Optimized toolkit for the manipulation of immortalized axolotl fibroblasts

Abstract

The axolotl salamander model has broad utility for regeneration studies, but this model is limited by a lack of efficient cell-culture-based tools. The Axolotl Limb-1 (AL-1) fibroblast line, the only available immortalized axolotl cell line, was first published over 20 years ago, but many established molecular biology techniques, such as lipofectamine transfection, CRISPR-Cas9 mutagenesis, and antibiotic selection, work poorly or remain untested in AL-1 cells. Innovating technologies to manipulate AL-1 cells in culture and study their behavior following transplantation into the axolotl will complement in-vivo studies, decrease the number of animals used, and enable the faster, more streamlined investigation of regenerative biology questions. Here, we establish transfection, mutagenesis, antibiotic selection, and in-vivo transplantation techniques in axolotl AL-1 cells. These techniques will enable efficient culture with AL-1 cells and guide future tool development for the culture and manipulation of other salamander cell lines.

Graphical Abstract

Introduction

Axolotl AL-1 cells were first isolated in the Gardiner laboratory in 2000 [1]. Amputated limbs were placed on Petri dishes in L-15 media, and AL-1 cells spontaneously migrated onto the plate, adhered, and proliferated [1]. This immortalization event appears to have been spontaneous, and its molecular underpinnings remain uncharacterized. In culture, AL-1 cells resemble and behave like dermal fibroblasts [2]. This is useful for regenerative studies, as dermal fibroblasts are the main source of cells for the regenerative limb blastema, a mass of multipotent progenitor cells which forms at the amputation site and generates all of the mesenchymal structures of the regenerated limb [3–5]. Since their initial isolation, AL-1 cells have been used, in combination with radiolabeling, western blots, and luciferase assays, to characterize TGF-beta signaling in axolotl cells, the unique biology of salamander p53 pathway [2,6,7], as well as to pioneer retroviral transfection with MMLV [8] and baculovirus [9]. AL-1 cells have also been used for reporter validation, overexpression tests, and comparative cell culture [6–23]. More recently, experimentally-provoked senescent AL-1 cells have been transplanted into regenerative limbs, demonstrating that senescent cells can promote regeneration [22]. Cumulatively, this body of work demonstrates the utility AL-1 cells for development of molecular tools in the axolotl, the examination of salamander cell biology and signaling, and even in-vivo regenerative experiments in axolotls.

AL-1 cells are often used to validate tools before use in whole organism studies, but they can also be used to directly test regenerative hypotheses. As AL-1 cells resemble dermal fibroblasts [2], several aspects of regenerative biology such as proliferation, migration, and multipotency could be explored in-vitro, yielding results more quickly than resource intensive organismal experiments. Importantly, AL-1 cells can survive implantation and contribute to the regenerative limb, allowing researchers to evaluate the phenotype of manipulated AL-1 cells in-vivo. In combination with mutagenesis, AL-1 implantation studies could bypass embryonic lethal effects seen when making whole animal mutants and could significantly accelerate the large-scale analysis of genes by directed mutagenesis. An expanded repertoire of robust molecular techniques in AL-1 will also make the axolotl system more accessible to cell culture labs, where AL-1 would be useful for exploring the key differences between mammalian cells, salamander cells, and other vertebrate cells that may underlie regenerative outcomes [6,7,15]. Of course, AL-1 cells will never replace organismal axolotl experiments, but AL-1 cells promise to be a faster, more scalable, and higher throughput complement to these studies.

Despite this promise, practical obstacles limit the use of AL-1 cells: (1) AL-1 cells grow slowly when compared to commonly used mammalian cell lines, taking more than 5 days to grow from 50% to 100% confluence, and (2) do not tolerate being split at ratios greater than 1:2 [23]. Because of this, experiments in AL-1 typically take much longer than they would in mammalian cells. AL-1 cells are also much larger than mammalian cells; we have measured the average, detached AL-1 cell to be 32 μm in diameter, demanding more culture surface area and media than mammalian cells. We have found that a T75 flask holds up to ~2.5 million AL-1 cells, while the same flask can hold approximately 8.6 million HeLa cells. We accommodate the large size and relatively slow growth rate by continuously maintaining large, active AL-1 cultures. This runs counter to the standard cell culture recommendation to keep cells frozen and only expand them immediately before use and is resource intensive. Despite these limitations AL-1 cells can grow at room temperature in the open air and have an impressive tolerance for over-confluence. Importantly, the limitations of AL-1 cell culture remain preferable in comparison to the time, reagents, animal, and human labor cost of organismal axolotl research.

A critical obstacle to the more widespread use of AL-1 cells is the lack of efficient, available molecular techniques. Electroporation, lipofection, and nucleofection have all been reported in AL-1 cells but with very low efficiency and/or viability [23]. Lipofectamine is reported as having an efficiency of 5%, while the most efficient method, electroporation, shows 50% efficiency and 50% viability for an overall efficiency of only 25% [23]. Moreover, there are no published examples of CRISPR mutagenesis, nor are there any published modified AL-1 cell lines. In this study, we demonstrate novel and optimized methods for AL-1 transfection, CRISPR mutagenesis, the generation and establishment of modified AL-1 cell lines, as well as in-vivo experimentation with AL-1 through transplantation, dissociation, and Fluorescence-activated Cell Sorting (FACS). We also generated several unique AL-1 lines, including an mCherry AL-1 cell line, a Cas9-expressing cell line, and a line of limb-recovered AL-1 which has been transplanted to, and recovered from an axolotl limb to select for better engraftment.

Many of the obstacles in working with AL-1 transfection and mutagenesis are shared by axolotl cells more generally, and we believe the tools described here will be of wide use for the field. The presented protocols will inform enhanced primary cell work, immortalized salamander cell work, and fully in-vivo transfection experiments.

