Organelle size scaling over embryonic development

Abstract

Cell division without growth results in progressive cell size reductions during early embryonic development. How do the sizes of intracellular structures and organelles scale with cell size and what are the functional implications of such scaling relationships? Model organisms, in particular Caenorhabditis elegans worms, Drosophila melanogaster flies, Xenopus laevis frogs, and Mus musculus mice, have provided insights into developmental size scaling of the nucleus, mitotic spindle, and chromosomes. Nuclear size is regulated by nucleocytoplasmic transport, nuclear envelope proteins, and the cytoskeleton. Regulators of microtubule dynamics and chromatin compaction modulate spindle and mitotic chromosome size scaling, respectively. Developmental scaling relationships for membrane-bound organelles, like the endoplasmic reticulum, Golgi, mitochondria, and lysosomes, have been less studied, although new imaging approaches promise to rectify this deficiency. While models that invoke limiting components and dynamic regulation of assembly and disassembly can account for some size scaling relationships in early embryos, it will be exciting to investigate the contribution of newer concepts in cell biology such as phase separation and interorganellar contacts. With a growing understanding of the underlying mechanisms of organelle size scaling, future studies promise to uncover the significance of proper scaling for cell function and embryonic development, as well as how aberrant scaling contributes to disease.

Graphical/Visual Abstract and Caption

How do the sizes of intracellular organelles adapt to reductions in cell size that occur during early embryogenesis? We address mechanisms of developmental organelle size scaling in a variety of model organisms and, if known, the functional significance of scaling.

Introduction

Early development is generally characterized by rapid cell divisions with little to no cell growth. As cell number increases, cell size decreases during early embryogenesis in a wide variety of organisms, including model species such as zebrafish (Danio rerio), frogs (Xenopus laevis), worms (Caenorhabditis elegans), and mice (Mus musculus) (Tadros & Lipshitz, 2009). A fundamental cell biological question concerns how these reductions in cell size influence the size and function of intracellular structures. From a physical perspective, it reasons that the sizes of certain organelles must become smaller as cell size decreases. For instance, the nucleus of an early stage X. laevis embryo would barely fit within the confines of a later stage blastomere (Jevtic & Levy, 2015) (Fig. 1A). Beyond physical considerations, scaling of organelle size with cell size might also be critical to ensure that organelle function is appropriately matched to cell size. In this review, we will present an overview of which organelles and intracellular structures are known to scale with cell size during early development in a variety of different model organisms, focusing on the nucleus, mitotic spindle, and chromosomes, and touching on cilia, endoplasmic reticulum (ER), Golgi, mitochondria, and lysosomes/vacuoles. Where it is known, we will discuss the mechanistic basis for these scaling relationships and the functional consequences of disrupted organellar scaling. The advantages of different model organisms have provided unique insights into organelle scaling, for example genetics in worms and flies, biochemistry in frogs, and imaging in zebrafish and sea urchins. Where appropriate, we will briefly discuss organelle scaling in single cell model organisms like yeast and Tetrahymena as well as in the processes of cell differentiation and aging, as mechanisms identified in these systems may be relevant to developmental organelle size scaling. We conclude with what we view as important outstanding questions and potential approaches to addressing these questions in the future.

Figure 1: Developmental size regulation of intracellular structures in Xenopus.

(A) Blastomeres were isolated from different stage X. laevis embryos and stained for the NPC. Scale bar, 50 μm. Adapted with permission from (Jevtic & Levy, 2015). (B) Nuclei were assembled in X. laevis egg extract in microfluidic channels of varying dimensions. Channel dimensions are indicated as height × width. Membranes, DNA, and incorporated dUTP are labeled green, blue, and red, respectively. Scale bar, 20 μm. Adapted with permission from (Hara & Merten, 2015). (C) Nuclei and cytoplasm from stage 10 X. laevis embryos were encapsulated in droplets of varying volume and allowed to reach steady-state sizes. Droplet volumes are indicated above each panel. Nuclei are visualized by import of GFP-NLS. Scale bar, 50 μm. Adapted with permission from (P. Chen et al., 2019). (D) Different stage X. laevis embryos were fixed and stained for tubulin (yellow) and DNA (red). Scale bar, 20 μm. Adapted with permission from (Wuhr et al., 2008). (E) Spindles were assembled in different volumes of X. laevis egg extract. Tubulin, DNA, and NuMA are labeled red, blue, and green, respectively. Droplet diameters are indicated above each panel. Scale bar, 25 μm. Adapted with permission from (Hazel et al., 2013). (F) Condensed mitotic chromosomes from different stage X. laevis embryos are shown. For comparison, unreplicated (Unrep) and replicated (Rep) sperm chromosomes were incubated in X. laevis egg extract. DNA and kinetochores are labeled blue and red, respectively. Scale bar, 5 μm. Adapted with permission from (Kieserman & Heald, 2011). (G) As cells become smaller during early Xenopus development, palmitoylated importin α is increasingly partitioned to the plasma membrane. In the case of the nucleus, this reduces nuclear import and size. In the case of the spindle, reduced kif2a inhibition promotes MT depolymerization and spindle shortening. Adapted with permission from (Brownlee & Heald, 2019).

As already mentioned, reductions in cell size over early development are driven by rapid cell divisions without cell growth, so cell cycle timing is a key determinant of cell size in the early embryo. There is also evidence that cell size and cytoplasmic volume determine cell division timing, with cell size reductions leading to increased cell cycle lengths (Arata & Takagi, 2019; Guan et al., 2018; Masui & Wang, 1998). Characteristic of many early developmental programs, asymmetric cell divisions give rise to different sized cells within the embryo (Cadart, Zlotek-Zlotkiewicz, Le Berre, Piel, & Matthews, 2014). Positioning of the interphase nucleus and mitotic spindle often determine the division site. Such positioning is dictated by both length-dependent microtubule (MT) forces and polarity domains at the cell cortex that alter such forces (Hasley, Chavez, Danilchik, Wuhr, & Pelegri, 2017; Minc & Piel, 2012; Pierre, Salle, Wuhr, & Minc, 2016; Xiao, Tong, Yang, & Wu, 2017). Other general regulators of cell size include nutrient availability (Leitao & Kellogg, 2017; Lucena et al., 2018; Vadia et al., 2017), signalling (Acebron, Karaulanov, Berger, Huang, & Niehrs, 2014; Luo, Liu, & Nassel, 2013), mechanical force (Perez Gonzalez et al., 2018), pressure (Pham et al., 2019), and expression and localization of cell cycle regulators (Keifenheim et al., 2017; Pan, Saunders, Flor-Parra, Howard, & Chang, 2014; Patterson, Rees, & Nurse, 2019; Schmoller & Skotheim, 2015; Zapata et al., 2014). Open questions remain about how these factors might regulate cell size in the early embryo. Maintenance of proper cell size is clearly important for intracellular function, for instance inappropriate cell enlargement can lead to cytoplasm dilution if nucleic acids and protein biosynthesis do not scale proportionately with cell size, potentially contributing to cellular senescence (Neurohr et al., 2019).

