Chromosomal Number Aberrations and Transformation in Adult Mouse Retinal Stem Cells In Vitro

Abstract

Purpose.: The potential of stem cells (SCs) as a source for cell-based therapy on a wide range of degenerative diseases and damaged tissues such as retinal degeneration has been recognized. Generation of a high number of retinal stem cells (RSCs) in vitro would thus be beneficial for transplantation in the retina. However, as cells in prolonged cultivation may be unstable and thus have a risk of transformation, it is important to assess the stability of these cells.

Methods.: Chromosomal aberrations were analyzed in mouse RSC lines isolated from adult and from postnatal day (PN)1 mouse retinas. Moreover, selected cell lines were tested for anchorage-dependent proliferation, and SCs were transplanted into immunocompromised mice to assess the possibility of transformation.

Results.: Marked aneuploidy occurred in all adult cell lines, albeit to different degrees, and neonatal RSCs were the most stable and displayed a normal karyotype until at least passage 9. Of interest, the level of aneuploidy of adult RSCs did not necessarily correlate with cell transformation. Only the adult RSC lines passaged for longer periods and with a higher dilution ratio underwent transformation. Furthermore, we identified several cell cycle proteins that might support the continuous proliferation and transformation of the cells.

Conclusions.: Adult RSCs rapidly accumulated severe chromosomal aberrations during cultivation, which led to cell transformation in some cell lines. The culture condition plays an important role in supporting the selection and growth of transformed cells.

Stem cells (SCs), including those residing in the retina, have the capacity to self-renew and generate the specific cells that compose an organ. Single cells isolated from dissociated retinas of mouse postnatal pups show remarkable self-renewal ability and multipotentiality. ^1–3 The eyes of adult mouse and human also contain retinal stem cells (RSCs) in the pigmented ciliary margin (PCM). ^4–6 Furthermore, extensive transplantation studies of neural SCs, RSCs, and retinal progenitor cells show promising approaches for therapy of regenerative vision defects (reviewed in Ref. 7). Nonetheless, few studies have been conducted to investigate the potential of adult retinal cells. 

Extensive culturing may increase cellular genomic instability, making cells unsafe for transplantation. It has been proposed that tumors originate from cancer SCs based on the observation that only a small percentage of cancer cells form tumors and that some tumor cells exhibit stem cell properties (continuous proliferation of a subpopulation, differentiation potentials, and stem cell–associated signaling pathways). Cancer SCs are reported in such diseases as leukemia, breast cancer, and brain tumors. ^8,9  

Extensively cultured neural SCs (NSCs) from fetal and adult brains were shown to undergo genetic and epigenetic alterations. ^10,11 Profound biological modifications were observed only in fetal NSCs. Recently, several laboratories reported spontaneous transformation of adult SCs from different organs during cultivation. ^12–14 Besides adult SCs, embryonic and fetal cells have been found to be tumorigenic depending on environmental conditions. Indeed, tumor formation was observed after transplantation of embryonic SC–derived neural precursor cells into mouse subretinal space. ¹⁵ This tumor formation may be promoted by the unlimited self-renewal of embryonic SCs that may contaminate the culture during differentiation. 

In this study, we analyzed RSCs isolated from the retinal PCM of adult (≥11 week-old) mice and from the retinas of neonates (postnatal day [PN]1). Adult RSCs showed aneuploidy from the early passages on, and the late-passage cells (P31–P33) demonstrated tumorigenicity. In addition, our data indicate preference for milder aneuploidy (diploidy/pseudodiploidy) versus polyploidy in late-passaged adult transformed cells, both in vitro and within the tumors formed. We observed that such instability is in part influenced by the splitting ratio during passaging. Moreover, we found indications of chromosome missegregation and formation of anaphase bridges during mitosis, as well as loss of several cell cycle proteins that may affect the proliferation and tumorigenicity of the RSCs. Neonatal RSCs exhibited an abnormal number of chromosomes after more than nine passages; however, no tumor formation was observed after late-passage (P32) cell transplantation into nude mice. 

Materials and Methods

Mouse model

Wild-type PN1 mouse pups (DBA/2J), adult NMRI and C57/BL6 mice (≥11 week-old), Hsd:Athymic-nude mice (5 to 8 week-old), and NOD.CB17-Prkdc^scid /J mice (7 to 8 week-old; JAX Mice Strain: NOD SCID; Charles River Laboratories, L'Abresle, France) were used. The animals were housed within the facilities of the Jules-Gonin Eye Hospital or the Swiss Institute for Experimental Cancer Research (ISREC). Protocols used in this study complied with institutional guidelines and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

RSC Isolation and Cultivation

Adult RSCs were isolated as described. ⁵ Briefly, RSCs were isolated from the PCM region of adult (≥11 week-old) mouse eyes and 150,000 cells/well were seeded into six-well plates. We used three different conditions: MHM+EGF; MHM+FGF2; MHM+EGF+FGF2 (20 ng/mL for each growth factor; Peprotech, Rocky Hill, NJ). ^1,3 The MHM medium is composed of DMEM/F12, 5.5 mM HEPES, 0.66% glucose, 1 mM l-glutamine, 0.12% NaHCO₃, 0.1 mg/mL transferrin, 0.025 mg/mL insulin, 9.6 μg/mL putrescin, 0.125 μg/mL progesterone, and 0.1 μg/mL selenium (all from Sigma-Aldrich, Buchs, Switzerland). Throughout this article, the last 1 or 2 letters of the cell line names indicate the growth factor(s) used (i.e., E for EGF, F for FGF2, EF for EGF+FGF2, and Ø for no growth factor). In all conditions, penicillin-streptomycin (50 units-50 μg/mL; Sigma-Aldrich) was added. After 3 days, the cells started to form spheres. We collected individual spheres between days 7 and 10, and seeded each sphere into an individual well (24-well-plate) using the same condition (MHM+growth factor[s]) with the addition of 10% FBS. Overnight, these spheres attached to the bottom of the wells and the cells grew out of the spheres to form a monolayer. The cells were passaged at 80% to 100% confluence, with trypsin/EDTA (Sigma-Aldrich). Cell lines NM1/EF, NM2/E, and NM1/Ø were passaged with a high dilution ratio (>1:10), whereas the other the cell lines (NM3/F, NM4/F, NM5/F, NM6/F, NM7/EF, BL1/F, and BL3/F) were passaged with ratios of 1:2 or 1:4. 

