A protocol is described that uses laser microdissection to isolate individual nematode tissues for RNA-sequencing. The protocol does not require species-specific genetic toolkits, allowing gene expression profiles to be compared between different species at the level of single-tissue samples.
Single-cell methodologies have revolutionized the analysis of the transcriptomes of specific cell types. However, they often require species-specific genetic "toolkits," such as promoters driving tissue-specific expression of fluorescent proteins. Further, protocols that disrupt tissues to isolate individual cells remove cells from their native environment (e.g., signaling from neighbors) and may result in stress responses or other differences from native gene expression states. In the present protocol, laser microdissection (LMD) is optimized to isolate individual nematode tail tips for the study of gene expression during male tail tip morphogenesis.
LMD allows the isolation of a portion of the animal without the need for cellular disruption or species-specific toolkits and is thus applicable to any species. Subsequently, single-cell RNA-seq library preparation protocols such as CEL-Seq2 can be applied to LMD-isolated single tissues and analyzed using standard pipelines, given that a well-annotated genome or transcriptome is available for the species. Such data can be used to establish how conserved or different the transcriptomes are that underlie the development of that tissue in different species.
Limitations include the ability to cut out the tissue of interest and the sample size. A power analysis shows that as few as 70 tail tips per condition are required for 80% power. Tight synchronization of development is needed to obtain this number of animals at the same developmental stage. Thus, a method to synchronize animals at 1 h intervals is also described.
Nematodes—particularly the rhabditid nematodes related to the model system Caenorhabditis elegans—are a wonderful group of animals for evolutionary developmental biology (EDB) for many reasons1,2. Advantages include their small number of cells, defined and consistent cell lineages, transparency, and ease of culture and husbandry. There are also many resources available, including high-quality genomes for multiple species, and for C. elegans, extensive molecular genetic tools and knowledge about development, genetics, anatomy, and physiology3,4,5,6.
As with many other organisms, the ability to characterize transcriptome dynamics in single tissues or single cells has revolutionized the analysis of development in C. elegans7,8,9,10. Being able to compare single-cell transcriptomes across nematodes would similarly transform EDB using these organisms. For example, such comparisons would provide insight into how gene regulatory networks have evolved for characters (traits) that have been conserved, for characters that have diverged, or for characters that evolved independently.
However, isolating particular tissues or cells from nematodes is one of the big challenges. For many organisms, single cells can be dissociated from tissues and harvested in an unbiased way or can be labeled with tissue-specific expression of a fluorescent protein and sorted by fluorescence-activated cell sorting (FACS)11. In C. elegans, high-throughput (HTP) isolation of cells has been limited mostly to embryos because the tough outer cuticle (and hydrostatic skeleton) has hampered cell isolation from larvae and adults. To get around this challenge, some methods have employed genetic tools in whole C. elegans worms, such as tissue-specific mRNA-tagging12, and differential expression comparisons between wild-type and mutants affecting a cell type13. More recent methods have overcome the challenge by dissolving the cuticle to isolate nuclei14 or entire cells8,9,15. Cell isolation and cell culture have the obvious disadvantages, however, that cells are removed from their natural developmental or anatomical context—e.g., away from cell-cell signaling and contact with the extracellular matrix—which are expected to impact the gene expression profile15. Moreover, the genetic tools and tissue-specific markers are species-specific (i.e., they can only be used in C. elegans).
LMD provides an alternative method for isolating tissues without disrupting the natural context of cells. Significantly for EDB, LMD also allows transcriptomes from homologous tissues of different species to be compared without the need for species-specific genetic toolkits if genome or whole transcriptome sequences of these species are available. LMD involves targeting tissues by direct microscopical observation and using a laser microbeam—integrated into the microscope's optics—to cut out and harvest (capture) the tissue of interest16. Limitations of LMD are that it is not conducive to very HTP approaches (although the transcription profiles for tail tips, as described in this protocol, were robust with ~70 samples), certain samples might be difficult to dissect out, and cuts are limited to the precision of the laser and what can be visualized in the microscope.
