Cellular barcoding: A technical appraisal

Cellular barcoding involves the tagging of individual cells of interest with unique genetic heritable identifiers or barcodes and is emerging as a powerful tool to address individual cell fates on a large scale. However, as with many new technologies, diverse technical and analytical challenges have emerged. Here, we review those challenges and highlight both the power and limitations of cellular barcoding. We then illustrate the contribution of cellular barcoding to the understanding of hematopoiesis and outline the future potential of this technology. Cellular barcoding involves the tagging of individual cells of interest with unique genetic heritable identifiers or barcodes and is emerging as a powerful tool to address individual cell fates on a large scale. However, as with many new technologies, diverse technical and analytical challenges have emerged. Here, we review those challenges and highlight both the power and limitations of cellular barcoding. We then illustrate the contribution of cellular barcoding to the understanding of hematopoiesis and outline the future potential of this technology. When Rudolph Virchow wrote in 1858 “omnis cellula e cellula” (every cell from a pre-existing cell), cell theory was established [1Mazzarello P.A. unifying concept: The history of cell theory.Nat Cell Biol. 1999; 1: E13-E15Crossref PubMed Scopus (116) Google Scholar]. Along with the work of Louis Pasteur and others, theories of spontaneous generation were discarded, initiating the search into how complex life originates from a single cell. Pre-occupation with the single cell has waxed and waned over the years; however, the study of how individual stem and progenitor cells make fate decisions to generate complex tissues is currently at the forefront of biology. Although much progress has been made in lower organisms, perennial questions surrounding single cell fate in higher-order animals still dominate. Do all progenitors contribute equally in cell numbers? At what stage is diversity generated? Is the diversity the result of intrinsic or extrinsic processes? Do these processes involve stochastic or deterministic regulation? What are the factors responsible for the generation of diversity? One of the most well studied systems addressing such questions is the hematopoietic system. With some variations on the theme, the current paradigm states that hematopoietic stem cells divide to give rise to multipotent progenitors, which then broadly restrict into lymphoid, myeloid and megakaryocyte/erythroid progenitors en route to the more differentiated sublineages [2Luc S. Buza-Vidas N. Jacobsen S.E. Delineating the cellular pathways of hematopoietic lineage commitment.Semin Immunol. 2008; 20: 213-220Crossref PubMed Scopus (38) Google Scholar, 3Ema H. Morita Y. Suda T. Heterogeneity and hierarchy of hematopoietic stem cells.Exp Hematol. 2014; 42: 74-82.e72Abstract Full Text Full Text PDF PubMed Scopus (92) Google Scholar]. This sequence can be seen in the often-drawn branching diagrams of hematopoiesis and assumes that progenitors lose multipotency as hematopoiesis proceeds with division and differentiation. Evidence to back this assumption includes the ability of a single stem cell to reconstitute the entire hematopoietic system [4Osawa M. Hanada K. Hamada H. Nakauchi H. Long-term lymphohematopoietic reconstitution by a single CD34-low/negative hematopoietic stem cell.Science. 1996; 27: 242-245Crossref Scopus (1689) Google Scholar] and the ability of downstream progenitors to make some but not all subtypes in vivo [5Reya T. Morrison S.J. Clarke M.F. Weissman I.L. Stem cells, cancer, and cancer stem cells.Nature. 2001; 414: 105-111Crossref PubMed Scopus (7653) Google Scholar]. Over the last decades, our understanding of hematopoiesis has evolved with leaps that often coincided with changes in assays and technology. Originally, Till and McCulloch investigated the ability of transferred progenitors to generate, from a single cell, colony-forming units in the spleens of irradiated recipients [6McCulloch E.A. Till J.E. Perspectives on the properties of stem cells.Nat Med. 2005; 11: 1026-1028Crossref PubMed Scopus (106) Google Scholar]. The time at which colonies were harvested from recipients and the typology of cells re-transferred into new recipients allowed researchers to identify short-, intermediate-, and long-term reconstituting cells in the bone marrow, thus establishing the concept of a stem cell. The advent of in vitro soft agar colony-forming assays, developed by Don Metcalf, recapitulated some of the in vivo findings and led to the discovery of colony-stimulating factors and progenitors able to generate differentiated cells of many lineages [7Metcalf D. Moore M.A.S. Haemopoietic cells. North-Holland, Amsterdam1971Google Scholar]. Later, the ability to stain for specific cell surface markers and sort cells by flow cytometry revolutionized the dissection of intermediate progenitor stages [5Reya T. Morrison S.J. Clarke M.F. Weissman I.L. Stem cells, cancer, and cancer stem cells.Nature. 