Analysis of RNA-containing compartments by hybridization and proximity labeling in cultured human cells

Summary This protocol describes a hybridization-proximity labeling (HyPro) approach for identification of proteins and RNAs co-localizing with a transcript of interest in genetically unperturbed cells. It outlines steps required for purification of a recombinant HyPro enzyme, hybridization of fixed and permeabilized cells with digoxigenin-labeled probes, HyPro enzyme binding, proximity biotinylation, and downstream analyses of the biotinylated products. Although the protocol is optimized for relatively abundant noncoding transcripts, recommendations are provided for improving the signal-to-noise ratio in case of scarcer RNA “baits.” For complete details on the use and execution of this protocol, please refer to Yap et al. (2021).

. Purification and quality control of recombinant HyPro enzyme (A) Two sequentially connected HisTrap columns used for the first step of HyPro protein purification. The APEX2 moiety of HyPro is a heme-containing enzyme co-purifying with this prosthetic group from bacteria and making concentrated HyPro protein appear visibly brown-red. In the photograph, imidazole-eluted HyPro peak has reached the bottom column. (B) HyPro protein eluted from HisTrap and ready to be loaded onto a size-exclusion column. (C) HyPro elution profile from a Superdex 75 size-exclusion column monitored by UV absorbance at 280 nm. (D) SDS-PAGE analysis of the peak fractions from the size-exclusion step. Purified fractions pooled for further analyses are highlighted in red. (E) Peroxidase activity assay of purified HyPro enzyme. (F) Spot assay showing that HyPro enzyme can bind digoxigenin-labeled oligonucleotides while retaining its peroxidase activity. 10. Wash the column with 20 mL of buffer A at 1 mL/min. 11. Elute His-tagged HyPro protein with a 50%-50% mixture of buffer A and buffer B (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 500 mM imidazole, and 14 mM b-ME) at 0.5 mL/min ( Figures 1A  and 1B). 12. Load the protein peak from HisTrap (6 mL) onto a HiLoad 26/60 Superdex 75 size-exclusion column (GE Healthcare) equilibrated with buffer C (20 mM Tris, pH 8.0, 100 mM NaCl, and 1 mM DTT) at 1 mL/min and elute with buffer C at 2 mL/min. 13. Monitor protein elution by UV absorbance at 280 nm and collect 1-mL fractions ( Figure 1C).
Pause point: Protein fractions can be stored in closed tubes at 4 C for 24-48 hours while performing the analyses described in steps 14 and 15. 14. Analyze the fractions by SDS-PAGE and stain with Coomassie R-250 ( Figure 1D). 15. Measure the protein concentration using a Pierce BCA Kit (Thermo Fisher Scientific) as recommended and pool the peak fractions containing large amounts of HyPro protein and no major contaminants ( Figure 1D). 16. Aliquot purified HyPro protein, snap-freeze in liquid nitrogen, and store at À80 C.
Note 1: First-time users may consider performing a pilot small-scale expression experiment and comparing IPTG-induced and non-induced total and clarified lysates by SDS-PAGE. IPTG-induced samples should contain a prominent HyPro protein band migrating between 37 and 50 kDa.
Note 2: We typically obtain 12-18 mg of purified HyPro protein from 600 ml of IPTG-induced bacterial culture.
Assaying peroxidase activity of purified HyPro enzyme Timing: 5-10 min 17. Mix 1 mL of purified HyPro protein with 20 mL of reconstituted enhanced chemiluminescence (ECL) reagent (e.g., from Thermo Fisher Scientific or Millipore). 18. Incubate for 1 min at 20 C-24 C. 19. Spot onto a piece of filter paper and image immediately using an Odyssey Fc system (LI-COR) ( Figure 1E).
Note: Use an equal amount of bovine serum albumin (BSA) as a negative control.