The purpose of the present protocol is to describe how LMD, followed by single-tissue RNA-Seq, can be used to obtain stage- and tissue-specific transcriptome data from nematodes. Specifically, it demonstrates LMD for isolating tail tips from fourth-stage larvae (L4) of C. elegans. However, this method can be adapted to other tissues and, of course, different species.
In C. elegans, there are 4 cells that make the tail tip in both males and hermaphrodites. During the L4 stage in males—but not in hermaphrodites—the tail tip cells change their shape and migrate anteriorly and inwardly. This process also occurs in some but not all other rhabditid nematode species. Therefore, the tail tip is a good model for the evolution of sexual dimorphic morphogenesis. Because of its position, the tail tip is also easy to isolate by LMD.
To obtain transcriptome profiles from tail tips, the present protocol uses CEL-Seq2, an RNA-seq method developed for single cells17,18. This method has several advantages for LMD-derived tissues. CEL-Seq2 is highly sensitive and efficient, using unique molecular identifiers (UMIs) to allow straightforward quantification of mRNA reads, in vitro transcription to ensure linear amplification, and barcoding that allows multiplexing of individual tissue samples. The only limitation of CEL-Seq2 is that recovered reads are biased to the 3' end of mRNAs, and most isoforms thus cannot be distinguished.
1. Worm synchronization
NOTE: Two methods are described below to synchronize the development of C. elegans and other rhabditid species.
2. Collecting L4 males and hermaphrodites and fixation
3. Laser microdissection
NOTE: From here on, use RNase-free reagents and consumables; use filter tips.
4. Single-tail RNA sequencing with CEL-Seq2
NOTE: For full details about the CEL-Seq2 protocol, see Yanai and Hashimshony18.
Following laser capture microdissection, individual tail tips of males and hermaphrodites at 4 time points (L3 22 h after hatch; L4 24, 26, and 28 h after hatch) were prepared for RNA sequencing using the CEL-Seq2 protocol. CEL-Seq2 primers contain unique barcodes that enable sequencing reads from a particular sample (in this case an individual tail tip) to be identified bioinformatically. Sequencing data were generated with this method for a total of 557 tail tips (266 hermaphrodites and 291 males across 4 developmental time points, 59-78 tails per sex and time point). CEL-Seq2 barcodes were recovered for 97% (i.e., 543) of these tail tips (Supplemental Table S2). For most libraries, the recovery rate was 99-100%; however, it was 88% for one male time point. It is worth noting that about half of the male tail tips from the 22, 24, and 28 h time points were stored at -80 °C for ~4 months due to COVID-19-related delays. This demonstrates that, while it is ideal to prepare sequencing libraries shortly after sampling, it is possible to store dissected samples for a longer time before library preparation.
The CEL-Seq2 primers also add a UMI to each mRNA transcript. This enables PCR duplicate removal and precise quantification of gene expression in the sample. The number of UMIs varied dramatically across tail tips (Figure 4; male mean = 92,560; male min. = 155; male max. = 1,183,998; hermaphrodites mean = 67,597; hermaphrodites min. = 132; hermaphrodites max. = 630,427). For UMI counts per tail tip, see Supplemental Table S3. Due to the low amount of input RNA for single-cell library preparation, single-cell sequencing data are known to have a large amount of technical noise. Hence, it is recommended to filter samples that have very low or very high UMI counts before analysis24.
