Mass measurements of polyploid lymphocytes reveal that growth is not size limited but depends strongly on cell cycle

Cell size is believed to influence cell growth and metabolism. Consistently, several studies have revealed that large cells have lower mass accumulation rates per unit mass (i.e. growth efficiency) than intermediate sized cells in the same population. Size-dependent growth is commonly attributed to transport limitations, such as increased diffusion timescales and decreased surface-to-volume ratio. However, separating cell size and cell cycle dependent growth is challenging. To decouple and quantify cell size and cell cycle dependent growth effects we monitor growth efficiency of freely proliferating and cycling polyploid mouse lymphocytes with high resolution. To achieve this, we develop large-channel suspended microchannel resonators that allow us to monitor mass of single cells ranging from 40 pg (small diploid lymphocyte) to over 4000 pg, with a resolution ranging from ~1% to ~0.05%. We find that mass increases exponentially with respect to time in early cell cycle but transitions to linear dependence during late S and G2 stages. This growth behavior repeats with every endomitotic cycle as cells grow in to polyploidy. Overall, growth efficiency changes 29% due to cell cycle. In contrast, growth efficiency did not change due to cell size over a 100-fold increase in cell mass during polyploidization. Consistently, growth efficiency remained constant when cell cycle was arrested in G2. Thus, cell cycle is a primary determinant of growth efficiency and increasing cell size does not impose transport limitations that decrease growth efficiency in cultured mammalian cells. Significance statement Cell size is believed to influence cell behavior through limited transport efficiency in larger cells, which could decrease the growth rate of large cells. However, this has not been experimentally investigated due to a lack of non-invasive, high-precision growth quantification methods suitable for measuring large cells. Here, we have engineered large versions of microfluidic mass sensors called suspended microchannel resonators in order to study the growth of single mammalian cells that range 100-fold in mass. This revealed that the absolute size of a cell does not impose strict transport or other limitations that would inhibit growth. In contrast to cell size, however, cell cycle has a relatively large influence on growth and our measurements allow us to decouple and quantify the growth effects caused by cell cycle and cell size.


Introduction
The extent to which cell cycle and cell size affect cell growth efficiency (growth rate per unit mass) is not known. In cultured and proliferating animal cells, mass increases exponentially with time, except in the largest cells which display decreased growth efficiency and proliferation rates (1-4). One explanation for the decreased growth in largest cells is that when cells grow beyond a certain size their growth becomes constrained by transport limitations (5)(6)(7)(8)(9)(10)(11)(12)(13)(14). Most notably, larger 35 cells have longer diffusion distances and lower surface-to-volume ratios, both of which could reduce the maximal rate at which large cells can transfer metabolites and information. Importantly, such transport limitations can exist even when cellular components scale isometrically with cell size. In a developmental setting, growth-influencing transport limitations could have a major impact on cell physiology, possibly explaining why most fast growing and proliferating cell types are small (<20 40 µm in diameter) (5,6,8). Transport limitations are also considered to result in allometric scaling of metabolism, a phenomenon where larger animals display lower metabolic and growth rates (10,11).
However, whether increasing cell size fundamentally imposes transport limitations that result in decreased growth efficiency is not known.
Alternatively, the non-linear correlation between cell mass and growth efficiency could 45 reflect cell cycle dependent growth, where each specific cell cycle stage has differential growth signaling and metabolism. This growth regulation can be entirely independent of cell size or can be coupled to size-dependent titration/dilution effects, where the concentration of cellular components is lowered as cells grow larger. Such dilution effects often depend on DNA content and, consequently, the dilution effects should be most prominent when cells grow during a cell cycle arrest (14)(15)(16)(17)(18). In 50 support of cell cycle dependent growth, cell cycle regulators are known to influence protein synthesis machinery (19-22), and growth rates in G1 have been shown to depend on cell size (23,24), presumably due to dilution effects. However, as cell cycle stage changes with cell size in most proliferating cell types, cell size and cell cycle effects must be decoupled to understand their individual contributions to cell growth.

55
To quantify the extent of cell size dependent growth, one would need to examine cells of vastly different sizes. Cultured cells maintain size homeostasis and display little size variability.
Yet, cell size can increase significantly when cells undergo repeated cell cycles in the absence of cell division (polyploidization). Polyploidization and the associated cellular hypertrophy is normal and critical in many tissues during development (5,12,25,26), and also commonly observed in cancers 60 (25,27). Although the physiological importance of polyploidy is well established, method limitations have prevented high-resolution single-cell measurements of growth in large polyploid cells. Several methods, including quantitative phase microscopy (3,28), fluorescence exclusion microscopy (23,29) and suspended microchannel resonators (SMRs) reported thus far (2,30), are capable of noninvasively quantifying single-cell growth rates of small cells (diameter range from <5 µm to 15 µm 65 in spherical cells). However, for the large cell sizes observed in polyploid cells, these techniques become imprecise or even infeasible, depending on the method. Here, we expand the analytical range of SMRs by engineering large-channel versions of the devices. We then use the large-channel SMRs together with previously published small-channel SMRs to monitor the growth of vastly different sized single cells (ranging from ~7 to ~32 µm in diameter) and to quantify the extent to which cell 70 size and cell cycle influence growth.

