Growth of the Mammalian Golgi Apparatus during Interphase

Abstract

During the cell cycle, genetic materials and organelles are duplicated to ensure that there is sufficient cellular content for daughter cells. While Golgi growth in interphase has been observed in lower eukaryotes, the elaborate ribbon structure of the mammalian Golgi apparatus has made it challenging to monitor. Here we demonstrate the growth of the mammalian Golgi apparatus in its protein content and volume during interphase. Through ultrastructural analyses, physical growth of the Golgi apparatus was revealed to occur by cisternal elongation of the individual Golgi stacks. By examining the timing and regulation of Golgi growth, we established that Golgi growth starts after passage through the cell growth checkpoint at late G₁ phase and continues in a manner highly correlated with cell size growth. Finally, by identifying S6 kinase 1 as a major player in Golgi growth, we revealed the coordination between cell size and Golgi growth via activation of the protein synthesis machinery in early interphase.

INTRODUCTION

The Golgi apparatus is a major glycosylation site of the cell and plays an essential role in the secretory pathway. During interphase, the mammalian Golgi apparatus is organized as a single, elaborate structure of interconnected stacks termed the Golgi ribbon (1). As the cell progresses through the cell cycle, the Golgi apparatus, like other organelles, is thought to double in size or number prior to equal partitioning between daughter cells (2). Although the literature on mammalian Golgi growth during interphase is limited, its inheritance during mitosis has been extensively studied (3).

While less is known about Golgi growth during interphase in mammalian cells, elegant work has been conducted in lower organisms. Protozoan parasites with easily traceable single Golgi stacks and short cell cycles have facilitated the elucidation of core Golgi growth mechanisms that may be utilized by all eukaryotes (4). During interphase, the single Golgi stack in Toxoplasma gondii grows laterally, followed by medial fission with each half partitioned to a daughter cell (5). Unlike T. gondii, Trypanosoma brucei forms a new Golgi stack at a distinct site during interphase, indicating de novo biogenesis, though some materials from the existing Golgi apparatus have been suggested to contribute to the new one (6). Contrary to the case for the parasites discussed, the Golgi apparatus in the budding yeast Pichia pastoris exists as several dispersed stacks in the cytoplasm. In this model system, individual Golgi stacks form de novo during interphase along with transitional endoplasmic reticulum (tER) sites, which are specialized ER domains involved in producing coat protein complex II (COPII) transport vesicles targeted to the Golgi apparatus (7). Finally, more quantitative assessments of Golgi growth have been performed in plant and insect cells. In Arabidopsis thaliana apical meristem cells, the numbers of dispersed Golgi stacks are similar in G₁- and S-phase cells whereas G₂-phase cells have double the number of Golgi stacks (8). In Drosophila melanogaster S2 cells, the dispersed Golgi stacks duplicate in content during G₁ and S phase, forming paired inheritance structures which separate during G₂ phase (9).

While it has been postulated that the mammalian Golgi apparatus must duplicate during interphase, this has not been conclusively demonstrated. Furthermore, it is unclear whether Golgi growth occurs continuously throughout interphase or at specific cell cycle phases (2). Evidence to date for Golgi growth in mammalian cells includes the near doubling of tER sites in cells at G₂ versus G₁ phase (10). Mammalian Golgi growth has also been linked to cell size, as enlarged cells stalled in S phase had increased numbers of mitotic Golgi fragments, indicating higher Golgi content, compared to those in untreated cells (11). The aim of our research was to determine conclusively whether the mammalian Golgi apparatus grows during interphase and to understand when and how this process may be regulated. Difficulties in visualizing mammalian Golgi growth have been attributed to its intricate and complex structure (9). Using flow cytometry, spinning-disk confocal microscopy, and transmission electron microscopy (TEM), we demonstrate the near doubling of the mammalian Golgi apparatus in its protein content and physical size during interphase. Through ultrastructural analyses, we reveal that the physical growth of the Golgi apparatus is achieved by cisternal elongation of the individual Golgi stacks. By stalling cells at various cell cycle phases, we show that continuous Golgi growth and cell size growth are initiated at late G₁ phase. Finally, our findings indicate that similar to overall cell size growth, Golgi growth is modulated by the “cell growth checkpoint” at late G₁ phase through the activities of S6 kinase 1 (S6K1). Together, we have followed the dynamic changes in the composition and structure of the mammalian Golgi apparatus and have elucidated the regulation of its growth during interphase.

MATERIALS AND METHODS

Cell culture, cell cycle synchronization, and transfection.

HeLa and NIH 3T3 cells (American Type Culture Collection) were cultured in Dulbecco modified Eagle medium (DMEM) (Wisent) containing 10% heat-inactivated fetal bovine serum (FBS) (Wisent), maintained at 37°C and supplied with 5% CO₂. To synchronize cells to early S phase, HeLa cells (one million per T-75 flask or 150,000 cells per 25-mm coverslip) were arrested using a double-thymidine method adapted from that described by Whitfield et al. (12). HeLa cells were exposed to 2 mM thymidine (Sigma-Aldrich) for 20 h, followed by release for 9 h by washing out thymidine. A second exposure to 2 mM thymidine was then carried out for 20 h prior to release. To acquire synchronous cells at various stages of the cell cycle, cells were collected at 2-h intervals for up to 20 h. For live cell sorting, HeLa cells were collected and incubated in medium with 5 μM Vybrant Dye Cycle (Life Technologies) for 30 min to label DNA. G₁- and G₂-phase cells were then sorted using a FACSAria III (BD Biosciences) according to the gating profile displayed in Fig. 3A. Cells were then spun onto coverslips coated with poly-l-lysine prior to analysis. For cell cycle stalling experiments, cells (one million per T-75 flask) were stalled in early G₁, late G₁, S, and G₂ phases through incubations in DMEM with 50 μM lovastatin (Sigma-Aldrich), 400 μM mimosine (Sigma-Aldrich), 2 mM thymidine (Sigma-Aldrich), or 5 μM aphidicolin (Sigma-Aldrich) and 10 μM RO3306 (Enzo Life Sciences), respectively, for 24 and 48 h prior to collection. To disconnect the Golgi ribbon into its individual stacks, cells were treated with 16.6 μM nocodazole (Sigma-Aldrich) for 30 min prior to collection. Measurement of nascent protein synthesis was performed with the Click-iT HPG kit (ThermoFisher Scientific) according to the manufacturer's instructions. For transient overexpression of S6K1-HA (a gift from John Blenis; Addgene plasmid 8984) (13), cells were transfected using FuGENE-HD (Promega) for 48 h prior to collection. For small interfering RNA (siRNA) knockdown, cells were transfected with S6K1 siRNA (Cell Signaling Technologies, catalog no. 6566) or control siRNA (Cell Signaling Technologies, catalog no. 6568), using Oligofectamine (Invitrogen) for 72 h, with fresh transfection reagents and medium replenished at 48 h posttransfection. To inhibit S6K1 activity, cells were treated with 10 μM PF-4708671 (Sigma-Aldrich) for 24 h prior to collection. Finally, to inhibit nascent protein synthesis, cells were treated with 71 μM cycloheximide (Sigma-Aldrich) for 24 h prior to further experimentation.

