Dissecting T cell lineage relationships by cellular barcoding - PubMed (original) (raw)

Dissecting T cell lineage relationships by cellular barcoding

Koen Schepers et al. J Exp Med. 2008.

Abstract

T cells, as well as other cell types, are composed of phenotypically and functionally distinct subsets. However, for many of these populations it is unclear whether they develop from common or separate progenitors. To address such issues, we developed a novel approach, termed cellular barcoding, that allows the dissection of lineage relationships. We demonstrate that the labeling of cells with unique identifiers coupled to a microarray-based detection system can be used to analyze family relationships between the progeny of such cells. To exemplify the potential of this technique, we studied migration patterns of families of antigen-specific CD8(+) T cells in vivo. We demonstrate that progeny of individual T cells rapidly seed independent lymph nodes and that antigen-specific CD8(+) T cells present at different effector sites are largely derived from a common pool of precursors. These data show how locally primed T cells disperse and provide a technology for kinship analysis with wider utility.

PubMed Disclaimer

Figures

Figure 1.

Figure 1.

Cellular barcoding strategy. (A) To allow the analysis of lineage relationships between cell populations, a retroviral plasmid library was constructed containing a GFP marker gene and a semirandom stretch of 98 bp of noncoding DNA (the barcode). The barcodes that are present in this library were amplified by PCR and individually spotted on microarray slides. In addition, from this plasmid library, a retroviral library was generated that is used to transduce T cells or other cells. Barcode-labeled cells are introduced into mice to allow proliferation and differentiation, and at different time points after infusion, populations of cells can be isolated for kinship analysis by comparing the barcodes present within the different cell populations. Barcode analysis is performed by hybridization on microarray slides containing the individual barcodes. (B) Analysis of a hybridization with PCR products of 273 individual E. coli clones of the master library. The PCR products were pooled into three pools (A, B, and C), each containing PCR products of 91 clones. The x-axes show nonnormalized fluorescence intensities of a Cy3-labeled sample that contains PCR products of pools A and B, and the y-axes show nonnormalized fluorescence intensities of a Cy5-labeled sample that contains PCR products of pools B and C. Barcodes that are present in pools A, B, and C are plotted on the left and are labeled green, blue, and red, respectively. Barcodes that are not present in pools A, B or C are plotted on the right. Colored numbers indicate the percentage of barcodes present in the respective gates (black rectangles) as a percentage of the total number of barcodes that is present in the corresponding pool.

Figure 2.

Figure 2.

Proof of principle and sensitivity of kinship analysis by cellular barcoding. (A) Barcode analysis of indicated samples (at day 7 after challenge) from mice that received 1,000 barcode-labeled OT-I T cells and that were subsequently challenged by i.v. LM-OVA infection. Plots represent the fluorescence intensities of the barcode microarray spots. Left and middle: Dot plots of barcode analysis of T cells isolated from two half-samples from the same spleen for two individual mice. Right: Dot plot of barcode analysis of T cells isolated from spleens of two different mice. Numbers indicate the number of barcodes present in each quadrant (black rectangles; cutoff used, P < 0.0005). With the assumption that ∼3,300 of the 4,743 barcodes can effectively be used for lineage analysis (supplemental Materials and methods, available at

http://www.jem.org/cgi/content/full/jem.20072462/DC1

) and that each barcode has the same probability to participate in the T cell response, the expected overlap between barcodes present in the two unrelated cell populations in A is 200/3,300 × 200 = 12 (in which 200 is the approximate number of barcodes within a half-sample of a spleen), which is close to the observed overlap between the two unrelated samples (i.e., 17 barcodes). (B) Barcode analysis of genomic DNA mixtures of the two spleen samples shown in A. Plots represent the fluorescence intensities of the barcode microarray spots. The y-axes show fluorescence intensities of a sample that contains a 1:1 mixture of genomic DNA of spleen 1 and 2. The x-axes show fluorescence intensities of samples that contain a 1:1, 3:1, 7:1, 15:1, or 1:0 mixture of genomic DNA of spleen 1 and 2, respectively. Barcodes uniquely present in spleen 1 and spleen 2 are labeled green and red, respectively. Numbers above each plot indicate the level at which the cells of spleen 2 are overrepresented in the sample on the y-axis as compared with the sample on the x-axis. R values represent the correlation between hybridization signals of all barcodes present in either spleen 1 or 2.