A variety of different, non-mutually exclusive models might account for the scaling of organelle size with cell size (Marshall, 2015, 2016). Limiting component models posit that the early embryo is preloaded with enough organelle building blocks to sustain a certain number of cell divisions (Fig. 2A). As components limiting for organelle size are partitioned into greater numbers of smaller cells, organelle size becomes smaller (Goehring & Hyman, 2012). In a related model, enzymes responsible for organelle assembly might also become limiting. An intriguing new area in cell biology is the notion of phase separation, in which specific proteins separate from the bulk cytoplasm and coalesce into membraneless organelles (Shin & Brangwynne, 2017). Reductions in cytoplasmic volume over development might affect the size and behavior of such phase-separated organelles (Brangwynne, 2013). Changes in cytoplasmic composition, for instance changes in ribosome concentrations, can influence the phase separation properties of membraneless organelles (Delarue et al., 2018). Ruler models implicate the size of constituent components or templates in determining final organelle size, and reductions in component or template size over development could account for organelle size scaling (Fig. 2B). Given that most subcellular structures are orders of magnitude larger than their protein constituents, the ruler model prompts the question of how component number is regulated to determine the final size of the structure. In more dynamic models, organelle growth and disassembly rates determine steady-state size (Fig. 2C). Developmentally-regulated reductions in growth and/or increases in disassembly could therefore give rise to smaller organelles. A related model is one where organelle functional output could feedback on organelle growth and/or disassembly. Given that many organelles are connected through contiguous membrane systems, it is easy to imagine how active mechanisms regulating the size of one organelle might passively influence the size of other interconnected organelles (Fig. 2D). Throughout this review, we will refer to these various general organelle size-scaling models in cases where sizing mechanisms are known.

Figure 2: Models for developmental organelle size scaling.

(A) Small red circles represent components that limit assembly and/or growth of the larger red organelle. Over development, limiting components are partitioned into greater numbers of smaller cells, such that the absolute amounts of limiting components per cell decrease leading to smaller organelle size. (B) Green rectangles assemble to form a linear structure whose length is dictated by the blue ruler. Developmental changes in the size of the ruler or building blocks might lead to reductions in the overall size of the structure. (C) A balance of assembly and disassembly sets steady-state organelle size. Developmental changes in assembly and disassembly rates would lead to organelle size scaling. (D) The orange and blue organelles are connected through membranes. Developmental reductions in the size of one organelle might lead to concomitant size scaling of the interconnected organelle. Another possible scenario is that total membrane amount remains constant but membrane distribution changes such that one organelle increases in size and the other becomes smaller.

Nucleus

Perhaps one of the most well documented examples of organelle size scaling during development is nuclear scaling. It has been known for over a century that nuclear size generally scales proportionately to cell size (Conklin, 1912; Wilson, 1925). While DNA amount correlates with nuclear size across species and likely sets a minimum nuclear size, experimentally manipulating DNA content in yeast and Xenopus did not lead to large changes in nuclear size, suggesting that DNA amount may not directly contribute to nuclear size determination (Jevtic & Levy, 2017; Levy & Heald, 2010; Neumann & Nurse, 2007). Consistent with these findings, nuclear size generally scales smaller during early development, while nuclear DNA content remains constant. In Xenopus, the first few cell divisions occur synchronously and are symmetric, after which asymmetric divisions divide the embryo into larger vegetal pole cells packed with yolk and smaller animal pole cells. The Drosophila early embryo is not cellularized, consisting instead of a syncytium of nuclei. After 14 nuclear divisions, nuclei migrate to the periphery of the embryo and form the cellular blastoderm, at which time damaged nuclei are eliminated (Iampietro et al., 2014). Cell divisions in C. elegans are neither synchronous nor symmetric during early developmental stages and short cell migrations reorganize the embryo. In zebrafish, embryonic development starts with largely symmetric, synchronous cell divisions that become progressively less synchronous, eventually resulting in smaller animal pole cells (Kimmel, Ballard, Kimmel, Ullmann, & Schilling, 1995). In spite of these variations in cell division patterns across these organisms, reductions in nuclear size during embryonic development are almost always observed. In Xenopus, reductions in nuclear size over development are accompanied by an increasing nuclear-to-cytoplasmic (N/C) volume ratio because cell size decreases more rapidly than nuclear size (Jevtic & Levy, 2015) (Fig. 1A). In zebrafish, nuclear size scales with developmental progression and thus cell size (Reisser et al., 2018). In C. elegans, nuclear and cell sizes scale from the 2-cell stage onward (Arata, Takagi, Sako, & Sawa, 2014; Hara, Iwabuchi, Ohsumi, & Kimura, 2013) (Fig. 3A), and nuclear size scales hypoallometrically with cell size in developing intestinal cells (Uppaluri, Weber, & Brangwynne, 2016). Though direct measurements have not been reported, Drosophila nuclear size decreases with each replication cycle during embryogenesis (Kotadia, Crest, Tram, Riggs, & Sullivan, 2010) (Fig. 4A), and during cellularization the farnesylated inner nuclear membrane protein Kugelkern induces nuclear lengthening (Brandt et al., 2006). Nuclear volume decreases over ten-fold in mouse embryos from the 1-cell stage to blastocyst with concomitant increases in the N/C ratio (Tsichlaki & FitzHarris, 2016) (Fig. 4B). While absolute N/C volume ratios differ in different species, the general trend is an increase in the N/C ratio during developmental progression (Fig. 5). Past early embryonic stages, N/C volume ratios tend to stabilize, and it has been proposed that maintenance of proper N/C ratios is important for cell function (Walters, Bommakanti, & Cohen-Fix, 2012).

Figure 3: Developmental size regulation of intracellular structures in C. elegans.

(A) Different stage C. elegans embryos with nuclei labeled green for H2B and plasma membrane labeled red. Scale bar, 10 μm. Images adapted from (Fickentscher & Weiss, 2017) and made available under a Creative Commons Attribution 4.0 International License. (B) Different stage C. elegans embryos expressing GFP-tagged β-tubulin. Scale bar, 20 μm. Adapted with permission from (Lacroix et al., 2018). (C) Different stage C. elegans embryos expressing GFP-tagged γ-tubulin and H2B. Scale bar, 10 μm. Adapted with permission from (Greenan et al., 2010). The cartoon model in the “C. elegans spindle size regulators box” was adapted with permission from (Lacroix et al., 2018).

Figure 4: Developmental size regulation of intracellular structures in other model organisms.

(A) Different stage D. melanogaster embryos showing labeled nuclei. Numbers denote the nuclear cycle. Note that nuclei are present in a syncytium until cellularization at nuclear cycle 14. Images adapted with permission from (Kotadia et al., 2010). (B) Different stage M. musculus embryos with DNA labeled green and cell cortex labeled red with phalloidin. Images adapted from (Tsichlaki & FitzHarris, 2016) and made available under a Creative Commons Attribution 4.0 International License. (C) Different stage P. lividus sea urchin embryos microinjected with ATTO 565-labelled tubulin. Scale bar, 20 μm. Adapted with permission from (Lacroix et al., 2018).

Figure 5: N/C volume ratios generally increase during early development in different species.

C. elegans worm data are from (Hara & Kimura, 2013). Fission yeast data are from (Neumann & Nurse, 2007). M. musculus data are from (Tsichlaki & FitzHarris, 2016). H. pulcherrimus sea urchin data are from (Masui & Kominami, 2001). A. pectinifera starfish data are from (Masui, Yoneda, & Kominami, 2001). D. rerio zebrafish data are from (Joseph et al., 2017). X. laevis frog data are from (Jevtic & Levy, 2015).

