We studied the mtDNA bottleneck in zebrafish to elucidate size, timing, and variation in germline and non-germline cells. Mature zebrafish oocytes contain, on average, 19.0 × 106 mtDNA molecules with high variation between oocytes. During embryogenesis, the mtDNA copy number decreases to ∼170 mtDNA molecules per primordial germ cell (PGC), a number similar to that in mammals, and to ∼50 per non-PGC. These occur at the same developmental stage, implying considerable variation in mtDNA copy number in (non-)PGCs of the same female, dictated by variation in the mature oocyte. The presence of oocytes with low mtDNA numbers, if similar in humans, could explain how (de novo) mutations can reach high mutation loads within a single generation. High mtDNA copy numbers in mature oocytes are established by mtDNA replication during oocyte development. Bottleneck differences between germline and non-germline cells, due to early differentiation of PGCs, may account for different distribution patterns of familial mutations.

Introduction

Various mechanisms have evolved to manage the high mtDNA mutation rate (Lynch et al., 2006). In animals, a high mtDNA copy number in cells dilutes the effect of mtDNA mutations (Otten and Smeets, 2015). Heteroplasmic mutations can only manifest above a tissue- and mutation-specific threshold. Another mechanism is the mtDNA bottleneck during maternal inheritance: a limited amount of mtDNA of the oocytes repopulates the cells of the next generation, thereby filtering out low-level mtDNA mutations. As a result, individuals are usually homoplasmic (Lee et al., 2012), which is the healthiest situation (Sharpley et al., 2012). However, in the case of familial pathogenic mutations, the bottleneck can cause high and unpredictable shifts in the mtDNA mutation load and disease risks in the offspring (Howell et al., 1992).

Despite extensive research, the mtDNA bottleneck is still not fully understood. Data from Holstein cows (Hauswirth and Laipis, 1982) indicated that the bottleneck was caused by a sharp decrease in mtDNA copy number during early development, most likely, followed by a clonal expansion of these founder molecules during oogenesis. In mice, a similar decrease in the mtDNA copy number between oocytes (range of 105; Cree et al., 2008 and Wai et al., 2008) and primordial germ cells (PGCs) was reported, with ∼200 mtDNA molecules in a single PGC at the bottom of the bottleneck, in line with previous estimates (185; Jenuth et al., 1996). In contrast, in another study, ∼2,000 mtDNA molecules were measured in a single PGC (Cao et al., 2007), and a small effective number of segregational units was proposed to explain the rapid segregation, either by assembly of mtDNA molecules into nucleoids or due to preferential replication of a subpopulation of the mtDNA genome (Cao et al., 2007 and Wai et al., 2008). For salmons, it was reported that the bottleneck occurred during oogenesis, with a size of about 85 mtDNA copies (Wolff et al., 2011). In humans, indirect estimations of the bottleneck size have been described, ranging from only 1–5 (Marchington et al., 1997) and 30–35 (Rebolledo-Jaramillo et al., 2014) to ∼90 (Pallotti et al., 2014) and 180 (Howell et al., 1992) copies.

We measured mtDNA copy number and variation in zebrafish oocytes and in larval germline and non-germline cells during embryonic development. The zebrafish model allowed relatively easy collection of mature oocytes from individual female fish, allowing assessment of the individual variation. Furthermore, PGCs were specifically visualized with GFP, followed by fluorescence-activated cell sorting (FACS) to isolate both PGCs and non-PGCs during embryogenesis (Goto-Kazeto et al., 2010). Lastly, oocytes from different stages of development were isolated. In this way, we characterized the mtDNA bottleneck in zebrafish in germline and non-germline cells, providing a better understanding of the underlying mechanism and possible consequences.