Materials and Methods

1. Recommended JetOPTIMUS^® protocol

Our jetOPTIMUS^® protocol was optimized utilizing a CMV-GFP-Puro plasmid (Addgene 45561), purified with the QIAGEN HI-Speed Maxiprep kit (QIAGEN 12165), and subsequently cleaned via phenol-chloroform extraction with pH8 phenol (Sigma-Aldrich P2069). Prior to transfection, one 100% confluent T75 (VWR 10062–860) was seeded across four 24 well plates and allowed to reattach for 24 hours, resulting in ~50–60% confluence in each well. At this point, media was removed and replaced with Opti-MEM (Thermo 11058021) diluted in ultrapure water to 70% to match amphibian cell osmolarity (70% OMEM), to starve the cells for an additional 24 hours at room temperature and without supplemental CO2. This starvation step is optional, but we found it to consistently improve transfection efficiency (see Results). On the third day, the OMEM was removed and replaced with complete AL-1 media. AL-1 complete media consists of: L-15 (Sigma Aldrich L5520) diluted to 60%, Fetal Bovine Serum (FBS) (VWR-1500–500) diluted to 5%, 100x Penn/Strep (Thermo15-140-122) diluted to 1x, 100x Antibiotic-Antimycotic (Thermo 15240062) diluted to 1x, 100x L-Glutamine (Thermo25-030-081) diluted to 1x, and 100x Insulin-Transferrin-Selenium (Thermo 41400045) diluted to 1x [2,23]. This results in final concentrations of 200U/mL penicillin, 200μg/mL streptomycin, 250ng/mL amphotericin B, 2mM L-Glutamine, 10μg/mL Insulin, 5.5μg/mL transferrin, and 6.7ng/mL sodium selenite.

In accordance with manufacturer’s instructions, plasmid was diluted in jetOPTIMUS^® buffer at a ratio of 0.1μg plasmid per 10μL buffer and briefly vortexed, before the addition of 1μL jetOPTIMUS^® reagent per 1μg plasmid. This mixture was then vortexed and incubated at room temperature for at least 10 minutes, before being added dripped to the wells. The amount of plasmid added reflects the size and number of wells transfected, in this paper we used 0.5μg plasmid per well in 24-well plates, 1μg plasmid per well in 12-well plates, 2μg plasmid per well in 6-well plates, and 10μg plasmid for entire T-75’s. This corresponds roughly to 0.2μg/cm culture area. After a 24-hour transfection period, transfection media was removed from the wells and replaced with fresh complete AL-1 media. Both the starvation period and transfection period were varied while optimizing this protocol.

2. Quantification of AL-1 fluorescence using the EVOS

For JetOPTIMUS^® optimizations, the evaluation of GFP CRISPR, and the quantification of puromycin selection, we imaged GFP fluorescence with an automated EVOS M7000 protocol. Cells were first stained with Hoechst 33342 (ApeBio A3472): a stock concentration of 12.5mg/mL was diluted 1:10,000 in AL-1 complete media [23] and added to the cells for 30 minutes. The cells were then briefly washed with 70% PBS (APBS) and returned to unmodified complete AL-1 media. To make APBS, Phosphate Buffered Saline (PBS) (CORNING 21–040-CV) is diluted with ultrapure water to 70% to reflect ampibian osmolarity [23]. After resting for several hours, we scanned the plates with the EVOS. We then segmented the nuclei and recorded fluorescence intensity with a Fiji macro (FluorescenceQuantification.ijm included in supplemental files) and consolidated the data with a Matlab script (RedandGreenQuantificationBasic.m included in supplemental files). Data was then interpreted in both Matlab and Excel.

3. Electroporation

Our Electroporation protocol was derived from Denis et al. [23], but reoptimized for our NepaGene Nepa21. 100% confluent T75’s were briefly washed with 5mls 1xPBS (CORNING 21–040-CV), trypsinized with 1ml 0.05% trypsin (Cytiva 25300054) for 2 minutes at 37°C, neutralized with 4mls complete media, and spun down for 5 mins at 200xg and 4°C, before being resuspended in APBS. Cells were resuspended in a volume of 600μL per T75 (~350,000 cells/μL). Cells were then pipetted into cuvettes, with 100μL of resuspended cells being added to 2mm gap cuvettes (Bulldog Bio 12358–346) or 250uL cells being added to 4mm gap cuvettes (Bulldog Bio 12358–347) and electroporated (see Table 3). After electroporation, 1ml of 10% FBS complete media (made as previously described in the jetOPTIMUS^® section, but with 10% FBS instead of 5%), was immediately added to the cuvette. The entire contents of the cuvette were then transferred with a glass pipette to either a single well of a 12-well plate (Celltreat 229111) in the case of a 2mm cuvette, or a single well of a 6 well plate (Celltreat 229106) in the case of a 4mm cuvette. For 2mm cuvettes, an additional 1ml 10% FBS AL-1 complete media was added to each transfected well, bringing the final volume to 2.1mls (100μL cells in PBS + 1mL 10% FBS complete media added to cuvette, + 1mL 10% FBS complete media added to well). For 4mm cuvettes, an additional 2mls 10% FBS AL-1 complete media were added to each transfected well, bringing the final volume to 3.25mls (250μL cells in PBS + 1mL 10% FBS AL-1 complete media added to cuvette, + 2mL 10% FBS AL-1 complete media added to well. Because the CRISPR components add up to a slightly larger volume than the transfection volume, 102μL vs 100μL in 2mm cuvettes, or 255μL vs 250μL in 4mm cuvettes, we were often left with a small number of untransfected cells which we could use to seed a spare well or T25 as an untransfected control. Cells were allowed to reattach for up to a week following electroporation before feeding with normal 5% complete AL-1 media, although in some cases the cells attached rapidly and quickly reached high confluence; these were fed as early as 24 hours after electroporation.

