Combinatorial programming of human neuronal progenitors using magnetically-guided stoichiometric mRNA delivery

Sayyed M Azimi; Steven D Sheridan; Mostafa Ghannad-Rezaie; Peter M Eimon; Mehmet Fatih Yanik

doi:10.7554/eLife.31922

. 2018 May 1;7:e31922. doi: 10.7554/eLife.31922

Combinatorial programming of human neuronal progenitors using magnetically-guided stoichiometric mRNA delivery

Sayyed M Azimi ¹, Steven D Sheridan ^1,², Mostafa Ghannad-Rezaie ^1,³, Peter M Eimon ¹, Mehmet Fatih Yanik ^1,^3,^✉

Editor: Sacha B Nelson⁴

PMCID: PMC5959718 PMID: 29714688

Abstract

Identification of optimal transcription factor expression patterns to direct cellular differentiation along a desired pathway presents significant challenges. We demonstrate massively combinatorial screening of temporally-varying mRNA transcription factors to direct differentiation of neural progenitor cells using a dynamically-reconfigurable magnetically-guided spotting technology for localizing mRNA, enabling experiments on millimetre size spots. In addition, we present a time-interleaved delivery method that dramatically reduces fluctuations in the delivered transcription factor copy numbers per cell. We screened combinatorial and temporal delivery of a pool of midbrain-specific transcription factors to augment the generation of dopaminergic neurons. We show that the combinatorial delivery of LMX1A, FOXA2 and PITX3 is highly effective in generating dopaminergic neurons from midbrain progenitors. We show that LMX1A significantly increases TH-expression levels when delivered to neural progenitor cells either during proliferation or after induction of neural differentiation, while FOXA2 and PITX3 increase expression only when delivered prior to induction, demonstrating temporal dependence of factor addition.

Research organism: Human

Introduction

The human nervous system contains at least hundreds of distinct subtypes of neurons that are necessary to create complex circuits. Generation of these specific neuronal cell types from progenitor cell populations in vitro is important for basic science, drug screening, and potentially for cell-based therapeutics.

Significant effort has shown that known extracellular trophic factor cocktails, although useful, do not enable sufficient control and specificity to efficiently reprogram progenitor populations to a desired differentiated state (Kriks et al., 2011; Maroof et al., 2013; Shi et al., 2012a; Shi et al., 2012b). It should be possible to systemically screen for transcription factor regimes that steer the fate of differentiating cells by ectopically introducing factors in a combinatorial manner. In fact, the pioneering work of Yamanaka demonstrated the potential of such an approach by reprogramming somatic cells to pluripotency (Takahashi and Yamanaka, 2006).

Although neurons inherit the same genes, cell fate decisions during neurogenesis are mediated through the unique and highly coordinated temporal and spatial expression of hundreds of transcription factors. The efficiency of lineage-specific differentiation can likely be greatly increased if key transcription factor combinations are delivered to cells in a stoichiometrically and temporally optimized manner over the course of differentiation. For instance, while the expression of early genes is necessary for differentiation and expansion of neuronal progenitors, these same factors may lead to malformations if not silenced later. Similarly, terminal differentiation factors might have deleterious effects if introduced too early.

One could envision using plasmid or non-replicative viral vectors to deliver transcription factors transiently, relying on cell division to dilute out undesired factors with time. However, this approach has severe limitations: (1) it affords only crude temporal control over transcription factor levels and stoichiometries while depending on a proliferative cellular state, (2) the copy numbers delivered to the cells are highly variable, and (3) the use of any DNA or integrative viral vector entails the risk of integration-induced mutagenesis leading to functional genetic alterations. Any protocol based on such techniques could not be risk free or cost effectively translatable to clinical practice. Chemically inducible vectors such as doxycycline-regulated systems also face similar challenges.

In vitro synthesized mRNAs represent an important alternative to the use of DNA vectors (Karikó et al., 2005; Karikó and Weissman, 2007; Karikó et al., 2008; Angel and Yanik, 2010; Karikó et al., 2012; Kormann et al., 2011; Mandal and Rossi, 2013; Warren et al., 2010). By suppressing the innate immune response to reduce the toxicity of exogenous RNA, we have previously been able to achieve high expression levels of transcription factors in cells with repeated delivery while avoiding the risk of genomic insertion (Angel and Yanik, 2010). In addition, since the lifetime of mRNA is shorter than that of DNA-based vectors, the temporal availability of transfected RNAs can be much more precisely controlled over repeated delivery.

Although several transcription factor libraries exist, and significant amounts of microarray and in situ hybridization data are now available, even for the relatively well-studied neuron types, testing all possible stoichiometries and temporal delivery patterns of transcription factors suspected to be involved in lineage specification would far exceed the capabilities of existing laboratories and technologies that use even large-scale formats such as 96- or 384- well plates.

To overcome this bottleneck, we developed a technology (Figure 1) that can massively parallelize the combinatorial transfection of nucleic acids in order to rapidly explore vast search spaces of transcription factor stoichiometries and validated it in an RNA-based screen on human neural progenitor cells (NPCs). By focusing transcription factors onto 1.5 mm spots using dynamically reconfigurable magnetic-field patterns, we significantly reduced the footprint of the experiments to enable screening of transcription factor cocktails over large combinatorial spaces. Importantly, with our technology the amount of RNA and transfection reagent required for a given transcription factor is proportional only to the number of spots actually transfected with that factor, rather than to the total number of possible spots in the array (i.e. to the surface area of transfected cells, not to the total surface area of the plate). This is possible because we focus almost all of the RNA in the medium to the target spots using an array of rare-earth magnets (Figure 1c). The number of cell medium changes we use also scale linearly with the number of transcription factors delivered, rather than growing exponentially with the number of combinatorial possibilities. For example, testing every possible combination of just 5 transcription factors results in 31 unique conditions (i.e. all 5 factors alone, 10 possible 2-factor cocktails, 10 possible 3-factor cocktails, 5 possible 4-factor cocktails, and 1 possible 5-factor cocktail) but requires only 5 medium changes when our platform is used.

Standard transfection methods cause considerable fluctuations in the copy numbers of delivered genes, making it impossible to achieve maximal reprogramming efficiency even once optimal transcription factor combinations have been identified. Thus, we also demonstrate here an interleaved RNA transfection technique that dramatically reduces fluctuations in the delivered mRNA copy numbers and stoichiometries. Using our magnetically-guided spotting platform and interleaved transfection protocol, we evaluated the temporal contributions of transcription factor cocktails by treating human NPCs with them during the proliferative stage and/or during the induction of neurogenesis (i.e. after mitogen withdrawal) to generate human dopaminergic neurons with high purity.

Results

Modified mRNA constructs and magnetofection

We constructed mRNA expression vectors to address common issues in mRNA delivery and protein expression, such as vector instability, poor translation, and cytotoxicity (Figure 1a, see also Materials and methods). The mRNA template was constructed with a T7 promoter and was engineered with untranslated regions (UTRs) of highly stable Human β–globin (Hbb) at both 5’ and 3’ ends (Angel and Yanik, 2010), since UTRs play a significant role in mRNA stability (Yu and Russell, 2001; Jiang et al., 2006). Polyadenylated transcripts with a cap-1 structure were synthesized with complete substitution of uridine and cytidine with the modified nucleosides pseudouridine-5-triphosphate (pseudo-UTP) and 5-methylcytidine-5’-triphosphate (5-methyl-CTP), respectively to increase stability and reduce cytotoxicity by evading the innate immune system (Karikó et al., 2005; Karikó and Weissman, 2007; Karikó et al., 2008; Stepinski et al., 2001; Jemielity et al., 2003; Motorin and Helm, 2011).

Magnetofection has been used for DNA and RNA transfection of cultured cells and yields faster and higher transfection efficiencies than most lipid-based transfection methods, especially for hard-to-transfect cell types such as primary neurons (Sapet et al., 2011; Plank et al., 2011). Standard transfection approaches rely on stochastic diffusion and collision of transfection reagents with cells, which is a slow process that transfects cells non-specifically over the entire surface of the cell culture plate (leftmost panels in Figure 1b and Figure 1c). In magnetofection, transfection vehicles complexed with magnetic nanoparticles (e.g. CombiMag) are drawn towards the target cells using an external magnet positioned underneath the plate. However, due to the non-uniform nature of the magnetic field around a single magnet, magnetofection of cells is not spatially limited by the magnet's physical boundaries and therefore results in partial transfection of cells that lie outside the target area (middle panels in Figure 1b and Figure 1c). This is a significant limitation for a screening platform intended to precisely restrict transfection to isolated spots using a compact array of small magnets. Magnetotransfection has been traditionally used only in standard bench-top or low-throughput assays.

Focused delivery of magnetic nanoparticles by dual-magnetic array

To overcome these limitations, we first introduce a simple dual-magnet configuration which focuses the magnetic field to millimeter-size spots in between pairs of magnets positioned above and below the surface of the plate with opposite poles facing one another (rightmost panel in Figure 1b). This configuration enables spatially restricted transfection of cells with minimal background transfection outside the boundaries of the magnets (rightmost panels in Figure 1c). To make an array of such focused magnetic fields, we note that only the bottom magnets need to be small and spatially defined. Therefore, we use large top magnets that cover the entire wells of the plate (e.g. ~22 mm diameter magnets for a 12-well plate) (Figure 1d). This significantly reduces the complexity of setup, requiring us to program the spatial arrangement of only the bottom magnets.

We used neodymium (NdFeB) rare-earth magnets because of their high magnetic field strength compared to other types of magnets. They possess high magnetic anisotropy (i.e. they preferentially align their magnetic moment along an ‘easy’ axis) and high magnetic coercivity (the ability to resist demagnetisation under an external magnetic field). Both properties are highly desirable in our screening platform, where numerous small magnets are positioned in close proximity. In addition to NdFeB magnets, we tested Alnico five magnets and electromagnets and found both to be inadequate due to insufficient magnetic field (electromagnet; B ~ 10 mT) or failure to withstand demagnetisation (Alnico 5). The top magnetic plate is comprised of 12 large 22 mm diameter NdFeB disk magnets (Supplementary file 1) positioned in a 3 × 4 matrix. The precise configuration and dimension of the top magnets is dictated by the desired cell culture plate format (standard 12-well cell culture plates are used in this study). The strength of the magnetic field was measured ~500 mT on the surface of top magnets. The bottom magnetic plate is a matrix of miniature 1.5 mm-diameter NdFeB disk magnets (Supplementary file 1) positioned in close proximity (3 mm center-to-center spacing) inside a compact array [an 8 × 12 cm Teflon substrate accommodating a matrix of 25 × 40 (i.e. 1000) magnets]. The magnetic field density on the surface of each magnet is B ~ 300 mT in isolation and B ~ 100 mT when inside the array.

Programming of the dual-magnet array

In order to drag the mRNA/Lipid/Magnetic particles onto user-defined spot patterns, we program the magnetic field pattern on the substrate by moving each bottom magnet independently closer to (‘active’) or further away from (‘inactive’) the bottom of the cell culture plate (Figure 1d, see also cross-section A-B there). To achieve this, all bottom magnets are first pushed upwards into the active position by a ‘reset-plate’ (via the pins mounted underneath each magnet) (Figure 1d, Step 1). Afterwards, the reset-plate retracts downwards (Figure 1d, Step 2). Since the pins move inside a teflon substrate filled with high viscosity grease, they remain in the elevated active position (5 mm above) after retraction of the reset-plate. Next, selected magnets are pushed down (away from the cell culture plate) using an array of pneumatic actuators located above the magnets (Figure 1d, Step 3). This eliminates the magnetic field of these bottom magnets and prevents the transfection of cells at these spots.

The top pneumatic head is comprised of 24 pneumatic actuators assembled in a 6 × 4 matrix. The actuators are sub-miniature stainless steel 5 mm air cylinders (McMaster Carr) with 1 mm diameter piston pins, a 12 mm stroke length, and a spring return mechanism. By activating the actuator of each piston, the piston pin rapidly pushes the desired magnet down and retracts back. In order to initiate programming, the top pneumatic actuator arm moves down along z-axis until it is in close proximity with bottom magnetic plate (x-y stage sets the initial position of magnetic plate). The pneumatic actuators program (strike) 24 spots simultaneously and then move to the next x-y position. The process automatically proceeds through multiple cycles until all 1000 bottom magnets are programmed. When programming is complete, the top pneumatic actuator arm retracts to its rest position until the next programming event.

Prior to programming the magnets as described above (Steps 1–4 in Figure 1d), the cell culture plate is kept off the platform stage in order allow the pneumatic actuators access to the bottom magnets. After programming, the cell-culture plate containing with the desired mRNA complex, along with the top magnets, is moved into the platform for transfection and afterwards moved out (Steps 5 and six in Figure 1d). Steps 1–6 are then repeated using a new mRNA complex until all transcription factors have been delivered to the specified spots. mRNA transfection of human NPCs under a user-defined pattern is shown in Figure 1e. With respect to cell mobility and cell-cell interactions, the spatial stability of the spots post-transfection and during differentiation could be of concern. However, Figure 1f shows that the GFP-transfected spots of proliferating and differentiating NPCs remain isolated from each other even after 7 days post-transfection and our results in the subsequent sections show the stability of differentiated cell spots even after 17 days (i.e. when the differentiation protocol is completed).