FIG 3.

TEM serial sections reveal Golgi growth during interphase. (A) HeLa cells were labeled with Vybrant Dye Cycle, a live DNA stain, followed by fluorescence-activated cell sorting. The gating profile shows the criteria to which cells in G₁ or G₂ phase were collected prior to being processed for TEM. (B) Serial sections images were obtained and manually aligned along the z axis. (C) Golgi structures were manually traced for each section in G₁- and G₂-phase cells. (D) 3D reconstruction of G₁- and G₂-phase Golgi apparatuses. Scale bars = 5 μm.

Cell staining and flow cytometry.

Cells were collected using 0.25% trypsin-EDTA (Wisent), followed by resuspension in 2% FBS in phosphate-buffered saline (PBS). Cells were then fixed with 80% ethanol (−20°C) on ice for 30 min. Afterwards, cells were blocked in 5% FBS in PBS for 30 min. All primary antibodies were applied for 1 h. To label the Golgi apparatus in HeLa cells, 1:200 GM130 antibody (BD Biosciences, catalog no. 610823) or 1:200 MannII antibody (Abcam, catalog no. ab12277) was used. To label the Golgi apparatus in NIH 3T3 cells, 1:200 GM130 antibody (Abcam, catalog no. ab52649) or 1:200 MannII antibody (Abcam, catalog no. ab12277) was used. To label S6K1, 1:200 S6K1 antibody (Cell Signaling Technology, catalog no. 9202) was used. To label the hemagglutinin (HA) tags of S6K1-HA-transfected cells, 1:200 HA tag antibody (Cell Signaling Technology, catalog no. 2367) was used. To label MannII in S6K1-HA transfected cells, 1:200 MannII antibody (EMD Millipore, catalog no. ab3712) was used. To label ceramide transfer protein (CERT) COL4A3BP, cells were first fixed with 4% paraformaldehyde (PFA), followed by permeabilization and blocking for 1 h with 10% FBS and 0.3 M glycine in 0.1% PBS–Tween 20. Anti-COL4A3BP antibody (1:200) (Abcam, catalog no. ab72536) was then applied for 24 h. All secondary staining were carried out with Cy2-, Cy3-, or Cy5-conjugated antibodies (Jackson ImmunoResearch) for 1 h. To label DNA, propidium iodide (PI) (Life Technologies) at 20 mg ml⁻¹ was applied to cells with RNase A (Sigma-Aldrich) at 10 mg ml⁻¹ in PBS for 30 min. Cells were then subject to flow cytometry using LSRFortessa (BD Biosciences). Flow cytometry analysis was carried out using FlowJo. To assess only single cells, clumped cells and debris were gated out using the forward scatter (FSC) versus side scatter (SSC) dot plot. To gate asynchronous cell populations into G₁, S, and G₂ phases based on DNA levels (PI fluorescence), FlowJo's built-in cell cycle fitting model was used. For all experiments involving cell cycle synchronizations, the asynchronous subsets that were unresponsive to synchronization drug treatments were gated out based on DNA levels. Finally, cells overexpressing S6K1-HA were gated from those nontransfected based on HA tag fluorescence. Golgi content and cell size were measured using Golgi protein fluorescence intensity and FSC, respectively, in all flow cytometry experiments. Unless otherwise specified, protein fluorescence intensity and cell size were normalized to the respective values in untreated G₁-phase cells to calculate the fold change.

Spinning-disk confocal microscopy and three-dimensional (3D) reconstruction.

Cells cultured on glass coverslips were fixed with 4% PFA in PBS for 20 min. Cells were then permeabilized with 0.1% Triton X-100 in PBS containing 100 mM glycine for 20 min. Cells were blocked with 5% FBS in PBS for 1 h, followed by primary antibody incubation for 1 h. Primary antibodies used to stain HeLa cells included 1:200 GM130 antibody (BD Biosciences, catalog no. 610823), 1:200 MannII antibody (Abcam, catalog no. ab12277), 1:200 MannII antibody (EMD Millipore, catalog no. ab3712), 1:200 HA tag antibody (Cell Signaling Technology, catalog no. 2367), and 1:200 S6K1 antibody (Cell Signaling Technology, catalog no. 9202). Labeling with 1:200 anti-COL4A3BP antibody (Abcam, catalog no. ab72536) was carried out as described above. Cells were then incubated in Cy2- or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch) for 1 h. DAPI (4′,6′-diamidino-2-phenylindole) was used to stain the nucleus before mounting. Single-slice images were acquired at 40× (numerical aperture [NA], 1.4) or 60× (NA, 1.4) using an inverted Zeiss AxioObserver Z1 epifluorescence microscope and Axiovision software. However, Z stacks at 0.2 μm per slice were acquired using a WaveFX-X1 spinning-disk confocal microscope (Quorum Technologies). A 63× objective (NA, 1.4) was used, and all imaging parameters, including exposure, laser intensity, and gain, were kept constant across samples in each trial. The images were captured with MetaMorph (Molecular Devices) and were then imported into Volocity image analysis (PerkinElmer) for deconvolution and Golgi volumetric analyses. To quantify the numbers of individual stacks in G₁- and G₂-phase cells following nocodazole treatment, cells were immunostained for GM130 and MannII and imaged by spinning-disk confocal microscopy as described above. The numbers of objects positive for GM130 and MannII staining were quantified using Volocity image analysis (PerkinElmer).

TEM analyses.

Cells on coverslips were fixed in 2% glutaraldehyde in 0.1 M Sorenson's phosphate buffer at pH 7.2 for 2 h. Cells were postfixed in 1% osmium tetroxide and 1.25% potassium ferrocyanide in sodium cacodylate buffer at room temperature for 45 min and stained for 30 min with 1% uranyl acetate in water. Dehydration and embedding into Epon resin were then carried out. Serial sections of approximately 90-nm thickness (light gold color) were collected once Golgi structures became apparent. Serial sections were collected on Formvar-coated copper grids, followed by staining with uranyl acetate and lead citrate. Sections were then imaged using a Tecnai 20 transmission electron microscope (FEI). In order to image the entire cross-section of the cell, multiple images were captured at each section, followed by 2D image stitching using DigitalMontage (Gatan). Using Reconstruct software (14), section images were aligned along the z axis. For each section, any cellular structures that resembled the Golgi stacked cisternae were traced, along with any vesicles and tubules at their immediate periphery. Quantification of the total Golgi volume across all sections was then determined using the Reconstruct software. Finally, to examine the ultrastructural changes in individual Golgi stacks, sections were imaged using an H-7500 transmission electron microscope (Hitachi). While the mammalian Golgi apparatus is composed of a single structure of interconnected stacks (1), we quantified any Golgi stacks with cisternae that were continuous from one dilated rim to the other as individual stacks. Using ImageJ (NIH), we quantified the individual Golgi stacks in terms of their numbers per cell, the number of cisternae per stack, the thicknesses of the stacks, and finally the lengths of the cisternae of each stack (see Fig. 4B). As several Golgi stacks were typically observed in each cell, the average stack number, stack thickness, and cisternal length of each cell were used for our statistical analysis.

