Author response: Extensive age-dependent loss of antibody diversity in naturally short-lived turquoise killifish
Year: 2022
Type: peer-review
Abstract: Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Appendix 1 Data availability References Decision letter Author response Article and author information Metrics Abstract Aging individuals exhibit a pervasive decline in adaptive immune function, with important implications for health and lifespan. Previous studies have found a pervasive loss of immune-repertoire diversity in human peripheral blood during aging; however, little is known about repertoire aging in other immune compartments, or in species other than humans. Here, we perform the first study of immune-repertoire aging in an emerging model of vertebrate aging, the African turquoise killifish (Nothobranchius furzeri). Despite their extremely short lifespans, these killifish exhibit complex and individualized heavy-chain repertoires, with a generative process capable of producing millions of distinct productive sequences. Whole-body killifish repertoires decline rapidly in within-individual diversity with age, while between-individual variability increases. Large, expanded B-cell clones exhibit far greater diversity loss with age than small clones, suggesting important differences in how age affects different B-cell populations. The immune repertoires of isolated intestinal samples exhibit especially dramatic age-related diversity loss, related to an elevated prevalence of expanded clones. Lower intestinal repertoire diversity was also associated with transcriptomic signatures of reduced B-cell activity, supporting a functional role for diversity changes in killifish immunosenescence. Our results highlight important differences in systemic vs. organ-specific aging dynamics in the adaptive immune system. Editor's evaluation This study introduces the killifish as a short-lived vertebrate model for immune aging and immunosenescence and characterizes the changes in the immune-repertoire during aging. This work provides an important first step in understanding how aging impacts the immune system in this model organism and will set the stage for many future studies. https://doi.org/10.7554/eLife.65117.sa0 Decision letter Reviews on Sciety eLife's review process Introduction The adaptive immune system undergoes a severe and systemic decline in proper function with age, resulting in higher susceptibility to a wide range of infections and decreased efficacy of vaccination in elderly individuals (Ademokun et al., 2010; Kogut et al., 2012; Dunn-Walters and Ademokun, 2010). In the humoral immune system, aging is accompanied by a decline in naïve B-cell output from the primary lymphoid organs; impaired production of specific antibodies in response to antigenic challenge; and a decline in antibody quality (Ademokun et al., 2010; Kogut et al., 2012; Sasaki et al., 2011; Aberle et al., 2013), as well as impairments in the establishment of novel immune memory (Aberle et al., 2013). These changes are major contributors to a generalized immunosenescent phenotype that significantly impairs health and quality of life in the elderly. The efficacy of the humoral immune system rests on its ability to generate an enormous array of different antibody sequences, with a correspondingly vast range of antigen specificities, and to progressively adjust the composition of this antibody population in response to antigen exposure (Schatz and Swanson, 2011; Di Noia and Neuberger, 2007; Magor, 2015; Elhanati et al., 2015). Sampling the resulting repertoire of antibody sequences in an individual using high-throughput sequencing can yield important insights into the diversity, clonal composition, and history of antibody-mediated immunity in that organism, as well as the effect of age, antigen exposure, and other factors on the diversity and functionality of the adaptive immune system (Weinstein et al., 2009; de Bourcy et al., 2017; Miho et al., 2018). In humans, antibody-repertoire sequencing has uncovered a number of important age-related changes, including reduced numbers of clones and unique sequences, increased baseline mutation, more frequent and larger clonal expansions, impaired B-cell selection, and a shift toward the memory compartment (de Bourcy et al., 2017; Jiang et al., 2013; Wang et al., 2014). The responsiveness of the peripheral repertoire to vaccination is also impaired during aging (de Bourcy et al., 2017; Wang et al., 2014). While within-individual repertoire diversity declines with age, between-individual variability increases, with repertoires from older individuals differing more from one another than those from young