By focally localizing all transfection components (i.e. mRNA, lipid-based transfection reagent, and magnetic particles) onto a very small footprint, our magnetically-guided spotting platform achieves a significant savings in reagents relative to 96-well plate screens. Unless otherwise noted, we deliver 0.6 ng of mRNA per 1.5 mm diameter spot (total surface area per spot: ~1.8 mm²) for all of our experiments. Achieving a comparable level of transfection in a 96-well format (total surface area per well: ~32.2 mm²) requires ~18 fold more reagents. Although the reagent requirements for multiwell plate screens can be further reduced with 384- or 1536-well plate assays, these smaller formats suffer significantly from variation due to edge effects, which also cause problems for even 96-well screens (Lundholt et al., 2003). Our magnetically-guided spotting platform offers the space and reagent savings characteristic of small multiwell assays in larger-format plates, thus ensuring highly uniform culture conditions for accurate comparison of all combinatorial conditions. Uniformity of delivered RNA copy-numbers and stoichiometries

The fact that cell fate decisions are highly dependent on the precise expression levels of a limited set of genes demands that each reprogramming factor in a given cocktail needs to be delivered with minimal cell-to-cell variation to achieve maximal efficiency. Variable expression and low overall efficiency remain major drawbacks of existing DNA transfection protocols, particularly when working with difficult to transfect cell types such as primary and postmitotic cells (Hansson et al., 2015; Landi et al., 2007). DNA-based delivery methods are most vulnerable to low/variable efficiency because transfected molecules must be transcribed in order to be effective and therefore must cross both the plasma membrane and the nuclear envelope. In contrast, mRNA-based methods tend to yield much higher transfection efficiencies since transfected molecules need to cross only the plasma membrane to be expressed. For example, a comparison of plasmid- and mRNA-based transfection approaches in human embryonic stem cell-derived retinal pigmented epithelial cells shows that DNA efficiency is only ~10% while RNA efficiency is ~90% (Hansson et al., 2015). Similarly, an extensive comparison of DNA and RNA transfection protocols using immature dendritic cells suggests that DNA-based approaches can achieve transfection efficiencies of 10–20% whereas RNA-based approaches can achieve efficiencies of 40–80% (Landi et al., 2007). Due to the low expression efficiency of DNA-based transfection methods, the stochastic noise in the expression is large. Higher expression results in correspondingly lower cell-to-cell variation as the Poisson distribution approaches a normal distribution. This is supported by the low variability of RNA transfection results with respect to DNA transfection (Hansson et al., 2015; Landi et al., 2007).

Here, we use our magnetically-guided spotting platform to show that we are able to further reduce variation in the copy number of transfected mRNAs by breaking the transfection process into multiple temporally-segregated transfections, where each transfection is done using a reduced concentration of the transfection complex (Figure 2a). Human neural progenitor cells were cultured in 12-well plates and spotted with mRNAs encoding GFP (Green) or mCherry (Red) using either a standard single transfection process (32 ng of each mRNA) or a quadruple transfection protocol (four transfections consisting of 8 ng of each mRNA per transfection). In both cases the total amount of mRNA delivered to the spots was 64 ng and the medium was changed after each transfection. As shown in Figure 2b, although both methods result in a majority of all cells being transfected, cell-to-cell expression is considerably more variable when using the single-transfection method. To quantify co-localization efficiency, we compared the ratio of mCherry vs. GFP fluorescence intensity for each cell following single-step and multi-step (‘interleaved’) transfection protocols using the Pearson correlation coefficient (PCC) and Mander’s overlap coefficient (MOC) (Figure 2c) (Li et al., 2004; Manders et al., 1993). Although a single transfection results in co-localization (PCC >0.5) with 79% overlap (MOC = 0.79), interleaved transfection improves the PCC to 0.85 and MOC to 0.92, which is indicative of highly uniform and co-localized transfection of cells with both mRNAs. While the interleaved protocol results in superior co-expression efficiency (i.e. transfection uniformity across all cells), the single-transfection method is more than adequate for large-scale screens where the priority is to compare the largest number of transcription factor combinations head-to-head under identical culture conditions while simultaneously minimizing the number of medium changes. The best performing mRNA combinations can then be retested using the interleaved protocol to precisely quantify their maximum efficiency.

Precise dosage control and transfection gradients can be achieved as shown in Figure 2d where NPCs were spotted with GFP and mCherry mRNAs at three different dosages horizontally and vertically, respectively. This resulted in a matrix of spots with varying combinatorial dosages of two mRNAs ranging from 4x/4x (top left spot) to 1x/1x (bottom right spot).

Screening of transcription factors for dopaminergic neurons

Parkinson's disease is one of the most common neurodegenerative disorders resulting from the functional loss of dopaminergic neurons in substantia nigra pars compacta of midbrain (Lees et al., 2009). Dopamine replacement therapy and deep-brain stimulation can improve the quality of life, however the long-term shortcomings of these treatments make the alternative option of cellular replacement attractive. Therefore, deriving dopaminergic neurons in vitro from pluripotent stem cells or progenitors either for cell therapy or for basic research is of critical importance, since other cell resources are quite limited and unreliable. However, deriving sufficiently pure cultures of dopaminergic neurons remains a key challenge.

To directly assess the contributions of many exogenous gene expression cocktails on the efficiency of dopaminergic neuron differentiation, we performed a combinatorial screen of transcription factors delivered by modified mRNA. We used proliferative NPCs, due to their relative ease to expand and culture as monolayers in the presence of growth factor mitogens (bFGF and EGF), and conducted the screen using a pool of midbrain-specific transcription factors reported in the literature to be involved in either the fate specification, survival, or maintenance of ventral midbrain dopaminergic cells (OTX2, LMX1A, FOXA2, ASCL1, NGN2, NURR1, and PITX3) (Hegarty et al., 2013; Abeliovich and Hammond, 2007; Ang, 2006). While OTX2, ASCL1, and NGN2 are essential during early neural development for the correct positioning of isthmus organizer and the development of the midbrain (Martinez-Barbera et al., 2001; Puelles et al., 2003; Kele et al., 2006), FOXA2 and LMX1A are necessary for the correct positioning of dopaminergic cell types along the ventral-caudal axis of the midbrain (Nakatani et al., 2010). PITX3 has been shown to be expressed in all dopaminergic neurons in the CNS during maturation and, together with NURR1, is necessary for their survival and maintenance (Saucedo-Cardenas et al., 1998; Peng et al., 2011; Martinat et al., 2006). Based on these findings, we hypothesized that temporally controlled delivery of these transcription factor combinations might be important for dopaminergic neuron differentiation.

We used two different mediums: (1) an initial expansion medium containing the mitogens bFGF and EGF (proliferative NPC stage) and (2) a subsequent differentiation induction medium lacking the mitogens in order to induce neurogenesis (induction stage; Figure 3a). On day 4, the differentiation medium was supplemented with BDNF, GDNF, and ascorbic acid to improve cell survival, and also with dibutyryl-cAmp and TGF-β3 to enhance differentiation. Neural differentiation media formulated with these supplements have been shown to support dopaminergic differentiation (Kriks et al., 2011; Maroof et al., 2013).

Figure 3. — (a) Timeline of the mRNA-induced differentiation process (proliferative NPC stage from day −2 to 0; early induction stage after day 0). Red arrows indicate time points of transfection before mitogen growth factor removal, transfection after growth factor removal (induction stage), and the final analysis at 17 days after growth factor removal. (b) Immunocytochemistry at day 17 for TH (green) showing upregulation of TH when *FOXA2* is delivered to NPCs (Fo^N), when *PITX3* is delivered to NPCs (Pt^N), and when *LMX1A* is delivered either to NPCs (Lm^N) or during induction (Lm^I). Combinations of these three conditions also increase the number TH⁺ neurons. (c) Double staining (MAP2⁺/TH⁺) results from cells transfected with the most effective combination (Fo^N +Pt^N + Lm^I) compared to non-transfected cells (Diff. Medium). (**d and e**) Quantitative TH gene expression analysis at day 17 comparing all single factors delivered at NPC (^N) and early induction (^I) stages (d), and temporal transfection of the selected combinations of double and triple factors (e). All quantifications were done with n = 3 independent experiments (mean ±s.e.m), ***(p<0.001), **(p<0.01), *(p<0.05) (compared to control, Dunnett's test). Abbreviations: TH, tyrosine hydroxylase; Ot, *OTX2*; Lm, *LMX1A*, Fo, *FOXA2*; As, *ASCL1*; Ng, *NGN2*; Nr, *NURR1*; Pt, *PITX3*; Superscript ‘^N’, proliferative NPC stage; Superscript ‘^I’, induction stage after mitogen withdrawal. Scale bars: 1 mm in (b), 100 µm in (c).

Figure 3—source data 1. Quantitative TH gene expression analysis at day 17 comparing factors delivered at NPC (superscript N) and early induction (superscript I) stages.

elife-31922-fig3-data1.xlsx^{(18.4KB, xlsx)}

DOI: 10.7554/eLife.31922.006

Transcription factors were delivered individually either during the proliferative NPC stage (Day −2, before mitogen removal) or the induction stage (Day 0, upon mitogen removal) as indicated by the superscripts N or I on each factor, respectively. Immunofluorescence results at day 17 indicated a significant increase in tyrosine hydroxylase (TH)⁺ neuron yield only in spots transfected with LMX1A, FOXA2, or PITX3 (Figure 3b and d). Interestingly, among spots transfected during induction stage, LMX1A (Lm^I) was the only factor that significantly increased the yield of TH⁺ neurons on its own (p<0.01, 40 ± 5% TH⁺/MAP2⁺ cells compared to the control, 22 ± 4%). On the other hand, significantly more TH⁺ neurons were counted in spots transfected during the NPC stage with either FOXA2 (Fo^N) (TH⁺/MAP2⁺, 27 ± 4%, p<0.05), PITX3 (Pt^N) (TH⁺/MAP2⁺, 35 ± 3%, p<0.01), or LMX1A (Lm^N) (TH⁺/MAP2⁺, 32 ± 3%, p<0.01), albeit with less effect than post-induction transfection with LMX1A (Lm^I) (Figure 3b). These results suggest that FOXA2 and PITX3 are more efficient when delivered to NPCs whereas LMX1A is more efficient when delivered during induction of differentiation after mitogen removal. Although OTX2, ASCL1, NGN2 are thought to play a role in dopaminergic differentiation, in our experiment their direct overexpression did not increase TH⁺ neurons. Since we used human NPCs, previous results reporting that ASCL1 enhances TH +neurons derived from IPSCs (Theka et al., 2013)may indicate that ASCL1 is only required during pre- or immediate early-neurogenesis stages. OTX2 expression may also have a similar role, since it is known to be necessary for the early patterning of midbrain/hindbrain regional identity.

We next investigated the temporal requirement of the selected transcription factors using combinatorial screening (Fo^N, Pt^N and Lm^I). While FOXA2 (Fo^N) and PITX3 (Pt^N) were only used for pre-induction NPCs, post-induction was performed only with LMX1A (Lm^I) on differentiating cells. Although all dual combinations increased the number of converted cells, triple combination of these factors (Fo^N +Pt^N + Lm^I) produced significantly more TH⁺ cells (p<0.001, 68 ± 5% TH⁺/MAP2⁺) than any other combination (Figure 3c). Gene expression analysis at day 17 of differentiation (Figure 3e) confirms robust seven-fold over-expression of TH in response to the Fo^N +Pt^N + Lm^I triple combination of factors when compared to non-transfected cells.

We then proceeded to characterize the identity of the TH⁺ cells derived from the Fo^N +Pt^N + Lm^I combination in greater detail using gene expression analysis for selected mid-brain specific neuronal makers. We used standard lipid-based transfections (Stemfect reagent; 12-well microplates) for this experiment and performed parallel gene expression and immunocytochemical analysis (staining for DAPI, MAP2, FOXA2 and TH) (Figure 4a and b; a list of TaqMan assays used in this research is given in Supplementary file 2). Neuronal identity (MAP2⁺) was observed in ~98% of both Fo^N +Pt^N + Lm^I transfected cells and the non-transfected controls. Double staining indicated robust expression of FOXA2 in both transfected (p<0.05, 95 ± 2%, FOXA2⁺/DAPI⁺) and non-transfected (70 ± 4%, FOXA2⁺/DAPI⁺) cells, suggesting the differentiated cells in both populations adopted a ventral midbrain identity (Figure 4b). However, transfected cells yielded significantly higher FOXA2⁺ neurons co-expressing TH (p<0.001, 75 ± 4%, TH⁺/FOXA2⁺) compared to non-transfected cells (30 ± 3% FOXA2⁺/TH⁺). Gene expression analysis also confirmed the robust up-regulation of numerous midbrain-specific markers compared to non-transfected controls (Figure 4c). Among the transfected transcription factors, LMX1A was up-regulated by eleven-fold, FOXA2 by nine-fold and PITX3 by five-fold 2 weeks after transfection with the Fo^N +Pt^N + Lm^I combination. Surprisingly, two of the non-effective transcription factors used in our initial screen (OTX2 and NURR1) were also up-regulated by eight- and four-fold respectively. A well-known midbrain regional marker expressed in substantia nigra DA neuron sub-types (VMAT2) was also up-regulated by more than five-fold, further confirming that our optimized triple combination of transcription factors (Fo^N +Pt^N + Lm^I) gives rise to dopaminergic neurons with midbrain identity.

Figure 4. — (a) Immunocytochemistry at day 17 for TH (green), MAP2 (red), FOXA2 (red) and DAPI (blue) performed in microwell plates; comparing triple staining DAPI⁺/MAP2⁺/TH⁺ and DAPI⁺/FOXA2⁺/TH⁺ between non-transfected cells in differentiation medium alone (top) to Fo^N + Pt^N + Lm^I transfected cells (bottom). (b) Quantification of data presented in (a). The percentage of cells expressing FOXA2 increases from 72% in non-transfected cells to 95% in cells transfected with the combination of Fo^N + Pt^N + Lm^I (left panel, FOXA2/DAPI). The percentage of the *FOXA2* positive cells that also co-express TH increases from 32% in non-transfected cells to 75% in cells transfected with the combination of Fo^N + Pt^N + Lm^I (right panel, TH/FOXA2). (c) Gene expression analysis at day 17 for the indicated midbrain specific markers. Quantifications were done with n = 3 independent experiments (mean ± s.e.m), ***p<0.001, **p<0.01, *p<0.05 (compared to control, Student’s t-test). Abbreviations: TH, tyrosine hydroxylase; Lm, *LMX1A*; Fo, *FOXA2*; Pt, *PITX3*; Superscript ‘^N’, proliferative NPC stage; Superscript ‘^I’, induction stage after mitogen withdrawal. Scale bars, 100 µm.

