Human mid-trimester amniotic fluid (stem) cells lack expression of the pluripotency marker OCT4A

Expression of OCT4A is one of the hallmarks of pluripotency, defined as a stem cell’s ability to differentiate into all the lineages of the three germ layers. Despite being defined as non-tumorigenic cells with high translational potential, human mid-trimester amniotic fluid stem cells (hAFSCs) are often described as sharing features with embryonic stem cells, including the expression of OCT4A, which could hinder their clinical potential. To clarify the OCT4A status of hAFSCs, we first undertook a systematic review of the literature. We then performed extensive gene and protein expression analyses to discover that neither frozen, nor fresh hAFSCs cultivated in multipotent stem cell culture conditions expressed OCT4A, and that the OCT4A positive results from the literature are likely to be attributed to the expression of pseudogenes or other OCT4 variants. To address this issue, we provide a robust protocol for the assessment of OCT4A in other stem cells.

www.nature.com/scientificreports www.nature.com/scientificreports/ It is expressed in various human pluripotent stem cells where it may play a role in the regulation of the cell cycle in pluripotent stem cells and is downregulated upon induction of differentiation 7 .
Eight OCT4 pseudogenes have also been identified, all displaying high homology to OCT4A 8 , and with several of them being transcribed and translated into proteins in various cancer cell lines and tissues 16,17 . Pseudogene 1 (1080 bp) is encoded by OCT4-Pg1 on chromosome 8, pseudogene 3 (1081 bp) is encoded by OCT4-Pg3 on chromosome 12 and pseudogene 4 (1083 bp) is encoded by OCT4-Pg4 on chromosome 1 8 . During differentiation of hESCs, OCT4A expression is silenced through Dnmt3a/3b-dependent methylation of its promoter and may be replaced by the expression of pseudogenes 8 .
Ectopic expression of OCT4A alone is sufficient to revert human fetal neural stem cells 18 , human primary keratinocytes 19 and human amniotic fluid stem cells (hAFSCs) 20,21 to functional pluripotency. For example, we previously showed that the reactivation of OCT4A expression in hAFSCs cultivated in hESC-like conditions is sufficient to progressively revert the cells to the pluripotent state, as evidenced by their ability to form well-differentiated teratomas 20,21 . First described by Prusa et al. in 2003, hAFSCs were later identified as a stem cell type of choice for regenerative applications, due to their fast growth kinetics, small size, and long telomeres 22,23 . Sharing similar features with other tissue-derived fetal stem cells 24,25 , hAFSCs 26 can be easily isolated and expanded from mid-trimester amniotic fluid obtained by routine amniocentesis or from term amniotic fluid obtained at delivery without ethical restriction. However, a number of reports define undifferentiated hAFSCs cultivated in non-hESC-like conditions as sharing features of hESCs and expressing OCT4A 27,28 . If correct, this potentially hinders the clinical use of primary undifferentiated hAFSCs and could render the cells capable of behaving unpredictably and with harm if transplanted in vivo in their undifferentiated state. It is therefore of paramount importance to carefully examine the expression of OCT4A in hAFSCs 14 .
Here, we present a systematic review of the literature to investigate whether published studies of hAFSCs distinguished OCT4A from other OCT4 isoforms. Our findings suggest that previous reports of OCT4A expression in hAFSCs may be due to cross-reaction with other isoforms and/or to a non-specific signal. Using reverse transcription-polymerase chain reaction (RT-PCR), immunocytochemistry and western blotting, we were unable to detect any population of OCT4A+ cells existing within the primary hAFSC population. The findings reported below therefore confirm that hAFSCs, either fresh or frozen, do not express OCT4A.

Results
Systematic review of studies on OCT4A in hAFSCs. OCT4A expression in hAFSCs is a subject of controversy and we believe that paying careful attention when designing primers should clarify this. Since exon 1 is unique to the OCT4A transcript, the forward primer should lie in exon 1 when detecting gene expression using RT-PCR (Fig. 1, Supplementary Fig. 1a), as recommended by Wang et al. 14 . www.nature.com/scientificreports www.nature.com/scientificreports/ The criteria for OCT4A expression includes nuclear but not cytoplasmic localization by immunohistochemistry using antibodies specific for OCT4A, a molecular weight of 48-53 kDa, use of primers that do not amplify any of the pseudogenes for RT-PCR, and use of appropriate positive (pluripotent cells) and negative (differentiated cells or cell line expressing OCT4 transgene such as HEK293T, HeLa and HepG2 cell lines) controls for western blotting, flow cytometry, RT-PCR and immunohistochemistry.