individuals (de Bourcy et al., 2017). Previous work in humans, however, has been limited by small sample sizes, a lack of temporal resolution, or a restriction to peripheral blood samples, which are known to systematically underrepresent the majority of B-cells resident in other organs and tissues (Siegrist and Aspinall, 2009; Tabibian-Keissar et al., 2016). Very little is known about how repertoire aging differs between distinct organs; in particular, almost nothing is known about how aging affects antibody repertoires at mucosal surfaces, which represent a crucial interface between the body and its microbial environment (Belkaid and Hand, 2014; Magadan et al., 2019). Even less is known about how aging might affect the antibody repertoires of vertebrates other than mice and humans. In this study, we introduce the naturally short-lived turquoise killifish (Nothobranchius furzeri) (Bradshaw and Valenzano, 2020; Hu and Brunet, 2018; Poeschla and Valenzano, 2020; Reichwald et al., 2015) as a model for adaptive immunosenescence. Here, we perform the first immune-repertoire sequencing experiments in this species, demonstrating that adult killifish express diverse and individualized heavy-chain repertoires that undergo rapid loss of diversity with age. The age-dependent loss of the antibody-repertoire diversity primarily affects the composition of expanded clones, with small naïve clones exhibiting much smaller age-related changes. By sequencing the repertoires of isolated intestinal samples, we further find that the killifish intestinal antibody repertoire exhibits much more dramatic age-dependent diversity loss than the body as a whole, possibly due to a much higher prevalence of expanded clones in the intestine, and that this loss of diversity is associated with gene expression changes indicating reduced B-cell activity. Taken together, our results reveal substantial differences between whole-body and organ-specific immune-repertoire aging, and establish the turquoise killifish as a powerful model for studying adaptive immune senescence. Results Establishing immunoglobulin sequencing in the turquoise killifish To investigate the effect of age on the B-cell receptor repertoire diversity and composition in turquoise killifish, we implemented an RNA-based repertoire-sequencing protocol based on the published protocol of Turchaninova et al., 2016, using template switching (Zajac et al., 2013) to add unique molecular identifiers (UMIs) to each RNA transcript of the immunoglobulin heavy chain (Figure 1; Bradshaw and Valenzano, 2020) to correct for errors and biases in abundance arising during PCR and Illumina sequencing (Vollmers et al., 2013). To test the validity and replicability of results obtained using this protocol, we performed three replicate library preps on whole-body total RNA samples from four adult (8-week-old) adult male turquoise killifish from the short-lived GRZ strain (Figure 1—figure supplement 1). Independent repertoires from the same individual showed a high degree of similarity in their clonal composition, with an average inter-replicate correlation in clone size of r = 0.89 (Figure 1—figure supplement 2). Inter-repertoire divergences computed with the published repertoire dissimilarity index (RDI) metric (Bolen et al., 2017) consistently identified replicates from the same individual as much more similar than repertoires from different individuals (Figure 1—figure supplement 3), demonstrating that this protocol is capable of accurately and reproducibly reconstructing the expressed heavy-chain repertoires of individual killifish. Figure 1 with 3 supplements see all Download asset Open asset Immunoglobulin sequencing from turquoise-killifish total RNA samples. Each sample undergoes reverse transcription with template switching to attach a 5' adaptor sequence and unique molecular identifier (UMI), followed by multiple rounds of PCR amplification and addition of Illumina sequencing adaptors. Libraries are then pooled, undergo size selection, and are sequenced on an Illumina MiSeq sequencing machine. Aging in whole-body killifish repertoires To investigate the effect of age on the structure and diversity of killifish antibody repertoires, we performed whole-body immunoglobulin sequencing on 32 adult male turquoise killifish from the short-lived GRZ strain (Hu and Brunet, 2018) at four different ages from early adulthood to late life (Figure 2A and Supplementary file 2a-d). The repertoire of each individual comprised some number of unique heavy-chain sequences, each of which could be classified by clonal identity (the ‘clonal repertoire’) and V/J usage (the ‘VJ repertoire’). Figure 2 with 4 supplements see