Translating ribosome affinity purification (TRAP) enables rapid and efficient isolation of cell type-specific translating mRNA. Here, we demonstrate a method that combines hydrodynamic tail-vein injection in a mouse model of liver repopulation and TRAP to examine the expression profile of repopulating hepatocytes.
Liver repopulation after injury is a crucial feature of mammals which prevents immediate organ failure and death after exposure of environmental toxins. A deeper understanding of the changes in gene expression that occur during repopulation could help identify therapeutic targets to promote the restoration of liver function in the setting of injuries. Nonetheless, methods to isolate specifically the repopulating hepatocytes are inhibited by a lack of cell markers, limited cell numbers, and the fragility of these cells. The development of translating ribosome affinity purification (TRAP) technology in conjunction with the Fah-/- mouse model to recapitulate repopulation in the setting of liver injury allows gene expression profiling of the repopulating hepatocytes. With TRAP, cell type-specific translating mRNA is rapidly and efficiently isolated. We developed a method that utilizes TRAP with affinity-based isolation of translating mRNA from hepatocytes that selectively express the green fluorescent protein (GFP)-tagged ribosomal protein (RP), GFP:RPL10A. TRAP circumvents the long time period required for fluorescence-activated cell sorting that could change the gene expression profile. Furthermore, since only the repopulating hepatocytes express the GFP:RPL10A fusion protein, the isolated mRNA is devoid of contamination from the surrounding injured hepatocytes and other cell types in the liver. The affinity-purified mRNA is of high quality and allows downstream PCR- or high-throughput sequencing-based analysis of gene expression.
As the main metabolic organ in vertebrates, the liver is responsible for glucose homeostasis, serum protein synthesis, bile acid secretion, and xenobiotic metabolism and detoxification. The liver possesses an extraordinary capacity to regenerate the injured parenchyma upon exposure to toxins to prevent immediate liver failure1. However, failure of regeneration can occur in the setting of acetaminophen or alcohol overconsumption, which can lead to acute liver failure2. Furthermore, chronic liver injury caused by viral hepatitis infection, fatty liver disease, and steatohepatitis can cause liver fibrosis, cirrhosis, and hepatocellular carcinoma3. The only available curative treatment for end-stage liver disease is transplantation but is limited by organ shortage, preventing efficient treatment for all patients4. A better understanding of the recovery process after toxic liver injury is therefore crucial for the development of treatments to stimulate regeneration sufficient to rescue function in the diseased organ.
The most broadly applied model system for the study of liver regeneration is partial hepatectomy in rodents, in which a large proportion of the liver is resected to stimulate rapid hepatocyte expansion5. However, partial hepatectomy does not recapitulate hepatocyte expansion following toxic liver injury due to the lack of immune cell infiltration and hepatocyte cell necrosis often observed in the setting of acute liver injury in humans6. A more suitable system to model this form of organ renewal is the Fah-/- mouse, which lacks functional fumarylacetoacetate hydrolase (FAH) required for proper tyrosine metabolism and develops severe liver damage leading to death7. These mice can be maintained in a healthy state indefinitely by treatment with the drug nitisinone in the drinking water. Alternatively, FAH expression can be restored by transgene delivery to a subset of hepatocytes, which will expand to repopulate the liver upon nitisinone removal8.
To profile the gene expression changes of repopulating hepatocytes, a tool to specifically isolate replicating hepatocytes in the Fah-/- mouse without contamination from the neighboring injured hepatocytes and other cell types is required. Unfortunately, fluorescence-assisted cell sorting (FACS) of hepatocytes is difficult since (1) the fragility of repopulating hepatocytes leads to poor recovery after liver perfusion, (2) replicating hepatocytes are highly variable in size, making isolation of a pure population by FACS difficult, and (3) the procedure time from liver perfusion to RNA isolation is greater than 2 h, hence gene expression profiles may undergo substantial artificial changes before samples are acquired9.
