alization view of Waddington’s epigenetic landscape, fates of somatic cells are progressively determined as cellular differentiation proceeds, like going downhill. It is generally believed that reversal of differentiated status requires artificial physical or genetic manipulation of nuclear function such as nuclear transfer, or the introduction of multiple transcription factors. Here we investigated the question of whether somatic cells can undergo nuclear reprogramming simply in response to external triggers without direct nuclear manipulation. This type of situation is known to occur in plants—drastic environmental changes can convert mature somatic cells (for example, dissociated carrot cells) into immature blastema cells, from which a whole plant structure, including stalks and roots, develops in the presence of auxins.
A challenging question is whether animal somatic cells have a similar potential that emerges under special conditions. Over the past decade, the presence of pluripotent cells (or closely relevant cell types) in adult tissues has been a matter of debate, for which conflicting conclusions have been reported by various groups. However, no study so far has proven that such pluripotent cells can arise from differentiated somatic cells.
Haematopoietic cells positive for CD45 (leukocyte common antigen) are typical lineage-committed somatic cells that never express pluripotency-related markers such as Oct4 unless they are reprogrammed. We therefore addressed the question of whether splenic CD45+ cells could acquire pluripotency by drastic changes in their external environment such as those caused by simple chemical perturbations.
CD45+ cells were sorted by fluorescence-activated cell sorting (FACS) from the lymphocyte fraction of postnatal spleens (1-week old) of C57BL/6 mice carrying an Oct4-gfp transgene, and were exposed to various types of strong, transient, physical and chemical stimuli (described below). We examined these cells for activation of the Oct4 promoter after culture for several days in suspension using DMEM/F12 medium supplemented with leukaemia inhibitory factor (LIF) and B27 (hereafter called LIF+B27 medium).
Spontaneous neural conversion from salamander animal caps by soaking the tissues in citrate-based acidic medium below pH 6.0 has been demonstrated previously.
Without exposure to the stimuli, none of the cells sorted with CD45 expressed Oct4-GFP regardless of the culture period in LIF+B27 medium. In contrast, a 30-min treatment with low-pH medium (25-min incubation followed by 5-min centrifugation; Fig. 1a; the most effective range was pH 5.4–5.8; Extended Data Fig. 1a) caused the emergence of substantial numbers of spherical clusters that expressed Oct4-GFP in day-7 culture (Fig. 1b). Substantial numbers of GFP+ cells appeared in all cases performed with neonatal splenic cells (n = 30 experiments).
The emergence of Oct4-GFP+ cells at the expense of CD45+ cells was also observed by flow cytometry (Fig. 1c, top, and Extended Data Fig. 1b, c). We next fractionated CD45+ cells into populations positive and negative for CD90 (T cells), CD19 (B cells) and CD34 (haematopoietic progenitors18), and subjected them to low-pH treatment. Cells of these fractions, including T and B cells, generated Oct4-GFP+ cells at an efficacy comparable to unfractionated CD45+ cells (25–50% of surviving cells on day 7), except for CD34+ haematopoietic progenitors19, which rarely produced Oct4-GFP+ cells (<2%; Extended Data Fig. 1d).
Among maintenance media for pluripotent cells, the appearance of Oct4-GFP+ cells was most efficient in LIF+B27 medium, and did not occur in mouse epiblast-derived stem-cell (EpiSC) medium (Extended Data Fig. 1e). The presence or absence of LIF during days 0–2 did not substantially affect the frequency of Oct4-GFP+ cell generation on day 7 (Extended Data Fig. 1f), whereas the addition of LIF during days 4–7 was not sufficient, indicating that LIF dependency started during days 2–4.
Most of the surviving cells on day 1 were still CD45+ and Oct4-GFP−. On day 3, the total cell numbers were reduced to between one-third to one-half of the day 0 population (Fig. 1d; see Extended Data Fig. 1g, h for apoptosis アナリシス), and a substantial number of total surviving cells became Oct4-GFP+ (Fig. 1d), albeit with relatively weak signal intensity. On day 7, a significant number of Oct4-GFP+CD45− cells (one-half to two-thirds of total surviving cells) constituted a distinct population from the Oct4-GFP−CD45− cells (Fig. 1c, top, day 7, and Fig. 1d). No obvious generation of Oct4-GFP+CD45− populations was seen in non-treated CD45+ cells cultured similarly but without low-pH treatment (Fig. 1c, bottom).