all Download asset Open asset Aging in whole-body killifish IGH repertoires. (A) Experimental design. Adult male GRZ-strain turquoise killifish were sacrificed at 39, 56, 73, and 128 days post-hatching, flash-frozen and homogenized. (B–E) Alpha-diversity spectra, indicating average within-individual repertoire diversity for each age group and diversity order (*: 0.05 ≤ 0.01, **: 0.01≤ p ≤ 0.001, Kruskal-Wallis permutation test, Appendix 1—note 7). (B) Clonal alpha-diversity spectra. (C) VJ alpha-diversity spectra, all clones. (D) VJ alpha-diversity spectra, large clones (>4 unique sequences) only. (E) VJ alpha-diversity spectra, small clones (<5 unique sequences) only. (F) VJ alpha-diversity ratios for old vs. young killifish at each diversity order, for small (dashed lines) or large (solid lines) clones. Color indicates the older age group being compared to young (39 days) fish. (G) Normalized VJ beta-diversity spectra, indicating between-individual variability in repertoire composition for each age group and diversity order. (H) Distributions of pairwise repertoire dissimilarity index (RDI) distances between individuals in each age group (***: p ≤ 0.001, Mann-Whitney U tests for pairwise age differences), based on the VJ composition of each individual’s repertoire. (I) Principal coordinate analysis (PCoA) of pairwise RDI distances for each age group, visualizing the progressively greater dispersion seen at later ages. Each curve in (B–G) represents the mean across 2000 bootstrap replicates (Appendix 1—note 7); shaded regions indicate 95% confidence intervals over the same. The diversity of a population is a measure of the number (a.k.a. the richness) and relative frequency of different subdivisions within that population. For B-cell repertoires, diversity can be calculated over the different clonal lineages detected in a sample, or over different variable-region gene segment combinations. To quantify immune-repertoire diversity in killifish heavy-chain repertoires, we computed Hill diversity spectra (Appendix 1—note 1), which provide a holistic overview of the diversity structure of a population (in this case, a repertoire) (Miho et al., 2018; Hill, 1973; Jost, 2006; Jost, 2007). Briefly, each Hill spectrum reports the ‘effective richness’ of a repertoire across a range of diversity orders; higher effective richness corresponds to a more diverse repertoire. At low diversity orders, species of different sizes are weighted more equally in the diversity calculation, while at higher orders, less-abundant species are progressively downweighted relative to more-abundant groups (Appendix 1—note 1). For the clonal repertoire, each clonotype (set of unique sequences descended from a single naïve ancestor B-cell) was designated a separate species; for the VJ repertoire, clonotypes with the same V/J identity were grouped together. Separate spectra were computed for the clonal and VJ repertoires of each individual, and these individual spectra were used to compute averaged alpha-diversity spectra for each age group and repertoire type. For each diversity order, we tested for a significant age effect on repertoire diversity using a permutation test on the Kruskal-Wallis H statistic (Materials and methods). The clonal diversity (Figure 2—figure supplement 1) of the whole-body killifish repertoire exhibited a significant decline with age (p < 0.05) at high diversity orders (Figure 2B, Figure 2—figure supplements 2–3), indicating a significant and extremely rapid age-related decline in the diversity of the largest B-cell clones. In contrast, lower-order clonal diversity exhibited no significant change with age, suggesting that the overall composition of the whole-body repertoire remains relatively unchanged. Since the B-cell clonal repertoire is overwhelmingly dominated by small, predominantly naïve clones (Figure 2—figure supplement 4A), low-order clonal diversity measurements are primarily driven by changes in the diversity of small clones. As such, these results indicate that the composition of small clones in the killifish antibody repertoire is much less sensitive to the effects of aging than that of large, expanded clones. In contrast with the rapid age-related declines observed in high-order clonal diversity, the VJ diversity of the killifish repertoire exhibited no significant age-related change at any diversity order (Figure 2C). Examining the clone-size distribution of each V/J combination (Figure 2—figure supplement 4B) revealed that even the largest V/J combinations in each age group are overwhelmingly dominated by small clones, suggesting that the observed lack of an age effect on VJ diversity was due to the