Discussion

During the past decade, there have been significant efforts to generate different cell types, such as dopaminergic neurons, from human pluripotent stem cells (hPSCs) (Kriks et al., 2011; Theka et al., 2013; Friling et al., 2009; Lee et al., 2010), typically by supplementing the cell medium with various growth factors. In addition to growth factor-based differentiation protocols, exogenous overexpression of transcription factors (such as the single transcription factor LMX1A [Friling et al., 2009] or the combined overexpression of FOXA2 and NURR1 [Lee et al., 2010]) using DNA vectors can partially enhance the number of TH positive dopaminergic neurons during differentiation of pluripotent stem cells under various culture conditions. However, directly comparing differentiation efficiencies across laboratories has been impractical due to the differences in experimental protocols. Although various yields and purities of dopaminergic neural differentiation from hPSCs and/or mutipotent NPCs have been reported—ranging from low (i.e. 10%) to high (80%)—much of this variation can be attributed to differences in starting cell lines, growth factor cocktails, use of recombinant vs. purified proteins, cell purification procedures, and/or co-culture with other cells such as astrocytes (Engel et al., 2016). Our high-throughput magnetically-guided spotting platform, where many factors and conditions can be simultaneously tested side-by-side and compared under the same laboratory conditions, can in the future allow for more accurate and quantitative comparisons.

Our high-throughput technology enables the rapid exploration of large combinatorial spaces of transcription factors by massively parallelizing the uniform delivery and transfection of mRNAs to unusually small spot sizes. In the present study, we demonstrate the combinatorial and temporal delivery of a pool of midbrain-specific transcription factors to generate dopaminergic neurons. We show that combinatorial delivery of LMX1A, FOXA2, and PITX3 is highly effective in generating dopaminergic neurons from neural progenitors, with LMX1A significantly increasing TH-expression levels when delivered during both the proliferating NPC stage and in the early neural induction stages upon mitogen withdrawal, while FOXA2 and PITX3 only exhibit high efficacy before neural induction.

Materials and methods

Key resources table.

Reagent type (species) or resource	Designation	Source or reference	Identifiers
Gene (Homo sapiens)	OTX2	NA	HGNC:8522
Gene (H. sapiens)	LMX1A	NA	HGNC:6653
Gene (H. sapiens)	FOXA2	NA	HGNC:5022
Gene (H. sapiens)	ASCL1	NA	HGNC:738
Gene (H. sapiens)	NGN2	NA	HGNC:13805
Gene (H. sapiens)	NURR1	NA	HGNC:7981
Gene (H. sapiens)	PITX3	NA	HGNC:9006
Cell line (H. sapiens)	hNP1	EMD Millipore	EMD Millipore:SCR055; RRID:CVCL_GS51
Antibody	anti-FOXA2 (mouse monoclonal)	Abcam	Abcam:ab60721
Antibody	anti-MAP2 (chicken polyclonal)	Abcam	Abcam:ab5392
Antibody	anti-Tyrosine Hydroxylase (rabbit polyclonal)	Abcam	Abcam:ab112
Sequence-based reagent	VMAT2 (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs00996835_m1
Sequence-based reagent	GAPDH (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs02758991_g1
Sequence-based reagent	OTX2 (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs00222238_m1
Sequence-based reagent	PITX3 (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs01013935_g1
Sequence-based reagent	FOXA2 (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:HS00232764_m1
Sequence-based reagent	LMX1A (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs00892663_m1
Sequence-based reagent	TH (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs0016594_m1
Sequence-based reagent	MAP2 (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs00258900_m1
Sequence-based reagent	NURR1 (TaqMan assay)	Thermo Fisher Scientific	Thermo Fisher Scientific:Hs00443062_g1
Other	DAPI stain	Thermo Fisher Scientific

Open in a new tab

Cell culture

The H9-derived human neural progenitor cell line (hNP1) was obtained from Aruna Biomedical (Athens, GA; now distributed as ENStem-A, EMD Millipore #SCR055). Each lot of ENStem-ATM Human Neural Progenitor Cells has been validated for high levels of expression of Nestin and Sox2 and low level expression of Oct4. The ability of ENStem-ATM cells to differentiate into multiple neuronal phenotypes and maintain a normal karyotype after multiple passages has been verified by the manufacturer, and the cells have been confirmed to be negative for mycoplasma. Cells were expanded on Matrigel-coated (Corning Inc., Corning, NY) six-well plates with the supplier’s expansion medium. Starting from passage two, the expansion medium was gradually changed (25% medium replacement per passage) to N2/B27 in DMEM/F12 supplied with bFGF/EGF (20 ng/ml, Thermo Fisher Scientific, Waltham, MA). Cells were passaged at the split ratio of 1:2 by cell scraper. All experiments were done with cells in passages 7–11.

Synthesis of modified synthetic transcription factor mRNAs

Our mRNA synthesis methodology has been previously described (Angel and Yanik, 2010). Briefly, dsDNA templates were linearized from cDNA clones in pCMV6 vectors for OTX2, LMX1A, FOXA2, ASCL1, NGN2, NURR1, PITX3, EGFP, and mCherry (OriGene, Rockville, MD). To maximize the stability of mRNA transcripts and increase protein translation, we added 5’ and 3’ untranslated regions from human Beta-globin gene (Hbb) to the templates by ligating with E. coli DNA ligase (New England Biolabs, Ipswich, MA). A T7 promoter was also added to the 5’ UTR to facilitate in vitro transcription. Assembled templates were cloned into the pCR-Blunt II TOPO vector and inserted in TOP10 chemically competent E Coli to propagate using the Zero Blunt TOPO PCR Cloning Kit (Thermo Fisher Scientific). Plasmids were purified using EndoFree Plasmid Maxi Kit (Qiagen, Hilden, Germany). Templates were linearized from plasmids then amplified by high-fidelity PCR using KAPA HiFi PCR Kit (KAPA Biosystems, Wilmington, MA). PCR products were separated by agarose gel electrophoresis and purified using QIAquick Gel Extraction Kit (Qiagen). Cap 1-capped, poly(A) tailed mRNA was synthesized from the purified templates by in vitro transcription using the mScript mRNA Production Kit (CellScript, Madison, WI). All mRNAs were synthesized with complete substitution of uridine and cytidine with the modified nucleosides pseudouridine-5-triphosphate (pseudo-UTP) and 5-methylcytidine-5’-triphosphate (5-methyl-CTP), respectively. The mRNAs were purified using an RNeasy Mini kit (Qiagen). To ensure the transcripts were produced with right poly(A) tail, we analysed the samples both before and after tailing using formaldehyde-agarose gel electrophoresis. SUPERase In RNase Inhibitor (Thermo Fisher Scientific) was added to mRNAs at concentration of 1 ug/20 ug.

RNA transfection and differentiation

For directed differentiation experiments without spotting, cells from passages 7–11 were plated on 12-well plates in expansion medium (DMEM/F12/N2/B27/bFGF/EGF). For temporal analysis, growth factors (bFGF and EGF) were kept in the medium for two additional days, which led us to seed the cells at the appropriate split ratio to reach the optimal surface confluency of 30–40% the next day and final cell density of 90% after removal of growth factors. For mRNA transfection, we used lipid-based Stemfect RNA transfection kit (Stemgent, Cambridge, MA) and complexed it with each RNA separately with 10 min of incubation. For all transfections, 100 ng total RNA was complexed with 1 µl Stemfect reagent in 20 µl Stemfect buffer. Cell culture medium was changed to N2/B27 without growth factors, and complexed RNA was resuspended in the medium followed by 4 hr of incubation. After transfection, for transfection at the proliferative NPCs stage (^N), the medium was changed back to N2/B27/bFGF/EGF for two more days. For transfection post-induction (^I), the medium was changed to B27 without growth factors 4 hr prior to transfection, and followed by the addition of RNA/lipid complex, 4 hr of incubation, and changing to fresh B27 medium. For both pre- and post-neural induction (with or without mitogens, respectively) transfection protocols, differentiation was started 72 hr after plating the cells by removing the growth factors and changing the medium to B27. The medium was changed every other day. On day 4, medium was supplemented with brain-derived neurotrophic factor (BDNF, 20 ng/ml; R and D Systems, Minneapolis, MN), glial-derived neurotrophic factor (GDNF, 20 ng/ml; R and D Systems), ascorbic acid (0.2 mM; Tocris), dibutyryl cyclic adenosine monophosphate (Dc-Amp, 0.5 mM; Sigma-Aldrich, St. Louis, MO), and transforming growth factor type β3 (TGF-β3, 1 ng/ml; R and D Systems). Cells were cultured in this medium until they were ready for end-point assays at day 17 post-differentiation, and were either fixed for image analysis or collected for qRT-PCR analysis.

Magnetic spotting methodology

For magnetofection of cells with spotted synthetic mRNA, cells from passages 7–11 were transferred to 12-well glass-bottom plates (Cellvis, Mountain View, CA) at optimised density of 30–40% in expansion medium (DMEM/F12/N2/B27/bFGF/EGF). With the 12-well plate format, a total number of 36 magnetic hotspots could be accommodated under each well. In order to deliver the RNA to the magnetic hotspots, we used CombiMag magnetofection transfection reagent (OZBiosciences, Marseille, France). CombiMag contains magnetic nanoparticles that can be mixed with all transfection reagents and improves their efficiency by means of magnetic field. We mixed mRNA with magnetic nanoparticles and complexed it with Stemfect reagent during a 5–10 min incubation according to the manufacturers’ recommended ratios. Localized transfection was initiated by replacing the medium in the wells with N2/B27 medium in which the complexed mRNA/lipid/nanoparticles had been resuspended. The concentration of the mRNA complex was adjusted based on the number of spots to be transfected per well (0.6 ng RNA per 1.5 mm diameter spot), ensuring that each spot received an equivalent amount of mRNA. Plates were then moved onto the spotting system with the desired magnetic patterns pre-programmed as described. The incubation time for magnetofection was 2 min per transfection. This transfection process was repeated for each mRNA factor until all desired combination were delivered to the spots. The medium was changed between each transfection with new N2/B27 containing the appropriate mRNA complex. The number of medium changes scale linearly with the number of transcription factors delivered to each well. For temporal transfection during the proliferative NPC stage, growth factors were added to the medium after the final round of spotting. For differentiation, the medium was changed to B27 with the remainder of the protocol similar to the standard differentiation protocols discussed in the previous section.

Immunocytochemistry

Cell were fixed in 4% paraformaldehyde in TBST for 15 min, permeabilized for 10 min in 0.2% Triton X-100, blocked with 1% casein for 60 min, and incubated overnight at 4°C with appropriate antibodies in 50/50 1% casein/TBST. Fixed cells were then incubated with secondary antibodies for 60 min (Alexa Fluor 488, 555 and 647 Dye, Molecular Probes, Eugene, OR). DAPI (Thermo Fisher Scientific) was used as nuclear counterstain. Antibodies used are as follows: mouse Foxa2 (ab60721, Abcam, Cambridge, MA), chicken MAP2 (ab5392, Abcam) and rabbit TH (ab112, Abcam)

Gene expression analysis (qRT-PCR)

Total RNA was collected and purified using an RNeasy Mini Kit (Qiagen), measured using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific), and gene expression analysis was performed using commercially available TaqMan gene expression assay (Applied Biosystems, Foster City, CA; a list of TaqMan assays is given in Supplementary file 2). The qRT-PCR was done in one-step, 20 µl reactions with 15 min reverse transcription at 50°C, 2 min initial denaturing at 95°C followed by 40 cycles of 15 s/95°C and 1 min/60°C. Three individual samples with three replicates each were used for gene expression analysis, and the data were normalized to GAPDH.

Cell counting and statistical analysis

For image acquisition, we used a high-performance laser-based confocal imaging system (INCell 6000, GE Healthcare, Chicago, IL). For non-spotting experiments, a total of 27 random images were taken for each condition from three independent experiments. For spotting experiments, images were taken from spotted areas in three independent experiments. Cell counting was performed using CellProfiler (Carpenter et al., 2006). To count the cells, we first counted the number of DAPI-positive cells, followed by counting the number of cells expressing the marker of interest. Student's t-test (comparing two groups) was used for statistical analysis. Co-localization analysis was performed using the WCIF-ImageJ software package and the Image Correlation Analysis plugin (Schneider et al., 2012). The plugin uses Pearson’s Correlation Coefficient (PCC) for quantifying the correlation between two channels, as well as calculating Mander’s Overlap Coefficient (MOC).

Code availability

LabView files for programming of the dual-magnet array are available online from GitHub (https://github.com/rezaie99/ELIFE-050518; copy archived at https://github.com/elifesciences-publications/Yanik_et_al_2018) (Ghannad-Rezaie, 2018).

Acknowledgement

This project was funded by NIH Director’s Pioneer Award (DP1 OD006782) and Packard Award for Science and Engineering.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Mehmet Fatih Yanik, Email: yanik@ethz.ch.

Sacha B Nelson, Brandeis University, United States.

Funding Information

This paper was supported by the following grants:

NIH Office of the Director NIH Director's Pioneer Award 1DP1OD006782 to Mehmet Fatih Yanik.
David and Lucile Packard Foundation Packard Fellowship for Science and Engineering to Mehmet Fatih Yanik.

Additional information

Competing interests

No competing interests declared.

Author contributions

Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing, Designed and performed the experiments, Analyzed the results, Was involved in the writing of the manuscript.

Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Provided advice and training to SM Azimi, Participated in the design and analysis of the experiments, Was involved in the writing of the manuscript.

Methodology, Writing—original draft, Writing—review and editing, Developed the spotting hardware with SM Azimi, Was involved in the writing of the manuscript.

Supervision, Methodology, Writing—review and editing, Worked on developing RNA-mediated differentiation protocols and on editing and preparing the manuscript for publication.