Many insights into the mechanisms of developmental nuclear size scaling have been gleaned from Xenopus experiments, in particular using Xenopus egg extracts that support de novo assembly of interphase nuclei and mitotic spindles (P. Chen & Levy, 2018; Field & Mitchison, 2018). One such mechanism involves the balance between nuclear import and export. Two nuclear transport factors, importin α and NTF2, contribute to differences in nuclear size between two Xenopus species (Levy & Heald, 2010; Vukovic, Jevtic, Zhang, Stohr, & Levy, 2016). In early stage embryos, while a fraction of palmitoylated importin α is associated with the plasma membrane, a large pool of importin α is cytoplasmic and available to drive nuclear growth through import. As cell size decreases over development and the blastomere surface area-to-volume ratio increases, an increasingly larger fraction of palmitoylated importin α is partitioned to the plasma membrane. With less available cytoplasmic importin α, import rates and nuclear growth both decrease, and increasing the levels of cytoplasmic importin α is sufficient to increase nuclear size (Brownlee & Heald, 2019; Levy & Heald, 2010; Wilbur & Heald, 2013) (Fig. 1G). While reductions in nuclear import contribute to nuclear size scaling in early Xenopus embryos, nuclear export appears to not play a major role (Edens & Levy, 2014), although in other systems reducing nuclear export can increase nuclear size (Jevtic, Schibler, et al., 2019; Kume et al., 2017; Neumann & Nurse, 2007). In Tetrahymena thermophila, macronucleus and micronucleus sizes are determined by nucleus-specific nuclear pore complex (NPC) components, importins, and histones (Iwamoto et al., 2009; Iwamoto et al., 2017; Malone et al., 2008; Shen, Yu, Weir, & Gorovsky, 1995). How does nuclear import regulate size? Nuclear growth is known to depend on importin α-mediated import of nuclear lamins (Jenkins et al., 1993; Newport, Wilson, & Dunphy, 1990), intermediate filament proteins that are part of the nuclear lamina underlying the inner nuclear membrane (Dittmer & Misteli, 2011). Lamins are one nuclear imported cargo that contributes to nuclear size scaling (Jevtic et al., 2015; Levy & Heald, 2010), and it is likely that other cargos are involved. Nuclear import and lamins also control nuclear size in C. elegans (Ladouceur, Dorn, & Maddox, 2015; Meyerzon et al., 2009). Given that NPCs and lamins exhibit age-dependent deterioration, changes in import capacity could contribute to altered nuclear morphology in aging cells (H. Chen, Zheng, & Zheng, 2014; D’Angelo, Raices, Panowski, & Hetzer, 2009; Flint Brodsly et al., 2019; Frost, Bardai, & Feany, 2016; Toyama et al., 2019).

While nuclear import represents a mechanism that drives nuclear growth, are there balanced nuclear disassembly activities that determine steady-state nuclear size? Using Xenopus embryo extracts, it was shown that late stage embryonic cytoplasm possesses an activity that can cause large nuclei to shrink in size. This activity is dependent on conventional protein kinase C (PKC), and altering PKC activity in vitro and in vivo affects nuclear size. Consistent with PKC being a developmental nuclear size regulator, PKC activity and nuclear localization increase during Xenopus development, and this leads to changes in the phosphorylation and dynamic nuclear envelope (NE) association of lamin B3 (Edens, Dilsaver, & Levy, 2017; Edens & Levy, 2014). PKC has also been implicated as a regulator of nuclear morphology in mouse embryos and neurons (Nishimura et al., 2014; Tsichlaki & FitzHarris, 2016). The fate of nuclear membrane during nuclear shrinking is unknown, however it could stream back into the ER, fold into invaginations, be degraded by nucleophagy (Mochida et al., 2015), or perhaps give rise to nuclear lipid droplets recently described in yeast (Romanauska & Kohler, 2018).

There is also evidence that changes in cytoplasmic volume during development scale nuclear size through titration of limiting components. When nuclei assembled in X. laevis egg extract were confined in microfluidic channels, nuclei reached different steady-state sizes depending on the dimensions of the channel. Nuclei in small channels exposed to less cytoplasm were smaller than nuclei exposed to larger cytoplasmic volumes in larger channel (Fig. 1B). The explanation for this was that channel size dictates the size of microtubule (MT) asters that form next to the nuclei. Small asters lead to less dynein-mediated accumulation of membrane around the nuclei and less nuclear growth, while large asters in larger channels allow for greater accumulation of perinuclear membrane and more nuclear growth (Hara & Merten, 2015), consistent with work showing that a proper tuning of MT polymerization is required for normal nuclear assembly and size (Xue, Woo, Postow, Chait, & Funabiki, 2013). Consistent with perinuclear membrane playing a role, altering the ER tubule-to-sheet ratio and membrane flow can alter nuclear size (Anderson & Hetzer, 2007, 2008; Jevtic & Levy, 2015; Kume, Cantwell, Burrell, & Nurse, 2019). Experiments in which X. laevis embryonic extract and nuclei were encapsulated in microfluidic droplets of defined volume also support a limiting component model. Nuclei grew more in larger droplets (Fig. 1C), and biochemical fractionation identified the histone chaperone nucleoplasmin as one factor that limits nuclear growth (P. Chen et al., 2019). During development, absolute nucleoplasmin amounts per cell decrease, correlating with reduced nuclear histone levels and chromatin compaction and consistent with the idea that nuclear volume is limited by nucleoplasmic factors rather than NE availability (Walters et al., 2019). The proposed mechanism is that increased nuclear import of histones by nucleoplasmin leads to increased chromatin compaction, and stiffer chromatin imparts intranuclear pushing forces on the NE that promote nuclear growth. While extreme chromatin decondensation can increase nuclear size (Bustin & Misteli, 2016), more subtle physiological increases in chromatin compaction might also drive nuclear growth. An open question is the relative contributions of DNA amount and chromatin compaction to nuclear size determination. Future studies will address whether these varied mechanisms of developmental nuclear size scaling are generally conserved in other organisms and in differentiation of various tissues.

An issue related to nuclear morphology and unique to early development in many organisms concerns the formation of a single nucleus. In many embryos, early cell divisions are so rapid that nuclear assembly initiates around individual chromatin masses, forming small nuclei called karyomeres. This allows DNA replication to begin quickly and complete before the next mitosis. In zebrafish, the protein brambleberry ensures that these karyomeres fuse into a single nucleus (Abrams et al., 2012). In C. elegans embryos, a paired nuclei phenotype results when Polo-like kinase PLK-1 or the lipin activator CNEP-1 is inhibited (Bahmanyar et al., 2014; Rahman et al., 2015), bilobed nuclei form in yeast secretion mutants (Walters et al., 2019), and in human cells BAF and KIF18A are important for the formation of a single nucleus after mitosis (Fonseca et al., 2019; Samwer et al., 2017). Failure of these mechanisms can have dire consequences, for instance many cancers exhibit micronuclei in which massive genomic rearrangements called chromothripsis can drive cancer progression (Crasta et al., 2012; Hatch, Fischer, Deerinck, & Hetzer, 2013; C. Z. Zhang et al., 2015). Another unique developmental change in nuclear morphology occurs in Xenopus tail fin cells that become extremely branched through an actin- and lamin-dependent process (Arbach, Harland-Dunaway, Chang, & Wills, 2018).

Recent studies on nuclear structure and function promise to inform novel mechanisms of developmental nuclear size scaling. One such area is nuclear and chromatin mechanics (Cho, Irianto, & Discher, 2017; Heo et al., 2016; Lele, Dickinson, & Gundersen, 2018; Shimamoto, Tamura, Masumoto, & Maeshima, 2017; Stephens, Banigan, & Marko, 2019). For instance, embryonic heart stiffening during development coincides with nuclear lamin A accumulation to prevent nuclear rupture (Cho et al., 2019), stretching forces can lead to NPC opening and increased nuclear import (Elosegui-Artola et al., 2017), and shear force can induce nuclear shrinking (Jetta, Gottlieb, Verma, Sachs, & Hua, 2019). How changes in intranuclear pressure over development might contribute to size scaling is largely unexplored (T. J. Mitchison, 2019). Studies in cultured cells and mouse embryos suggest that intranuclear f-actin contributes to nuclear growth (Baarlink et al., 2017), and whether a similar mechanism scales nuclear size during development is an open question, although nuclear actin levels are known to change during Drosophila oogenesis (Kelpsch, Groen, Fagan, Sudhir, & Tootle, 2016) and early Xenopus development (H. Oda, Shirai, Ura, Ohsumi, & Iwabuchi, 2017). Linker of nucleoskeleton and cytoskeleton (LINC) complexes should be explored with respect to developmental nuclear size scaling (Cantwell & Nurse, 2019; D’Alessandro et al., 2015; Lu et al., 2012; Sun et al., 2019; Tan et al., 2018), as should lipid synthesis at the NE (Barbosa et al., 2019; Haider et al., 2018). In early Drosophila embryos, pre-assembled NPCs are incorporated into nuclei, and developmental regulation of this process could potentially contribute to scaling of nuclear import and size (Hampoelz et al., 2016).