RESULTS

Golgi protein levels increase during interphase.

The Golgi apparatus, like other organelles, is thought to double in size or number prior to equal partitioning between daughter cells (2). To analyze Golgi growth in mammalian cells, we began by measuring Golgi protein levels in HeLa cells at various cell cycle phases. GM130, a Golgi structural protein, and MannII, a Golgi enzyme, were used to represent proteins involved in maintaining Golgi integrity and Golgi function, respectively (15, 16). Using flow cytometry, asynchronous HeLa cell populations were divided into G₁, S, and G₂ phases based on DNA levels (PI fluorescence) as described in Materials and Methods. The Golgi protein levels at each cell cycle phase were assessed simultaneously (Fig. 1A). Fluorescence quantifications revealed that the mean GM130 and MannII levels increased significantly from G₁ to S and subsequently G₂ phase, peaking at an approximate 40% increase relative to those in G₁ phase (Fig. 1B). To confirm that our observations were not specific to cell type, we repeated the experiment with the NIH 3T3 fibroblast cell line and acquired similar results (Fig. 1C and D).

FIG 1.

Golgi protein levels increase during interphase. (A) Flow cytometry assay of HeLa cells displaying DNA and Golgi protein fluorescence after cell cycle fitting. GM130 and MannII levels at G₁, S, and G₂ phases were compared individually. (B) Quantifications of GM130 and MannII at G₁, S, and G₂ phases in HeLa cells. *, P < 0.05. Error bars represent standard errors of the means (SEM) from three independent experiments (n = 10,000 cells/experiment). (C) NIH 3T3 cells were similarly stained for DNA and Golgi proteins and analyzed based on cell cycle phase. (D) Quantifications of GM130 and MannII at G₁, S, and G₂ phases in NIH 3T3 cells. *, P < 0.05. Error bars represent SEM from three independent experiments (n = 10,000 cells/experiment). (E) Schematic for the double-thymidine S-phase synchronization and release method. (F) Flow cytometry assay plotting cell number against DNA levels, showing cells collected at various time points after release from double-thymidine block. (G) Flow cytometry assay plotting cell number against Golgi protein fluorescence, with time points grouped based on cell cycle phase. (H) Linear regression analyses revealed a significant increase in Golgi protein levels from early G₁ to late G₂ phase/mitosis for GM130 (r² = 0.77, F [1, 31] = 103.1, P < 0.0001) and MannII (r² = 0.74, F [1, 31] = 87.86, P < 0.0001). Error bars represent SEM from three independent experiments (n = 110,000 cells/experiment).

For both the HeLa and NIH 3T3 cell lines, neither GM130 nor MannII reached doubling levels by G₂ phase, compared to G₁. Since we measured only the mean Golgi protein levels at G₁ phase, any growth within G₁ phase would not be discerned. To address this limitation, we synchronized cells to allow measurement of the Golgi protein levels throughout G₁ phase (Fig. 1E to G). Using cells in early G₁ phase at 12 h postrelease as the normalization standard, we found that in HeLa cells both GM130 and MannII gradually increased by approximately 70 to 80% by the time cells were at late G₂ phase, compared to those in early G₁ (Fig. 1H). The two Golgi proteins increased at similar rates during interphase based on their linear regression slopes (GM130, 0.04412 ± 0.004346; MannII, 0.04493 ± 0.004793). These results suggest that in mammalian cells, the Golgi protein content increases throughout interphase, reaching near-doubling levels by late G₂, relative to that in early G₁ phase.

The Golgi cisternal volume increases during interphase.

Since Golgi proteins increased over the course of the cell cycle, we next examined whether the Golgi apparatus also increased physically in size. To this end, we measured the volume of the Golgi apparatus during interphase. HeLa cells at various stages of the cell cycle were collected after double-thymidine synchronization and release, as described above. The Golgi apparatus was then visualized by immunostaining GM130 and MannII, which are membrane-bound Golgi proteins localized to the cis cisternae and medial/trans cisternae, respectively (15, 16). Our spinning-disk confocal images showed an increasingly expanded Golgi apparatus as the cell cycle progressed (Fig. 2A). Higher-magnification images revealed that signals for GM130 and MannII originated from separate cisternal compartments (cis and medial/trans), with only a slight overlap. To assess volumetric changes, we collected z-stack images and reconstructed the Golgi apparatus in 3D, followed by measurements of its volume (Fig. 2B). Though we were able to measure the abundance of Golgi proteins in cells within mitosis using flow cytometry, we were unable to examine the Golgi volumes of these cells using microscopy due to the Golgi apparatus fragmentation at mitosis (17). By measuring only the volumes of intact Golgi apparatuses, the time point for “10 h postrelease” more appropriately represented cells “at late G₂ phase” rather than “at mitosis” as in our flow cytometry experiments. Using cells in early G₁ phase at 12 h postrelease as the normalization standard, we found that the Golgi volume reconstructed from GM130 and MannII fluorescence signals increased gradually by 120 to 150% by the time cells were at late G₂ phase, relative to those at early G₁ (Fig. 2C). These findings indicate that the mammalian Golgi apparatus also grows in its physical size throughout interphase. Comparing the levels of Golgi proteins (Fig. 1H) with Golgi volumes during interphase (Fig. 2C), we saw a high correlation between GM130 and MannII protein levels and the Golgi volumes reconstructed from their fluorescence signals (Fig. 2D). Golgi protein levels are hence predictive of the Golgi volume.

FIG 2.

