aggregation. To test this, we replaced the remaining FG loop (Val64 to Val78) with a single Gly residue in pDeg variant2 and termed this pDeg variant2(del loop). We then coexpressed EYFP-pDeg variant2(del loop) with and without circular PP7 in HEK293T cells. Without PP7, only minimal yellow fluorescence above background fluorescence was detected. However, when PP7 was coexpressed, cells exhibited a 21-fold increase of yellow fluorescence (Fig. 2). Furthermore, we did not observe any yellow-fluorescent puncta in the EYFP-pDeg variant2(del loop)-PP7 complex (Supplementary Fig. 7). Biochemical characterization showed that pDeg variant2 (del loop) binds PP7 with a moderate affinity (K_d = 426 ± 24 nM) (Supplementary Fig. 3). Furthermore, PP7-pDeg variant2 (del loop) can regulate the stability of EYFP in different cell types (Supplementary Fig. 8). Together, these results indicate that the fluorescent puncta of EYFP-pDeg variant2 was caused by the remaining residues in the FG loop. Since the stability of pDeg variant2 (del loop) can be regulated by PP7, we therefore termed this PP7-regulated destabilization domain, ‘pDeg’.

Fig. 2 ∣. Development of pDeg, a PP7 RNA-regulated protein destabilization domain.

a, Schematic drawing of a PP7 RNA-regulated protein destabilization domain, pDeg. b, PP7 RNA stabilizes EYFP fused to pDeg in HEK293T cells. Widefield fluorescence images of HEK293T cells expressing EYFP-pDeg with and without circular PP7. Shown are representative images from three independent cell cultures. All cells were stained with Hoechst. Scale bar: 100 μm. c, Summary data of normalized fluorescence of EYFP-pDeg with and without circular PP7 as in b. Normalized average cellular yellow fluorescence of individual cells from three independent cell cultures is plotted. (n_control = 526 cells; n_PP7 = 489 cells). Values are means ± s.d. ****P = 3.0 x 10⁻¹²⁶ using an unpaired, two-tailed, Student’s t-test.

MS2-mDeg and PP7-pDeg can regulate the stability of various FPs

To determine whether MS2-mDeg and PP7-pDeg can regulate FPs other than EYFP, we fused mDeg to the C terminus of tdStayGold²², mCherry²³, and HaloTag²⁴, and pDeg to the C terminus of mEGFP²⁵, mScarlet-I3²⁶, and iRFP670²⁷. Each of these proteins was transiently expressed in HEK293T cells with and without its cognate RNA (that is, MS2 or PP7). In addition, we incubated cells expressing HaloTag-mDeg with the JF646 ligand²⁸ before imaging. The expression levels of each protein were measured by fluorescence. In each case, there was a considerable increase in fluorescence of the mDeg-tagged and pDeg-tagged protein only when their cognate RNA was coexpressed in cells (Fig. 3 and Extended Data Fig. 4). Together, these results showed that MS2-mDeg and PP7-pDeg can regulate a diverse array of FPs and the HaloTag.

Fig. 3 ∣. MS2-mDeg confers MS2-dependent regulation to various FPs and HaloTag.

a, Schematic of plasmids that encode protein of interest fused to mDeg, and circular MS2 RNA, respectively. b-d, MS2-mDeg can regulate the stability of various FPs and HaloTag. Widefield fluorescence images of HEK293T cells expressing tdStayGold (b), mCherry (c), and HaloTag (d) fused to mDeg with and without circular 1XMS2. For imaging HaloTag-mDeg, cells were incubated in media with the JF646 ligand. Shown are representative images from three independent cell cultures. All cells were stained with Hoechst. Scale bar: 100 μm. e-g, Summary data of normalized fluorescence of mDeg-fused FPs and HaloTag with and without circular MS2 as in b (e), c (f) and d (g). Normalized average cellular fluorescence of individual cells from three independent cell cultures is plotted. (For each bar from left to right, n = 306, 305, 313, 310, 544, and 468 cells, respectively). Values are means ± s.d. ****P_{tdStayGold-mDeg} = 7.1 x 10⁻¹⁰¹; ****P_mCherry-mDeg = 1.4 x 10⁻⁹³; ****P_HaloTag-mDeg = 6.2 x 10⁻¹²⁵ using an unpaired, two-tailed, Student’s t-test.

MS2-mDeg and PP7-pDeg enable live-cell single RNA imaging with improved SNR

Since fluorogenic proteins containing mDeg and pDeg can be rapidly degraded in cells when not bound to their cognate RNA, this system provides a potential alternative for mRNA imaging without the need for nuclear localization signal (NLS) and stable cell lines. We therefore asked if mDeg-fused and pDeg-fused fluorogenic proteins can image mRNA in living cells by transient transfection. To test this, we first transiently expressed (mNeonGreen)₄-mDeg and (mNeonGreen)₄-pDeg in U2OS cells to image an ER-targeted mTagBFP2 reporter mRNA (CyTERM-mTagBFP2)^Ref,32 containing a previously reported 24XMS2 tag and 24XPP7 tag in the 3’UTR^33,34, respectively. In both cases, we observed green-fluorescent puncta, likely reflecting single mTagBFP2 mRNA (Supplementary Fig. 9). Moreover, these green-fluorescent puncta were not observed in control cells expressing the same mTagBFP2 reporter mRNA without any 24XMS2 tag or 24XPP7 tag (Supplementary Fig. 9). In addition, when compared to the conventional MS2-MCP and PP7-PCP methods, MS2-mDeg and PP7-pDeg showed a higher signal-to-noise ratio (SNR) for imaging the CyTERM-mTagBFP2 reporter mRNA (Supplementary Fig. 10). Together, these results indicate that MS2-mDeg and PP7-pDeg can be used for imaging mRNA with an improved SNR compared to the conventional MS2-MCP and PP7-PCP methods.

Comparison of MS2-mDeg, PP7-pDeg, and Pepper-tDeg for mRNA imaging

To evaluate whether mDeg-MS2 and pDeg-PP7 can detect mRNA in living cells as effectively as by the previously reported Pepper-tDeg method¹³, we generated DNA plasmids encoding the same ER-targeted mTagBFP2 reporter mRNA containing 24XMS2, 24XPP7, and (F30-2xPepper)₁₀, respectively, in the 3’UTR. We imaged these tagged mTagBFP2 mRNA by coexpressing (mNeonGreen)₄ fused to the cognate RNA-regulated destabilization domain in U2OS cells. To assess the performance of each RNA imaging system, we quantified the percentage of transfected cells showing green-fluorescent puncta in the cytosol. Among these three systems, we observed the highest percentage of transfected cells showing green-fluorescent puncta using mDeg-MS2 94% (107 out of 114 cells) (Supplementary Fig. 9). Compared to mDeg-MS2, we observed a lower percentage of transfected cells showing green-fluorescent puncta with pDeg-PP7 (67%, 68 out of 102 cells) and tDeg-Pepper (64%, 62 out of 97 cells) (Supplementary Fig. 9). Together, these results suggest that mDeg-MS2 can detect mRNA most effectively, while pDeg-PP7 performed on par with the previously reported tDeg-Pepper system.