Our systematic review, last updated on 19.11.2018, combined searches from the Web of Science and  MEDLINE and found 1873 publications, 488 of which were duplicates and 1286 were non-relevant papers. Out of  the 114 publications left, 5 were literature reviews, 17 publications were about cells from fetal tissue only, 6 were not in not written in English and 27 were about cellular reprogramming. The remaining 59 publications included 51 reporting positive RT-PCR results for OCT4 (only 45 gave primers information), 6 used flow cytometry and 20 used immunocytochemistry (19 gave antibody information) ( Table 1). Table 2 lists the primers used in the studies identified. Of these, 27 did not use a forward primer lying in exon 1, which is uniquely contained in the 5′ sequence of the OCT4A transcript, but not in the OCT4B and OCT4B1 transcript variants (primer sets 1, 3,5,6,7,8,0,10,12,13,16,18,19,21,22,23,24,25,26,27,28,30 and 31 (Table 2)). Out of the 21 papers using a forward primer lying in exon 1, 20 used primers that also bind to pseudogene sequences (OCT4-Pg1 and OCT4-Pg3 for primer set 2; OCT4-Pg1 only for prier set 4,15,17 and 29; OCT4-Pg1, OCT4-Pg3 and OCT4-Pg4 for primer set 11 and 14) whilst only one publication (PMID 24375948) used primers that detects OCT4A exclusively (primer set 20). Despite this, appropriate positive and negative controls must be used to avoid potential bias. This should be performed in conjunction with the sequence of the amplification product to confirm detection of the isoform OCT4A.
A total of 20 publications reported OCT4 expression using specific antibodies against OCT4 for either flow cytometry, western blotting and/or immunofluorescence applications (listed Table 1, antibodies used listed in Table 3). The description of the methodology from these studies was deemed insufficient during the systematic review. For example, the antibodies used were not always specified, positive and negative controls were not systematically included to confirm antibody specificity, fluorescence was not exclusively present in the nucleus. No study used controls to confirm exclusive nuclear staining and OCT4A-specificicity, for example using a cell line expressing pseudogene OCT4-Pg1. This is particularly important for western blotting since the OCT4A protein has the same molecular weight (39 kDa) as OCT4-Pg1 8 . Similarly, primers amplifying both OCT4A and OCT4-Pg1 (Table 2, primer sets 3, 4, 6, 7, 14, 16 and 18) do not allow determination of OCT4 expression, especially in the absence of immunostaining to confirm nuclear localization. Thus, we concluded that given our current state of knowledge and list of OCT4A expression criteria, positive OCT4A expression in hAFSCs cultivated in non-pluripotent conditions remained inconclusive in all studies listed in Table 1.
Validation of the specificity of OCT4 antibodies to detect the OCT4A isoform immunofluorescence. To examine the specificity of OCT4 antibodies for the OCT4A isoform, we selected the eight antibodies most commonly used to assess OCT4A isoform expression in the literature (listed in Table 4). We used hESCs as positive control since OCT4A is expressed in human pluripotent stem cells. The osteoblast cell line MG63 cells and hAFSCs differentiated down the osteogenic lineage for 3 weeks (Diff cells) were selected as negative controls because OCT4A is downregulated upon differentiation. To confirm OCT4A specificity and the absence of OCT4 pseudogene expression, we used HeLa, HEK293T (293T) and HepG2 cell lines which express pseudogenes Pg1, Pg3 and Pg4 29 , as additional negative controls. We used immunofluorescence, flow cytometry and/or western blotting to determine whether each antibody fulfilled the criteria for specific detection of OCT4A (nuclear localisation, the absence of cytoplasmic staining and a molecular weight of 48-53 kDa as opposed to 39-45 kDa for OCT4B).