The Golgi volume increases during interphase. HeLa cells at various stages of the cell cycle were collected after release from double-thymidine S-phase synchronization. Cells were immunostained for GM130 and MannII to label the Golgi apparatus and imaged using spinning-disk confocal microscopy. (A) Representative images of the Golgi apparatus at G₁, Ss and G₂ phases. The inset image shows that GM130 and MannII are localized to different Golgi cisternae with a slight overlap. Scale bar = 10 μm. (B) Serial z-stack images of the Golgi apparatus were collected and processed for 3D reconstruction, followed by measurements of their volumes. (C) Linear regression analyses revealed a significant increase in volume from early G₁ to late G₂ for the Golgi apparatus reconstructed based on GM130 (r² = 0.57, F [1, 31] = 40.77, P < 0.0001) and MannII (r² = 0.32, F [1, 31] = 14.66, P < 0.001) signals. Error bars represent SEM from three independent experiments (n = 550 cells/experiment). (D) By combining the results from Fig. 1H and 2C, Golgi protein levels are highly correlated with the reconstructed Golgi volumes derived from their respective fluorescence: GM130, r² = 0.84, F [1,9] = 46.97, P < 0.0001; MannII, r² = 0.61, F [1,9] = 14.29, P < 0.005. Error bars represent results from three independent experiments (n = 110,000 cells/experiment for Golgi protein levels and n = 550 cells/experiment for Golgi volumes).

Our fluorescent imaging results demonstrated that the mammalian Golgi apparatus increased volumetrically during interphase. However, the calculations of the Golgi volumes were based solely on fluorescent signals emitted by immunostained Golgi proteins. This carried an assumption that the probed proteins were equally distributed throughout the Golgi compartments they occupied. To confirm our fluorescence-based observations of Golgi size changes, we employed transmission electron microscopy (TEM) to examine Golgi membrane growth. Live asynchronous HeLa cells were sorted into G₁ and G₂ phases using Vybrant Dye Cycle prior to fixation, serial sectioning, and TEM processing (Fig. 3A). Two series of sections which traversed through a G₁-phase cell and a G₂-phase cell were imaged. Since sections varied in orientation during imaging, manual alignment along the z axis was carried out for both cells (Fig. 3B). The Golgi structure was then manually traced at each section (Fig. 3C). The Golgi apparatus in the G₂-phase cell had more elongated cisternae, and the Golgi stacks were densely concentrated together in the cell, compared to the case for the Golgi structures in in the G₁-phase cell. Complete G₁- and G₂-phase Golgi apparatuses were reconstructed based on these traces (Fig. 3D; see Movies S1 and S2 in the supplemental material). Quantitative analyses of these reconstructions confirmed that the volume of the Golgi apparatus at G₂ phase was indeed larger than that at G₁ phase (9.83 μm³ versus 4.15 μm³, respectively) (Fig. 3D), confirming our fluorescence-based observations.

Golgi cisternae elongate during interphase.

Having demonstrated that the Golgi apparatus grows in its volume using fluorescence and electron microscopy, we were next interested in understanding the expansion of the Golgi apparatus at an ultrastructural level. To begin, G₁- and G₂-phase cells were collected by sorting as described above prior to fixation and TEM processing. To best capture the bulk of the ribbon structure, we imaged sections closest to the middle of the cell. Initial inspection revealed a more prominent Golgi network in cells in G₂ phase than in those in G₁ (Fig. 4A). To quantify Golgi growth between G₁- and G₂-phase cells, several parameters of the Golgi apparatus were considered (Fig. 4B). The numbers of Golgi stacks observed in each cell as well as the numbers of cisternae that made up each stack were counted (Fig. 4C and D). The thicknesses and the lengths of the individual Golgi stacks were also measured (Fig. 4E and F). Between the cell populations, a significant difference was observed only in cisternal length with G₂-phase cells having individual Golgi stacks with cisternae that were 43% longer that those in G₁-phase cells (Fig. 4F). To further confirm that the number of Golgi stacks does not change between G₁ and G₂ phases, we collected G₁- and G₂-phase cells by sorting and administered nocodazole treatment for 30 min to disconnect the Golgi ribbon. Cells were then fixed, immunostained for GM130 and MannII, and imaged with spinning-disk confocal microscopy. Similar numbers of GM130- and MannII-labeled stacks were observed in G₁- and G₂-phase cells (Fig. 4G). Together these results suggest that the mammalian Golgi apparatus grows by cisternal elongation during interphase.

FIG 4.

Golgi cisternae elongate during interphase. (A) TEM images of individual Golgi stacks from cells sorted into G₁- and G₂-phase populations. Scale bars = 500 nm. (B) Schematic for the Golgi parameters that were quantified. (C to F) Quantifications of G₁- and G₂-phase cells for number of individual Golgi stacks per cell (C) and, for each individual Golgi stack, the number of cisternae (D), stack thickness (E), and cisternal length (F). *, P < 0.05. Error bars represent SEM (n = 100 cells). (G) Cell cycle-sorted cells were treated with nocodazole for 30 min, immunostained for GM130 and MannII, and imaged throughout the cell by spinning-disk confocal microscopy. Golgi-labeled stacks were quantified based on GM130 and MannII fluorescence. Error bars represent SEM (n = 60 cells).

CERT levels increase during interphase.

Our flow cytometry and electron microscopy results demonstrated the growth of the Golgi apparatus in its protein content and membranes, respectively. To further investigate the changes in the lipid content of the Golgi apparatus during interphase, we also examined the levels of ceramide transfer protein (CERT) (also called COL4A3BP), an enzyme involved in the transfer of ceramide from the ER to the Golgi apparatus (18). COL4A3BP immunostaining overlapped with the Golgi apparatus, labeled with GM130 (Fig. 5A), and also displayed cytoplasmic staining consistent with its localization in the cytosol and ER (18). The double-thymidine synchronization and release method was used to collect cells in G₁ and G₂ phases, as the COL4A3BP antibody staining protocol prohibited the use of PI for staining DNA. G₁- and G₂-phase cells were collected following synchronized release, followed by fixation with 4% PFA and permeabilization with Tween 20. G₁- and G₂-phase cells were then immunostained for COL4A3BP and subjected to flow cytometry. COL4A3BP levels significantly increased during interphase (Fig. 5B), suggesting that there is enrichment of Golgi lipids as cells progress through interphase.

FIG 5.

Ceramide transfer protein (CERT/COL4A3BP) levels increase during interphase. (A) Fluorescent images of cells immunostained for COL4A3BP and GM130. Scale bar = 20 μm. (B) Quantification of COL4A3BP levels in G₁- and G₂-phase cells collected following release from double-thymidine synchronization using flow cytometry. *, P < 0.05. Error bars represent three independent experiments (n = 10,000 cells/experiment).

Continuous Golgi growth is initiated in mid- to late G₁ phase.