Conceptualization, Resources, Supervision, Funding acquisition, Investigation, Writing—original draft, Project administration, Writing—review and editing, Was the Principle Investigator, Conceived the study, Worked on analysis of the experiments, and was involved in the writing of the manuscript.

Additional files

Supplementary file 1. List and details of top and bottom magnets.

elife-31922-supp1.docx^{(48.9KB, docx)}

DOI: 10.7554/eLife.31922.008

Supplementary file 2. List of TaqMan qRT-PCR assays used in this research.

elife-31922-supp2.docx^{(56KB, docx)}

DOI: 10.7554/eLife.31922.009

Transparent reporting form

elife-31922-transrepform.docx^{(1.1MB, docx)}

DOI: 10.7554/eLife.31922.010

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files.

References

Abeliovich A, Hammond R. Midbrain dopamine neuron differentiation: factors and fates. Developmental Biology. 2007;304:447–454. doi: 10.1016/j.ydbio.2007.01.032. [DOI] [PubMed] [Google Scholar]
Ang SL. Transcriptional control of midbrain dopaminergic neuron development. Development. 2006;133:3499–3506. doi: 10.1242/dev.02501. [DOI] [PubMed] [Google Scholar]
Angel M, Yanik MF. Innate immune suppression enables frequent transfection with RNA encoding reprogramming proteins. PLoS One. 2010;5:e11756. doi: 10.1371/journal.pone.0011756. [DOI] [PMC free article] [PubMed] [Google Scholar]
Carpenter AE, Jones TR, Lamprecht MR, Clarke C, Kang IH, Friman O, Guertin DA, Chang JH, Lindquist RA, Moffat J, Golland P, Sabatini DM. CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome Biology. 2006;7:R100. doi: 10.1186/gb-2006-7-10-r100. [DOI] [PMC free article] [PubMed] [Google Scholar]
Engel M, Do-Ha D, Muñoz SS, Ooi L. Common pitfalls of stem cell differentiation: a guide to improving protocols for neurodegenerative disease models and research. Cellular and Molecular Life Sciences. 2016;73:3693–3709. doi: 10.1007/s00018-016-2265-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
Friling S, Andersson E, Thompson LH, Jönsson ME, Hebsgaard JB, Nanou E, Alekseenko Z, Marklund U, Kjellander S, Volakakis N, Hovatta O, El Manira A, Björklund A, Perlmann T, Ericson J. Efficient production of mesencephalic dopamine neurons by Lmx1a expression in embryonic stem cells. PNAS. 2009;106:7613–7618. doi: 10.1073/pnas.0902396106. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ghannad-Rezaie M. Rezaie99. Github. 2018 https://github.com/rezaie99/ELIFE-050518
Hansson ML, Albert S, González Somermeyer L, Peco R, Mejía-Ramírez E, Montserrat N, Izpisua Belmonte JC. Efficient delivery and functional expression of transfected modified mRNA in human embryonic stem cell-derived retinal pigmented epithelial cells. Journal of Biological Chemistry. 2015;290:5661–5672. doi: 10.1074/jbc.M114.618835. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hegarty SV, Sullivan AM, O'Keeffe GW. Midbrain dopaminergic neurons: a review of the molecular circuitry that regulates their development. Developmental Biology. 2013;379:123–138. doi: 10.1016/j.ydbio.2013.04.014. [DOI] [PubMed] [Google Scholar]
Jemielity J, Fowler T, Zuberek J, Stepinski J, Lewdorowicz M, Niedzwiecka A, Stolarski R, Darzynkiewicz E, Rhoads RE. Novel "anti-reverse" cap analogs with superior translational properties. Rna. 2003;9:1108–1122. doi: 10.1261/rna.5430403. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jiang Y, Xu XS, Russell JE. A nucleolin-binding 3' untranslated region element stabilizes beta-globin mRNA in vivo. Molecular and Cellular Biology. 2006;26:2419–2429. doi: 10.1128/MCB.26.6.2419-2429.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
Karikó K, Buckstein M, Ni H, Weissman D. Suppression of RNA recognition by Toll-like receptors: the impact of nucleoside modification and the evolutionary origin of RNA. Immunity. 2005;23:165–175. doi: 10.1016/j.immuni.2005.06.008. [DOI] [PubMed] [Google Scholar]
Karikó K, Muramatsu H, Keller JM, Weissman D. Increased erythropoiesis in mice injected with submicrogram quantities of pseudouridine-containing mRNA encoding erythropoietin. Molecular therapy : the journal of the American Society of Gene Therapy. 2012;20:948–953. doi: 10.1038/mt.2012.7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Karikó K, Muramatsu H, Welsh FA, Ludwig J, Kato H, Akira S, Weissman D. Incorporation of pseudouridine into mRNA yields superior nonimmunogenic vector with increased translational capacity and biological stability. Molecular therapy : the journal of the American Society of Gene Therapy. 2008;16:1833–1840. doi: 10.1038/mt.2008.200. [DOI] [PMC free article] [PubMed] [Google Scholar]
Karikó K, Weissman D. Naturally occurring nucleoside modifications suppress the immunostimulatory activity of RNA: implication for therapeutic RNA development. Current Opinion in Drug Discovery & Development. 2007;10:523–532. [PubMed] [Google Scholar]
Kele J, Simplicio N, Ferri AL, Mira H, Guillemot F, Arenas E, Ang SL. Neurogenin 2 is required for the development of ventral midbrain dopaminergic neurons. Development. 2006;133:495–505. doi: 10.1242/dev.02223. [DOI] [PubMed] [Google Scholar]
Kormann MS, Hasenpusch G, Aneja MK, Nica G, Flemmer AW, Herber-Jonat S, Huppmann M, Mays LE, Illenyi M, Schams A, Griese M, Bittmann I, Handgretinger R, Hartl D, Rosenecker J, Rudolph C. Expression of therapeutic proteins after delivery of chemically modified mRNA in mice. Nature Biotechnology. 2011;29:154–157. doi: 10.1038/nbt.1733. [DOI] [PubMed] [Google Scholar]
Kriks S, Shim JW, Piao J, Ganat YM, Wakeman DR, Xie Z, Carrillo-Reid L, Auyeung G, Antonacci C, Buch A, Yang L, Beal MF, Surmeier DJ, Kordower JH, Tabar V, Studer L. Dopamine neurons derived from human ES cells efficiently engraft in animal models of Parkinson's disease. Nature. 2011;480:547–551. doi: 10.1038/nature10648. [DOI] [PMC free article] [PubMed] [Google Scholar]
Landi A, Babiuk LA, van Drunen Littel-van den Hurk S. High transfection efficiency, gene expression, and viability of monocyte-derived human dendritic cells after nonviral gene transfer. Journal of Leukocyte Biology. 2007;82:849–860. doi: 10.1189/jlb.0906561. [DOI] [PubMed] [Google Scholar]
Lee HS, Bae EJ, Yi SH, Shim JW, Jo AY, Kang JS, Yoon EH, Rhee YH, Park CH, Koh HC, Kim HJ, Choi HS, Han JW, Lee YS, Kim J, Li JY, Brundin P, Lee SH. Foxa2 and Nurr1 synergistically yield A9 nigral dopamine neurons exhibiting improved differentiation, function, and cell survival. Stem Cells. 2010;28:501–512. doi: 10.1002/stem.294. [DOI] [PubMed] [Google Scholar]
Lees AJ, Hardy J, Revesz T. Parkinson's disease. Lancet. 2009;373:2055–2066. doi: 10.1016/S0140-6736(09)60492-X. [DOI] [PubMed] [Google Scholar]
Li Q, Lau A, Morris TJ, Guo L, Fordyce CB, Stanley EF. A syntaxin 1, Galpha(o), and N-type calcium channel complex at a presynaptic nerve terminal: analysis by quantitative immunocolocalization. Journal of Neuroscience. 2004;24:4070–4081. doi: 10.1523/JNEUROSCI.0346-04.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lundholt BK, Scudder KM, Pagliaro L. A simple technique for reducing edge effect in cell-based assays. Journal of Biomolecular Screening. 2003;8:566–570. doi: 10.1177/1087057103256465. [DOI] [PubMed] [Google Scholar]
Mandal PK, Rossi DJ. Reprogramming human fibroblasts to pluripotency using modified mRNA. Nature Protocols. 2013;8:568–582. doi: 10.1038/nprot.2013.019. [DOI] [PubMed] [Google Scholar]
Manders EMM, Verbeek FJ, Aten JA. Measurement of co-localization of objects in dual-colour confocal images. Journal of Microscopy. 1993;169:375–382. doi: 10.1111/j.1365-2818.1993.tb03313.x. [DOI] [PubMed] [Google Scholar]
Maroof AM, Keros S, Tyson JA, Ying SW, Ganat YM, Merkle FT, Liu B, Goulburn A, Stanley EG, Elefanty AG, Widmer HR, Eggan K, Goldstein PA, Anderson SA, Studer L. Directed differentiation and functional maturation of cortical interneurons from human embryonic stem cells. Cell Stem Cell. 2013;12:559–572. doi: 10.1016/j.stem.2013.04.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
Martinat C, Bacci JJ, Leete T, Kim J, Vanti WB, Newman AH, Cha JH, Gether U, Wang H, Abeliovich A. Cooperative transcription activation by Nurr1 and Pitx3 induces embryonic stem cell maturation to the midbrain dopamine neuron phenotype. PNAS. 2006;103:2874–2879. doi: 10.1073/pnas.0511153103. [DOI] [PMC free article] [PubMed] [Google Scholar]
Martinez-Barbera JP, Signore M, Boyl PP, Puelles E, Acampora D, Gogoi R, Schubert F, Lumsden A, Simeone A. Regionalisation of anterior neuroectoderm and its competence in responding to forebrain and midbrain inducing activities depend on mutual antagonism between OTX2 and GBX2. Development. 2001;128:4789–4800. doi: 10.1242/dev.128.23.4789. [DOI] [PubMed] [Google Scholar]
Motorin Y, Helm M. RNA nucleotide methylation. Wiley interdisciplinary reviews. RNA. 2011;2:611–631. doi: 10.1002/wrna.79. [DOI] [PubMed] [Google Scholar]
Nakatani T, Kumai M, Mizuhara E, Minaki Y, Ono Y. Lmx1a and Lmx1b cooperate with Foxa2 to coordinate the specification of dopaminergic neurons and control of floor plate cell differentiation in the developing mesencephalon. Developmental Biology. 2010;339:101–113. doi: 10.1016/j.ydbio.2009.12.017. [DOI] [PubMed] [Google Scholar]
Peng C, Aron L, Klein R, Li M, Wurst W, Prakash N, Le W. Pitx3 is a critical mediator of GDNF-induced BDNF expression in nigrostriatal dopaminergic neurons. Journal of Neuroscience. 2011;31:12802–12815. doi: 10.1523/JNEUROSCI.0898-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
Plank C, Zelphati O, Mykhaylyk O. Magnetically enhanced nucleic acid delivery. Ten years of magnetofection-progress and prospects. Advanced Drug Delivery Reviews. 2011;63:1300–1331. doi: 10.1016/j.addr.2011.08.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
Puelles E, Acampora D, Lacroix E, Signore M, Annino A, Tuorto F, Filosa S, Corte G, Wurst W, Ang SL, Simeone A. Otx dose-dependent integrated control of antero-posterior and dorso-ventral patterning of midbrain. Nature Neuroscience. 2003;6:453–460. doi: 10.1038/nn1037. [DOI] [PubMed] [Google Scholar]
Sapet C, Laurent N, de Chevigny A, Le Gourrierec L, Bertosio E, Zelphati O, Béclin C. High transfection efficiency of neural stem cells with magnetofection. BioTechniques. 2011;50:187–189. doi: 10.2144/000113628. [DOI] [PubMed] [Google Scholar]
Saucedo-Cardenas O, Quintana-Hau JD, Le WD, Smidt MP, Cox JJ, De Mayo F, Burbach JP, Conneely OM. Nurr1 is essential for the induction of the dopaminergic phenotype and the survival of ventral mesencephalic late dopaminergic precursor neurons. PNAS. 1998;95:4013–4018. doi: 10.1073/pnas.95.7.4013. [DOI] [PMC free article] [PubMed] [Google Scholar]
Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nature Methods. 2012;9:671–675. doi: 10.1038/nmeth.2089. [DOI] [PMC free article] [PubMed] [Google Scholar]
Shi Y, Kirwan P, Livesey FJ. Directed differentiation of human pluripotent stem cells to cerebral cortex neurons and neural networks. Nature Protocols. 2012a;7:1836–1846. doi: 10.1038/nprot.2012.116. [DOI] [PubMed] [Google Scholar]
Shi Y, Kirwan P, Smith J, Robinson HP, Livesey FJ. Human cerebral cortex development from pluripotent stem cells to functional excitatory synapses. Nature Neuroscience. 2012b;15:477–486. doi: 10.1038/nn.3041. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stepinski J, Waddell C, Stolarski R, Darzynkiewicz E, Rhoads RE. Synthesis and properties of mRNAs containing the novel "anti-reverse" cap analogs 7-methyl(3'-O-methyl)GpppG and 7-methyl (3'-deoxy)GpppG. Rna. 2001;7:1486–1495. [PMC free article] [PubMed] [Google Scholar]
Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell. 2006;126:663–676. doi: 10.1016/j.cell.2006.07.024. [DOI] [PubMed] [Google Scholar]
Theka I, Caiazzo M, Dvoretskova E, Leo D, Ungaro F, Curreli S, Managò F, Dell'Anno MT, Pezzoli G, Gainetdinov RR, Dityatev A, Broccoli V. Rapid generation of functional dopaminergic neurons from human induced pluripotent stem cells through a single-step procedure using cell lineage transcription factors. STEM CELLS Translational Medicine. 2013;2:473–479. doi: 10.5966/sctm.2012-0133. [DOI] [PMC free article] [PubMed] [Google Scholar]
Warren L, Manos PD, Ahfeldt T, Loh YH, Li H, Lau F, Ebina W, Mandal PK, Smith ZD, Meissner A, Daley GQ, Brack AS, Collins JJ, Cowan C, Schlaeger TM, Rossi DJ. Highly efficient reprogramming to pluripotency and directed differentiation of human cells with synthetic modified mRNA. Cell Stem Cell. 2010;7:618–630. doi: 10.1016/j.stem.2010.08.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yu J, Russell JE. Structural and functional analysis of an mRNP complex that mediates the high stability of human beta-globin mRNA. Molecular and Cellular Biology. 2001;21:5879–5888. doi: 10.1128/MCB.21.17.5879-5888.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.31922.016

Decision letter

Editor: Sacha B Nelson¹

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Combinatorial Programming of Human Neuronal Progenitors Using Magnetically-Guided Stoichiometric mRNA Delivery" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we wish to convey that we find the manuscript of potential interest for eLife, but we need to reject the manuscript in its present form because there are too many unresolved technical issues. These issues have been discussed separately in both reviews. Should you be able to address these substantial concern, we would be interested in receiving a revised version of the manuscript.

