Amniotic Fluid Stem Cells Increase Embryo Survival Following Injury

Abstract

Although amniotic fluid cells can differentiate into several mesenchymal lineages and have been proposed as a valuable therapeutic cell source, their ability to undergo terminal neuronal differentiation remains a cause of controversy. The aim of this study was to investigate the neuronal differentiation ability of the c-Kit-positive population from GFP-transgenic rat amniotic fluid, amniotic fluid stem (AFS) cells, and to assess how they affected injury response in avian embryos. AFS cells were found to express several neural stem/progenitor cell markers. However, no overt neuronal differentiation was apparent after either treatment with small molecules known to stimulate neuronal differentiation, attempts to differentiate AFS cell-derived embryoid body-like structures, or grafting AFS cells into environments known to support neuronal differentiation (organotypic rat hippocampal cultures, embryonic chick nervous system). Nonetheless, AFS cells significantly reduced hemorrhage and increased survival when grafted at the site of an extensive thoracic crush injury in E2.5 chick embryos. Increased embryo survival was induced neither by desmopressin treatment, which also reduced hemorrhage, nor by grafting other mesenchymal or neural cells, indicating a specific effect of AFS cells. This was found to be mediated by soluble factors in a transwell coculture model. Altogether, this study shows that AFS cells reduce tissue damage and increase survival in injured embryos, providing a potentially valuable tool as therapeutic agents for tissue repair, particularly prenatal/perinatal repair of defects diagnosed during gestation, but this effect is mediated via paracrine mechanisms rather than the ability of AFS cells to fully differentiate into neuronal cells.

Introduction

T he amniotic fluid contains cells that are believed to originate from both embryonic and extraembryonic tissues such as skin, urinary tract, gastrointestinal tract, respiratory system, and placental membranes [1 –4]. Although routinely used in prenatal genetic testing, the precise origin and properties of the cells present in the amniotic fluid have been only recently started to be investigated. Much interest has recently stemmed from several findings suggesting that cells derived from amniotic fluid and placenta contain a population of pluripotent stem cells that can give rise to cells of all 3 embryonic germ layers [5 –10]. Furthermore, the c-Kit-positive cell population isolated from the amniotic fluid of human and rodent was found to express markers of both embryonic and adult stem cells and could be maintained in culture without a feeder layer for at least 250 population doublings without any apparent karyotypic alteration [9]. This suggested that the c-Kit selected population of amniotic fluid cells had stem cell properties, and for convenience of presentation we will refer to this population as amniotic fluid stem (AFS) cells.

There is much evidence that cells present in the amniotic fluid have the potential to give rise to several differentiated mesenchymal tissues, such as bone, cartilage, and adipocytes [9,11,12]. In contrast, their ability to undergo neural differentiation and produce functional neurones has been a matter of contention [4,7,9,13 –16]. In vitro, amniotic fluid cells appeared to differentiate into neurones expressing voltage-gated Na(+)channel with properties similar to those found in native neurones [17]. When transplanted into the rat brain, human amniotic fluid cells were able to migrate in the brain parenchyma and express immature neuronal markers, and AFS cells primed toward neuronal differentiation prior to transplantation seemed to integrate within host tissues and apparently differentiated into dopaminergic neurones [9,13]. Interestingly, amniotic fluid cells transplanted into animals following induction of an ischemic stroke not only were found to upregulate the neural markers Dcx and GFAP, but also appeared to ameliorate behavior in ischemic animals [13,18]. A more recent study, however, did not find any significant evidence that amniotic fluid cells could differentiate into mature dopaminergic neurones and suggested that these cells did not survive long term following transplantation [14].

Given the great interest recently emerged in the potential use of amniotic fluid cells in clinical contexts not only for mesenchymal tissue repair but also for neural repair [19], it is important to better understand their true potential and possible limitations for their use. The main aims of this study were to establish whether AFS cells have a significant differentiation potential along the neural lineage and whether they aid repair in a severe embryonic injury model. AFS cell differentiation ability was studied both in culture following treatments successfully used to induce neural differentiation in other cell types with neural potential and by grafting them in environments where neural development normally occurs, to ensure exposure to appropriate cues for differentiation over time. To this purpose we studied the behavior of AFS cells challenged with small molecules, grown as embryoid body-like (EB-like), or grafted onto organotypic rat hippocampal slices and in the chick embryo. This is a valuable model given its accessibility in vivo and previous reports that mammalian hematopoietic and embryonic stem cells could integrate in the chick developing nervous system [20 –23]. The chick was also used as an embryonic injury model to investigate whether AFS cells affected the response to injury. Here we show that while no overt neural differentiation could be observed in AFS cells under different culture conditions and following transplantation in vivo, grafted cells greatly increased embryo survival following severe injury and that this effect was specific to AFS cells and mediated via paracrine mechanisms.

Materials and Methods

Cell lines

All cells were grown at 37°C in a humidified incubator with 5% CO₂. The tissue culture reagents, unless otherwise specified, were from Gibco. A c-Kit-positive population of AFS cells isolated from amniotic fluid of E16 rats expressing GFP was used. AFS cells were grown in bacteriological dishes (BD Bioscience) in complete Chang medium [α-MEM supplemented with 15% fetal bovine serum (FBS), 20% Chang B and 2% Chang C (both from Metachem Diagnostics), and 1% penicillin/streptomycin] [9] (De Coppi et al., in preparation). The C17.2 neural cell line (kind gift of C. Cepko) derived from postnatal mouse cerebellum by v-myc immortalization was maintained in DMEM containing 10% FBS, 5% horse serum, 2 mM glutamine, and 1% penicillin/streptomycin. Mouse NIH3T3 fibroblasts were grown in DMEM containing 10% FBS, 2 mM glutamine, and 1% penicillin/streptomycin. For grafting experiments, C17.2 and NIH3T3 cells were labeled with the green fluorescent dye, 5-chloromethylfluorescein diacetate (CMFDA), as previously described [24].