Preparing antisense oligonucleotide labeled with digoxigenin
Timing: 45 min-1 h 20. For non-repetitious RNA targets, design 24-48 (48 is preferred) antisense DNA oligonucleotides using Stellarisâ probe designer program (LGC Biosearch Technologies; https://www. biosearchtech.com/support/tools/design-software/stellaris-probe-designer). Order them in a 96-well non-modified format (e.g., from IDT; https://eu.idtdna.com/; dissolved at 100 mM in 10 mM Tris-HCl and 0.1 mM EDTA, pH 8.0) and proceed with the 3 0 digoxigenin labeling steps 21-23. Oligonucleotide mixtures used to label 45S and NEAT1 RNAs are described in Table S1. For RNA targets containing short tandem repeats, such as PNCTR (Yap et al., 2018), a single repeat-specific oligonucleotide might be sufficient to produce a strong and specific HyPro signal (Yap et al., 2021). In this case, it may be cheaper to order the oligonucleotide with the 3 0 -terminal digoxigenin modification (e.g., /3Dig_N/; https://eu.idtdna.com/) and use it for hybridization directly. 21. Dilute aliquots of non-modified oligonucleotide stocks to 10 mM with nuclease-free water and pool in a 1.5-mL microcentrifuge tube. 22. Label the oligonucleotide mixture using a 2nd generation DIG Oligonucleotide 3 0 -End Labeling Kit (Sigma Aldrich) as recommended to yield 5-mM digoxigenin-labeled probe pool. 23. Aliquot and store at À20 C. Avoid repeated freezing and thawing.
Note: A scrambled version of target-specific probes designed using an appropriate online program (e.g. https://www.genscript.com/tools/create-scrambled-sequence) provides a good negative control for HyPro labeling experiments.
Note: This assay may be also used to monitor the performance of DIG Oligonucleotide 3 0 -End Labeling Kit (steps 20-23 above).
Validating probe specificity by RNA-FISH

Timing: 3-4 days
We typically validate specificity of newly designed digoxigenin-labeled probes by RNA-FISH before using them for HyPro labeling. The following steps describe RNA-FISH analysis of the 45S and NEAT1 RNAs in HeLa cells. We have also used this protocol to analyze human induced pluripotent stem (iPS) and ARPE-19 cells. a. Incubation in CSK (10 mM PIPES-KOH, pH 6.8, 3 mM MgCl 2 , 100 mM NaCl, 300 mM sucrose) with 0.5% Triton X-100 followed by fixation with 4% formaldehyde. b. Fixation with 4% formaldehyde followed by permeabilization with 0.1% Triton X-100.
Note: If necessary, fixed and permeabilized cells can be immunostained with an antibody against a compartment-specific protein marker, post-fixed with 4% formaldehyde, and washed 3 times with 13PBS before proceeding with the following steps.
35. Rinse the cells with 23SSC, 10% formamide for 1-2 min. 36. Dilute the digoxigenin-labeled oligos in the hybridization buffer (see below) to a final concentration of 125 nM. You will need 20-30 mL hybridization mixture for each coverslip.
Note: It may be necessary to increase the final probe concentration for some targets. The highest concentration we tried is 400 nM.
37. Spread a layer of parafilm on a flat surface (e.g., a plastic or glass plate) keeping the clean side up (Figure 2A). 38. Spot 20-30 mL of the probe-containing hybridization mixture (Figure 2A).  Note: it is possible that the hybridization conditions (e.g. the temperature and the concentration of formamide) will have to be optimized for RNA targets with unusually high or unusually low GC content.
Note: If necessary, this step can be carried out at 20 C-24 C for 1-2 h. However, room-temperature incubation may increase the background staining compared to 4 C.
Alternatives: Alexa Fluor 647 can be substituted with another bright fluorophore detectable by your imaging system.  47. Wash at 20 C-24 C once with 1 mL 43SSC, once with 1 mL 43SSC and 0.1% Triton X-100, and once with 1 mL 43SSC, 10 min each wash. 48. Rinse with 13 PBS for 1 min. 49. Counterstain with DAPI (0.5 mg/mL) in 13 PBS for 3 min and wash briefly with 13PBS. 50. Mount on a clean glass slide with ProLong Gold Antifade mountant (Thermo Fisher Scientific; avoid trapping air bubbles) and cure for 16-24 h at 20 C-24 C in the dark. 51. Seal glass coverslips with transparent nail polish. 52. Image the cells using an epifluorescence microscope ( Figures 2F and 2G).
Note: We use a ZEISS Axio Observer 7 system equipped with an a Plan-Apochromat 1003/ 1.46 Oil DIC M27 objective, a Colibri 7 light source, and a Hamamatsu ORCA-Flash4.0 V3 Digital CMOS camera. To image Alexa Fluor 647-stained samples, we use 5%-20% of the maximal light source intensity and 0.5-2 s exposure time, depending on the RNA target abundance.  Weigh out single-use aliquots and store them in the powder form in 1.5-mL tubes at À80 C ( Figure 3A)

STEP-BY-STEP METHOD DETAILS
The protocol describes HyPro labeling of the 45S and NEAT1 RNAs in fixed and permeabilized HeLa cells followed by fluorescence imaging of proximity-labeled foci (HyPro-FISH) or purification of biotinylated proteins and RNAs for mass-spectrometry and RNA-seq (HyPro-MS and HyPro-seq). We have also used this protocol to analyze the 45S RNA in human iPS cells, NEAT1 in ARPE-19 cells and PNCTR in HeLa cells.