Neonatal RSCs were cultured in the presence of EGF and FGF2, in the absence of serum, as has been reported. ^1,3 All cells were directly seeded on a plate, dish, or flask (Nalge Nunc, Roskilde, Denmark) and kept in a 37°C incubator. Table 1 summarizes the cell lines used for the different analyses in the study. 

Table 1.

 

Cell Lines Used for the Indicated Analyses

Cell Lines	Splitting Ratio	Chromosome Aberration		Soft Agar Proliferation		Telomerase Activity		Tumor Formation
								In Nude Mice		In NOD SCID
		P	Result	P	Result	P	Result	P	Result	P	Result
Neonatal
NeoAA/EF	1:10	9	+	10	—	13	+			8	—
		23	++
		36	+++	34	+	34	+	32	—	35	—
NeoN/EF	1:10	10	+
Adult
NM1/EF	>1:10	12	+++	37	++	41	++	31	+++
NM2/E	>1:10	11	+++
NM3/F	1:2 to 1:4	5	+
		15	+	19	++			16	—	20	—
NM4/F	1:2 to 1:4	5	++
		15	++
NM5/F	1:2 to 1:4	5	+++
		15	+++
NM7/EF	1:2 to 1:4	5	+++
		15	+++	18	++			16	—	19	—
NM1/ϕ	>1:10	36	+++	37	++	41	++	33	+++
BL1/F	1:2 to 1:4	12	++
		20	++	23	—			23	—	23	—
BL3/F	1:2 to 1:4	4	+++
		20	+++	22	++			21	—	23	—

 Chromosomal aberration: low (+), medium (++), and high (+++) levels of aberration. Soft agar proliferation: no (—), few (+), and many (++) colonies were formed in semi-solid medium. Telomerase activity: low (+) and high (++) levels of activity. Tumor formation: no (—) and big (+++, reaching ∼1 cm³) tumors were detected.

Immunocytochemistry and RT-PCR

Immunocytochemistry and RT-PCR were performed as described. ^1,3 The antibodies and primers are listed in Tables 2 and 3, respectively. 

Table 2.

 

Antibodies Used for Immunocytochemistry

Antibodies	Antigens	Dilution	Source*
Anti-nestin RAT401	Rat Nestin	1:10	The Developmental Studies Hybridoma Bank
Anti-Bmi1	Bmi1	1:50	Dr. Van Lohuisen (The Netherlands Cancer Institute)
Anti-doublecortin	Doublecortin	1.500	Abcam
Anti-map2 MAB378	Map2	1:500	Chemicon International-Millipore
Anti-β-tubulin-III T8660	β-Tubulin-III	1:1000	Sigma-Aldrich

 * Developmental Studies Hybridoma Bank, Iowa University, Iowa City, IA; Abcam, Cambridge, UK; Chemicon International-Millipore, Zug, Switzerland; Sigma-Aldrich, Buchs, Switzerland.

Table 3.

 

Primers Used for RT-PCR

Primers	Sequence	Size (bp)	Annealing Temp. (°C)
Sox1	F:5′-TGCAGGAGGCACAGCTGGCCTAC-3′	260	55
	R:5′-TGCCGCCACCGCCGAGTTCTGG-3′
Bmi1	F:5′-CAGCAATGACTGTGATGC-3′	240	52
	R:5′-CTCCAGCATTCGTCAGTC-3′
Pax6	F:5′-TCTAATCGAAGGGCCAAATG-3′	334	58
	R:5′-CATAACTCCGCCCATTCACT-3′
FGF2	F:5′-CAAGCAGAAGAGAGAGGAGTTGTGTC-3′	199	66
	R:5′-CAGTTCGTTTCAGTGCCACATACC-3′
Miff	F:5′-GGAACAGCAACGAGCTAAGG-3′	170	49
	R:5′-TGATGATCCGATTCACCAGA-3′
Bmp4	F:5′-CCTCTTCAACCTCAGCAGCATCC-3′	794	72
	R:5′-CACACCCCTCTACCACCATCTCC-3′
GAPDH	F:5′-ACCACAGTCCATGCCATCAC-3′	450	62
	R:5′-TCCACCACCCTGTTGCTGTA-3′

Metaphase Preparation and Chromosome Painting Analysis

The metaphase spreads were prepared as described. ¹⁶ The chromosomes were stained with DAPI and 15 to 30 spreads/cell-line were counted. Chromosome pseudopainting was done by Chrombios GmbH (Raubling, Germany). 

Soft Agarose Assay

Cells (500 cells/well) were seeded in semisolid medium-agarose mix (final concentration: 0.35% low-melting agarose (SeaPlaque BioWhittaker Molecular Application, Rockland, ME). EGF and/or FGF2 (20 ng/mL, final concentration) and 10% FBS were added only to the cells that had been cultured with these supplements. 

Telomere Repeat Amplification Protocol (TRAP) Assay

The TRAP Assay was performed as recommended by the manufacturer (TRAPeze Telomerase Detection Kit; Chemicon-Millipore, Zug, Switzerland). TRAP assay products were size-fractionated in polyacrylamide gel followed by silver staining. Dried gels were scanned, and assay quantitation was performed as described. ¹⁶  

Subcutaneous Cell Transplantation into Immunocompromised Mice

One million cells per site were injected subcutaneously into the left and right flanks of the mice. Seven groups of nude mice (n = 3 to 4 mice/group = 6 to 8 injection sites/group) were injected with the following RSC lines: one line of neonatal RSCs (NeoAA/EF [P32]), three lines of adult NMRI RSCs cultured with growth factor(s) (NM1/EF [P31], NM3/F [P16], NM7/EF [P16]), one line of adult NMRI RSCs cultured without growth factor (NM1/Ø [P33]), and two lines of adult C57/BL6 RSCs grown with FGF-2 (BL1/F [P23], BL3/F [P21]). Six groups of NOD mice (n = 2 mice/group = four injection sites/group) were injected with the following RSC lines: two cell batches of neonatal RSCs (NeoAA/EF [P8 and 35]), two lines of adult NMRI RSCs grown with growth factor(s) (NM3/F [P20], NM7/EF [P19]), and two lines of adult C57/BL6 RSCs grown with FGF-2 (BL1/F [P23], BL3/F [P23]). The mice with tumors were killed when the tumors reached the size of 1 cm³ (≤8 weeks after transplantation). 