Formation of EB-like spheres was induced by plating AFS cells in complete Chang medium either onto PLL-LM (100 μg/mL poly-l-lysine and 10 μg/mL laminin)-coated coverslips or by seeding 1,000 cells in 30 μL drops as hanging drop (HD) on the lid of a 100-mm dish as previously described [25]. EB-like structures were collected after either 3 days (HD) or 15 days (cultures seeded on coverslips) and either induced to differentiate or processed for RNA extraction or immunocytochemistry.

Induction of differentiation

AFS cells were seeded at a plating density of 10,000 cells per 13-mm coverslip in complete Chang medium. The medium was changed to N2 medium (1% N2 supplement, 2 mM glutamine, and 1% penicillin/streptomycin in DMEM/F12) containing the compounds to be tested; 0.5 mM 3-isobutyl-1-methylxanthine (IBMX) and 1 mM dbcAMP (N6, 2'-O-dibutyryl-cAMP) were used either alone or in combination. In some experiments, after 48 h of chemical treatment, the medium was switched to Chang medium or DMEM containing 10% FBS. Cultures were monitored over 7 days and fixed for immunohistochemistry at days 1 and 7.

HD-derived EB-like spheres were plated onto LM-coated coverslips and maintained in either Chang medium or N2 medium (5 days) and N2 medium plus 10% FBS (1–10 days).

Organotypic cultures

Three-hundred-micrometer-thick entorhino-hippocampal slices were prepared from 5-day-old Wistar rats. Slices were grown individually onto a 0.45-μM white Millipore^® FHLC membrane on a 0.4-μM Millipore®-CM culture plate insert according to Stoppini and coworkers [26]. Entorhino-hippocampal slices (n=9) were grown in serum-containing medium (MEM with Glutamax, 5% EBSS, 36 mM d-glucose, 25% horse serum, and 1% penicillin/streptomycin). AFS cell pellets were deposited onto 8±1-day-old hippocampal slices using a glass capillary with very gentle air pressure under a fluorescent LEICA MZ FL III stereomicroscope.

Two-hundred-micrometer-thick E15 chick spinal cord cross sections were cultured onto a 0.4-μM Milipore^®-CM culture plate insert in AFS cell medium (complete Chang medium) with or without AFS cells (100,000 cells/well) present in the lower chamber. Propidium iodide (PI, 5 μg/mL) was added to the medium to assess cell death at specific time points and some slices were processed for immunohistochemistry. PI staining intensity per organotypic slice area was measured at 24 h using ImageJ software available at http://rsb.info.nih.gov/ij (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD); statistical significance was assessed by Student's t-test.

In vivo transplantation into chick embryos

Fertilized Brown Leghorn eggs (Needle Farm) were incubated at 37°C in a humidified forced flow incubator. All procedures were carried out under the Animals Scientific Procedures Act 1986. The total number of animals used for each experimental condition is indicated in the Results section and each experiment was carried out at least twice using controls from the same batch of eggs. Cells were grafted in the developing chick nervous system at different Hamburger and Hamilton stage (H&H) using a glass capillary with very gentle air pressure under a fluorescent LEICA MZ FL III stereomicroscope. Cells were grafted either in the anterior neuropore (H&H9-10) or in normal and injured spinal cords (E2–E2.5). Injury was induced by crushing the spinal cord and surrounding tissues cranially to the vitelline blood vessels (∼2 somites above).

In some experiments, injured embryos were treated with desmopressin, an inhibitor of fibrinolysis, which has been used to treat patients with factor XI deficiency and was recently used to induce hemostasis in an established chick model of spinal cord injury [27]. Injured E2.5 chick embryos were treated with 100 μg desmopressin/embryo at the time of injury. Embryos were observed daily to monitor gross anatomy changes, heart beating, vascularization, and GFP- or CMFDA-positive cells for up to 5 days following surgery and surviving embryos were processed for immunohistochemistry.

Significance of changes in embryo survival over 5 days was assessed by the Log-rank (Mantel-Cox) test and Gehan-Breslow-Wilcoxon test using Prism5 software; P<0.05 was taken to be significant.

RT-PCR

Total RNA was extracted from rat AFS cell monolayers, EB-like cultures, and E16 rat brains (positive controls) using TRIzol^® reagent (Invitrogen) according to the manufacturer's instruction. Five hundred nanograms of the extracted RNA was used for reverse transcription using Superscript II (Invitrogen) with random hexamers. Primers (Appendix Table A1.) were designed using Primer3 software [28]. Cycler was programmed for either 31 or 40 cycles as follows: 1 cycle at 95°C for 4 min, 1 cycle at 95°C, 62°C, and 72°C for 30 s, 1 cycle at 95°C, 60°C, and 72°C for 30 s, 29 or 40 cycles at 95°C, 58°C, and 72°C for 30 s, and finally, 1 cycle at 72°C for 5 min. Following RT-PCR, reaction products were resolved using a 1.5% agarose gel.

Immunocytochemistry and immunohistochemistry

Immunostaining of cells, slices, and frozen sections fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) was carried out essentially as previously described [29,30]. Monolayer cells and EB-like spheres were washed with PBS and fixed in 4% PFA at room temperature for 20 min. They were permeabilized and blocked in 3% bovine serum albumin (BSA), 10% FBS, and 0.2% Triton X-100 in PBS. Chick embryos were fixed in 4% PFA overnight, cryoprotected by incubation in 30% sucrose containing 0.02% sodium azide in PBS at 4°C for approximately 24 h, and OCT embedded. Cryosections (14–16 μm) were permeabilized and blocked in 3% BSA, 10% FBS, and 1% Triton X-100 in PBS. Hippocampal slices were fixed in 4% PFA for 5 min at room temperature and postfixed in 20% cold methanol in PBS for 30 min. Slices were permeabilized in 0.5% Triton X-100 in PBS for at least 12 h at 4°C and blocked in 20% BSA in PBS either for 4 h at room temperature or overnight at 4°C.

The primary and secondary antibodies used are listed in Supplementary Table S2. Primary and secondary antibodies were diluted in blocking solution according to the type of culture to be stained. Incubation was either for 3 h at room temperature or overnight at 4°C for primary antibodies and for 1–3 h at room temperature for secondary antibodies. Stained specimens were visualized and digitally scanned using an Axiophot 2 (Zeiss) with Hamamatsu ORCA-ER digital camera, or an Axiovert 135 (Zeiss) with a ProgRes C14 digital camera, or by confocal laser scanning microscopy using an LSM 710 (Zeiss). Image collection and analysis were performed using Openlab, Volocity (Improvision), or ImageJ software. Complete “z” series optical sections were collected and projected onto a single plane using ImageJ software [31]. The percentage of cells positive for each antibody was evaluated by counting at least 400 cells in different representative fields.

