Expanded skin virome in DOCK8-deficient patients

Abstract

Human microbiome studies have revealed the intricate interplay of host immunity and bacterial communities to achieve homeostatic balance. Healthy skin microbial communities are dominated by bacteria with low viral representation^1-3, mainly bacteriophage. Specific eukaryotic viruses have been implicated in both common and rare skin diseases, but cataloging skin viral communities has been limited. Alterations in host immunity provide an opportunity to expand our understanding of microbial-host interactions. Primary immunodeficient patients manifest with various viral, bacterial, fungal, and parasitic infections, including skin infections⁴. Dedicator of cytokinesis 8 (DOCK8) deficiency is a rare primary human immunodeficiency characterized by recurrent cutaneous and systemic infections, as well as atopy and cancer susceptibility⁵. DOCK8, encoding a guanine nucleotide exchange factor highly expressed in lymphocytes, regulates actin cytoskeleton, which is critical for migration through collagen-dense tissues such as skin⁶. Analyzing deep metagenomic sequencing data from DOCK8-deficient skin samples demonstrated a notable increase in eukaryotic viral representation and diversity compared with healthy volunteers. De novo assembly approaches identified hundreds of novel human papillomavirus genomes, illuminating microbial dark matter. Expansion of the skin virome in DOCK8-deficient patients underscores the importance of immune surveillance in controlling eukaryotic viral colonization and infection.

Fig. 1. Skin microbiome of DOCK8-deficient patients is dominated by eukaryotic viruses.

a, Clinical spectrum of dermatologic features of DOCK8-deficient patients: herpes simplex viral infection on the lips of D8_Pt16 (top left); warts on the hand of D8_Pt17 (bottom left); severe eczema behind the knee of D8_Pt12 (middle); and large molluscum contagiosum lesions on the knee of D8_Pt18 (right). b, Skin microbiome of DOCK8-deficient patients (n = 27) and healthy volunteers (HV) (n = 5) classified at the kingdom level. Shotgun metagenomics data presented as mean relative abundance of total mapped microbial reads (normalized to genome length) from all sampled body sites per patient. c, Mean relative abundance of viral families identified on skin of DOCK8-deficient patients compared with HVs. d, Number of DOCK8-deficient patients harboring specific eukaryotic DNA virus families.

Fig. 2. HPV diversity on skin of DOCK8-deficient patients.

a, A total of 405 HPV types on the skin of all DOCK8-deficient patients (n = 27), categorized as reference (155) or novel (250) types classified by species and genus. Each genus is depicted as a different color; pie slices represent different species within a genus and lighter shades represent new HPV types within the species. Numbers around the pie refer to the HPV type count. b, Skin microbiome of individual DOCK8-deficient patients: the mean relative abundances of HPV types incorporating novel HPV genomes. Relative abundance scale is on the axis of D8_Pt26. Color code as in a. c, Boxplots showing that mapping skin shotgun metagenomics reads to reference database, which includes novel HPV genomes, significantly increases the number of HPV types detected in children (n = 10) and adults (n = 17) throughout all sampled skin sites (n = 188 biologically independent samples, two-sided Wilcoxon rank-sum test). d, Number of HPV types classified at the genus level detected in each patient. e, Boxplots showing mean theta similarity index of inter- and intrapersonal pairwise similarity of skin sites of DOCK8-deficient patients (n = 188 biologically independent samples from all skin sites, two-sided Wilcoxon rank-sum test), comparing HPV species detected at relative abundance > 0.1%. Theta value of 1 indicates identical HPV community structure. All boxplots are the median with the interquartile range, and error bars are the 1.5 times interquartile range (whiskers).

Fig. 3. Longitudinal stability of the skin virome of DOCK8-deficient patients.

a, Mean relative abundances of two skin sites for the three most abundant viral families (Papillomaviridae, Polyomaviridae, Poxviridae) shown at the species level in patients (n = 7) sampled longitudinally. Baseline and follow-up time points (first and second visit) are represented by 1 and 2, respectively. Patients ordered by months between samplings (from shortest to longest). b, Absence (visit 1, baseline) and presence (visit 2, follow-up, 60 months) of molluscum contagiosum on the face of D8_Pt4. c, Boxplots showing mean theta similarity index of inter- and intrapersonal pairwise similarity of patients at baseline and follow-up time points (n = 28 biologically independent samples, two-sided Wilcoxon rank-sum test), comparing HPV species detected in relative abundances >0.1%. All boxplots are the median with the interquartile range, and the error bars are the 1.5 times interquartile range (whiskers) and outliers (points). d, Longitudinal changes include gain and loss of HPV types (for types with relative abundance > 0.1%) and are patient specific. Duration of time between sampling time points (measured in months) is shown in parentheses next to each patient. Numbers on bars represent count of HPV types at first baseline sampling.

Fig. 4. Human RNA viruses in DOCK8-deficient patients and comparison between viral presence in sequencing data and cutaneous lesions.

a, Relative abundance heatmap of human RNA viral species detected in the nares of DOCK8-deficient patients. b, Boxplots comparing mean percentage of HPV reads and patients that presented with any warts on their skin (n = 27 patients, two-sided Wilcoxon rank-sum test). c, Boxplots comparing mean percentage of MCV reads and patients that presented with any molluscum contagiousum lesions on their skin (n = 27 patients, two-sided Wilcoxon rank-sum test). All boxplots are the median with the interquartile range, and error bars are the 1.5 times interquartile range (whiskers).