Table 3:

Recommended Parameters for Nepa21 Electroporator

	Poring Pulse
Cuvette Size	Voltage (V)	Pulse Length (ms)	Pulse Interval (ms)	Number of Pulses	Decay Rate %	Polarity	Transfer Pulse
2mm	100	10	50	3	0	+	None
4mm	200	10	50	3	0	+	None

4. Tol2 protocol

Tol2 integration plasmids were generated through a combination of Infusion-HD (Takara Bio) and Gateway cloning (Thermo 12538120). Tol2 Inverted terminal repeats were provided by pDestTol2pA [24], Cas9 and mCherry were obtained from pME-cas9-2AmCherry (Addgene 107592), the CMV promoter came from p5E-CMV min (Addgene 80802), and p3E-poly A [24] was used for the 3’ entry vector. Snapgene files for each plasmid are included in supplemental materials. Integration plasmids were maxi-prepped (Qiagen12663 or Zymo D4202 or Takara 740424.50) and introduced to AL-1 cells via jetOPTIMUS^® (as detailed protocol above), alongside the Tol2-transposase expressing PT2K plasmid (Addgene 114725). In the case of the mCherry-TVA cell line, plasmids were added simultaneously: 250ng of each plasmid were added to each transfected well of a 24-well plate (Corning 3524), totaling to 0.5μg per well as recommended by manufacturer. Cells were expanded for 6 passages before sorting with FACS (as described below) and have been propagated to date, without loss of fluorescence.

5. FACS

We counted and sorted fluorescent AL-1 cells using the Harvard Bauer Core’s Becton Dickinson AriaII. We used a 100μm nozzle, and a 2.0 ND filter for forward scatter. Prior to sorting, cultured cells were trypsinized, neutralized, spun down at 200xg for 5mins at 4°C, washed in APBS, spun down again and resuspended in AL-1 FACS Buffer (70% PBS, 2%FBS) (more details in supplemental step-by-step protocol). Limb sample preparation for FACS is detailed in the dissociation section below. Calcein blue (Invitrogen C1429) was dissolved at 1mg/mL in EDTA and diluted 1:500 to evaluate viability. Cells were sequentially gated. First with Forward Scatter area vs side scatter area to isolate cells from smaller debris, followed by forward scatter area vs forward scatter area to isolate single cells from doublets. Live cells were gated using the appropriate viability dye channel before being evaluated for experimental dyes and fluorophores. Gates for viability dye were calibrated every sort using dead control samples. Gates for experimental dyes and fluorophores were also calibrated every sort using appropriate negative control samples. Cytometry data was initially collected with BD FACSDiva software and reanalyzed with FlowJo_v10.10.0.

6. Puromycin selection

AL-1 cells in a single T75 were transfected with the CMV-GFP-Puro plasmid (Addgene 45561), which expresses Puromycin N-acetyl transferase, capable of imparting puromycin resistance. We transfected with jetOPTIMUS^® (polyPlus) according to the manufacturer’s instructions, replacing the transfection media with fresh media after 24 hours; however, we did not include a 24-hour OMEM starvation step before transfection as we intentionally sought a lower transfection efficiency. 7 days after transfection, the cells were split, and a subset was plated onto a 24 well plate. Cells were subjected to multiple concentrations of puromycin dihydrochloride (Santa Cruz CAS 58-58-2) in complete AL-1 media. Cells were scanned daily for 16 days with the EVOS M7000, and the number of GFP+ and GFP− cells were quantified as previously described in “Quantification of AL-1 fluorescence using the EVOS”. Media was changed immediately before scanning; consequently, both media and antibiotics were refreshed frequently.

7. AL-1 implant via injection

Cells were trypsinized with 0.05% trypsin (Cytiva 25300054), Trypsin was neutralized with complete AL-1 media, cells were spun down at 200xg and resuspended in APBS (more details in the supplemental, step-by-step protocol). Cells were counted at this step with a hemacytometer. In cases where DiI staining was used, cells were resuspended in 1mL APBS, live-stained with 10 μL 1mg/mL DiI (TargetMol T15129) dissolved in 100% ethanol, and incubated for 15 minutes, before being spun down at 200xg and washed twice with APBS.

Cells were resuspended in APBS for injection. The volume of this final injection suspension was adjusted according to the experimental condition, but we found injecting 100μL of AL-1 cells resuspended at 45,000 cells/μL provided the best recovery of AL-1 from the regenerative stump and the regenerative blastema. This approximates to about three confluent T75 flasks (or 4,500,000 cells) per injection. Higher concentrations of cells were difficult to inject as they frequently aggregated. For most of our injections, we injected the limbs of ~15 cm Leucistic animals. Both sexes were used, and animals ranged in age from 8 to 16 months. For these injections, resuspended cells were loaded into 31-gauge insulin syringes (BD 328419) and injected into limbs 100μL at a time. Larger volumes were found to leak from the injection site and sometimes into the body cavity. The limbs of larger (23cm+) animals, however, could comfortably accommodate at least 400μL injections. Smaller volumes were difficult to measure with the insulin syringe. We also tested injections on smaller ~7cm animals. Finding that the insulin syringe was too large for these animals, we instead injected them using glass capillaries (World Precision Instruments T100F-4) and a pico injector (Warner Instruments PLI-90A). We pulled these needles using a Sutter needle puller (Sutter Instruments P-97). While quantifying a consistent volume with the capillary method is challenging, we found the limbs of these smaller animals could accommodate 5–10μL of resuspended cells.

Before injection and amputation, animals were anesthetized in a solution of 0.1% tricaine (Syndel MS 222). When injecting, we aimed for the region within the stylopod, between the muscle and the bone, as this produced the most consistent results. Amputations were typically performed 5 days after injection to allow the wound to heal and allow the implanted cells to settle. When amputating, we attempted to cut through the distal end of the injection site to guarantee there were always AL-1 cells close to or within the amputation plane. After both injections and amputations, animals were placed in 0.5% sulfamerazine for 24 hours to prevent infection during initial wound healing and recovery. All animals were housed separately and closely monitored throughout the experiment.