Alternatively, the expression of epitope-tagged ribosomes specifically in repopulating hepatocytes allows for the rapid isolation of actively translating mRNA bound by ribosomes using affinity purification immediately after organ harvest, with bulk liver tissue lysates. Here, we describe a protocol to perform translating-ribosome affinity purification (TRAP)10 followed by high-throughput RNA-sequencing (TRAP-seq), to specifically isolate and profile mRNA in repopulating hepatocytes in the Fah-/- mouse9. Coexpression of green fluorescent protein-tagged ribosomal protein (GFP:RPL10A) with FAH allows affinity purification of translating mRNA bound by polysomes containing GFP:RPL10A. This method avoids any cell dissociation steps, such as liver perfusion to isolate fragile repopulating hepatocytes. Instead, it utilizes lysis of whole organ tissue and antibodies to rapidly extract the RNA specifically from the target cells. Finally, isolation of abundant, high-quality mRNA via TRAP-seq enables downstream applications such as sequencing analysis to profile the dynamic change of gene expression during the repopulation process.
All methods that involve the use of mice are consistent with the guidelines provided by the IACUC of the Penn Office of Animal Welfare at the University of Pennsylvania.
1. Reagent Preparation
2. Buffer Preparation
3. Conjugation of Antibodies to Magnetic Beads
4. Liver Tissue Lysis
5. Immunoprecipitation
6. RNA Isolation
7. Optional RNA Quality Analysis (Recommended)
8. Downstream Applications
NOTE: Total RNA isolated by the TRAP protocol can be used in a number of standard downstream applications, including RNA-seq (TRAP-seq). Standard reverse transcription and quantitative PCR protocols can also be used following TRAP.
To profile gene expression in repopulating hepatocytes of the Fah-/- mouse, Gfp:Rpl10a fusion and Fah transgenes are co-delivered within a transposon-containing plasmid8 (TRAP vector) to livers by hydrodynamic injection (Figure 1A). The removal of nitisinone induces a toxic liver injury that creates a selection pressure for hepatocytes stably expressing FAH to repopulate the injured parenchyma9. Immunofluorescence staining confirms the co-expression of FAH and the GFP:RPL10A fusion protein in repopulating hepatocytes after two weeks of liver repopulation (Figure 1B).
In the following representative experiment, TRAP-seq was performed using quiescent and repopulating mouse hepatocytes. First, to obtain GFP-tagged ribosomes from quiescent hepatocytes, transgenic RosaLSL-GFP-L10A mice were injected with AAV8-TBG-Cre 7 days prior to sacrifice to induce GFP:RPL10A expression in all hepatocytes14. We also processed a liver sample collected from a wild type mouse as a negative control to ensure isolation of translating mRNA was specific, meaning RNA could only be extracted from mice expressing GFP:RPL10A. The concentration of isolated RNA correlated with the number of cells expressing the fusion protein, with the quiescent sample producing the highest yield since all hepatocytes express GFP:RPL10A after AAV8-TBG-Cre injection (Figure 2A). Conversely, barely any RNA was detectable in wild type controls that did not have the GFP:RPL10A transgene, indicating the TRAP procedure is highly specific and has a low background. When TRAP was used on liver tissue undergoing repopulation with GFP:RPL10A-transduced hepatocytes, abundant, high-quality RNA was obtained (Figure 2B). In contrast, no RNA trace was detected via bioanalyzer for the negative control sample.
Downstream gene expression analysis can be carried out via reverse transcription and quantitative PCR or RNA-seq on TRAP-isolated RNA. Gsta1 encodes glutathione S-transferase that plays an important role for the metabolism of glutathione, the main detoxifying peptide to protect the liver against oxidative stress damage15. Gsta1 expression is induced by over 10-fold in repopulating hepatocytes as compared to quiescent hepatocytes, while no threshold cycle (Ct) value was detected with TRAP-isolated RNA from the wild type mouse due to the lack of input RNA (Figure 3A). Note that RNA quality can greatly impact the gene expression analysis. In the case of RNA-seq experiments, assessment of RNA quality should be performed according the recommendations of the library preparation kit and the sequencing platform (Figure 3B). A bioanalyzer is often used to determine the RNA integrity number (RIN), with a high RIN correlating with a higher rate of mRNA alignments to the genome (Figure 3B, left), whereas a lower RIN leading to a higher rate of ribosomal reads, indicating mRNA degradation (Figure 3B, right). Figure 3C,D demonstrates that TRAP-seq can identify differential gene expression in quiescent and repopulating hepatocytes. For instance, Alb expression is inhibited and Afp expression is upregulated during liver repopulation, reflecting that the replicating hepatocytes assume a less differentiated state to inhibit liver metabolic functions during repopulation9,16.