Tumor Sample Handling

After removal from the mice, the tumors were divided into three pieces. Two pieces were fixed in ice-cold 4% PFA and one was embedded in paraffin. Paraffin-embedded sections (5 μm) were stained with hematoxylin-eosin for histopathologic assessment and with the Feulgen method to stain the DNA. One tumor piece was minced and digested with trypsin/EDTA to attain single cells, cultured, and passaged once before metaphase preparation. 

Adult mRSC Differentiation

Single cells (20,000 cells/well) were seeded on coverslips coated with polyornithin and laminin. The cells were cultured in MHM medium with FGF2 (20 ng/mL) and heparan sulfate (2 mg/mL). After 6 to 9 days, colonies started to form in the shape of neurospheres. Five spheres were then transferred onto a new coverslip treated with polyornithin, and incubated in six different conditions: (1) MHM+1% FBS, (2) MHM+1% FBS+Noggin (200 ng/mL; Peprotech), (3) MHM+1% FBS+2% B27 (Gibco/Invitrogen, Basel, Switzerland), (4) MHM+1% FBS+FGF2, (5) MHM+1% FBS+FGF2+2% B27, and (6) MHM+1% FBS+FGF2+Noggin. After ∼6 days of cultivation in these conditions, the cells were fixed for immunocytochemistry. 

Western Blot Analysis

Nuclear proteins were extracted (NE-PER Nuclear and Cytoplasmic Extract Kit; Pierce Biotechnology, Rockland, IL), size-fractionated in polyacrylamide gel, blotted to PVDF membranes, and probed with antibodies (Table 4). The signals were detected on chemiluminescent film (ECL Plus Kit; GE Healthcare; Piscataway, NJ). 

Table 4.

 

Antibodies Used for Western Blot Analyses

Antibodies	Antigens	Dilution	Source*
Anti-Rb1(IF-8) sc-102	Rb fusion protein	1:200	Santa Cruz
Anti-CDK6	Human CDK6 (amino acids 1–230)	1:200	Santa Cruz
Anti-CDKN2A/p19ARF [ab80]	Mouse p19ARF (residues 50–150)	1:1000	Abcam
Anti-p53 [PAb 240]	Murine p53 (amino acids 14–289)	1:1250	Abcam
Anti-Cdkn1ap21Kip1 [ab7960]	Human p21 (amino acids 156–164)	1:250	Abcam
Anti-TATA binding protein (1TB18)	TATA binding protein	1:2500	Abcam
Anti-p27 Kip1	Mouse p27	1:1000	Cell Signal
Anti-Cyclin D1	Human Cyclin D1 (residues 151–170)	1:2000	Cell Signal
Anti-Cyclin D3	Human Cyclin D3 (residues 241–260)	1:2000	Cell Signal
Anti-CDK4	Human CDK4 (amino acids 270–290)	1:2000	Cell Signal
HRP-conjugated anti-mouse IgG		1:2000	GE Healthcare
HRP-conjugated anti-goat IgG		1:2000	GE Healthcare

 Santa Cruz Biotechnology, Heidelberg, Germany; Abcam, Cambridge, UK; Cell Signaling Technology, Beverly, MA; GE Healthcare, Basel, Switzerland.

Statistical Analysis

Student's t-test was used to calculate P-value. P < 0.05 was considered statistically significant (Prism; GraphPad Software, Inc., San Diego, CA). 

Results

High Proliferative Capacity of Adult mRSCs

As previously observed in neonatal RSCs, ^1,3 adult RSCs are also highly proliferative (Fig. 1A). Two cell lines isolated from adult NMRI mice (named NM1/EF and NM2/E) showed an increase in proliferation between days 25 and 75, and these cells could generate up to 10⁴⁰ cells in less than 200 days (>116 population doubling). The proliferation was maintained after freezing and thawing (data not shown). When serum was withdrawn from the medium of the monolayer culture, within 1 to 2 days, a fraction of single cells started forming secondary spheres. This finding was similar to our observations in human RSCs. ⁵  

Figure 1.

 

(A) Two adult RSC lines showing high proliferative ability. One line (NM1/EF) was cultured with EGF and FGF2 and the other one (NM2/E) with EGF only. There was no difference in the proliferation potential between the two different conditions. (B) Expression of Nestin (left) in 100% of the cells and BMI1 (right) in 86% of the cells. Arrows: show that not all cells in the population express BMI1. (C) RT-PCR analysis of the two adult RSC lines showing expression of genes needed to maintain stemness and genes known to be expressed during eye development (PD, population doubling; MEFs, mouse embryonic fibroblasts, used as a control).