Results and discussion
The SMR is a microfluidic mass measurement device where a cell is flown through a vibrating cantilever and the change in the cantilever's vibration frequency is used to quantify the 75 buoyant mass of the cell. To overcome previous size-range limitations, we developed large-channel SMRs, which have 60 × 60 µm microfluidic channel inside the vibrating cantilever (Fig. 1A). These large-channel devices operate in the first vibration mode and utilize a new image-based hydrodynamic trapping approach to repeatedly measure the buoyant mass of a single particle/cell (SI Appendix, Fig. S1, Materials and Methods). The image-based hydrodynamic trapping provided 80 additional stability for long-term mass monitoring by allowing us to maintain a cell or a bead in a specified region within the microfluidic channels between measurements. Using polystyrene beads, we quantified each large-channel SMR mass measurement to have a resolution (standard deviation) ranging between 0.24 and 1.25 pg for particles ranging from 10 to 50 µm in diameter, respectively . This corresponds to a 85 measurement coefficient of variation range from 1.1 % to 0.05 %, respectively. When monitoring single-cell growth, we were able to carry out mass measurement every ~30 s without affecting cell viability, allowing us to average multiple mass measurements when monitoring mass changes that take place over longer time periods (Fig. 1C).
To validate that the large-channel SMRs provide data comparable to previous 15 × 20 90 µm SMRs (from here on referred to as small-channel SMRs), we measured single-cell buoyant mass accumulation rate (from here on referred to as growth rate) of mouse lymphocyte L1210 cells expressing the mAG-hGeminin cell cycle reporter (FUCCI). The interphase L1210 cell growth rates obtained from small and large-channel SMRs were similar ( Fig. 1D and E). It is known that growth rate, cell density and cell stiffness display dynamic changes in mitosis (19, 31, 32). As these changes 95 are unlikely to reflect cell size-dependent effects, we have excluded mitosis from all future analyses.
While the small-channel SMR has better measurement resolution than the large-channel SMR when measuring normal sized L1210 cells (stdev of 0.026 pg and 0.24 pg for a 10 µm diameter bead, respectively) (19), the large-channel SMR increases the maximum spherical cell volume that can be measured 64-fold. Importantly, the large-channel SMR is also able to monitor growth of a single cell 100 over multiple cell cycles (randomly following one of the daughter cells following each division, Fig.   1E), which is previously achieved by only a few cell size measurement methods (33).
We first studied the size-dependency of cell growth by monitoring unperturbed L1210 cells using the small-channel SMRs. Our data revealed that when cells are small (G1 and early Sstage cells), growth rate increases linearly with size (as cell cycle proceeds), indicative of exponential  Next, to examine cell cycle stage independent growth effects, we measured growth efficiency specifically in newborn G1 cells, in cells at G1/S transition and in cells at late G2. This revealed little to no correlation between growth efficiency and cell mass (Fig. 2C). Considering our measurement resolution (19), the lack of correlation is unlikely to be caused by noise in our measurement. Thus, these results suggest that cell size does not have a major influence on L1210 cell hypothesis, we induced polyploidy in L1210 cells using 50 nM Barasertib (also known as AZD1152-HQPA), a selective inhibitor of Aurora B, which is critical for cytokinesis (34,35). This resulted in 125 several endomitotic cycles where ploidy increased from 2N up to 128N (Fig. 3B) with corresponding increases in cellular hypertrophy ( Fig. 3C and D), suggesting that DNA-to-cell size ratio remained comparable to control cells. Importantly, the cells remained spherical with a single, multilobed nucleus (Fig. 3C). Prolonged drug treatments also resulted in cell death, which manifested in mass measurements as sudden transition to zero or negative growth (SI Appendix, Fig. S2A-C). These data 130 were excluded from our analysis (Materials and Methods).
When examining growth over larger size scales using the polyploid cells, mass increased exponentially over time ( Fig. 3D and SI Appendix, Fig. S2D). Remarkably, the non-linear growth efficiency behavior that was observed in control cells (Fig. 2B) repeated in every successive cell cycle during polyploidization independently of cell size (Fig. 3E). This non-linear growth 135 behavior cannot be explained by the DNA-to-cell size ratio alone, as growth efficiency decreased towards the end of each cell cycle, but started to increase immediately following endomitosis before the subsequent S-stage. Furthermore, the low growth efficiency in newborn G1 cells (Fig. 2B and C), cannot reflect too small absolute size, as polyploid G1 cells display similarly low growth efficiency.
To validate that the observed growth behavior cannot be attributed to drug specific effects, we 140 induced polyploidy using an alternative cytokinesis inhibitor, 10 µM H-1152, which targets the Rhokinase (ROCK) (36). This resulted in similar growth behavior as Aurora B inhibition (SI Appendix, Overall, we quantified growth efficiency for L1210 lymphocytes over an approximately 100-fold mass range spanning from 40 pg to 4000 pg. In spherical L1210 cells this corresponds to a 145 diameter range from <7 µm to >32 µm resulting in estimated 4.5-fold reduction in surface-to-volume scaling. This size range covers most proliferating cell types in the human body. Unlike cells in vivo, cultured cells are constantly selected for the highest growth rate, allowing us to assume that the measured growth rates reflect maximal growth rates possible for the cells. Size scaling typically follows a power law = , where is the observable biological feature, is a normalization 150 constant, is the mass of the organisms (or a cell), and is the scaling exponent which typically has values close to ¾ when studying metabolic rate (10, 11). We observed a minor decrease in growth efficiency in largest cells when plotting data obtained across multiple measurement systems and conditions (Fig. 3E). We therefore quantified size-dependent growth and the allometric scaling exponent from our growth rate data using only Barasertib-treated cells monitored with the large- This corresponds to each doubling of cell mass changing growth efficiency by -0.1 ± 1.1% (mean ± s.e.m.).

