We describe a technique for profiling microRNAs in early mouse embryos. This protocol overcomes the challenge of low cell input and small RNA enrichment. This assay can be used to analyze changes in miRNA expression over time in different cell lineages of the early mouse embryo.
MicroRNAs (miRNAs) are important for the complex regulation of cell fate decisions and developmental timing. In vivo studies of the contribution of miRNAs during early development are technically challenging due to the limiting cell number. Moreover, many approaches require a miRNA of interest to be defined in assays such as northern blotting, microarray, and qPCR. Therefore, the expression of many miRNAs and their isoforms have not been studied during early development. Here, we demonstrate a protocol for small RNA sequencing of sorted cells from early mouse embryos to enable relatively unbiased profiling of miRNAs in early populations of neural crest cells. We overcome the challenges of low cell input and size selection during library preparation using amplification and gel-based purification. We identify embryonic age as a variable accounting for variation between replicates and stage-matched mouse embryos must be used to accurately profile miRNAs in biological replicates. Our results suggest that this method can be broadly applied to profile the expression of miRNAs from other lineages of cells. In summary, this protocol can be used to study how miRNAs regulate developmental programs in different cell lineages of the early mouse embryo.
A central question of developmental biology is how a single undifferentiated cell can give rise to an entire organism with numerous complex cell types. During embryogenesis, the developmental potential of cells becomes progressively restricted as the organism develops. One example is the neural crest lineage, which progressively differentiates from a multipotent cell population into various terminal derivatives, such as peripheral neurons, glia, cranial bone, and cartilage. Neural crest cells are specified from the ectoderm during gastrulation and then undergo an epithelial to mesenchymal transition and migrate through the embryo to discrete locations throughout the body where they will terminally differentiate1. Decades of work have uncovered a transcriptional gene regulatory network, but far less is known about the mechanisms of post-transcriptional regulation that control the timing of neural crest development.
Previous work suggests that microRNAs (miRNAs) repress gene expression for proper developmental timing and cell fate decisions2,3,4,5,6. Studies of miRNAs in neural crest development have largely focused on later stages of craniofacial development. For example, miR-17~92 and miR-140 are critical for palatogenesis during craniofacial development in mouse and zebrafish, respectively7,8. The contribution of miRNAs to the earliest neural crest fate decisions of the embryo has not been thoroughly investigated. Studies of miRNAs in early fate decisions have been limited by technical challenges such as the low cell number present in early embryos.
MiRNAs have been profiled in vitro from cell lines using embryoid bodies at different stages of differentiation to model early mouse development9. The investigation of small RNAs in vivo during early mammalian development has been relatively limited. Previous methods to profile miRNAs have led to bias as a known sequence is used to analyze expression of a specific miRNA in methods such as qPCR, microarrays, and northern blots10. Next generation sequencing and ever improving molecular tools now allow for relatively unbiased analysis of miRNA expression to study their contribution to early mammalian development and cell fate decisions.
Here, we report a technique to harvest and sequence small RNAs expressed in neural crest cells from early mouse embryos spanning gastrulation (E7.5) to the beginning of organogenesis (E9.5). This technique is straightforward and combines lineage tracing, cell sorting, and gel-based size selection to prepare small RNA sequencing libraries from a minimal number of cells for next generation sequencing. We highlight the importance for strict somite stage matching of embryos to resolve 6-hour time intervals to obtain a comprehensive view of miRNAs during the rapid changes of early development. This method can be widely applied to genetic and developmental studies and avoids the pooling of embryos. We describe a way to overcome challenges of current methods such as miRNA enrichment using gel-based purification, library quantification, and minimizing bias introduced from PCR. This method has been used to identify miRNA expression patterns over time to study how miRNAs control developmental timing in the neural crest lineage of mouse embryos.
All research and animal care procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee and housed in the Association for Assessment and Accreditation of Laboratory Animal Care-approved animal facility at Baylor College of Medicine. All strains were maintained on C57BL6 background.
1. Embryo dissection (E7.5-E9.5)
2. Embryo dissociation and cell sorting
3. RNA extraction
NOTE: The protocol is adapted from the RNA isolation kit; see Table of Materials.
