Spontaneous emergence of cell-like organization in Xenopus egg extracts

See allHide authors and affiliations

Science  01 Nov 2019:
Vol. 366, Issue 6465, pp. 631-637
DOI: 10.1126/science.aav7793

Order in the cytoplasm

Extracts of the very large eggs of the African clawed frog, Xenopus laevis, have proven a valuable model system for the study of cell division. Cheng and Ferrell found that after homogenization, such cytoplasm can reorganize back into cell-like structures and undergo multiple rounds of division (see the Perspective by Mitchison and Field). This reorganization apparently occurs without the usual factors that are known to lead to such structural changes during cell division, such as F-actin, myosin II, various individual kinesins, aurora kinase A, or DNA. What is required is energy from adenosine triphosphate, microtubule polymerization, cytoplasmic dynein activity, and a specific kinase-involved cell cycle progression. Nongenetic information in the cytoplasm is apparently sufficient for basic spatial organization of the cell.

Science, this issue p. 631; see also p. 569


Every daughter cell inherits two things from its mother: genetic information and a spatially organized complement of macromolecular complexes and organelles. The extent to which de novo self-organization, as opposed to inheritance of an already organized state, can suffice to yield functional cells is uncertain. We used Xenopus laevis egg extracts to show that homogenized interphase egg cytoplasm self-organizes over the course of ~30 minutes into compartments 300 to 400 micrometers in length that resemble cells. Formation of these cell-like compartments required adenosine triphosphate and microtubule polymerization but did not require added demembranated sperm nuclei with their accompanying centrosomes or actin polymerization. In cycling extracts with added sperm, the compartments underwent multiple cycles of division and reorganization, with mother compartments giving rise to two daughters at the end of each mitotic cycle. These results indicate that the cytoplasm can generate much of the spatial organization and cell cycle function of the early embryo.

If an organism were homogenized, most of us would predict that it would not spontaneously reform its original structure and return to life (1). Indeed, organismal development unfolds the body plan in a specific sequence that is not likely to be reproduced by simply mixing its underlying chemical ingredients. However, it is not immediately clear whether the same is to be expected for the cytoplasm of a cell. If a cell is mechanically homogenized, will the resulting cytoplasm remain in the homogenized state?

We used Xenopus laevis egg extracts (Fig. 1A) to address this question, in part because Xenopus extracts can be prepared without substantial dilution and in part because they have been shown to be remarkably functional (29). Interphase-arrested Xenopus extracts were prepared by standard methods (10) and supplemented with demembranated Xenopus sperm nuclei plus one or more of the following probes to allow visualization of the nucleus, microtubules, endoplasmic reticulum (ER), and mitochondria: green fluorescent protein or mCherry with a nuclear localization signal (GFP-NLS or mCherry-NLS), fluorescently labeled tubulin or SiR-tubulin, ER-Tracker dye, and MitoTracker dye. The resulting extracts were mixed thoroughly and imaged over time in a fluorinated ethylene propylene (FEP)–clad chamber by bright-field and fluorescence microscopy (Fig. 1B).

Fig. 1 Homogenized X. laevis egg extracts self-organize into cell-like compartments.

(A) Schematic diagram of the experimental procedure. (B) The design of the chamber used to image extracts. (C) Bright-field microscopy images showing that homogenized X. laevis egg cytoplasmic extracts spontaneously organized into cell-like compartments in a multimillimeter-scale field. Pattern formation dynamics in bright-field images as well as tubulin and ER channels are presented in movies S1 and S2. (D) Spatial organization of microtubules, ER, nuclei, and mitochondria in the cell-like compartments, visualized with added HiLyte 647–labeled porcine tubulin (T), ER-Tracker Red (E), GFP-NLS (N), and MitoTracker Red CMXRos (M). Pattern formation dynamics are presented in movie S3. The still images shown in (D) are from the last frame of movie S3, at 60.6 min after imaging began. (E) Confocal images of tubulin (top) and a merge of the ER and nuclear channels (bottom) during pattern formation. SiR-tubulin, a docetaxel derivative, was used to monitor microtubules because of its superior signal-to-background ratio. Control experiments verified that the same basic features revealed by SiR-tubulin, including asters and junctional complexes, were also seen with labeled tubulin (for example, fig. S5). Images are maximum intensity projections from five confocal z-slices 10 μm apart. A more detailed time course with channels for all probes is presented in fig. S2. (F) Quantification of the spatial distributions of labeled tubulin, ER-Tracker, MitoTracker, and GFP-NLS or mCherry-NLS signal across the compartments. For the top plot, data are from five separate experiments with 15 compartments quantified from each experiment. For the bottom plot, data are from two separate experiments, with 30 compartments quantified from one and 45 from the other. Shaded regions are the interdecile range. a.u., arbitrary units. (G) Quantification of the width of the border region. The left plot shows how border width changes with time. The green line shows the average border width between neighboring microtubule compartments and the red line the average border width between ER compartments. The error bars indicate standard errors (n = 9). The right plot shows the relationship between compartment size and the surrounding border width. Circles correspond to compartments containing at least one nucleus, and triangles correspond to compartments with no nuclei. Data are from a single representative experiment.