160
In contrast to cell size, cell cycle displays a strong influence over cell growth efficiency.
To validate that cell cycle progression causes the oscillating growth behavior within each cell cycle, we arrested L1210 cells to G2 stage with 2 µM RO-3306, an inhibitor of cyclin-dependent kinase 1 (CDK1) (38) (SI Appendix, Fig. S3A and B). Prolonged RO-3306 treatment resulted in cell death, and to avoid this toxicity, we only analyzed growth for the first 40 pg increase (corresponding to a typical mass increase in a cell cycle) from the normal mitotic size. This revealed that the decrease in growth efficiency that was observed in large control cells stopped as cells were arrested in G2 and the growth efficiency remained constant for G2 arrested cells even as their sizes increased (SI Appendix, Fig. S3C and D). Thus, as suggested by previous work in budding yeast (39), our results show that cell cycle has a major influence on mammalian cell growth efficiency. We quantified this 170 cell cycle dependent growth to be 29 ± 3% (mean ± s.e.m.) of the average growth efficiency in untreated L1210 cells. In addition, the steady growth efficiency observed in G2 arrested cells validates that increasing cell size does not automatically result in decreasing growth efficiency even in a model where DNA content does not scale with cell size. Furthermore, these results suggest that G2 growth efficiency is not regulated by dilution of components produced in earlier cell cycle stages.

175
Finally, using the polyploidy cell data collected using the large-channel SMR, we also analyzed how cell size increase and cell cycle duration scale with cellular hypertrophy and the associated polyploidy. This revealed that with each successive endomitotic cycle the L1210 cells approximately doubled their size independently of the cell size at the start of that cell cycle (Fig. 3G,   top). Cell cycle duration also remained constant regardless of cell size (Fig. 3G, bottom). This 180 suggests that massive cellular hypertrophy and the associated polyploidy do not interfere with the mechanism(s) ensuring that cells double their size during each cell cycle.
In conclusion, increasing cell size does not impose strict transport limitations that would lower growth efficiency in cultured mammalian cells. This conclusion was reached when observing freely proliferating lymphocytes in specific cell cycle stages (Fig. 2C), when examining cells across Methodologically, we anticipate that the large channel SMRs will have important uses outside this study. The ability to monitor the mass of unlabeled large samples will enable growth (2,19), drug-response (41) and nutrient uptake (42) studies in various models. These include extremely large single-cells such as adipocytes or megakaryocytes, as well as individual organoids or tumor 205 spheroids, where adherent cell mass accumulation can now be monitored in a preserved 3D microenvironment.

210
Please see SI Appendix, Materials and Methods.   Data represents individual mass measurements without any averaging. At each division, one daughter cell is discarded. The mAG-Geminin signal (green) detection was only carried out in small-channel devices, and its increase indicates G1/S transition and loss indicates metaphase/anaphase transition (blue arrows).