SFARI workshop highlights challenges and best practices for induced pluripotent stem (iPS) cell models in autism research

Neurons derived from neural stem cellsYirui Sun. Attribution 4.0 International (CC BY 4.0)

The Simons Foundation Autism Research Initiative (SFARI) hosted a virtual workshop on March 14 and 18, 2022, to discuss the best practices for the development and use of induced pluripotent stem (iPS) cell models for autism research. The workshop was organized by SFARI senior scientists Marta Benedetti and Julia Sommer and the director and chief scientific officer of the Nancy Lurie Marks Family Foundation (NLMFF) Clarence Schutt.

 Key take-aways from the workshop include:

  • The importance of a robust and standardized quality control pipeline
  • A need for detailed and standardized donor phenotyping
  • The value of establishing systems to support resource access and data sharing
  • The importance of choosing measurements in cultured tissue that can potentially be correlated to participant phenotypes
  • The potential benefits of creating a consortium to leverage research community strengths

iPS cells can be grown and differentiated into many different brain cell types, as well as 3D brain organoids1,2. IPS cells derived from individuals with neurodevelopmental conditions provide a unique tool to examine early developmental changes in molecular and cellular mechanisms within the human genetic context, thus complementing studies that use human postmortem brain tissue and animal models.

To facilitate this type of research in the autism field, SFARI has supported the generation of numerous iPS cell lines from individuals with genetic variants in high-confidence autism risk genes, who enrolled in the Simons Simplex Collection (SSC) and Simons Searchlight studies. In 2021, SFARI and the NLMFF joined forces to expand these efforts. As of July 2022, the collaboration between SFARI and NLMFF has led to the generation of iPS cell lines from an additional ~60 participants, with plans to establish iPS cell lines from approximately 200 individuals with a variety of autism-linked neurodevelopmental conditions by the end of 2022. These lines are especially valuable, as many participating donors also received detailed clinical and genetic characterizations, and are available to researchers through SFARI Base.

At the workshop, leaders in the field, including several SFARI Investigators, discussed best practices for the generation and characterization of the SFARI iPS cell lines, and they brainstormed ways in which iPS cell models can be used to aid our understanding of autism spectrum disorder (ASD) and other neurodevelopmental conditions.

Participants included Anne Bang (Sanford Burnham Prebys), Aparna Bhaduri (University of California, Los Angeles), Michael Boland (Columbia University), Adam Cohen (Harvard University), Eric Courchesne (University of California, San Diego), Emanuel DiCicco-Bloom (Rutgers University), Fred Etoc (Rumi Scientific), Sami Farhi (Broad Institute of MIT and Harvard), Kyle Fink (University of California, Davis), Daniel Geschwind (University of California, Los Angeles), Joseph Gleeson (University of California, San Diego, n-Lorem Foundation), Joachim Hallmayer (Stanford University), Jacob Hooker (Massachusetts General Hospital, Harvard Medical School), Phil Jensik (Southern Illinois University School of Medicine), Eugenia Jones (Fujifilm Cellular Dynamics, Inc.), Jürgen Knoblich (Institute of Molecular Biotechnology of the Austrian Academy of Sciences), Madeline Lancaster (MRC Laboratory of Molecular Biology), James Millonig (Rutgers University), Guo-Li Ming (University of Pennsylvania), Rick Monsma (New York Stem Cell Foundation Research Institute), Eric Morrow (Brown University), Jennifer Moore (Sampled), Nael Nadif Kasri (Radboud University Medical Centre), Ralda Nehme (Broad Institute of MIT and Harvard), Tomasz Nowakowski (University of California, San Francisco), Sergiu Paşca (Stanford University), David Panchision (National Institute of Mental Health), Daniel Paull (New York Stem Cell Foundation Research Institute), Kathrin Plath (University of California, Los Angeles), Brock Roberts (Allen Institute), Lee Rubin (Harvard Stem Cell Institute), Mustafa Sahin (Boston Children’s Hospital), Neville Sanjana (New York Genome Center, New York University), Michael Sheldon (Sampled), Sally Temple (Neural Stem Cell Institute), Flora Vaccarino (Yale University), Matthijs Verhage (Vrije Universiteit Amsterdam), Michael Wells (University of California, Los Angeles), Gene Yeo (University of California, San Diego).

“IPS cells are the only model that allows longitudinal and mechanistic analyses within relevant human genetic backgrounds that can be related back to current phenotypes in living individuals.”

- Flora Vaccarino

Highlighting the importance of this iPS cell collection, Vaccarino started off the first session saying that “iPS cells are the only model that allows longitudinal and mechanistic analyses within relevant human genetic backgrounds that can be related back to current phenotypes in living individuals.”