Over the course of ~30 min, the homogenized extract self-organized into a sheet of 300- to 400-μm compartments that resembled cells (Fig. 1C and movies S1 and S2). The added sperm nuclei migrated during compartment formation, and their initial localization did not entirely determine the final pattern (movie S2). Pattern formation did not appear to be coordinated by a propagating spatial signal, because the emergence of compartments across a large 2.6-mm field was approximately synchronous (movie S1). Because interphase-arrested extracts contain cycloheximide, which inhibits protein translation, the pattern formation appeared to be independent of genomic input from the added nuclei or translation of preexisting maternal mRNA.

The localization of the nuclei, microtubules, mitochondria, and ER was reminiscent of that in a typical interphase cell (11) (Fig. 1D). The dark gray regions observed in bright-field images contained microtubules, ER, and mitochondria (Fig. 1D). The center of the compartment was populated by a single nucleus or a cluster of nuclei, as indicated by the GFP-NLS or mCherry-NLS signal. Typically, the microtubules were denser around the periphery of the compartments and were organized in a wreath-like structure (Figs. 1, D and E; 2, A to D; 3A; and 4A; and movie S2) that closely resembled the microtubule structures seen in intact Xenopus embryos (4). The ER and mitochondria were denser near the centers of the compartments (Figs. 1, D to F, 3A, and 4A; fig. S3; and movie S2). Higher-magnification confocal microscopy corroborated these findings and allowed the evolving texture of the microtubules and ER to be better appreciated (Fig. 1E and fig. S2).

Fig. 2 The formation of cell-like compartments in egg extracts does not require added demembranated sperm nuclei.

(A) Time-lapse montage of cell-like compartment formation in an interphase egg extract with approximately 160 demembranated X. laevis sperm nuclei added per microliter of extract. (B) Time-lapse montage of compartment formation in an extract from the same experiment as (A) but with no sperm nuclei added. Dynamics of pattern formation in (A) and (B) are presented in movie S4. (A) and (B) share the scale bar located at the bottom of (B). “T + E + N” denotes a merge of the fluorescent tubulin (T), ER-Tracker (E), and mCherry-NLS (N) images. The individual fluorescent images are shown in fig. S3. (C) Confocal microscopy time-lapse images of microtubule dynamics during compartment formation in an interphase extract without added sperm nuclei. Images are maximum intensity projections from five confocal z-slices 10 μm apart. More-detailed time-lapse images comparing SiR-tubulin to labeled tubulin are presented in fig. S5. The time courses of extracts with and without added sperm are compared in movie S5. (D) Interphase egg extracts supplemented with different concentrations of sperm nuclei. Microtubules are shown in green, ER in red, and nuclei in cyan. The dynamics of pattern formation are presented in movie S6. (E) Size measurements of cell-like compartments from the same experiment as (D). Blue symbols indicate measured areas of individual nucleated compartments and red symbols non-nucleated ones. Areas are taken as the area of the tubulin compartment plus half of the associated border zone.

Fig. 3 Perturbing microtubule polymerization and cytoplasmic dynein activity disrupts pattern formation, whereas changes in actin polymerization have little effect.

