Towards an Optimized Culture Medium for the Generation of Mouse Induced Pluripotent Stem Cells*

Generation of induced pluripotent stem cells from somatic cells using defined factors has potential relevant applications in regenerative medicine and biology. However, this promising technology remains inefficient and time consuming. We have devised a serum free culture medium termed iSF1 that facilitates the generation of mouse induced pluripotent stem cells. This optimization of the culture medium is sensitive to the presence of Myc in the reprogramming factors. Moreover, we could reprogram meningeal cells using only two factors Oct4/Klf4. Therefore, iSF1 represents a basal medium that may be used for mechanistic studies and testing new reprogramming approaches.

like morphology. Picked colonies were subsequently expanded and maintained as ESCs.
Quantification of Reprogramming Efficiency-For the GR method we counted GFP ϩ colonies under a fluorescent microscope at day 14 postinfection, for iSF1 at day 8 or day 10 postinfection. For confirming these results, infected MEFs were also trypsinized between days 7 and 9 postinfection and then analyzed using a FACS Calibur machine without any gate on the SSC/FSC channels. GFP ϩ cells were gated with a control signal from the PE channel, and a minimum of 10,000 events were recorded. Cells infected with pMX-FLAG were used as negative control.
Quantitative RT-PCR (qPCR)-Total mRNA was isolated using TRIzol, and 2 g was used to synthesize cDNA using ReverTra Ace (Toyobo) and oligo(dT) (Takara). qPCR was performed using Premix Ex Taq TM (Takara) and analyzed with an ABI 7300 machine. Primers sequences are shown in supplemental Table 1.
Teratoma Formation and Blastocyst Injection-Teratomas were produced by injecting 2 million cells subcutaneously into SCID mice. Tumor samples were collected within 4 weeks, fixed in 4% paraformaldehyde, and processed for paraffin embedding and hematoxylin and eosin staining following standard procedures. Chimeras were produced by injecting iPSCs into blastocysts derived from ICR mice, followed by implantation into pseudopregnant ICR mice.
Bisulfite Sequencing-Genomic DNA (700 ng) from various cell lines was exposed overnight to a mixture of 50.6% sodium bisulfite (Sigma) and 10 mM hydroquinone (Sigma). Afterward, a region from the Nanog proximal promoter was amplified by PCR using primers described previously (11). The PCR products were cloned into pMD18-T vector (Takara), propagated in DH5␣, and sequenced.
Whole Genome Expression Analysis-Total RNA was isolated from cells and purified using the RNeasy mini kit (Qiagen, Valencia, CA). Three micrograms of total RNA was used for the reverse transcription reaction primed with T7-Oligo(dT) pro-  (1,10). D indicates day. B, MEFs infected with Oct4/Sox2/Klf4/Myc (OKSM) were cultured in mES medium, fSF1, and GR protocol for 7 days. The percentage of Oct4-GFP ϩ cells was measured by FACS; n ϭ 4. ***, p Ͻ 0.001. C, representative pictures are shown at day 14 postinfection in OKSM-and OKS-infected MEFs with GR strategy compared with infected MEFs in control mES medium. Scale bar, 250 m. D, MEFs infected with OKSM or OKS were cultured in modified fSF1 medium eliminating different components. Oct4-GFP ϩ colonies were counted at day 7 postinfection; n ϭ 3. E, MEFs infected with OKSM or OKS were cultured in modified iSF1 medium containing different basal media. Oct4-GFP ϩ colonies were counted at day 8 postinfection; n ϭ 2. F, growth curve of MEFs in fibroblast medium containing 10% FBS or iSF1 medium is shown; n ϭ 3. G, representative images of OKSM and OKS iPSC colonies produced in iSF1 at day 14 postinfection are shown. Cells were not split on feeders. Scale bar, 500 m. H, MEFs infected with OKSM or OKS were cultured in mES or iSF1. The number of GFP ϩ colonies (green bars) and GFP Ϫ colonies (blue bars) were scored at day 8 postinfection; n ϭ 3. All error bars indicate S.D. OCTOBER 1, 2010 • VOLUME 285 • NUMBER 40 moter (Affymetrix, Santa Clara, CA) using Superscript II (Invitrogen). Biotin-labeled cRNA was synthesized by in vitro transcription using Bioarray RNA Transcript Labeling kit (Affymetrix). After being fragmented, the cRNA were hybridized to a mouse Affymetrix (mouse 430 2.0). Array were scanned with a GeneArray Scanner 7G (Affymetrix). Data in the form of CEL files were background-subtracted and normal-ized with the Robust Multi-chip Average method and ArrayAssist 5.0 software (Stratagene). Data of microarray are available on Gene Expression Omnibus GSE15267.