Low-pH-treated CD45+ cells, but not untreated cells, gradually turned on GFP signals over the first few days (Fig. 1e, Supplementary Videos 1 and 2 and Extended Data Fig. 2a), whereas CD45 immunoreactivity became gradually reduced in the cells that demonstrated Oct4-GFP expression (Fig. 1f and Extended Data Fig. 2b). By day 5, the Oct4-GFP+ cells attached together and formed clusters by accretion. These GFP+ clusters (but not GFP− cells) were quite mobile and often showed cell processes on moving (Supplementary Video 1).
The Oct4-GFP+ cells demonstrated a characteristic small cell size with little cytoplasm and also showed a distinct fine structure of the nucleus compared with that of parental CD45+ lymphocytes (Fig. 1g). The Oct4-GFP+ cells on day 7 were smaller than non-treated CD45+ cells (Fig. 1g, h and Extended Data Fig. 2c) and embryonic stem (ES) cells (Fig. 1h), both of which are generally considered to be small in size. The diameter of low-pH-treated CD45+ cells became reduced during the first 2 days, even before they started Oct4-GFP expression (Fig. 1f), whereas the onset of GFP expression was not accompanied by cell divisions. Consistent with this, no substantial 5-ethynyl-2′-deoxyuridine (EdU) uptake was observed in the Oct4-GFP+ cells after the stressor (Extended Data Fig. 2d).
The lack of substantial proliferation argues against the possibility that CD45− cells, contaminating as a very minor population in the FACS-sorted CD45+ cells, quickly grew and formed a substantial Oct4-GFP+ population over the first few days after the low-pH treatment. In addition, genomic rearrangements of Tcrb (T-cell receptor gene) were observed in Oct4-GFP+ cells derived from FACS-purified CD45+ cells and CD90+CD45+ T cells (Fig. 1i, lanes 4, 5, and Extended Data Fig. 2e–g), indicating at least some contribution from lineage-committed T cells. Thus, Oct4-GFP+ cells were generated de novo from low-pH-treated CD45+ haematopoietic cells by reprogramming, rather than by simple selection of stress-enduring cells.
expressed pluripotency-related marker proteins (Oct4, SSEA1, Nanog and E-cadherin; Fig. 2a) and marker genes (Oct4, Nanog, Sox2, Ecat1 (also called Khdc3), Esg1 (Dppa5a), Dax1 (Nrob1) and Rex1 (Zfp42); Fig. 2b and Extended Data Fig. 3a) in a manner comparable to those seen in ES cells. Moderate levels of expression of these pluripotency marker genes were observed on day 3 (Fig. 2b and Extended Data Fig. 3b). Notably, the Oct4-GFP+ cells on day 3, but not on day 7, expressed early haematopoietic marker genes such as Flk1 (also called Kdr) and Tal1 (Extended Data Fig. 3c), indicating that Oct4-GFP+ cells on day 3, as judged by their expression pattern at the population level, were still in a dynamic process of conversion.
On day 7, unlike CD45+ cells and like ES cells, low-pH-induced Oct4-GFP+ cells displayed extensive demethylation at the Oct4 and Nanog promoter areas (Fig. 2c), indicating that these cells underwent a substantial reprogramming of epigenetic status in these key genes for pluripotency.
In vitro differentiation assays demonstrated that low-pH-induced Oct4-GFP+ cells gave rise to three-germ-layer derivatives (Fig. 2d) as well as visceral endoderm-like epithelium (Extended Data Fig. 3d). When grafted into mice, low-pH-induced Oct4-GFP+ cell clusters formed teratomas (40%, n = 20) (Fig. 2e and Extended Data Fig. 4a–c) but no teratocarcinomas that persistently contained Oct4-GFP+ cells (n = 50). Because some cellular variation was observed in the signal levels of Oct4-GFP within the clusters, we sorted GFP-strong cells (a major population) and GFP-dim cells (a minor population) by FACS on day 7 and separately injected them into mice. In this case, only GFP-strong cells formed teratomas (Extended Data Fig. 4d).