To further characterize the dynamics of the mRNA being tagged by the fluorogenic protein-RNA pairs, we constructed plasmids encoding fluorogenic proteins comprising two AausFP1⁴³ in tandem with a C-terminal mDeg, pDeg, and tDeg, respectively. Each of these fluorogenic proteins was coexpressed with an ER marker, ER-miRFP670 and a reporter mRNA of CyTERM-mTagBFP2 tagged with its cognate RNA tag, respectively. In each case, we observed a similar number of green-fluorescent puncta localized on the ER with limited mobility (apparent diffusion coefficients D for MS2-mDeg, PP7-pDeg, and Pepper-tDeg are 0.0364 ± 0.0011 μm² s⁻¹, 0.0333 ± 0.0010 μm² s⁻¹, and 0.0415 ± 0.0015 μm² s⁻¹, respectively), suggesting that the mRNAs were tethered on the ER membrane (Fig. 4, Supplementary Fig. 11, and Supplementary Videos 1-3). After treatment of a translation inhibitor, harringtonine (HT), we observed a considerable increase in the mobility of the green-fluorescent puncta (D for MS2-mDeg, PP7-pDeg, and Pepper-tDeg are 0.2752 ± 0.0070 μm² s⁻¹, 0.2684 ± 0.0093 μm² s⁻¹, and 0.2449 ± 0.0093 μm² s⁻¹, respectively), which is in agreement with previous reported values^36-38 (Fig. 4 and Supplementary Videos 1-3). Together, these results suggest that MS2-mDeg, PP7-pDeg, and Pepper-tDeg can track the dynamics of RNA in living cells.

Fig. 4 ∣. MS2-mDeg, PP7-pDeg, and Pepper-tDeg can track the dynamics of RNA in living cells.

a, Schematic drawing of a CyTERM-mTagBFP2 reporter mRNA tagged with an RNA imaging tag. b, MS2-mDeg, PP7-pDeg, and Pepper-tDeg can track the dynamics of RNA in living U2OS cells. Widefield fluorescence images of U2OS cells expressing an ER marker (ER-miRFP670) and an mRNA reporter fused to different RNA tags (24XMS2, 24XPP7, and (F30-2xPepper)₁₀, respectively) along with their cognate fluorescent proteins, shown with and without HT treatment. Shown are representative images from three independent cell cultures. Scale bar: 20 μm. c, Summary data of the diffusion coefficient (D) of the puncta before and after HT treatment from three independent cell cultures. (From left to right, n = 866, 869, 620, 564, 455, and 429 puncta, respectively.) Values are means ± s.e.m. ****P_AausFP1-mDeg = 1.7 x 10⁻⁷³; ****P_AausFP1-pDeg = 3.2 x 10⁻⁷⁹; ****P_AausFP1-tDeg = 1.2 x 10⁻⁷² using an unpaired, two-tailed Student’s t-test.

MS2-mDeg and PP7-pDeg can be incorporated into stable cell lines for RNA imaging

Since the 24XMS2 and 24XPP7 tags have been widely used for RNA imaging, we therefore sought to test whether mDeg and pDeg can be used for detecting 24XMS2-tagged and 24XPP7-tagged mRNAs that are stably expressed in mammalian cell lines. To test this, we generated U2OS cell lines stably expressing NLS-mTagBFP2–24XMS2 and CyTERM-mCherry-24XPP7, respectively. We then transiently transfected (AausFP1)₂-mDeg and (AausFP1)₂-pDeg to the cell line stably expressing the reporter mRNA with the cognate tag, respectively. In the U2OS cells stably expressing NLS-mTagBFP2–24XMS2, we observed mobile green-fluorescent puncta in the cytosol, reflecting the NLS-mTagBFP2–24XMS2 mRNA (Extended Data Fig. 5a and Supplementary Video 4). Furthermore, in the U2OS cells stably expressing CyTERM-mCherry-24XPP7, we observed ER-localized green-fluorescent puncta with limited mobility (Extended Data Fig. 5b and Supplementary Video 5), consistent with what we observed when the 24XPP7-tagged reporter mRNA was transiently transfected (Fig. 4 and Extended Data Fig. 5b). Together, these results indicate that fluorogenic proteins containing mDeg and pDeg can be used for imaging 24XMS2-tagged and 24XPP7-tagged mRNA that are stably expressed in U2OS cells.

To further demonstrate the utility of the MS2-mDeg and PP7-pDeg systems, we sought to test whether fluorogenic proteins containing mDeg and pDeg can be incorporated into stable cell lines for RNA imaging. To test this, we generated U2OS cell lines stably expressing (AausFP1)₂-mDeg and (AausFP1)₂-pDeg, respectively. We then transiently transfected plasmids encoding CyTERM-mTagBFP2-24XMS2 and CyTERM-mTagBFP2-24XPP7 to the stable cell line expressing the cognate fluorogenic proteins, respectively. In both cases, we observed ER-localized green-fluorescent puncta with limited mobility (Extended Data Fig. 5c and 5d, and Supplementary Videos 6 and 7), consistent with what we observed when the fluorogenic proteins were transiently transfected (Fig. 4). As a control, the same stable cell lines transiently transfected with plasmids encoding CyTERM-mTagBFP2 without any tags did not show any puncta (Extended Data Fig. 5c and 5d), suggesting that the green-fluorescent puncta were coming from the 24XMS2-tagged and 24XPP7-tagged mRNA, respectively. Together, these results indicate that (AausFP1)₂-mDeg and (AausFP1)₂-pDeg can be incorporated into stable cell lines for RNA imaging.

MS2-mDeg and PP7-pDeg for tethering RNA to the plasma membrane

To image translation, mRNA transcripts are often tethered to the inner surface of the plasma membrane to limit their three-dimensional diffusion^39,40. To test whether mDeg and pDeg can be used for tethering mRNA on the inner surface of the plasma membrane, we chose a transmembrane protein Stargazin⁴¹, which has been used for tethering proteins to the plasma membrane by optogenetic methods⁴². After validating Stargazin’s plasma membrane localization in HEK293T cells (Supplementary Fig. 12), we constructed plasmids expressing Stargazin-tdTomato-mDeg and Stargazin-tdTomato-pDeg, respectively. When each of these proteins was coexpressed with an NLS-mTagBFP2 mRNA reporter fused the cognate RNA tag, we observed red-fluorescent puncta with limited mobility (D for MS2-mDeg and PP7-pDeg are 0.0064 ± 0.0001 μm² s⁻¹ and 0.0042 ± 0.0001 μm² s⁻¹, respectively) (Extended Data Fig. 6 and Supplementary Videos 8 and 9), reflecting mRNA being tethered to the inner surface of the plasma membrane. The mobility of these plasma membrane-tethered mRNAs is in agreement with previous mobility measurements on mRNA tethered to the plasma membrane by the CAAX sequence⁴⁰. Together, these results indicate that mDeg and pDeg can be combined with Stargazin for tethering mRNA on the inner surface of the plasma membrane.