Does nuclear size scaling influence cell function and embryonic development? Specific to early embryogenesis in many species, the midblastula transition (MBT) is characterized by upregulated zygotic gene expression and altered cell cycle kinetics (Blythe & Wieschaus, 2015; Jukam, Shariati, & Skotheim, 2017; Schulz & Harrison, 2019; Vastenhouw, Cao, & Lipshitz, 2019). Altering the N/C ratio in Xenopus embryos changed the timing of MBT onset with concomitant effects on later development (Jevtic & Levy, 2015, 2017; Jevtic, Mukherjee, Chen, & Levy, 2019), and the timing of zygotic genome activation (ZGA) is strongly correlated with cell size (H. Chen, Einstein, Little, & Good, 2019). In Drosophila multinucleate muscle cell development, global scaling of nuclear size with cell size is important for muscle function, while individual nuclear size is spatially regulated (Windner, Manhart, Brown, Mogilner, & Baylies, 2019), as has also been observed in a multinucleate fungus (Dundon et al., 2016). Lamin levels influence nuclear morphology and embryonic heart and retina development (Razafsky et al., 2016; Tran, Zheng, & Zheng, 2016). Increasing ploidy in various amphibians leads to reduced numbers of larger cells such that animal size and the morphology of internal structures are relatively unchanged. For example, higher ploidy and larger cells mean that kidney cell morphology is more drastically altered to ensure proper dimension tubules, while lens epithelial cells in the eye become more flattened to reach the correct thickness (Fankhauser, 1939, 1945a, 1945b; Fankhauser & Watson, 1942). Similar observations have been reported in plants (Robinson et al., 2018). There are also fascinating correlations between cell size and brain morphology in amphibians (G. Roth & Walkowiak, 2015), and polyploidization-induced cell size increases contribute to would healing in Drosophila (Losick, Fox, & Spradling, 2013).

Some nuclei must adopt unique morphologies appropriate for cell function. A classic example is the multi-lobulated neutrophil nucleus that imparts nuclear flexibility to facilitate migration through narrow constrictions (Manley, Keightley, & Lieschke, 2018), regulated in part through altered expression of nuclear lamina proteins and genome organization (Rowat et al., 2013; Zhu et al., 2017). Changes in nuclear shape and mechanics are also important for constricted cell migration during development and cancer metastasis (Bone, Chang, Cain, Murphy, & Starr, 2016; Denais et al., 2016; Jayo et al., 2016; Pfeifer et al., 2018; Raab et al., 2016; Smith, Irianto, Xia, Pfeifer, & Discher, 2019). At the extreme, erythrocytes in some small amphibian species undergo partial or complete enucleation to allow for their circulation through narrow vessels (Mueller, 2015; Mueller, Gregory, Gregory, Hsieh, & Boore, 2008), and lens fiber cells must degrade their nuclei (Chaffee et al., 2014). Many questions remain about the relationship between nuclear size and function, in the context of both embryonic development and disease, particularly laminopathies where altered nuclear lamina composition contributes to a varying array of pathologies (Dobrzynska, Gonzalo, Shanahan, & Askjaer, 2016; Osmanagic-Myers & Foisner, 2019; Worman & Schirmer, 2015) and cancers with enlarged nuclei (Chow, Factor, & Ullman, 2012; Jevtic & Levy, 2014; Zink, Fischer, & Nickerson, 2004).

Mitotic spindles

Mitotic spindles are composed of microtubules (MTs), polymers of repeating α/β tubulin heterodimers. MTs undergo phases of growth and shrinkage known as dynamic instability (T. Mitchison & Kirschner, 1984). A variety of MT binding and regulatory proteins affect MT dynamics and consequently spindle length (Rieckhoff, Ishihara, & Brugues, 2019). Spindles are responsible for equal segregation of chromosomes between daughter cells during cell division. They also interact with the cell cortex, playing a role in the initiation of contractile ring formation and cytokinesis (Akhshi, Wernike, & Piekny, 2014). It is, therefore, important that spindle size scales with cell size to ensure these processes are faithfully executed. As with nuclei, there is evidence that DNA amount has a relatively small impact on spindle size (Brown et al., 2007; Wuhr et al., 2008), although chromatin does modulate MT dynamics and spindle morphology (Carazo-Salas et al., 1999; Dinarina et al., 2009; Hara & Kimura, 2013).

Spindle scaling has been studied extensively during Xenopus, Mus musculus, and C. elegans development. In early Xenopus embryos, reductions in cell size correlate with decreased spindle length in different stage blastomeres (Wuhr et al., 2008) (Fig. 1D). Reaching a plateau in early stages, spindle length scales roughly linearly with cell size beyond the 32-cell stage. This suggests that spindle size is uncoupled from cell size during the early stages of development, with spindle length reaching an upper limit. Similar results are observed in developing C. elegans embryos where spindle size scales in 20 μm-diameter and smaller cells, while such scaling is less evident in larger, earlier stage cells (Hara & Kimura, 2009) (Fig. 3B). In mouse embryos spindle size remains constant during the first three divisions and only scales smaller with cell size after the fourth division (Courtois, Schuh, Ellenberg, & Hiiragi, 2012), although fusion of early stage blastomeres can lead to increased spindle length (Novakova et al., 2016). Cytoplasm removal or embryo fusion showed that spindle scaling is dependent on cytoplasmic volume and not developmental stage (Courtois et al., 2012; Kyogoku & Kitajima, 2017; Lane & Jones, 2017). Interestingly, in sea urchin embryos, spindle length scaling is evident from the first cell division (Lacroix et al., 2018) (Fig. 4C). In asymmetrically dividing C. elegans embryos, cell and spindle sizes correlate at the same developmental stage, providing further support for cytoplasmic volume regulating spindle size (Hara & Kimura, 2009; Lacroix et al., 2018). In addition, the rate of spindle elongation scales with cell size in worms (Hara & Kimura, 2009).

Mechanisms of spindle length scaling have been elucidated using cell-free in vitro approaches that recapitulate spindle assembly. The contribution of cytoplasmic volume to spindle size scaling was explicitly tested by coupling Xenopus egg extracts with microfluidic approaches to encapsulate cytoplasm in droplets of defined size (Oakey & Gatlin, 2018). Similar to results in Xenopus embryos, spindle size scaled with cytoplasmic volume in droplets ranging from 30–80 μm in diameter, with an upper limit to spindle size being reached in larger droplets (Good, Vahey, Skandarajah, Fletcher, & Heald, 2013; Hazel et al., 2013) (Fig. 1E). Because droplet volume and dimensions were both manipulated in these experiments, spindle size might be sensitive to volume and/or boundary sensing. To distinguish these two models, isovolumetric droplets of differing shape were generated. Spindle size was observed to scale with droplet volume and to not be affected by droplet shape (Good et al., 2013; Hazel et al., 2013), supporting the idea that cytoplasmic components may limit spindle size (Goehring & Hyman, 2012).