Results

To assess the AFS cell neural potential, we first investigated the expression of neural stem cell markers and markers of neural differentiation by RT-PCR (Fig. 1A) and immunocytochemistry (Fig. 2 and Table 1). AFS cells expressed the neural stem cell markers Sox2, Pax6, and nestin and the neural crest marker Snai1, but not FoxD3 (Fig. 1A). Expression of Sox2, Pax6, and nestin was also detected at the protein level (Fig. 2A–C and Table 1). The transcript for βIII-tubulin, which is usually considered a marker of neuronal differentiation, was detected in AFS cells (Fig. 1A), but the protein was hardly detectable by immunocytochemistry. The neural marker A2B5, however, was clearly expressed by a subset of AFS cells (Fig. 2D–F and Table 1).

FIG. 1.

Analysis of gene expression in amniotic fluid stem (AFS) cells grown as either monolayers or embryoid body (EB)-like spheres by RT-PCR. (A) Undifferentiated AFS cells. (B) Comparison of confluent AFS cell monolayers (cAFS) and EB-like spheres (EB). Rat embryo cDNA was used as a positive control.

FIG. 2.

Expression of neural progenitor and neural differentiation markers in rat AFS cells grown as either monolayers (A–H, O) or EB-like spheres (I–N, P). (A, B) Sox2 (red) is detected in the majority of GFP-AFS cells (green) grown as monolayers, although to different extents. (C) Nestin expression is detected in all AFS cells grown as monolayers. (D) The marker of glial differentiation A2B5 is strongly expressed in a subset of AFS cells. (E) βIII-Tubulin, a marker of neuronal differentiation, is not detectable in AFS cells. (F) Same field as in E, showing GFP expression. (G) βIII-Tubulin expression can be detected in some AFS cells following treatment with dbcAMP and 3-isobutyl-1-methylxanthine for 24 h. (H) Phase image of G. (I, J) EB-like structures are observed when AFS cells are grown on nonadherent substrate. The inset in I shows strong nestin protein expression in an EB-like structure. (K) Vimentin is strongly expressed in EB-like spheres. (L) GFAP immunostaining is detected in EB-like spheres cultured in neural differentiation medium for 6 days. (M) Dcx immunostaining of EB-like spheres cultured in differentiation medium for 6 days. (N) No βIII-tubulin is detected in EB-like spheres cultured in neural differentiation medium for 6 days. (O) A low level of cytokeratin immunoreactivity and a few strongly labeled cells are observed in AFS cell monolayers. (P) Cytokeratin immunoreactivity is higher in EB-like spheres than in monolayers. Scale bars: 50 μm.

Table 1.

Summary of the Expression of Cell Lineage Markers Detected by Immunocytochemistry in Amniotic Fluid Stem Cell Monolayers

Pluripotency		Neural markers		Mesodermal markers		Others
+	−	+	−	+	−	+	−
Sox2^1,2	Oct4¹	Pax6²	MBP⁸	Vimentin^a	Runx2	Flk-1	CD11
(100)		(100)		(100)		(100)
SSEA3¹	Nanog¹	Dcx/DCLK³	O4⁸	Snai1	Stro1	VEGF	CD45
(100)		(100)		(80)		(100)
	c-Myc¹	A2B5⁶	HuC/D⁵	PEB2αA	Iba-1	Clusterinβ	vWF
		(50)		(80)		(100)
	SSEA1¹	NGFRP75^2,4,9	βIII-tubulin⁴	FGFR2-IIIb^a	Myosin	Patched
		(100)		(80)		(100)
	SSEA4¹	Nestin^2,10	MAP2⁵		Myogenin	pH3
		(100)				(25)
		CRMP2^2,5	GFAP^2,7		Pax7	TGF β1
		(100)				(100)
			HNK-1⁹			TGF β3
						(100)

Approximate percentages of positive cells are given in brackets. Bold indicates strong staining and underline indicates weak staining.

Not only in mesoderm (eg, also in neural stem cells).

Embryonic stem cells; ²neural stem cells; ³neuroblast; ⁴early neurones; ⁵neurones; ⁶glial progenitors; ⁷astrocytes; ⁸oligodendrocyte; ⁹neural crest cells; ¹⁰mesenchymal stem cells.

As the expression profile of neural markers in AFS cells was suggestive of a neural differentiation potential, we performed a number of treatments using small molecules known to induce neural differentiation (not shown). The most significant change in cell morphology was observed following treatment with agents that increase cAMP levels such as 1 mM IBMX and 0.5 mM dbcAMP. Treated cultures were still nestin positive (not shown) and some of the cells (<5%) displayed a neuronal morphology and expressed βIII-tubulin (Fig. 2G, H), but this was lost when the drug was removed after 1-week treatment, suggesting that no bona fide neuronal induction might have occurred.

AFS cells plated onto PLL-LM, but not onto other substrates (not shown), rapidly underwent morphological changes and EB-like sphere appearance was observed within 5 days (Fig. 2I, J). The same type of structures also formed when AFS cells were grown in suspension as hanging drops. We tested whether these aggregates were indeed EBs containing all 3 germ layers and could undergo neural differentiation. Expression of markers of the 3 germ layers (endoderm: Sox17, Gata4; mesoderm: Brachyury, Sm22-a; neuroectoderm and derivatives: Pax6, βIII-tubulin) was assessed by RT-PCR. Sm22, Pax6, and βIII-tubulin could be easily detected in AFS-derived EB-like spheres, whereas the expression of Sox17, Gata4, and Brachyury was very low and could be detected only when a high cycle number (40 cycles) was used (Fig. 1B). Under these conditions, Sox17, Gata4, and Brachyury PCR products were not detected in AFS monolayers at either confluency or low density (Fig. 1B and Supplementary Fig. S1). At the protein level (Fig. 2J–P and Table 1), the EB-like spheres continued to express the stem cell markers vimentin and nestin and the neuroblast marker Dcx, but no significant βIII-tubulin (Fig. 2I, K, M, N). GFAP staining was also observed (Fig. 2L), but unlike that of other intermediate filaments (eg, nestin and vimentin) was rather granular, suggesting that though GFAP protein was synthesized, it did not become properly organized in a filamentous structure. Cytokeratin reactivity appeared to be more extensive in EB-like spheres grown either under normal or differentiating culture conditions (not shown) than in AFS monolayers, where patches of cytokeratin-positive cells were sometimes observed (Fig. 2O, P).