MS2-mDeg, PP7-pDeg, and Pepper-tDeg are orthogonal to each other

We next asked whether MS2-mDeg, PP7-pDeg, and Pepper-tDeg are orthogonal to each other since orthogonality between the three RNA-protein pairs is a prerequisite for using them to image multiple RNA simultaneously in the same cell. To test pairwise orthogonality, we coexpressed all nine possible combinations of EYFP-mDeg, EYFP-pDeg, and EYFP-tDeg with their cognate and non-cognate RNA in HEK293T cells, respectively (Fig. 5a, b). We observed strong yellow fluorescence only when an EYFP-fused destabilization domain was coexpressed with its cognate RNA (Fig. 5a, b). Coexpression of each EYFP-fused destabilization domain with a noncognate RNA showed minimal yellow fluorescence in cells (Fig. 5a, b). Together, these results indicate that MS2-mDeg, PP7-pDeg, and Pepper-tDeg are fully orthogonal in mammalian cells.

Fig. 5 ∣. MS2-mDeg, PP7-pDeg, and Pepper-tDeg are orthogonal to each other and they enable three-color RNA imaging in living cells.

a, MS2-mDeg, PP7-pDeg, and Pepper-tDeg are orthogonal to each other. Shown are representative images from three independent cell cultures. All cells were stained with Hoechst. Scale bar: 100 μm. b, Summary data of normalized fluorescence of a from three independent cell cultures. Average cellular fluorescence of EYFP-fused destabilization domain with non-cognate RNA is normalized to the same protein with cognate RNA. (For each bar from left to right, n = 359, 359, 406, 364, 375, 409, 442, 441, and 395 cells, respectively.) Values are means ± s.d. ****P_mDeg-PP7 = 1.8 x 10⁻⁹⁰; ****P_mDeg-Pepper = 6.8 x 10⁻⁹¹; ****P_pDeg-MS2 = 1.9 x 10⁻⁹¹; ****P_pDeg-Pepper = 2.5 x 10⁻⁹¹; ****P_tDeg-MS2 = 1.3 x 10⁻¹⁰¹; ****P_tDeg-PP7 = 2.0 x 10⁻¹⁰¹ using one-way ANOVA. c, MS2-mDeg, PP7-pDeg, and Pepper-tDeg enable three-color RNA imaging in living cells. We observed mobile fluorescent puncta in the cytosol reflecting cellular mRNAs as shown in Supplementary Video 10. Shown are representative images from three independent cell cultures. Scale bar: 20 μm.

MS2-mDeg, PP7-pDeg, and Pepper-tDeg enable three-color RNA imaging in living cells

We next sought to image multiple different RNAs in living cells at the same time using the orthogonal RNA-regulated destabilization domains. We first started by imaging two different RNAs in living mammalian cells. We coexpressed two reporter plasmids encoding a Nanoluc tagged with 24XMS2 and a CyTERM-mTagBFP2 tagged with 24XPP7, respectively. In the same cells, we coexpressed (HaloTag)₄-mDeg (with the JF646 ligand) and (AausFP1)₂-pDeg as fluorogenic proteins. We observed distinct but nonoverlapping fluorescent puncta in both the GFP channel and Cy5 channel, respectively, reflecting individual mRNAs (Supplementary Fig.13). These results indicate that MS2-mDeg and PP7-pDeg can be used for tracking the dynamics of two RNA species in the same cell at the same time.

After validating that MS2-mDeg and PP7-pDeg can be used for two-color RNA imaging, we then sought to simultaneously image three different RNA species in living cells by combining them with the Pepper-tDeg system. To test this, we coexpressed three reporter plasmids expressing a Nanoluc mRNA tagged with 24XMS2, a CDK6 mRNA tagged with 24XPP7, and a CyTERM-mTagBFP2 tagged with (F30–2xPepper)₁₀ tag in U2OS cells. To visualize these reporter mRNAs, we coexpressed (HaloTag)₄-mDeg (with the JF646 ligand), tdTomato-pDeg, and (mNeonGreen)₄-tDeg as fluorogenic proteins. In this case, we again observed fluorescent puncta in the cytosol reflecting cellular mRNAs (Fig. 5c and Supplementary Video 10). Together, these results indicate that MS2-mDeg, PP7-pDeg, and Pepper-tDeg can be used for simultaneous imaging of three different mRNA species in living cells.

MS2-mDeg combined with photostable fluorogenic proteins

As MS2-mDeg has the best performance for single RNA imaging among the three pairs, we asked whether mDeg can be combined with photostable fluorophores for continuous RNA imaging for an extended period of time. To test this, we chose (HaloTag)₄-mDeg with the JF646 ligand as the fluorogenic protein due to its high photostability. We used this fluorogenic protein to image a NanoLuc-24XMS2 mRNA in living U2OS cells with continuous excitation and image acquisition. We observed mobile far-red fluorescent puncta with continuous image acquisition for 120 seconds with limited photobleaching (Supplementary Video 11). These results indicate that MS2-mDeg can be combined with photostable fluorogenic proteins for live-cell RNA imaging with continuous image acquisition for an extended period of time.

Development of short RNA imaging tags with minimal perturbation of RNA stability

A recent study has shown that the 24XMS2 tag can destabilize the tagged mRNA through nonsense-mediated mRNA decay due to its excess length¹⁴. Since MS2-mDeg and PP7-pDeg are fluorogenic and can detect mRNA in living cells, we reasoned that we might be able to construct a shorter RNA tag to image mRNA without perturbing its stability.

To design short MS2 and PP7 tags with minimal perturbation to RNA stability, we first sought to identify the minimal repeat number of MS2 that does not destabilize the tagged mRNA. To test this, we constructed reporter plasmids expressing mCherry mRNA with different repeat numbers of MS2 at the 3’UTR, including 3X, 6X, 9X, 12X, and 24X, respectively, upstream of a bovine growth hormone polyadenylation signal. We used quantitative real-time PCR with reverse transcription (RT-qPCR) to quantify the levels of the reporter mRNA transcripts in U2OS and HEK293T cells. Consistent with previous report^13,14, 24XMS2 destabilized the tagged mCherry mRNA levels by over 50% compared to an untagged reporter mRNA (Extended Data Fig. 7a). However, we also found that 9XMS2 has a minimal effect on the stability of the tagged mCherry mRNA (Fig. 6a and Extended Data Fig. 7a). Using the same assay, we also found that 6XPP7 does not significantly destabilize the tagged mCherry RNA (Extended Data Fig. 7b). Additionally, we observed the same phenomenon with reporter mRNAs containing different 3’UTRs (GAPDH, RPL13A, and RPL27) (Extended Data Fig. 8). Together, these results indicate that a short imaging tag containing 9XMS2 or 6XPP7 has minimal perturbation on the tagged mRNA’s stability in living cells.

Fig. 6 ∣. 9XMS2, a short RNA imaging tag with minimal perturbation of RNA stability, for single mRNA imaging.

a, 9XMS2 has minimal perturbation on the tagged mCherry mRNA stability in living U2OS cells. RT-qPCR results of the levels of mCherry mRNA transcripts tagged with different repeat numbers (n = 0, 3, 6, 9, 12, and 24, respectively) of MS2 in U2OS cells (n = 3 independent cell cultures). Values are means ± s.d. **P_12XMS2 = 3.1 x 10⁻³; ****P_24XMS2 = 6.3 x 10⁻⁵ using one-way ANOVA. b, 9XMS2 enables single mRNA imaging when combined with (mStayGold)₈-mDeg. We observed mobile green-fluorescent puncta in the cytosol of the U2OS cells as also shown in Supplementary Video 12. Shown are representative images from three independent cell cultures. Scale bar: 20 μm.