One obvious candidate limiting component is tubulin itself, since it is the building block of MTs. Indeed, a tubulin-limiting computational model accurately predicted Xenopus developmental spindle scaling data, although supplementing extract droplets with porcine brain tubulin did not affect spindle size (Good et al., 2013). Several lines of evidence suggest that tubulin may not be the primary determinant of spindle length scaling (Lacroix et al., 2018). Another promising candidate is XMAP215, a MT polymerase that binds tubulin and MT plus ends (Reber et al., 2013). Depletion of XMAP215 from Xenopus egg extracts resulted in short spindles, and titrated addition of wild-type XMAP215 led to graded increases in MT growth velocity and spindle length. Addition of XMAP215 mutants with reduced affinity for tubulin revealed a correlation between MT polymerase activity and spindle length (Reber et al., 2013). Consistent with these in vitro data, microinjection of XMAP215 into Xenopus embryos led to increased spindle size beyond the usual upper limit, indicating structural components are not limiting, while a polymerase-deficient mutant decreased spindle length (Milunovic-Jevtic, Jevtic, Levy, & Gatlin, 2018). Taken together, these studies identify XMAP215 as a putative limiting component in developmental spindle size scaling and highlight how factors that affect MT dynamics and growth rates can regulate spindle size, as opposed to spindle structural components.

Following up on these Xenopus studies, developmental changes in MT dynamics were carefully examined in C. elegans embryos (Lacroix et al., 2018). Concomitant with spindle size reductions, MT growth rates also decreased, but no changes in MT shrinkage rates or the frequencies of MT catastrophe and rescue were observed. In addition, the spindle assembly rate was decreased in smaller cells. In mutant 2-cell stage embryos in which cytokinesis was blocked, MT growth rates were similar to 1-cell wild-type embryos, suggesting that MT growth rates and spindle size are dictated by cytoplasmic volume rather than developmental stage or timing. Similar correlations between MT growth rates and cell and spindle sizes were observed during early development of much larger sea urchin embryos, suggesting that scaling of MT growth rates may represent a conserved mechanism of spindle size scaling. Based on computational modelling, the proposed mechanism is that the absolute amounts of positive regulators of MT growth are reduced upon being partitioned into smaller and smaller cells over development, leading to reduced MT growth rates and spindle sizes (Fig. 3). Candidate limiting components for spindle size scaling include Xenopus XMAP215 and the C. elegans CLASP homolog CLS-2 that promotes MT assembly (Lacroix et al., 2018).

There is also evidence that MT depolymerization can contribute to mitotic spindle scaling. Xenopus developmental progression is accompanied by increased spindle recruitment of the MT-depolymerizing kinesin kif2a, leading to increased MT catastrophe frequency and reduced spindle length (Wilbur & Heald, 2013). This developmental change in kif2a localization is regulated by importin α. Kif2a is inhibited by bound importin α, leading to reduced MT depolymerization. As discussed in the context of nuclear scaling, in early embryos there is a large pool of cytoplasmic importin α that acts to inhibit kif2a. As palmitoylated importin α exhibits reduced cargo binding and is partitioned to the plasma membrane over development, there is less cytoplasmic importin α available to inhibit kif2a, so activated kif2a relocalizes to spindles leading to progressive reductions in spindle size (Brownlee & Heald, 2019; Wilbur & Heald, 2013). With a similar importin α-mediated mechanism contributing to developmental nuclear size scaling in Xenopus, this represents a fascinating example of coordinated size regulation of two organelles by a similar mechanism (Fig. 1G). Studies in mouse embryos also suggest that the nuclear-to-cytoplasmic ratio might influence spindle size scaling (Novakova et al., 2016).

In addition to developmental spindle scaling, correlations between embryo, cell, and spindle sizes have been characterized across a wide variety of different species (Crowder et al., 2015; Farhadifar et al., 2015; Valfort, Launay, Semon, & Delattre, 2018). X. laevis spindles are larger than X. tropicalis spindles (Brown et al., 2007). Two regulators that contribute to this difference are the MT-severing protein katanin and TPX2, a MT-associated protein that binds the kinesin-5 motor Eg5. Although katanin levels are similar in the two species, X. tropicalis katanin possesses higher MT severing activity due to the absence of an inhibitory phosphorylation site present in X. laevis katanin (Loughlin, Wilbur, McNally, Nedelec, & Heald, 2011). Higher TPX2 levels in X. tropicalis lead to increased polar recruitment of Eg5, altering MT structure at the poles to reduce spindle length (Helmke & Heald, 2014). Recently described egg extracts from a third Xenopus clawed frog species, X. borealis, promise to provide additional new insights into interspecies organelle size scaling (Kitaoka, Heald, & Gibeaux, 2018).

In summary, spindle size generally scales with cell size during early development. A number of positive and negative regulators of MT dynamics contribute to the phenomenon of spindle size scaling, and it is likely that multiple mechanisms must be invoked to fully account for the magnitude and different regimes of scaling (Brownlee & Heald, 2018). Positive regulators include XMAP215 and CLS-2, while negative regulators include katanin and kif2a. Apart from regulators of MT dynamics, other aspects of cell biology that deserve further consideration with respect to their potential impact on spindle size scaling include: 1) liquid-liquid phase separation that has been shown to enrich MT regulatory factors in acentrosomal spindle assembly (So et al., 2019), possibly akin to the previously described spindle matrix (Jiang et al., 2015; Johansen, Forer, Yao, Girton, & Johansen, 2011), 2) feedback from altered MT dynamics on tubulin expression levels (Gasic, Boswell, & Mitchison, 2019), 3) the influence of actin on MT dynamics (Colin, Singaravelu, Thery, Blanchoin, & Gueroui, 2018; Inoue et al., 2019; Kita et al., 2019; Mogessie & Schuh, 2017), and 4) interphase MT bridges between sister cell pairs in early mammalian embryos (Zenker et al., 2017). Computational models of spindle size regulation may also inform scaling mechanisms (Jiang, 2015; Li & Jiang, 2017). Further investigations into how these and other factors coordinately regulate MT dynamics during development promise to provide a more complete picture of spindle scaling mechanisms. Another exciting area for future research will be to utilize this new mechanistic information to manipulate spindle size scaling in embryos to assess effects on chromosome segregation, cell division site determination, developmental timing, and embryonic morphology and function.

Centrosomes

Centrosomes nucleate the MTs of interphase asters and mitotic spindles. Consisting of a pair of centrioles and a number of proteins referred to as pericentriolar material, nucleus-juxtaposed centrosomes duplicate during interphase to give rise to mitotic spindle poles. In C. elegans, centrosomes begin to mature in interphase, grow during mitosis, and scale smaller with cell size over development (Decker et al., 2011; Greenan et al., 2010) (Fig. 3C). Reducing centrosome size was sufficient to decrease spindle length (Greenan et al., 2010). In the first two cell divisions, wild-type centrosomes continue to grow while a smaller steady-state centrosome size was reached in a small embryo mutant, suggesting centrosome growth rate depends on cell size but not developmental stage. While centrosome growth rates are slower in smaller cells, total centrosome volume remains constant up to the 4-cell stage and in mutants with increased or decreased numbers of centrosomes. These data support a limiting component model where cell size determines the amount of available centrosome components, thereby setting centrosome volume. One putative limiting component is Spd-2, a conserved protein key to centrosome assembly, which when overexpressed increased centrosome growth rate and volume (Decker et al., 2011). Additional factors that may contribute to developmental reductions in centrosome size include Polo-like kinase 1 and Spd-5 (Wueseke et al., 2016), and recent work showing centrosome formation by phase separation may also inform the scaling mechanism (Woodruff et al., 2017).