To rule out that failure to induce clear neural differentiation in AFS cells might have been due to nonoptimal in vitro conditions, an organotypic culture system was used. We grafted AFS cells onto hippocampal slices, as neurogenesis in the dentate gyrus continues postnatally and therefore provides a permissive neurogenic environment (Fig. 3). Cells seeded onto the dentate gyrus of rat hippocampal slices cultured either in serum-containing medium (n=9; Fig. 3) or in serum-free media (n=22; not shown), but not onto the enthorinal cortex (not shown), were able to migrate and proliferate (Appendix Table A2) and underwent morphological changes. However, neither 2 weeks (not shown) nor 4 weeks after grafting, costaining of GFP and βIII-tubulin (Fig. 3A, B) or GFAP (not shown) was observed.

FIG. 3.

AFS cells grafted either onto hippocampal slices (A) or within the developing spinal cord in sham-operated and injured E2.5 chick embryos (B–H). (A) Immunohistochemistry of hippocampal slice grafted with rat AFS cells after 30 days in culture. AFS cells are observed in the dentate gyrus (DG) of the hippocampal slice. (A′) Higher magnification of the AFS cells shown in A. (B) Detection of AFS cells (green) following injection into E2.5 chick embryo spinal cord. (C) Image of the vascular network at 24 h after AFS cell injection into an E2.5 chick embryo; note that the vascular network is not affected by the procedure. (C′) Fluorescent image of region boxed in C, showing grafted cells. (D) Detection of AFS cells (green) following injection into E2.5 chick embryo spinal cord and βIII-tubulin (red); nuclear counterstain is blue. AFS cells are observed within the spinal cord. (E) Detection of AFS cells (green) and βIII-tubulin (red) at 4 days after grafting into normal E2 embryos. (F) Another embryo from the same experimental group as in E, showing (arrow) AFS cells close to the dorsal root ganglion (DRG). (G) Detection of AFS cells (green) at 3 days after injection into an injured E2.5 chick embryo and of βIII-tubulin (red); nuclear counterstain is blue. The arrow points at cells close to the DRG. (H) Phase contrast at low magnification of the section shown in G; healing at the injured site is observed, although in this embryo morphological abnormalities are observed. Scale bars: (A′, C, E, F, G)=100 μm; (A)=300; (B, C′)=240 μm; (D, H)=200 μm.

It is conceivable that the postnatal organotypic hippocampal slices/culture conditions used might not have provided the factors required for neural differentiation of AFS cells. Therefore, we grafted these cells in different locations within the chick central nervous system (CNS) in vivo at different developmental stages when neurogenesis is occurring (Fig. 3). No obvious damage appeared to be caused by grafting AFS cells and the vascular network was not affected (Fig. 3B, C). Cells injected in the anterior neuropore at H&H 9–10 (n=6; not shown), in E2 (H&H 14–15; n=10; Fig. 3E, F), and in E2.5 (H&H 15–16; n=14; Fig. 3D) spinal cords were monitored for up to 5 days, initially in vivo and at the later stages in tissue sections. AFS cells were observed in the E2.5 spinal cord at 1 day after grafting (Fig. 3C) and were still detectable after 4 days (E6), although they were mainly found in proximity of the dorsal root ganglia rather than in the spinal cord (Fig. 3D, E). These cells did not express βIII-tubulin.

Finally, we assessed the effect of AFS cells and their behavior in an embryonic injury model. AFS cells were grafted into E2.5 chick embryo spinal cords at the time of a thoracic crush injury. As at early developmental stages chick embryos can repair small injuries, we performed a lesion that resulted in severing of the spinal cord and nearby tissues that did not repair itself. The response to injury in control and grafted embryos was monitored over 5 days (Figs. 3F and 4).

FIG. 4.

Effect of AFS cell grafts on chick embryos injured at E2.5. (A) Injured chick embryos photographed at 0, 24, 48, and 72 h. (B) Injured chick embryos grafted with AFS cells photographed at 0, 24, 48, and 72 h (h). The arrows point at the lesion site. The inset in A (control embryo at 24 h) shows extensive tissue damage in a section stained with haematoxylin and eosin. The fluorescent images ( insets in B) show presence of AFS cells. Scale bars: whole mounts and insets in B=500 μm; inset in A=100 μm.

Although no significant hemorrhage was usually observed shortly after injury in control injured embryos (n=27), by 24 h diffuse heavy bleeding was apparent. In contrast, in AFS cell-grafted injured embryos (n=18), hemorrhage was more localized to the injury site (Fig. 4A, B). By 48 h, injured chick embryos had slower heart rate than noninjured embryos and by day 3 high mortality was observed in this group. In contrast, in the cell graft group, 60% of the embryos appeared to be developing normally, although in some cases morphological defects were observed at the microscopic level at the site of injury, they were always less severe than those observed in control embryos at earlier stages after injury (Fig. 3F and inset in Fig. 4A). Altogether, AFS cell-grafted embryo survival was significantly higher than in injured controls (P<0.02), particularly at the earlier time points; 43% of the grafted embryos appeared normal at day 4 (Fig. 5).

FIG. 5.

Effect of grafting AFC cells (AFSCs) or different cell types and of desmopressin treatment on injured chick embryos. Survival of embryos injured at E2.5 was monitored over 5 days following different treatments at the time of injury and the cumulative survival probability with 95% confidence intervals was assessed. Survival was significantly increased (*) in embryos grafted with AFS cells when compared to injury alone (log-rank Mantel-Cox test, P=0.009; Gehan-Breslow-Wilcoxon test, P=0.0136). In contrast, no significant difference in survival was observed in embryos grafted with 3T3 or C17.2 cells or treated with the antihemorrhage drug desmopressin.