2001; 414: 105-111Crossref PubMed Scopus (7653) Google Scholar]. Subsequently, the molecular mechanisms governing these processes were identified using knockdowns, knockouts, and reporter mice. Combined, these technologic progressions have generated our current models of hematopoiesis. Some of the questions listed earlier regarding individual cell fate cannot be fully addressed by these assays. Even if in vitro assays can track individual cell fate, they can bias cell differentiation and may only partially reproduce the complexity of lineage fates. In other words, although these assays reveal what a single cell can produce under the experimental conditions chosen, they do not reveal what the output of a given progenitor would have been in vivo. Finally, in vivo population-based assays, by definition, miss levels of complexity of individual cell commitment. Thus, methods that are able to track single cell fate in vivo are ultimately required to establish true lineage fate. In vivo tracking of single cell fate has been achieved through hematopoietic reconstitution from a single cell [4Osawa M. Hanada K. Hamada H. Nakauchi H. Long-term lymphohematopoietic reconstitution by a single CD34-low/negative hematopoietic stem cell.Science. 1996; 27: 242-245Crossref Scopus (1689) Google Scholar], the use of retrovirus-tagged progenitors, with clonal output extrapolated from Southern blots [8Lemischka I.R. What we have learned from retroviral marking of hematopoietic stem cells.Curr Top Microbiol Immunol. 1992; 177: 59-71PubMed Google Scholar, 9Kreso A. O'Brien C.A. van Galen P. et al.Variable clonal repopulation dynamics influence chemotherapy response in colorectal cancer.Science. 2013; 339: 543-548Crossref PubMed Scopus (512) Google Scholar], or through multiplexed expression of fluorophores [10Livet J. Weissman T.A. Kang H. et al.Transgenic strategies for combinatorial expression of fluorescent proteins in the nervous system.Nature. 2007; 450: 56-62Crossref PubMed Scopus (1297) Google Scholar, 11Schepers A.G. Snippert H.J. Stange D.E. et al.Lineage tracing reveals Lgr5+ stem cell activity in mouse intestinal adenomas.Science. 2012; 337: 730-735Crossref PubMed Scopus (819) Google Scholar, 12Rios A.C. Fu N.Y. Lindeman G.J. Visvader J.E. In situ identification of bipotent stem cells in the mammary gland.Nature. 2014; 506: 322-327Crossref PubMed Scopus (359) Google Scholar]. However, these methods are either restricted to dozens of clones (e.g., single cell transfer, multiplexed fluorophore expression) or have a restricted dynamic range (Southern blot analyses). Some labs have valiantly scaled up single cell transfer experiments [13Sieburg H.B. Cho R.H. Dykstra B. Uchida N. Eaves C.J. Muller-Sieburg C.E. The hematopoietic stem compartment consists of a limited number of discrete stem cell subsets.Blood. 2006; 107: 2311-2316Crossref PubMed Scopus (168) Google Scholar, 14Dykstra B. Kent D. Bowie M. et al.Long-term propagation of distinct hematopoietic differentiation programs in vivo.Cell Stem Cell. 2007; 1: 218-229Abstract Full Text Full Text PDF PubMed Scopus (416) Google Scholar, 15Yamamoto R. Morita Y. Ooehara J. et al.Clonal analysis unveils self-renewing lineage-restricted progenitors generated directly from hematopoietic stem cells.Cell. 2013; 154: 1112-1126Abstract Full Text Full Text PDF PubMed Scopus (428) Google Scholar], yet the process remains labor intensive and limited to the detection of more common output patterns. Although these studies represent landmarks in single cell fate tracking, systems that allow higher throughput, better quantitation, and tracking of large numbers of individual cells simultaneously would benefit the field. A new technology, termed cellular barcoding, originally developed by our group [16Schepers K. Swart E. Jeroen W.J. et al.Dissecting T cell lineage relationships by cellular barcoding.J Exp Med. 2008; 205: 2309-2318Crossref PubMed Scopus (90) Google Scholar, 17Van Heijst J.W. Gerlach C. Swart E. et al.Recruitment of antigen-specific CD8+ T cells in response to infection is markedly efficient.Science. 2009; 325: 1265-1269Crossref PubMed Scopus (109) Google Scholar], with more recent versions from our group [18Gerlach C. Rohr J.C. Penié L. et al.Heterogeneous differentiation patterns of individual CD8+ T cells.Science. 2013; 340: 635-639Crossref PubMed Scopus (240) Google Scholar, 19Naik S.H. Perié L. Swart E. et al.Diverse and heritable lineage imprinting of early haematopoietic progenitors.Nature. 2013; 496: 229-232Crossref PubMed Scopus (260) Google Scholar] and others [20Gerrits A. Dykstra B. Kalmykowa O.J. et al.Cellular barcoding tool for clonal analysis in the hematopoietic system.Blood. 2010; 115: 2610-2618Crossref PubMed Scopus (161) Google Scholar, 21Lu R. Neff N.F. Quake S.R. Weissman I.L. Tracking single hematopoietic stem cells in vivo using high-throughput sequencing in conjunction with viral genetic barcoding.Nat Biotechnol. 