How might centrosome size contribute to spindle length control? One model involves a gradient of centrosome-derived TPXL-1 (C. elegans homolog of TPX2), in which a larger centrosome loads more TPXL-1 onto spindle MTs, producing a wider TPXL-1 gradient and longer spindles (Greenan et al., 2010). Species-specific differences in TPXL-1 and TPX2 may account for their different effects on spindle size in C. elegans and Xenopus (Bird & Hyman, 2008; Helmke & Heald, 2014). Chemical waves emanating from centrosomes have also been proposed to organize MT assemblies in Xenopus (Ishihara et al., 2014), and nuclear size can affect centrosome attachment and separation (Boudreau, Chen, Edwards, Sulaimain, & Maddox, 2019; Meyerzon et al., 2009).

Chromosomes and chromatin

From a physical perspective, it reasons that mitotic chromosome length might scale over development. If chromosome length were to exceed cell length in small cells, this could result in chromosome fragmentation and/or aneuploidy. Indeed, artificially increasing chromosome length in fava bean resulted in growth and developmental defects (Schubert & Oud, 1997), and nuclei in closed yeast mitoses must grow sufficiently to ensure faithful chromosome segregation (Takemoto et al., 2016). In Xenopus, mitotic chromosome length scaling, though not evident in early development, is apparent from the blastula-to-neurula stage (Kieserman & Heald, 2011) (Fig. 1F). Because blastomeres are quite large in the early embryo, it is perhaps expected that chromosome scaling would only be necessary when cells become significantly smaller post-blastula. Mitotic chromosome length also scales with cell size during C. elegans development (Hara et al., 2013; Ladouceur et al., 2015). It is unknown if developmental chromosome length scaling is mediated through active sensing of cell size. Candidate chromosome scaling factors include topoisomerase-II, the histone H3 variant CENP-A, and condensins (Ladouceur et al., 2017; Shintomi & Hirano, 2011). Intriguingly, a new microscopy technique that integrates correlative light and electron microscopy revealed that chromatin constitutes a relatively small amount of the metaphase chromosome mass, begging the question of what other chromatin-associated proteins might contribute to chromosome scaling (Booth et al., 2016).

Another interesting context to consider chromosome scaling is the increase in genome size that occurred during eukaryotic evolution, where the insertion of arginine residues into the N-terminus of histone H2A led to greater chromatin compaction of larger genomes (Macadangdang et al., 2014). Because there is a general correlation between nuclear and chromosome size and because reducing nuclear growth led to smaller mitotic chromosomes (Hara et al., 2013; Ladouceur et al., 2015), one proposal is that nuclear size and/or nucleocytoplasmic transport of chromosome scaling factors during interphase might determine chromosome size (Heald & Gibeaux, 2018). It is tempting to speculate that limiting nuclear import may represent a coordinated mechanism for the control of nuclear and chromosome size as well as MBT timing.

While some mechanisms responsible for mitotic chromosome size scaling have been elucidated, much less is known about interphase chromosome size, in part because quantifying decondensed chromosome length is more difficult. Nonetheless, interphase chromosome size scaling may be functionally relevant. There is evidence in cell culture that cell geometry can influence the 3D positioning of chromosome territories within the nucleus, thereby influencing gene expression (Y. Wang, Nagarajan, Uhler, & Shivashankar, 2017). Perhaps stereotyped changes in chromatin organization and gene expression during embryogenesis might, at least in part, be regulated by cell size and shape (Andrey & Mundlos, 2017; Borsos et al., 2019; Buchwalter, Kaneshiro, & Hetzer, 2019; Flyamer et al., 2017; Perino & Veenstra, 2016; Stadhouders, Filion, & Graf, 2019; Wu et al., 2018). Changes in chromatin architecture, potentially linked to chromosome scaling, are important drivers of early developmental programs. For example, studies in Xenopus, Drosophila, and zebrafish have shown that limiting histone pools and competition for transcription factor binding, whose concentrations can be influenced by nuclear size, are key determinants of ZGA (Amodeo, Jukam, Straight, & Skotheim, 2015; S. H. Chan et al., 2019; Joseph et al., 2017; Reisser et al., 2018; Shindo & Amodeo, 2019; Wilky, Chari, Govindan, & Amodeo, 2019). At least in Drosophila, changes in chromatin architecture associated with ZGA are independent of transcription, suggesting that chromosome re-positioning occurs upstream of, and is required for, ZGA (Hug, Grimaldi, Kruse, & Vaquerizas, 2017). Subsequent to the MBT, cell cycle slowing regulates the establishment of constitutive heterochromatin in Drosophila and C. elegans, through the increased accumulation of histone methyltransferases on the DNA (Mutlu et al., 2018; Seller, Cho, & O’Farrell, 2019). At the extreme, improper chromosome scaling can lead to inviability. For instance, when X. tropicalis eggs are fertilized with X. laevis sperm containing nearly twice the amount of genomic DNA, hybrid embryos die at late blastula stages due to chromosome mis-segregation and unbalanced gene expression (Gibeaux et al., 2018). Although requiring future investigation, chromosome scaling may underlie several key events in early development.

Endoplasmic reticulum and interconnected organelles

The endoplasmic reticulum (ER) is an extensive membrane network composed of tubules that intersect at three-way junctions and ribosome-studded rough ER sheets (N. Wang & Rapoport, 2019; H. Zhang & Hu, 2016). While developmental scaling of ER size and structure have not been studied in great detail, one possibility is that ER is equally partitioned into dividing blastomeres through MT-dependent positioning (S. Wang, Romano, Field, Mitchison, & Rapoport, 2013), resulting in progressive scaling of ER size, although this scaling relationship could be altered if new membrane synthesis occurs. In C. elegans, ER compartmentalization is evident as early as the one-cell stage (Lee, Prouteau, Gotta, & Barral, 2016), consistent with separate ER domains roughly delineating future daughter cells in budding yeast and neural stem cells (Clay et al., 2014; Luedeke et al., 2005; Moore, Pilz, Arauzo-Bravo, Barral, & Jessberger, 2015; Pina, Fleming, Pogliano, & Niwa, 2016). ER structural proteins including REEP3/4 help to clear tubulated ER membrane from metaphase chromatin during mitosis and may facilitate equal partitioning of ER to daughter cells (Kumar, Golchoubian, Belevich, Jokitalo, & Schlaitz, 2019; Schlaitz, Thompson, Wong, Yates, & Heald, 2013). In a subset of cells in gastrulating Drosophila embryos, asymmetric ER partitioning occurs dependent on the conserved ER membrane protein Jagunal (Eritano et al., 2017).

Steady-state size and number of many membrane-bound organelles are determined by balanced fusion and fission that are regulated by diverse activities including lipid signalling, expression of coat and adaptor proteins, and cytoskeletal interactions. This mode of size regulation applies to lysosomes (Saffi & Botelho, 2019), mitochondria (Miettinen & Bjorklund, 2017), and Golgi (Dippold et al., 2009; Sengupta & Linstedt, 2011). In a related model, the functional requirements of an organelle may dictate its size, as is the case for the Golgi where cisternal size is regulated by membrane trafficking kinetics (Bevis, Hammond, Reinke, & Glick, 2002; Bhave et al., 2014). By and large, developmental size regulation of these various organelles has not been thoroughly investigated. Simply determining if it is organelle surface area or volume that scales with cell size may provide clues to the underlying mechanisms. One might imagine surface area to scale for membrane-limiting reactions, for instance protein synthesis in the rough ER, while volume might scale for organelles like lysosomes.