We first tested the eight antibodies by immunofluorescence (IF) (Fig. 2). IF staining using sc-5279 and 130-105-606 showed positive nuclear localisation in the positive control hESCs and the absence of staining in all negative controls, confirming the specificity of these antibodies for the nuclear detection of the OCT4A isoform. IF staining with sc-8628 showed nuclear localisation in hESCs but also strong nuclear staining in a subset of HeLa cells, suggesting this antibody detects OCT-Pg1, 3 and/or 4. IF staining with sc-9081, MAB17591 and IC1759P showed positive nuclear staining of hESCs but also positive nuclear staining for all five negative controls, revealing the non-specificity of the antibody for the OCT4A isoform and its cross-reactivity with isoform OCT4B and pseudogenes OCT4-Pg1, 3 and/or 4. IF staining with MAB17591 also showed positive cytoplasmic staining in osteoblast-differentiated hAFSCs and IC1759P also showed positive cytoplasmic staining in HEK293T and osteoblast-differentiated hAFSCs. IF staining with MAB4419 showed positive nuclear staining of hESCs but also positive nuclear staining in HeLa cells and HepG2 cells, and positive cytoplasmic staining in osteoblast-differentiated hAFSCs, indicating the importance of including several negative control cell lines to test for cross-reactivity. IF staining with Ab19857 showed positive nuclear staining in hESCs but also strong positive nuclear and cytoplasmic staining of 293T and MG63 cells and faint cytoplasmic staining in osteoblast-differentiated hAFSCs, HeLa and HepG2 cells. In conclusion, our results indicate that only sc-5279 and 130-105-606 antibodies are suitable for the detection of OCT4A isoform by immunofluorescence.
We next used the antibodies sc-5279 and 130-105-606 to determine whether hAFSCs express OCT4A. Immunofluorescence revealed the absence of staining of both hAFSC subsets SS-hAFSCs and RS-hAFSCs cultivated either in Chang C or D10 medium (Fig. 3a,b). However, these cells have been previously expanded, frozen www.nature.com/scientificreports www.nature.com/scientificreports/ Continued www.nature.com/scientificreports www.nature.com/scientificreports/ and thawed for analysis. We therefore hypothesized that OCT4 might be expressed in freshly-isolated cells and progressively lost during ex vivo expansion or that freshly-isolated populations contain a small number of cells expressing OCT4A that do not undergo clonal expansion. To test this hypothesis, we analysed freshly-isolated passage 1 SS-hAFSCs and RS-hAFSCs cultivated in either D10 or Chang culture medium immediately after isolation that had not been expanded in culture beyond the first passage. Results indicated the absence of staining using the sc-5279 antibody (Fig. 3c) and the 130-105-606 antibody (data not shown) in both cell subsets.
Western blotting. As the sc-5279 antibody is suitable for western blot analysis, we next confirmed the expression of the OCT4A protein isoform in hESCs but its absence in the negative control MG63 cells and in freshly-isolated passage 1 SS-hAFSCs and RS-hAFSCs cultivated in D10 or Chang medium (Fig. 3d), with a faint non-specific band present in all cell lines (Fig. 3d).
Flow cytometry. We next used flow cytometry to confirm the results obtained using immunofluorescence.
We tested the eight different antibodies listed in Table 4, with hESCs as positive control and MG63 cells as negative control. Results showed positive expression in hESCs for all antibodies (Fig. 4). For all antibodies, the peak of fluorescence obtained for the negative control MG63 was distinct from the peak corresponding to the primary antibody-only control, indicating that autofluorescence could be interpreted as false-positive in the absence of positive controls. hAFSCs do not express most pluripotency markers. Since the nuclear OCT4A isoform is exclusively expressed in pluripotent cells, we first assessed the expression of other pluripotency-associated markers in SS-hAFSCs and RS-hAFSCs cultivated either in D10 or Chang medium. We found that REX1 was present in the nucleus of both cell subsets in either culture medium. However, NANOG, SOX2, KLF4 and DNMT3b were only expressed in the positive control (hESCs) but not in hAFSCs cultivated either in D10 or Chang medium (Fig. 5), confirming that both SS and RS hAFSCs do not express pluripotency-associated markers except REX1. hAFSCs express transcripts for OCT4-Pg1 and OCT4-Pg4 but not OCT4A. Although OCT4A protein was not detected in freshly-isolated