Molecular Characteristics of Adult mRSCs

The PCM of mouse eye contains RSCs that have the capacity to proliferate in vitro, with the potential to produce all cell types of the neural retina, including rod photoreceptors, bipolar neurons, and Müller glia. ^4,6 By immunocytochemistry, we observed the expression of nestin (Fig. 1B left), a neural progenitor protein, and BMI1 (Fig. 1B right), an oncoprotein known to be needed for self-renewal of neural SCs. ^17,18 These positive expressions resemble our in vitro results from mouse neonatal RSCs. ¹  

We investigated whether the expanded cells maintain molecular characteristics of retinal progenitor/stem cells. RT-PCR analyses (Fig. 1C) of adult RSCs cultured in the presence of EGF and FGF2 showed the expression of Bmi1, which may be the key player in maintaining cell stemness. ^17,18 We observed the expression of Sox1, one of the earliest transcription factors to be expressed in neural progenitor cells. ¹⁹ In addition, Pax6, known to be a key regulator of eye development and essential in maintaining the multipotentiality of RPCs. ²⁰ , was also expressed. Of interest, we noted strong expressions of Pax6 and Sox1 in the cell line grown with both EGF and FGF2 (NM1/EF), and only a low expression from the late-passaged cell line grown with EGF alone (NM2/E). We believe this indicates that both Pax6 and Sox1 expressions depend on exogenous FGF2, especially since both cell lines showed endogenous Fgf2 expression, albeit to different levels (Fig. 1C). The expression of Fgf2 suggested that endogenous FGF2 was needed to maintain cell proliferation, as previously demonstrated. ⁶ We also detected the expression of Bmp4 and Mitf1, known to be regulated during the early stages of retinogenesis. ²¹ This array of gene expressions indicates that the cultured adult RSCs express genes normally active during eye cup formation and characteristic of retinal progenitor and SCs of these developmental stages. NM1/EF had relatively stable expressions of the studied genes at both early and late passage, whereas NM2/E showed higher expressions in late passages, possibly indicating a selection of the cells that express these genes by extensive passaging. 

Continuous Proliferation of Adult mRSCs, but Not of Neonatal mRSCs, in the Absence of Growth Factor

In neonatal RSCs, withdrawal of EGF and FGF2 from the medium leads to cell death. ¹ In adult RSCs, withdrawal of growth factors from the medium did not seem to affect the NM1/EF or NM2/E cells, which showed proliferation kinetics similar to the original RSC lines grown with growth factors (Figs. 2A, 2B). This finding raised our concern about the possibility of transformation of these cells into cancerous cells. To assess this, we chose adult NM1/EF cells to be cultured long term without growth factor, named this line NM1/Ø, and used it for further analyses. 

Figure 2.

 

Some cell lines retained proliferation without growth factor(s) and showed chromosome aberrations. (A) Proliferation of adult RSCs NM1/EF (grown with EGF and FGF2, gray line) and NM1/Ø (grown without growth factor, black dashed line). (B) Proliferation of adult RSCs NM2/E (grown with EGF, gray line) and NM2/Ø (grown without growth factor, black dashed line). Note that NM1/Ø was originally isolated and grown with EGF+FGF2 for 44 days and NM2/Ø with EGF for 100 days before growth factors were withdrawn from the medium. (C) Aberrant chromosome numbers in adult NMRI RSCs. (D) Chromosome count variations in neonatal RSCs: NeoAA/EF at three different passages (P9, P23, and P36) and NeoN/EF. (E) Representative microphotographs of metaphases with normal (n = 40, top) and aberrant chromosome number (n = 80, bottom). (F) Pseudopainted chromosomes of NeoAA/EF at P9 showing a normal karyotype. (G) Pseudopainted chromosome of NeoAA/EF at P23. Note the loss of chromosome 7 and translocation in chromosome 14. (H, I) Pseudopainted chromosomes of NeoAA/EF at P36. Note that all cells had a translocation of chromosome 16 to chromosome 9, besides other aberrations such as loss of chromosomes and trisomy.

Chromosomal Aberration in Adult and Neonatal mRSCs

To check for indications of chromosomal instability, we performed chromosome counting. Both adult NM1/EF at P12 and adult NM2/E at P11 were polyploid with a median chromosome count of 65 and 68.5, respectively (Fig. 2C). Of interest, late-passaged NM1/Ø (P36) consisted of mostly pseudodiploid cells (>56% with 41 chromosomes), despite being derived from NM1/EF, indicating a preference of pseudodiploidy over polyploidy after extensive passaging. 

We also analyzed two samples of neonatal RSCs derived from two different preparations: NeoAA/EF ^1,3 and NeoN/EF (Fig. 2D). In contrast to the adult RSCs, the majority (83%) of neonatal NeoAA/EF cells in early passage (P9) had normal diploid chromosome number (n = 40). In later passages (P23 and P36), the number of cells with 40 chromosomes decreased (to 67% and 33%, respectively). However, the median chromosome count remained at 40 (Fig. 2D). 

Pseudopainted chromosomes of NeoAA/EF showed that at P9, most cells (73%) displayed a normal karyotype (Fig. 2F). At a later passage (P23), 67% of the cells had a normal karyotype and the rest of the cells had either lost 1 chromosome or undergone a translocation (Fig. 2G and not shown). At P36 (the latest passage analyzed) no cells showed the normal karyotype. All cells (including the ones with diploid 40 chromosomes) exhibited a translocation of chromosome 16 to chromosome 9 (Figs. 2H, 2I) and some cells showed other rearrangements such as translocation, partial deletion, and trisomy (Figs. 2H, 2I, and not shown). 

Independent Cell Lines from NMRI Background

To confirm the results from the adult RSCs, which had already shown a severe aberration in the number of chromosomes at early passages, we isolated five independent cell lines from adult NMRI mouse eyes (NM3/F, NM4/F, NM5/F, NM6/F, and NM7/EF). For these cell lines, we used a low splitting ratio during passaging (1:2 for the first 2 to 3 passages, and then 1:4), to investigate whether this method has any effect on chromosomal stability and cell transformation. These cells also showed high proliferation (Fig. 3A). We monitored the number of chromosomes in four cell lines at an early passage (P5, ∼population doubling [PD] 8–12) and a late passage (P15, ∼PD 39–55). Of note, at the early passage (P5), all lines had aberrant chromosome counts; however, at different levels (Fig. 3B). In the NM3/F cells, 43.3% of the cells had 36 to 39 chromosomes, 40% had 40 chromosomes, and 16.7% had 77 to 80 chromosomes (median = 40). No significant difference was found between chromosome counts of the NM3/F cells at P5 and P15. The three other cell lines showed more variation. The NM4/F cells had a median chromosome count of 39.5 at early passages and down to 37.5 at late passages, indicating loss of chromosomes in a portion of the population. The NM5/F cells showed a high degree of polyploidy at early passages (median = 71.5) that stayed in a similar level at the late passage (median = 69). NM7/EF cells showed a larger range (P < 0.0001) of aneuploidy at the late passage compared with the early passage, suggesting a possible impact of culturing these cells with two growth factors (EGF and FGF2) on the aneuploidy level (Fig 3B). 