Figure 1: Implementation of TRAP with Fah-/- to profile gene expression change of repopulating hepatocytes. (A) Schematic of expressing the GFP:RPL10A fusion protein with FAH in the Sleeping Beauty transposon system followed by injection into the Fah-/- mouse. Green hexagons indicate repopulating hepatocytes with stable expression of FAH and GFP:RPL10A, whereas black hexagons represent injured, dying hepatocytes. (B) Representative immunofluorescence staining demonstrates coexpression of GFP-tagged ribosomal protein L10A (green) and FAH (red) in the repopulating hepatocytes. Scale bar = 50 µm. Please click here to view a larger version of this figure.
Figure 2: TRAP allows cell type-specific isolation of high-quality RNA. (A) The yield of RNA is positively correlated with the number of hepatocytes expressing GFP:RPL10A. The low yield of RNA from a wild type mouse demonstrates the specificity of TRAP from sources without the expression of GFP:RPL10A. (B) Bioanalyzer traces of total RNA isolated from repopulating livers expressing GFP:RPL10A and from wild type livers demonstrate the specificity of TRAP. Total RNA isolated from wild type liver tissue devoid of the GFP:RPL10A transgene shows that minimal RNA has been collected, whereas transgene-expressing tissue provides ample high-quality RNA. Note that ribosomal RNA peaks are present following successful TRAP10. FU, fluorescence unit. RIN, RNA integrity number. Please click here to view a larger version of this figure.
Figure 3: TRAP-isolated RNA can be used for downstream gene expression analysis. (A) Representative reverse transcription and quantitative PCR results of Gsta1 in quiescent and repopulating hepatocytes. No Ct value was detected with RNA isolated from wild type animals. (B) Alignment analysis of isolated RNA after high-throughput sequencing, demonstrating the importance of determining RNA integrity after isolation. High-quality RNA results in a higher percentage of mRNA reads (green), while low-quality RNA leads to a much higher percentage of ribosome reads (red), as most mRNA is degraded. RIN, RNA integrity number. (C,D) Integrative Genomics Viewer (IGV) tracks of RNA-seq reads of mRNA affinity-purified from quiescent and repopulating hepatocytes at the (C) Alb and (D) Afp loci. Note that the 3' read bias is typical of a poly(A) selection pipeline. Please click here to view a larger version of this figure.
TRAP-seq is a technique for cell type-specific isolation of translating mRNA via epitope-tagged ribosomes and presents an alternative to FACS approaches, as it circumvents limitations such as time requirements of FACS9. Instead, TRAP allows rapid and efficient isolation of RNA directly from bulk tissues, helping to avoid any alterations in gene expression. TRAP-seq is especially well-suited for use in the repopulating Fah-/- mouse liver, as hepatocyte expansion following removal of nitisinone is cell autonomous and enables gene expression profiling of the subset of hepatocytes with integrated transgenes. The TRAP vector can also be coexpressed with gene-activating or -silencing molecules17, including cDNA, short-hairpin RNA, and guide RNA, to study the effects on global gene expression of activation or inhibition of a specific gene. Alternatively, the RosaLSL-GFP-L10A transgenic mouse provides the ability to profile gene expression in any cell with Cre recombinase activity. Since GFP:RPL10A can be specifically expressed in any cells that express Cre, the role of other cell types in the liver during liver injury and repopulation could be studied. For instance, crossing CK19-Cre mouse with the TRAP transgenic mouse could be used to express GFP:RPL10A in cholangiocytes followed by TRAP-seq to study the change of gene expression in the biliary epithelium during the repopulation process.