(A) Nocodazole, a microtubule polymerization inhibitor, abolished the formation of cell-like compartments. Dynamics of the process are shown in movie S7. (B) Ciliobrevin D, an inhibitor of the minus-end–directed microtubule motor cytoplasmic dynein, led to a pattern with microtubules at the compartment boundaries and disorganized ER. mCh-NLS in (A) and (B) denotes mCherry-NLS. (C to E) Compartment formation in extracts for which actin polymerization was permitted (actin-intact) or inhibited. (C) Latrunculin A treatment abolished F-actin but had no effect on compartment formation. Images shown are at 54 min. BF, bright field. (D) F-actin became enriched at the compartment cortex upon extended incubation in an actin-intact extract. (E) After cortical enrichment of F-actin, a compartment further contracted and rolled on its side, revealing a cage-like actin cortex that contained an actin-rich nuclear cluster at its center. Data for (C) to (E) are from a single representative experiment. More details are presented in fig. S9.

Fig. 4 Cell-like compartments are capable of mitotic division.

(A) Time-lapse images of cell-like compartments performing a mitotic cell cycle in a sperm-supplemented cycling X. laevis egg extract. Formation of the mother compartments became apparent at 15 min and was completed by 23 min, when duplicated centrosomes could also be seen. A mitotic spindle subsequently formed at 33 min. The mother compartments completed their divisions at 40 min, each giving rise to two daughter compartments. Detailed dynamics of the process are shown in movie S9. (B) Time-lapse images of cell-like compartments undergoing five consecutive cell division cycles in a sperm-supplemented cycling extract. In this experiment, the daughter compartments formed after every division contained a single nucleus each. Bright-field images are shown in gray, and mCherry-NLS signal, which visualizes interphase nuclei, is shown in red. The corresponding time-lapse video is presented as movie S10.

The border zones between the cell-like compartments were largely free of microtubules and organelles (Fig. 1D and movies S2 and S3). Typically, the borders were ~40 μm wide, as measured by the distance between neighboring microtubule structures (Fig. 1G). Overall, the borders constituted 25 ± 3% (mean ± SD, from six independent experiments) of the total cross-sectional area. Organelles initially extended to near the tips of the microtubules but, upon extended incubation, moved toward the center, so that the distance between adjacent organelle compartments became greater than the distance between the outermost parts of the microtubule structures (quantified in Fig. 1G for microtubules versus ER). There was no apparent correlation between the size of a compartment and the size of the border surrounding it (Fig. 1G, right). The simplest hypothesis is that the border regions consist of cytosol—cytoplasm largely depleted of organelles and microtubules—although it is possible that they are enriched or depleted in particular cytosolic components. Sometimes a dark boundary line in the middle of the border regions could be seen in the bright-field images (Fig. 1D and fig. S1). This boundary line weakly stained with ER-Tracker, MitoTracker, and a plasma membrane stain (fig. S1). In addition, bright foci of microtubules could often be seen at the junctions between the cell-like units (Fig. 1E and fig. S2). Similar junctional complexes have been reported for extracts with intact actin filaments (5, 6), but in the experiments shown here, actin polymerization was suppressed by addition of cytochalasin B (vide infra).

The self-organization of the extract raised the question of which cytoplasmic constituents mediate the process. Sperm possess centrosomes, and the earliest step in the self-organization appeared to be the formation of large microtubule asters from the centrosomes of the added sperm (Fig. 1E, fig. S2, and movie S2). The asters were always detectable by the first frame of the video, 5 to 10 min after the extracts were taken off ice. We therefore tested whether the added sperm, with their accompanying centrosomes, were required for pattern formation. To our surprise, grossly normal cell-like compartments still formed in the absence of sperm (Fig. 2, A and B; fig. S3; and movie S4). Even though some maternal nuclear DNA might be expected to be present in the no-sperm extracts, in the vast majority of the self-organizing fields, no DNA was detected by Hoechst staining (fig. S4). Compared with sperm-supplemented extracts, extracts without sperm took longer to form cell-like compartments, and the initial dynamics of pattern formation were distinctly different—the ultimate wreath-like microtubule structures did not emerge from asters. Nevertheless, the final patterns and compartment appearances in these two types of extract were almost the same (Fig. 2, A and B, and movie S4). Therefore, the self-organization phenomenon does not require the presence of a centrosome or chromatin and does not need to begin from a large microtubule aster.