observed age-insensitivity of small clones (Figure 2B). To test this hypothesis, we filtered the repertoire dataset to separate sequences from small and large clones and computed VJ diversity repertoires for each subset (Figure 2D–E). While both small and large clones in isolation showed a significant age effect on VJ diversity, the relative reduction in VJ alpha diversity with age was dramatically stronger for large clones, an effect observed across all diversity orders (Figure 2F). As suggested by the clonal-diversity results, therefore, the repertoire diversity of large (expanded) clones in the killifish whole-body repertoire appears to be far more age-sensitive than that of small (predominantly naïve) clones. In addition to the average within-individual diversity of a population (alpha diversity), the between-individual variation in composition (beta diversity) can provide important insights into repertoire development and evolution. Previous studies of human peripheral blood repertoires have suggested a decrease in alpha diversity but an increase in beta diversity with age (Gibson et al., 2009; de Bourcy et al., 2017). In our dataset, VJ beta-diversity spectra (Appendix 1) indicated a large age-related increase in beta diversity across a wide range of diversity orders (Figure 2G), indicating a similar pattern of progressive individualization in repertoire composition with age. Concordantly, older killifish also exhibited significantly greater pairwise RDI distances (Bolen et al., 2017), indicating progressive divergence in repertoire composition (Figure 2H–I). As in humans, therefore, younger killifish exhibit antibody repertoires that are significantly more similar to one another, which then become increasingly distinct and individualized as the cohort increases in age. The killifish generative repertoire The naïve sequence diversity of the antibody heavy-chain repertoire depends on the molecular processes underlying the generation of novel sequences in developing B-cells: random selection of V, D, and J segments during VDJ recombination; deletions and palindromic (P-) insertions at the ends of conjoined segments; and nonpalindromic (N-) insertions between segments (Schatz and Swanson, 2011; Schroeder and Cavacini, 2010). Each of these contributes diversity to the overall generative process, increasing the variety of sequences that can be generated. Excluding nonfunctional sequences, the human generative process has an estimated Shannon entropy of roughly 70 bits, corresponding to a first-order Hill diversity of roughly 1021 possible unique sequences (Elhanati et al., 2015). However, little is known about how this generative diversity varies across species, or how it changes during aging. To gain insight into these generative processes in the turquoise killifish, we used IGoR (Marcou et al., 2018) to infer models of sequence generation from killifish repertoire data. In training these models, we restricted the dataset to nonfunctional naïve sequences, in order to avoid distortions introduced by positive and negative selection in the primary lymphoid organs (Elhanati et al., 2015; Marcou et al., 2018). As is often the case with RNA data, the number of naïve nonfunctional sequences available per individual was frequently low; hence, to better capture low-probability events in the generative process, we inferred models from pooled data from multiple individuals in the same age group. As the parameters of the generative model are typically very similar across conspecific individuals (Figure 3—figure supplements 1–2), especially in an inbred line, pooling data like this is a useful way to infer more robust generative models using IGoR (Marcou 2019, personal communication). To model the generative process in its baseline state, we first inferred a model of the killifish generative repertoire from the four 8-week-old adult male individuals used in the pilot study (Figure 1—figure supplement 1). Using this model, we estimated a total raw entropy for the killifish generative repertoire of roughly 33 bits (Figure 3A). Of these 33 bits, roughly 8 arise from variability in VDJ segment choice, 12 from variability in the number and composition of junctional N-insertions, and 11 from P-insertions and deletions. Accounting for convergent production of identical sequences via different recombination events, and for events that give rise to nonfunctional nucleotide sequences (e.g. due to frame shift) reduced this initial raw estimate by 10 bits. Figure 3 with 5 supplements see all Download asset Open asset The killifish generative repertoire. (A) Entropy composition of the generative process from four 8-week-old GRZ-strain adult