A Forensic Path to RGC-5 Cell Line Identification: Lessons Learned

Abstract

In 2001, a transformed cell line RGC-5 was developed from the rat retina that was thought to be of retinal ganglion cell origin. Since that time many investigators have used this line in a wide variety of studies to understand better retinal ganglion cell activity, cell signaling, and neuroprotection. Recently, a publication emerged that claimed that this RGC-5 cell line was derived from mouse and not rat, and other studies also indicated the expression of certain proteins that typically were not associated with retinal ganglion cells. This certainly came as a shock not only to the originators of this cell line, but also to others who have been using this as an in vitro model of rat retinal ganglion cells. As a result, we undertook experiments to determine if the RGC-5 cell line currently in use may have been mischaracterized. We, indeed, found that the RGC-5 cell line was of mouse and not rat origin, as was claimed originally in the original research report. We further determined whether these cells were of retinal ganglion origin. Our findings showed conclusively that RGC-5 cells were, indeed, of mouse origin and, using additional cytogenetic profile testing, karyotyping, and genetic and protein profiling, we concluded that these cells were not of retinal ganglion cell origin, but were the cell line 661W, a mouse SV-40 T antigen transformed photoreceptor cell line. The 661W cell line also was present in the laboratory of the originating laboratory and probably resulted in cross-contamination. The present study reviews some of the errors that were made in misidentifying the RGC-5 cell line and offers some insight as to how this may have happened, and ways one can avoid mischaracterization of a potentially important cell line.

Introduction To The Origin Of Rgc-5 Cells

In 2001, a research article was published describing a new transformed rat retinal ganglion cell line, RGC-5. ¹ This cell line originally was thought to have been derived from isolated rat retinal cells that were transformed with Ψ2 E1A virus. The methods describing the steps in the production and isolation of this transformed cell line appeared in the original report, along with characterizing cell markers consistent with their identification as retinal ganglion cells. ¹ More than 220 published reports have used this cell line. Recently, a report emerged that has reclassified this cell line claiming it was derived from mouse not rat. ² This certainly came as a shock not only to the originators of this cell line, but to others who have been using this as an in vitro model of rat retinal ganglion cells. As a result, we undertook experiments to determine if the RGC-5 cell line currently in use was from mouse and not rat as originally was claimed in the original research report, ¹ and we further determined if they were of retinal ganglion origin. 

New Observations on Species Identification

We were able to obtain several passages of the RGC-5 cell line from cryopreserved RGC-5 cells from the laboratory of origin. Two cell vials from early passages (P2 and above) were sent to Charles River Laboratories, Inc. (Wilmington, MA) for species determination. Sex determination and microsatellite results supported the finding that the RGC-5 cell line was, indeed, of mouse origin, not rat (Fig. 1) as reported by Van Bergen et al. ² We also sent additional samples of RGC-5 cells and a sample of a rat tissue to the University of Iowa for additional genotype testing. The laboratory at the University of Iowa sequenced 500 base pairs (bp) of the coding sequence for the MKKS gene, which has a 15 bp difference between mouse and rat. The laboratory ran rat and mouse controls with each sequence run of the DNA test samples (i.e., RGC-5 passages and rat controls). The analysis showed that the RGC-5 cells were of mouse, not rat, origin. To characterize the origin of these cell lines further, additional cytogenetic profile testing was done at the University of Colorado Cancer Center, Aurora, Colorado. 

Figure 1. 

 

Passages of P6 and P24 of RGC-5 cells were tested for sex determination and microsatellite results to identify origin of cells. The RGC-5 passaged cells were compared to C57 mouse and Brown Norway rats. Both tests revealed that RGC-5 cells were of mouse and not origin.