The R package powsimR25 was used to assess the statistical power and sample size requirements for reliably detecting differentially expressed (DE) genes in single-cell or bulk RNA-seq experiments. Parameters for the simulations were based on a sequencing dataset of 70 individual male tail tips (at the 24 h time point) obtained with the method described here. Expected log-fold changes were based on results from a separate RNA-seq experiment that pooled 80-100 tail tips. The simulations determined that the single-tail-tip data have sufficient power (True Positive Rate = TPR) to detect DE genes, except for genes that have a very low mean expression value (top of Figure 5; dashed line represents 80% TPR). Adding more simulated tail tips per time point increased the power somewhat for lowly expressed genes. A similar pattern is seen for the False Discovery Rate (FDR). FDR is high (>0.10) for the lowly expressed genes; however, for more highly expressed genes, it falls at or below the nominal 0.10 cutoff (dashed line for FDR in the bottom of Figure 5). In summary, increasing the number of tails sampled per time point above 70 would do little to lower the FDR or increase power. However, 70 tail tips provide a much lower FDR and stronger power than 30 tail tips.
Figure 1: Procedure overview for synchronization of Caenorhabditis elegans with the hatch-off method and laser microdissection of tail tips. Abbreviations: L1-L4 = larval stages 1 to 4; PEN = polyethylene naphthalate; LMD = laser microdissection. Please click here to view a larger version of this figure.
Figure 2: Appearance of C. elegans L3 hermaphrodites and males under a dissection microscope. Hermaphrodites (A, B) and males (C, D) at 21-23 h after hatch can be distinguished under a dissection microscope (~50x magnification) by the morphology of their tails (arrows). The tail of hermaphrodites is narrow, while that of males is swollen and appears clear. Scale bars = 0.1 mm. Please click here to view a larger version of this figure.
Figure 3: Appearance of the PEN membrane slide structure and worm tail. Focus is correct for dissection of the tissue viewed with the 20x (A) and 40x (B) lens at the microscope. (C) Dissected tail and partially cut out PEN membrane. After closing the gap in the cut, the membrane piece will drop into the tube cap below the slide. (D) Tube cap with a PEN membrane section containing a dissected tail tip. Scale bars = 0.1 mm (A-C), 1 mm (D). Please click here to view a larger version of this figure.
Figure 4: Natural log-transformed UMI counts per individual tail tip for different time points and sexes. RNA from individual tails was prepared for sequencing using the CEL-Seq2 method; 557 tails were sequenced in total, with 59-78 tails per sex and time point. Extremely low and high UMI outliers would be removed from the data before analysis. Abbreviation: UMI = unique molecular identifier. Please click here to view a larger version of this figure.
Figure 5: Results of an a posteriori power analysis using simulations with powsimR. The powsimR software determines the number of independent samples required to detect DE genes at various expression levels. Genes are binned by mean expression transformed as the natural log of UMI counts. (A) Power (TPR) to detect DE genes between two conditions (here, male vs hermaphrodite) for four different simulations (different colored graphs) incorporating different sample sizes (numbers of individual tail-tips) per condition. Dashed line indicates 80% TPR. (B) FDR in the same four simulations as in (A), dashed line indicating 10% FDR. The graphs show that a sample size of 70 tail tips (green) per condition is sufficient for detecting DE genes, except for genes with very low expression levels. That is, the power and false discovery rate for such genes cannot be greatly improved by increasing the sample size beyond 70. Abbreviations: DE = differentially expressed; UMI = unique molecular identifier; TPR = true positive rate; FDR = false discovery rate. Please click here to view a larger version of this figure.
Supplemental Table S1: Sequences of primers used in the CEL-Seq2 protocol. Please click here to download this File.
Supplemental Table S2: Sampled versus recovered individual tail tips. CEL-Seq2 primers contain unique barcodes that enable identification of sequencing reads from each sample to be identified bioinformatically. Reads were recovered from 97% of the tail tip samples prepared for sequencing. Please click here to download this File.
Supplemental Table S3: UMI counts of all samples with recovered barcodes, as noted in Supplemental Table S2. Abbreviation: UMI = unique molecular identifier. Please click here to download this File.