Optimized Medium for Mouse Somatic Cell Reprogramming
Compound Screening-MEFs were seeded on 12-well plates with 20,000 cells/well and infected as described above. One day after infection various compounds were added until cells were analyzed by FACS. The following compounds were used:

RESULTS
Typically, murine iPSCs can be generated from MEFs in about 2 weeks, and this requires FBS and feeders to reach 0.01ϳ0.5% efficiency (1)(2)(3)(4)6). This inefficient process and undefined culture conditions are a roadblock for both drug screening and mechanistic insights (6,12,13). Thus, it is desirable to use a serum-free medium such as medium containing SR (supplemental Table 2). KSR medium cannot support the proliferation of MEFs (supplemental Fig.  1A). However, iPSC colonies can be obtained efficiently when FBS-containing medium (mES) is switched to KSR at day 4 postinfection (10). After testing candidate factors systematically, we developed a medium (fSF1) containing SR, basic FGF, and N2 (supplemental Table 2). fSF1 allows MEFs (from OG2/Rosa26 mice carrying the Oct4-GFP marker) to grow well and reprogram more efficiently than using mES based on the number of Oct4-GFP ϩ cells detected by FACS (Fig. 1, A and B, and supplemental Fig. 1, A and B). We also tested a GR of fSF1 with KSR ( Fig. 1A), and this GR approach further improved reprogramming as measured by FACS or counting GFP ϩ colonies using either MEFs or skin fibroblasts as donor cells (Fig. 1, B and C, and supplemental Fig. 1, C and D). As previously reported (14), we failed to reprogram MEFs using N2B27 medium supplemented with PD0325901 (MEK inhibitor), CHIR99021 (GSK3 inhibitor), and LIF (supplemental Fig. 1, E and F).
We then simplified the GR protocol by formulating a medium that does not require a gradual replacement. For this we tested the relative contribution of all components present in fSF1 and KSR. As shown in Fig. 1D, we discovered that the selection of basal medium is critical. Of eight basal media screened, we found that DMEM, especially the high glucose DMEM, was the most suitable basal medium for reprogramming ( Fig. 1E and supplemental Fig. 2, A and B). We then replaced the basal medium of fSF1 with high glucose DMEM and renamed it iSF1 (supplemental Table 2). iSF1 supports MEF growth as well as FBS-containing medium ( Fig. 1F) but allows very high efficiency of reprogramming induced either by OKSM (Oct4/Klf4/Sox2/ Myc) or OKS (Oct4/Klf4/Sox2) ( Fig.  1, D, E, and G). When used throughout the reprogramming protocol, iSF1 also improves the reprogramming kinetics over the GR protocol (supplemental Fig. 2C). By adding or eliminating basic FGF in iSF1, we confirmed that basic FGF is important for optimal iPSC generation efficiency (supplemental Fig. 2D). By comparing the number of incomplete reprogrammed colonies (GFP Ϫ ) and GFP ϩ colonies, we found that the improvement induced by iSF1 may differ mechanistically between OKSM and OKS (Fig. 1H). In OKSM-infected MEFs, high efficiency appears to be achieved through the conversion of GFP Ϫ colonies to fully reprogrammed ones, whereas in OKS the total number of colonies increased, and most of them were GFP ϩ , thus suggesting that an increased number of cells have initiated the reprogramming.
To demonstrate that iSF1 also accelerates the conversion into iPSCs we detected the expression of pluripotent markers in the reprogramming process. iSF1 enhances the expression of Nanog and endogenous Oct4 significantly in MEFs infected with OKSM and OKS ( Fig. 2A). iSF1 also allows the appearance of small numbers of GFP ϩ cells at day 4 postinfection that later on transform into GFP ϩ colonies (Fig. 2B). Reac- Values from infected MEFs (isolated at day 6 postinfection) were set to 1. Uninfected MEFs were used as negative control. G, Teratomas differentiated from iPSCs (S5-3C5) containing all three embryonic germ layers: central nervous system (CNS) and skin (ectoderm), muscle and cartilage (mesoderm), columnar epithelium and tubular gland (endoderm). Scale bars, 100 m. H, karyotype of iPSCs produced with iSF1. We examined three iPSC lines, and all of them showed normal karyotype. I, chimeric mice generated using iPSCs produced with iSF1. iPSC-derived cells are responsible for the agouti coat color. The mouse on the right is the control. OCTOBER 1, 2010 • VOLUME 285 • NUMBER 40 tivation of pluripotent genes (such as Nanog and Rex1) can be detected by immunofluorescence in iPSC colonies as early as day 10 postinfection (Fig. 2C). Retroviral silencing has been defined as a late event that indicates full reprogramming (15,16). We used DsRed as a reporter for viral silencing and observed Oct4-GFPϩ/DsRed-colonies emerging at day 9 postinfection (Fig. 2D).