Figure 3.

 

(A) Proliferation of independent RSC lines isolated from adult NMRI mouse eyes. There were no significant differences between cell lines. (B) Variations in chromosome counts in four different adult NMRI RSC lines at P5 and P15. (C) Proliferation of RSC lines isolated from C57/BL6 mouse eyes. (D) Proliferation of C57/BL6 RSC lines BL1/F (grown with FGF2, black line), BL1/Ø (grown without growth factor, black dashed line), BL3/F (grown with FGF2, gray line), and BL3/Ø (grown without growth factor, gray dashed line). Note that BL1/Ø and BL3/Ø cells were cultured in the presence of FGF2 for 135 days, before FGF2 was withdrawn from the medium. †Cells stopped proliferating, showed a differentiated morphology, and eventually died. (E) Variation in chromosome counts in BL6 RSCs in early and late passages.

Adult mRSC Lines from a C57/BL6 Background

To check whether the aberration in chromosome count is linked to certain genetic backgrounds, we derived RSCs from adult mice of a different background, the C57/BL6. These RSC lines (named BL1/F and BL3/F) showed a long, slow growth period that lasted ∼105 days before a sudden sharp increase in proliferation (Fig. 3C). These cells were cultured with a low splitting ratio (1:2–1:4) as explained earlier. Up to the last time the cells were cultured, they had undergone at least 45 population doublings in 146 days. It is important to note that the increase in proliferation of BL3/F was very sharp (Fig. 3C) which could be an indication of transformation. Also of interest, when we withdrew the growth factor from the medium after the cells had been cultured for 135 days, only the BL3/F cells continued to grow, whereas proliferation of the BL1/F cells stopped after 12 days (3 passages; Fig. 3D). Chromosome count in the BL1/F and BL3/F cells showed that aberrations also occurred in adult RSCs derived from the C57/BL6 background. The medians of the chromosome counts, however, were maintained at (pseudo)diploid 40 and 41 at later passages for the BL1/F and BL3/F cells, respectively (Fig 3E). Despite the indications of transformation by the sharp increase in proliferation, we observed no tumor growth after transplantation of these two lines into both the nude and NOD mice, showing that these cells were not transformed. 

Telomerase Activity and Anchorage-Independent Proliferation in mRSCs

To check for telomerase activity that may facilitate cell immortalization, we performed a TRAP assay. All cells analyzed showed telomerase activity (Fig. 4A). Neonatal RSCs exhibited lower telomerase activity compared to adult RSCs (Fig. 4A); however, no significant differences were found among the samples. 

Figure 4.

 

RSCs showed telomerase activity and an anchorage-independent manner of proliferation in semisolid medium that correlated with the variation in chromosome counts. (A) Telomerase activity in neonatal and adult RSCs (MEFs, mouse embryonic fibroblasts). Each sample was normalized to telomerase activity of MEFs (used as basic 100% activity level). (B, C) Percentage of colonies formed by neonatal (B) and adult (C) RSC lines in semisolid medium (0.35% medium-agarose mix). Histograms represent the mean number of colonies in three wells from two independent experiments. Error bars, SEM. (D–G) Representative microphotographs of the colonies formed by: NeoAA/EF, P10 (D); NeoAA/EF, P34 (E); NM1/EF, P37 (F); and NM1/Ø P37 (G).

To check for other signs of cell transformation, we analyzed the anchorage-dependent proliferation competence of these cells by seeding 500 cells into semisolid growth medium and counted the colonies formed after 1 week. The neonatal NeoAA/EF cell line showed an increase in colony formation with increasing passages, from 0.5% formed by P10 to 9% by P34 (Fig. 4B). The adult RSC lines formed a similar high number of colonies (between 32.3% and 50.2%), except for BL1/F (Fig. 4C). Of interest, BL1/F was the cell line that did not proliferate in the absence of growth factor (Fig. 3D). We took into account that although an anchorage-dependent proliferation can be an indication of transformation (i.e., when a high percentage of cells form colonies), this capacity may also be a reflection of the ability of neural SCs to proliferate as spheres. ²² However, only a low percentage of SCs is expected in the culture, as previously observed for neonatal RSCs. ¹  

Tumor Formation in Nude Mice

To determine the tumorigenicity of these cells, we subcutaneously transplanted 10⁶ cells/site on the right and left flanks of 24 nude mice (7 groups: NeoAA/EF [P32], adult NM1/EF [P31], NM3/F [P16], NM7/EF [P16], NM1/Ø [P33], BL1/F [P23], BL3/F [P21]; n = 3–4 mice/group = 6–8 injection sites/group). None of the mice injected with neonatal RSCs (P32) showed any tumor growth up to 8 months later (Fig. 5A). We observed four tumors from eight injection sites in mice injected with NM1/EF (P31) (Fig. 5B) and eight tumors from eight injection sites in mice injected with NM1/Ø (P33) (Fig. 5C). The other cell lines showed no tumor in nude mice up to 6 to 7 months after transplantation. 

Figure 5.

 

(A–C) Two adult RSC lines were transformed and formed tumors in nude mice. Representative photographs of nude mice after subcutaneous transplantation of (A) neonatal RSCs NeoAA/EF (n = 0 tumors/8 injection sites), (B) adult RSCs NM1/EF (n = 4 tumors/8 injection sites), and (C) adult RSCs NM1/Ø (n = 8 tumors/8 injection sites). (D) Variations in chromosome counts between original cell lines and cells isolated from tumors. (E, F) Aberrant chromosome counts may originate from the unequal distribution of the chromosome during mitosis. Representative micrographs of mitotic cell undergoing equal distribution (E) and unequal distribution (F) of chromosomes. Arrows: anaphase bridges indicating the presence of fused chromosomes (F). (G) The formation of anaphase bridges (F, arrows) may lead to fragmented chromosomes (white arrows) often found in metaphase spreads.