We then asked whether these short tags can be used for mRNA imaging in living cells. To test this, we first tested fluorogenic proteins comprising one, two, four or eight tandem mStayGold⁴⁴ with a C-terminal mDeg for imaging a CyTERM-mTagBFP2 mRNA tagged with 9XMS2 in living U2OS cells. We observed the highest percentage of transfected cells showing green-fluorescent puncta using (mStayGold)₈-mDeg, 83% (150 out of 180 cells), compared to (mStayGold)₁-mDeg, 56% (77 out of 138 cells) (Supplementary Fig. 14). Furthermore, mDeg showed a higher SNR compared to the conventional MS2-MCP method with (mStayGold)₈ as the fluorogenic proteins (Supplementary Fig. 15). Together, these results demonstrate that when combined with (mStayGold)₈-mDeg, a short MS2 tag (9XMS2) can be used for mRNA imaging with minimal perturbation of RNA stability.

To test whether 6XPP7 can be used for single mRNA imaging in living cells, we constructed a plasmid expressing a reporter mRNA of CyTERM-mTagBFP2-6XPP7 and used (mStayGold)₈-pDeg as the fluorogenic protein. Unlike using the 9XMS2 tag with (mStayGold)₈-mDeg, we could not reliably observe green-fluorescent puncta with the 6XPP7 tag and (mStayGold)₈-pDeg (Supplementary Fig. 16), likely due to the decreased PP7 repeat number and the moderate binding affinity between pDeg and PP7. Thus, these results indicate that, when combined with (mStayGold)₈-pDeg, 6XPP7 is not able to detect single mRNA in living cells.

Since mStayGold is a highly photostable FP, we then sought to test whether (mStayGold)₈-mDeg can be used for tracking mRNA tagged with 9XMS2 for an extended period of time. To test this, we used (mStayGold)₈-mDeg to image an mCherry mRNA tagged with 9XMS2 in living U2OS cells. We observed mobile green-fluorescent puncta in the cytosol of the U2OS cells (Fig. 6b and Supplementary Video 12), likely reflecting single mRNAs. These mobile green-fluorescent puncta were not observed in cells coexpressing (mStayGold)₈-mDeg and an mCherry mRNA without any MS2 tag (Fig. 6b), suggesting that the green-fluorescent puncta are coming from the (mStayGold)₈-mDeg bound to 9XMS2 in the tagged mCherry mRNA. In addition, due to the high photostability of mStayGold, we were able to continuously track the dynamics of the 9XMS2 tagged mCherry mRNA for up to 40 seconds (Supplementary Video 12). Together, these results indicate that, when combined with (mStayGold)₈-mDeg, 9XMS2 can be used for continuous mRNA imaging for an extended period.

DISCUSSION

Overall, this study describes the development of two orthogonal RNA-regulated destabilization domains for live-cell three-color RNA imaging. These two RNA-regulated destabilization domains (termed mDeg and pDeg, respectively) were engineered from the most widely used RNA-RNA binding protein pairs, MS2-MCP and PP7-PCP, respectively. We showed that MS2-mDeg and PP7-pDeg can regulate the stability of a broad array of FPs and HaloTag for imaging RNA in different cellular localizations. Due to their fluorogenicity and orthogonality, we showed that mDeg and pDeg can be combined with tDeg for simultaneous imaging the dynamics of three different mRNA species in living mammalian cells by transient transfection. Furthermore, we identified 9XMS2 as a short tag for mRNA imaging without perturbing the tagged RNA’s stability.

During the development of pDeg, we unexpectedly found that both EYFP-stdPCP and all EYFP-fused circularly permuted stdPCP variants showed yellow-fluorescent puncta, suggesting protein aggregation when expressed in cells (Supplementary Fig. 5). We found that this aggregation problem was not caused by binding to PP7 or circular permutation of the stdPCP protein. We hypothesized that the aggregation of stdPCP is due to the remaining residues in the FG loop. By removing these remaining FG loop residues (Val64 to Val78), we observed diffused yellow fluorescence without any observable aggregates, further confirming the remaining FG loop primarily contributes to the aggregation of stdPCP.

Compared to the Pepper-tDeg system, MS2-mDeg can detect single mRNA in living cells more effectively. This suggests a major improvement on the reliability for RNA detection using MS2-mDeg. On the other hand, the performance of PP7-pDeg is on par with the current system, Pepper-tDeg. Biochemical characterization showed that PP7-mDeg has a modest binding affinity of (K_d = 426 ± 24 nM) compared to the affinity of MS2-mDeg (K_d = 35 ± 6 nM). We reasoned that the higher affinity of MS2-mDeg may be able to recruit more fluorogenic proteins to the tagged mRNA, which leads to a higher percentage of cells showing fluorescent puncta (Supplementary Fig. 9). Future optimization of the binding affinity of PP7-pDeg may improve its performance for RNA imaging.

mRNA transcripts are often tethered to the inner surface of the plasma membrane to limit their mobility for imaging translation and their interactions with proteins. To tether MS2- or PP7-tagged mRNA to the plasma membrane, it requires coexpression of stdMCP or stdPCP with a C-terminal CAAX sequence^39,40. Since mDeg and pDeg contain a C-terminal degron, further appending the CAAX sequence to mDeg and pDeg will likely inhibit the C-terminal degron’s function. To address this, we showed that mDeg and pDeg can be fused to the C-terminus of a transmembrane protein, Stargazin, for tethering MS2-tagged and PP7-tagged mRNA to the plasma membrane by transient transfection. Conceivably, this method can be readily combined with the SunTag method⁴⁶ for imaging translation.

Long RNA tags have been shown to destabilize the tagged RNAs in cells^14,15. Having shorter RNA tags for imaging is critical to faithfully recapitulate the behavior of endogenous mRNA. We showed that 9XMS2 can be used as a short tag for RNA imaging with minimal perturbation of the tagged mRNA’s stability. This is achieved through taking advantage of the fluorogenic property of MS2-mDeg by using a highly fluorescent and photostable fluorogenic protein, (mStayGold)₈-mDeg.

Although these new RNA-regulated destabilization domains were used for live-cell RNA imaging in this study, the ability to use already widely adapted RNA aptamers, such as MS2 and PP7, to control protein stability would likely enable numerous applications in the field of RNA biology and synthetic biology. It is conceivable that the MS2-mDeg and PP7-pDeg systems described here can be used for tethered function assays⁴⁷ to elucidate the function of RNA-binding proteins without potential artifacts due to protein aggregations from overexpression of RNA-binding proteins with low complexity domains. In addition, MS2-mDeg and PP7-pDeg can be readily combined with the CRISPR-Cas technology for gene expression regulation, genome imaging, and other synthetic biology applications.

Methods

General methods and materials

Gene fragments used for constructing DNA plasmids used in this study were purchased from Twist Bioscience and GenScript. For routine PCR amplifications, single-stranded synthetic DNA oligonucleotides were purchased from Integrated DNA Technologies, Phusion High-Fidelity PCR Master Mix (M0531L) was purchased from New England Biolabs. For restriction digest of PCR products and DNA plasmids, restriction endonucleases were purchased from New England Biolabs, and used according to the manufacturer’s recommended protocol. PCR products and digested DNA plasmids were purified using 0.8% TAE (Tris–acetate–ethylenediaminetetraacetic acid (EDTA) agarose gels and Thermo Scientific Gel Extraction kit (FERK0692). For DNA ligation reactions, Quick Ligation Kit (M2200L) was purchased from New England Biolabs. All DNA plasmids used in this study were propagated using chemically competent Escherichia coli (Agilent, 200315) then extracted using Thermo Scientific GeneJET Plasmid Miniprep Kit (FERK0503). Purified DNA plasmids were sent to Azenta Life Sciences or Plasmidsaurus for DNA sequencing to verify sequence identity.