Optimized Medium for Mouse Somatic Cell Reprogramming
iPSC clones established from cultures in iSF1 maintained strong Oct4-GFP expression (Fig. 3A) and were alkaline phosphatase, Nanog, Rex1, and SSEA-1 positive (Fig. 3, A  and B). These cell lines also displayed a demethylated Nanog proximal promoter (Fig. 3C). DNA microarray analysis demonstrated as well that these iPSCs have an expression profile similar to that of mouse ESCs (Fig. 3D). The pluripotent markers Nanog and Rex1 were reactivated and the transgenes potently silenced in all four iPS cell lines tested as measured by qPCR (Fig. 3, E and F). These iPSCs could produce teratomas containing tissues derived from all three germ layers and had normal karyotypes (Fig.  3, G and H). Moreover, when injected into blastocysts, they efficiently produced chimeric mice (Fig. 3I).
Given the above shown results, iSF1 may be useful as a basal medium for screening compounds (13) that enhance or reduce reprogramming or replace one of the reprogramming factors (17). We tested 16 compounds known to affect reprogramming or pluripotency (5, 12, 14, 18 -20), but most of them could not enhance it further in our optimized condition (Fig. 4A). Only trichostatin A and valproic acid, two histone deacetylase inhibitors, were able to increase the reprogramming substantially (Figs. 4A and supplemental Fig. 2, E  and F). Interestingly, we also observed that OKS-infected MEFs are more sensitive to the starting culture conditions, such as density and passage, than OKSM-infected MEFs (Fig. 4, B and C). A possible explanation is that Myc may antag- onize senescence and apoptosis triggered by late passage or low density. On the other hand, the efficiency of OKSM-mediated iPSC generation is more susceptible to changes in the basal medium (Fig. 1E), again suggesting that the reprogramming process induced by OKSM and OKS may follow different routes. In the most optimal condition and using Oct4-GFP ϩ colony number at day 8 postinfection as criterion, the efficiency of reprogramming with OKSM is 2.89 Ϯ 1.05%, and with OKS is 0.595 Ϯ 0.135% (Fig. 4B), which is ϳ300-fold and 500-fold higher than standard mES medium, respectively.
Considering that iSF1 enhances MEFs reprogramming significantly, we wondered whether meningeal cells, which express endogenous Sox2 and reprogram more efficiently than fibroblasts (11), can be transformed into iPSCs with fewer factors in iSF1. Meningeal cells failed to be reprogrammed by Oct4 and Klf4 in standard mES medium (data not shown), which might be because the expression of endogenous Sox2 is lower than neural stem cells (data not shown). However, we could produce iPSCs using only Oct4 and Klf4 in iSF1, whereas no colonies emerged in mES medium in four independent experiments (Fig. 4D). These OK meningeal iPSC clones express Oct4-GFP, SSEA-1, and Nanog, indicating that they are pluripotent (Fig. 4D). Examination of retroviral integration in the genomic DNA confirmed that these clones contained only Oct4 and Klf4 (Fig. 4E). When injected into blastocysts, OK meningeal iPSCs contributed to germ line cells of chimeric mouse as determined by activation of the Oct4-GFP reporter (Fig. 4F). Therefore, optimization of the culture conditions not only accelerates reprogramming, but also reduces the need for some reprogramming factors.

DISCUSSION
Here, we describe an optimized tissue culture condition that allows efficient generation of mouse iPSCs. The combination of SR, basic FGF, and N2 can significantly improve the reprogramming efficiency and also the kinetics. SR contains abundant vitamin C, which enhances reprogramming remarkably especially when added as the more stable 2-phospho-L-ascorbic acid trisodium salt (21). Lipids in SR are also known to regulate ESC self-renewal and might contribute to the reprogramming as well (22). On the other hand, basic FGF and N2 seem to act by promoting cell proliferation (supplemental Fig. 1B), which may allow the acquisition of stochastic changes during the reprogramming (23). SR and basic FGF are routinely used in human iPSC generation, which in general yields much lower efficiencies than the mouse (7,8). This may be related to different requirements of mouse and human somatic cells to achieve the induced pluripotent status.
Compared with other reported reprogramming media, iSF1 represents a good option for several reasons. It eliminates many undefined factors present in serum although some others still remain in SR. For example the lipid-rich albumin is not chemically defined (22), but its elimination can serve as a screening system for formulating a better defined medium. Furthermore, given the speed of the process and the high efficiency, iSF1 may prove very useful for mechanistic research perhaps using enriched populations sorted by FACS. iSF1 may prove useful as well to test new reprogramming techniques, especially nonin-tegrating approaches, as for example we could effectively reprogram meningeal cells using only two factors.
One surprising finding is that the basal medium appears to be of critical importance because DMEM is permissive but other media are not. This offers the possibility of screening among the components of such media for elements that influence the reprogramming. In addition, consistent with previous reports (15,24,25), many incompletely reprogrammed colonies were observed in OKSM-mediated reprogramming even in the presence of iSF1, whereas OKS mostly generated fully reprogrammed iPSCs. This difference suggests that our protocol is more optimal for OKS than OKSM. The latter may help highlight differential roles of Myc during the reprogramming, which may be both positive and negative depending on the context. Our laboratory is currently testing this idea as well as others mentioned above.