Hybridization and proximity labeling
Timing: 3 days For HyPro-FISH, grow cells in 12-well plates on 18-mm round coverslips.
To isolate HyPro-labeled proteins and RNAs, use the 10-cm dish format.
Note: Optimization of cell seeding density may be needed if a different cell line is used.
CRITICAL: The working solution of DSP must be prepared immediately before the fixation and be completely transparent.  24. Mount on a clean glass slide with ProLong Gold Antifade mountant avoiding air bubbles and cure for 16-24 h at 20 C-24 C in the dark. 25. Seal glass coverslips with transparent nail polish. 26. Visualize the signal using a widefield epifluorescence microscope (Figure 4). Troubleshooting 2.
CRITICAL: test a range of digoxigenin probe concentrations for newly designed and labeled probe sets.
Note: we use 3% of the maximal light source intensity and 150 ms exposure time for both 45S and NEAT1 HyPro-FISH.

Cell lysis and decrosslinking
Timing: 2 h The following four steps are common for both protein and RNA isolation parts of the protocol.
27. Aspirate the solution and lyse cells directly in the 10-cm dish with 600 mL High-SDS RIPA buffer.
Spread the buffer over the entire dish and incubate on ice for 5 min. 28. Scrape the lysed material off the plate and incubate for 10 min on ice. 29. Split the lysate into %300-mL aliquots in 1.5-mL microfuge tubes and sonicate using a Bioruptor (Diagenode) set on ''high'', for 7 cycles of 30 s ON/30 s OFF.
Note: these settings may need to be modified for a different cell line or/and sonicator model.
30. Incubate at 37 C for 30 min to reverse DSP crosslinks.

Isolation of biotinylated proteins
Timing: 2 days Follow this part of the protocol to capture proximity-biotinylated proteins for subsequent immunoblotting and label-free mass-spectrometry analyses. To purify biotinylated RNAs, proceed to step 50.
31. Spin decrosslinked lysate at 15,0003g for 10 min at 4 C. 32. Transfer the supernatant to a new tube.
Pause point: Protein lysates can be stored at À80 C.
Optional: Set aside 10% of the lysates as the input fraction for subsequent analyses.
33. Wash 60 mL of streptavidin magnetic beads twice with Regular-SDS RIPA buffer. 34. Resuspend the beads in 3 mL Regular-SDS RIPA buffer, combine with $600 mL of lysates, and incubate at 20 C-24 C for 2 h with rotation Alternatives: The bead-lysate slurry can be alternatively incubated overnight ($16 h) at 4 C.
Note: The 6-fold dilution of the lysate is required to improve binding of biotinylated proteins to streptavidin.
35. Pellet the beads using a DynaMagä-2 magnet and remove the supernatant   Table  S1 for probe sequences). Hybridizations in the second from the bottom and the bottom rows were carried out without probes. Cells in the bottom row were additionally infused with diluted HyPro enzymes immediately before the proximity biotinylation step. Scale bars, 10 mm. Note: we prefer overnight 4 C incubations with streptavidin-HRP since they tend to produce a better signal-to-noise ratio.
43. Pellet the beads, remove the supernatant and wash the beads three times with 50 mM ammonium bicarbonate. 44. Resuspend in 45 mL of 50 mM ammonium bicarbonate containing 1.5 mg of Trypsin/Lys-C mix (7.5 mL of 0.2 mg/mL stock diluted with 37.5 mL ammonium bicarbonate). Incubate overnight ($16 h) at 37 C, with rotation. 45. The next day, add an additional 0.75 mg Trypsin/Lys-C mix in 50 mM ammonium bicarbonate (3.75 mL of 0.2 mg/mL stock diluted with 11.25 mL ammonium bicarbonate) and incubate for another 2 h at 37 C, with rotation. 46. Pellet the beads and transfer the supernatant to a fresh tube. 47. Wash the beads twice with 45 mL of mass-spec grade water (90 mL in total) and combine the washes with the $60 mL of supernatant collected at the previous step. 48. Centrifuge the combined solution at 14,000-16,0003g for 10 min to remove any remaining beads, and transfer the supernatant to a fresh tube. 49. Submit the samples for label-free mass spectrometry analysis.
CRITICAL: use gloves to avoid contaminating protein samples with keratin.

Isolation of biotinylated RNAs
Timing: 2 h Follow this part of the protocol to capture biotinylated RNAs for RT-qPCR and RNA-seq analyses.
50. Incubate decrosslinked lysate from step 30 with 20 mL of 20 mg/mL proteinase K at 37 C for 20 min and then at 50 C for 60 min. (E) A dot blot assay detects biotinylated RNA species in both total and streptavidin bead pull-down fractions from 45S-and NEAT1-labeled and HyPro-infused samples, but not from a no-probe control.
Note: For 45S-and Neat1-labeled RNA samples, we recommend using 18 cycles at the "PCR cycling enrichment" step.