2011; 29: 928-933Crossref PubMed Scopus (283) Google Scholar, 22Cheung A.M. Nguyen L.V. Carles A. et al.Analysis of the clonal growth and differentiation dynamics of primitive barcoded human cord blood cells in NSG mice.Blood. 2013; 122: 3129-3137Crossref PubMed Scopus (69) Google Scholar, 23Nguyen L.V. et al.Clonal analysis via barcoding reveals diverse growth and differentiation of transplanted mouse and human mammary stem cells.Cell Stem Cell. 2014; 14: 253-263Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar], is emerging as a powerful tool to address individual cell fates on a large scale. The basic principle underlying cellular barcoding involves the tagging of individual cells of interest with unique heritable identifiers or barcodes (Fig. 1). The barcode collection (or “barcode library”) is constructed artificially from semirandom, noncoding stretches of DNA and is delivered into the genome of progenitor cells of interest using a lenti- or retroviral vector. Barcode-labeled cells are then transferred into recipient mice and allowed to develop in vivo into the various lineages. As the barcode is integrated into the genome, each subsequent daughter cell also inherits this genetic tag. In this way, different cell types can be isolated later, and their genomic DNA assessed for its barcode signature (Fig. 1). Originally, the detection of barcodes was achieved using a custom DNA microarray [16Schepers K. Swart E. Jeroen W.J. et al.Dissecting T cell lineage relationships by cellular barcoding.J Exp Med. 2008; 205: 2309-2318Crossref PubMed Scopus (90) Google Scholar, 17Van Heijst J.W. Gerlach C. Swart E. et al.Recruitment of antigen-specific CD8+ T cells in response to infection is markedly efficient.Science. 2009; 325: 1265-1269Crossref PubMed Scopus (109) Google Scholar], but this technique has now been replaced by next-generation sequencing [18Gerlach C. Rohr J.C. Penié L. et al.Heterogeneous differentiation patterns of individual CD8+ T cells.Science. 2013; 340: 635-639Crossref PubMed Scopus (240) Google Scholar, 19Naik S.H. Perié L. Swart E. et al.Diverse and heritable lineage imprinting of early haematopoietic progenitors.Nature. 2013; 496: 229-232Crossref PubMed Scopus (260) Google Scholar, 20Gerrits A. Dykstra B. Kalmykowa O.J. et al.Cellular barcoding tool for clonal analysis in the hematopoietic system.Blood. 2010; 115: 2610-2618Crossref PubMed Scopus (161) Google Scholar, 21Lu R. Neff N.F. Quake S.R. Weissman I.L. Tracking single hematopoietic stem cells in vivo using high-throughput sequencing in conjunction with viral genetic barcoding.Nat Biotechnol. 2011; 29: 928-933Crossref PubMed Scopus (283) Google Scholar, 22Cheung A.M. Nguyen L.V. Carles A. et al.Analysis of the clonal growth and differentiation dynamics of primitive barcoded human cord blood cells in NSG mice.Blood. 2013; 122: 3129-3137Crossref PubMed Scopus (69) Google Scholar, 23Nguyen L.V. et al.Clonal analysis via barcoding reveals diverse growth and differentiation of transplanted mouse and human mammary stem cells.Cell Stem Cell. 2014; 14: 253-263Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar]. The latter affords better quantitation than microarray and allows massively parallel processing of samples by the use of index primers for different samples, which are then pooled for sequencing. By comparing the shared and distinct barcodes between cell types, one can establish progenitor fates at the single cell level on a large scale. Using this technology is akin to doing hundreds of single cell assays simultaneously in one mouse and clearly has power in addressing many questions in the field of single cell development. As with many new technologies, however, diverse technical and analytical challenges have emerged with cellular barcoding. Here, we review those challenges and highlight both its power and limitations, give examples of experimental results, and outline the future potential of this technology. We have summarized a number of frequently asked questions about cellular barcoding experiments in Box 1. Our advice to research groups contemplating the use of cellular barcoding is to pay specific attention to three experimental factors: (1) library size, (2) the transduction step, and (3) technical replicates.Box 1Frequently asked questions1.What happens if cells have multiple barcodes? In general, most protocols aim for a low transduction efficiency to reduce the probability of multiple integrations within individual cells. The progenitors with multiple integrations will read as multiple progenitors with the same fate (Fig. 2). This is something to take into account for interpretation of the data, but will have only marginal influence on the proportion of progenitors assigned with a particular output. For example, a double integration results in one false-positive progenitor of the same fate. However, the output patterns themselves are not