hAFSCs, there is a possibility that the gene was transcribed but not translated. To test this hypothesis, we used primers specifically amplifying OCT4A, OCT4A-Pg1, OCT4A-Pg3 and OCT4A-Pg4 sequences (sequences in Supplementary Fig. 1b) and assessed gene expression in hESCs (positive control for OCT4A), osteoblast-differentiated hAFSCs and MG63 cells (negative control for OCT4A), HeLa, HepG2, and 293T (positive control for OCT4A-Pg1, OCT4A-Pg3 and OCT4A-Pg4). Results confirmed the expression of OCT4A-Pg1, OCT4A-Pg3 and OCT4A-Pg4 in HeLa, HepG2, and 293T, as well as expression of OCT4A-Pg1 in hESCs, osteoblast-differentiated hAFSCs and MG63 cells and the expression of OCT4A-Pg3 in hAFSCs and in osteoblast-differentiated hAFSCs (Fig. 6). We also confirmed the expression of the OCT4A isoform in hESCs and its absence in all negative control cell lines. Finally, our results confirmed that OCT4A is not transcribed in freshly-isolated (never frozen) hAFSCs, but that OCT4A-Pg1 and OCT4A-Pg4 are transcribed in these populations, which could somehow be related to their plasticity and their ability to reach pluripotency without any genetic manipulation 31 . This may further explain the reports of positive OCT4A expression in hAFSCs.    Continued www.nature.com/scientificreports www.nature.com/scientificreports/ Cell isolation and culture. AF was collected from six different donors as a routine procedure during the mid-trimester (15-18 weeks of gestation) from pregnant women undergoing prenatal diagnosis for possible chromosomal abnormalities. Each AF sample was immediately centrifuged before the cell pellet was resuspended into a single cell suspension in StemMACS expansion media and seeded under xeno-and serum-free conditions at low density (10 2 cells/cm 2 ) onto a plastic culture dish. The cell cultures were then allowed to expand until clones >50 cells formed. Some clones contained round-shaped cells and others spindle-shaped cells. Clones presenting similar morphology were pooled, centrifuged and resuspended into a single cell suspension that was plated (at   www.nature.com/scientificreports www.nature.com/scientificreports/ and HEK293T cells were cultured in D10 as described above. HeLa cell lines were obtained from the American type culture collection (ATCC, Manassas, USA) and cultured in D10.

Methods
Osteogenic differentiation. Cells were seeded at 10,000 cells/cm 2 in D10. When confluent, the expansion medium was replaced with freshly-prepared osteogenic medium (D10 supplemented with 0.2 mM ascorbic acid, 10 mM ß-glycerophosphate and 10 −8 M dexamethasone (all Sigma-Aldrich)). Osteogenic medium was freshly prepared and replaced twice weekly for 21 days.
Flow cytometry. Cells (n = 3 for SS-hAFSCs and n = 3 for RS-hAFSCs) were detached, washed in flow buffer (PBS + 3% BSA, Sigma) and centrifuged at 5000 g for 2 minutes before 1 × 10 5 cells were resuspended in the appropriate primary antibody (listed in Table 3) at its optimal dilution (1:10) in flow buffer and incubated for 1 hour at 4 °C. For unconjugated antibodies, cells were then washed and resuspended in a 1:10 dilution of FITC-conjugated donkey anti-mouse (FITC conjugated donkey anti-rabbit IgG, FITC donkey anti-goat IgG and donkey anti-mouse IgG all from ImmunoResearch) for 30 minutes at 4 °C. Cells were then analysed using a Becton Dickinson FACScalibur flow cytometer (BD Biosciences) using Cell Quest Pro and FlowJo software.   www.nature.com/scientificreports www.nature.com/scientificreports/ Immunofluorescence. Cells (n = 3 for SS-hAFSCs and n = 3 for RS-hAFSCs) were washed, fixed in 4% paraformaldehyde (PFA, Sigma) and permeabilized. Cells were then blocked for 30 min with blocking buffer (PBS supplemented with 2% bovine serum albumin (BSA) and 0.1% Tween) and incubated overnight in the dark with primary antibodies at their optimal dilution, i.e. SC-5279 (Santa Cruz  www.nature.com/scientificreports www.nature.com/scientificreports/

RT-PCR.
Total RNA was extracted from 3-5 × 10 6 cells (n = 3 for SS-hAFSCs and n = 3 for RS-hAFSCs) using the RNeasy Mini RNA kit (Qiagen) and cDNA was synthesised from 1 µg RNA using Pd(N)6 random hexamers (Amersham Pharmacia Biotech) and 1 ml of 200U M-MLV Reverse Transcriptase in the presence of dNTPs (Promega), according to the manufacturer's instructions. The generated cDNA was amplified using the ABI StepOne Sequence Detector system (Applied Biosystems) and primers listed in Supplementary Fig. 1b. The results were then analysed by gel electrophoresis.