Cell culture and transfection

All mammalian cell lines used in this study, including HEK293T/17 (American Type Culture Collection (ATCC), CRL-11268), U2OS (ATCC, HTB-96), COS-7 (ATCC, CRL-1651), and HeLa (ATCC, CCL-2) cells, were cultured in Dulbeco’s modified Eagle’s medium (Thermo Fisher Scientific, 11995-065) supplemented with 10% fetal bovine serum (Corning 35-010-CV), 100 U ml⁻¹ of penicillin and 100 μg ml⁻¹ of streptomycin (Thermo Fisher Scientific, 15140122) under 37 °C with 5% CO₂. For cell passage, TrypLE Express (Thermo Fisher Scientific, 12604013) was used for dissociating the adherent mammalian cells from culture flasks. All transfections in this study were carried out using FuGENE HD (Promega, E2312) according to the manufacturer’s recommended protocol. Cell culture media was changed to FluoroBrite^™ DMEM (A1896701) prior to live-cell imaging.

Stable cell line generation

U2OS cells were transfected with DNA expression constructs as shown in Extended Data Fig. 15 and stable lines were isolated by fluorescence-activated cell sorting with assistance from the UMASS Amherst Core Facilities (Cell Culture, SSRID:SRC023477, Flow Cytometry). Before transfection, all expression plasmids (CMV-NLS-mTagBFP2-24XMS2, CMV-CyTERM-mCherry-24XPP7, UbC-(AausFP1)₂-mDeg, and UbC-(AausFP1)₂-pDeg) were linearized by endonuclease digestion with MluI (NEB #R3198) to enhance stable integration.

To generate stable cell lines, 5 μg plasmid DNA was delivered to 1 x 10⁶ U2OS cells for each plasmid using Lonza’s 4D-Nucleofector X Unit and following Lonza’s 4D-Nucleofector Protocol. Briefly, 1 x 10⁶ cells were pelleted and resuspended in 100 μL Nucleofector Solution with supplement from the SE Cell Line 4D-Nucleofector Kit (V4XC-1012). Five micrograms of DNA was added and the solution transferred to a Nucleovette and placed in the instrument. Program CM-104 was run on the 4D-Nucleofector unit. Cells were immediately recovered to fresh, warm media and incubated overnight. On day 2, cells were expanded to a larger vessel and subjected to stable cell selection with 1 mg mL⁻¹ Geneticin (Gibco 10131035). Four days later, clonal pools of low and high expression were sorted by FACS. Cells with low levels of (AausFP1)₂-mDeg/pDeg, high levels of NLS-mTagBFP2-24XMS2, and high levels CyTERM-mCherry-24XPP7 were used for experiments.

Live-cell fluorescence imaging of RNA in mammalian cells

DNA plasmids expressing all the mDeg- and pDeg-tagged fluorogenic proteins and HaloTag, and Stargazin-EGFP used in this study are based on a pcDNA3.1(+) vector backbone containing a miniCMV promoter followed by Kozak sequence (5’-GCCACC-3’). DNA plasmids expressing ER-miRFP670, fluorescent proteins fused to NLS-stdMCP and NLS-stdPCP are based on the same pcDNA3.1 (+) vector backbone with a UbC promoter as previously described³³ followed by Kozak sequence (5’-GCCACC-3’). Gene fragments encoding all the fluorogenic proteins and HaloTag were purchased from Twist Bioscience. These gene fragments were cloned in the pcDNA3.1(+) vector using standard restriction enzyme cloning procedures.

DNA plasmids expressing different circular RNAs are based on a pAV U6+27 vector backbone containing the Tornado expression system, which was a gift from Samie Jaffrey (Addgene plasmid # 129405). Double-stranded synthetic DNA oligonucleotides encoding MS2 and PP7 were purchased from Integrated DNA Technologies and cloned into this circular RNA expression vector backbone containing an F30 folding scaffold⁴⁸ using standard restriction enzyme cloning procedures.

DNA plasmids expressing reporter mRNA are based on a pcDNA3.1(+) vector backbone containing a CMV promoter followed by Kozak sequence (5’-GCCACC-3’). The (F30–2xPepper)₁₀ tag was a gift from Samie Jaffrey (Addgene plasmid # 129403). The 24xMS2 tag was a gift from Robert Singer (Addgene plasmid # 84561). The 24xPP7 tag was a gift from Robert Singer (Addgene plasmid # 40652).

For fluorescence imaging of mDeg- and pDeg-tagged proteins with circular RNA, HEK293T (3.2 × 10⁵), U2OS (2.5 × 10⁵), COS-7 (2.5 × 10⁵), HeLa (2.5 × 10⁵) cells were seeded into 35 mm glass bottom dishes precoated with poly-D-lysine (Metek, P35GC-1.5-14-C), respectively, and cultured overnight. On the next day, cells were transfected using FuGENE HD according to the manufacturer’s recommended protocol. Specifically, 1400 ng of DNA plasmids encoding for fluorogenic protein fused to mDeg or pDeg were mixed with 1400 ng of DNA plasmids encoding for circular RNA for transfecting each 35 mm dish. The cells were imaged two days after transfection. Cell culture media was changed to FluoroBrite^™ DMEM (A1896701) prior to live-cell imaging. For fluorogenic proteins containing HaloTag, a manufacturer’s recommended amount of the JF646 ligand was incubated with the transfected cells for at least one hour prior to imaging.

For fluorescence imaging of mRNAs, U2OS (2.5 × 10⁵) cells were seeded into 35 mm glass bottom dishes precoated with poly-D-lysine (Metek, P35GC-1.5-14-C) and culture overnight. On the next day, cells were transfected using FuGENE HD according to the manufacturer’s recommended protocol. Specifically, a total of 2800 ng of DNA plasmids were used for transfecting each 35 mm dish. For cells that were co-transfected with more than one DNA plasmid, the amount of each plasmid equals to 2800 ng divided by the number of DNA plasmids used. The cells were imaged two days after transfection. Cell culture media was changed to FluoroBrite^™ DMEM (A1896701) prior to live-cell imaging. For fluorogenic proteins containing HaloTag, a manufacturer’s recommended amount of the JF646 ligand was incubated with the transfected cells for at least one hour prior to imaging.

