Physical plasticity of the nucleus in stem cell differentiation J. David Pajerowski*, Kris Noel Dahl���, Franklin L. Zhong*, Paul J. Sammak���, and Dennis E. Discher*�� *Molecular and Cell Biophysics Laboratory, 129 Towne Building, University of Pennsylvania, Philadelphia, PA 19104 ���Departments of Chemical and Biomedical Engineering, 5000 Forbes Avenue, Carnegie Mellon University, Pittsburgh, PA 15213 and ���Division of Developmental and Regenerative Medicine, University of Pittsburgh, Pittsburgh, PA 15213 Edited by L. B. Freund, Brown University, Providence, RI, and approved August 10, 2007 (received for review March 19, 2007) Cell differentiation in embryogenesis involves extensive changes in gene expression structural reorganization within the nucleus, including chromatin condensation and nucleoprotein immobiliza- tion. We hypothesized that nuclei in naive stem cells would therefore prove to be physically plastic and also more pliable than nuclei in differentiated cells. Micromanipulation methods indeed show that nuclei in human embryonic stem cells are highly de- formable and stiffen 6-fold through terminal differentiation, and that nuclei in human adult stem cells possess an intermediate stiffness and deform irreversibly. Because the nucleo-skeletal com- ponent Lamin A/C is not expressed in either type of stem cell, we knocked down Lamin A/C in human epithelial cells and measured a deformability similar to that of adult hematopoietic stem cells. Rheologically, lamin-deficient states prove to be the most fluid- like, especially within the first 10 sec of deformation. Nuclear distortions that persist longer than this are irreversible, and fluo- rescence-imaged microdeformation with photobleaching confirms that chromatin indeed flows, distends, and reorganizes while the lamina stretches. The rheological character of the nucleus is thus set largely by nucleoplasm/chromatin, whereas the extent of deformation is modulated by the lamina. chromatin remodeling cell mechanics nuclear plasticity Dto evelopmental ������plasticity������ generally refers to a cell���s ability modulate its gene expression (1) and often reflects changes in chromatin structure (2). Stem cell nuclei are therefore said to be more plastic than fully differentiated cells (3). Major differences are indeed found in chromatin conformation (4, 5), nuclear protein expression (6���8), and also DNA and histone modification (9). Differentiation further entails immobilization of some nucleoplasmic proteins and is said to functionally ������rigidify������ the genome, setting a relatively permanent preference for particular expression profiles (10, 11). Motivated by such remodeling, we sought to directly assess collective physical properties of the nucleus in differentiation. In an embryo, more deformable nuclei could confer a signif- icant advantage to less differentiated cells that must squeeze their way through developing tissues. Remarkably large nuclear deformations are seen, for example, during the migration of stem cell-like progenitor cells in brain tissue (Fig. 1A Upper) (12). Although such data illustrate deformability, we hypothesized that the nuclei of stem cells would prove more deformable, and it is indeed shown by micropipette aspiration that stem cell nuclei deform to a greater extent under a fixed stress compared with typical differentiated cell types. We go on to characterize the time-dependent deformation or creep under a constant applied stress and then also once the stress is removed from the nucleus. In the latter phase, we provide evidence of physical plasticity in which irreversible deformations persist without fracture. Single chromatin fibers in forced extension also exhibit plasticity, based perhaps on dissociation of histones from DNA (13���15). Our recent work on adult stem cells has additionally shown that multipotent cells are highly contractile and can generate signif- icant cytoskeletal stress (16), representing a potential driving force not only for cell motility (e.g., Fig. 1A) but also for nuclear remodeling. Here we assess some of the macromolecular deter- minants of nuclear deformability by fluorescence-imaged micro- deformation (Fig. 1A), combined with quantitative stress mea- surements, photobleaching, and knockdown of lamin proteins at the periphery of the nucleus. Results and Analysis Nuclei in Stem Cells Deform More Readily than in Differentiated Cells. To controllably assess the deformability of nuclei as stem cells differentiate, live cell microaspiration of nuclei within primary human embryonic stem cells (ESCs) was performed. Pluripotent cells (day-0) with fluorescently labeled nuclei were aspirated at constant pressure and compared with cells at days 2 and 6 after induction to neurectodermal differentiation (17). The ratio of nuclear projection length to cell membrane extension (Lnuc/Lcell) after 1 min of fixed pressure aspiration provides a first simple metric of relative nuclear deformability (Fig. 1B) this ratio is near unity for highly deformable nuclei, and it is low for very stiff nuclei. We find that ESC nuclei are highly deformable and that they stiffen over several days in culture, approaching a 6-fold higher relative stiffness of 0.15 that is typical of differentiated cells such as embryonic fibroblasts (Fig. 1C). An empirically fit exponential yields an effective time constant of 2.7 days for differentiation-associated stiffening of the nucleus relative to the cell body this timescale appears roughly consistent with global changes in nucleoprotein expression such as lamins in ESCs (7, 17). The limiting ratio for Lnuc/Lcell (i.e., nuclear:cytoplasmic stiffness) of 0.15 is also consistent with past measurements of relative nuclear compliance for other differentiated cells (18). Hematopoietic stem cells (HSCs) obtained from bone marrow are less multipotent than ESCs, but HSCs can still differentiate into all of the mature blood cell types (19) as well as a few solid tissue cells that include epithelial cells (20). Micropipette aspi- ration of human-HSC nuclei subjected to a step increase in pressure show progressive flow into the pipette, and these stem cell nuclei again exhibit greater deformation than fibroblast nuclei at the same stress (Fig. 2B and C): HSC nuclei deform more than twice as much after 200 sec of aspiration. A relative deformability for HSC nuclei of Lnuc/Lcell 0.5 identifies HSCs as roughly equivalent to ESCs at day 3 1 (Fig. 1C). On the basis of this physical measure, HSCs would be rightly viewed as partially differentiated. Author contributions: J.D.P., K.N.D., P.J.S., and D.E.D. designed research J.D.P., K.N.D., and F.L.Z. performed research P.J.S. contributed new reagents/analytic tools J.D.P., K.N.D., P.J.S., and D.E.D. analyzed data and J.D.P., K.N.D., P.J.S., and D.E.D. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Abbreviations: ESC, embryonic stem cell HSC, hematopoietic stem cell shRNA, small hairpin RNA. ��To whom correspondence should be addressed. E-mail: discher@seas.upenn.edu. This article contains supporting information online at www.pnas.org/cgi/content/full/ 0702576104/DC1. �� 2007 by The National Academy of Sciences of the USA www.pnas.org cgi doi 10.1073 pnas.0702576104 PNAS October 2, 2007 vol. 104 no. 40 15619���15624 DEVELOPMENTAL BIOLOGY ENGINEERING

We find that fits of Eq. 2 to the HSC results as well as to the Lamin A/C-knockdown results split into two phases below and above 10 sec [Fig. 2C and supporting information (SI) Fig. 9]. For consistency we also separate the remaining aspiration data into two regimes (see Fig. 5A and SI Tables 1 and 2). In the initial phase, HSC results fit with i 0.60, indicating greater fluidity than fibroblasts with i 0.41 (SI Table 1). Beyond 10 sec, the creep slows for both types of nuclei, and 0.2 indicates a more solid-like behavior. Relating these power law exponents to specific fractal dimensions is beyond the scope here and is a general challenge even in simple systems (27). Network of Lamin A/C Modulates Nuclear Compliance. Differentia- tion of ESCs entails changes in multiple structural components of the nucleus, including altered expression of Lamin proteins (A/C and B) (7), as well as changes in the extent of heterochro- matinization (10). Neither ESCs nor HSCs have detectable levels of Lamin A/C (28) (confirmed by immunofluorescence and blotting), and so we hypothesized that observed changes in nuclear rheology reflect altered Lamin A/C expression. To mimic this aspect of the stem cells with the human-derived epithelial cell line A549, we knocked down Lamin A/C by 85% with plasmid-based small hairpin RNA (shRNA) mediated interference (SI Fig. 10A). While immunofluorescence detected little to no change in either Lamin B expression or key chromatin modification markers of the cell���s differentiated state (SI Fig. 10B), the A549 Lamin A/C knockdowns exhibit a nuclear rheology that is indistinguishable from that of HSCs (Fig. 2C), whereas the rheological properties of untreated A549 cells prove very similar to those of committed fibroblasts (Fig. 2C, black and gray symbols). Changes in Lamin A/C are therefore sufficient to cause a 2-fold difference in stiffness (Fig. 2 B���D). Also, although the