Reviewer #1:

The manuscript by Azimi et al. describes the development of a new technique for focused transfection of cultured cells that is compatible with combinatorial gene expression studies. They demonstrate the utility of this approach by testing combinations of transcription factors known to be involved in the specification of ventral midbrain dopaminergic neurons. They identified LMX1A, PITX3 and FOXA2 as the most optimal combination of factors and show that PITX3 and FOXA2 factors need to be delivered to dividing progenitors while LMX1A can be delivered to postmitotic neurons. While these three factors were previously shown to promote dopaminergic neuronal identity, this study adds by probing temporal requirements of individual factors.

The major concern with the described method is that it requires the preparation of transfection reagents in large volumes as the entire plate is filled with the solution while only few spots are being transfected. Besides being costly, it might be challenging to perform multiple complete media replacements for truly large scale combinatorial analysis of gene function without disturbing the plated cells. It is not immediately obvious what the principal advantages of the described assay are, compared to preforming the same experiment in a 96 well plate format. The manuscript seems more appropriate for methods journal than eLife.

1) The magnetic transfection device and methods need to be described in a greater detail to allow reproduction of the experiments in other labs. For example, the types and provenance of magnets is not obvious. What magnetic fields were tested to achieve optimal transfections and focusing of magnetic particles? Are the top 20mm magnets circular? How does it affect focusing of the particles in the dots aligned with the centers vs spaces between these large magnets?

2) The authors argue that mRNA transfection is superior to plasmid transfection as it allows for better temporal control of gene expression. However no data are provided measuring the kinetics of gene expression – analysis of destabilized GFP or short-lived proteins following magnetofection. Furthermore, the authors should complement their data with a direct comparison of the mRNA vs. plasmid delivery. Demonstrating the use of this technique for simple plasmid transfection is important as long term/stable gene expression might be preferable in many instances. Moreover, the simplicity of plasmid preparation compared to mRNA synthesis would make the technique more generally accessible and interesting to a broader audience.

3) Uniformity of delivered RNA copy-numbers and stoichiometries: Figure 2B shows that the 4 pulses of transfection achieve higher level of GFP and RFP expression. The apparent increase of transfection homogeneity shown in Figure 2C could therefore simply be a consequence of higher reporter expression with more cells reaching saturation. The authors need to compensate the exposure to achieve comparable overall intensity and saturation (i.e. shorter exposure of the 4x experiment) and replot GFP:RFP ratios to more realistically determine whether better homogeneity of expression has been achieved.

4) The authors should comment on the way they switch media between magnetofections – does 4x mean 4x change of media or 4x magnet application with the same media? Do media changes result in neuronal detachment and might that influence final cell counts? Overall it is not clear whether this tool, as described, could be used for "massively combinatorial screening". Experiments performed with few transcription factors do not demonstrate massive parallel screening and therefore references the language should be appropriately toned down.

5) Related to point 2, how long are exogenous FOXA2 and PITX3 proteins detectable in transfected cells and at what point are endogenous factors induced? The authors should epitope tag transfected factors to distinguish them from the induced endogenous factors and probe kinetics of their expression.

6) The text is in places too speculative and hard to follow, there are several typos and panel in Figure 3B does not agree with Figure 3D (see values for PtM vs. PtP, FoM vs. FoP) – very confusing…

Reviewer #2:

This is an interesting technical development for the field. Transcription factors control cell differentiation during embryonic development, thus an appealing tool to guide in vitro cell differentiation for clinical applications. However, these transcription factors work not only in combination but also at different times during development. Thus, to achieve proper and precise in vitro cell differentiation guided by transcription factors it is not only necessary to express combinations of factors but also to do so at different differentiation stages. The method presented by Azimi and co-workers has the potential to accomplish both tasks. Moreover, it can do so in a scalable and multiplexing manner conducive for genetic screens. The value of this paper is the technology and its potential. Thus, I believe that it deserved a better description of the method. The proof of principle for generating TH positive neurons is extremely welcome and elevates the manuscript. However, it will require more experiment to convincingly show that these cells are actual midbrain dopaminergic that are beyond the main innovation presented here. However, a minimal extra quantification will elevate the claims.

1) A better description of the magnet control is necessary for a general audience. Also, a picture or a more desirable video in supplementary data will help to visualize the device and its function. This is a key development and should be understandable by all possible future users.

"'inactivating' unwanted magnets via pushing them significantly down along the z-axis". Pushing the top magnet down does not displace them away from the substrate. Unless this refers to the control of the bottom magnets before the plate is inserted, I am confused by this statement: "The displacement of a magnet as little as 5 mm in z-axis away from the substrate surface is sufficient to eliminate the effect of magnet's magnetic field on the spotting field." Is there a single z-axis manipulator per bottom magnet? Or a single manipulator that moves in the x-y space pushes one magnet at the time? The figure suggests a big plate pushing all the pins.

2) The transfection efficiency and consistency is a key factor in the success of this type of approaches. Thus, an important aspect of this technology to be scalable.

It is clear that a large fraction of the cells express both fluorescent proteins in the improved protocol (Figure 2B). It would be nice to complement Figure 2C with a quantification of the percentage of transfected cells. The authors mention "90%" in the text, thus I assume they have measured. Reporting the actual% over DAPI and its error will provide an estimation of the variability of the method. By the same token, is the PCC calculated from a single transfection or with data from multiple days?

3) I could not find any experiment that validates the expression of other transcription factor other than an increase in FoxA2 over the already high percentage of cells that endogenously express it during differentiation. This manuscript will benefit from a simple validation of at least a couple of them with by antibody staining.

4) This comment is related to the comment above. A potential transformative feature of this new technology is the ability to introduce transcription factors at different states of differentiation, in particular postmitotic stage. Since current methods have limitations, the ability to transfect postmitotic neuros at high efficiency is a highly desirable feature of any protocol. However, I see no evidence that this method actually transfects postmitic neurons. Thus, although the temporal claims in Figure 3 are supported by the experiment, the labeling between mitotic and postmitotic are not. There are very simple experiments to address this point.

5) Dopaminergic neurons are characterized by the expression of several genes. Although required, TH expression by no means is indicative of dopaminergic fate. Thus, additional stainings such as TH in combination with Dbh or Tph2 will demonstrate better that these cells are DA neurons. Also, since the authors have the cDNA, qPCR quantification of other monoaminergic genes will enhance the claims of DA fate over other cell types (i.e. Gad1, Vglut, Sert, Dat, Net, Dbh, Tph).

6) Figure 3B and Figure 3D do not agree. In Figure 3B Fom and Ptm increase TH staining while in Figure 3D Fop and Ptp increase TH. This discrepancy needs clarification since it concerns the conclusion of each factor's activity period to induce TH. I believe this discrepancy also extends to Lmx1b.

7) Figure 2D is beautiful. Please add the concentration for each axis and describe the method. Is this with the single transfection per concentration or it requires interleaved (4x) per concentration? Or alternatively, 4x for GFP is 4x interleaved vs. 1x was a unique event?

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Combinatorial Programming of Human Neuronal Progenitors Using Magnetically-Guided Stoichiometric mRNA Delivery" for consideration by eLife. Your article has been favorably evaluated by a Senior Editor and two reviewers, one of whom, Sacha B Nelson, is a member of our Board of Reviewing Editors.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. In considering the nature of this contribution, we suggest this should be viewed as a Tools and Resources paper rather than in the category of a Research Article. That will not affect any aspect of the way this paper is published should the revised version be accepted by the Board.

Summary:

The authors describe a new magnetic spotting technology that permits much more reproducible transfection of cultured cells. By directly transfecting mRNA, they greatly reduce the variability typically associated with plasmid transfection, and by using a device that permits simultaneous magnetotransfection in a multi well format they present a system suitable for high throughput studies. Using this system they demonstrate optimized combinations of transcription factors for generating dopaminergic neurons from stem cells. Although technical in nature, this seems an important development that will help put the field of stem cell biology on a more quantitative footing, a development that is sorely needed.

Essential revisions:

1) Both reviewers felt the comparison with standard transfection methods was unclear. There is no direct comparison to other methods. An image presented in the reply to reviewers makes the essential point, but this point is likely to be lost on most readers of the paper, even including many who routinely use transfection, but do not try to quantify its reproducibility. The correlation coefficients and MOC values obtained seem impressive, but many readers may have no idea how this compares to other methods. The authors should either use some of their own experiments to document the improvement that results from a) RNA over DNA transfection and b) magnetotransfection, or they should use some published data on reproducibility and then analyze their own data in such a manner as to allow comparison. I do not think it necessary to quantitatively assess the two components (RNA and the controllable magnets) separately, although some indication of their relative importance should be provided. Presumably the main reproducibility factor is the RNA and the magnets just allow the neat trick of multiplexing factors and/or dosages.

The fact that the RNA is more reproducible because it is not subject to the multiplicative effect of transcription was not made directly. The focus on numbers of molecules transfected is confusing, since as pointed out by one reviewer, plasmids are not so much larger than the modified RNA molecules.

Along the same lines, it would be helpful to provide benchmarks for readers unfamiliar with current state of the art in differentiating stem cells into dopamine neurons. How do the current results compare? How much of an improvement is this? This is very hard to glean without going through all of the cited papers and other relevant literature.

2) There is not much in the way of consideration of limitations. Most notably, is the method known to be applicable to postmitotic cells? If not, the authors should be clearer on this point. The authors seem to imply that they do not really know if the cells are postmitotic or not when the "postmitotic" transfection is performed. Does the nuclear membrane need to break down for high efficiency? This won't affect experiments on stem cells and cell lines but the methods are potentially also applicable to differentiated cells and whether or not this is the case should be communicated. This point was also raised by a prior reviewer, but was not directly addressed.

The authors do not need to do additional experiments to address this point. They only need to state more clearly what they do and do not know about the transfection. If they have no existing data on truly post mitotic cells, they should take care not to imply that the cells they are transfecting are post mitotic unless they are sure that they are.

[Editors’ note: minor issues and corrections have not been included, so there is not an accompanying Author response.]

Thank you for submitting your article "Combinatorial Programming of Human Neuronal Progenitors Using Magnetically-Guided Stoichiometric mRNA Delivery" for consideration by eLife. Your article has been reviewed by a Reviewing Editor in consultation with a Senior Editor.

The many changes made in the manuscript adequately address all of the concerned raised in earlier reviews with one exception. The opening statement "The human nervous system contains hundreds of distinct subtypes of neurons" is in my opinion a vast underestimate and should minimally be qualified with "at least". In addition, the reference cited (#1) does not make any statement about the numbers of cell types in the human or any other nervous system, hence I believe citing it in this context is incorrect.

eLife. 2018 May 1;7:e31922. doi: 10.7554/eLife.31922.017

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

The manuscript by Azimi et al. describes the development of a new technique for focused transfection of cultured cells that is compatible with combinatorial gene expression studies. They demonstrate the utility of this approach by testing combinations of transcription factors known to be involved in the specification of ventral midbrain dopaminergic neurons. They identified LMX1A, PITX3 and FOXA2 as the most optimal combination of factors and show that PITX3 and FOXA2 factors need to be delivered to dividing progenitors while LMX1A can be delivered to postmitotic neurons. While these three factors were previously shown to promote dopaminergic neuronal identity, this study adds by probing temporal requirements of individual factors.

The major concern with the described method is that it requires the preparation of transfection reagents in large volumes as the entire plate is filled with the solution while only few spots are being transfected.

Our method actually substantially reduces the amount of transfection reagents required per experiment and allows many more experiments to be performed per area. We now explain this more clearly in the manuscript:

"Importantly, with our technology the amount of RNA and transfection reagent required for a given transcription factor is proportional only to the number of spots actually transfected with that factor, rather than to the total number of possible spots in the array (i.e. to the surface area of transfected cells, not to the total surface area of the plate). […] The number of cell medium changes we use also scale linearly with the number of transcription factors delivered, rather than growing exponentially with the number of combinatorial possibilities."

Thus, very small amounts of reagents are spotted specifically onto defined millimeter size spots of cells as opposed to treating entire wells with only a single combination of factors and reagents (for instance, in the experiments shown in our manuscript we use only 0.6 ng RNA for each 1.5 mm diameter spot). Figure 1C and Figure 2D show how the reagents are focally localized onto these very small regions for transfection, which is much smaller than 96- or even 384-well plate formats. In addition, uniform cell culture in 96/384-well plates is often not achievable due to surface effects, a problem that is well known within the high-throughput cell culture screening field.

Besides being costly, it might be challenging to perform multiple complete media replacements for truly large scale combinatorial analysis of gene function without disturbing the plated cells.

We do indeed demonstrate that we can perform multiple rounds of transfections and media exchanges without disturbing the cells or spotting pattern (Figure 2 and Figure 3). We have now made this point clearer in the manuscript: As noted in our response to the previous question, the number of media changes scale linearly with the number of transfection factors while the number of transfected combinations (hence spots) scale exponentially. Thus, for even larger-scale experiments, the number of exchanges would not be much different than what we already do. In addition, industrial systems that change cell-culture mediums from multi-well plates use automated slowly tilting/suction mechanisms for many sequential medium. Although we did not need to implement this in our screen, such systems could be used if cell displacement becomes an issue during manual medium exchange.