In the absence of added sperm, the first step in pattern formation appeared to be a local depletion of microtubules (Fig. 2C, fig. S5, and movie S5). The depleted zones began as circular, tubulin-depleted regions with a small bright focus of microtubules at their centers (Fig. 2C and fig. S5), and the depleted zones ultimately extended and fused to produce the final border region, with the small foci of microtubules persisting at junctions between the compartments. As mentioned above, similar junctional microtubule complexes could also be seen in extracts with added sperm (Fig. 1E and fig. S2), and they appeared at approximately the same time as they did in the no-sperm extracts (movie S5). They were detected by both labeled tubulin and SiR-tubulin, indicating that they are not an artifact of possible SiR-tubulin–induced microtubule stabilization (fig. S5).

At a high sperm concentration (800 nuclei/μl), some patterning still occurred, but discrete cell-like compartments failed to form (Fig. 2D and movie S6). At an intermediate concentration of sperm (160 nuclei/μl), almost all of the compartments contained nuclei and appeared relatively uniform in size, with an average cross-sectional area of 0.129 ± 0.044 mm2 (mean ± SD). However, at lower sperm concentrations (16 to 40 nuclei/μl), the global patterns were a mix of nucleated and non-nucleated cell-like compartments of different sizes. For each sperm concentration in this range, the nucleated compartments were larger than the non-nucleated ones, possibly because they formed earlier and thus had more time to grow (Fig. 2, D and E, and movie S6). In the complete absence of nuclei, the compartments were fairly uniform in size and were smaller than those seen in the presence of nuclei, with an average area of 0.060 ± 0.025 mm2 (Fig. 2E and movie S6). These findings suggest that competition may be at play in determining the final length scale of the compartments.

We tested various inhibitors for their effects on pattern formation (Table 1). Two treatments abolished pattern formation: the microtubule depolymerizer nocodazole (Fig. 3A and movie S7) and the adenosine triphosphate (ATP)–depleting enzyme apyrase (Table 1). Several other inhibitors changed the character of pattern formation without completely abolishing it. Ciliobrevin D, an inhibitor of cytoplasmic dynein, which is the main minus-end–directed microtubule motor protein, had several effects. It largely abolished the formation of normal nuclei from added sperm (Fig. 3B and movie S8), and the nuclei that formed did not become properly centered within the compartments (fig. S6). Ciliobrevin D also largely abolished the formation of discrete ER compartments (Fig. 3B). Moreover, microtubules in treated samples moved from within the compartments to the borders and became mixed with microtubules from neighboring compartments (Fig. 3B and movie S8). Despite these overall abnormalities, the length scale of the patterning remained close to normal. Thus, ATP and microtubule polymerization are essential for pattern formation, and the minus-end–directed motor dynein is required for partitioning the ER and for properly localizing the microtubules.

Table 1 The effect of perturbations on compartment formation.

ADP, adenosine diphosphate; AMP, adenosine monophosphate.

View this table:

Two inhibitors of the polo-like kinase Plk1 also affected pattern formation (Table 1 and fig. S7). At maximal concentrations, they abolished microtubule aster formation and resulted in a grossly abnormal pattern of ER localization (fig. S7). Inhibitors of aurora kinase A, Eg5, kinesin spindle protein, centromere-associated protein-E (CENP-E), and myosin II had little or no effect on pattern formation.

In standard Xenopus extracts, including the interphase extracts used here, the actin-polymerization inhibitor cytochalasin B is present (10, 12), and, therefore, actin polymerization is expected to be minimal. To further test for a possible dependence of pattern formation on actin polymerization, we prepared extracts in the absence of cytochalasin B. As assessed by the F-actin probe SiR-actin, actin filaments were present throughout the compartments and were concentrated around the nuclei and in junctional complexes (Fig. 3, C to E, and fig. S8). After extended incubation, F-actin became more concentrated at the periphery of the compartments, and the compartments contracted, with the intercompartment border zones expanding correspondingly (Fig. 3D). One example was found where a contracted compartment appeared to roll, revealing a cage-like structure of its actin (Fig. 3E and fig. S9). Treatment of actin-intact extracts with the actin-polymerization inhibitor latrunculin A abolished the F-actin staining and the late-stage contraction of the compartments but otherwise did not affect pattern formation (Fig. 3C and fig. S10). Thus, microfilaments are not required for the formation of cell-like compartments.