8. Limb Harvest and Dissociation

Our dissociation protocol was modified from Gerber et al. [25]. All dissociation experiments in this study were performed on limbs harvested 21 days past amputation when the blastema is particularly prominent [5]. Animals were anesthetized with 0.1% tricaine. Limbs were briefly rinsed in distilled water to reduce red blood cell contribution [26] and left dry on Petri dishes on ice while amputations continued. In many, but not all, cases, we euthanized the animals after limb harvest. Once all amputations were complete, blastemas were surgically separated from the proximal stump. For this dissection, the boundary of the blastema was defined by the distal edge of the bone in the amputated limb. Blastemas, stumps, and intact limbs were subsequently pooled by condition and minced together on Petri dishes, first with surgical scissors (F.S.R 14060–09) and subsequently with a razor (VWR 55411–050), until they were minced into ~1mm diameter fragments.

Limbs, blastemas, and stumps are complex, heterogenous samples comprising multiple distinct tissue types, each requiring different amounts of enzyme and mechanical stress in order to fully dissociate. The digestion conditions required to dissociate the most stubborn tissue types risk damaging the more easily liberated and delicate cells. To address this challenge, after mincing, we subjected the samples to three sequential 10-minute digests with escalating liberase concentrations, straining the samples between the digests in order to separate dissociated cells from undissociated tissue, and spare them from further processing. Similar methods have been used to dissociate other stubborn tissue types [27,28], and we found this method consistently dissociated almost all limb material and yielded more live cells than our initial protocol which, used a constant liberase concentration [25].

The initial lysis buffer contained 0.1mg/mL (1x) concentration of liberase (Sigma Aldrich 5401127001) in APBS with 0.5U/μL DNAse I (Thermo 18047019). 350μL of this initial lysis buffer per limb was added to minced samples and used to wash/transfer these samples from the petri dish to 50mL conical tubes. Samples were then pipetted up and down for 1 minute with a p1000 pippette, with the tip intentionally widened using a razor so that larger clumps would pass through (more details on tip-widening in the supplement). Samples were then rocked for 10 minutes at room temperature. After this initial rocking, samples were again pipetted up and down for one minute with a widened p1000 tip. At this point, samples were passed through a filter assembly containing a 400μm strainer (pluriSelect 43-50400-03) and a 100μm strainer (pluriSelect 43-50100-51) into a liberase-stopping solution containing 70% PBS 30% FBS (details on filter assembly and stopping solution in the supplement). Dissociated cells that passed through both strainers fell into the stopping solution, where they were protected from further liberase digestion and maintained on ice for the remainder of the dissociation. Undissociated material remaining on the 100μm and 400μm strainers were washed into separate 50 ml conical flasks and spun down at 300xg for 5 minutes at 4°C (washing details in the supplement). The supernatant was removed, and the material on the 100μm strainer was resuspended in 0.1mg/mL of liberase. In comparison, the material on the 400μm strainer was resuspended in 0.5mg/mL (5x) of liberase for further dissociation (more details on the concentrations and volumes in the supplement). These samples were again pipetted up and down with widened p1000 tips for an additional 1 minute, before being placed on the rocker for a second, 10-minute room temperature incubation. The entire filtration process was repeated for this material. Cells which passed through both strainers were added to the initial stopping solution tube, and material remaining on the 100μm strainer was resuspended in 0.1mg/mL concentration of liberase, while material remaining on the 400μm strainer was resuspended in 2.5mg/mL concentration (25x) of liberase. These conical flasks were incubated on the rocker for a final 10 minutes at room temperature. After this incubation, all material remaining in both tubes was poured directly into the stopping solution. The resulting samples were then strained through a final 70μm strainer. At this point, all non-dissociated material remaining on the strainer was discarded.

After the 70μm filtration, all samples were spun down at 300xg and resuspended in 1ml APBS. At this point, ~100μL of one of the control samples was removed and heat-killed (5 minutes at 65°c, 5 minutes on ice). Cells were then stained with 2μg/mL Calcein green (Invitrogen C3099), or 2μg/mL Calcein blue (Invitrogen C1429) for 20 minutes on ice. Cells were subsequently washed twice with APBS and resuspended at a ratio of ~100μL AL-1 FACS staining Buffer (70% PBS + 2% FBS) per limb for stump and intact samples, or 50μL AL-1 FACS staining Buffer per limb for blastema samples. Cells were kept on ice until sorting.

Results

Section 1 – Improved AL-1 cell transfection with the JetOPTIMUS^® reagent

With an efficiency of ~50%, electroporation is currently the most commonly used method for AL-1 transfection [23]. Because electroporation causes high cell death, we sought to develop alternative AL-1 transfection approaches. The reported efficiency for lipofectamine transfection in axolotl AL-1 cells is only 5% [23], and this protocol requires heating the cells to 37°C for several hours, which significantly reduces the survival and growth rate of AL-1. We have also tested the lipofectamine alternative Fugene[29] and found it to have similarly low transfection efficiencies.

In our search for an improved protocol, we trialed the lipofectamine-related, lipid-cation based reagent JetOPTIMUS^® (PolyPlus), developed for difficult-to-transfect mammalian cells [30–34]. Our initial experiments closely followed the vendor instructions, adding the premixed reagent and plasmid directly to cell culture media at the recommended concentration (discussed in the methods section). We replaced the transfection media with fresh media after 7 days, which is our normal feeding schedule. We found efficiencies ranging from 1–10% in these early trials. These numbers did not greatly outperform optimized lipofectamine transfection at 37°C, but we pursued optimization of this protocol because of its simplicity and the fact it worked at the AL-1-friendly temperature of 26°C. To quantify our transfection efficiency, we developed an automated scanning, segmentation, and quantification protocol using the EVOS M7000, Fiji, and MATLAB Figure 1A. We found that GFP expression increased rapidly for the first 8 days after transfection, before it plateaued, and compared the efficiency of different protocols at this time point. We also found that 24-hour transfection outperformed longer transfections (2 and 7 days). We subsequently tested whether we could increase the efficiency of the protocol by shortening the duration of transfection or by adding a serum starvation step Figure 1B. We also tested whether various endotoxin removal strategies could increase the efficiency of maxiprepped plasmids Figure 1C.