Figure 3.

Figure 3.

Simultaneous priming of OT-I T cells in tumor DLN and lung DLN. Four groups of B6 Ly5.1+ mice that received 3 × 106 CFSE-labeled mock-transduced OT-I T cells (Ly5.2+) on day 0 were challenged intranasally with the indicated influenza virus on day 0 and challenged s.c. with the indicated tumor cells on day 1. Plots show representative flow cytometric analyses of tumor-draining axillary/inguinal LNs (TDLN) and lung-draining mediastinal LNs (LDLN) at different time points after initial challenge. In each plot the same amount of living CD8+ lymphocytes is depicted. Plots are representative for three to four mice per experiment, out of two experiments.

Figure 4.

Figure 4.

Redistribution of T cell families over LN beds through time. Barcode analysis of T cell populations isolated from draining LNs in mice that received 10,000 barcode-labeled OT-I T cells and that were subsequently challenged intranasally with WSN-OVA influenza virus (day 0) and s.c. with EL4-OVA cells (day 1). At indicated days after challenge, TDLN and LDLN were each split into two half-samples that were separately cultured for 4 d and used for barcode analysis. The results represent data from two to three mice analyzed separately within one experiment. (A and B) Representative dot plots of the fluorescence intensities of the barcode microarray spots at days 5 (A) and 10 (B). Left and middle, top two rows: dot plots of barcode analysis of two pools of T cells isolated from the same tissue for two individual mice. Right, top two rows: dot plots of barcode analysis of T cells isolated from the TDLN versus T cells isolated from the LDLN for two individual mice. Bottom row: evaluation of background overlap by comparison of samples from mouse 1 with samples from mouse 2. Numbers indicate the barcodes that are present within each quadrant (black rectangles; cutoff used, P < 0.005). (C) Percentage of barcode overlap between different tissue samples. Each sample was compared with a second sample generated from the same tissue (TDLN or LDLN) or from a different tissue (TDLN vs LDLN), either of the same mouse (intra) or of a different mouse (inter). The percentages indicate the number of barcodes that is present within the top right quadrant as a fraction of the total number of barcodes present in the top left, top right, and bottom right quadrant (as indicated in A and B). By normalizing for background and maximal overlap, data can be converted to percentage identity between T cell families at both sites (numbers in text). Percentage identity of T cell families in the control comparisons (TDLN-X vs. LDLN-Y) was 1, 0, −1, and 0% on days 5, 6, 8, and 10, respectively.

Figure 5.

Figure 5.

T cells present at lung and tumor effector sites are derived from the same precursors. Barcode analysis of T cell populations from effector sites in mice that received 10,000 barcode-labeled OT-I T cells and that were subsequently challenged intranasally with WSN-OVA influenza virus (day 0) and s.c. with EL4-OVA cells (day 1). At day 8 after challenge, T cells were isolated from tumor and lung tissue for barcode analysis. Results are representative of two independent experiments in which seven mice were analyzed separately in total. (A) Representative dot plots of the fluorescence intensities of the barcode microarray spots. Left and middle, top two rows: dot plots of barcode analysis of two pools of T cells isolated from the same tissue for two individual mice. Right, top two rows: dot plots of barcode analysis of T cells isolated from tumor versus T cells isolated from lung for two individual mice. Bottom row: evaluation of background overlap by comparison of samples from mouse 1 with samples from mouse 2. Numbers indicate the barcodes that are present within each quadrant (black rectangles; cutoff used, P < 0.00005). (B) Percentage of barcode overlap between different tissue samples. Each sample was compared with a second sample generated from the same tissue (tumor or lung) or from a different tissue (tumor vs. lung), either of the same mouse (intra) or of a different mouse (inter). The percentages indicate the number of barcodes that is present within the top right quadrant as a fraction of the total number of barcodes present in the top left, top right, and bottom right quadrant (as indicated in A). By normalizing for background and maximal overlap, data can be converted to percentage identity between T cell families at both sites (numbers in text). Percentage identity of T cell families in the control comparison (tumor-X vs. lung-Y) was 0%.