For live-cell fluorescence imaging, a Nikon Ti2-E epifluorescence inverted microscope equipped with a Prime BSI Express sCMOS monochrome camera, a Lumencor SOLA V-NIR Light Engine for Epifluorescence light source, and a temperature and CO₂ stage top incubator (Tokai Hit) was used. The NIS-Elements Advanced Research software (Nikon) was used for controlling the microscope and camera. For live-cell imaging, a 20×/0.80-NA (numerical aperture) or a 60×/1.42-NA oil immersion objective (Nikon) with TYPE 37 immersion oil was used. A GFP filter cube (with excitation filter 470 ± 20 nm, dichroic mirror 495 nm (long pass), and emission filter 525 ± 25 nm) was used for detecting fluorogenic proteins with green fluorescence emissions. A YFP filter cube (with excitation filter 500 ± 10 nm, dichroic mirror 515 nm (long pass), and emission filter 535 ± 15 nm) was used for detecting fluorogenic proteins with yellow fluorescence emissions. A dsRed filter cube (with excitation filter 545 ± 15 nm, dichroic mirror 570 nm (long pass), and emission filter 620 ± 30 nm) was used for detecting fluorogenic proteins with red fluorescence emissions. A Cy5 filter cube (with excitation filter 620 ± 30 nm, dichroic mirror 660 nm (long pass), and emission filter 700 ± 37.5 nm) was used for detecting fluorogenic proteins with far-red fluorescence emissions. A DAPI filter cube (with 395 ± 12.5 nm excitation filter, 425 nm (long pass) dichroic mirror, and 460 ± 25 nm emission filter) was used for detecting the Hoechst-stained nuclei and mTagBFP2. ImageJ was used for analyzing fluorescence images. Cellular fluorescence intensity was calculated by measuring the mean fluorescence signal in a cell’s area and subtracting background based on average signal of culture media. Normalized fluorescence was calculated by dividing the cell fluorescence intensity of each cell to the averaged cell fluorescence of the whole cell population. The mean fluorescence intensity of a cell or a punctum or cytosolic background was calculated using ImageJ by measuring the mean fluorescence signals in a cell’s area or a punctum and subtracting the background based on the average signal of the culture medium. SNR of the fluorescent puncta was calculated by dividing the mean fluorescence intensity of the punctum by the mean fluorescence intensity of the adjacent cytosolic background fluorescence.

For RNA imaging experiments with harringtonine (HT) treatment, U2OS cells were prepared as described above and imaged with pre-warmed FluoroBrite^™ media with recorded xy-coordinates for monitoring the same cells before and after HT treatment. Specfically, images were captured in three channels including DAPI, GFP, and Cy5, respectively, and a 10-second nonstop movie was recorded for each field of view in the GFP channel, before and 30 minutes after HT treatment (final concentration of 4 μM HT). To analyze the diffusion coefficients of particles, General Analysis 3 (GA3) in NIS Elements was used. Single particle tracking was generated using Tracking Binaries with maximum of 3 gap size closing between tracks, minimum of 3 frames in tracks, and a standard deviation multiplication factor of 2. Mean-squared displacement values and the diffusion coefficient of each particle was calculated by GA3.

Protein expression and purification

Constructs for the E. coli expression of pDeg and mDeg were cloned in a pET15b vector with an N-terminal 6X-His-tag followed by a thrombin cleavage site. These plasmids were transformed into BL21(DE3) electrocompetent E. coli and plated on LB agar plates supplemented with ampicillin (100 μg/mL).

A 50-mL culture was grown from a single colony in LB with ampicillin (100 μg/mL) at 37°C overnight. 3 mL of the overnight culture was inoculated into 1 L LB supplemented with ampicillin (100 μg/mL) at 37°C until the OD₆₀₀ reached 0.6 and was induced with 1 mM IPTG (Gold Biotechnology) for 3 hours. Cultures were subjected to centrifugation and pelleted cells stored at −80°C.

Cell pellets were resuspended in lysis buffer (50 mM sodium phosphate pH 8.0, 1 M NaCl, 10 mM imidazole) containing one SIGMAFAST^™ EDTA-free protease inhibitor cocktail tablet (Sigma Aldrich) and lysed with a microfluidizer (Microfluidics^™). The lysate was clarified by centrifugation at 27,000 x g for 50 minutes at 4°C. The clarified lysate was loaded onto a pre-equilibrated 5 mL HisTrap^™ HP Ni Sepharose column (Cytiva Life Sciences). The column was washed with 10 column volumes of lysis buffer before being developed with elution buffer (50 mM sodium phosphate pH 8.0, 1 M NaCl, 250 mM imidazole) for 5 column volumes. Fractions containing the protein of interest were pooled together and diluted with Buffer A (25 mM HEPES pH 7.0, 1 mM DTT, 1 mM EDTA) to dilute the NaCl concentration to 50 mM. The diluted protein was loaded onto a HiTrap^™ SP HP cation exchange column (Cytiva Life Sciences). The column was washed with 5 column volumes of Buffer A, and developed with Buffer B (25 mM HEPES pH 7.0, 1 M NaCl, 1 mM DTT, 1 mM EDTA) using a linear gradient of 0-70% NaCl. The pDeg protein eluted at 570 mM NaCl. mDeg protein eluted at 480 mM NaCl. Fractions of each protein were assessed for purity using SDS-PAGE. The purest protein fractions were pooled together and dialyzed against a buffer with 25 mM HEPES pH 7.0, 25 mM NaCl, 1 mM EDTA. Protein concentration was measured by absorbance at 280 nm. The concentration of the proteins was validated using a BCA protein assay kit (Thermo Fisher Scientific) according to manufacturer’s instructions.

Electrophoretic mobility shift assay

The affinity between MS2-mDeg and PP7-pDeg was measured by electrophoretic mobility shift assay as previously described⁴. The MS2 RNA (5’-ACAUGAGGAUCACCCAUGU-3’) with a 5’-conjugated fluorescein and the PP7 RNA (5’-GCGCACAGAAGAUAUGGCUUCGUGCGC-3’) with a 3’-conjugated fluorescein were purchased from Horizon Discovery/Dharmacon. Two-fold serial dilution of mDeg (from 1.6 μM to 0.8 nM) and pDeg (8.9 μM to 25 nM) were incubated with 2 nM MS2 and pDeg, respectively, in a 20 μL system containing 10 mM Tris, 100 mM NaCl, 0.1 mM EDTA, 5 mM DTT, 100 U RNaseOUT, 0.01 mg/mL tRNA, 0.01% CA 630, 50 μg/mL Heparin. After 2 hours incubation at 4°C, each sample was mixed with a 9X loading dye (30% glycerol and 0.01% bromocresol green) and loaded into precast 6% polyacrylamide gels in 0.5X TBE buffer (Thermo Fisher Scientific). The RNA-protein complex and free RNA were separated by 120V electrophoresis for 40 minutes. Gels were then scanned by a Typhoon 9500 fluorescence imager. The resulting images were analyzed and quantified by ImageJ, then fitted into the Hill1 equation in OriginPro.

Quantitative PCR with reverse transcription

In the RT-qPCR experiments, DNA plasmids expressing reporter mRNA are based on a pcDNA3.1(+) vector backbone containing a CMV promoter followed by Kozak sequence (5’-GCCACC-3’). Gene fragments encoding the different UTRs were purchased from Twist Bioscience. Gene fragments encoding the different repeats of PP7 were purchased from Genscript. These gene fragments were cloned in the pcDNA3.1(+) vector using standard restriction enzyme cloning procedures.