Figure 1:

A) Schematic of AL-1 transfection and quantification using the EVOS M7000 and ImageJ macros. B) A 24-hour jetOPTIMUS^® transfection yields higher efficiency than a 4-hour transfection, this efficiency is further increased by starving the cells for 24 hours in 70% OMEM before transfection. C) Cleaning maxi-prepped plasmid with an additional phenol chloroform purification step results in increased jetOPTIMUS^® transfection efficiency when compared to concentration-matched, uncleaned, maxi prepped plasmid, or plasmid prepped with an endotoxin free kit. D) Recommended starvation protocol for optimal transfection efficiency. E) A representative image of a very high efficiency jetOPTIMUS^® transfection of AL-1 cells. All scale bars are 275μm. Efficiency data is collected 8 days after transfection. Data shown in panels B and C reflects the average of 4 independently transfected replicate wells per condition, error bars represent the standard deviation.

Ultimately, we found that a 24-hour starvation in 70% OMEM before transfection, followed by a 24-hour transfection with JetOPTIMUS^® reagent and phenol-chloroform-purified plasmid, consistently outperformed other conditions Figure 1D. When implementing this protocol, we observe between 20% and 50% transfection efficiency with little mortality; however, there is great variability between transfections in different batches of cells. It is not unusual to have transfections as low as 5%, and on occasion, we have achieved efficiencies as high as 90% Figure 1E. Larger plasmids (>10kb) typically have lower transfection efficiencies. Despite this variability, we find these recommended adjustments greatly improve efficiency and outperform lipofectamine, with little observed mortality. We foresee researchers benefiting from both electroporation and cation-based transfection for different applications, and we anticipate that several of the applications discussed here can be achieved with both strategies.

Section 2 – CRISPR mutagenesis in Axolotl AL-1 cells

Targeted mutagenesis is one of the most effective and rigorous ways to determine gene function. Several studies have demonstrated the efficacy of CRISPR-Cas9 in axolotl embryos and in mature in-vivo tissues [35–50], but it remains undemonstrated in AL-1 cells. Ultimately, by combining mutagenesis with in-vitro and, as discussed later, in-vivo assays, interrogation of axolotl gene function could be greatly accelerated.

To develop a CRISPR-Cas9 mutagenesis protocol in AL-1, we combined an axolotl embryo CRISPR mutagenesis protocol [45], with our electroporation protocol optimized from [23], and an electroporation-based mammalian cell CRISPR mutagenesis protocol [51] Figure 2A, Tables 1–2. To validate this approach, we targeted the genes GJC1, SSBP3, SRFS1, and UBE21 with synthetic guide RNAs (gRNAs), which had already been designed and tested in the lab through in-vivo experimentation. We confirmed mutagenesis of AL-1 cells with these gRNAs through PCR amplification at the target site and T7 endonuclease digest Figure 2B, Supplemental Figure 1. T7 endonuclease targets and cuts heteroduplex DNA resulting from the insertions and deletions caused by successful CRISPR mutagenesis [52]. We found that all four target sites were mutagenized in the presence of Cas9 Figure 2B.

Figure 2:

A) A step by step schematic of AL-1 CRISPR electroporation protocol. B) Demonstration of CRISPR activity with T7 endonuclease digest. CRISPR target site was amplified with PCR and purified with AmPure beads before T7 endonuclease digest. Insertions and deletions caused by successful mutagenesis lead to heteroduplex DNA formation at the CRISPR target site. T7 cleaves this heteroduplex DNA, resulting in two bands in mutagenized conditions. The uncleaved product is highlighted with a white arrow, while cleavage products caused by heteroduplex are highlighted with yellow arrows. The mutagenesis percentage is calculated as in (Guschin et al 2010) and shown in yellow for each mutagenized lan (all other lanes have no mutagenesis).

Table 1.

Becton-Dickinson AriaII Instrument Configuration
Laser	Parameter Name	Dye/Fluorophore	Long Pass Filter	Bandpass Filter
355nm	DAPI	Calcein Blue/ Hoechst	N/A	450/50
488	FITC	GFP/Calcein Green	505	530/30
561	PE	DiI	N/A	586/15
586//561	PE Texas Red	mCherry	600	610/20

Table 2:

Electroporation Reagent Proportions

	CRISPR Components				Cellular Component				Plating
Cuvette Size	Volume crRNA (100mM)	Volume tracrRNA (100mM)	Volume Cas-9 (5.6μg/μL)	Volume Ultrapure H2O	Number of Cells	Volume PBS	Volume Ultrapure H2O	Electroporation volume	Well Size	Volume 10% FBS AL-1 Media
2mm	1.5μL	1.5μL	2μL	7μL	~400,000	70μL	20μL	100μL	12-well	2ml
4mm	3.75μL	3.75μL	5μL	17.5μL	~1,000,000	175μL	50μL	250μL	6-well	3ml