EXPECTED OUTCOMES
Expected yield of purified HyPro enzyme is $20-30 mg from 1 L of IPTG-induced bacterial culture ( Figures 1A-1D). Typical results of the peroxidase and digoxigenin binding assays used to control quality of HyPro enzyme preparations are shown in Figures 1E and 1F. RNA-FISH validation of 45S-and NEAT1-specific digoxigenin-labeled oligonucleotide probes should produce characteristic nucleolar and paraspeckle signals, while scrambled controls, no or little staining ( Figures 2F  and 2G).
When optimizing working concentrations of digoxigenin-labeled probe sets by HyPro-FISH, it is useful to remember that less probe is typically needed in this case compared to RNA-FISH. For example, we use 125 nM of both 45S-and NEAT1-specific probe sets for RNA-FISH, and 5 nM and 25 nM of these probes, respectively, for HyPro-FISH ( Figure 4). HyPro-FISH staining is expected to be similar to RNA-FISH, with only minimal blurring of nuclear body outlines. HyPro-infusion control is expected to stain the cell homogeneously ( Figure 4).
For abundant RNA targets, streptavidin-HRP immunoblotting of HyPro-labeled proteins should show a clear difference between RNA-specific and scrambled or no-probe controls ( Figures 5A  and 5B). Since this analysis detects predominantly the most abundant proteins, the difference between different RNA-specific samples tends to be more subtle, with only a few distinct bands. Label-free mass spectrometry is much more sensitive in identifying sample-specific proteins (Yap et al., 2021).
Finally, target RNAs are expected to show some enrichment in corresponding HyPro-labeled samples compared to other RNAs ( Figures 5C and 5D). Detecting this effect by relatively cheap RT-qPCR assays is recommended before performing more expensive RNA-seq analyses (Yap et al., 2021).

LIMITATIONS
A major limitation of all APEX-based techniques, including HyPro-MS and HyPro-seq, is a relatively large labeling radius that may result in false-positive biotinylation of abundant cellular proteins and RNAs. This problem can be tackled, at least in part, by comparing proteomes and transcriptomes associated with an RNA of interest and other RNAs showing similar intracellular localization (e.g., 45S vs. NEAT1 or 45S vs. PNCTR) instead or in addition to the HyPro-infusion controls. Another strategy for reducing the spread of reactive biotin from the HyPro recruitment sites may involve increasing viscosity of the labeling solution, an approach previously used in a genome mapping protocol called TSA-seq (Chen et al., 2018).
Our protocol is optimized for relatively abundant RNAs (R50 molecules per cell; (Yap et al., 2018)) localizing to membraneless compartments. It is possible that HyPro analyses of less abundant targets will require a more stringent optimization of the ratio between specific and nonspecific signals.
We therefore recommend validating specificity of all newly designed probes by RNA-FISH and HyPro-FISH with appropriate negative and positive controls.
A previous study reported that APEX2 must be concentrated locally or expressed above a certain level to achieve detectable proximity labeling (Tan et al., 2020). If this is a general feature of APEX2-based protocols, HyPro-labeling of rare or/and diffusely distributed RNAs may require the Finally, oligonucleotide probes used in our method may compete with cellular proteins recruited to overlapping target sequences. It is therefore recommended to design antisense oligonucleotides against multiple positions in the target sequence. Designing more than one probe set against the same RNA target transcript may be also considered for improved detection sensitivity and specificity.

Potential solution
Prepare fresh DSP by adding it drop wise to the PBS with intermittent mixing. Ideally, it should be prepared immediately before use since it may begin precipitating after $30 min.

Problem 2
No or weak RNA-FISH or/and HyPro-FISH signals (step 26)

Potential solution
Consider (1) increasing the concentration of digoxigenin-labeled probes; (2) increasing biotin phenol and hydrogen peroxide concentration in the HyPro labeling step; or/and (3) extending the duration of the HyPro labeling step to 2-5 min.

Potential solution
Assuming that the probe set has been validated by RNA-FISH and HyPro-FISH, scale up the number of cells used for pulldown. Consider redesigning the probe set if the performance of RNA-FISH and HyPro-FISH is poor.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources should be directed to the lead contact, Eugene Makeyev (eugene.makeyev@kcl.ac.uk).

Materials availability
All reagents used in this study are described in the key resources table.
Data and code availability Not applicable.