We next sought to understand the dependence of Golgi growth on cell cycle progression. Cell cycle phases are governed by molecular “checkpoints” that determine whether the cell is fit to progress to the next phase (19). We hypothesized that if the cell cycle directed Golgi growth, Golgi growth would cease if cell cycle progression was experimentally halted. To stall cells in early G₁, late G₁, S, and G₂ phases, treatments of lovastatin, mimosine, thymidine, or aphidicolin, and RO3306 were administered, respectively, followed by the assessment of Golgi protein levels using flow cytometry (20, 21, 22, 23). Treatment durations of 24 and 48 h were chosen to ensure that all cells would have sufficient time to complete at least one cell cycle and be stalled in their respective phases. Prolonged stalling of cells in early G₁ phase did not lead to significant increases in total GM130 and MannII levels, compared to those in untreated cells in G₁ phase from an asynchronous population (Fig. 6A and B). Interestingly, cells stalled in late G₁ phase for 24 h had significantly higher GM130 and MannII levels than untreated cells in G₁ phase (Fig. 6C and D). When late G₁ phase was prolonged to 48 h, GM130 and MannII continuously increased to levels beyond the 24 h treatment and reached levels that were even higher than those in untreated cells in S phase (Fig. 6D). To determine whether this continuous increase in Golgi protein content occurred at later cell cycle phases, we stalled cells in S phase using thymidine (Fig. 6E). Our results showed that GM130 and MannII levels did not increase significantly when cells were stalled in S phase for 24 h (Fig. 6F). However, when S phase was prolonged to 48 h, GM130 and MannII levels were significantly higher than those in untreated cells in S and G₂ phases (Fig. 6F). To confirm that these observations were not specific to thymidine treatments, we also used another S-phase synchronization agent, aphidicolin (Fig. 6G). Cells stalled in S phase with aphidicolin had significantly higher GM130 and MannII levels after 24 h of treatment than untreated cells in S phase (Fig. 6H). Moreover, levels of GM130 and MannII were even higher than those in untreated cells in G₂ phase when S phase was extended to 48 h (Fig. 6H). Under the same treatment durations, cells stalled in S phase with aphidicolin had higher levels of GM130 and MannII than those stalled in S phase with thymidine. Aphidicolin inhibits S-phase progression by directly inhibiting DNA polymerase α (24). Thymidine, on the other hand, inhibits S-phase progression by feedback inhibition following the disruption in nucleotide pools (25). Based on their respective mechanisms of function, we speculate that aphidicolin is more rapid and potent in its inhibition effect than thymidine. Hence, given the same duration of incubation, cells treated with aphidicolin possibly achieved S-phase stalling much sooner than those treated with thymidine, consequently giving them a head start in continuous Golgi growth. Finally, we assessed the levels of Golgi proteins in cells stalled in G₂ phase for extended periods (Fig. 6I). GM130 levels were significantly higher than those in untreated cells in G₂ phase when the G₂ phase was prolonged for 48 h using RO3306 treatment (Fig. 6J). MannII levels, on the other hand, were already higher than those in untreated cells when G₂ phase was stalled for 24 h (Fig. 6J). Taking the results together, Golgi growth was limited in cells stalled in early G₁ phase, while continuous Golgi growth occurred during prolonged late G₁, S, and G₂ phases. This suggests that Golgi growth may be initiated in mid- to late G₁ phase and remains active until late G₂ phase.

FIG 6.

Continuous Golgi growth is initiated in mid- to late G₁ phase. (A) HeLa cells were stalled in early G₁ phase through lovastatin treatment. Flow cytometry profiles for PI fluorescence (DNA) revealed a synchronous populations of early-G₁-phase cells after both 24- and 48-h treatments (blue) compared to the asynchronous cell population (black outline). (B) Quantification of Golgi protein levels in cells stalled in early G₁ phase versus control cells. (C) Cells were stalled in late G₁ phase for 24 and 48 h through mimosine treatments. A synchronous population of late-G₁-phase cells (purple) was acquired. (D) Quantifications of Golgi protein levels in cells stalled in late G₁ phase versus control cells. (E) Cells were stalled in S phase for 24 and 48 h through thymidine treatments (red). (F) Quantifications of Golgi protein levels in cells stalled in S phase (thymidine) versus control cells. (G) Cells were also stalled in S phase for 24 and 48 h through aphidicolin treatments (orange). (H) Quantifications of Golgi protein levels in cells stalled in S phase (aphidicolin) versus control cells. (I) Cells were stalled in G₂ phase for 24 and 48 h by RO3306 treatment (green). (J) Quantification of Golgi protein levels in cells stalled in G₂-phase versus control cells. *, P < 0.05 for all data presented. Error bars represent SEM from three independent experiments (n = 30,000 cells/experiment).

Golgi growth is initiated concomitantly with cell size growth.

We next examined the relationship between Golgi growth and cell size. Since the Golgi apparatus either maintained its size or continuously grew when the cell cycle progression was impeded at various cell cycle phases, we assessed whether the size of these cells followed a similar trend. Using the forward scatter (FSC) of the cell as a measurement for cell size by flow cytometry (26), we compared our inhibitor-stalled cells with asynchronous populations. Cells stalled in early G₁ phase for 24 and 48 h showed FSC results similar to those for untreated cells in G₁ phase (Fig. 7A). Remarkably, cells stalled in late G₁ phase for 24 h had significantly higher FSC, and hence a larger cell size, than untreated cells in G₁ phase (Fig. 7B). Extending late G₁ phase for 48 h, however, did not give rise to further cell size growth, compared to that of cells stalled for 24 h (Fig. 7B). Following thymidine treatments, cells stalled in S phase for 24 h were no larger than untreated cells in S phase from an asynchronous population (Fig. 7C). However, a significant increase in cell size was observed in cells when S phase was extended to 48 h, compared to that of untreated cells in S phase (Fig. 7C). With aphidicolin treatments, cells stalled in S phase for 24 h were larger than the untreated cells in S phase (Fig. 7D). Extending aphidicolin treatment to 48 h gave rise to cells that were larger than cells treated for 24 h and those untreated in G₂ phase (Fig. 7D). Finally, no significant differences in cell size were observed in cells stalled in G₂ phase for 24 and 48 h, compared to untreated cells in G₂ phase (Fig. 7E). Evaluating our cell size results against Golgi protein levels in cell cycle-stalled cells (Fig. 6) revealed similar patterns of changes under most conditions. The Golgi content and cell size remained unchanged when cells were stalled at early G₁ phase, while both parameters continued to increase when cells were stalled at late G₁ and S phases. Furthermore, under the same treatment durations, cells stalled in S phase by aphidicolin treatment had sizes larger than those treated with thymidine. Discrepancies in these results are likely due to the mechanisms of action of aphidicolin and thymidine described in the previous section. Cells stalled in G₂ phase were the only exception, in which the Golgi content continued to increase while the cell size remained unchanged. Taken together, these results indicate that similar to Golgi growth, cell size growth is initiated in mid- to late G₁ phase, suggesting that both may be controlled by the same mechanism.

FIG 7.