To clarify that the absence of tumor formation was not a result of clearance by the natural killer cells in nude mice, we reassessed the non–cancer-forming RSC lines by subcutaneous injection into NOD mice (six groups: NeoAA/EF [P8 and P35], NM3/F [P20], NM7/EF [P19], BL1/F [P23], BL3/F [P23] ; n = 2 mice/group = 4 injection sites/group). We observed the mice for 5 weeks and observed no tumor growth from these lines, confirming the earlier results with the nude mice. 

All together, the data show that only the NM1/EF and NM1/Ø lines were transformed. Of interest, these were the two adult RSC lines passaged for longer time with a high dilution ratio (>1:10). The other adult RSCs, all of which were passaged with 1:2 to 1:4 dilution ratios, showed no evidence of transformation. Moreover, NM1/Ø tumors appeared earlier and grew faster, inferring that cultivation without growth factor had facilitated the selection and accumulation of malignant cells over the non- or less malignant cells. Of note, the tumor cells were pseudodiploid (Fig. 5D). Since the original cell line (NM1/EF) was polyploid, this again suggests that the pseudodiploidy is favored over polyploidy. It also indicates that the addition of only a few chromosomes was sufficient to transform the cells. 

Histopathologic examination showed that tumors (n = 6) were aggressive and malignant, without any specific differentiated cell morphology (not shown). We found that the tumors from NM1/Ø had more necrotic regions than did those from the NM1/EF line (data not shown), a characteristic of very aggressive and highly proliferative tumors. Analyses of Feulgen-stained histologic sections revealed that although some mitotic cells showed a similar proportion of chromosome distribution during anaphase and telophase (Fig. 5E), some cells showed an unequal distribution of chromosomes to the daughter cells (Fig. 5F). Anaphase bridges were also observed in some of these cells (Fig. 5F, arrows), which might be the cause of the fragmented chromosomes frequently found in the metaphase spreads (Fig. 5G, arrows). 

Potential of Adult RSCs to Differentiate into Cells with Neuronal Fate

We then wanted to determine whether the transformed adult RSCs could still differentiate into retinal cells. Single cells were seeded onto polyornithin and laminin substrate, and cultured in the MHM medium containing FGF2 and heparan sulfate, but in the absence of FBS. Such conditions promoted secondary sphere formation. Five secondary spheres were then plated on polyornithin-coated coverslips and exposed to a combination of FGF-2, B27, and/or Noggin (inhibitor of Bmps), to commit the cells to a neuronal phenotype. We found that even the late-passaged adult RSCs could be committed to a neuronal fate (Figs. 6A–M), as seen by immunostaining for MAP2 (microtubule-associated protein 2) and Doublecortin, both known to be expressed by differentiating neurons. ^23,24 However, these adult cells revealed only immature morphology (Figs. 6E, 6F, 6H, 6I, 6K, 6L). The cells exposed to medium only, to FGF2, or to B27 did not show a differentiated morphology (not shown). In the presence of FGF2 and FBS, certain cells adopted a neuroblast morphology expressing Doublecortin and Map2. This number was increased by the addition of B27 or Noggin (Fig. 6M). Nonetheless, it is important to note that the total number of adult RSCs committed to a neuronal fate was low. By contrast, neonatal RSCs showed a robust differentiation capacity (Fig. 6N) as previously observed with four different cell lines (until P20) ³ , and even at late passages (P32, not shown). 

Figure 6.

 

Adult RSC lines retained the potential to form early neurons even with an aberrant chromosome count. (A–L) Representative photographs of the cells after differentiation: DAPI-stained cells (A–C), Doublecortin-immunostained cells (D–F), Map2-immunostained cells (G–I), and phase-contrast micrographs showing that cells that expressed Doublecortin and Map2 had an early neuronal morphology (J–L). (M) Number of cells (from a total from five spheres/condition) expressing Doublecortin and Map2 with (an immature) neuronal morphology after differentiation protocol with the indicated factors (n = 5–6). (N) β-Tubulin-III expressing cells (red) and GFAP-positive cells (green) derived from neonatal mRSCs after differentiation in FGF-2 and B27. Nuclei were counterstained with DAPI (blue).

Possible Effect of the Loss and/or Misregulation of Cell Cycle Proteins on Cell Transformation

To determine whether the regulation of some important cell cycle proteins may be the reason behind the chromosomal aberrations or cell transformation, immunoblot analysis was performed. We found that the expressions of early cell cycle proteins such as cyclin D1, cyclin D3, and their targets CDK4 and CDK6, were maintained in all cell lines analyzed, albeit to different levels between cell lines (Fig. 7A). These cyclin-CDK complexes are known to phosphorylate Rb1, which when hyperphosphorylated results in the de-repression of E2F, enforcing cell cycle progression. ^25–28 We could not detect Rb1 expression in all RSC lines tested (Fig. 7B), which may create a condition conducive to continuous cell proliferation, and can serve as one genetic hit toward cell transformation. 

Figure 7.

 

(A–E) Immunoblots of nuclear protein extracts (20 or 50 μg of total nuclear protein was used). Proteins were extracted from cell lines NeoAA/EF (P39), NM1/EF (P38), NM1/Ø (P40), NM3/F (P19), NM7/EF (P18), BL1/F (P23), and BL3/F (P22). (A) All cell lines maintained the expression of early cell cycle proteins Cyclin D1, Cyclin D3, CDK4, and CDK6. (B) There was a loss of Rb1 in all RSC lines and a loss of Cdkn2a^p19ARF in some cell lines that showed high level of chromosomal aberrations. U2OS and SaOS are osteosarcoma cell lines used as positive and negative controls for Rb1, respectively. (C) Low expression level of Trp53 in some RSC lines. (D, E) Protein expression of Cdkn1a^p21Cip1 and Cdkn1b^p27Kip1. CDK, cyclin-dependent kinase; Cdkn, cyclin-dependent kinase inhibitor; Rb1, retinoblastoma; TBP, TATA binding protein (loading control).