While everyone in the meeting acknowledged the importance of iPS cell model systems, they also noted potential challenges. Across the course of the two-day workshop, the participants discussed best practices needed to overcome such hurdles.

A robust and standardized quality control pipeline is critical

Heterogeneity both across lines and within lines was highlighted as a key point of concern — with many of the participants stressing the need for robust quality controls. Temple stressed that we first need to solve the reproducibility and reliability issuesand noted that the Tau Consortium had established additional quality control measures while panelist Eugenia Jones of Fujifilm Cellular Dynamics, Inc., stressed the importance of ongoing and routine line testing. Katherin Plath of the University of California, Los Angeles, also brought up the need to consider X-inactivation (Xi) when working with female-derived lines. As a whole, the workshop participants argued that the SFARI iPS cell resource would benefit from the inclusion of:

  • Multiple lines per individual
  • A diversity of donors with respect to sex and ancestry
  • Control isogenic iPS cell lines and lines from unaffected family members
  • Deep characterization of established lines, including whole genome sequencing of donors to provide a reference for the assessment of genome stability of the iPS cell lines on an ongoing basis
  • Monitoring of Xi, due to Xi erosion in female lines over time

The group also noted that establishing a standardized set of procedures for iPS cell derivation and differentiation would help ensure a robust resource and simplify the process of data sharing. Etoc noted the emphasis that RUMI Scientific has placed on micropatterning technologies3,4, which can standardize stem cell differentiation and the geometry of organoids by providing a lattice often comprised of extracellular matrix components. Verhage echoed this theme in stressing the importance of developing optimized methodologies and testing parameters to ensure standardization, power and scalability of established iPS cell lines. Vaccarino also highlighted her lab’s work developing controlled organoid growth methods to standardize 3D organoid development. Panchision noted that automating routine handling steps during reprogramming, as the New York Stem Cell Foundation has done with the “Global Stem Cell Array5, may be another way to introduce common standards across labs and reduce variability.

“The correct ‘model’ is human, and we should take measurements in humans that you can analogously take in iPS cell models.”

- Gene Yeo

Provide detailed and standardized participant phenotyping

An important aim of autism research is to understand how the brain of affected individuals works and to leverage that understanding to potentially inform intervention strategies. Courchesne and Nehme argued that drawing connections between findings in iPS cell models and behavioral phenotypes requires access to deep clinical phenotypic information from participants.

Courchesne stated that “we should start with clinical phenotypes and use these to inform the readouts taken in iPS cell cultures,” and further suggested that “this should be built in at the program design stage.” Such measures, he argued, should encompass both CNS and non-CNS clinical presentations, neuroimaging and neurophysiological data. Many participants also pointed out the particular utility of collecting phenotypic information that could be ‘recapitulated’ or at least approximated in iPS cell models, the closest maybe being EEG recordings. Yeo noted that “the correct ‘model’ is human, and we should take measurements in humans that you can analogously take in iPS cell models.”

The iPS cells generated through the collaboration between SFARI and NLMFF are mostly derived from participants in the Simons Searchlight cohort, which recruits individuals based on their genetic diagnosis with the goal to better understand how changes in variants in the same gene lead to their particular phenotype(s). To this end, medical and phenotypic data is collected from the participants and is available to researchers through SFARI Base. Importantly, Simons Searchlight participants can be recontacted to participate in additional studies and deeper phenotyping through Research Match, a service supported by SFARI that facilitates participant recruitment.

Despite general agreement among the panelists that it would be ideal to identify causal relationships between changes at the cellular level and human behavior, some raised concerns that this goal may be difficult to achieve and that looking for correlations is likely to be a more feasible approach. Given the mechanistic gap between behavioral phenotypes and cellular measures in iPS cell models, the phenotypic heterogeneity of autism and the technical variability of the cellular measures, large human cohorts and iPS cell collections will be needed to establish solid correlations.

The readouts matter

Building on from the discussions of participant phenotyping, workshop attendees continued this discussion of how to translate culture findings into an understanding of phenotypes. To do so, Geschwind argued that it “requires that we clearly define the cellular phenotypes that show a correlation with the behavioral phenotypes we see in affected individuals.” In his presentation, Geschwind showed how transcriptional networks can be used as a sensor of genetic perturbations and how reversal of perturbed networks to the unaffected state may be a readout to identify therapeutics — under the condition that transcriptional changes also correlate with cellular changes or even human behavioral or neurological phenotypes.

Along these lines, Kasri has shown that neural network activity in human iPS cell-derived neurons grown on micro-electrode arrays demonstrate syndrome-specific activity patterns6. Lancaster stressed that organoid phenotypes should similarly match up with donor clinical measures. The potential for this was highlighted in talks from Knoblich demonstrating that 3D organoids can model relatively late developmental processes7, and Temple’s work8 with the Tau Consortium, which nicely demonstrates that, while still relatively immature, brain organoids can even recapitulate aspects of the aging brain.