Golgi growth is initiated concomitantly with cell size growth. Forward scatter (FSC) of the cell was measured to assess cell size using flow cytometry. Cell size quantifications of cells stalled in early G₁ phase (A), late G₁ phase (B), S phase (C and D), and G₂ phase (E) are shown. *, P < 0.05. Error bars represent SEM from three independent experiments (n = 30,000 cells/experiment).

Golgi size, cell size, and protein translation increase concurrently during interphase.

To eliminate the possibility of artifacts from drug treatments, we repeated our assessments of Golgi protein content and cell size in untreated asynchronous cells. Using flow cytometry, we measured the levels of GM130, MannII, DNA, and FSC. As cells progressed through the cell cycle, both Golgi protein content and cell size increased in a highly correlational manner (Fig. 8A). Thus, Golgi and cell size growth appear to be coupled during interphase in normally growing cells. Cell enlargement is a result of protein and rRNA accumulation, as the dry mass of the cell is dependent on the cell's protein content, while the amount of protein per cell is dependent on the number of rRNAs (27). To verify that protein translation changes during interphase, we employed the Click-iT HPG assay. The rate of nascent protein synthesis was measured based on the fluorescence of incorporated HPG in G₁-, S-, and G₂-phase cells. The levels of protein translation were highly correlated with both the cell size and the levels of GM130 and MannII during interphase (Fig. 8B). Thus, Golgi and cell size growth occur concurrently with increasing levels of total protein translation during interphase in naturally dividing cells.

FIG 8.

Golgi size, cell size, and overall protein translation increase concurrently during interphase. (A) Golgi protein levels (GM130 and MannII), DNA (PI), and cell size (FSC) were assessed in asynchronous populations. A correlational relationship between Golgi and cell size growth during interphase was discovered (Pearson's r = 0.67 for both GM130 and MannII levels versus cell size, P < 0.0001 for both comparisons, n = 20,000 cells/experiment). (B) The Click-iT HPG assay was used to measure the rate of nascent protein synthesis. Fluorescence of incorporated HPG was measured along with DNA levels, cell size, and GM130 or MannII levels using flow cytometry. Correlational relationships were observed between levels of protein translation and Golgi and cell size growth during interphase (Pearson's r = 0.73 for cell size/protein translation, 0.69 for GM130 levels/protein translation, and 0.77 for MannII/protein translation; P < 0.0001 for all comparisons; n = 20,000 cells/experiment).

Golgi growth is modulated by S6K1.

Mammalian Golgi growth is likely linked to a mechanism which initiates protein translation and cell size growth during mid- to late G₁ phase. Upon nutritional sensing at the cell growth checkpoint in late G₁ phase (28), activation of the phosphatidylinositol 3-kinase (PI3K)/AKT/mTOR pathway enhances global protein translation, which contributes to cell size growth (29). We examined the involvement of S6 kinase 1 (S6K1) in Golgi growth, as it is one of the most important downstream effectors in the PI3K/AKT/mTOR pathway. Overexpression of S6K1 increases protein translation and causes the enlargement of cells (30). To test for a role of this protein in Golgi growth using flow cytometry, we first confirmed S6K1 antibody specificity by immunostaining cells overexpressing S6K1 with an HA tag (S6K1-HA) (Fig. 9A). Transfected cells were identified by HA staining and showed intense cytosolic anti-S6K1 staining in transfected cells compared to the endogenous levels in nontransfected cells (Fig. 9A). Conversely, siRNA knockdown of SK61 caused a dramatic reduction in detectable endogenous S6K1 protein (Fig. 9B), confirming the specificity of this antibody. We next measured the levels of endogenous S6K1 in asynchronous cells. S6K1 levels increased significantly from G₁ to S and subsequently to G₂ phase (Fig. 9C). We also confirmed that the levels of endogenous S6K1 are indeed correlated with cell size and levels of protein translation during interphase (Fig. 9D). To see if the levels of S6K1 were modulated at the critical mid- to late G₁ phase that initiates Golgi and cell size growth, we measured the levels of S6K1 in cells experimentally stalled in early G₁ versus late G₁ phase (Fig. 9E). Cells stalled in early G₁ phase had levels of S6K1 similar to those in untreated G₁-phase cells in asynchronous populations (Fig. 9E). Cells stalled in late G₁ phase, however, had significantly higher S6K1 levels than both lovastatin-treated (early-G₁-phase) cells and untreated G₁-phase cells (Fig. 9E). These results revealed that S6K1 levels increase after mid- to late G₁ phase, which coincides with Golgi and cell size growth. Before testing whether Golgi growth is controlled by S6K1, we transiently overexpressed S6K1-HA and first monitored cell cycle progression and levels of protein translation using flow cytometry. Cells overexpressing S6K1-HA were distinguished from the nontransfected population by detection of the HA tag (Fig. 9F). The relative abundances of cells within G₁, S, and G₂ phases were similar in transfected and nontransfected cells, indicating normal cell cycle progression after transient S6K1 overexpression (Fig. 9G). S6K1-HA-transfected cells were indeed overexpressing S6K1, as our anti-S6K1 antibody detected increased S6K1 protein levels in transfected G₁-, S-, and G₂-phase cells relative to their nontransfected counterparts (Fig. 9H). Importantly, transfected cells overexpressing S6K1-HA exhibited increased levels of overall protein translation across all cells cycle phases compared to those in nontransfected cells (Fig. 9I).

FIG 9.

S6K1 protein levels, protein translation levels, and cell size are correlated. (A) HeLa cells transiently overexpressing S6K1-HA were immunostained to determine antibody specificity for S6K1. (B) HeLa cells were transfected with S6K1 siRNA or control siRNA, followed by immunostaining for S6K1 to further confirm antibody specificity of the S6K1 antibody. Scale bars = 20 μm. (C) S6K1 levels at the G₁, S, and G₂ phases of asynchronous cell populations were quantified using flow cytometry. (D) Endogenous S6K1 protein levels, cell size, and levels of protein translation were assessed in asynchronous cells. A correlational relationship was observed between S6K1 protein levels and cell size during interphase (Pearson's r = 0.38, P < 0.0001, n = 20,000 cells/experiment). S6K1 protein levels also correlated with levels of protein translation during interphase (Pearson's r = 0.36, P < 0.0001, n = 20,000 cells/experiment). (E) Cells were stalled at early G₁ and late G₁ phase through lovastatin and mimosine treatments, respectively. S6K1 levels in cells stalled in early G₁ and late G₁ phase were quantified relative to those in control cells. (F) Following S6K1-HA overexpression, cells were gated based on immunostained HA tag fluorescence. (G to I) Comparisons were made between S6K1-HA-overexpressing cells and nontransfected cells in terms of their cell cycle distributions (G), S6K1 protein levels (H), and levels of protein translation (I). *, P < 0.05. Error bars represent three independent experiments (n = 40,000 cells/experiment).