It is not immediately obvious what the principal advantages of the described assay are, compared to preforming the same experiment in a 96 well plate format.

As noted above and also made clearer in the revised manuscript, our method’s advantages with respect to 96-well plate formats are tremendous and multifold:

1) Performing experiments on much smaller spots of cells results in a significant savings in reagents. To deliver proportional amount of RNA to each spot comparable to the standard non- spotting transfection, we use an RNA amount linearly proportional to only the number of spots to be transfected and to the surface area of each spot (for instance, 0.6 ng RNA was used for each spot with 1.5 mm diameter). The number of cell medium changes we use also scale linearly with the number of transcription factors delivered, rather than growing exponentially with the number of combinatorial possibilities.

Figure 1c shows how the reagents are focally localized onto a very small footprint for transfection, which is much smaller than 96-well plate formats. In addition, uniform cell culture in 96-well format plates is often not achievable due to surface effects, as known by most people with expertise in high-throughput cell culture screens. Our method thus saves large amounts of transfection reagents by allowing many more experiments to be performed per area.

Here is an estimate of the gain with our technology:

Amount of RNA needed per condition:

96-well plate: 10.9ng (= 0.6ng*(6.4mm/1.5mm)²) where 6.4mm is the diameter of a 96- well plate and 1.5mm is the diameter of a spot.

Spotting technology: 0.6ng per spot/condition

Improvement: 18.2 (i.e. ~18 fold)

Amount of transfection reagents (lipid/magnet nano-mixture) needed per combination: Transfection reagents needed scale with the amount of RNA that needs to be transfected. Thus:

Improvement: 18.2 (i.e. ~18 fold)

Amount of cell-culture medium + growth factors needed per combination:

96-well plate: ~6mm x pi*(6.4mm/2)² assuming 200ul per well.

Spotting technology: ~6mm x (3mm)² where 3mm is the spot to spot distance (This estimate includes the unused volume in between spots).

Improvement: 3.6 (i.e. 3.6 fold)

2) The mRNA transfection procedure we developed dramatically reduces fluctuations in copy numbers and expression of delivered factors as we explained in the manuscript in great detail. Author response image 1 clearly shows the variation in gene expression when a 1536-well plate is transfected with DNA (unpublished image provided by Inglese, J. and Jang, S.-W. from NIH Screening Center). Note that each well of a 1536-well plate is even larger than our spots. The left panel shows plating density of a clonal line stably expressing bioluminescent-reporter gene (luciferase) to highlight the well-to-well variation in gene expression on the right panel with transient transfection by luciferase. The reason for this significant variability with DNA transfection is that the number of DNA molecules delivered is typically very low (unlike mRNA transfection). As a result, the Poisson noise in the delivered copy numbers (assuming each transfection process is independent) is huge because of the low mean DNA copy number per cell. This is in contrast to mRNA transfection, where typically hundreds to thousands of copies of the mRNA are delivered during each transfection.

3) The cell cultures in 96-well or smaller formats significantly suffer from edge effects and the culture becomes no longer homogeneous in properties. This challenge is well known within the high-throughput cell culture screening field as shown in Figure 2C from Lundholt et al. (Journal of Biomolecular Screening 8(5); 2003).

The manuscript seems more appropriate for methods journal than eLife.

We believe our manuscript is of great interest not only to the neuroscience and stem cell community, but also to any field of biology involving high-throughput experiments. Many colleagues in the field have expressed great interest in our work. Indeed, NIH Director Francis Collins has highlighted our platform’s blueprint in his presentations among the top examples of high-impact research areas that they fund (This technology is the core of our NIH Director’s Pioneer Award (DP1)).

In addition, we note that eLife has a “Tools and Resources” category. According to the eLife website: “This category highlights tools or resources that are especially important for their respective fields and have the potential to accelerate discovery. For example, we welcome the submission of significant technological or methodological advances.”

1) The magnetic transfection device and methods need to be described in a greater detail to allow reproduction of the experiments in other labs. For example, the types and provenance of magnets is not obvious. What magnetic fields were tested to achieve optimal transfections and focusing of magnetic particles? Are the top 20mm magnets circular? How does it affect focusing of the particles in the dots aligned with the centers vs spaces between these large magnets?

We have included additional details in the revised manuscript and highlighted the type of magnets used and their shapes:

"We used neodymium (NdFeB) rare-earth magnets because of their high magnetic field strength compared to other types of magnets. […] The magnetic field density on the surface of each magnet is B~300 mT in isolation and B~100 mT when inside the array."

In addition, we have added Supplementary file 1, which contains vendor names and catalog numbers for both top and bottom NdFeB magnets.

2) The authors argue that mRNA transfection is superior to plasmid transfection as it allows for better temporal control of gene expression. However no data are provided measuring the kinetics of gene expression – analysis of destabilized GFP or short-lived proteins following magnetofection.=

With standard DNA delivery methods, genes are forcibly expressed for as long as the DNA template is present in the cell, thus overriding the cell's endogenous mRNA/protein degradation mechanisms. There are chemically controlled gene expression systems that can overcome these concerns, however these are not scalable for high-throughput combinatorial screening. In addition, when using DNA plasmids and/or viral vectors, there is greater uncertainty about the time lag between the transfection and when the desired levels of expression have actually been achieved. With the synthetic mRNAs, both the timing and the relative amounts (i.e. stoichiometry) of each factor are very precise.

Author response image 2 comes from a previous publication by our lab (Figure 1C from Angel and Yanik PLoS ONE 2010) and shows the lifetime of several proteins following transfection of cells with mRNAs that use the same 5’- and 3’ UTRs as those in our current manuscript. These data show the kinetics of protein expression and confirm that tight temporal regulation is possible for most proteins. The exact kinetics of expression will, of course, always vary to some extent between different proteins. We do not believe that quantifying additional specific cases (e.g. destabilized GFP or other proteins) will add significantly to the generality of our platform.

To further address the kinetics of transfected mRNAs, in Author response image 3 we provide unpublished RT-PCR data on the lifetime of EYFP mRNA following transfection (again with the same 5’- and 3’- UTRs used in the present manuscript and in Angel and Yanik, 2010). These data show that the mRNAs significantly degrade within 48 hours, a shorter period than the timescale of our cellular programming experiments.

Furthermore, the authors should complement their data with a direct comparison of the mRNA vs. plasmid delivery. Demonstrating the use of this technique for simple plasmid transfection is important as long term/stable gene expression might be preferable in many instances. Moreover, the simplicity of plasmid preparation compared to mRNA synthesis would make the technique more generally accessible and interesting to a broader audience.

Reliable DNA transfection is not feasible when scaled down to the dimensions we use for spotting due dramatic fluctuations in expression. Each well of the 1536-well plate on the right in Author response image 1 (unpublished data provided with permission of Inglese, J. and Jang, S.-W. from NIH Screening Center) has been transfected with DNA encoding luciferase, clearly showing the variation in gene expression that results under these conditions. The left panel shows a clonal line stably expressing luciferase to better highlight the well-to-well variation in expression on the right panel. Each of our spots is even smaller than the surface area of a single well of a 1536-well plate. The reason for this significant variability with DNA transfection is that the number of DNA molecules delivered is typically very low (unlike mRNA transfection). As a result, the Poisson noise in the delivered copy numbers (assuming each transfection process is independent) is huge because of the low mean DNA copy number per cell. This is in contrast to mRNA transfection, where typically hundreds to thousands of copies of the mRNA are delivered during each transfection.

3) Uniformity of delivered RNA copy-numbers and stoichiometries: Figure 2B shows that the 4 pulses of transfection achieve higher level of GFP and RFP expression. The apparent increase of transfection homogeneity shown in Figure 2C could therefore simply be a consequence of higher reporter expression with more cells reaching saturation. The authors need to compensate the exposure to achieve comparable overall intensity and saturation (i.e. shorter exposure of the 4x experiment) and replot GFP:RFP ratios to more realistically determine whether better homogeneity of expression has been achieved.

This is a good point and we have now explained in the text why this is not the case. Among all conditions in all the images, our camera software automatically finds the brightest spot (across all the well images) and reduces exposure to prevent saturation according to that maximum intensity spot. Thus, the exposure durations in the wells/images used in the generation of the statistics in Figure 2C were not saturated. In addition to clarifying this in the manuscript, we also provided a histogram of pixel intensity distributions of the raw image data in Author response image 4.

4) The authors should comment on the way they switch media between magnetofections – does 4x mean 4x change of media or 4x magnet application with the same media? Do media changes result in neuronal detachment and might that influence final cell counts? Overall it is not clear whether this tool, as described, could be used for "massively combinatorial screening". Experiments performed with few transcription factors do not demonstrate massive parallel screening and therefore references the language should be appropriately toned down.

The 4x indicates that the medium was changed 4 times. Although we performed many medium exchanges, we do not observe any significant cell detachment in the experiments reported here. We do indeed demonstrate that we can perform multiple rounds of transfections and media exchanges without disturbing the cells or spotting pattern (Figure 2 and Figure 3). The number of media changes scale only linearly with the number of transfection factors but the number of transfected combinations (hence spots) scale exponentially with the number of transfection factors. Thus, for even larger-scale experiments, the number of exchanges would not be much greater than what we already do.

In addition, industrial systems that change cell-culture mediums from multi-well plates use automated slowly tilting/suction mechanisms for many sequential medium exchanges. Although we did not need to implement this in our screen, such systems could be used if cell displacement becomes an issue during manual medium exchange.

We respectfully disagree that our screen was small: we have screened every combination (up to triplets) of 7 transcription factors at different time points. Besides this, we had to repeat/replicate many steps of our experiments during the development of the methodology and also to show the robustness of our results. A much larger screening set-up would be beyond our resources. Our results are sufficient to demonstrate that our platform can be expanded for larger and more elaborate screening procedures.

5) Related to point 2, how long are exogenous FOXA2 and PITX3 proteins detectable in transfected cells and at what point are endogenous factors induced? The authors should epitope tag transfected factors to distinguish them from the induced endogenous factors and probe kinetics of their expression.

Such tracking of endogenous vs. exogenous protein degradation kinetics is very rarely demonstrated in reprogramming studies in the literature (unless it is necessary for the biological question under investigation). Such a study would be beyond the scope of our manuscript.

6) The text is in places too speculative and hard to follow, there are several typos and panel in Figure 3B does not agree with Figure 3D (see values for PtM vs. PtP, FoM vs. FoP) – very confusing.

We have fixed the two typos in the figures and edited the text throughout to improve clarity and readability. If the reviewer could specifically point out any other hard-to-follow or speculative points, we will be happy to further improve them.

Reviewer #2:

This is an interesting technical development for the field. Transcription factors control cell differentiation during embryonic development, thus an appealing tool to guide in vitro cell differentiation for clinical applications. However, these transcription factors work not only in combination but also at different times during development. Thus, to achieve proper and precise in vitro cell differentiation guided by transcription factors it is not only necessary to express combinations of factors but also to do so at different differentiation stages. The method presented by Azimi and co-workers has the potential to accomplish both tasks. Moreover, it can do so in a scalable and multiplexing manner conducive for genetic screens. The value of this paper is the technology and its potential. Thus, I believe that it deserved a better description of the method. The proof of principle for generating TH positive neurons is extremely welcome and elevates the manuscript. However, it will require more experiment to convincingly show that these cells are actual midbrain dopaminergic that are beyond the main innovation presented here. However, a minimal extra quantification will elevate the claims.

1) A better description of the magnet control is necessary for a general audience. Also, a picture or a more desirable video in supplementary data will help to visualize the device and its function. This is a key development and should be understandable by all possible future users.

"'inactivating' unwanted magnets via pushing them significantly down along the z-axis". Pushing the top magnet down does not displace them away from the substrate. Unless this refers to the control of the bottom magnets before the plate is inserted, I am confused by this statement: "The displacement of a magnet as little as 5 mm in z-axis away from the substrate surface is sufficient to eliminate the effect of magnet's magnetic field on the spotting field." Is there a single z-axis manipulator per bottom magnet? Or a single manipulator that moves in the x-y space pushes one magnet at the time? The figure suggest a big plate pushing all the pins.

We have rewritten the description of the setup in the manuscript in order to address these concerns. We have also significantly modified Figure 1D to clearly show the sequence of operations. We now use arrows to indicate the direction of motion and the sequence of operation for every moving component in the system.

2) The transfection efficiency and consistency is a key factor in the success of this type of approaches. Thus, an important aspect of this technology to be scalable.

It is clear that a large fraction of the cells express both fluorescent proteins in the improved protocol (Figure 2B). It would be nice to complement Figure 2C with a quantification of the percentage of transfected cells. The authors mention "90%" in the text, thus I assume they have measured. Reporting the actual% over DAPI and its error will provide an estimation of the variability of the method. By the same token, is the PCC calculated from a single transfection or with data from multiple days?

We calculated Pearson’s Correlation Coefficient (PCC) and Mander’s Overlap Coefficient (MOC) from different wells with single transfection, and we averaged it over 3 different experiments, which is now indicated in the revised manuscript.

For the 1X transfection protocol 90.4% (236/261) of the detected cells express GFP and 92.3% (241/261) express mCherry. For the 4X interleaved transfection protocol 100% (235/235) of the detected cells express both GFP and mCherry (at varying ratios). The average 90% transfection efficiency was based on determining the number of cells that only express one of the two fluorescent proteins (GFP and mCherry) above a minimum threshold; these cells were considered "untransfected" with respect to the low-expressing protein. Specifically, cells were considered untransfected for a given fluorescent protein if they failed to express it at a level ≧10% of the strongest-expressing cell in the total population analyzed.

3) I could not find any experiment that validates the expression of other transcription factor other than an increase in FoxA2 over the already high percentage of cells that endogenously express it during differentiation. This manuscript will benefit from a simple validation of at least a couple of them with by antibody staining.