4. Library preparation
NOTE: The protocol is adapted from small RNA library preparation kit handbook; see Table of Materials.
5. Size selection
Using the procedure demonstrated here, we have harvested embryos at E7.5, E8.5, and E9.5. Extraembryonic structures were removed from all embryos and then embryos were somite staged to resolve 6-hour time intervals (Figure 1A-1B). Using principle component analysis to group samples based on similarity, we find that samples cluster by age, highlighting the variation as a result of embryonic age and the need for careful somite matching for biological replicates (Figure 1C). We profiled the pluripotent epiblast at E7.5, lineage traced premigratory and migratory neural crest cells using Wnt1-Cre at E8.5 and migratory neural crest cells using Sox10-Cre at E9.5 (Figure 1D). Here we specifically harvest the cranial neural crest by decapitating the embryo just above the otic placode. A comparison of the two Cre-drivers (Wnt1 and Sox10) that are frequently used to label neural crest cells confirms they mark different populations in early mouse embryos (Figure 1E). Gating strategies were used to obtain live single RFP positive cells which were sorted directly into RNA extraction lysis solution (see Table of Materials) and stored at -80 °C (Figure 1F). It is important to keep track of how many cells were harvested from each sample.
RNA isolation was performed using a modified version of the RNA kit protocol. Specifically, we use the mini RNA columns that are to be stored at 4 °C. These columns are useful for eluting in a small volume (11 µL) to obtain the highest possible concentration of RNA from samples where cell input is limiting. For this same reason, it is important to obtain all of the aqueous phase after the phenol chloroform extraction. Slow pipetting and tilting the tube to one side while collecting are critical to maximize yield. In this procedure, we quantify the RNA using both a spectrophotometer, to obtain information on salt/protein contamination, and a capillary electrophoresis to measure concentration (Figure 2). The spectrophotometer trace reveals that RNA was isolated with no contaminating proteins but has high salt content (Figure 2A-2B). Ideally the 260/280 ratio should be ~1.8 and the 260/230 ratio should be >2.0. The total RNA yield as measured by the spectrophotometer did not increase with age between E8.5-E9.5. This is due to the resolution of the spectrophotometer not being sensitive enough to detect changes in RNA concentration from the number of cells that we are harvesting, and we recommend using the concentration information obtained from the capillary electrophoresis (Figure 2C). The capillary electrophoresis trace can be used to estimate concentration and size of the RNA fragments. Peaks <1000 nucleotides are indicative of degradation. The representative trace is consistent with RNA that is not degraded. Peaks at 2000 nucleotides and 5100 nucleotides are 18s and 28s rRNA, respectively. The small RNA region is located at ~150 nucleotides (Figure 2D).
Small RNA sequencing libraries were prepped using the small RNA library prep kit described in the Table of Materials. Here we use just less than half of the RNA obtained from each sample (4 µL of the 11 µL elution, ~80-120 ng) that enables RNA-sequencing libraries to be synthesized from the remaining RNA from each sample. Here we dilute the 3’ and 5’ small RNA adapters ¼ to lower the amount of adapter dimers present and suggest that the adapter ligation, cleanup, and reverse transcription steps are completed as much as possible in a continuous manner. We have used 16 PCR cycles to amplify the libraries and suggest that the minimum number of PCR cycles for any experiment be empirically determined. By completing excess PCR cycles, one could artificially inflate the read count of a lowly expressed miRNA.