While T7 endonuclease demonstrates that some cells within the population of AL-1 cells are mutagenized, it does not demonstrate the cell-by-cell frequency of mutation across the entire population of cells. To test this, we used an unpublished GFP-AL-1 line transfected with the foamy virus [53] and expanded from a single transfected cell (Tanaka lab, unpublished). In this line, GFP expression is driven by a Ubiquitin C promoter. We electroporated the cells with gRNAs targeting GFP, and after two weeks, we observed a significant drop in fluorescence relative to cells transfected with non-GFP targeting gRNAs, Cas9 protein only, or untransfected control cells Figure 3A–C, Supplemental Figure 2. 80% of naïve GFP-AL-1 cells show GFP fluorescence when sorted by FACS Figure 3C. This may reflect variable activity from the Ubiquitin C promoter across the population of cells. It may also reflect the loss of the cassette through genomic instability. However, we have sorted these cells multiple times, and this ratio appears to be stable across passages. While control conditions containing Cas-9 protein only, or mCherry targeting gRNAs, do not impact levels of GFP fluorescence, GFP targeting guides reduce the proportion of GFP+ cells to ~10%. The successful targeting of the GFP locus was confirmed through T7 endonuclease digest Supplemental Figure 2A.

Figure 3:

A) GFP+ AL-1 cells 14 days after electroporation with either GFP targeting gRNAs or control (mCherry) targeting gRNAs. Only conditions with GFP targeting gRNAs show a reduction in fluorescence. All scale bars are 275μm. B) Quantification of the fluorescence from panel A with a swarm plot of GFP intensity per cell after mutagenesis with listed gRNAs. Cells electroporated with GFP targeting gRNAs show a notable drop in fluorescence. mCherry control shows baseline intensities from cells with no GFP expression. Cells from three independently mutagenized replicates were pooled in silico for each condition. C) Two months after mutagenesis, most cells have lost GFP fluorescence as quantified by flow cytometry. Each column represents an independently mutagenized replicate, with replicates clustered by condition.

Section 3 – Creation and establishment of Novel axolotl cell lines with Tol2 transposition

Both MMLV retrovirus and foamy virus are capable of producing transgenic AL-1 cell lines, but these protocols require the generation of virus [8,53]. AL-1 experimentation would thus benefit from faster, plasmid-based methods for genomic cassette integration. The medaka-derived Tol2 transposon is a promising tool and has already been used to generate in-vivo axolotl reporter lines [54,55]. With this approach, genomic integration is achieved through the expression of two plasmids—an inverted terminal repeat (ITR) containing integration plasmid and pCAGGS-T2TP (Addgene #114725), a plasmid which expresses Tol2 transposase [56–59].

To test if Tol2 would work to create AL-1 cell lines, we transfected AL-1 cells with an integration plasmid containing mCherry-TVA receptor cassette and a Tol2 transposase expressing plasmid, Figure 4A–B. After transfection, we expanded these cells for 6 passages and isolated mCherry expressing cells with FACS Figure 4B–C. After the initial sort, the cells maintained their fluorescence for 30 passages, suggesting the integration is stable. The TVA receptor allows cells to be transfected by ASLV-pseudotyped retrovirus, and as an avian gene, it is not endogenously expressed in axolotl cells [8,60]. This feature may be particularly useful following transplantation of cells in vivo since they could be specifically targeted later, in the limb, using pseudotyped retrovirus. We confirmed that this mCherry-TVA expressing AL-1 cell line is competent to transfection by ASLV-pseudotyped MMLV retrovirus, Figure 4D. We have since used Tol2 and jet OPTIMUS^® to generate a CMV-Cas9-mCherry AL-1 with constitutive Cas9 expression Figure 4E. This line should simplify CRISPR mutagenesis, as the transfection of these Cas9-mCherry cells with gRNA can now be accomplished using JetOPTIMUS^®. The mCherry reporter will allow for the isolation of these cells after in-vivo implant and for experiments where two experimental groups, labeled with different fluorophores, are mixed together.

Figure 4:

A) Plasmid schematic for Tol2-mCherry integration plasmid, including Inverted terminal repeats (ITRs) flanking the inserted segment and a CMV promoter driving mCherry-P2A-TVA cassette. B) Schematic outlining the transfection, passaging and sorting of CMV-mCherry-TVA AL-1 cells. C) EVOS image of sorted mCherry-TVA AL-1 cells. D) mCherry-TVA AL-1 cells can be transfected with ASLV-A pseudotyped MMLV expressing NLS-GFP. E) Representative image of Cas9-mCherry AL-1 cells. F-H) Cells in a single T75 were transfected with a GFP-Puromycin resistant plasmid before being distributed across a 24 well plate, treated with multiple concentrations of puromycin, and scanned over a period of 16 days. F) Representative images of AL-1 with and without GFP-Puromycin resistance at different concentrations of puromycin. G) The quantified counts of these cells, showing the reduction in population caused by different puromycin doses. H) The percentage of GFP+ cells at different puromycin concentrations, showing that selection improves at higher doses. All scale bars are 275μm. In bar graphs each column represents an independent replicate well, wells are clustered by experimental condition.

Another tool missing from the AL-1 toolkit is antibiotic selection. Antibiotic selection is a practical way to isolate transfected cells from their untransfected neighbors and could simplify the generation of transgenic AL-1 cell lines. We transfected AL-1 cells, using jetOPTIMUS^®, with a GFP expressing puromycin resistance plasmid (Addgene 4556) and tested several concentrations of puromycin dihydrochloride (Santa Cruz CAS 58-58-2) Figure 4F–H. Preliminary experiments suggested that concentrations typically recommended in mammalian cell culture (~1μg/mL) [61], had little selective power for AL-1 cells, leading us to test higher concentrations of 10μg/μL, 100μg/μL, and 200μg/μL. After 16 days of incubation, we found that the best selection was achieved with incubation with 100μg/mL Puromycin. Higher concentrations of puromycin also reduced the number of GFP+ cells, while lower concentrations did not sufficiently select for visibly GFP+ cells Figure 4G–H.

