Optical imaging of cell mass and growth dynamics

Abstract

Using novel interferometric quantitative phase microscopy methods, we demonstrate that the surface integral of the optical phase associated with live cells is invariant to cell water content. Thus, we provide an entirely noninvasive method to measure the nonaqueous content or "dry mass" of living cells. Given the extremely high stability of the interferometric microscope and the femtogram sensitivity of the method to changes in cellular dry mass, this new technique is not only ideal for quantifying cell growth but also reveals spatially resolved cellular and subcellular dynamics of living cells over many decades in a temporal scale. Specifically, we present quantitative histograms of individual cell mass characterizing the hypertrophic effect of high glucose in a mesangial cell model. In addition, we show that in an epithelial cell model observed for long periods of time, the mean squared displacement data reveal specific information about cellular and subcellular dynamics at various characteristic length and time scales. Overall, this study shows that interferometeric quantitative phase microscopy represents a noninvasive optical assay for monitoring cell growth, characterizing cellular motility, and investigating the subcellular motions of living cells.

Fig. 1.

A: optical phase shift φ is produced by a cell of refractive index n and thickness h surrounded by a medium of refractive index n₀. B: the same cell shrinking in a hypertonic solution assumes a new refractive index n′ and thickness h′, resulting in a new optical phase shift φ′. Thick arrows indicate the direction of the incident light beam. Cell projected areas A and A′ are indicated by dashed arrows on the substrate.

Fig. 2.

Refractive index n of mixtures of water (n₀ = 1.33) with nonaqueous material (n′ = 1.4) as a function of volume fraction f of the nonaqueous material as predicted by the Maxwell Garnett theory (○) and approximated by a linear function (solid line). Inset: relative error of the linear approximation. As the excellent fit shows, when n₀ − n′ is small, the n versus f dependence can be adequately modeled as a straight line.

Fig. 3.

A: representative snapshots of quantitative phase images of an isolated live HeLa cell at various time points (t) during volume regulation (the color bar indicates the phase in radians). B: for the same cell shown in A, the temporal variations of A′, φ′, calculated cell dry mass (M′), and volume (V′) are shown. Data were normalized to the initial values of A, φ, M, and V, respectively, to emphasize the relative magnitudes of change. As detailed in the text, cell dry mass stayed essentially constant, whereas projected area, optical phase shift, and cell volume changed as physiologically expected.

Fig. 4.

A: dry mass density image σ(x,y) of a monolayer of live HeLa cells obtained using Fourier phase microscopy. The color scale bar has units of pg/μm². 1–4, Cells 1–4. B: nanometer stability of the measurements demonstrated by the histogram of the pathlength SD corresponding to the pixels within a 15 × 15-μm² acellular area shown in the inset. The color scale bar of the inset indicates optical pathlength in units of nm. C: temporal evolution of the total dry mass content for each of the 4 representative cells shown in A. The solid lines for cells 1 and 4 indicate linear fits. Slopes (m) of the corresponding fits are shown, indicating more rapid growth for cell 4 compared with cell 1. In contrast, cell 2 was essentially static during these measurements, whereas cell 3 showed an oscillatory behavior of unknown origin.

Fig. 5.

Histograms of mesangial cell dry mass for identical cell monolayers grown at two different glucose levels. A: normal glucose levels show cell mass distribution under “physiological” conditions. B: high glucose levels (mimicking diabetes) show cellular hypertrophy as indicated by histogram broadening and shifting to the right.

Fig. 6.

A: mean squared displacements (MSD) versus time lag for cellular dry mass of a live HeLa cell monolayer at various spatial wavelengths (Λ), as indicated on the right. The dotted line indicates a region of linear dependence with exponent β approaching 1 with increasing spatial wavelengths. B: for the linear region highlighted in A, the average exponent of the power law fit β is plotted for various spatial wavelengths (Λ). For size scales of less than ∼6 μm, exponent β is <1, indicating subdiffusive motion. In contrast, β approaches 1 for larger spatial wavelengths, suggesting diffusive motion for larger structures such as whole cells and cell nuclei.