To ensure accurate profiling of gene expression, it is critical to prepare all buffers and the affinity matrix prior to tissue dissection. All steps should be performed on ice with cold buffers unless otherwise specified to ensure polysome stabilization10 and prevent RNA degradation. All buffers should be prepared with RNase-free reagents and the TRAP-seq protocol should be carried out in an RNase-free environment to prevent RNA degradation and low yield of immunoprecipitated RNA. The affinity matrix can be prepared up to 2 weeks prior to use with gentle resuspension on a tube rotator overnight. Special care should be taken to not vigorously shake the matrix to prevent disruption of the antibody-conjugated, protein L-coated magnetic beads. The methods to prepare the affinity matrix includes conjugation of magnetic beads to biotinylated protein L followed by incubation with anti-GFP antibodies. However, commercially available protein A/G magnetic beads can be substituted; if used, skip the initial conjugation step and proceed directly to antibody binding. Furthermore, alternative epitope tags are presumably feasible with the above protocol with appropriate modification.
There are various points in which the RNA isolation and purification step can be paused (see protocol above). However, once liver samples have been harvested, continuing to the immunoprecipitation is recommended, as the yield of isolated RNA could drop by ~50% with freezing at this step10. Tissues should be quickly rinsed with dissection buffer that contains cycloheximide to inhibit mRNA translation. Insufficient tissue lysis could also contribute to low RNA yield. It is critical to homogenize tissues on ice until no tissue chunks are visible with the motor homogenizer while ensuring minimal aeration10. Additionally, sufficient washing with high-salt buffer is crucial to ensure removal of nonspecific binding of ribosomal proteins to the affinity matrix. Including a wild type negative control helps to assess the specificity of the immunoprecipitation and the efficiency of the wash steps. Additionally, using a commercial RNA purification kit that includes RNase-free DNase treatment will increase RNA purity.
Moreover, it is recommended to verify the expression and abundance of the GFP:RPL10A fusion protein and assess the amount of tissue required to obtain ample RNA for downstream analysis. Tissue sections or lysates could be used for immuno-based detection methods to validate the expression of GFP:RPL10A. The amount of RNA isolated can vary by: (1) the number of cells expressing GFP:RPL10A, (2) the expression level of the transgene, and (3) the size and ploidy of the cells expressing the transgene. A pilot experiment using half and double the amount of the recommended amount of tissue could be useful in determining the optimal input lysate for TRAP-seq. In our hands, we could obtain ~150 ng of RNA with as little as 1-2% of hepatocytes expressing GFP:RPL10A from 200 mg of the repopulating Fah-/- liver, representing ~2 x 105 polyploid hepatocytes with transgene expression9.
The TRAP-seq methodology isolates ribosome-bound mRNA to profile a cell's translating mRNA pool. The resulting sequencing reads therefore correspond to the 'translatome' rather than the transcriptome. Note that translating ribosome footprints will not be collected, as TRAP is performed on native rather than crosslinked complexes. If footprinting analyses are desired, the above protocol should be modified with relevant cross-linking followed by immunoprecipitation (CLIP) methodologies18. Another limitation of TRAP is the requirement of a sufficient number of cells expressing the GFP:RPL10A fusion protein. For experiments in which the cell type of interest is small, combining multiple biological samples may be required to isolate sufficient RNA to enable RNA-seq19. Furthermore, TRAP-seq requires the presence of GFP:RPL10A in the cell type of interest. This could pose a challenge if there is no specific delivery system to the cells or if a cell-type specific promoter to drive Cre expression is not available.