Size selection is imperative to enrich for libraries of miRNAs and not adapter dimers, which are typically present. The library product size (150 bp) and adapter dimers (130 bp) are very similar in size. Gel extraction is used to isolate the small RNA sequencing libraries away from the adapter dimers (Figure 3). An image of a PAGE gel before excision shows that a multitude of product sizes are present in each sample (Figure 3A). It is important to leave one lane open between any two samples or a ladder that is loaded onto the gel. The migration front at the bottom of the gel is slightly curved indicating that running at a slower speed may be necessary if the 150 bp band is difficult to distinguish for all samples. This representative image also shows an abundance of 150 bp product from the higher concentration of positive control RNA (total RNA from the brain of a rat) that went into the library prep as compared to that which went into each embryonic sample. The negative control reveals that the reagents were free of contaminating nucleic acid species (Figure 3A). Excision of the 150 bp band should be done with a clean razor blade and the area collected is shown in Figure 3B. Capillary electrophoresis traces before and after size selection show the dramatic improvement of the purity of the 150 bp library product with gel purification (Figure 3C-3D). It is important to note that the traces in Figure 3C-3D have sample peaks higher than the markers, indicative of overloading. In these cases, the size of the fragments can be accurate, however the concentration may not. Quantification can be obtained by diluting the libraries into the range of the markers or using alternative methods. We find some variation across alternative methods of library quantification and recommend that multiple methods of quantification be empirically tested. Accurate quantification of concentration is essential when maximizing the number of samples to be sequenced at one time. Libraries will be diluted down to 1.3 pM for sequencing and generally anywhere from 15-25 samples can be sequenced at one time on a 150 cycle sequencing kit that provides ~140 million reads. This results in about 5 million reads per sample. Generally mapping rates are between 60-80%.
Figure 1: Harvesting cells from mouse embryos for small RNA sequencing. (A) E8.5 mouse embryo to highlight the removal of the yolk sac and somites used to determine the stage of the embryo (B) Somite staging of mouse embryos to capture the stages of neural crest development (C) Principle component analysis of libraries prepped from sorted wildtype neural crest cells showing that samples group by age (D) Schematic showing how samples were harvested using Wnt1-Cre at E8.5 to label premigratory and migratory neural crest cells and Sox10-Cre at E9.5 to label only migratory neural crest cells (E) Principle component analysis of libraries prepped from sorted wildtype neural crest cells showing samples group together by Cre-driver regardless of age (F) Gating strategy used to isolate RFP+ neural crest cells from E8.5 and E9.5 embryos using FACS sorting. Please click here to view a larger version of this figure.
Figure 2: RNA isolation from E7.5-E9.5 mouse embryos. (A) Representative spectrophotometer trace of an RNA isolation from the youngest stage of embryo that we have applied this protocol. (B) Table showing the representative values of RNA quality obtained from the spectrophotometer. (C) Example of the RNA harvested from sorted neural crest cells from embryos of each age (D) Capillary electrophoresis trace showing good quality RNA. Please click here to view a larger version of this figure.
Figure 3: miRNA libraries before and after size selection. (A) TBE-PAGE gel showing representative small RNA libraries for four samples and controls before and (B) after 150 bp band excision. The arrows represent the 150 bp band to be excised (C) Capillary electrophoresis trace of small RNA library before and (D) after size selection. Please click here to view a larger version of this figure.
Developmental processes can proceed rapidly, and cells are undergoing many sequential specifications such that to capture a comprehensive view of the miRNAs contributing to early fate decisions, more specific staging is needed than the widely used half-day increment. A recent study has performed RNA sequencing from Theiler stage 12 embryos which range from having 3 to 6 somites11. We find that during this period of time, the neural crest cells are specified (3 somites of age), delaminate (4 somites of age), and migrate (5-6 somites of age). We also find that besides the Cre-driver used to lineage trace cell populations, age is the largest source of variation between biological samples, and only somite matched embryos should be considered as replicates. This should also be taken into consideration when comparing transgenic embryos to wildtype controls.
Previous methods to profile miRNAs during early development have used between 10-100 ng of RNA input for small RNA sequencing library preparation and have pooled multiple embryos into one sample13,14. We demonstrate RNA isolation and library preparation from a single E7.5 embryo or from sorted neural crest cells at E8.5 and E9.5 using approximately 100 ng of total RNA as input. When dissociating embryos for sorting, one should take care to watch the dissociation under a microscope and quench the reaction to observe when single cells are obtained. We find that dissociation of the cranial region of E8.5-E9.5 embryos is almost instant with gentle manual pipetting as described in the protocol. For larger tissues and increasingly older embryos, dissociation time may be longer depending on the portion of the embryo being dissociated. For E7.5-E9.5 embryos, clumps of cells are easily visible under the microscope and the dissociation should continue until no more clumps are visible. Single cells are visible in solution if you adjust the focus through the solution in your well anywhere from 5-10x. Previous methods sort cells directly into lysis buffer for RNA sequencing to prep bulk RNA from a low number of cells14. Here we sort directly into RNA extraction lysis solution so that RNA can be isolated before the start of library preparation. Use of mini columns with 11 µL elution volume allowed for a high enough RNA concentration such that a single RNA prep could be split between small RNA and bulk RNA sequencing.