We next determined the impact of manipulating S6K1 on Golgi and cell size. Cells overexpressing S6K1-HA were gated from nontransfected cells (Fig. 10A), and cell size and Golgi protein levels were measured. S6K1-HA-transfected cells were larger across all cell cycle phases than nontransfected cells (Fig. 10B), in agreement with the results reported by Fingar et al. (30). GM130 and MannII protein levels were higher in S6K1-HA-transfected G₁-, S-, and G₂-phase cells than in nontransfected cells (Fig. 10C). We next looked at the effect of S6K1-HA overexpression on Golgi volume. Since we could not distinguish cell cycle phases using microscopy, we first confirmed that the entire transfected population showed increased overall levels of GM130 and MannII compared to those in nontransfected cells (Fig. 10D). Spinning-disk confocal microscopy imaging was then carried out to quantify the Golgi volume. Golgi volumes in S6K1-HA-positive and nontransfected cells were then calculated from 3D Golgi reconstructions of optical Z sections of GM130 and MannII fluorescence. Cells overexpressing S6K1-HA indeed had increased Golgi volumes compared to those in nontransfected cells (Fig. 10E).

FIG 10.

Golgi growth is modulated by S6K1 levels. (A) Transfected cells overexpressing S6K1-HA were gated from nontransfected cells based on immunostained HA tag fluorescence. (B and C) Both nontransfected and transfected cells were gated by cell cycle phase based on PI fluorescence (DNA) and quantified in terms of their cell size (B) and Golgi protein levels (GM130 and MannII) (C). (D) Overall levels of Golgi proteins in nontransfected and transfected cells were examined. *, P < 0.05. Error bars represent three independent experiments (n = 40,000 cells/experiment). (E) Spinning-disk confocal microscopy was used to examine the Golgi volume in nontransfected and transfected cells based on 3D Golgi reconstructions from GM130 and MannII fluorescence. *, P < 0.05. Error bars represent three independent experiments (n = 60 cells/experiment). (F) Golgi and cell size following S6K1 inhibitor and cycloheximide treatment. Cells were treated with the S6K1 inhibitor PF-4708671 and cycloheximide for 24 h, followed by the examination of Golgi protein levels and cell size using flow cytometry. *, P < 0.05. Error bars represent three independent experiments (n = 10,000 cells/experiment). (G) HeLa cells transfected with control siRNA or S6K1 siRNA were immunostained for S6K1 and GM130. Decreased S6K1 fluorescence confirmed S6K1 knockdown. Scale bar = 20 μm. (H) Spinning-disk confocal microscopy was used to image the Golgi apparatus in control and scrambled siRNA- or S6K1 siRNA-transfected cells. Golgi volumes were then measured and compared based on 3D Golgi reconstructions from GM130 fluorescence. *, P < 0.05. Error bars represent three independent experiments (n = 60 cells/experiment).

Finally we tested the impact of downregulating S6K1 activity and levels on Golgi and cell size. To start, we exposed cells to the S6K1 inhibitor PF-4708671 (31) for 24 h, followed by examination of Golgi protein levels and cell size. To directly block the protein translation machinery, we also treated cells with cycloheximide (32) for 24 h. S6K1 inhibitor treatment induced decreasing trends in Golgi protein levels and cell size across all cell cycle phases, though only reductions in MannII levels at S and G₂ phase were statistically significant (Fig. 10F). For cycloheximide-treated cells, cell size and MannII were significantly reduced across all cell cycle phases, while GM130 levels were significantly reduced only in S- and G₂-phase cells (Fig. 10F). We speculate that the discrepancy in these results may be due to the mechanisms and potency with which PF-4708671 and cycloheximide function. Specifically, PF-4708671 suppresses the phosphorylation of established S6K1 substrates, which consequently disrupts downstream signaling for protein translation (31), while cycloheximide directly blocks translation elongation (32). Additionally, the potentially different half-lives or stabilities of GM130 and MannII may also be a contributing factor. To directly evaluate the effects of reducing S6K1 protein levels on Golgi volume, S6K1 siRNA knockdown experiments were performed. Endogenous levels of S6K1 were noticeably lower in cells subjected to S6K1 siRNA knockdown than in control siRNA-treated cells (Fig. 10G). Confocal imaging and 3D Golgi reconstructions from GM130 staining revealed a significant reduction in Golgi volume following S6K1 siRNA knockdown compared to that in control siRNA-treated cells (Fig. 10H). Collectively, these results for interphase cells, cell cycle-stalled cells, and S6K1-HA-overexpressing cells suggest that the levels of S6K1 impact protein translation and direct both Golgi and cell size.

DISCUSSION

Golgi growth during interphase has been examined in multiple model organisms, though for mammalian systems these data have been scarce. To analyze organelle changes during interphase, criteria to distinguish cells based on their cell cycle phases must first be outlined. For the assessment of plant Golgi growth, cell cycle phases were determined by calculating the nuclear-to-cytoplasmic ratios in TEM slices (8). For insect Golgi growth, cells in G₁ phase were distinguished from those in G₂ phase by centrosome numbers, while bromodeoxyuridine (BrdU) fluorescence identified S-phase cells in confocal images (9). Though valuable insights on Golgi growth were revealed by these studies, these sampling methods can be subjective and laborious. We approached these limitations by utilizing flow cytometry, as it could rapidly and quantitatively measure multiple physical and fluorescent parameters on a cell-to-cell basis at a considerably high throughput. DNA labeling followed by flow cytometry and cell cycle fitting distinguished asynchronous cell populations into G₁-, S-, and G₂-phase cells, making this a considerably bias-free categorization strategy compared to those used in previous Golgi studies. Likewise, the measurements of DNA content by flow cytometry confirmed the cell cycle phases of cells released from double-thymidine synchronization and also those undergoing experimentally prolonged G₁, S, and G₂ phases. To examine the Golgi apparatus, we labeled two abundant matrix and enzyme proteins located in different cisternae with the underlying assumption that other Golgi proteins show the same trend. Using this approach, we captured, for the first time, the near doubling of the Golgi protein content from early G₁ to late G₂ phase in mammalian cells. To extend our findings to the Golgi apparatus volumetric and organizational changes during interphase progression, we next turned to microscopy.