Cultures were harvested for cytogenetic analyses using a standard protocol. Cell suspensions were fixed in methanol–acetic acid 3:1 and dropped onto microscope glass slides. One slide of each specimen was stained with 4′6-diamidino-2-phenylindole (DAPI, 0.3 μg/mL in Vectashield mounting medium; Vector Laboratories, Burlingame, CA). Another slide from each specimen was hybridized with a mixture of rat Hyblock DNA (Applied Genetics Laboratories, Melbourne, FL) labeled in SpectrumRed (SR) and mouse Cot-1 DNA (Invitrogen, Life Technologies, Grand Island, NY) labeled in SpectrumGreen (SG). Dual-color FISH assays were performed according to standard protocols. Chromatin was counterstained with DAPI (same solution listed above). Analysis was performed by epifluorescence microscopy using single interference filters sets for green (FITC), red (Texas red), blue (DAPI), dual (red/green), and triple (blue, red, green) band pass filters. For each interference filter, monochromatic images were acquired and merged using CytoVision (Applied Imaging, Inc., Santa Clara, CA). 

The quality of each preparation was optimal, and all specimens were highly mitotic. All 6 specimens showed similar chromosomal patterns, represented by a hyper-3n chromosomal complement with all acrocentric chromosomes (Figs. 2–6). The DAPI-stained spreads, illustrated in Figures 2A to 6A, allow the identification of mouse banding patterns in several chromosomes. Figures 2B to 6B, and 2C to 6C show, respectively, interphase and metaphase cells subjected to the dual-color FISH assays. The mouse DNA probes labeled in green were the only probes that hybridized in these preparations. 

Figure 2. 

 

Chromosomal pattern of RGC-5 cells (p2). (A) Metaphase spread stained with DAPI showing mouse polyploid chromosome pattern. Interphase (B) and metaphase (C) cells hybridized with rat Cot1 DNA labeled with SpectrumRed, and mouse Cot1 DNA labeled with SpectrumGreen showing specimen carries mouse DNA.

Figure 3. 

 

Chromosomal pattern of RGC-5 cells (p3). (A) Metaphase spread stained with DAPI showing mouse polyploid chromosome pattern. Interphase (B) and metaphase (C) cells hybridized with rat Cot1 DNA labeled with SpectrumRed, and mouse Cot1 DNA labeled with SpectrumGreen showing line exhibits mouse DNA.

Figure 4. 

 

Chromosomal pattern of RGC-5 cells (p4). (A) Metaphase spread stained with DAPI showing mouse polyploid chromosome pattern. Interphase (B) and metaphase (C) cells hybridized with rat Cot1 DNA labeled with SpectrumRed and mouse Cot1 DNA labeled with SpectrumGreen showing specimen carries mouse DNA.

Figure 5. 

 

Chromosomal pattern of RGC-5 cells (p5). (A) Metaphase spread stained with DAPI showing mouse polyploid chromosome pattern. Interphase (B), and metaphase (C) cells hybridized with rat Cot1 DNA labeled with SpectrumRed and mouse Cot1 DNA labeled with SpectrumGreen showing specimen carries mouse DNA.

Figure 6. 

 

Chromosomal pattern of RGC-5 cells (p12). (A) Metaphase spread stained with DAPI showing mouse polyploid chromosome pattern. Interphase (B) and metaphase (C) cells hybridized with rat Cot1 DNA labeled with SpectrumRed, and mouse Cot1 DNA labeled with SpectrumGreen showing specimen carries mouse DNA.

In conclusion, these three experiments showed no evidence that any RGC-5 specimen carried DNA of rat origin. Now that we had confirmation using several approaches that the cell line was not of rat but of mouse origin, we needed to identify what was the phenotypic origin of the RGC-5 cells. Are they, indeed, retinal ganglion cells? 

The original RGC-5 cell line reportedly was transformed using the ΨE1A virus isolated from the NIH 3T3 cell line. Therefore, we ran a Western blot to determine if the ΨE1A protein was expressed in the RGC-5 cell line. At the time this cell line was created, the laboratory of RGC-5 origin also was working with 661W cells, a mouse cone photoreceptor cell line developed by Tan et al. ³ In our Western blot study, we used RGC-5 cells, 661W cells, and National Institutes of Health 3T3 (NIH 3T3) cells harboring the ΨE1A virus. Only the NIH 3T3 cell line expressed ΨE1A virus protein in the nucleus and in the postnuclear supernatant fraction, and none was detected in RGC-5 or 661W cells (Figs. 7A, 7B). Since the 661W cells were SV40 T-antigen transformed, we used PCR to determine if the SV40 large T-antigen was expressed in these cells. To our surprise, 661W and RGC-5 cells expressed the SV40 large T-antigen (Figs. 8A, 8B), indicating that the RGC5 cells were not transformed with ΨE1A as claimed originally, but were transformed similarly to 661W cells. We also ran PCR to verify mRNA expression. The results again demonstrated a lack of ΨE1A in any RGC-5 cell lines or 661W cells (Fig. 9). 