To assess the specificity of the survival effect induced by the AFS cells, we repeated the injury experiment grafting either neural progenitor cells, C17.2 (n=6), or fibroblasts (n=12). Both cell types showed no significant effect on chick survival; by day 3 after surgery, less than 25% of the embryos grafted with C17.2 cells or fibroblast were alive, versus 58% of AFS cell-grafted embryos (Fig. 5).

As the extent of hemorrhage greatly differed in control and AFS cell-grafted injured embryos, we investigated whether a reduction in hemorrhage would be sufficient to reduce mortality. To this purpose we treated injured embryos (n=8) with the hemostatic drug desmopressin, which we had previously shown to reduce hemorrhage and spinal cord injury at later developmental stages [27]. Although reduced hemorrhage was observed in desmopressin-treated embryos, survival in these embryos was lower than under all the other conditions tested, as almost (n=6) 80% of the animals had died by day 2 after surgery (Fig. 5).

We then tested whether this AFS cell rescue effect was due to neural differentiation by staining for Dcx, GFAP (not shown), and Tuj1 (Fig. 3F); the grafted cells were negative for these neural markers, although most of them continued to express nestin (not shown). As the rescue effect did not appear to be due to AFS cell differentiation, we further investigated whether AFS cells could increase cell survival and whether this was mediated via paracrine mechanisms using organotypic cultures (Fig. 6). Chick spinal cord slices were cocultured in a transwell system either with or without AFS cells (n≥15/group). Control slices grown in medium alone displayed extensive cell death at 2 days in culture as indicated by PI staining of live cultures (Fig. 6A). In contrast, hardly any PI-positive cell was observed when slices were cocultured with AFS cells (Fig. 6B). Consistent with this observation, both NeuN and βIII-tubulin staining showed a healthier neuronal population in spinal cord slices maintained in the presence of AFS cells (Fig. 6C–F); beading, which is indicative of axonal degeneration, was hardly observed in spinal cords cocultured with AFS cells.

FIG. 6.

AFS cells increase survival of E15 chick spinal cord organotypic slices. Spinal cords were either cultured alone (A–A′′, C, E) or with AFS cells in the lower chamber of a transwell culture system (B–B′′, D, F) and stained and photographed live for propidium iodide (PI) at 2 days in culture or fixed either at 2 and 4 days for neuronal marker staining. (A–A′′) Examples of spinal cord slices cultured without AFS cells; note strong PI staining. (B–B′′) Examples of spinal cord slices cultured with AFS cells. PI staining is hardly detectable. The inset shows semiquantitative analysis of PI staining intensity per slice area in no cell control (C) and AFS cell (AFS) cocultures; the difference between the 2 groups is statistically significant (P<0.001). (C, D) Spinal cord slices cultured either without (C) or with (D) AFS cells stained for the neuronal marker NeuN; note the better preservation of the gray matter in the presence of AFS cells. (E, F) Spinal cord slices cultured either without (E) or with (F) AFS cells stained for the neuronal marker βIII-tubulin; note that axon beading is much more extensive in the absence of AFS cells. (E′, F′) High magnification of axons shown in E and F, respectively. Scale bars: (A, C)=200 μm (A, B and C, D are at the same magnification); (E, F)=50 μm.

Discussion

In this study, we have shown that AFS cells are of potential value for aiding repair, although this ability does not correlate with a significant potential to terminally differentiate along neural lineages.

AFS cells have limited neural differentiation potential

Although a number of published studies have suggested a neural potential for AFS cells from different sources, the evidence that they have full neural differentiation potential has not been extensively tested and has been mainly based on neuronal markers and morphological changes [9,14,16,32]. Electrophysiological analysis has indicated presence of some channel/neuronal responses, but the density of these channels is much lower than in mature neurones [17].

Most of the markers, and neural markers in particular, that we have detected in AFS cell monolayers are expressed by the majority of the cells, although to different extents, with A2B5 being the exception. Altogether, this suggests an AFS cell population relatively homogeneous, at least in the context of the markers investigated. This is not surprising given that our AFS cells had been initially selected on the basis of their c-kit expression, whereas other authors have studied unsorted amniotic fluid cells [4,15]. Notwithstanding AFS cells express several neural markers, consistent with work published while this manuscript was in preparation [4], none of the in vitro and in vivo approaches we used revealed a terminal differentiation potential along the neural lineage. Of the many treatments tested, only IBMX and dbcAMP in combination induced AFS cells to acquire a neuronal-like morphology and express βIII-tubulin protein. These 2 agents, which increase intracellular cAMP availability, have been shown to induce neuronal differentiation in various stem cell types [33,34]. However, the fact that the morphological changes and increase in βIII-tubulin were seen within 24 h after treatment, and that neuron-like morphology could be reversed by removing the treatment, suggest that the changes observed reflect more induction of a stress response than of a differentiation process. This is a crucial point to be considered for the interpretation of results from neural differentiation studies.

Another possible explanation of the phenotypic changes observed following treatment of AFS cells with small molecules could be that they induce a labile phenotype that needs to be “stabilized” by other factors, as recently suggested for differentiation of Schwann cell from mesenchymal stem cells [35]. However, the fact that AFS cells grafted in permissive environments did not undergo neuronal differentiation is not consistent with this hypothesis. Interestingly, when grafted onto hippocampal slices, AFS cells survived only when placed onto the dentate gyrus, where neurogenesis occurs during development and in the adult brain, but not on the enthorinal cortex (not shown), indicating a response of AFS cells to specific local cues. Nonetheless, although the grafted cells changed morphology and extended processes, they did not differentiate into neurones.

Therefore, the neurogenic niche appears to provide factors that support survival of rat GFP-AFS cells, but it does not seem to provide differentiation factors. We do not know at this stage which type of interactions is established between AFS cells and hippocampal cells, and whether there is any effect of the former on the latter.