influenced. Only in a case in which one subset of progenitors has a much higher likelihood of receiving multiple integrations would the relative occurrence of particular output patterns be influenced.2.What happens if multiple progenitors have the same barcode? If two different progenitors receive, by chance, the same barcode (repeat use), false lineage relationships can be assigned (Fig. 2). To prevent this, the output of a fraction of barcoded progenitors much smaller than the total library complexity must be measured. Furthermore, by transfer of progenitors from the same transduction batch into a minimum of two separate mice, and subsequent comparison of the barcodes between those mice, it is possible to estimate the frequency of repeat use.3.Can integration site affect the outcome? Insertional mutagenesis has successfully been used by a number of groups to identify oncogenes, so it is apparent that retroviral insertions can alter cell fate. Importantly though, these experiments rely on a high multiplicity of infection and use large numbers of progenitors to increase the probability of hitting a relevant locus. Furthermore, in these experiments, analysis is started only once a hyperproliferative event is seen [36Ranzani M. Cesana D. Bartholomae C.C. et al.Lentiviral vector-based insertional mutagenesis identifies genes associated with liver cancer.Nat Methods. 2013; 10: 155-161Crossref PubMed Scopus (62) Google Scholar]. In barcoding experiments, we typically have a single integration per cell and only use hundreds of progenitors per experiment, making insertion site effects unlikely. In line with this, progenitor output patterns are reproducible between mice, and we have never observed any long-term differences in cellular output relative to the co-transferred non-transduced cells (except for T cells, see point 5). Others have noted that integration is benign with no genomic regions associated with any particular hematopoietic stem cell subtype bias [35Kim S. Kim N. Presson A.P. et al.Dynamics of HSPC repopulation in nonhuman primates revealed by a decade-long clonal-tracking study.Cell Stem Cell. 2014; 14: 473-485Abstract Full Text Full Text PDF PubMed Scopus (65) Google Scholar]. Thus, although we cannot formally exclude the possibility of an incidental integration that influences cell fate, there are no data to suggest that this is a significant factor under the conditions tested.4.Can exposure of a progenitor to a virus change its fate? If pattern recognition machinery (e.g., toll-like receptors) operates to recognize and thereby change a cell’s fate, then the fate readout could be influenced by the barcode labeling procedure and may thereby no longer reflect the progenitor’s unperturbed fate. The only way we currently assess this is by comparing the output to each lineage of green fluorescent protein (GFP)-positive versus GFP-negative donor cells. With the exception of T cells (see point 5 below), we do not find a large difference in this proportion in our experiments, making it unlikely that cell fate is altered by the tagging procedure for those cell types that we analyze. However, we cannot formally exclude that cell fate would be influenced by viral modification in other settings.5.What if particular progenitors are not amenable to transduction? This can be an issue. For example, our experience is that T cells do not develop from barcoded GFP+ donor lymphoid-primed multipotent progenitors (LMPPs), even though donor GFP– LMPPs can yield T cells. Furthermore, the lack of detection of GFP+ T cells from barcode-labeled LMPPs appears not to be caused by promoter silencing, as T cells can develop from barcoded hematopoietic stem cells (HSC). Conceivably, virus exposure or integration diverts fate or kills the progenitor. Alternatively, the existence of a subset of progenitors within LMPPs that are specialized in T-cell production and are resistant to transduction may be postulated.6.Do barcoding experiments reveal what a progenitor “can” make? No, barcoding experiments measure cellular outcome rather than cellular potency. After transfer into a host, a progenitor’s fate may be influenced by its site of engraftment even if it had potency for alternative fates. Notably, clone-splitting assays in conjunction with cellular barcoding can reveal whether the daughters of a single cell have a similar output pattern, providing information on cellular potency at that level (see Box 2). This strategy was used, for instance, to demonstrate that the variability in the output patterns of LMPPs is at least in part caused by intrinsic differences in potency. 