One current limitation of most small RNA sequencing methods is the PCR amplification of converted cDNA. Our method does not overcome this limitation, but we were able to minimize the number of PCR cycles from the 25-maximum recommended down to 16 cycles. This reduction in amplification decreases artificial amplification bias introduced by PCR. Another source of bias is the ligation of adapters, where it is known that specific sequences located at the ends of adapters and miRNAs can ligate together with greater efficiency than other sequences. To avoid this, the adapters used in this protocol have 4 random bases incorporated at the end of each adapter to prevent bias in ligation reactions. Additionally, another common issue is the amount of adapter dimers that form when the RNA input is low. The library preparation kit does include steps to reduce adapter dimer formation such as adapter inactivation and bead cleanups to remove excess adapters after each ligation. We also diluted the 3’ and 5’ 4N adapters by ¼ to reduce the amount of adapter dimer that can form. We found that when not diluted, the 130 bp band intensity increases making it difficult to distinguish from the 150 bp band containing the desired small RNA libraries on a gel.
Another current challenge of preparing sequencing libraries is the accurate quantification of product prior to sequencing. We have found that different methods give varying results on the same library. We suggest that the researchers use multiple methods of quantification to get an accurate estimation of concentration.
This protocol can be widely applied to genetic, developmental studies, or other applications where RNA is being harvested from a low number of cells. This approach simplifies temporal studies by avoiding the pooling of embryos and can easily be applied to both non-sorted and sorted cells.
The authors have nothing to disclose.
This project was supported by Andrew McDonough B+ Foundation and the NIH (R01-HD099252, R01-HD098131. R.J.P. is a CPRIT Scholar in Cancer Research (RR150106) and V Scholar in Cancer Research (V Foundation). The authors would also like to acknowledge the Cytometry and Cell Sorting Core at BCM for providing equipment necessary for this project.
#5 Forceps Fine Science Tools | Fisher Scientific | NC9277114 | |
0.4% Trypan blue | ThermoFisher | 15250061 | |
10 cm Petri dishes | Fisher Scientific | 07-202-011 | |
2-Propanol | Miilapore Sigma | I9516-500ML | |
5 % Criterion TBE Polyacrylamide Gel, 12+2 well, 45 µL | Bio-Rad | 3450047 | |
5200 Fragment Analyzer | Agilent | M5310AA | |
96 well PCR plates high profile semi skirted clear/clear | Bio-Rad | HSS9601 | |
BD FACSAria II | bdbiosciences | ||
Bovine Serum Albumin, fraction V | Fisher Scientific | BP1600100 | |
Dulbecco's Phosphate-Buffered Saline (DPBS) | Gendepot | CA008-300 | |
Eppendorf tubes | Fisher Scientific | 05-408-129 | |
Ethanol Pure, 200 proof anhydrous | Sigma Aldrich | E7023-500ml | |
FC-404-2001 | Illumina | FC-404-2001 | |
HS NGS Fragment kit | Agilent | DNF-474-0500 | |
HS RNA Kit | Agilent | DNF-472-0500 | |
miRNeasy Mini Kit (50) | Qiagen | 217004 | |
NEXTflex Small RNA-Seq Kit v3 (48 barcodes) | Fisher Scientific | NC1289113 | |
NextSeq 500/550 Mid Output Kit v2 (150 cycles) | Illumina | FC-404-2001 | |
NGS Fragment kit | Agilent | DNF-473-1000 | |
Papain | Worthington | LK003176 | resuspend in 5 mL of Ca/Mg free PBS for a final concentration of 27 U/mL |
SYBR Gold nucleic acid gel stain | ThermoFisher | S11494 | |
Trizol LS | ThermoFisher | 10296028 | |
UVP Benchtop UV Transilluminators: Single UV | Fisher Scientific | UVP95044701 | |
Vannas Scissors Straight Fine Science Tools #91500-09 | Fisher Scientific | NC9609583 |