Total cellular RNAs were extracted with a GeneJET RNA Purification Kit (Thermo Fisher Scientific, K0731) according to the manufacturer’s instructions. The purified RNAs were then reverse-transcribed to the cDNAs using SuperScript^™ IV First-Strand Synthesis kit (Invitrogen, 18080051) with an oligo-dT and random hexamer primers according to the manufacturer’s instructions. Quantitative PCR was performed to measure the relative mRNA level by mixing the cDNA with SYBR Green Supermix (Thermo Fisher Scientific, A25741). The amplification conditions were as follows: 50°C for 2 mins, 95°C for 2 min, 40 cycles of 95°C for 15 s, 56°C for 15 s and 72°C for 1 min. qPCR primers used for the amplification were listed in Supplementary Table 1. Gene of interest (GOI) expression measured were normalized to that of the housekeeping gene GAPDH using 2^{−(Cq GOI - Cq GAPDH)}, where $\text{Cq}$ is the amplification cycle measured by qPCR. Biological replicates were tested as indicated in figure legends. For the qPCR experiments in Supplementary Figure 19, we constructed plasmids encoding a non-mammalian synthetic gene in the coding region followed by the 3’UTR indicated in the figure. Specifically, we construct: NanoLuc (coding region)-GAPDH (3’UTR), ecDHFR (coding region)-RPL13A (3’UTR), and TetR (coding region)-RPL27 (3’UTR).

Statistical analysis

All data are expressed as either means ± s.d. or means ± s.e.m. with sample sizes (n) listed for each experiment as indicated in figure legends. Statistical analyses were performed using Excel (Microsoft) and Prism (Graphpad). One-way ANOVA and unpaired, two-tailed, Student’s t-test were used to analyze significant differences between the group means.

Extended Data

Extended Data Fig. 1. Initial designs of an MS2-regulated destabilization domain.

Shown is the crystal structure of MS2-stdMCP (PDB: 2BU1) (a) and our initial designs for converting stdMCP to an MS2-regulated destabilization domain (b). (a) The MS2 RNA is depicted in orange and the stdMCP protein is depicted in grey with side chains of Arg49 and Arg83 highlighted in each MCP monomer. To engineer an MS2-regulated destabilization domain, we circularly permuted stdMCP at Arg49 and Arg83, respectively, and inserted “Arg-Arg-Gly” at the new C-terminus to generate the full “Arg-Arg-Arg-Gly” degron. We chose Arg49 and Arg83 for circular permutation because they are close to the MS2 binding site, and thus, binding of the MS2 RNA may block the Arg-Arg-Arg-Gly degron from recruitment of the proteasomal machinery needed for proteolysis. (b) Partial sequence of our designs for an MS2-regulated destabilization domain. Amino acid sequence of circularly permutated stdMCP is highlighted in grey, the “Arg-Arg-Arg-Gly” degron sequence is in red.

Extended Data Fig. 2. Initial design of a PP7-regulated destabilization domain.

Shown is the crystal structure of PP7-stdPCP (PDB: 2QUX) (a) and our initial design for converting stdPCP to a PP7-regulated destabilization domain (b). (a) The PP7 RNA is depicted in tiffany green and the stdPCP protein is depicted in grey with side chains of Arg45 (equivalent to Arg49 in stdMCP) highlighted in each PCP monomer. Similar to the initial design of mDeg, we circularly permuted stdPCP at Arg45 and inserted “Arg-Arg-Gly” at the new C-terminus to generate the full “Arg-Arg-Arg-Gly” degron. We termed this stdPCP-CP45-RRRG, also known as pDeg variant0 in Extended Data Figure 3. The idea is that the binding of the PP7 RNA to stdPCP-CP45-RRRG may block the Arg-Arg-Arg-Gly degron from recruitment of the proteasomal machinery needed for proteolysis. (b) Partial sequence of our design for a PP7-regulated destabilization domain. Amino acid sequence of circularly permutated stdPCP is highlighted in grey, the “Arg-Arg-Arg-Gly” degron sequence is in red.

Extended Data Fig. 3. The stability of pDeg variants can be regulated by the PP7 RNA.

To construct a PP7-regulated destabilization domain, we used a similar strategy used for developing mDeg. Specifically, we titrated linker length from one to six amino acids between stdPCP-CP45 and the Arg-Arg-Arg-Gly degron, resulting in seven pDeg variant (0-6) as shown in (a). To test whether PP7 can regulate the stability of these different pDeg variants (0-6), we coexpressed each EYFP-pDeg variant with and without circular 1XPP7 in HEK293T cells, respectively. We found that stdPCP-CP45-RRRG (pDeg variant0) showed limited degradation efficiency as it exhibited 76% of yellow fluorescence compared to EYFP. The addition of linkers between stdPCP-CP45 and the Arg-Arg-Arg-Gly degron led to decreases of yellow fluorescence in the absence of circular PP7. When PP7 was coexpressed, we found that pDeg variant2 containing a ‘Gly-Arg’ linker showed the highest fold change in yellow fluorescence of 22-fold, as shown in (b) and (c). Shown are representative images from 3 independent cell cultures. All cells were stained with Hoechst. Scale bar, 100 μm. Normalized average cellular yellow fluorescence of individual cells is plotted in (c). (For each bar from left to right, n = 348, 377, 330, 400, 369, 376, 370, 386, 394, 339, 355, 395, 347, 376, 332, and 295 cells, respectively). Values are means ± s.d. ****P_variant0 = 2.6 x 10⁻²⁶; ****P_variant1 = 2.4 x 10⁻⁴¹; ****P_variant2 = 1.2 x 10⁻¹¹⁰; ****P_variant3 = 3.6 x 10⁻¹¹⁰; ****P_variant4 = 9.5 x 10⁻⁷⁵; ****P_variant5 = 8.0 x 10⁻⁶⁶; ****P_variant6 = 1.1 x 10⁻⁷¹ using unpaired, two-tailed, Student’s t-test.

Extended Data Fig. 4. PP7-pDeg confers PP7-dependent regulation to various fluorescent proteins.

(a) Schematic of plasmids that encode protein of interest fused to pDeg, and circular 1XPP7 RNA, respectively. Proteins of interest include mEGFP, mScarlet-I3, and iRFP670.

(b-d) PP7-pDeg can regulate the stability of various fluorescent proteins. To test whether the PP7 RNA stabilizes different proteins fused to pDeg, we imaged HEK293T cells expressing mEGFP (b), mScarlet-I3 (c), and iRFP670 (d) fused to pDeg with and without circular 1XPP7, respectively. In each case, there was a considerable increase in fluorescence. Shown are representative images from 3 independent cell cultures. All cells were stained with Hoechst dye. Scale bar, 100 μm.

(e-g) Summary data of normalized fluorescence of pDeg-fused fluorescent proteins with and without circular PP7 as in (b-d). Normalized average cellular fluorescence of individual cells is plotted. For (e), n_−PP7 = 298 cells, n_+PP7 = 317 cells; for (f), n_−PP7 = 308 cells, n_+PP7 = 304 cells; for (g), n_−PP7 = 369 cells, n_+PP7 = 319 cells. Values are means ± s.d. ****P_mEGFP-pDeg = 3.4 x 10⁻⁹⁵; ****P_{mScarletI3-pDeg} = 1.3 x 10⁻⁸⁴; ****P_iRFP670-pDeg = 1.2 x 10⁻⁷⁹ using an unpaired, two-tailed, Student’s t-test.