The developing chick embryo provides a neurogenic environment easily accessible in vivo and has been previously used to graft embryonic and somatic stem cells to study their neural potential [20 –23]. As both rodent and human cells were shown to differentiate into neurones when grafted in chick CNS, it is most unlikely that lack of terminal differentiation of AFS cells into neural cell types is due to species differences. Although grafted AFS cells expressed neither βIII-tubulin nor GFAP, some of the grafted cells became nestin negative. As these cells were found in the mesenchyme rather than in the spinal cord, loss of nestin may be indicative of differentiation along mesenchymal lineages, as the AFS cells we used could undergo skeletogenic differentiation (not shown). This is consistent with their extensive mesenchymal differentiation capability [9,36,37].

While our work was in progress, formation of EBs from amniotic fluid cells was reported [38]. That study did not provide any indication of the levels of expression of the 3 germ layer markers found in the EBs and did not assess the EB differentiation potential. As Valli et al., we could detect some markers of the 3 germ layers in EB-like spheres generated under either of the culture conditions used, but in most cases only when a very high cycle number was used in the RT-PCR studies, although expression of some of these markers was never observed in the AFS cell monolayers. However, there was no evidence that EB-like spheres could undergo terminal neural differentiation. Altogether, our work suggests that if any true reprogramming of rat AFS cells occurs spontaneously in a culture condition-dependent manner, this is a rare event and not sufficient to obtain overt differentiation along a neural lineage. Inconsistency in the results reported in the literature may reflect the presence of very small subsets of pluripotent cells, or their absence, in different AFS cell preparations. In contrast to AFS cell neural potential, differentiation along mesenchymal lineages is much more robust, and hence, it might be more valuable for certain clinical applications, for example, for intrauterine repair when damaged tissue/organs are identified prenatally and autologous cells from the amniotic fluid could be used.

AFS cells display a specific repair potential

Considering the size of injury induced in our model and the several tissues affected, the beneficial effect of grafted AFS cells is particularly striking. Interestingly, reduction of hemorrhage with desmopressin alone is not sufficient to reduce mortality, suggesting that the effect of AFS cells goes well beyond this initial response to injury and preservation of the vasculature.

It is clear that AFS cell integration/differentiation is not increased in the injury environment and cannot underlie improved recovery. Further, the effect of AFS cells on survival of operated chick embryos is an early effect, and it should be noted that later mortality seems to correlate with disappearance of injected cells at 5 days. As other cell types do not support survival, it is unlikely that the ability of AFS cells to decrease injury response is simply due to a “cement” effect that fills the gap created by the injury. The observed effect might be due to their ability to either increase survival within the injured tissues or boost endogenous repair mechanisms, or both. The extensive difference in tissue damage observed in live embryos with or without grafted AFS cells within the first 24 h after surgery suggests that the AFS cell graft plays an important trophic role. This is further supported by the great reduction in cell death and parallel increased maintenance of healthy neurones in spinal cords cocultured with AFS cells. This effect is clearly due to factors secreted by the AFS cells, as there is no direct contact between these cells and the spinal cord slices. The paracrine mechanism modulating the AFS cell-dependent survival effect in severe embryonic injury identified here is consistent with reports suggesting that amniotic fluid cells, particularly when transduced with growth factors, can facilitate peripheral nerve regeneration and wound healing [39 –43]. The identity of the factors involved in the increased survival observed both in vivo and in vitro has yet to be established. Several growth factors known to be expressed by AFS cells (eg, VEGF, Tgfβ) are also expressed by fibroblasts, which do not reduce cell death in our in vivo model. It is most likely that the AFS cell-induced effect reported here is due to specific combinations and levels of growth factors. A comparative analysis of secreted factors in AFS cells and cells that do not support survival following injury will have to be carried out to identify such “cocktail.”

Conclusions

Although we cannot rule out that differences in the literature are due to selective properties of different AFS cell lines, the overall difficulty encountered by many researchers in inducing AFS cell neural differentiation suggests that it is at the least premature to consider using these cells for cell therapies aimed at repopulating specific neural populations in the CNS. AFS cells, however, could be a very useful cell source for aiding recovery in injured tissues, given their ability to reduce tissue damage via paracrine mechanisms. They could provide a particularly useful tool for in-uterus repair of defects diagnosed during gestation, as autologous cells could be easily obtained. It will therefore be important to investigate the molecular basis underlying the effects reported here as this could lead to the development of novel therapeutic approaches for neural and nonneural tissue damage.

Footnotes

Acknowledgments

The authors thank the Office of the Commission on Higher Education of the Royal Thai Government, Thailand (Higher Educational Strategic Scholarship for Frontier Research to W.P.), Great Ormond Street Hospital Children's Charity and BBSRC for financial support and Bertrand Vernay for help with confocal imaging.

Author Disclosure Statement

The authors have no conflicting financial interests.

Appendix

Table A2.