1.What happens if cells have multiple barcodes? In general, most protocols aim for a low transduction efficiency to reduce the probability of multiple integrations within individual cells. The progenitors with multiple integrations will read as multiple progenitors with the same fate (Fig. 2). This is something to take into account for interpretation of the data, but will have only marginal influence on the proportion of progenitors assigned with a particular output. For example, a double integration results in one false-positive progenitor of the same fate. However, the output patterns themselves are not influenced. Only in a case in which one subset of progenitors has a much higher likelihood of receiving multiple integrations would the relative occurrence of particular output patterns be influenced.2.What happens if multiple progenitors have the same barcode? If two different progenitors receive, by chance, the same barcode (repeat use), false lineage relationships can be assigned (Fig. 2). To prevent this, the output of a fraction of barcoded progenitors much smaller than the total library complexity must be measured. Furthermore, by transfer of progenitors from the same transduction batch into a minimum of two separate mice, and subsequent comparison of the barcodes between those mice, it is possible to estimate the frequency of repeat use.3.Can integration site affect the outcome? Insertional mutagenesis has successfully been used by a number of groups to identify oncogenes, so it is apparent that retroviral insertions can alter cell fate. Importantly though, these experiments rely on a high multiplicity of infection and use large numbers of progenitors to increase the probability of hitting a relevant locus. Furthermore, in these experiments, analysis is started only once a hyperproliferative event is seen [36Ranzani M. Cesana D. Bartholomae C.C. et al.Lentiviral vector-based insertional mutagenesis identifies genes associated with liver cancer.Nat Methods. 2013; 10: 155-161Crossref PubMed Scopus (62) Google Scholar]. In barcoding experiments, we typically have a single integration per cell and only use hundreds of progenitors per experiment, making insertion site effects unlikely. In line with this, progenitor output patterns are reproducible between mice, and we have never observed any long-term differences in cellular output relative to the co-transferred non-transduced cells (except for T cells, see point 5). Others have noted that integration is benign with no genomic regions associated with any particular hematopoietic stem cell subtype bias [35Kim S. Kim N. Presson A.P. et al.Dynamics of HSPC repopulation in nonhuman primates revealed by a decade-long clonal-tracking study.Cell Stem Cell. 2014; 14: 473-485Abstract Full Text Full Text PDF PubMed Scopus (65) Google Scholar]. Thus, although we cannot formally exclude the possibility of an incidental integration that influences cell fate, there are no data to suggest that this is a significant factor under the conditions tested.4.Can exposure of a progenitor to a virus change its fate? If pattern recognition machinery (e.g., toll-like receptors) operates to recognize and thereby change a cell’s fate, then the fate readout could be influenced by the barcode labeling procedure and may thereby no longer reflect the progenitor’s unperturbed fate. The only way we currently assess this is by comparing the output to each lineage of green fluorescent protein (GFP)-positive versus GFP-negative donor cells. With the exception of T cells (see point 5 below), we do not find a large difference in this proportion in our experiments, making it unlikely that cell fate is altered by the tagging procedure for those cell types that we analyze. However, we cannot formally exclude that cell fate would be influenced by viral modification in other settings.5.What if particular progenitors are not amenable to transduction? This can be an issue. For example, our experience is that T cells do not develop from barcoded GFP+ donor lymphoid-primed multipotent progenitors (LMPPs), even though donor GFP– LMPPs can yield T cells. Furthermore, the lack of detection of GFP+ T cells from barcode-labeled LMPPs appears not to be caused by promoter silencing, as T cells can develop from barcoded hematopoietic stem cells (HSC). Conceivably, virus exposure or integration diverts fate or kills the progenitor. Alternatively, the existence of a subset of progenitors within LMPPs that are specialized in T-cell production and are resistant to transduction may be postulated.6.Do barcoding experiments reveal what a progenitor “can” make? No, barcoding experiments measure cellular outcome rather than cellular potency. After transfer into a host, a progenitor’s fate may be influenced by its site of engraftment even if it had potency for alternative fates. Notably, clone-splitting assays in conjunction with cellular barcoding can reveal whether the daughters of a single cell have a similar output pattern, providing information on cellular potency at that level (see Box 2). This strategy was used, for instance, to demonstrate that the variability in the output patterns of LMPPs is at least in part caused by intrinsic differences in potency. One of the very first steps in cellular barcoding technology is choosing the size of the library, that is, the number of different DNA barcodes that should be available. The length of the DNA stretch predicts maximal theoretical diversity. In practice, however, library diversity depends on other variables, in particular the cloning of the library into delivery vectors. Both the length of the barcodes and the size of the libraries differ between the current versions of cellular barcoding [18Gerlach C. Rohr J.C. Penié L. et al.Heterogeneous differentiation patterns of individual CD8+ T cells.Science. 