Figure 7. 

 

Western blot identification of Ψ2 E1A protein in several passages of RGC-5 cells, 661W (P22) and NIH 3T3 cells. Only NIH 3T3 cells had the Ψ2 E1A protein present in the nuclear (A) or cytoplasmic (B) fractions.

Figure 8. 

 

Western blot identification of SV40 T-Antigen protein in several passages of RGC-5 cells, other RGC isolates (RGC-2, RGC-8, RGC-9), 661W (P22), and NIH 3T3 cells. Only NIH 3T3 cells lacked the presence of the SV40 T-Antigen protein in the nuclear (A) or cytoplasmic (B) fractions.

Figure 9. 

 

PCR identification of Ψ2 E1A mRNA in NIH 3T3 cells, but not in several passages of RGC-5 cells.

Therefore, we did a further comparison between 661W cells and RGC5 cells using G-banding karyotyping, and comparison between these two cell lines. We again used the Cytogenetics Core of the University of Colorado Cancer Center to run these tests. One culture vial for each cell line was received by the Cytogenetics Core. RGC-5 and 661W cells were incubated with Colcemid (0.05 μg/mL final concentration) for 3 hours, and specimens were harvested according to standard protocol. Cell pellets were exposed to hypotony in 0.075M KCl at 37°C for 20 minutes and fixed in methanol–acetic acid 3:1. Cell suspensions were dropped onto cleaned slides and carefully air dried. GTL-banding was performed after artificial aging overnight at 56°C using approximately 40 seconds of trypsin digestion and 2 hours 30 minutes of Leishman's staining. Images of well spread metaphase chromosomes were captured using a CCD camera and karyotyping was done using BandView software (Applied Spectral Imaging, Inc., Carlsbad, CA). Chromosome classification followed the method of Nesbitt and Francke, ⁴ and the guidelines of the International Committee on Standardized Genetic Nomenclature for mice (available in the public domain at http://www.informatics.jax.org/mgihome/nomen/anomalies.shtml). A composite karyotype was defined for each specimen assuming all gains and losses present in at least 4 cells (25% of karyotyped cells) as clonal events. 

The mitotic index of these specimens was good. In total, 20 metaphases were counted for each specimen, 11 were karyotyped for 661W and 12 were karyotyped for RGC-5. Both specimens showed a hypertriploid karyotype, with one cell in 661W being hypotriploid. There was intercellular heterogeneity in both specimens and the composite karyotypes are listed below: 

•  
  Specimen 661-W: 55 ∼ 67, XXYY,−3,+5,−6,−7,−8, −11,+15,+16,+17,+18,+19,+mar2(cp11)
•  
  Specimen RGC-5: 63 ∼ 68,XXYY,+4,+5,+6,−7, −9,−11,−13,14,+15,+16,+17,+18,+19,+mar1,+mar2(cp12)

Images are provided in Figure 10 for specimen 661W and in Figure 11 for RGC-5. The two specimens had similar karyotypes and shared numerous numeric changes that were clonal, such as gain for chromosomes 5, 15, 16, 17, 18, and 19, and loss for chromosomes 7 and 11. The two marker chromosomes (mar1, mar2) were small acrocentrics and looked similar in both specimens. Single-cell structural abnormalities were observed in 5 cells in 661W, represented by unknown segments attached to chromosomes 5, 6, and X, and a small acrocentric that could not be recognized. Large chromosomes with rearrangements were not detected in RGC-5, but at least 3 cells carried small acrocentrics different from mar1 and mar2. It is important to note that the length of chromosomes in the metaphase spreads was relatively short, which compromises chromosome classification. This adds to the intrinsic limitations of the G-banding technique in accurately identifying abnormalities. Despite these weaknesses, it is possible to conclude that the results are compatible with the hypothesis that the two specimens are related genetically. The differences shown between RGC-5 and 661-W cells are likely due to the intercellular heterogeneity in both samples reported in this study. This finding is consistent with the transformed cell phenotype, which also can explain the different results for protein expressions reported by different labs. 