extent of deformation is dramatically altered by changes in Lamin A/C expression, the characteristic power law in creeping flow of nuclei is unaffected by Lamin A/C knock- down. Underlying rheological properties of nuclei might there- fore be lamin-independent. Our recent compliance measurements of the Xenopus oocyte (XO) nucleus provide a useful comparison because the mechan- ics of XO nuclei are dominated by the lamin network that surrounds dilute chromatin (29). We find here that JXO approx- imates the difference in nuclear compliance ( J) between HSCs and fibroblasts as well as between A549 controls and A549- Lamin-knockdowns (Fig. 2C Inset). A stiff lamina thus seems likely to explain nuclear stiffening in late differentiation. Chromatin Is Stiff When Condensed but Otherwise Flows. The re- markable compliance of nuclei in pluripotent ESCs (Fig. 1A) seems likely to reflect high accessibility of chromatin (10). Some terminally differentiated nuclei have, in contrast, highly con- densed chomatin that might be expected to be stiff (30). Divalent cations such as Ca2 and Mg2 will condense nuclei (22, 31), and so we introduced these ions into nuclei via selective permeabi- lization of the cell membrane by the detergent digitonin, which does not permeabilize the nuclear membrane. Condensation by divalent cations results in extremely stiff nuclei with small values for the creep compliance factors A and (Fig. 3). Similar effects were seen with the ionophore A23187, which more selectively permeabilizes the plasma membrane to Ca2 . The lamin-independent nature of nuclear flow (i.e., 0.2), combined with the strong effects of condensation, implicates chromatin in the physical plasticity of the nucleus. We therefore tracked chromatin with GFP-histones (H2B), as well as GFP Lamin A in TC7 epithelial cells. A dark stripe of GFP-H2B was made by laser photobleaching (FRAP/FIMD, fluorescence re- covery after photobleaching/fluorescence-imaged microdefor- mation). The parabolic flow profile of fluorescence proves consistent with a strong but sheared linkage between chromatin and the lamina (Fig. 4A and SI Fig. 11). Axial intensity profiles show that chromatin is progressively compacted toward the tip as the bleached stripe convects during aspiration. In some nuclei, chromatin bundles align and visibly extend within the pipette, indicative of physical remodeling (SI Fig. 11, asterisk). GFP-Lamin A stretches during aspiration (Fig. 4B), with widening of the bleached stripe and lengthening of the un- bleached tip area, which also thins as it stretches. The lack of recovery of the photobleached region provides direct evidence that large, sustained strains do not significantly increase Lamin A incorporation into the nuclear lamin over the experimental timescale. In unbleached nuclei, the observed gradient in density is similar to that of other elastic networks, particularly the red cell spectrin network (32) and the nuclear lamina of both HeLa cells and mouse embryonic fibroblasts (23). Labeling nucleoli with GFP-B23 (nucleophosmin) reveals slow nucleolar extrusion into the pipette during aspiration (Fig. 4C), as quantitated by the decreased distance from nucleolus- to-tip (L Lneol). Further kinematic analyses (SI Figs. 12 and 13) suggest nucleoli are stiffer than nucleoplasm, consistent perhaps with higher electron density in EM imaging as well as reduced un- treated salt- Condensed (+digitonin) �� A [kPa-1] �� 1.0 A [kPa-1] 0.2 exponent prefactor stiff Fig. 3. Divalent salts condense chromatin and stiffen the nucleus. Compared with untreated nuclei, quantification reveals decreases in both the creep prefactor, A, and the creep exponent, , due to cation-induced condensation. (Scale bar: 3 m.) L - Lneol Nucleoli follow B Lamina stretches Chromatin flows L ��� L l o e n [ �� ] m �� neol = 21 sec 0 50 100 0 5 time [sec] axial Position [ �� m] 0 5 10 15 20 y t i s n e t n I l a i x a ] . U . A [ 0 2 4 6 8 10 A C Fig. 4. Distinctive mechanics of nuclear components. (A) FRAP of GFP-H2B chromatin (pseudocolored yellow) reveals chromatin flow within the pipette. Chromatin compaction at the tip is quantified in the axial intensity profiles, and the arrow indicates an increase in intensity with compaction. (B) The lamina (green) is stably stretched into the pipette, and the arrow indicates a decrease in intensity with dilation. (C) Nucleoli slowly follow chromatin to- ward the nuclear tip, as GFP permeates the intact envelope (arrow into magnified view in bottom panel). (Scale bar: 3 m.) Pajerowski et al. PNAS October 2, 2007 vol. 104 no. 40 15621 DEVELOPMENTAL BIOLOGY ENGINEERING