“Creating a consortium will drive collaborations between clinical work and basic analyses being done in different labs.”

- Emanuel DiCicco-Bloom

Numbers and controls matter

During the discussion on roadblocks that need to be overcome if iPS cell models for autism are to be used in high throughput assays for biomarker discovery, Monsma emphasized the need for larger numbers of iPS cell lines in diverse genetic backgrounds. Etoc pointed out the convergence of genetic variability with that of technical variability and suggested that it would be most rigorous to have multiple isogenic pairs of lines from affected individuals and CRISPR-generated control lines in diverse genetic backgrounds. Nehme suggested that one may be able to do with fewer isogenic controls by instead using iPS cell lines from a large number of individuals with the same genetic variant in diverse genetic backgrounds. The number of iPS cell lines needed for such a study design would need to be based on power calculations that take the variability of the cellular or molecular readout into account. Panelists agreed that SFARI could play a central role in coordinating among multiple disparate iPS cell resource efforts to increase the number of available cell lines.

Put systems in place to ensure resource access and data sharing

Creating a high-quality iPS cell resource is the first important step. Ensuring researchers can access this resource and share their findings is equally important. During the collection of tissue samples, SFARI ensures that participants are made aware of the planned usage of these materials and that appropriate consents are signed. SFARI ensures that participants of its cohorts know about its strong commitment to data and resource sharing through the consent process and has material transfer agreements for use with academia and industry in place.

Participants also argued for the establishment of a set of reference lines, ideally with diverse genetic backgrounds, and for centralized repositories for protocols and data sharing, in particular for those lines created by SFARI. Rounding out the session was a discussion of the potential utility of establishing a consortium to take advantage of the range of techniques, modalities, samples and levels of questions being addressed by researchers using the SFARI/NLMFF iPS cell resource. DiCicco-Bloom noted that “creating a consortium will drive collaborations between clinical work and basic analyses being done in different labs.” Sahin mentioned that in addition to driving collaboration, a consortium would help standardize protocols and achieve larger scale studies, which are both means to conquer the experimental variability intrinsic to working with human iPS cell models.

A consortium may take many shapes, but at a minimum the workshop participants agreed it could provide a central database where participating labs could individually deposit protocols and raw data derived from SFARI iPS cell collections for data sharing. At a more advanced level, it could facilitate the systematic collaboration between labs and use of shared protocols for differentiation and interrogation, with the goal to conduct a comprehensive and well-controlled characterization of the iPSC resource — including multiple stages of neuronal differentiation, different neuron types and brain organoid models — which can help to better understand cellular, molecular, developmental or functional changes linked to autism risk genetic variants.

“SFARI is uniquely positioned to provide a cell repository that far surpasses other repositories out there.”

- Madeline Lancaster

Moving forward

The workshop aimed to gather input from leaders in the field on best practices for the collection and development of the SFARI iPS cell resource. “The workshop has nicely highlighted what is needed to ensure this is the ‘gold standard’ for iPS cell repositories,” said Lancaster, who also noted that “SFARI is uniquely positioned to provide a cell repository that far surpasses other repositories out there.”

While the workshop was only two days long, “the wealth of insights and ideas that came out of it will help drive concrete actions at SFARI and NLMFF for some time to come,” said Kelsey Martin, director of SFARI and the Simons Foundation Neuroscience Collaborations. One immediate outcome of the workshop will be the addition of sex-matched familial control iPS cell lines to SFARI’s iPS cell resource. In addition, SFARI is considering generating whole genome sequencing data on the source cell material (cryopreserved lymphocytes) to provide a reference data set for the assessment of genomic integrity of the iPS cell lines after several rounds of expansion.


  1. Arlotta P. and Paşca S.P. Curr. Opin. Neurobiol. 56, 194-198 (2019) PubMed
  2. Zhou H. et al. Stem Cell Rev. Rep. 11, 652-665 (2015) PubMed
  3. Velasco V. et al. Microsyst. Nanoeng. 6, 76 (2020) PubMed
  4. Warmflash A. et al. Nat. Methods 11, 847-854 (2014) PubMed
  5. Paull D. et al. Nat. Methods 12, 885-892 (2015) PubMed
  6. Frega M. et al. Nat. Commun. 10, 4928 (2019) PubMed
  7. Eihmüller O.L. et al. Science 375, eabf5546 (2022) PubMed
  8. Bowles K.R. et al. Cell 184, 4547-4563 (2021) PubMed
Recent Workshop and Meeting Reports