T cell dynamics and response of the microbiota after gene therapy to treat X-linked severe combined immunodeficiency

Abstract

Background: Mutation of the IL2RG gene results in a form of severe combined immune deficiency (SCID-X1), which has been treated successfully with hematopoietic stem cell gene therapy. SCID-X1 gene therapy results in reconstitution of the previously lacking T cell compartment, allowing analysis of the roles of T cell immunity in humans by comparing before and after gene correction.

Methods: Here we interrogate T cell reconstitution using four forms of high throughput analysis. (1) Estimation of the numbers of transduced progenitor cells by monitoring unique positions of integration of the therapeutic gene transfer vector. (2) Estimation of T cell population structure by sequencing of the recombined T cell receptor (TCR) beta locus. (3) Metagenomic analysis of microbial populations in oropharyngeal, nasopharyngeal, and gut samples. (4) Metagenomic analysis of viral populations in gut samples.

Results: Comparison of progenitor and mature T cell populations allowed estimation of a minimum number of cell divisions needed to generate the observed populations. Analysis of microbial populations showed the effects of immune reconstitution, including normalization of gut microbiota and clearance of viral infections. Metagenomic analysis revealed enrichment of genes for antibiotic resistance in gene-corrected subjects relative to healthy controls, likely a result of higher healthcare exposure.

Conclusions: This multi-omic approach enables the characterization of multiple effects of SCID-X1 gene therapy, including T cell repertoire reconstitution, estimation of numbers of cell divisions between progenitors and daughter T cells, normalization of the microbiome, clearance of microbial pathogens, and modulations in antibiotic resistance gene levels. Together, these results quantify several aspects of the long-term efficacy of gene therapy for SCID-X1. This study includes data from ClinicalTrials.gov numbers NCT01410019, NCT01175239, and NCT01129544.

Conflict of interest statement

Ethics approval and consent to participate

Written informed consent for all subjects described in this study was obtained as previously described [19, 33] and under IRB 12-010072. For the previous studies, local institutional review boards at the participating sites (Paris, Boston, Cincinnati, and Los Angeles) approved enrollment and study protocols. All human subject research was performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures

Fig. 1

Sampling schedule for 14 SCID-X1 gene-corrected subjects studied for integration site distributions and TCRB CDR3 sequence composition. a Times of sample acquisition after cell infusion. b Population sizes of inferred progenitor cells deduced from marking with unique sites of vector integration. Integration site data is in Additional file 1: Table S2, and summaries of genes near sites of vector integration are in Additional file 2: Figure S1. The x-axis shows the time since cell infusion. The y-axis shows the population size reconstructed using Chao1 from the numbers of unique integration sites and replicate sampling. Samples are named for the site of gene correction (U indicates USA, F indicates France; the next digit indicates the trial 1 = SCIDn1, 2 = SCIDn2, and the next two digits indicate the patient number within that trial). Patients F107 and F110 suffered severe adverse events at months 68 and 33, respectively. Both recovered by the final timepoints listed

Fig. 2

Analysis of TCRB CDR3 sequences. a The unique numbers of rearranged genes detected are shown. The colors indicate in frame rearrangements (blue), frameshifts (tan), and stop codons (red). b Richness and evenness of the inferred TCRB CDR3 populations. Patients are color coded as indicated on the right, and all replicates for each patient timepoint displayed. The ranges of healthy adults and children are shown by the light and dark gray diamonds, respectively. c Clustering of the samples sequenced using Bray-Curtis similarity and t-SNE. The association of patients with samples is shown by the key at the right. d V gene usage. The patient of origin is marked at the top of each panel. The V genes shown were the top most commonly found V genes in healthy children, ordered by prevalence; not all V genes used by each patient are shown. All panels show the same x-axis of V genes. Chi-squared test used to assess differences between gene distribution in patients compared to healthy children; red label indicates significant difference from healthy (p < 0.05). e J gene usage, with J genes determined by the same manner as in D. f Heat map summarizing the frequencies of utilization of the most common V and J pairs. Subjects studied are marked at the top. Time of sampling is shown on the right

Fig. 3

Minimum numbers of cell divisions between progenitors and daughter T cells. The x-axis shows time after corrected cell infusion. The y-axis shows the estimated number of cell divisions calculated as described in the text. The subjects studied are indicated beneath the figure as indicated by the color code. Circles indicate subjects from the second trial, and squares indicate subjects from the first. Stars next to the points indicate PBMCs were sequenced rather than sorted CDR3+ cells. Replicates were not available for U205b and so this subject was not analyzed

Fig. 4

Longitudinal analysis of the microbiome during SCID-X1 gene correction. a Timing of sample acquisition. b Longitudinal analysis of the nasopharyngeal microbiome. Each column indicates a sample. Samples are grouped by subject as indicated at the top. Each row summarizes the proportions of a specific microbial taxa inferred using Kraken. Abundance is color coded as indicated to the right and reflects the number of reads assigned to that taxa as a proportion of all non-human reads. c As in B, but oropharyngeal samples. d As in B, but stool samples

Fig. 5

Microbiome community analysis. a Sample clustering using Bray-Curtis dissimilarity. Different sample types are marked by the colors, healthy versus SCID are shown by the shapes. b Comparison of stool samples for each patient queried to healthy controls. Samples were clustered using Bray-Curtis dissimilarity. Each panel compares one SCID subject samples (indicated at the top) to healthy control samples (shown in gray). Elapsed time is shown using the color code (bottom). c The longitudinal species richness of each patient’s oral, nasal, and stool samples is shown in separate panels. The 95% CI of the healthy child richness for that sample type is shown in tan. d Representation of selected antibiotic resistance genes in the stool samples studied. Each column indicates a metagenomic data set from the subject listed at the top. Each row summarizes the abundance of an antibiotic resistance gene class. The tiles are colored by reads per kilobase of target per million sequence reads (RPKM); the color code is to the right of the panel

Fig. 6

Virome analysis. a Heat map summarizing RNA viruses detected. Each column indicates a sample from the patient indicated at the top of the heatmap, each row indicates a type of virus. The tiles are colored according to abundance. b Heat map summarizing DNA viruses detected. Markings as in A