Figure 10

 

GTL-banding karyotype of a 661W cell showing hypertriploid male mouse chromosomal complement.

Figure 11. 

 

GTL-banding karyotype of a RGC-5 cell showing hypertriploid male mouse chromosomal complement.

These data suggested that the 661W cell line and the RGC5 cell line may be identical or at least closely related. To examine further if these two cell lines were identical, Therion International, LLC (Saratoga Springs, New York) ran a DNA profile comparing the two lines. The two 661W and RGC-5 cell lines were compared to the inbred mouse strain C57BL/6. The purpose of this study was to estimate the degree of genetic similarity between the two cell lines and verify the strain identity of both cell lines. It was believed that the two lines potentially were genetically identical and derived from the inbred mouse strain C57BL/6. Two multilocus DNA probe/enzyme combinations were used to produce DNA profiles. 

DNA was isolated from cell line samples using an organic extraction procedure. The DNA from each sample was cleaved with the restriction enzyme Hinf I. The DNA from a control C57BL/6 mouse also was processed in this fashion. DNA was digested with various restriction enzymes, and fragments from the two test and single control samples were separated by size using gel electrophoresis, transferred to a nylon membrane, and then hybridized sequentially with radioactively labeled multilocus DNA probes OPT-02 and OPT-05. Each restriction enzyme cuts the DNA into a different family of fragments. Each time a different probe is used, an independent subset of the fragments (hence the genome) is assayed. Thus, each probe/enzyme combination produces an entirely different and independent DNA profile for each test sample. Individuals are found to share genetic information if they display a genetic marker (band) at the same relative positions in their respective DNA profiles. The more bands two individuals share the greater their genetic relatedness. The fewer shared bands, the less related they are. Because of the great similarity of the resulting DNA profiles, we were able to use visual interpretation methods to compute similarity measures. 

A total of 63 genetic markers (bands), or an average of 31 markers per sample, was detected in the DNA profiles of the test and control samples by the two combined probe/enzyme assays. The combined Similarity Indexes (SI = number of genetic markers shared/total number of genetic markers) for the two probe/enzyme combinations are listed below for samples tested. 

 

These two cell lines, 661W (passage 6) and RGC-5 (passage 3), share approximately 95% (60/63 of total) of their genetic markers. This level of variation is consistent with amounts of genetic variation observed previously in C57BL/6 mice from within and among colonies from a given supply house stock. In addition, when compared to a standard C57BL/6 mouse from one of the supply houses, the two cell lines shared approximately 87% and 89% of the assayed genetic markers. In previous studies, anywhere from 5% to 15% genetic variation has been observed among C57BL/6 mice collected from various supply houses. Based on these results, we concluded that both of these cell lines were derived from C57BL/6 mice. 

Lessons Learned

The article by Van Bergen et al., identifying the RGC-5 cell line as mouse and not rat origin, ² was pivotal in our study to verify their findings and to identify the actual cell type of origin. In addition, this group followed with a description of neuronal markers present following different differentiation agents. ⁵ They found that the differentiation resulted in a number of neuronal markers being expressed that were not RGC-specific. This further supports the above finding that RGC-5 cells do not appear to be retinal ganglion cells. 