Only two cell lines (NeoAA/EF and BL1/F) maintained Cdkn2a^p19ARF expression, with NeoAA/EF showing a much stronger Cdkn2a^p19ARF band compared with the adult BL1/F line (Fig. 7B). Both cell lines were the ones that could not proliferate without exogenous growth factor(s) (Ref. 1 and Fig. 3D). Cdkn2a^p19ARF is known to promote MDM2 degradation and thus stabilize Trp53. ^29–31 In accordance with this, the expression of Trp53 during normal proliferative conditions was observed at different levels, with the highest level expressed by NeoAA/EF (Fig. 7C), which also showed an intact Cdkn2a^p19ARF protein regulation (Fig. 7B). The low Trp53 expression levels in most of the cell lines are probably related to the deletion of one allele residing in a chromosome that was lost in those cell lines. The expression level of Cdkn1a^p21Cip1, a direct target of Trp53, did not reflect Trp53 expression (Figs. 7C, 7D). However it is known that Cdkn1a^p21Cip1 can also be regulated in a Trp53-independent manner. ^32–35 Another protein from the cip/kip family, Cdkn1b^p27Kip1, was expressed by all cell lines (Fig. 7E). Together, our data on cell cycle protein expression indicate some possible effects of the loss or misregulation of cell cycle proteins on chromosomal aberrations and/or transformation of RSCs. 

Discussion

Our current data on RSC proliferation potential confirm our previous reports on neonatal mouse ^1,3 and human RSCs. ⁵ In this study, more detailed analyses showed that the most of the adult mRSCs rapidly accumulated severe aberrations in chromosomal counts with increasing passaging. The degree of chromosomal aberrations differed between neonatal and fully adult RSCs. Neonatal RSCs appeared to be more stable and, although gaining or losing few chromosomes at later passages, remained in the diploid or pseudodiploid range. On the contrary, adult RSCs showed heterogeneity in chromosome counts, even at early passages. Of interest, in some adult RSC lines, late passages were dominated by diploids or pseudodiploids, even when the same cell lines showed a high degree of polyploidy in the earlier passage (NM1 in Fig. 2C and BL3/F in Fig. 3E). The majority of cells isolated from the tumors were also diploids or pseudodiploids (Fig. 5D). Together, these data indicate that in adult RSCs, the milder aneuploidy (diploidy or pseudodiploidy) was favored over polyploidy. It is known that a higher degree of aneuploidy may induce the process of transformation in early passages, but inhibit the tumor progression in later passages. ³⁶ In addition, cells with milder aneuploidy may be preferred during proliferation. Results of a study suggest that a larger number of centrosomes and/or chromosomes force cells to undergo a longer mitosis, ³⁷ thus cells with fewer extra chromosomes will proliferate and accumulate faster if the mutations favor cell division. 

The finding that the neonatal RSCs appeared to be more stable compared with the fully adult RSCs could be due to the presence of different cell populations. In addition, it is possible that the age of the tissue (PN1 vs. ≥11 weeks old) also had an effect. Adult human RSCs have a limited proliferation potential in vitro (up to P15 ⁵ ), showing the occurrence of cellular ageing. In the hematopoietic system, hematopoietic stem cells (HSCs) undergo senescence as do all mammalian cells. Indeed, the HSC pool was exhausted after five serial transplantations in a murine model. ³⁸ Furthermore, fetal HSCs differ from those of the adult by criteria such as surface phenotype and cell cycle status. ⁸ Adult SCs are kept in a reversible cell cycle arrest state known as quiescence. HSCs, for example, have a slow turnover, and only a few are activated at a time. ³⁹ HSCs from old mice accumulate more damage-induced γ-H2AX foci in their genomes than more actively proliferating progenitors. ⁴⁰ In addition, global genomic DNA repair was suppressed under experimental serum-starvation–induced quiescence. ⁴¹ Moreover, the diminishing function of neural SCs with age has been linked with increased genomic instability. ⁴² Together, accumulated evidence indicate that adult SCs may already have some degree of chromosomal aberrations when reactivated from quiescence. 

Adult neural stem cells derived from the hippocampus show a fraction of cells with abnormal chromosome number after extensive amplification (15 passages, PD35); however, most of the cells retain normal number of chromosomes. ¹¹ These studies suggest that cells have different susceptibility for transformation, but we cannot exclude that the culture conditions used may or may not favor cell stability. 

In this study, the cell aneuploidy was likely to be caused by abnormal segregation of the chromosomes during anaphase, often accompanied by the occurrence of anaphase bridges (Figs. 5E, 5F). Anaphase bridges often break, resulting in chromosomal rearrangements such as amplification, translocation, or deletions. ^43,44 It is possible that the loss and/or gain of chromosome(s) in the aneuploid cells affect the genes with roles in survival or proliferative advantage, such as genes acting on cell cycle regulation. A study using conditional Rb1-deficient MEFs showed that the acute loss of Rb1 induces centrosome amplification and aneuploidy. ⁴⁵ We were unable to detect Rb1 by Western blot analysis, leading to speculation on a possible relation between the aneuploidy observed and the loss of this protein, presumably in conjunction with other cell cycle regulating proteins, as the sole loss of Rb1 in mice has no effect on retinal cell proliferation. ⁴⁶  

The loss of Rb1 and Cdkn2a^p19ARF activity may also play an important role in the transformation of some of the studied cell lines. Disruption of Ink4a exon2 which encodes part of Cdkn2a^p19ARF is known to induce tumor development in mice. ⁴⁷ ARF^−/− cells have an increased proliferative capacity, grow faster, and are less responsive to inhibition by cell-to-cell contact. ⁴⁸ The loss of Cdkn2a^p19ARF may explain the continuous proliferation of the cell lines, which in some lines even without exogenous growth factors. Moreover, mouse cells express telomerase (Ref. 49, Fig. 4A), ensuring long-term cell renewal and easing cell immortalization. 