List of Antibodies Used for Immunocytochemistry

Primary antibody	Species	Specificity	Dilution	Source
Oct3/4 (c-10)	Mouse	IgG2b	1:100	Santa Cruz Biotechnology, Inc.
Nanog	Rabbit	Ig	1:100	Abcam
Sox2	Rabbit	IgG	1:1,000	Millipore
c-Kit	Mouse	IgG	1:100	Santa Cruz Biotechnology, Inc.
c-Myc	Mouse	IgG1	1:100	Developmental Study Hybrydoma Bank
SSEA1	Mouse	IgM	1:100	Developmental Study Hybrydoma Bank
SSEA3	Mouse	IgM	1:100	Developmental Study Hybrydoma Bank
SSEA4	Mouse	IgG3	1:100	Developmental Study Hybrydoma Bank
Nestin (Rat-401)	Mouse	IgG	1:200	Developmental Study Hybrydoma Bank
Vimentin	Mouse	IgG	1:100	DAKO
GFAP	Rabbit	IgG	1:1,000	Millipore
Pax6	Mouse	IgG1	1:100	DSHB
A2B5	Mouse	IgM	1:100	R&D Systems
Tuj1	Mouse	IgG	1:1,000	Promega
MAP2 (clone HM-2)	Mouse	IgG1	1:100	Sigma
CRMP2	Rabbit	Ig	1:200	Covalab
HuC/D	Mouse	IgG	1:100	Molecular Probe, Invitrogen
Dcx	Rabbit	IgG	1:200	Cell Signalling
O4	mouse	IgM	1:100	R&D Systems
MBP	Rat	IgG	1:100	Millipore
HNK-1	Mouse	IgG	1:100	Sigma
NGFR P75	Rabbit	IgG	1:500	Sigma
VEGF (VG-1)	Mouse	IgG1	1:100	Santa Cruz Biotechnology, Inc.
flk-1	Rabbit	IgG	1:400	Fisher Scientific
TGF β1	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.
TGF βIII	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.
vWF	Rabbit	Ig	1:500	Abcam
PH3 (ser10)	Rabbit	IgG	1:200	Millipore
Cit^* H3	Rabbit	Ig	1:250	Abcam
Iba-1 (C-20)	Goat	IgG	1:100	Santa Cruz Biotechnology, Inc.
Snail	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.
Runx2	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.
GFP	Rabbit	IgG	1:500	Molecular Probe, Invitrogen
GFP	Mouse	IgG	1:100	Cell Signaling
CD11	Mouse	IgG	1:100	Developmental Study Hybrydoma Bank
CD45	Mouse	IgG1	1:100	Developmental Study Hybrydoma Bank
Clusterin β (N-18)	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.
Patched	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.
Pax7	Mouse	IgG1	1:100	Developmental Study Hybrydoma Bank
Myosin	Mouse	IgG2a	1:100	Developmental Study Hybrydoma Bank
Myogenin	Mouse	IgG1	1:100	Developmental Study Hybrydoma Bank
PEB2αA	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.
FGFR2 (BEK)	Rabbit	IgG	1:100	Santa Cruz Biotechnology, Inc.

Secondary antibody	Species	Type	Dilution	Source
Anti-mouse IgG1a Alexa Fluor^® 568	Goat	IgG1a	1:400	Molecular probe, Invitrogen
Anti-mouse IgM Alexa Fluor^® 594	Goat	IgM	1:400	Molecular probe, Invitrogen
Anti-rabbit IgG Alexa Fluor^® 594	Goat	IgG	1:400	Molecular probe, Invitrogen
Anti-mouse IgG Alexa Fluor^® 488	Goat	IgG	1:400	Molecular probe, Invitrogen
Anti-goat IgG Alexa Fluor^® 594	Rabbit	IgG	1:400	Molecular probe, Invitrogen

Ig, immunoglobulin.

, citrullinated histone 3.

References

Underwood

, Gilbert

, Sherman

. 2005. Amniotic fluid: not just fetal urine anymore. J Perinatol, 25:341–348.

Sarkar

, Chang

, Porreco

, Jones

. 1980. Neural origin of cells in amniotic fluid. Am J Obstet Gynecol, 136:67–72.

Gosden

. 1983. Amniotic fluid cell types and culture. British medical bulletin, 39:348–354.

Jezierski

, Gruslin

, Tremblay

, Ly

, Smith

, Turksen

, Sikorska

, Bani-Yaghoub

. 2010. Probing stemness and neural commitment in human amniotic fluid cells. Stem Cell Rev, 6:199–214.

Prusa

, Hengstschlager

. 2002. Amniotic fluid cells and human stem cell research: a new connection. Med Sci Monit, 8:RA253–RA257.

Prusa

, Marton

, Rosner

, Bernaschek

, Hengstschlager

. 2003. Oct-4-expressing cells in human amniotic fluid: a new source for stem cell research? Hum Reprod (Oxford, England), 18:1489–1493.

Prusa

, Marton

, Rosner

, Bettelheim

, Lubec

, Pollack

, Bernaschek

, Hengstschlager

. 2004. Neurogenic cells in human amniotic fluid. Am J Obstet Gynecol, 191:309–314.

Cananzi

, Atala

, De Coppi

. 2009. Stem cells derived from amniotic fluid: new potentials in regenerative medicine. Reprod Biomed Online, 18,Suppl 1:17–27.

De Coppi

, Bartsch

Jr. , Siddiqui

, Xu

, Santos

, Perin

, Mostoslavsky

, Serre

, Snyder

, Yoo

, Furth

, Soker

, Atala

. 2007. Isolation of amniotic stem cell lines with potential for therapy. Nat Biotechnol, 25:100–106.

10.

Tamagawa

, Ishiwata

, Saito

. 2004. Establishment and characterization of a pluripotent stem cell line derived from human amniotic membranes and initiation of germ layers in vitro. Hum Cell, 17:125–130.

11.

Kolambkar

, Peister

, Soker

, Atala

, Guldberg

. 2007. Chondrogenic differentiation of amniotic fluid-derived stem cells. J Mol Histol, 38:405–413.

12.

Kim

, Lee

, Kim

, Hwang

, Kwon

, Kim

, Cho

, Kang

, You

. 2007. Human amniotic fluid-derived stem cells have characteristics of multipotent stem cells. Cell Prolif, 40:75–90.

13.

Cipriani

, Bonini

, Marchina

, Balgkouranidou

, Caimi

, Grassi Zucconi

, Barlati

. 2007. Mesenchymal cells from human amniotic fluid survive and migrate after transplantation into adult rat brain. Cell Biol Int, 31:845–850.

14.

Donaldson

, Cai

, Yang

, Iacovitti

. 2009. Human amniotic fluid stem cells do not differentiate into dopamine neurons in vitro or after transplantation in vivo. Stem Cells Dev, 18:1003–1012.

15.

Tsai

, Hwang

, Tsai

, Cheng

, Lee

, Chang

. 2006. Clonal amniotic fluid-derived stem cells express characteristics of both mesenchymal and neural stem cells. Biol Reprod, 74:545–551.

16.

Toselli

, Cerbai

, Rossi

, Cattaneo

. 2008. Do amniotic fluid-derived stem cells differentiate into neurons in vitro? Nat Biotechnol, 26:269–270author reply 270–261.

17.

Mareschi

, Rustichelli

, Comunanza

, De Fazio

, Cravero

, Morterra

, Martinoglio

, Medico

, Carbone

, Benedetto

, Fagioli

. 2009. Multipotent mesenchymal stem cells from amniotic fluid originate neural precursors with functional voltage-gated sodium channels. Cytotherapy, 11:534–547.

18.