Critical steps of the method
If performed correctly, the method described here will obtain robust RNA profiles with a relatively small number of laser-dissected samples (70 tail tips in this example). However, for samples from developing animals, tight synchronization is critical to reducing the variability between samples. For this reason, the protocol recommends the hatch-off method for worm-synchronization. Here, the researcher can determine and precisely control the age difference between individuals (1 h in the present protocol). In addition, the hatch-off method is applicable to any species, even if the embryos are sensitive to bleach, L1 do not arrest, or recovery from L1 arrest is variable. For a successful synchronization by hatch-off, the washing steps are crucial: all adults and larvae must be removed at the beginning of the hatching period, and no embryos should be washed off along with the newly hatched L1 at the end of the hatching period. This only succeeds if the agar surface of the plate is undamaged by cracks, holes, or bubbles, the bacterial lawn on the plate is fresh and not too thick, and the liquid is added and agitated only very gently.
If data are to be obtained separately for males and hermaphrodites/females, reliable identification of the sexes is also important. Distinguishing L3 larvae by sex (see Figure 2) requires experience. It is recommended to practice picking L3 males and hermaphrodites/females and check the success rate after the animals have developed into adults and the sexes are easily distinguished. After single-tissue RNA-Seq, the outliers can also be identified by principal component analysis and removed, if necessary.
For successful recovery of laser-cut samples, it is important to reduce static electricity as much as possible. Charged PEN-membrane pieces often do not drop into the tube cap but stick to the slide or any other part of the microscope. One remedy is raising the humidity in the room and specifically around the microscope by placing a small humidifier next to the stage. Additionally, the membrane slides can be treated with UV light. To do this, incubate slides in a UV-C (254 nm) crosslink chamber and deliver at least 1 joule of energy, or expose the slides to the UV light in a laminar air flow bench for 30 min.
Since the goal of the protocol is RNA-Seq, keeping an RNase-free working environment is critical. Beginning with the fixation solution, reagents, containers, and consumables should be RNase-free, the work surface should be decontaminated, and the researchers should wear clean gloves. The dissected samples should be frozen as soon as possible and kept at -70 °C until further processing. It is also recommended to use low-retention tubes and tips for the CEL-Seq2 part of the protocol.
The present article provides only a basic outline of the CEL-Seq2 protocol, which was previously published by its developers with helpful notes and tips17,18. It is recommended that these publications are consulted before using the CEL-Seq2 method.
The LMD-RNA-Seq data can be validated by single-molecule-RNA fluorescent in-situ hybridization (smRNA FISH)26,27,28. smRNA FISH has been extensively used in C. elegans and is amenable to other nematode species, different from immunostaining with existing antibodies (which may not crossreact) or the introduction of transcriptional reporters through transgenesis. The latter works well in C. elegans and some related Caenorhabditis species29, but transgenesis can be more challenging in other nematode species30,31.
Limitations of the method
The method described here works very well for collecting tail tips, a thin tissue at the end of a worm. Dissecting tissues in the thicker middle of older larvae or adults is more challenging. The software of the instrument used here includes a setting for multiple cuts placed at subsequently deeper levels in the tissue. This setting can be used for cutting thicker areas of the animal. Because the worms need to be fixed before dissection, structural details are difficult to see, which prevents the precise dissection of specific small structures. As mentioned above, LMD-RNA-Seq is not an HTP method. However, 50-70 samples can be dissected in one afternoon.
Significance of the method with respect to existing/alternative methods
LMD-RNA-Seq can be used in any species even if no transgenic tools are available. Other methods rely on FACS sorting of fluorescently labeled cells8,9,32 or isolation of labeled nuclei33,34 and thus require transgenic animals. Methods that dissociate and isolate cells in postembryonic C. elegans tend to miss the tissues at the two ends of the worm (Dylan Rahe, personal communication). These caveats are overcome by combining single-cell RNA-Seq with cryosectioning of entire worms (RNA tomography)35. This method was used to compare spatial gene expression between C. elegans and another rhabditid nematode, Pristionchus pacificus36. Alternatively, one can experiment with formalin-fixed paraffin-embedded (FFPE) worms. Such material has been successfully used for RNA-Seq following LMD of mammalian tissue samples37. However, RNA tomography and LMD of FFPE worms are limited to the analysis of only a handful of animals. They are, therefore, not as well suited for the study of dynamic gene expression in developing tissues as LMD-RNA-Seq.