male turquoise killifish. (B) Probability distributions of junctional N-insertions in the same dataset. (C) P-insertions and deletion distributions inferred from the same dataset, with P-insertions modeled as negative deletions. (D) Boxplots of total recombination entropy values for models inferred separately for each individual in the 32-individual aging cohort (p = 0.43, Kruskal-Wallis one-way analysis of variance [ANOVA] for an age effect). Before initial selection in the primary lymphoid organs, therefore, the killifish generative process has an estimated Shannon entropy of roughly 23 bits (Figure 3A), corresponding to a first-order Hill diversity of roughly 107 possible unique sequences. While, as in humans, only a small fraction of potential diversity will actually be generated in any single individual, this nevertheless represents a highly complex and sophisticated system capable of generating highly individualized IGH repertoires. While impressive, the potential generative diversity of the killifish repertoire is nevertheless vastly lower than in humans, with a difference in productive generative entropy of almost 50 bits (Elhanati et al., 2015). While all components of the generative process exhibit lower entropies in killifish than in humans, by far the greatest difference lies in the junctional N-insertions, which contribute almost 40 bits more to the generative entropy of the human repertoire than that of killifish. The difference in the productive generative entropy between killifish and human arises from the distributions of N-insertions inferred from killifish and human data: in humans, these distributions peak at around 5 nt per junction and often yield insertions of 10–20 nt (Elhanati et al., 2015), while in killifish the insertion distribution peaks at 0 nt per junction, and sequences with more than 5 nt of insertions at either junction are very rare (Figure 3B and Figure 3—figure supplement 1). Since N-insertions are the dominant source of sequence diversity in human repertoires, the large reduction in N-insertions in killifish relative to humans unsurprisingly results in a much lower overall generative diversity for the killifish adaptive immune system. The relative lack of change in the small-clone antibody repertoire in older turquoise killifish (Figure 2B–F) suggested to us that the diversity of the generative process in the primary lymphoid organs might remain relatively intact throughout the killifish lifespan. To test this hypothesis, we trained separate IGoR models for each individual in the 32-fish aging cohort (Figure 2A, Figure 3—figure supplements 3–4) and tested for an effect of age on the generative diversity inferred for each individual. As expected, no age effect was found in either total generative diversity (Figure 3D) or the contributions of different diversification processes (Figure 3—figure supplement 5). It therefore appears that, while some aspects of the killifish antibody repertoire certainly decline with age, the entropy of the generative process is not among them. Effect of age and microbiota transfer on killifish intestinal repertoires The populations of B-lymphocytes associated with mucosal epithelia play a crucial role in defending the body from pathogenic threats (Magadan et al., as well as in the composition of resident microbial populations (Belkaid and Hand, 2014). Despite the of these B-cell compartments, relatively little is known about the structure of their antibody repertoires (Magadan et al., and less about how these repertoires change with age. As the of the greatest microbiota diversity, the is of as an important and immune Previous work on the killifish microbiota et al., 2017) has that it declines in alpha diversity and increases in beta diversity with age, that the changes seen in the whole-body composition of the killifish antibody repertoire (Figure 2). of intestinal from young to has also been to et al., 2017). these and the between intestinal and (Belkaid and Hand, we the effect of aging and microbiota transfer on the immune repertoires of B-cell populations. Using intestinal total RNA isolated by et al., we sequenced the intestinal IGH repertoires of male GRZ-strain including four individuals and individuals from groups file and the effect of age and on repertoire diversity in the killifish to our the alpha beta diversity of the killifish intestinal repertoire were significantly by microbiota with no significant difference in clonal diversity, VJ diversity, or RDI (Figure supplements In contrast to the whole-body data, however, was a and significant decline in both clonal and VJ alpha diversity with age across all diversity orders (Figure