With the exception of NeoAA/EF, all studied cell lines expressed only a very low level of Trp53. Trp53 is a crucial tumor suppressor gene commonly found to be mutated or deleted in human cancers. ⁵⁰ Loss of Trp53 function has been described to give rise to spontaneously tetraploid cells, which are genetically unstable and tumorigenic. ^51,52 A more recent study shows that the loss of one allele of Trp53 results in haploinsufficiency of the cells to transactivate and transsuppress specific target genes, therefore perhaps explaining the increased tumorigenesis observed in mice and humans who have loss of one Trp53 allele. ⁵³ On the other hand, the intact Trp53 expression, albeit to a low level, may explain the functional and reversible cell cycle arrest observed when three of our cell lines (NeoAA/EF, NM1/EF, NM1/Ø) were exposed to G1/S-phase-blocker aphidicolin and M-phase-blocker nocodazole (not shown). 

Cell culture protocols have been linked to cytogenetic abnormalities. ^54–57 If a culture system is suboptimal, it creates a selection. Variants arising spontaneously with growth or survival advantage may overcome this selection barrier and dominate the cell population. In our case, a lower dilution during passaging (ratio, 1:2–1:4) may have prevented the outgrowth of transformed cells, as no tumor growth was observed when these cells were transplanted subcutaneously in immunocompromised mice. In our NeoAA/EF line, the translocation of a segment of chromosome 16 to chromosome 9 may have conferred a selective growth advantage, leading to rapid replacement of the normal diploid cell population by the abnormal cells, as shown by the observation that 100% of this cell line showed this aberration at late passage (P36). Further study is needed to determine which cytogenetic changes actually confer the selective advantage for this particular neonatal RSC line. Chromosomal instability has also been linked to how the cells are passaged. The use of enzymatic dissociation may select the cells based on cell adherence and sensitivity to the enzyme(s). In our culture system, a mixture of trypsin and EDTA was used. Enzymatic dissociation into single cells has been linked to appearance of trisomy in hESCs, whereas the normal karyotype was preserved with mechanical passaging. ^54–57 In mESCs, it has been suggested that chromosomal abnormalities are acquired when cells are serially cultured in serum-supplemented medium. ^58,59 There are also indications that high O₂ and high glucose levels may increase chromosomal abnormalities as a consequence of oxidative stress. ⁶⁰  

The soft agarose assay appeared to be a reliable method for evaluating the normal state of RSCs. The early-passage neonatal cells lines (NeoAA/EF) grown in this semisolid medium (Fig. 4B) showed a percentage of colony formation similar to the percentage described previously for neural SCs or RSCs grown in liquid medium. ¹ Moreover, the adult RSC line BL1/F that showed proliferation arrest after mitogen withdrawal and did not form tumors in nude mice was also the sole adult RSC line incapable of forming colonies in the soft agarose assay. Altogether, these data show that this assay can be used to identify RSCs undergoing abnormal expansion. 

Intriguingly, our findings showed that even the late-passage cells still expressed stem and progenitor cell markers (Figs. 1B, 1C) and retained the potential to generate cells committed to the neuronal lineage (Figs. 6A–M). However, only neonatal RSCs generated neurons with a neuronal morphology and retinal markers (Refs. 1, 3 and Fig. 6N) when adult RSCs failed to generate mature neurons (Figs. 6A–L), suggesting a marked alteration of the differentiation program. Studies have demonstrated that chromosomal abnormalities do not necessarily affect cell differentiation potential. ^61–63 In normal neural progenitor cells, mitotic segregation defects (lagging chromosomes, supernumerary centrosomes, and nondisjunction) have been found to account for aneuploidy in the central nervous system (CNS). ⁶⁴ Classic in vitro indicators of neural SC transformation (changes in adhesive properties, accelerated growth rate, loss of growth factor dependence, and altered patterns of gene expression) have been shown to have no effect on cell multipotency and indeed those cells could differentiate into neurons, astrocytes, and oligodendrocytes. ¹⁰ Nonetheless, the differentiated adult RSCs in this study never fully matured and appeared mostly like neuroblasts. 

In conclusion, adult mouse RSCs appeared to rapidly accumulate chromosomal changes and lacked the ability to robustly differentiate into neuronal cells. As a consequence, these cells, when cultivated in the present conditions, do not appear to be suitable for studies of retinogenesis or cell fate after transplantation. Similar analyses on adult human RSCs are needed to define whether such undesirable behavior is also inherent in adult human cells. The results would have an important impact on therapeutic applications. 

Summary

This study shows that adult and, to a lesser extent, neonatal RSCs, accumulated chromosomal aberrations during cultivation in vitro. The neonatal RSC karyotype was stable at least until P9, whereas adult RSCs showed aberrations at an earlier passage (P5). It is important to note that tumors were formed only in immunocompromised mice when adult RSC lines that had been passaged extensively with a high splitting ratio (>1:10) were used. In addition, our extensive transplantation experiments using these neonatal RSCs ^65,66 showed no abnormality after transplantation into mouse eyes. Together, the data support the use of neonatal cells (before P10) for biological and transplantation studies, especially with their robust capacity to generate neurons. ³  

An optimal culture condition, which includes passaging protocol (splitting ratio, detachment or dissociation method), medium composition, and cellular stress level (lowering exposure to chemicals, oxidation, toxin) thus enhancing genomic stability, might ensure safer batches of cells for transplantation. Our data highlight the importance of cytogenetic analysis throughout the culture period, especially when such lines are used for transplantation or differentiation studies. Finally, we found that the soft agarose assay may serve as a method of verifying the state and stability of RSC lines to aid in identification of cells adequate for in vitro retinogenesis studies. 

Footnotes

 Supported by the Swiss National Science Foundation and the Velux Foundation.

Footnotes

 Disclosure: M. Djojosubroto, None; F. Bollotte, None; P. Wirapati, None; F. Radtke, None; I. Stamenkovic, None; Y. Arsenijevic, None

Footnotes

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

The authors thank Dana Wanner and Meriem Tekaya for technical assistance; Corinne Kostic, Brigitte Angenieux, Michael Eberhardt, Dusan Zencak, Kriss Canola for valuable discussion; and Ute Koch and Giselle Ferrand for kind help with animal experiments.