Extended Data Fig. 5. MS2-mDeg and PP7-pDeg can be incorporated into stable cell lines for RNA imaging.

To test whether mDeg and pDeg can be used for detecting 24XMS2- and 24XPP7-tagged mRNAs that are stably expressed in mammalian cell lines, we generated U2OS cell lines stably expressing NLS-mTagBFP2-24XMS2 (a) and CyTERM-mCherry-24XPP7 (b), respectively. We then transiently transfected (AausFP1)₂-mDeg and (AausFP1)₂-pDeg to the cell line stably expressing the reporter mRNA with the cognate tag, respectively. In the U2OS cells stably expressing NLS-mTagBFP2-24XMS2, we observed mobile green-fluorescent puncta in the cytosol, reflecting the NLS-mTagBFP2-24XMS2 mRNA (a). Furthermore, in the U2OS cells stably expressing CyTERM-mCherry-24XPP7, we observed green-fluorescent puncta colocalized with the ER (b). These results suggest that fluorogenic proteins containing mDeg and pDeg can be used for imaging 24XMS2- and 24XPP7-tagged mRNA that are stably expressed in U2OS cells. Shown are representative images from 3 independent cell cultures. Scale bar, 20 μm. To test whether fluorogenic proteins containing mDeg and pDeg can be incorporated into stable cell lines for RNA imaging, we generated U2OS cell lines stably expressing (AausFP1)₂-mDeg (c) and (AausFP1)₂-pDeg (d), respectively. We then transiently transfected plasmids encoding CyTERM-mTagBFP2-24XMS2 (c) and CyTERM-mTagBFP2-24XPP7 (d) to the stable cell line expressing the cognate fluorogenic proteins, respectively. In both cases, we observed ER-localized green-fluorescent puncta colocalized to the ER. These results suggest that (AausFP1)₂-mDeg and (AausFP1)₂-pDeg can be incorporated into stable cell lines for RNA imaging. Shown are representative images from 3 independent cell cultures. Scale bar, 20 μm.

Extended Data Fig. 6. MS2-mDeg and PP7-pDeg can be used for tethering RNA to the plasma membrane.

To test whether Stargazin can be combined with mDeg and pDeg for tethering RNA to the plasma membrane, we constructed plasmids expressing Stargazin-tdTomato-mDeg and Stargazin-tdTomato-pDeg, respectively (a). When each of these proteins was coexpressed with an NLS-mTagBFP2 mRNA reporter fused the cognate RNA tag, we observed red-fluorescent puncta (b) with limited mobility (D for MS2-mDeg and PP7-pDeg are 0.0064 ± 0.0001 μm²/s and 0.0042 ± 0.0001 μm²/s, respectively) (c), which is in agreement with previous measurements on mRNA tethered to the plasma membrane by the CAAX sequence². These results suggest that mDeg and pDeg can be combined with Stargazin for tethering mRNA on the inner surface of the plasma membrane. Shown are representative images from 3 independent cell cultures. Scale bar, 20 μm. (n_24XMS2 = 2144 puncta; n_24XPP7 = 4526 puncta). Values are means ± s.e.m.

Extended Data Fig. 7. Development of short RNA imaging tags with minimal perturbation of RNA stability.

To design short tags with minimal perturbation to RNA stability, we performed RT-qPCR to quantify the levels of mCherry mRNA transcripts tagged with different repeat numbers (n = 3, 6, 9, 12, and 24, respectively) of MS2 (a) and PP7 (b), respectively, and compared these to the same mCherry mRNA transcript without any tag (n = 0) in HEK293T cells. Our results showed that 9XMS2 (a) and 6XPP7 (b) have minimal perturbation on the tagged mCherry mRNA stability in living cells. (n = 5 independent cell cultures for the MS2 tags; n = 3 independent cell cultures for the PP7 tags). Values are means ± s.d. **P_0xMS2-12xMS2 = 1.2 x 10⁻³; ****P_0xMS2-24xMS2 = 4.2 x 10⁻⁶; **P_0xPP7-9xPP7 = 2.6 x 10⁻³; ****P_0xPP7-12xPP7 = 1.2 x 10⁻⁵; ****P_0xPP7-24xPP7 = 8.0 x 10⁻⁵ using one-way ANOVA.

Extended Data Fig. 8. 9XMS2 and 6XPP7 do not significantly destabilize the tagged mRNA containing different 3’UTRs.

To demonstrate the generality of the short tags, we constructed plasmids expressing reporter mRNAs containing different 3’UTRs (GAPDH, RPL13A, and RPL27) fused to different repeats of MS2 (a) and PP7 (b), respectively. We performed RT-qPCR to quantify the levels of reporter mRNA transcripts tagged with different repeat numbers of MS2 (a) and PP7 (b), respectively. Our results showed that 9XMS2 (a) and 6XPP7 (b) have minimal perturbation on the stability of the tagged reporter mRNAs, which is consistent with what we observed in Extended Data Figure 7. However, reporter mRNAs tagged with 24XMS2 (a) or 24XPP7 (b) showed a considerable decrease in their RNA levels. Thus, these results suggest that 9XMS2 and 6XPP7 do not significantly destabilize the tagged mRNA containing different 3’UTRs. (n = 3 independent cell cultures). Values are means ± s.d. ****P_{0xMS2-24xMS2 GAPDH} = 2.0 x 10⁻⁶; ****P_{0xMS2-24xMS2 RPL13A} = 2.9 x 10⁻⁵; **P_{0xMS2-24xMS2 RPL27} = 4.7 x 10⁻³; ****P_{0xPP7-24xPP7 GAPDH} = 3.8 x 10⁻⁷; ****P_{0xPP7-24xPP7 RPL13A} = 4.5 x 10⁻⁶; ****P_{0xPP7-24xPP7 RPL27} = 1.6 x 10⁻⁵ using one-way ANOVA.

Supplementary Material

Supplementary Information is linked to the online version of the paper.

Data Availability

The sequence of the plasmids used in this study have been deposited on GenBank with the following accession number: miniCMV-EYFP-mDeg (PX401316), U6+27-Tornado-F30-MS2 (PX401332), miniCMV-EYFP-pDeg (PX401317), U6+27-Tornado-F30-PP7 (PX401333), CMV-CyTERM-mTagBFP2-24XMS2 (PX401318), miniCMV-(AausFP1)2-mDeg (PX401319), CMV-CyTERM-mTagBFP2-24XPP7 (PX401320), miniCMV-(AausFP1)2-pDeg (PX401321), CMV-NanoLuc-24XMS2 (PX401322), miniCMV-(HaloTag)4-mDeg (PX401323), CMV-CDK6-24XPP7 (PX401324), miniCMV-tdTomato-pDeg (PX401325), CMV-mCherry-9XMS2 (PX401326), miniCMV-(mStayGold)8-mDeg (PX401327), CMV-NLS-mTagBFP2-24XMS2 (PX401328), miniCMV-Stargazin-tdTomato-mDeg (PX401329), CMV-NLS-mTagBFP2-24XPP7 (PX401330), miniCMV-Stargazin-tdTomato-pDeg (PX401331). These plasmids are also available on Addgene according to the terms of the Uniform Biological Material Transfer Agreement. Data generated from this study are available through the corresponding author upon reasonable request. Source data are provided with this paper.

Code Availability

Not relevant.