The authors have nothing to disclose.
This work was funded by NIH (R01GM141395) and NSF (1656736) grants to DF and NIH fellowship (F32GM136170) to AW. Figure 1 was created with the help of BioRender.com.
0.5 µM PEN membrane glass slides RNase free | Leica | 11600288 | for LMD |
500 µL PCR tubes (nuclease-free) | Axygen | 732-0675 | to cut the tail tips into |
Compound microscope with 40x objective and DIC | any | to check age of worms | |
Desktop humidifier | any | ||
Dissection microscope with transmitted light base | any | for all worm work | |
glass pasteur pipets | any | handle of worm pick | |
glass slides and coverslips | any | to check age of worms | |
LMD6 microdissection system | Leica | multiple | to cut tail tips |
LoBind tubes 0.5 mL | Eppendorf | 22431005 | |
M9 Buffer | Recipe in WormBook | ||
Methanol 99.8% | Sigma | 322415 | to fix worms |
NGM growth medium | US Biological | N1000 | Buffers and salts need to be added: Recipe in WormBook |
P10 pipette variablle volume | e.g. Gilson | ||
P1000 pipette variable volume | e.g. Gilson | ||
P2 pipette variable volume | e.g. Gilson | ||
Pipette tips 1,000 µL | any | ||
Pipette tips 1-10 µL filtered | any | ||
platinum iridium wire | Tritech | PT-9010 | to make worm pick |
sterile and nuclease-free 1 mL centrfuge tubes | any | ||
Tween 20 | Sigma | P9416 | Add a very small amount to M9 buffer to prevent worms from sticking to the pipet tips |
vented 6 mm plastic Petri dishes | any | ||
For CEL-Seq2 | |||
4200 TapeStation System with reagents for high-sensitivity RNA and DNA detection | Aligent | automated electrophoresis system | |
AMPure XP beads | Beckman Coulter | A63880 | DNA cleanup beads |
Bead binding buffer 20% PEG8000, 2.5 M NaCl | |||
CEL-Seq2 primers (see Table S1) | Sigma Genosys Mastercycler Nexus GX2 Eppendorf | 6335000020 | Thermal cycler with programmable lid and block for 200 µl tubes. |
DNA Polymerase I (E. coli) | Invitrogen | 18052-025 | |
dNTP mix 10 mM | any | ||
E. coli DNA ligase | Invitrogen | 18052-019 | |
Ethanol | |||
ExoSAP-IT For PCR Product Clean-Up | Affymetrix | 78200 | exonuclease solution |
MEGAscript T7 Transcription Kit | Ambion | AM1334 | For step 4.6.1 |
Nuclease-free water | any | ||
Phusion High-Fidelity PCR Master Mix with HF Buffer | NEB | M0531 | PCR mix step 4.9.7 |
random hexamer RT primer GCCTTGGCACCCGAGAATTCCA NNNNNN |
IDT | a primer with 6 nucleotides that are random | |
RNA Fragmentation buffer | NEB | E6150S | |
RNA Fragmentation stop buffer | NEB | E6150S | |
RNA PCR Index Primers (RPI1–RPI48) | Illumina, NEB, or IDT | RPIX in protocol step 4.9.7, sequences available from Illumina | |
RNAClean XP beads | Beckman Coulter | A63987 | |
RNase AWAY Surface Decontaminant | Thermo Scientific | 7000TS1 | or any other similar product |
RNaseH (E. coli) | Invitrogen | 18021-071 | |
RNaseOUT Recombinant Ribonuclease Inhibitor | Invitrogen | 10777-019 | |
Second strand buffer | Invitrogen | 10812-014 | |
Superscripit II | Invitrogen | 18064-014 | reverse transcriptase |