Section 4 – In-vivo experimentation with immortalized AL-1 cells

In comparison to mammalian models, axolotls are remarkably tolerant of grafts from other individuals, with many studies grafting cells between GFP⁺ and GFP⁻ animals [22,26,62–68]. This tolerance extends to immortalized AL-1 cells [22]. We sought to further characterize the behavior of implanted AL-1 cells and develop robust implantation methods.

We injected AL-1 cells labeled with DiI, a stable, non-toxic membrane dye[22,69], or mCherry, or GFP into both blastemas and intact limbs and evaluated their ability to persist in the limb and migrate to the regenerative blastema Figure 5. We injected one forelimb and one hindlimb of ~15 cm animals. We performed our initial injections with a pico injector and did not control the volume or concentration of these injections Figure 5B. We later found that in animals of this size, it was easiest to deliver resuspended, labeled AL-1’s with 31-gauge insulin syringes (BD 328419), and that 100μL’s was the largest volume we could consistently inject without overflowing the injection site. We also found that 45,000 cells/uL was the highest concentration we could inject without clogging the needle. Like endogenous dermal fibroblasts [3,4], implanted AL-1 cells can participate in regeneration and persist in the regenerated limb for at least 7 weeks Figure 5A–B. Moreover, after residing in the limb for 49 days, the engrafted AL-1 cells and/or their descendants can migrate to a new blastema triggered by a second amputation event Figure 5A–B. Implanted AL-1 cells can also engraft into intact limbs Figure 5C–D. We injected AL-1 cells into intact limbs, amputating the injected limbs 5 days later. We allowed the limbs to regenerate, finding that cells from preamputation AL-1 grafts can migrate and contribute to the regenerative blastema Figure 5C–D. Upon migration to the blastema, implanted AL-1 cells express blastema markers such as Kazald2 and Axrnbp (formerly known as Cirbp) Figure 5E–F [35].

Figure 5:

A) Schematic showing the injection of AL-1 cells into the limb, subsequent integration with the limb, and migration to the blastema after a second amputation event. B) DiI labeled AL-1 cells integrate with the regenerative limb and migrate to a new blastema after a second amputation event. C) Schematic for the implant of AL-1 cells into intact limbs, and their migration to the blastema after a subsequent amputation event. D) mCherry AL-1 cells injected into the intact limb migrate to the blastema after a subsequent amputation event. E) DiI labeled AL-1 cells in the blastema express the blastema marker Kazald2 detected with HCR. F) GFP labeled AL-1 cells express the blastema marker Axrnbp after migration to the blastema at 21dpa. All injections were performed in animals ~15cm in length.

We then attempted to retrieve implanted AL-1 cells from the regenerative blastema. We injected AL-1 cells into the unamputated limbs of 15cm animals and allowed them to settle for 5 days Figure 6A. We then amputated the limbs and allowed them to regenerate for 21 days when the blastema is most prominent Figure 6A. At this point we harvested the entire limb and separated the blastema from the proximal limb stump. We dissociated these tissues with liberase and sorted with FACS Figure 6A. We successfully recovered DiI-labeled AL-1 cells from limbs Figure 6B. We have optimized this method, finding that equivalent injections into forelimbs and hindlimbs yield a similar number of cells from the blastema Figure 6B, Supplemental Figure 3A–B. Through further optimizations we have found that injecting larger numbers of AL-1 cells increases the number of cells we can recover from blastema and proximal stump samples Figure 6C–D, Supplemental Figure 4A–B.

Figure 6:

A) Schematic showing the injection of AL-1 cells into the limb and the subsequent collection of limbs for dissociation and cell sorting. Flow cytometry data showing B) the isolation of DiI labeled AL-1 cells from axolotl forelimb and hindlimb, blastema, stump and intact limb samples, and C) larger concentration injections result in a greater recovery of DiI+ AL-1 cells. D-F) A visual summary of our findings that D) Injecting more cells leads to greater recovery, E) injection of a larger volume of cells into larger animals leads to more recovery, and F) injection of small volumes into smaller limbs leads to a smaller, but more efficient recovery of labeled AL-1 cells. G) Both GFP+ and mCherry+ AL-1 cells can be recovered from axolotl limbs. All flow cytometry plots have been down-sampled to 30,000 live cells for purposes of comparison, though percentages reflect the entire sample. Extended data available in supplemental figures 2–4.

We also found an upper limit on the number of cells we can inject into 15cm animals. This is because AL-1 suspensions at concentrations higher than 4.5 million cells per 100μL, or 45,000 cells per μL, behave more like a solid than a liquid. As mentioned previously, injections greater than 100μLs overflow back into the environment or the body cavity. To circumvent this obstacle, we injected 200μL of cells into the limbs of much larger 25cm animals, finding that we could indeed recover more cells from the blastema Figure 6E. We also tested injections into smaller ~9cm animals. We were only able to inject these animals with ~10μL of cells per limb, but we were able to recover a larger portion of the injected cells Figure 6F. In our best recoveries to date, we have recovered ~6000 cells per blastema and 14,000 cells per stump (100μL of or 45,000 cells per μL in 15cm animals), although we find these numbers can vary between experiments, even when injecting the same volume and concentration of cells. In addition to DiI, we have recovered cells from both our GFP and mCherry cell lines from Axolotl limbs Figure 6G, Supplemental Figure 5A–B.

Discussion

The tools presented in this study greatly expand the salamander cell culture tool kit, improving transfection and allowing for loss-of-function studies in AL-1 cells for the first time. Importantly, these cells can also be implanted and evaluated in-vivo, allowing researchers to screen through the cellular phenotypes of candidate genes before investing in costly, time-consuming whole organism experiments. We anticipate a model where researchers can mutagenize or otherwise transfect populations of AL-1 cells, transplant these cells into the limb, and evaluate their regenerative phenotype. In the long term, one may be able to transfect cells with multiplexed libraries, enabling high-throughput overexpression screens or even multiplexed CRISPR screens [70–72]. These implant-based approaches should be enhanced by the fluorescent cell lines produced in this study, which obviate the need for dyes such as DiI, and allow for the use of sophisticated, multicolor experiments, and simplified gene editing.