Figure 6.

Figure 6.

T cells present at gut and tumor effector sites are derived from the same precursors. Barcode analysis of T cell populations in mice that received 10,000 barcode-labeled OT-I T cells and that were subsequently challenged orally with LM-OVA and s.c. with EL4-OVA cells (both on the same day). At day 7 after challenge, T cells were isolated from tumor and gut tissue for barcode analysis. Before analysis, T cells isolated from gut tissue from each mouse were split into two half-samples that were subsequently cultured for 4 d. Results represent data from four mice analyzed separately within one experiment. (A) Representative dot plots of the fluorescence intensities of the barcode-microarray spots. Left and middle, top two rows: dot plots of barcode analysis of two pools of T cells isolated from the same tissue for two individual mice. Right, top two rows: dot plots of barcode analysis of T cells isolated from the gut versus T cells isolated from tumor for two individual mice. Bottom row: evaluation of background overlap by comparison of samples from mouse 1 with samples from mouse 2. Numbers indicate the barcodes that are present within each quadrant (black rectangles; cutoff used, P < 0.0005. (B) Percentage of barcode overlap between different tissue samples. Each sample was compared with a second sample generated from the same tissue (gut or tumor) or from a different tissue (gut vs. tumor), either of the same mouse (intra) or of a different mouse (inter). The percentages indicate the number of barcodes that is present within the top right quadrant as a fraction of the total number of barcodes present in the top left, top right, and bottom right quadrant (as indicated in A). By normalizing for background and maximal overlap, data can be converted to percentage identity between T cell families at both sites (numbers in text). Percentage identity of T cell families in the control comparison (tumor-X vs. gut-Y) was −2%.

Figure 7.

Figure 7.

Pathways for intermingling of T cell families. (A) After priming, two local and genetically distinct populations of antigen-specific T cells develop. Upon LN exit, these T cell families accumulate at both effector sites, and mixing of T cell families within the DLN occurs as a secondary phenomenon upon migration or passive transport of T cells via afferent lymph. (B) After priming, two local and genetically distinct populations of antigen-specific T cells develop. Upon LN exit, these T cell families redistribute over the reactive LNs (in which the possibility of reprogramming of migration patterns may exist (11), and both effector sites are subsequently seeded by T cells from all families, regardless of the site of original priming. As evidence for both pathways of effector T cell entry into LNs exists (12, 13), it is possible that both processes operate in parallel.

Similar articles

Cited by

References

    1. Williams, D.A., I.R. Lemischka, D.G. Nathan, and R.C. Mulligan. 1984. Introduction of new genetic material into pluripotent haematopoietic stem cells of the mouse. Nature. 310:476–480. - PubMed
    1. Dick, J.E., M.C. Magli, D. Huszar, R.A. Phillips, and A. Bernstein. 1985. Introduction of a selectable gene into primitive stem cells capable of long-term reconstitution of the hemopoietic system of W/Wv mice. Cell. 42:71–79. - PubMed
    1. Lemischka, I.R., D.H. Raulet, and R.C. Mulligan. 1986. Developmental potential and dynamic behavior of hematopoietic stem cells. Cell. 45:917–927. - PubMed
    1. Golden, J.A., S.C. Fields-Berry, and C.L. Cepko. 1995. Construction and characterization of a highly complex retroviral library for lineage analysis. Proc. Natl. Acad. Sci. USA. 92:5704–5708. - PMC - PubMed
    1. Hamann, A., D.P. Andrew, D. Jablonski-Westrich, B. Holzmann, and E.C. Butcher. 1994. Role of alpha 4-integrins in lymphocyte homing to mucosal tissues in vivo. J. Immunol. 152:3282–3293. - PubMed

Publication types

MeSH terms

Substances

LinkOut - more resources