Rehni

, Singh

, Jaggi

, Singh

. 2007. Amniotic fluid derived stem cells ameliorate focal cerebral ischaemia-reperfusion injury induced behavioural deficits in mice. Behav Brain Res, 183:95–100.

19.

Parolini

, Soncini

, Evangelista

, Schmidt

. 2009. Amniotic membrane and amniotic fluid-derived cells: potential tools for regenerative medicine? Regen Med, 4:275–291.

20.

Sigurjonsson

, Perreault

, Egeland

, Glover

. 2005. Adult human hematopoietic stem cells produce neurons efficiently in the regenerating chicken embryo spinal cord. Proc Natl Acad Sci U S A, 102:5227–5232.

21.

Wichterle

, Lieberam

, Porter

, Jessell

. 2002. Directed differentiation of embryonic stem cells into motor neurons. Cell, 110:385–397.

22.

Goldstein

, Drukker

, Reubinoff

, Benvenisty

. 2002. Integration and differentiation of human embryonic stem cells transplanted to the chick embryo. Dev Dyn, 225:80–86.

23.

Demeter

, Herberth

, Duda

, Domonkos

, Jaffredo

, Herman

, Madarasz

. 2004. Fate of cloned embryonic neuroectodermal cells implanted into the adult, newborn and embryonic forebrain. Exp Neurol, 188:254–267.

24.

Santos-Ruiz

, Mowatt

, Marguerie

, Tukiainen

, Kellomaki

, Tormala

, Suokas

, Arstila

, Ashammakhi

, Ferretti

. 2007. Potential use of craniosynostotic osteoprogenitors and bioactive scaffolds for bone engineering. J Tissue Eng Regen Med, 1:199–210.

25.

Narita

, Bielinska

, Wilson

. 1997. Cardiomyocyte differentiation by GATA-4-deficient embryonic stem cells. Development (Cambridge, England), 124:3755–3764.

26.

Stoppini

, Buchs

, Muller

. 1991. A simple method for organotypic cultures of nervous tissue. J Neurosci Methods, 37:173–182.

27.

Whalley

, O'Neill

, Ferretti

. 2006. Changes in response to spinal cord injury with development: vascularization, haemorrhage and apoptosis. Neuroscience, 137:821–832.

28.

Rozen

, Skaletsky

. 2000. Primer3 on the WWW for general users and for biologist programmers. Bioinformatics Methods and Protocols: Methods in Molecular Biology. Krawetz

, Misener

. Humana Press: Totowa, NJ, 365–386.

29.

Whalley

, Gogel

, Lange

, Ferretti

. 2009. Changes in progenitor populations and ongoing neurogenesis in the regenerating chick spinal cord. Dev Biol, 332:234–245.

30.

Gogolla

, Galimberti

, DePaola

, Caroni

. 2006. Staining protocol for organotypic hippocampal slice cultures. Nat Protoc, 1:2452–2456.

31.

Rasband

. 1997–2009. ImageJ. http://rsbinfonihgov/ij/ U.S. National Institutes of Health: Bethesda, MD.

32.

Tamagawa

, Ishiwata

, Ishikawa

, Nakamura

. 2008. Induced in vitro differentiation of neural-like cells from human amnion-derived fibroblast-like cells. Hum Cell, 21:38–45.

33.

Deng

, Obrocka

, Fischer

, Prockop

. 2001. In vitro differentiation of human marrow stromal cells into early progenitors of neural cells by conditions that increase intracellular cyclic AMP. Biochem Biophys Res Comm, 282:148–152.

34.

Long

, Olszewski

, Huang

, Kletzel

. 2005. Neural cell differentiation in vitro from adult human bone marrow mesenchymal stem cells. Stem Cells Dev, 14:65–69.

35.

Shea

, Tsui

, Chan

, Shum

. 2010. Bone marrow-derived Schwann cells achieve fate commitment—a prerequisite for remyelination therapy. Exp Neurol, 224:448–458.

36.

Yeh

, Lee

, Yu

, Hwang

, Chung

, Hsu

, Chang

, Lin

, Tsai

, Wei

, Sung

. 2010. Cardiac repair with injectable cell sheet fragments of human amniotic fluid stem cells in an immune-suppressed rat model. Biomaterials, 31:6444–6453.

37.

Klemmt

, Vafaizadeh

, Groner

. 2010. Murine amniotic fluid stem cells contribute mesenchymal but not epithelial components to reconstituted mammary ducts. Stem Cell Res Ther, 1:20.

38.

Valli

, Rosner

, Fuchs

, Siegel

, Bishop

, Dolznig

, Madel

, Feichtinger

, Atala

, Hengstschlager

. 2010. Embryoid body formation of human amniotic fluid stem cells depends on mTOR. Oncogene, 29:966–977.

39.

Cheng

, Tai

, Sheu

, Chen

, Yang

, Su

, Ho

, Lai

, Pan

. 2010. Enhancement of regeneration with glia cell line-derived neurotrophic factor-transduced human amniotic fluid mesenchymal stem cells after sciatic nerve crush injury. J Neurosurg, 112:868–879.

40.

Pan

, Cheng

, Chen

, Lai

, Lee

, Yang

, Chang

, Ho

. 2007. Post-injury regeneration in rat sciatic nerve facilitated by neurotrophic factors secreted by amniotic fluid mesenchymal stem cells. J Clin Neurosci, 14:1089–1098.

41.

Pan

, Chen

, Cheng

, Ho

, Liu

, Hwang

, Chang

, Wang

. 2009. Combination of G-CSF administration and human amniotic fluid mesenchymal stem cell transplantation promotes peripheral nerve regeneration. Neurochem Res, 34:518–527.

42.

Yoon

, Moon

, Jun

, Kim

, Maeng

, Kim

, Lee

, Baik

, Kim

et al. 2010. Secretory profiles and wound healing effects of human amniotic fluid-derived mesenchymal stem cells. Stem Cells Dev, 19:887–902.

43.

Teodelinda

, Michele

, Sebastiano

, Ranieri

, Chiara

. 2011. Amniotic liquid derived stem cells as reservoir of secreted angiogenic factors capable of stimulating neo-arteriogenesis in an ischemic model. Biomaterials, 32:3689–3699.