In addition to novel AL-1 methods we make several observations about AL-1 biology which warrant further investigation. Implanted AL-1 cells migrate to the regenerative blastema and can persist in the regenerative limb for at least 8 weeks. However, the extent to which AL-1 cells follow the same molecular programs as endogenous blastema cells, and whether they contribute towards functional tissue in the regenerated limb, remain unclear. It is likely that implanted AL-1 cells recapitulate many, but not all, behaviors of endogenous axolotl cells, and in future studies researchers should experimentally confirm that implanted AL-1 cells do, in fact, replicate their behavior of interest before drawing regenerative conclusions.

While we have greatly improved AL-1 implant, it can likely be optimized further. We demonstrate methods to inject large volumes of AL-1 cells into medium sized animals (15cm), but we do not believe that we have reached the ceiling for injection volume in large animals, where it may be possible to inject up to 800μL of cells. Coupled with higher injection concentrations, one might recover between 10,000 and 20,000 cells per blastema. These large numbers would be especially powerful for multiplexed CRISPR screening and overexpression screening studies, where larger numbers would allow the screening of larger gene lists in a smaller number of animals [70,73]. Conversely, one could take the opposite approach and inject a larger number of smaller animals. Injecting 800μL of AL-1 cells at maximum concentration into a large animal would consume approximately 12 T75 flasks of AL-1 per limb, dramatically increasing the necessary volume of cell culture materials. In Figure 6F we observe a more efficient recovery of AL-1 cells per limb in smaller animals. This could be due to softer limb tissue in younger animals, which dissociates more efficiently.

Another strategy for improved recovery is the selection of AL-1 cells with better engraftment. We typically recover 10,000’s of cells from each limb (stump and blastema combined) from the millions of cells implanted (~1% recovery), suggesting injection, engraftment, and dissociation are major selective bottlenecks. In the cancer metastasis field it is common for researchers to serially transplant, or “passage”, immortalized cancer cell lines from animal to animal to select for specific metastatic behaviors [74,75]. To test this strategy, we expanded AL-1 cells recovered from an intact axolotl limb and tested whether they engrafted more efficiently than their implantation-naïve counterparts Figure 7, Supplemental Figure 6. Although inconclusive, the results are encouraging, and we continue to maintain this line of “limb-recovered” AL-1 (LRAL-1). A cell line with stronger blastema migration would not only be a useful tool for implant experiments, but genetic comparisons between the selected line and non-transplanted AL-1’s could inform blastema biology, although this approach risks selecting for traits that are absent from endogenous regeneration.

Figure 7:

Schematic for the injection, recovery, and growth of Limb-recovered (LRA-L1).

Our use of flow cytometry to evaluate the number of implanted AL-1 cells recovered from limbs has several caveats. The process of dissociation and the speed/pressure of flow cytometry likely results in the destruction of most of the implanted AL-1 cells, and the numbers presented here are likely an underestimate [76]. Clearing and imaging the limbs as in [77–80] and counting the number of positive nuclei may achieve more accurate numbers than FACS. An alternative method would be to transfect the AL-1 cells with a viral barcode library before implant, isolate the genomic DNA from the tissue sample at the experimental time point, PCR-amplify the barcodes, and count the number of barcodes retrieved from sequencing [81]. Unlike both FACS and imaging-based strategies, this method counts the number of clones rather than the total number of cells and is methodologically similar to multiplex screening techniques [70,81].

As spontaneously immortalized cells, AL-1 cells are likely aneuploid, have chromosomal instability, and may possess a significantly modified genome [82]. Resolving the AL-1 karyotype should be a top priority for future AL-1 studies. If AL-1 cells are significantly aneuploid, they will still be quite useful for regenerative and salamander biology. Aneuploid cell lines such as HeLa and Hek293T have been essential tools for mammalian research and cell biology despite their significant deviations from the diploid human genome [83–85]. AL-1 cells will remain superior to mammalian cells and other non-salamander cell lines for limb regeneration experiments, as they possess the axolotl genome, which contains genes important for regeneration, like Kazald2, which have been lost in mammals, and other potentially important regenerative genes such as AXRNBP and Prod1 which are unique to the salamander lineage [18,35]. Large salamander genomes also increase the size of the nucleus and the overall size of the cell, with profound impacts on salamander cell biology, which can only be explored in salamander cells such as AL-1 [17,86–90]. Finally, though axolotls easily tolerate intraspecies grafts, including immortalized AL-1 cell grafts, they do not readily tolerate xenografts from distantly related salamanders and non-salamanders such as Xenopus [62,91]; thus, AL-1 cells are at this time the only option for immortalized cell grafts in axolotls.

The utility of these techniques extends beyond AL-1 cell culture and beyond the limb regenerative field specifically. The techniques developed in this study can translate to other salamander cell lines, and AL-1 cells will continue to be a testing bed for molecular biology techniques in both salamander cell culture and organismal experiments. Moreover, expanding the number of tools available for AL-1 cell culture should make this cell line more accessible to non-axolotl labs, who may be interested in studying salamander cells for comparative biology.

Highlights.

1. An optimized transfection protocol for immortalized Axolotl AL-1 cells with JetOPTIMUS^® reagent, achieving relatively high transfection efficiencies.

2. A CRISPR-Cas9 mutagenesis protocol in AL-1 cells using electroporation.

3. Several transgenic AL-1 cell lines, including GFP AL-1, mCherry AL-1, and Cas-9-mCherry AL-1.

4. Protocols for the large-scale implant of AL-1 cells into axolotl limbs, as well as their recovery from regenerating limbs, allowing for the in-vivo analysis of AL-1 phenotypes.