Spinning-disk confocal imaging revealed a more expansive Golgi network, with its volume increasing up to 150% in late G₂-phase cells compared to early G₁-phase cells. Though the pronounced increase in Golgi volume could be a result of elevated cargo loads and membrane flexibilities, limitations associated with point spread function in conventional fluorescence microscopy may have overestimated the finer elaborations of the Golgi apparatus in cells at G₂ phase. Hence, we carried out serial sectioning followed by TEM analysis to confirm our fluorescence-based Golgi volumetric analyses. To understand the finer aspects of mammalian Golgi growth, we also analyzed the changes in Golgi membrane architecture during interphase by TEM. Most remarkably, the cisternal lengths of the individual Golgi stacks were 43% longer in G₂-phase cells than in those in G₁ phase, while the numbers of individual Golgi stacks remained similar. As additional support, the numbers of individual Golgi stacks were comparable in G₁- and G₂-phase cells following nocodazole treatment and confocal microscopy analysis. We did not observe a true doubling in cisternal length, possibly because our approach sorted out a mixed population of cells at various stages of the G₁ and G₂ phases. Nevertheless, this elongation of the Golgi cisternae is reminiscent of the lateral cisternal lengthening observed in the T. gondii parasite (5). Since plant and insect Golgi apparatuses grow by increasing their numbers of Golgi stacks (8, 9), the mammalian Golgi apparatus was proposed to grow in a similar fashion via ribbon extension by the attachment of additional Golgi stacks (33). While ribbon extension has been described in differentiating neurons and muscles (34, 35), our work suggests a different mechanism for mammalian Golgi growth during interphase. Since the majority of the cisternae that we observed and measured were continuous from one dilated rim to the other, with the total number of individual stacks unchanged from G₁ to G₂ phase, we conclude that the extension of the mammalian Golgi ribbon occurs by the cisternal elongation of the existing stacks rather than the addition of new stacks during interphase.

Overall Golgi size has been postulated to be maintained by a dynamic equilibrium between vesicle input in the cis cisternae and output from the trans cisternae (36). Inhibition of ER-to-Golgi trafficking causes the disintegration of the Golgi apparatus, while the inhibition of membrane budding from the Golgi apparatus enlarges the trans-Golgi network (TGN) but reduces the overall Golgi size (37, 38). Mammalian Golgi growth in interphase could result from a steady increase in protein translation, as supported by our HPG results. Additionally, though it is unknown whether the rate of output from the Golgi apparatus remains the same, increased trafficking of secretory cargo from additional tER sites in later stages of interphase (10) likely contributes to a larger Golgi apparatus. Our TEM observations of Golgi growth via cisternal elongation may have implications for the models of protein trafficking through the Golgi apparatus. An increase in tER sites during interphase likely results in enlarged ER-Golgi intermediate compartments (ERGICs). If we put this scenario into the vesicular transport model (39), increasing numbers of vesicles (containing both cargo and Golgi proteins) from the continuously growing ERGICs will enter the Golgi apparatus and cause the cis cisternae to enlarge, assuming that anterograde trafficking remains the same. We did not observe cisternal length asymmetries in our TEM analysis and instead saw cisternae with similar lengths across the entire stack in G₁- and G₂-phase cells. On the other hand, the cisternal maturation model entails that the cisternae themselves are the carriers for secretory cargo during trafficking (40). Therefore, an increasingly enlarged ERGIC during early interphase will sequentially mature into enlarged cisternae in a cis to trans direction, finally peeling off at the TGN for sorting. Since the overall length of the stack is dependent on the length of the ERGIC, which in turn depends on the number of tER sites, the coordinated cisternal lengthening we observed by TEM is best reconciled with this cisternal maturation model. Measurements of the ERGIC during interphase are required to definitively test this theory. While it is tempting to speculate about protein trafficking during Golgi growth, much work is needed to elucidate the precise means of passage. With technology advancements in superresolution imaging, fluorescent protein tags, and correlative microscopy, the tracking of protein movements through the Golgi apparatus during interphase will be an important area of future research.

Having observed the growth of the mammalian Golgi apparatus during interphase, our next aim was to investigate the mechanism driving this process. According to our cell cycle stalling experiments, Golgi growth is initiated at mid- to late G₁ phase along with cell size growth. In addition, S6K1 levels followed patterns similar to those for Golgi protein content and cell size under all cell cycle conditions tested. Also, transient overexpression of S6K1 alone caused an increase in Golgi protein content within larger cells. Finally, a role of a major protein translation regulator in modulating Golgi growth is supported by studies showing that gene expressions for Golgi proteins is unchanged throughout the cell cycle (12, 41). Putting the data together, we envision a model in which there is minimal Golgi and cell size growth at early G₁ phase, but upon passage through the cell growth checkpoint at late G₁ phase, S6K1 levels are elevated, and activated by the PI3K/AKT/mTOR pathway, triggering a global increase in protein translation. This consequently leads to an increase in Golgi protein content and in turn Golgi structural growth (Fig. 11). Supporting the “building blocks” expansion of the organelle, the overexpression of a single Golgi protein is sufficient to cause increased COPII assembly on the ER and subsequently growth of the Golgi apparatus (42). With Golgi growth beginning at the cell growth checkpoint and ending at the mitotic Golgi checkpoint at which the Golgi apparatus fragments (43), the duration of Golgi growth is limited to the time between late G₁ and late G₂ phase. A cell cycle timer described to control centrosome and nuclear growth (44) is hence in place for the Golgi apparatus. It is fascinating that this length of time is precisely sufficient for the Golgi apparatus to achieve near doubling, as we observed in our study. As a future direction, it will be interesting to assess the involvement of the PI3K/AKT/mTOR pathway in the growth of other complex membranous organelles during the mammalian cell cycle.

FIG 11.

Model of Golgi growth during interphase in mammalian cells. As the cell enters the cell cycle, minimal Golgi and cell size growth occurs during early interphase. Upon passage through the cell growth checkpoint at late G₁ phase, S6K1 levels are elevated, triggering an increase in global protein translation. As a result, the cell increases in its overall size, while the Golgi apparatus grows in its resident protein content and physical size. This concomitant growth then continues for the rest of interphase until mitotic Golgi fragmentation is initiated.

Organelle scaling entails that any dynamic alterations in cell size during development, division, or differentiation would require parallel changes in organelle abundance or architecture for maintenance of cell function (45). In other words, larger cells would have proportionately larger or more abundant organelles to accommodate the increased metabolic or organelle-specific functions required by a larger cell. As yet, scaling relationships have been revealed only in more structurally well-defined organelles such as the nucleus, vacuole, mitochondria, and centrosome, with limited examination of more dynamic and pleomorphic organelles such as the ER and Golgi apparatus. By examining cells undergoing interphase, we discovered that Golgi protein levels are predictive of Golgi volume. We also revealed a strong correlation between Golgi protein levels and cell size. Collectively these finding indicate that during interphase, a scaling relationship between Golgi and cell size exists such that any growth in cell size will give rise to corresponding growth in the Golgi's resident protein content and physical volume. We speculate that the functional capacity of the Golgi apparatus may also scale according to cell size during interphase, an important avenue of future research.