The recent development of single-cell RNA-seq (scRNA-seq) technology has allowed direct sequencing followed by in silico identification of various cell types, enabling sequencing without sorting for specific cell types of interests20,21,22. However, scRNA-seq still requires dissociation of cells from the organ. In the case of the Fah-/- repopulation model, liver perfusion and hepatocyte isolation is extremely difficult and inefficient due to the fragility of both the injured and replicating hepatocytes. In fact, we have not yet been able to isolate sufficient hepatocytes from Fah-/- mice undergoing repopulation after hydrodynamic injection of Fah plasmids. Additionally, in the time it takes to process tissues, gene expression levels could change. Protocols for liver perfusion take up to 30 min of warm ischemia time. Future methodologies to optimize liver perfusion to decrease processing time and increase isolation efficiency could allow scRNA-seq integration to the Fah-/- mouse model system and possibly to other injury and repopulation models. This would also enable the study of all liver cell types.
In conclusion, the integration of TRAP-seq with the Fah-/- mouse allows specific isolation and gene expression profiling of replicating hepatocytes to identify therapeutic targets that could promote liver repopulation. This method can be implemented to study other cell types in the liver and other organ systems for disease-specific identification of gene expression changes to identify potential drug targets or biomarkers. An analogous technique can be used to collect nuclei from repopulating hepatocytes using affinity purification, and then to perform an epigenetic analysis of these specific cells16.
The authors have nothing to disclose.
This work was supported by the following grants: F31-DK113666 (AWW), K01-DK102868 (AMZ), K08-DK106478 (KJW), and P30-DK050306 (Pilot grant to KJW).
10 mL Tissue Grinder, Potter-Elv, Coated | DWK Life Sciences (Wheaton) | 358007 | |
Absolutely RNA Miniprep Kit | Agilent | 400800 | |
Anti-GFP antibodies | Memorial Sloan-Kettering Antibody & Bioresource Core | GFP Ab #19C8 and GFP Ab #19F7 | |
Bovine Serum Albumin, IgG-Free, Protease-Free | Jackson ImmunoResearch | 001-000-162 | |
cOmplete, Mini, EDTA-free Protease Inhibitor Cocktail | Roche | 11836170001 | |
Cycloheximide | Millipore Sigma | C7698 | |
Deoxycholic acid, DOC | Millipore Sigma | D2510 | |
D-Glucose, Dextrose | Fisher Scientific | D16 | |
DL-Dithiothreitol | Millipore Sigma | D9779 | |
Dynabeads MyOne Streptavidin T1 | Thermo Fisher Scientific | 65602 | |
Fisherbrand Petri Dishes with Clear Lid | Fisher Scientific | FB0875712 | |
HBSS (10x), calcium, magnesium, no phenol red | Thermo Fisher Scientific | 14065-056 | |
HEPES, 1M Solution, pH 7.3, Molecular Biology Grade, Ultrapure, Thermo Scientific | Thermo Fisher Scientific | AAJ16924AE | |
Magnesium chloride, MgCl2 | Millipore Sigma | M8266 | |
Methanol | Fisher Scientific | A452 | |
NanoDrop 2000/2000c Spectrophotometer | Thermo Fisher Scientific | VV-83061-00 | |
NEBNext Poly(A) mRNA Magnetic Isolation Module | New England BioLabs | E7490S | |
NEBNext Ultra RNA Library Prep Kit for Illumina | New England BioLabs | E7530S | |
Nonylphenyl polyethylene glycol, NP-40. IGEPAL CA-630 | Millipore Sigma | I8896 | |
Nuclease-Free Water, not DEPC-Treated | Ambion | AM9932 | |
Overhead Stirrer | DWK Life Sciences (Wheaton) | 903475 | |
PBS Buffer (10x), pH 7.4 | Ambion | AM9625 | |
Pierce Recombinant Protein L, Biotinylated | Thermo Fisher Scientific | 29997 | |
Potassium chloride, KCl | Millipore Sigma | P4504 | |
RNA 6000 Pico Kit & Reagents | Agilent | 5067-1513 | |
RNaseZap RNase Decontamination Solution | Invitrogen | AM9780 | |
RNasin Ribonuclease Inhibitors | Promega | N2515 | |
Sodium azide, NaN3 | Millipore Sigma | S2002 | |
Sodium bicarbonate, NaHCO3 | Millipore Sigma | S6297 | |
SUPERase·In RNase Inhibitor | Invitrogen | AM2694 |