Figure supplements 2–3), even by clone This age-related decline in alpha diversity was consistently far more dramatic than that observed in the whole-body samples at any diversity order. The B-cells of the killifish intestine, therefore, exhibit a much stronger age-dependent decline in repertoire diversity than is seen in the killifish body Figure 4 with 4 supplements see all Download asset Open asset Aging in killifish intestinal repertoires. (A) Experimental design. Adult male GRZ-strain turquoise killifish were sacrificed at and post-hatching, and total RNA was from the Alpha-diversity spectra, indicating average within-individual repertoire diversity for each age group and diversity order (*: 0.05 ≤ 0.01, **: 0.01≤ p ≤ 0.001, Kruskal-Wallis permutation test, Appendix 1—note 7). (A) Clonal alpha-diversity spectra. (B) VJ alpha-diversity spectra, all clones. (D) Normalized VJ beta-diversity spectra, indicating between-individual variability in repertoire composition for each age group and diversity order. (E) of pairwise repertoire dissimilarity index (RDI) distances between killifish intestinal repertoires at different ages (***: p ≤ 0.001, Mann-Whitney U tests for pairwise age Each curve in represents the mean across 2000 bootstrap replicates (Appendix 1—note 7); shaded regions indicate 95% confidence intervals over the same. While results from beta-diversity spectra showed large increases in beta diversity with age at some diversity orders but not at (Figure the pairwise RDI individual repertoires increased and significantly with age (Figure and Figure supplement suggesting that, as in the killifish intestinal repertoires become increasingly distinct and individualized as age. potential for the stronger age-related in alpha diversity of intestinal samples is as a of the antigen exposure by intestinal as a of their with the This exposure could high of clonal resulting in a greater loss in repertoire diversity et al., not with the is that the has different clone-size distribution relative to the the whole-body repertoire, the not the primary lymphoid organs, and be to be far less dominated by small, naïve clones. Since the population of large clones appears to be more to in diversity with age than that of small clones (Figure the stronger overall age-related diversity loss found in the repertoire could be a of this greater relative prevalence of large clones. analysis of clonal in whole-body and intestinal repertoires showed that the far small clones, resulting in a much higher of large clones (Figure this difference in clonal composition, than some functional difference between intestinal and other is primarily for the difference in aging between whole-body and intestinal repertoires, we to find a of clonal diversity loss during aging in intestinal repertoires at low diversity orders are dominated by small clones in whole-body but not at high orders are dominated by large clones in both sample we to find loss in intestinal samples of V/J diversity over all clones, but not the V/J diversity is restricted to large clones Figure 5 with 1 supplement see all Download asset Open asset clonal composition and aging of whole-body and intestinal repertoires. (A) analysis of clonal composition of antibody repertoires from whole-body and intestinal samples, the average number of small unique sequences) and large 5 unique sequences) clones for each individual across replicates at each sample as well as the average of all clones in each repertoire which are large regions around each the within one of the mean (B) Boxplots of individual diversity measurements of repertoires from each age group in the whole-body and intestinal by the mean diversity of the age group in each dataset. the of a generalized model of diversity vs. age and sample for the whole-body and intestinal dataset, relative to the average diversity of the age group in each for a significant effect of sample on the of diversity change with age p p To test these we the diversity measurements from each dataset by the mean diversity of the group in that dataset, then generalized models for different diversity orders and methods of diversity (Figure and Figure supplement 1), for a significant between sample vs. and the effect of age on repertoire samples exhibited significantly higher of age-dependent diversity loss low-order clonal-diversity or total that those for which clones of all sizes were in the diversity was no significant difference in
Source: eLife
Language: English
Topic: T-cell and B-cell Immunology
Subfield: Immunology
Field: Immunology and Microbiology
Domain: Life Sciences
Open Access status: gold
Estimated APC paid: $2,500
OpenAlex ID: W4214669987