We have now clarified this point in this manuscript and in the legend for Figure 4B, which addresses this concern. Although the increase in total FoxA2 is marginal (FoxA2/DAPI in the left panel of Figure 4B), the percentage of FoxA2 cells that also co-express TH (FoxA2/TH in the right panel of Figure 4B) is more than doubled when we deliver the mix of transcription factors FoxA2m, Pitx3m, and Lmx1ap. The quantification based on FoxA2/TH co-expression has been used as the sole criteria in the seminal papers of Lorenz Studer and colleagues.

We have also attempted staining with Girk2, VMAT2, and Lmx1a antibodies in the past, however we found staining by these antibodies to be unreliable over large numbers of samples, even though we tested antibodies from several vendors.

4) This comment is related to the comment above. A potential transformative feature of this new technology is the ability to introduce transcription factors at different states of differentiation, in particular postmitotic stage. Since current methods have limitations, the ability to transfect postmitotic neuros at high efficiency is a highly desirable feature of any protocol. However, I see no evidence that this method actually transfects postmitic neurons. Thus, although the temporal claims in Figure 3 are supported by the experiment, the labeling between mitotic and postmitotic are not. There are very simple experiments to address this point.

"Mitotic" here more specifically refers to the presence of mitogens in the medium and "post- mitotic" specifically refers to the medium condition where mitogens have been withdrawn. To be more specific, we have now clarified what we mean by mitotic and post-mitotic in the manuscript:

"We used two different mediums: 1) an initial expansion medium containing the mitogens bFGF and EGF to sustain the mitotic (proliferative) state and 2) a subsequent differentiation medium lacking the mitogens, causing immediate exit from the cell cycle (post-mitotic state) and inducing neurogenesis."

It is very well known that NPCs require mitogens to proliferate and a standard method to induce differentiation of NPCs is mitogen withdrawal. Although we did not directly measure the mitotic index (e.g. ki-67 staining), we did not observe any additional growth of the cell culture after mitogen withdrawal.

5) Dopaminergic neurons are characterized by the expression of several genes. Although required, TH expression by no means is indicative of dopaminergic fate. Thus, additional stainings such as TH in combination with Dbh or Tph2 will demonstrate better that these cells are DA neurons. Also, since the authors have the cDNA, qPCR quantification of other monoaminergic genes will enhance the claims of DA fate over other cell types (i.e. Gad1, Vglut, Sert, Dat, Net, Dbh, Tph).

Lmx1a and FoxA2 expression are essential to ventral medial DA neuronal progenitors and Nurr1 and Pitx3 are required for the maturation of ventral medial DA neuronal progenitors to mature DA neurons. We chose these in addition to VMAT2 and OTX2 as markers to demonstrate increased dopaminergic fate in vitro by the delivery of RNA factors.

In addition to TH expression, we also observe co-expression of FoxA2 (FoxA2/TH in the right panel of Figure 4B). Although the increase in total FoxA2 is marginal (FoxA2/DAPI in the left panel of Figure 4B), the percentage of FoxA2 cells that also co-express TH (FoxA2/TH in the right panel of Figure 4B) is more than doubled when we delivered the mix of transcription factors FoxA2^m, Pitx3^m, and Lmx1a^p. The quantification based on FoxA2/TH co-expression and the factors we tested above have been used as the sole criteria in the seminal papers of Lorenz Studer and colleagues.

We have not commonly seen the use of co-staining for either Dbh or Tph2 with TH. We are also not aware of midbrain specific cell types that co-express Dbh or Tph2 with TH.

If the purpose were to exclude the possibility of some aberrant cell types that co-express other genes along with TH, this would require a very large-scale profiling which is beyond the scope of our study.

We had also attempted staining with Girk2, VMAT2, and Lmx1a antibodies in the past, however we found staining by these antibodies to be unreliable over large numbers of samples, even though we tested antibodies from several vendors.

6) Figure 3B and Figure 3D do not agree. In Figure 3B Fom and Ptm increase TH staining while in Figure 3D Fop and Ptp increase TH. This discrepancy needs clarification since it concerns the conclusion of each factor's activity period to induce TH. I believe this discrepancy also extends to Lmx1b.

We thank the reviewer for catching this error. We have now corrected it.

7) Figure 2D is beautiful. Please add the concentration for each axis and describe the method. Is this with the single transfection per concentration or it requires interleaved (4x) per concentration? Or alternatively, 4x for GFP is 4x interleaved vs. 1x was a unique event?

Figure 2D involves 24 medium exchanges [interleaved (4x) per concentration] to generate all the possible combinations. We have now clarified this in the manuscript. We now also indicate concentration on the figure axis.

[Editors' note: the author responses to the re-review follow.]

Essential revisions:

1) Both reviewers felt the comparison with standard transfection methods was unclear. There is no direct comparison to other methods. An image presented in the reply to reviewers makes the essential point, but this point is likely to be lost on most readers of the paper, even including many who routinely use transfection, but do not try to quantify its reproducibility. The correlation coefficients and MOC values obtained seem impressive, but many readers may have no idea how this compares to other methods. The authors should either use some of their own experiments to document the improvement that results from a) RNA over DNA transfection and b) magnetotransfection, or they should use some published data on reproducibility and then analyze their own data in such a manner as to allow comparison. I do not think it necessary to quantitatively assess the two components (RNA and the controllable magnets) separately, although some indication of their relative importance should be provided. Presumably the main reproducibility factor is the RNA and the magnets just allow the neat trick of multiplexing factors and/or dosages.

We now include references that directly compare RNA- vs. DNA-based transfection approaches head-to-head in various difficult to transfect cell lines with detailed efficiency assays. These references support our claim regarding the superior efficiency and reliability of RNA over DNA. As the reviewers and editor correctly hypothesize, the magnetic transfection reagent (CombiMag) functions to achieve precise localization/multiplexing of RNA molecules, while the transfection process itself relies on a commercial transfection reagent rather than the magnets. We have now re-written the paragraph discussing the benefits of RNA to emphasize well-known mechanistic differences impacting the expression of exogenous RNAs vs. DNAs:

"The fact that cell fate decisions are highly dependent on the precise expression levels of a limited set of genes demands that each reprogramming factor in a given cocktail needs to be delivered with minimal cell-to-cell variation to achieve maximal efficiency. […] This is supported by the low variability of RNA transfection results with respect to DNA transfection (Hansson et al., 2015; Landi, Babiuk and van Drunen Littel-van den Hurk, 2007)."

However, when using special cell lines that are easily transfected and/or transfection protocols that have been optimized for delivery of DNA, it may indeed be feasible to apply our platform to combinatorial screens that use plasmids or other DNA-based vectors. The primary emphasis of our manuscript is on validating our high-throughput platform for magnetically-guided combinatorial transcription factor screens, rather than exploring all possible cell types, nucleic acid vectors, and/or commercial transfection reagents that our platform may be compatible with. It has been established in the literature that CombiMag can be used to deliver DNA (Varro, Kenny et al., 2007), siRNA (Lee, Shim et al., 2011) miRNA (Zhang, Tang et al., 2016), retroviruses/lentiviruses (Fukushima, Tezuka et al., 2007), and even proteins (Watanabe, Tatebe et al., 2012) to a variety of cell types. In addition, it has been demonstrated to function with a variety of commercially available lipid-based transfection reagents.

The fact that the RNA is more reproducible because it is not subject to the multiplicative effect of transcription was not made directly. The focus on numbers of molecules transfected is confusing, since as pointed out by one reviewer, plasmids are not so much larger than the modified RNA molecules.

As noted above, although plasmids are not substantially larger than RNA molecules, the fact that they need to cross both the plasma membrane and the nuclear envelope can have a profound impact on the number of DNA molecules that can be functionally expressed. This is particularly true for difficult to transfect cell types [as noted in the newly incorporated Hansson et al. (2015) and Landi et al. (2007) references]. It is also a consideration that is generally acknowledged in the technical literature and online documentation accompanying commercial transfection reagents (for example: https://www.biocompare. com/Editorial-Articles/171593-What-to-Transfect-DNA-vs-RNA-vs-Protein/). Therefore, the variation observed between RNA and DNA transfections is likely not due exclusively to the fact that DNA is subject to transcriptional multiplication, but also due to the relatively smaller number of DNA molecules that are expressed.

Along the same lines, it would be helpful to provide benchmarks for readers unfamiliar with current state of the art in differentiating stem cells into dopamine neurons. How do the current results compare? How much of an improvement is this? This is very hard to glean without going through all of the cited papers and other relevant literature.

As requested, we have now included an overview of a range of differentiation efficiencies that have been reported in the literature as well as a discussion of why direct head-to-head comparisons between different publications/protocols is often challenging and how our platform can help address this challenge in the future:

"During the past decade, there have been significant efforts to generate dopaminergic neurons from human pluripotent stem cells (hPSCs) (Kriks et al., 2011; Theka et al., 2013; Friling et al., 2009; Lee et al., 2010), typically by supplementing the cell medium with various growth factors. In addition to growth factor-based differentiation protocols, exogenous overexpression of transcription factors (such as the single transcription factor LMX1A (Friling et al., 2009) or the combined overexpression of FOXA2 and NURR1 (Lee et al., 2010)) using DNA vectors can partially enhance the number of tyrosine hydroxylase (TH) positive dopaminergic neurons during differentiation of pluripotent stem cells under various culture conditions. […] Our high-throughput platform, where many factors and conditions can be simultaneously tested side-by-side and compared under the same laboratory conditions, can in the future allow for more accurate and quantitative comparisons."

2) There is not much in the way of consideration of limitations. Most notably, is the method known to be applicable to postmitotic cells? If not, the authors should be clearer on this point. The authors seem to imply that they do not really know if the cells are postmitotic or not when the "postmitotic" transfection is performed. Does the nuclear membrane need to break down for high efficiency? This won't affect experiments on stem cells and cell lines but the methods are potentially also applicable to differentiated cells and whether or not this is the case should be communicated. This point was also raised by a prior reviewer, but was not directly addressed.

The authors do not need to do additional experiments to address this point. They only need to state more clearly what they do and do not know about the transfection. If they have no existing data on truly post mitotic cells, they should take care not to imply that the cells they are transfecting are post mitotic unless they are sure that they are.

Consistent with the published literature, we observe very little proliferation once mitogen growth factors have been removed from the medium. However, in order to avoid these concerns, we have replaced all "mitotic/post-mitotic" terminology in the revised manuscript and now simply refer to whether transfections are being done on proliferative NPCs (i.e. cultured in the presence of growth factors) or after induction of neural differentiation (i.e. after grown factor removal). In the revised versions of Figure 3 and Figure 4 these conditions are indicated with an 'N' or an 'I' superscript, respectively.

Corresponding changes have been made to terminology throughout the manuscript where required. For example, from the Introduction:

"Using our magnetically-guided spotting platform and interleaved transfection protocol, we evaluated the temporal contributions of transcription factor cocktails by treating human NPCs with them during the proliferative stage and/or during the induction of neurogenesis (i.e. after mitogen withdrawal) to generate human dopaminergic neurons with high purity."

And from the Results:

"Transcription factors were delivered individually either during the proliferative NPC stage (Day -2, before mitogen removal) or the induction stage (Day 0, upon mitogen removal) as indicated by the superscripts N or I on each factor, respectively."

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure 2—source data 1. Ratio of red (mCherry) to green (GFP) fluorescence intensity per cell for both single and interleaved transfection protocols.

elife-31922-fig2-data1.xlsx^{(72.3KB, xlsx)}

DOI: 10.7554/eLife.31922.004

Figure 3—source data 1. Quantitative TH gene expression analysis at day 17 comparing factors delivered at NPC (superscript N) and early induction (superscript I) stages.

elife-31922-fig3-data1.xlsx^{(18.4KB, xlsx)}

DOI: 10.7554/eLife.31922.006

Supplementary file 1. List and details of top and bottom magnets.

elife-31922-supp1.docx^{(48.9KB, docx)}

DOI: 10.7554/eLife.31922.008

Supplementary file 2. List of TaqMan qRT-PCR assays used in this research.

elife-31922-supp2.docx^{(56KB, docx)}

DOI: 10.7554/eLife.31922.009

Transparent reporting form

elife-31922-transrepform.docx^{(1.1MB, docx)}

DOI: 10.7554/eLife.31922.010

Data Availability Statement

All data generated or analyzed during this study are included in the manuscript and supporting files.