Based on the above references and our current findings, we have concluded that the RGC-5 cells are, indeed, 661W cells. This is supported by the following data: The RGC-5 cell line was of mouse and not rat cells origin, and RGC-5 cells appear to be genetically identical to 661W cells, and possess the same transforming protein, SV40 large T-antigen. How this happened still is a mystery. The 661W cell line was in use in the laboratory of the RGC-5 cell origin and when it was attempted to develop a transformed RGC line, may have led to 661W cell contamination during the cell cloning experiments. Why was this possibility not caught? At the time the clone was selected, there were several important characterization procedures that were not done. Why? Let's examine the procedure that was used and see where this could have been improved. Retinas obtained from postnatal day 1 rats were dissociated with papain and the mixed retinal cell population was seeded on 100 mm tissue culture dishes. After 4 hours in culture, the retinal cells were overlaid with a Ψ2E1A viral suspension and allowed to incubate for 4 hours for viral transduction to occur. The cells then were washed free of the viral suspension and stably transduced cells were selected using Geneticin (a neomycin analog). The surviving cells were presumed to be transformed by the virus (since they were neomycin resistant) and were used to select various clones of putative retinal ganglion cell lines. The selected clones arising from well separated single cells were selected by placing small circles of Whatman paper dipped in trypsin, lifting them off the dishes, and propagating them in 6-well plates. The clones propagated in this manner were used to test for markers of retinal ganglion cells. A major fault in this process was that the investigators made no attempt to determine if the transformed cells expressed the Ψ2E1A protein, which was instrumental to the cellular transformation. As seen in the present report, these “RGC-5” cells do not have the Ψ2E1A viral protein, but instead express the SV40 large T-antigen protein. According to the lead investigator who established the RGC-5 cell line, the cells from the clone were neomycin-resistant, and had a growing rate that suggested successful transduction and transformation. Moreover, the RGC-5 cells expressed retinal ganglion cell markers, Thy 1.1., Brn 3B or 3c, and related growth factors. ¹ These characteristics also were upheld by others in the field. ^6,7 In addition, following treatment with various agents that appeared to differentiate these cells morphologically, they resembled retinal ganglion cells. ⁸ Some of these treatments resulted in glutamate-sensitive NMDA receptors expressed that responded to glutamate with increased calcium responses and increased cell death following increased dosing with glutamate. ^9,10 Early mistakes with insufficient characterization were not apparent because of the data of others suggested that these cells were, indeed, retinal ganglion cells. Recently Nieto et al. demonstrated that RGC-5 cells contained opsins and intrinsic light responses that were indicative of cone-like photoreceptors. ¹¹ The 661W cell line developed by Tan et al. is a mouse photoreceptor-derived cell line that was immortalized with the SV40 T-antigen and express cone proteins, but not those of rod photoreceptors. ³ The finding that RGC-5 cells are more like cone photoreceptors agrees with our contention that the RGC-5 cell line is, indeed, the 661W cell line. Recently, Wood et al. have shown that visible light exposure on cultured RGC-5 cells can activate cell death pathways leading to apoptosis. ⁵ The finding that “RGC-5” cells are susceptible to light is consistent with these cells being more like photoreceptor cells than ganglion cells. 

Unfortunately, some basic identification checks were not done in the original retinal ganglion cell isolation, including determining if the transformation vector was, indeed, present. In addition, characterization using Western blot analysis or using genetic analysis to determine if these cells, indeed, were of rat origin would have been beneficial for proper identification. The lesson learned is clear, we should not be cavalier about one's findings with regard to cell lines, but always should question whether the cells developed are, indeed, what you think they are based on well-established criteria. Recently, Nardone has called for a more restricted review of reports and grants that use cell lines, but they did not provide evidence of their identification. ¹² Cell lines are distributed readily across laboratories following numerous passages without the receiving laboratories checking the authenticity of the lines. The ATCC has a web site of misidentified cell lines that they have received previously and have now adopted authenticity technology, including STR analysis (DNA profiling) as part of their routine authentication procedures (available in the public domain at http://atcc.org/MisidentifiedCellLines/tabid/683/Default.aspx). However, even ATCC did not independently characterize the RGC-5 cells that were placed in their catalog of available cells (catalog #PTA-6600). The misidentification of cell lines is not new, ¹³ but still remains a major problem, as those using the misidentified cells are investing considerable time, effort, and funding using them to understand better a particular cellular function. There must be a greater scrutiny of cell culture data to determine if the laboratory using these cells has independently authenticated the origin of the cell lines. We must be more vigilant of our own labs and our colleagues requesting that cell line authenticity be performed before we present data on these cells to the scientific world. 

McMaster University has an on-line tutorial on basic practices to follow for verification of cell lines that may be useful for those wanting to adhere to current best practices (available in the public domain at http://www.mcmaster.ca/biosafety/documents/doc053_misidentification_contamination_2012.pdf). 

Acknowledgments

Disclosure: R.R. Krishnamoorthy, None; A.F. Clark, None; D. Daudt, None; J.K. Vishwanatha, None; T. Yorio, None