We thank the reviewers for their careful reading and constructive feedback on our manuscript. We appreciated that each reviewer offered support and enthusiasm for our work, but raised substantial concerns. We have therefore added substantial new experimental, analytical, and functional datasets to our manuscript. These new data are in agreement with our original conclusions, but we believe fully address the reviewer concerns and significantly strengthen the manuscript. We provide a point-by-point response to each reviewer concern below, but first summarize our new data and analyses, emphasizing points that were raised by multiple reviewers.Technical limitations of scRNA-seq data. Reviewers #2 and #3 both highlighted that the sequencing depth in our original Drop-seq data could limit the interpretation of our dataset, particularly our finding that mitotic progenitors do not show transcriptomic evidence of fate specification. To address this, we first more than doubled the sequencing depth of our Drop-seq dataset. In parallel, we used a version of the SMART-Seq2 protocol to obtain deep sequencing data (4-7,000 genes/cell) from 400 mitotic progenitors. Both new datasets strongly validate our original findings, producing nearly identical developmental trajectories, and exhibiting no evidence of distinct subpopulations of fate-restricted cells in the subventricular zone. As requested by Reviewer #2, we performed the SMART-Seq2 experiments with the FlashTag system, providing an independent validation that our inferred maturation trajectory is consistent with biological time. Finally, we added data produced using the 10X genomics system for 7,000 postmitotic progenitors with significantly increased sequencing depth (1,500-11,000 UMI/cell, 1,000-4,000 genes/cell), and observed identical branching patterns as in our original Drop-seq dataset. Taken together, these new data demonstrate that our original findings and interpretations were not skewed by technical limitations, and could be fully reproduced from complementary scRNA-seq techniques.Identifying the emergence of distinct subtypes. While our initial manuscript focused on our unexpected finding that committed progenitor subsets could not be identified in mitotic progenitors, all reviewers requested further analysis of when distinct interneuron subsets first emerge during development, and how heterogeneity in the eminences could be connected to later timepoints. Leveraging a Dlx6a; Cre ;Ai9 line that uniformly marks interneuron progenitors, We therefore collected a set of 15,522 single cells across a developmental timecourse (E13.5, E18, P10), obtaining 3,432 additional P56 cells from a public resource hosted by the Allen Brain Atlas. We next applied our recently developed 'alignment' tool for scRNA-seq datasets (Butler et al., 2017), to identify shared sources of variation that exist between developmental and adult stages. This allowed us to link the emergence of interneuron subtypes across embryonic, postnatal, and adult timepoints, to identify the earliest emergence of non-overlapping populations that we hypothesize are committed to Vip, Sst, Id2, and Pvalb interneuron fates, and the initial transcriptomic markers that mark these groups across all developmental stages. Remarkably, functional perturbation of one of the strongest discriminating genes between early Pvalb and Sst progenitors, the autism-associated transcription factor Mef2c, led to a complete yet specific ablation of Pvalb interneurons, validating our computational predictions. We believe that these aligned datasets will serve as a valuable resource, in particular by revealing rich lists of genes that likely play similarly important functional roles in the establishment and maintenance of interneuron fates.Biological validation of key computational predictions. All reviewers requested that complementary methods to scRNA-seq be used to broaden support for our key conclusions, and Reviewer #2 specifically requested that specific computational predictions from the sequencing data should be validated and functionally characterized. [NB from Rahul : This paragraph is repetitive, but meant to highlight how new experiments are in the revision in one place, most specificially for Reviewer 2. We could add more here, but may want to highlight techniques and data that are not more scRNA-seq, though we could add Lhx6 discussion here as well]. As discussed above, we generated data using the tagging system from the Jabaudon lab (FlashTag system) as a complementary technique to measure a cell's true biological birth date, and validate our inferred maturation trajectory. We leveraged trancriptomic markers of our three post-mitotic 'branches' to perform genetic fate mapping of MGE branch 3 cells based on specific expression of Lhx8, and find that Lhx8+ progenitors do not give rise to cortical interneurons as predicted, but instead of give rise to X. Finally, we validate a functional role for a key transcription factor, Mef2c, that we predict discriminates between early progenitors of MGE-derived interneuron subtypes, by performing immunohistochemistry on a conditional knockout mouse. These experiments demonstrate the high quality of our computational predictions from Drop-seq data, and more importantly, provide complementary support and validation for our key conclusions.Conceptual insights that advance our understanding of interneuron development. All reviewers expressed enthusiasm for our dataset and analysis, but requested greater clarification of our new insights and advances. In the revised manuscript, we have worked to clearly articulate the following advances : First, we discover that mitotic progenitors exhibit tightly conserved maturation programs across eminences, with downstream differences in fate seeded by only a few genes that have largely been previously described. We see no evidence for detailed fate restriction or non-overlapping subpopulations for individual cells within an eminence, an important insight that weighs heavily on outstanding questions in the field. Second, we discover that upon cell-cycle exit, cells branch into one of three distinct precursor states, representing fate-restricted populations of interneurons and projection neurons. Surprisingly, the markers defining the branches exhibit conservation across the MGE, CGE, LGE, although the proportion of cells in each state does vary across eminences, along with the exact expression programs activated along each branch. Lastly, we trace the emergence of interneuron heterogeneity across a developmental time series, and identify subtle heterogeneity in the postmitotic MGE that separates SST and Pvalb-committed progenitors. We validate that Mef2c, a transcriptomic marker that is enriched in PV -mapping progenitors at multiple time points, is functionally required for the production of Pvalb but not Sst interneurons. Similarly, we propose a list of early markers for early progenitors of Sst, Vip, and Id2 interneurons which are conserved throughout development, and may play similar functional roles.