While yeast is not a developmental system per se, mechanisms identified in yeast that regulate the sizes of membrane-bound organelles will undoubtedly inform scaling in higher eukaryotes. For example, ER stress is alleviated by membrane expansion (Schuck, Prinz, Thorn, Voss, & Walter, 2009), while ER-phagy selectively degrades excess ER membrane in the vacuole, the yeast equivalent of the lysosome (Schuck, Gallagher, & Walter, 2014). Mathematical modelling and experimental measurements in yeast showed that stochastic fluctuations in the abundance of Golgi, vacuoles, and peroxisomes must be taken into account when considering size scaling relationships (Mukherji & O’Shea, 2014). SNAREs and membrane trafficking contribute to yeast vacuole size control, and similar mechanisms may regulate lysosome size scaling in higher eukaryotes (Y. H. Chan & Marshall, 2014; Y. H. Chan, Reyes, Sohail, Tran, & Marshall, 2016; D’Agostino, Risselada, Endter, Comte-Miserez, & Mayer, 2018; Desfougeres, Neumann, & Mayer, 2016).

Interesting correlations between mitochondrial size, cell size, and metabolism may inform future scaling studies. Mitochondrial size scales with cell size in yeast, with new mitochondrial accumulation occurring in the bud and reduced mitochondrial content in aging mothers (Jajoo et al., 2016; Rafelski et al., 2012). In addition, the mitochondria-to-cytoplasm ratio remains constant during HeLa cell growth (Posakony, England, & Attardi, 1977). Assuming mitochondrial scaling is similar in yeast and higher eukaryotes, cell size-dependent mitochondrial scaling during development and differentiation might be expected, although future research will be necessary to acquire these measurements and to determine the underlying mechanisms. More is known about the scaling of mitochondrial function with cell size. In cultured Drosophila cells, mitochondrial mass scales with cell size, however mitochondrial function, measured by membrane potential and oxidative phosphorylation, is highest in intermediate sized cells. This optimal cell size also correlates with reduced apoptosis and increased proliferation (Miettinen & Bjorklund, 2016). Consistent with these findings, increasing hepatocyte size downregulates expression of mitochondrial genes and lipogenic transcription factors, and inhibiting mitochondrial function leads to increased cell size (Miettinen et al., 2014). How mitochondrial activity is correlated with cell size is not fully understood, though the mevalonate pathway has been implicated (Miettinen & Bjorklund, 2016). In the developing Drosophila intestine, cell size reductions are accompanied by autophagy-dependent clearance of mitochondria. This mitochondrial scaling is dependent on the ubiquitin binding activity of Vps13D, and large cells and mitochondria in Vps13D mutants are rescued by inhibiting mitochondrial fusion (Anding et al., 2018). In mammalian cells, Largen (gene PRR16) increases translation of mitochondrial proteins, leading to increased mitochondrial size, respiration, and cell size (Yamamoto et al., 2014). Thus multiple mechanisms contribute to the complex scaling relationships between cell size and mitochondrial size and function.

There is evidence that ER morphology controls the size and morphology of other organelles that contact the ER (Gottschling & Nystrom, 2017; H. Zhang & Hu, 2016), including ER-mitochondria contacts that coordinate mitochondrial DNA replication and division (Lewis, Uchiyama, & Nunnari, 2016). Of note, the outer nuclear member is continuous with the ER, and nuclear expansion in early embryos can occur through streaming of ER into the NE as well as by direct incorporation of annulate lamellae (Anderson & Hetzer, 2007; Hampoelz et al., 2016). Manipulations that increase ER tubulation generally decrease nuclear size while increasing ER sheet volume drives nuclear growth, suggesting a tug-of-war between ER and nuclear membranes (Anderson & Hetzer, 2008; Jevtic & Levy, 2015). The ER is also key in the biogenesis of other organelles, such as peroxisomes and lipid droplets (Joshi et al., 2016; Joshi, Zhang, & Prinz, 2017; Sugiura, Mattie, Prudent, & McBride, 2017).

Because relatively little is known about developmental scaling of membrane-bound organelles, many questions remain about its physiological significance. Still there is abundant evidence that the size and morphology of membrane-bound organelles affect functional output. Cells that synthesize and secrete large amounts of proteins, such as plasma cells and pancreatic acinar cells, are enriched in rough ER sheets. On the other hand, smooth ER tubules are abundant in hepatocytes, adrenal cortical cells, and muscle cells that are specialized for carbohydrate metabolism, steroid hormone synthesis, and calcium signalling, respectively (Black, 1972; Friedman & Voeltz, 2011; Goyal & Blackstone, 2013; Shibata, Voeltz, & Rapoport, 2006; West, Zurek, Hoenger, & Voeltz, 2011). The unfolded protein response and ER stress induced by excess fatty acids lead to expansion of ER sheets, which is energetically more favorable than ER tubule expansion and is thought to provide more space for protein folding (Friedman & Voeltz, 2011; Schuck, 2016; Schuck et al., 2009; Walter & Ron, 2011; Wikstrom et al., 2013). In developing Drosophila photoreceptors, rough ER sheets expand to produce the protein synthetic capacity necessary to build the photosensitive rhabdomere in a process requiring Ire1, a key component of the unfolded protein response (Xu, Chikka, Xia, & Ready, 2016). It is worth noting that changing ER morphology is not sufficient to alter exocytosis (Mukherjee & Levy, 2019), so further studies are required to understand which functional outputs of the ER are sensitive to organelle size and morphology. Disrupting mitotic peroxisome distribution leads to delayed mitosis and altered spindle positioning with concomitant defects in polarized cell division and daughter cell fate determination (Asare, Levorse, & Fuchs, 2017). Golgi and mitochondrial sizes also must scale with cell size and type to ensure the cell’s metabolic requirements are met (Miettinen & Bjorklund, 2017; Sengupta & Linstedt, 2011).

Flagella, cilia, stereocilia, and microvilli

Because flagella and cilia are roughly linear structures (in eukaryotes flagella and cilia refer to the same organelle), they represent somewhat simplified organelles for the study of size regulation. Flagellar length control in particular has been studied in Chlamydomonas and revealed mechanisms involving balanced rates of assembly and disassembly (Hilton, Gunawardane, Kim, Schwarz, & Quarmby, 2013; Ishikawa & Marshall, 2017; Marshall, 2015; Meng & Pan, 2016; Wemmer & Marshall, 2007) as well as molecular rulers (Hendel, Thomson, & Marshall, 2018; T. Oda, Yanagisawa, Kamiya, & Kikkawa, 2014). For instance, NIMA-related kinases have recently been implicated in equilibrium balance models of ciliary length control while the size of the structural FAP59/172 complex influences flagella length. Centrioles are critical for cilia formation, and centriole length varies between different species and cell types (Azimzadeh & Marshall, 2010). In Drosophila embryos, a variety of factors contribute to proper centriole length control, including Polo-like kinase 4 that acts as a homeostatic clock to regulate centriole growth rate (Aydogan et al., 2018) and Asterless that plays independent roles in centriole length control and sperm basal body function (Galletta, Jacobs, Fagerstrom, & Rusan, 2016). In terminally differentiated epithelia of the respiratory tract, centriole amplification produces 100–600 centrioles with a concomitant increase in the number of motile cilia. In airway progenitor cells, ablation of the two parental centrioles did not affect centriole amplification. Rather, by manipulating cell size by varying the extracellular collagen concentration, it was shown that cell surface area scales centriole and cilia number (Nanjundappa et al., 2019). In turn, fluid flow-mediated signalling in primary cilia can regulate epithelial cell volume through an autophagy-based mechanism (Orhon et al., 2016).

Linear actin-based structures that are developmentally regulated include stereocilia and microvilli. Stereocilia that constitute cochlear hair bundles are key to hearing. During hair cell differentiation, changes in the number, length, and organization of stereocilia are developmentally regulated by class III myosins and actin capping proteins, and defects in this process lead to hearing loss in mice (Avenarius et al., 2017; Lelli et al., 2016). During vertebrate oogenesis, oocytes grow larger and synthesize proteins and membranes necessary for early embryogenesis. In Xenopus oocytes, the Ca²⁺-activated chloride channel Ano1 contributes to this increase in oocyte surface area. Independent of its channel activity, Ano1 regulates microvilli length and scaffolding, leading to stabilization of larger membrane structures (Courjaret et al., 2016). Identifying other cellular systems where the lengths and/or numbers of these various MT- and actin-based linear structures change over development will provide new insights into scaling of these unique organelles.