Discussion
Mid-trimester amniotic fluid contains two sub-populations of cells that can be distinguished by their differential morphology and immunophenotype. The SS-hAFSC population has huge potential for regenerative medicine 27 . We, and others, have demonstrated therapeutic effects of these cells in mouse models of osteogenesis imperfecta 32 , hypoxic-ischemic encephalopathy 30 or kidney damage 33,34 for example. Moreover, they are currently under assessment in clinical trials targeting osteoarthritis, neuropathy and pelvic pain amongst others 35 . Conflicting reports exist regarding their expression of pluripotency markers. We establish here that the OCT4A isoform, which is exclusively present in the nucleus of pluripotent stem cells, is not expressed in human fetal stem cells isolated from mid-trimester amniotic fluid. Our results highlight the necessity of using appropriate positive controls not only for OCT4A but also for the OCT4 pseudogenes, as well as a combination of approaches to confirm the expression of OCT4A at the RNA and protein levels. With the advancement in the field there is now more robust techniques to address OCTA expression as established for other cell lines 28 . For immunostaining, flow cytometry and western blotting, the specificity of antibodies for OCT4A should be validated by positive expression in pluripotent cells and the absence of expression in multipotent cells, lineage differentiated cells and for cell lines expressing OCT4 pseudogenes. OCT4A protein should also be localized in the nucleus (as assessed by immunostaining) and should be the correct size (as determined by western blotting). For RT-PCR, it is necessary www.nature.com/scientificreports www.nature.com/scientificreports/ to use a forward primer located in OCT4A exon 1, primers that exhibit no specificity for the OCT4 pseudogene sequences, and to validate the specificity for OCT4A using positive and negative controls. To confirm positive expression, it would be strongly advised to sequence the amplification product, thus identifying it as OCT4A.
Critical examination of the studies identified by our systematic review and our robust hAFSC sample analysis led us to conclude that human amniotic fluid does not contain cells expressing OCT4A. However, it remains possible that some studies may have detected positive expression in some samples. Validating OCT4A specificity remains a technical challenge to unequivocally report the existence of a subset of cells present in the amniotic fluid expressing OCT4 or to document OCT4A reactivation upon in vitro expansion. For example, we previously reported that OCT4A expression in human mesenchymal stem cells (MSCs) from fetal-placental tissues was not attributable to different culture conditions, tissue sources, or gestational ages but instead to the techniques used 36 . www.nature.com/scientificreports www.nature.com/scientificreports/ More recently, we showed that the human amniotic fluid contains MSCs of renal origin, and that the presence of these cells increases with gestational age 37 . Despite some cells expressing the pluripotency-associated markers TRA-1-60 and TRA-1-81, OCT4 protein was localized in the cytoplasm and absent in the nucleus.
The possibility that OCT4A expression occurs only in hAFSCs upon fresh isolation has led to the hypothesis that the amniotic fluid might contain a sub-population of pluripotent cells. However, the data from our laboratory suggest that this is not the case, as freshly-isolated cells seeded in culture media that do not maintain pluripotency (D10 and Chang C) failed to show OCT4A positivity. In addition, publications claiming OCT4A expression in hAFSCs have not demonstrated the role of OCT4 in maintaining self-renewal, as it has been done in hESCs using siRNAs 38,39 .
Our results do not allow us to conclude the absence of OCT4A-positive stem cells in the amniotic fluid. Instead, they confirm that OCT4A is not expressed in hAFSCs expanded in vitro in non-pluripotent conditions. Interestingly, we have reported that SS-hAFSCs may be more plastic than their postnatal counterparts since when they are cultivated in pluripotent conditions in the presence of valproic acid (VPA), they up-regulated OCT4A expression and reverted to functional bona fide pluripotency by reactivating the OCT4-downstream pluripotent pathway 27,31,40 . This suggests that despite OCT4A being downregulated, DNA conformation at the epigenetic level might be permissive to OCT4A reactivation. Moreover, it remains possible that samples isolated during the first trimester of pregnancy contain OCT4A + cells. Our findings, if verified by reports of a lack of expression of additional pluripotency markers in hAFSCs along with their inability to form tumours in animal models despite being positive for REX1, could have important implications for the safe clinical use of mid-trimester hAFSCs. www.nature.com/scientificreports www.nature.com/scientificreports/