A special feature of Xenopus extracts is that they can be prepared with essentially no dilution. We tested how much dilution could be tolerated by the self-organizing cytoplasm. We mixed undiluted interphase extracts with various amounts of egg lysis buffer and monitored pattern formation in the diluted extracts. Nearly normal compartments still formed in extracts diluted to 70% of the original concentration (fig. S11). When cytoplasmic concentration was 50%, the pattern was noticeably abnormal, and when reduced to 30%, the compartments failed to form. Therefore, cell-like compartment formation requires a threshold concentration of the cytoplasm.

We wondered whether the cell-like compartments can carry out biological functions expected of an embryo. One of the hallmarks of the early embryo is its rapid reductive divisions, which allow the large egg to quickly develop into a 4000-cell embryo whose total volume is no greater than that of the original egg. We therefore tested whether the cell-like compartments could divide and reorganize if the extracts were allowed to progress through the cell cycle. We prepared a cycling egg extract (12) in the absence of cycloheximide, which permits cyclin synthesis and allows Cdk1 to be periodically activated and inactivated. The extract was supplemented with demembranated sperm nuclei (to provide centrosomes and DNA), plus ER-Tracker and fluorescently labeled tubulin, and the dynamics were tracked by bright-field and fluorescence microscopy (Fig. 4A and movie S9).

Initially, the cycling extract was in the first interphase after meiotic exit, and it appeared relatively homogeneous. By 15 min, cell-like compartments had begun to form, and by 23 min, they were indistinguishable from those seen in interphase-arrested extracts. The centrosomes separated at about this time, and, shortly thereafter, the texture of the ER changed, becoming coarser in appearance. At 33 min, well-defined mitotic spindles were visible. At this time, the ER-Tracker signal concentrated around the spindle poles, and a corresponding darker gray region was visible in the bright-field image. At 35 min, the division of the cell-like compartments began. Large microtubule asters emanated from the two daughter centrosomes, now pulled further away from each other, and a microtubule-enriched zone emerged near the bisecting point. ER appeared to be depleted from the same zone. By 40 min, the mother compartment had completely divided into two daughter compartments, each with the same pattern of microtubules and ER as had been present in the mother compartment before mitosis. Often the compartments underwent multiple consecutive cycles of such reductive divisions, yielding a population of much smaller offspring compartments (Fig. 4B and movie S10). These results demonstrate that the cell-like compartments are not only morphologically similar to cells but can also perform one of the most important functions of a living cell.

These experiments show that homogenized X. laevis egg cytoplasm can self-organize into spatially distinct compartments that resemble cells. The formation of these cell-like compartments does not require the presence of nuclei or centrosomes but requires energy, microtubule polymerization, cytoplasmic dynein function, and a threshold concentration of the cytoplasm. When supplemented with sperm, the cell-like compartments are capable of multiple rounds of division and reorganization. These results suggest that the cytoplasm can robustly generate the basic spatial organization of the cell and retains some of its distinctive functions.

Cytoplasmic ingredients self-organize on many different scales. Minimal sets of purified cytoplasmic macromolecules can self-organize into molecular assemblies such as the apoptosome (13) and the ribosome (14) and subcellular structures such as the centrosome (15) and ER (16). Cell-free extracts can autonomously assemble larger subcellular structures like the bipolar meiotic spindle (3) and the machinery of cytokinesis (4). The phenomena described here show that self-organization of the cytoplasm can occur on the cellular level in terms of both scale and complexity, underscoring the importance of nongenetic information inherent in the components of the cytoplasm.

Supplementary Materials

Materials and Methods

Figs. S1 to S11

References (1719)

Movies S1 to S10

References and Notes

Acknowledgments: We thank H. Funabiki and M. Dasso for providing the GST-GFP-NLS construct; T. Meyer, S. Pfeffer, T. Stearns, and J. Theriot for helpful discussions; O. Afanzar, Y. Chen, W. Huang, J. Kamenz, S. Liu, and C. Phong for comments on the manuscript; and the Stanford Cell Sciences Imaging Facility for use of their spinning disc confocal microscope. Funding: This work was supported by grants from the National Institutes of Health (R01 GM110564 and P50 GM107615). Author contributions: X.C. and J.E.F. jointly designed the studies, made the figures, and wrote the paper. X.C. carried out the experiments. Competing interests: None declared. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials.

Stay Connected to Science

Navigate This Article