Other intracellular structures

While not organelles per se, several other intracellular structures exhibit interesting size scaling relationships. The nucleolus is a membraneless organelle within the nucleus that contains rDNA repeats and is therefore key to ribosome biogenesis. In early C. elegans embryos, nucleolar size directly scales with cell and nuclear size. Interestingly, when embryo size is altered by various RNAi treatments, nucleolar size inversely scales with cell size, consistent with a fixed amount of nucleolar material becoming diluted or concentrated in larger or smaller RNAi-treated embryos, respectively (Weber & Brangwynne, 2015). In extremely large Xenopus oocytes, distinct nucleoli are maintained by an F-actin scaffold and disrupting this scaffold causes nucleoli to sediment under the force of gravity and fuse (Feric & Brangwynne, 2013). This fusion behaviour revealed that the nucleolus exhibits liquid-like properties and forms through phase separation, regulating nucleolar size and shape (Brangwynne, Mitchison, & Hyman, 2011). Transcription of rRNA is important for seeding nucleolus assembly and ensuring the proper number of nucleoli form at the correct time and location (Berry, Weber, Vaidya, Haataja, & Brangwynne, 2015; Falahati, Pelham-Webb, Blythe, & Wieschaus, 2016). Reconstitution experiments demonstrated that nucleoli are composed of subcompartments in which distinct RNA processing reactions might occur (Feric et al., 2016). Thus the phase separation properties of the nucleolus are important for both size control and function. It is also worth noting that nucleolar size increases with normal and disease-related aging (Buchwalter & Hetzer, 2017).

The sizes of clathrin-coated endocytic lattices change during cellular differentiation, mediated by altered expression of the AP2μ2 adaptor (Dambournet et al., 2018). Retinal pigment epithelial cells contain melanosomes, melanin-containing structures that reduce backscattered light and neutralize free radicals. Proper regulation of melanosome number and shape are required for normal vision (Burgoyne, O’Connor, Seabra, Cutler, & Futter, 2015). Drosophila embryos possess actin-rich structures called denticles that line the ventral epidermis and may be important for larval locomotion. Denticle number and spacing exhibit MT-dependent scaling with cell length over development (Spencer, Schaumberg, & Zallen, 2017). Acting through a ruler mechanism, the size of the large muscle protein nebulin dictates actin thin filament length and spacing, thereby influencing sarcomere size (Fernandes & Schock, 2014). The size of a unique secreted particle, the Weibel-Palade body, is determined by the size and number of Golgi ministacks, providing an interesting example of how the sizes of two organelles might be interrelated (Ferraro et al., 2014). In Arabidopsis, the speed of myosin-regulated cytoplasmic streaming affects cell and plant size but not cell number (Tominaga et al., 2013). In Plasmodium, sporozoite shape and infectivity are determined by MT number, which is reduced upon knocking down the level of α-tubulin expression (Spreng et al., 2019). Understanding how the sizes of these diverse structures are regulated may inform organelle-scaling mechanisms.

Conclusion

While size scaling of certain organelles with cell size has been known about for over a century, the last decade has witnessed a renewed interest in the phenomenology and mechanisms of organelle size scaling. Many of these recent studies have taken advantage of early developmental systems where reproducible reductions in cell size provide a powerful platform to study size scaling at the intracellular level. While our understanding of the developmental regulation of nuclear and spindle size has advanced, basic questions still remain about how the sizes of many membrane-bound organelles scale over early development, including the ER, Golgi, lysosomes, and mitochondria. Though these organelles exhibit complex morphologies, new microscopy techniques and image analysis approaches promise to facilitate quantification of developmental changes in their sizes and properties (Chung et al., 2013; Ghosh et al., 2019; Guo et al., 2018; Liu et al., 2018; Ou et al., 2017; Stegmaier et al., 2016; Valm et al., 2017). It is also worth considering new models of early development that might be amenable to studying size scaling of these organelles (Behringer, Johnson, & Krumlauf, 2009; Goldstein & King, 2016), as well as regeneration (Stocum, 2017). Developmental scaling processes in human cells might be studied using organoids or micropatterned stem cells (Deglincerti et al., 2016; Lehmann et al., 2019; Morgani, Metzger, Nichols, Siggia, & Hadjantonakis, 2018). Diverse mechanisms can account for organelle scaling in early embryos, including limiting components and dynamic regulation of assembly and disassembly (Fig. 2). It is clear that multiple, non-mutually exclusive mechanisms can contribute to organelle size regulation and more than one model may be required to account for the full extent of organelle size scaling. In that regard, it will be exciting to see how phase separation and interorganellar contacts contribute to developmental control of organelle size. In the way that importin α has been shown to regulate the sizes of both Xenopus spindles and nuclei, another exciting area will be to investigate how the sizes of distinct organelles might be coordinately regulated.

A variety of new techniques and approaches may lend themselves to studies of organelle size. Bottom-up cellular reconstitution allows for precise control of synthetic cell size and shape to address how these properties influence organelle size scaling (Bermudez, Chen, Einstein, & Good, 2017; Gopfrich, Platzman, & Spatz, 2018; Sonnen & Merten, 2019; Vahey & Fletcher, 2014). Identifying the parts list for a given organelle will facilitate studies of size because the presence and levels of individual components can be precisely modulated. In that vein, the tubular ER has been reconstituted in vitro (Powers, Wang, Liu, & Rapoport, 2017). CRISPR approaches now allow for precise genome modification in over 20 different plant and animal species, for minimally disruptive labelling of organelles for size quantification and manipulation of putative size scaling factors (Harrison, Jenkins, O’Connor-Giles, & Wildonger, 2014). Complementary methods allow for more acute, temporally regulated degradation of endogenous proteins (S. Roth, Fulcher, & Sapkota, 2019). Comprehensive transcriptomics and proteomics databases of early development in a variety of organisms provide a rich resource to identify putative developmental regulators of organelle size (Casas-Vila et al., 2017; Lucitt et al., 2008; Peshkin et al., 2015; Xia et al., 2018).

As mechanisms of developmental organelle scaling are elucidated, it will become increasingly feasible to manipulate organelle size in cells and alter normal scaling relationships to address the consequences for cell function and embryonic development. New optogenetic methods might allow for dynamic modulation of organelle size in living embryos (Buckley et al., 2016; Niopek, Wehler, Roensch, Eils, & Di Ventura, 2016). These functional studies will be critical to distinguish whether changes in organelle size over development are merely a secondary consequence of cell size reductions or if organelle size scaling is important for proper spatiotemporal patterning and development of the embryo. As one example, nuclear size may influence gene positioning and expression, and this hypothesis can now be addressed using new approaches to image 3D chromatin organization and transcription in embryos (Bothma, Norstad, Alamos, & Garcia, 2018; Cardozo Gizzi et al., 2019; Y. Chen et al., 2018; Stevens et al., 2017) and to manipulate the positions of specific genomic loci (H. Wang et al., 2018; Wijchers et al., 2016). Functional studies will also inform the aberrant organelle scaling that is often observed in cancer and certain developmental disorders (Chow et al., 2012; Jevtic & Levy, 2014; Zink et al., 2004), and model systems like Xenopus promise to inform disease mechanisms related to altered scaling (Hardwick & Philpott, 2015; Kakebeen & Wills, 2019; Sater & Moody, 2017). With many exciting open questions about the mechanisms and functional significance of developmental organelle size scaling, the next decade is sure to see further advances in this fascinating field.