2013; 340: 635-639Crossref PubMed Scopus (240) Google Scholar, 20Gerrits A. Dykstra B. Kalmykowa O.J. et al.Cellular barcoding tool for clonal analysis in the hematopoietic system.Blood. 2010; 115: 2610-2618Crossref PubMed Scopus (161) Google Scholar, 21Lu R. Neff N.F. Quake S.R. Weissman I.L. Tracking single hematopoietic stem cells in vivo using high-throughput sequencing in conjunction with viral genetic barcoding.Nat Biotechnol. 2011; 29: 928-933Crossref PubMed Scopus (283) Google Scholar, 22Cheung A.M. Nguyen L.V. Carles A. et al.Analysis of the clonal growth and differentiation dynamics of primitive barcoded human cord blood cells in NSG mice.Blood. 2013; 122: 3129-3137Crossref PubMed Scopus (69) Google Scholar]. Increasing the size of the barcode library can be beneficial, as it allows analysis of the fate of a larger number of cells within a single animal without “repeat use” of barcodes (see below). On the other hand, only the exact sequence composition of libraries with a lower diversity can accurately be described presently by second-generation sequencing. Such library sequencing allows the generation of a reference file of “true barcodes,” making it straightforward to filter out the large number of sequencing and polymerase chain reaction (PCR) errors in experimental data. Different transduction times, ranging from 6 to 48 hours, have also been used. We advocate a short transduction culture of 6 hours to reduce the chance of biasing the fate of the cells. During transduction, two types of events can occur that have consequences on the interpretation of the data: multiple integration and repeat use. Multiple integrations of barcodes into one cell (Fig. 2) would influence quantification, as they would read as multiple progenitors with the same fate, rather than one cell labeled with two different barcodes. To avoid such an issue, most protocols aim for low transduction efficiency, thereby reducing the probability of multiple integrations. The transduction efficiency has to be monitored for every system and depends on the amenability of progenitors for transduction. For example, Lu et al. established that at their transduction rate of 50%, >95% of cells had one integration [21Lu R. Neff N.F. Quake S.R. Weissman I.L. Tracking single hematopoietic stem cells in vivo using high-throughput sequencing in conjunction with viral genetic barcoding.Nat Biotechnol. 2011; 29: 928-933Crossref PubMed Scopus (283) Google Scholar]. In our system, we typically aim for 10–15% transduction efficiency to decrease this chance even further. Multiple integrations within one cell can be considered a relatively minor problem, which would lead to overestimation of the amount of data obtained, but would not influence the output patterns inferred (see Box 1). A second consideration during transduction is that the same barcode in different virus particles may integrate into different cells. This repeat usage of the barcode forms a more significant concern, as it would result in false assignments of lineage relationship (Fig. 2). Importantly, by transfer of progenitors from the same transduction batch into at least two separate mice, followed by subsequent comparison of the barcodes recovered from those mice, it is possible to estimate the frequency of repeat barcode use within one mouse. Repeat usage depends on the frequency of virus with the same barcode within the library, the transduction efficiency, the size of the library, the number of labeled cells, and the engraftment probability. For the design of an experiment, the number of labeled cells transferred needs to be optimized while the risk of multiple integrations and repeat use is limited. As a rule of thumb, the number of progenitors tracked per mouse should be equivalent to 10% of the library diversity to limit the issue of repeat use. Simulation tools may be used to further optimize parameters [21Lu R. Neff N.F. Quake S.R. Weissman I.L. Tracking single hematopoietic stem cells in vivo using high-throughput sequencing in conjunction with viral genetic barcoding.Nat Biotechnol. 2011; 29: 928-933Crossref PubMed Scopus (283) Google Scholar], and the development of a mathematical formalism to design future experiments would be attractive. To establish lineage relationships between cell types, there must be a high degree of confidence that a determined barcode signature is accurate and is not a reflection of barcode contamination or hampered by poor expansion of progenitors or poor recovery of daughters.