[bib1] Abeliovich A, Hammond R. Midbrain dopamine neuron differentiation: factors and fates. Developmental Biology. 2007;304:447–454. doi: 10.1016/j.ydbio.2007.01.032. [DOI] [PubMed] [Google Scholar]

[bib2] Ang SL. Transcriptional control of midbrain dopaminergic neuron development. Development. 2006;133:3499–3506. doi: 10.1242/dev.02501. [DOI] [PubMed] [Google Scholar]

[bib3] Angel M, Yanik MF. Innate immune suppression enables frequent transfection with RNA encoding reprogramming proteins. PLoS One. 2010;5:e11756. doi: 10.1371/journal.pone.0011756. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Carpenter AE, Jones TR, Lamprecht MR, Clarke C, Kang IH, Friman O, Guertin DA, Chang JH, Lindquist RA, Moffat J, Golland P, Sabatini DM. CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome Biology. 2006;7:R100. doi: 10.1186/gb-2006-7-10-r100. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib5] Engel M, Do-Ha D, Muñoz SS, Ooi L. Common pitfalls of stem cell differentiation: a guide to improving protocols for neurodegenerative disease models and research. Cellular and Molecular Life Sciences. 2016;73:3693–3709. doi: 10.1007/s00018-016-2265-3. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib6] Friling S, Andersson E, Thompson LH, Jönsson ME, Hebsgaard JB, Nanou E, Alekseenko Z, Marklund U, Kjellander S, Volakakis N, Hovatta O, El Manira A, Björklund A, Perlmann T, Ericson J. Efficient production of mesencephalic dopamine neurons by Lmx1a expression in embryonic stem cells. PNAS. 2009;106:7613–7618. doi: 10.1073/pnas.0902396106. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] Ghannad-Rezaie M. Rezaie99. Github. 2018 https://github.com/rezaie99/ELIFE-050518

[bib8] Hansson ML, Albert S, González Somermeyer L, Peco R, Mejía-Ramírez E, Montserrat N, Izpisua Belmonte JC. Efficient delivery and functional expression of transfected modified mRNA in human embryonic stem cell-derived retinal pigmented epithelial cells. Journal of Biological Chemistry. 2015;290:5661–5672. doi: 10.1074/jbc.M114.618835. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] Hegarty SV, Sullivan AM, O'Keeffe GW. Midbrain dopaminergic neurons: a review of the molecular circuitry that regulates their development. Developmental Biology. 2013;379:123–138. doi: 10.1016/j.ydbio.2013.04.014. [DOI] [PubMed] [Google Scholar]

[bib10] Jemielity J, Fowler T, Zuberek J, Stepinski J, Lewdorowicz M, Niedzwiecka A, Stolarski R, Darzynkiewicz E, Rhoads RE. Novel "anti-reverse" cap analogs with superior translational properties. Rna. 2003;9:1108–1122. doi: 10.1261/rna.5430403. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] Jiang Y, Xu XS, Russell JE. A nucleolin-binding 3' untranslated region element stabilizes beta-globin mRNA in vivo. Molecular and Cellular Biology. 2006;26:2419–2429. doi: 10.1128/MCB.26.6.2419-2429.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] Karikó K, Buckstein M, Ni H, Weissman D. Suppression of RNA recognition by Toll-like receptors: the impact of nucleoside modification and the evolutionary origin of RNA. Immunity. 2005;23:165–175. doi: 10.1016/j.immuni.2005.06.008. [DOI] [PubMed] [Google Scholar]

[bib13] Karikó K, Muramatsu H, Keller JM, Weissman D. Increased erythropoiesis in mice injected with submicrogram quantities of pseudouridine-containing mRNA encoding erythropoietin. Molecular therapy : the journal of the American Society of Gene Therapy. 2012;20:948–953. doi: 10.1038/mt.2012.7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Karikó K, Muramatsu H, Welsh FA, Ludwig J, Kato H, Akira S, Weissman D. Incorporation of pseudouridine into mRNA yields superior nonimmunogenic vector with increased translational capacity and biological stability. Molecular therapy : the journal of the American Society of Gene Therapy. 2008;16:1833–1840. doi: 10.1038/mt.2008.200. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Karikó K, Weissman D. Naturally occurring nucleoside modifications suppress the immunostimulatory activity of RNA: implication for therapeutic RNA development. Current Opinion in Drug Discovery & Development. 2007;10:523–532. [PubMed] [Google Scholar]

[bib16] Kele J, Simplicio N, Ferri AL, Mira H, Guillemot F, Arenas E, Ang SL. Neurogenin 2 is required for the development of ventral midbrain dopaminergic neurons. Development. 2006;133:495–505. doi: 10.1242/dev.02223. [DOI] [PubMed] [Google Scholar]

[bib17] Kormann MS, Hasenpusch G, Aneja MK, Nica G, Flemmer AW, Herber-Jonat S, Huppmann M, Mays LE, Illenyi M, Schams A, Griese M, Bittmann I, Handgretinger R, Hartl D, Rosenecker J, Rudolph C. Expression of therapeutic proteins after delivery of chemically modified mRNA in mice. Nature Biotechnology. 2011;29:154–157. doi: 10.1038/nbt.1733. [DOI] [PubMed] [Google Scholar]

[bib18] Kriks S, Shim JW, Piao J, Ganat YM, Wakeman DR, Xie Z, Carrillo-Reid L, Auyeung G, Antonacci C, Buch A, Yang L, Beal MF, Surmeier DJ, Kordower JH, Tabar V, Studer L. Dopamine neurons derived from human ES cells efficiently engraft in animal models of Parkinson's disease. Nature. 2011;480:547–551. doi: 10.1038/nature10648. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] Landi A, Babiuk LA, van Drunen Littel-van den Hurk S. High transfection efficiency, gene expression, and viability of monocyte-derived human dendritic cells after nonviral gene transfer. Journal of Leukocyte Biology. 2007;82:849–860. doi: 10.1189/jlb.0906561. [DOI] [PubMed] [Google Scholar]

[bib20] Lee HS, Bae EJ, Yi SH, Shim JW, Jo AY, Kang JS, Yoon EH, Rhee YH, Park CH, Koh HC, Kim HJ, Choi HS, Han JW, Lee YS, Kim J, Li JY, Brundin P, Lee SH. Foxa2 and Nurr1 synergistically yield A9 nigral dopamine neurons exhibiting improved differentiation, function, and cell survival. Stem Cells. 2010;28:501–512. doi: 10.1002/stem.294. [DOI] [PubMed] [Google Scholar]

[bib21] Lees AJ, Hardy J, Revesz T. Parkinson's disease. Lancet. 2009;373:2055–2066. doi: 10.1016/S0140-6736(09)60492-X. [DOI] [PubMed] [Google Scholar]

[bib22] Li Q, Lau A, Morris TJ, Guo L, Fordyce CB, Stanley EF. A syntaxin 1, Galpha(o), and N-type calcium channel complex at a presynaptic nerve terminal: analysis by quantitative immunocolocalization. Journal of Neuroscience. 2004;24:4070–4081. doi: 10.1523/JNEUROSCI.0346-04.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Lundholt BK, Scudder KM, Pagliaro L. A simple technique for reducing edge effect in cell-based assays. Journal of Biomolecular Screening. 2003;8:566–570. doi: 10.1177/1087057103256465. [DOI] [PubMed] [Google Scholar]

[bib24] Mandal PK, Rossi DJ. Reprogramming human fibroblasts to pluripotency using modified mRNA. Nature Protocols. 2013;8:568–582. doi: 10.1038/nprot.2013.019. [DOI] [PubMed] [Google Scholar]

[bib25] Manders EMM, Verbeek FJ, Aten JA. Measurement of co-localization of objects in dual-colour confocal images. Journal of Microscopy. 1993;169:375–382. doi: 10.1111/j.1365-2818.1993.tb03313.x. [DOI] [PubMed] [Google Scholar]

[bib26] Maroof AM, Keros S, Tyson JA, Ying SW, Ganat YM, Merkle FT, Liu B, Goulburn A, Stanley EG, Elefanty AG, Widmer HR, Eggan K, Goldstein PA, Anderson SA, Studer L. Directed differentiation and functional maturation of cortical interneurons from human embryonic stem cells. Cell Stem Cell. 2013;12:559–572. doi: 10.1016/j.stem.2013.04.008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] Martinat C, Bacci JJ, Leete T, Kim J, Vanti WB, Newman AH, Cha JH, Gether U, Wang H, Abeliovich A. Cooperative transcription activation by Nurr1 and Pitx3 induces embryonic stem cell maturation to the midbrain dopamine neuron phenotype. PNAS. 2006;103:2874–2879. doi: 10.1073/pnas.0511153103. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] Martinez-Barbera JP, Signore M, Boyl PP, Puelles E, Acampora D, Gogoi R, Schubert F, Lumsden A, Simeone A. Regionalisation of anterior neuroectoderm and its competence in responding to forebrain and midbrain inducing activities depend on mutual antagonism between OTX2 and GBX2. Development. 2001;128:4789–4800. doi: 10.1242/dev.128.23.4789. [DOI] [PubMed] [Google Scholar]

[bib29] Motorin Y, Helm M. RNA nucleotide methylation. Wiley interdisciplinary reviews. RNA. 2011;2:611–631. doi: 10.1002/wrna.79. [DOI] [PubMed] [Google Scholar]

[bib30] Nakatani T, Kumai M, Mizuhara E, Minaki Y, Ono Y. Lmx1a and Lmx1b cooperate with Foxa2 to coordinate the specification of dopaminergic neurons and control of floor plate cell differentiation in the developing mesencephalon. Developmental Biology. 2010;339:101–113. doi: 10.1016/j.ydbio.2009.12.017. [DOI] [PubMed] [Google Scholar]

[bib31] Peng C, Aron L, Klein R, Li M, Wurst W, Prakash N, Le W. Pitx3 is a critical mediator of GDNF-induced BDNF expression in nigrostriatal dopaminergic neurons. Journal of Neuroscience. 2011;31:12802–12815. doi: 10.1523/JNEUROSCI.0898-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] Plank C, Zelphati O, Mykhaylyk O. Magnetically enhanced nucleic acid delivery. Ten years of magnetofection-progress and prospects. Advanced Drug Delivery Reviews. 2011;63:1300–1331. doi: 10.1016/j.addr.2011.08.002. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib33] Puelles E, Acampora D, Lacroix E, Signore M, Annino A, Tuorto F, Filosa S, Corte G, Wurst W, Ang SL, Simeone A. Otx dose-dependent integrated control of antero-posterior and dorso-ventral patterning of midbrain. Nature Neuroscience. 2003;6:453–460. doi: 10.1038/nn1037. [DOI] [PubMed] [Google Scholar]

[bib34] Sapet C, Laurent N, de Chevigny A, Le Gourrierec L, Bertosio E, Zelphati O, Béclin C. High transfection efficiency of neural stem cells with magnetofection. BioTechniques. 2011;50:187–189. doi: 10.2144/000113628. [DOI] [PubMed] [Google Scholar]

[bib35] Saucedo-Cardenas O, Quintana-Hau JD, Le WD, Smidt MP, Cox JJ, De Mayo F, Burbach JP, Conneely OM. Nurr1 is essential for the induction of the dopaminergic phenotype and the survival of ventral mesencephalic late dopaminergic precursor neurons. PNAS. 1998;95:4013–4018. doi: 10.1073/pnas.95.7.4013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nature Methods. 2012;9:671–675. doi: 10.1038/nmeth.2089. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib37] Shi Y, Kirwan P, Livesey FJ. Directed differentiation of human pluripotent stem cells to cerebral cortex neurons and neural networks. Nature Protocols. 2012a;7:1836–1846. doi: 10.1038/nprot.2012.116. [DOI] [PubMed] [Google Scholar]

[bib38] Shi Y, Kirwan P, Smith J, Robinson HP, Livesey FJ. Human cerebral cortex development from pluripotent stem cells to functional excitatory synapses. Nature Neuroscience. 2012b;15:477–486. doi: 10.1038/nn.3041. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib39] Stepinski J, Waddell C, Stolarski R, Darzynkiewicz E, Rhoads RE. Synthesis and properties of mRNAs containing the novel "anti-reverse" cap analogs 7-methyl(3'-O-methyl)GpppG and 7-methyl (3'-deoxy)GpppG. Rna. 2001;7:1486–1495. [PMC free article] [PubMed] [Google Scholar]

[bib40] Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell. 2006;126:663–676. doi: 10.1016/j.cell.2006.07.024. [DOI] [PubMed] [Google Scholar]

[bib41] Theka I, Caiazzo M, Dvoretskova E, Leo D, Ungaro F, Curreli S, Managò F, Dell'Anno MT, Pezzoli G, Gainetdinov RR, Dityatev A, Broccoli V. Rapid generation of functional dopaminergic neurons from human induced pluripotent stem cells through a single-step procedure using cell lineage transcription factors. STEM CELLS Translational Medicine. 2013;2:473–479. doi: 10.5966/sctm.2012-0133. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] Warren L, Manos PD, Ahfeldt T, Loh YH, Li H, Lau F, Ebina W, Mandal PK, Smith ZD, Meissner A, Daley GQ, Brack AS, Collins JJ, Cowan C, Schlaeger TM, Rossi DJ. Highly efficient reprogramming to pluripotency and directed differentiation of human cells with synthetic modified mRNA. Cell Stem Cell. 2010;7:618–630. doi: 10.1016/j.stem.2010.08.012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib43] Yu J, Russell JE. Structural and functional analysis of an mRNP complex that mediates the high stability of human beta-globin mRNA. Molecular and Cellular Biology. 2001;21:5879–5888. doi: 10.1128/MCB.21.17.5879-5888.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Combinatorial programming of human neuronal progenitors using magnetically-guided stoichiometric mRNA delivery

Sayyed M Azimi

Steven D Sheridan

Mostafa Ghannad-Rezaie

Peter M Eimon

Mehmet Fatih Yanik

Roles

Abstract

Introduction

Figure 1. Magnetically-guided spotting platform enables localized reconfigurable stoichiometrically-defined mRNA transfection.

Results

Modified mRNA constructs and magnetofection

Focused delivery of magnetic nanoparticles by dual-magnetic array

Programming of the dual-magnet array

Figure 2. Interleaved transfection reduces fluctuations in delivered mRNA copy numbers and stoichiometries.

Screening of transcription factors for dopaminergic neurons

Figure 3. Magnetically reconfigurable spotting platform can screen and identify key transcription factors promoting dopaminergic cell fate.

Figure 4. Immunocytochemical and quantitative analysis confirms midbrain identity dopaminergic neurons generated from screen.

Discussion

Materials and methods

Key resources table.

Cell culture

Synthesis of modified synthetic transcription factor mRNAs

RNA transfection and differentiation

Magnetic spotting methodology

Immunocytochemistry

Gene expression analysis (qRT-PCR)

Cell counting and statistical analysis

Code availability

Acknowledgement

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Additional files

Data availability

References

Decision letter

Roles

Author response

Author response image 1. Source: Courtesy of Inglese, J. & Jang, S.-W. from NIH Screening Center.

Author response image 2. Western blots showing expression levels and lifetimes of Oct4, Sox2, Nanog, Lin28, and MyoD1 proteins in MRC-5 human fetal lung fibroblasts transfected with protein- encoding RNA, relative to levels in hES (H9) and rhabdomyosarcoma (Rh30) cells.

Author response image 3.

Author response image 4.

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases