Tunable Single-Cell Extraction for Molecular Analyses

Resource

Tunable Single-Cell Extraction for Molecular Analyses Graphical Abstract

Authors Orane Guillaume-Gentil, Rashel V. Grindberg, Romain Kooger, ..., Martin Pilhofer, Tomaso Zambelli, Julia A. Vorholt

Correspondence [email protected] (O.G.-G.), [email protected] (J.A.V.)

In Brief Extraction of sub-picoliter samples of nucleoplasm and cytoplasm from live cells allows cellular heterogeneity to be assessed without killing the cells.

Highlights d

Sampling of cytoplasmic and nucleoplasmic fractions from single live cells

d

Real-time monitoring of the cellular extraction at a subpicoliter resolution

d

Extract dispensing strategies adaptable to a broad range of analytical methods

d

Method allows for ready assessment of enzymatic activities and transcriptional readouts

Guillaume-Gentil et al., 2016, Cell 166, 506–516 July 14, 2016 ª 2016 Elsevier Inc. http://dx.doi.org/10.1016/j.cell.2016.06.025

Resource Tunable Single-Cell Extraction for Molecular Analyses Orane Guillaume-Gentil,1,* Rashel V. Grindberg,1 Romain Kooger,2 Livie Dorwling-Carter,3 Vincent Martinez,3 Dario Ossola,3 Martin Pilhofer,2 Tomaso Zambelli,3 and Julia A. Vorholt1,* 1Department

of Biology, Institute of Microbiology, ETH Zurich, 8093 Zurich, Switzerland of Biology, Institute of Molecular Biology and Biophysics, ETH Zurich, 8093 Zurich, Switzerland 3Laboratory of Biosensors and Bioelectronics, Institute for Biomedical Engineering, ETH Zurich, 8092 Zurich, Switzerland *Correspondence: [email protected] (O.G.-G.), [email protected] (J.A.V.) http://dx.doi.org/10.1016/j.cell.2016.06.025 2Department

SUMMARY

Because of cellular heterogeneity, the analysis of endogenous molecules from single cells is of significant interest and has major implications. While micromanipulation or cell sorting followed by cell lysis is already used for subsequent molecular examinations, approaches to directly extract the content of living cells remain a challenging but promising alternative to achieving non-destructive sampling and cell-context preservation. Here, we demonstrate the quantitative extraction from single cells with spatiotemporal control using fluidic force microscopy. We further present a comprehensive analysis of the soluble molecules withdrawn from the cytoplasm or the nucleus, including the detection of enzyme activities and transcript abundances. This approach has uncovered the ability of cells to withstand extraction of up to several picoliters and opens opportunities to study cellular dynamics and cell-cell communication under physiological conditions at the single-cell level. INTRODUCTION Recent technological advances have enabled biological studies to be scaled down to the single-cell level, thus revealing the heterogeneity in cell populations with implications in nearly all fields of biology and medicine. While tissues of multicellular organisms are made of multiple cells that originally share the same genetic information, the individual cells are unique in their structure, composition, and functionality. An unexpected level of somatic genomic variations has been shown in both normal and diseased tissues (Cai et al., 2014; O’Huallachain et al., 2012). Differential gene expression gives rise to a diversity of cell phenotypes, shaping individual cells toward highly specialized functions. In addition, the stochasticity of intracellular processes, as well as variations in the surrounding environment, further generates differential cell behaviors, even in apparently homogenous cell populations. Therefore, molecular analyses at the single-cell level are critical for dissecting the complexity of heterogeneous tissues, characterizing pathological conditions, or investigating 506 Cell 166, 506–516, July 14, 2016 ª 2016 Elsevier Inc.

the biological processes and cellular responses to perturbations, without the limitations of population averaging (Schmid et al., 2010; Wang and Bodovitz, 2010). While molecular analyses at the single-cell resolution are highly attractive, they present several challenges. They require the ability to manipulate micron-sized individual cells and to handle volumes in the picoliter range (reported HeLa cell volumes range between 1.2 and 4.3 pl [Zhao et al., 2008]), the detection of minute amounts and a broad variety of analytes, and the minimization of perturbations on the cell and the physiological process under investigation. In current single-cell studies, the cellular molecules are commonly retrieved through the consecutive isolation and lysis of individual cells. Fluorescence-activated cell sorting (FACS)-based methods and microfluidic devices are the predominant platforms that enable single cells to be handled and analyzed, and they provide highthroughput methods to examine numerous individual cells in parallel (Kovarik and Allbritton, 2011; Lo and Yao, 2015; Wu and Singh, 2012). However, both methods require the removal of the cell from its original environment, and doing so causes the loss of contextual information and may generate physiological perturbations. A recently developed microfluidic probe has the potential to partially address these issues, enabling the selective lysis of a cell in its tissue culture environment, followed by lysate aspiration into micro-chambers for single-cell enzyme activity measurements (Sarkar et al., 2014). However, gaining access to cellular contents through lysis of the cell under investigation limits single-cell studies to post-mortem analyses, and the lysis process itself may alter the cellular analytes. To overcome these limitations, a few studies have explored the insertion of minimally invasive sampling devices directly inside a cell for extraction. These approaches allowed direct access to the native intracellular fluid. Atomic force microscopy (AFM) tips have been inserted inside living cells without compromising their viability, enabling the analysis of specific mRNAs adsorbed onto the tips (Nawarathna et al., 2009; Osada et al., 2003). Micropipettes have been introduced inside living cells to aspirate nanoliters of cytoplasmic samples and analyze their mRNA (Van Gelder et al., 1990) or metabolite (Saha-Shah et al., 2015) content. However, micropipettes are relatively invasive because they impact cell survival and limit studies to large cells. A recent report presented the uptake of femtoliter samples of cytoplasm into a nanopipette using electrowetting. The technology enabled analyses of mRNAs and mitochondrial DNA,

Figure 1. FluidFM-Based Approach

Extraction

(A) Schematic of the FluidFM-based extraction procedure. The micro-channeled FluidFM cantilever is monitored by the AFM laser (red) and allows for handling femtoliter liquid volumes with nanometric spatial precision and piconewton force resolution. The FluidFM probe is aligned on top of the desired location of the cell and inserted inside the cell by force spectroscopy with a preset force. Once the force is attained, Z-piezo regulation allows maintenance of the tip inside the cell at constant preset force. During this time, underpressure is applied through the microchannel to flow the cellular content into the probe. The probe is then lifted off of the cell. The extract is then released onto a suitable substrate for further analysis by applying overpressure through the channel. The FluidFM setup is mounted onto an inverted optical microscope (not shown), which allows visual inspection of the entire process. (B) Scanning electron micrographs of a FluidFM probe for extraction. The probe consists of a hollow cantilever with a hollow pyramidal tip. A triangular aperture (400 nm in height and in base length) was milled by a focused ion beam on the front face of the pyramid, close to the apex. See also Figure S2.

while preserving cell viability (Actis et al., 2014). These few studies have shown that fluidic extraction holds great promise for single-cell molecular analyses, but the potential of extraction devices has not yet been exploited. Using fluidic force microscopy (FluidFM), we recently established an effective protocol for the minimally invasive insertion, stabilization, and withdrawal of the tip of a microchanneled probe inside a cell for solute injection (Guillaume-Gentil et al., 2013, 2014; Meister et al., 2009). Because the FluidFM is mounted on top of an inverted optical microscope, the cells of interest can be readily observed during manipulation in real time. The process relies on AFM to accurately and gently drive the probe through the cell membrane. In this study, we developed and explored a platform for the extraction of tunable amounts of intracellular fluid from live cells and the delivery of the withdrawn cell contents to a variety of analytical techniques. RESULTS Development of a Volume-Controlled Single-Cell Extraction Approach The insertion of miniaturized devices, followed by extraction, provides an attractive means by which to retrieve the intracellular contents of single cells for downstream analyses. This method circumvents the need to isolate and lyse the cell prior to analyte recovery. In this study, we developed a generic and versatile method for such extraction experiments using FluidFM, which addresses selected adherent live cells within tissue cultures (Figure 1A). The approach consists of inserting the tip of the FluidFM probe into the cell and filling the probe through the application of negative pressure, followed by withdrawal from the cell, local dispensing, and molecular analyses.

We used FluidFM probes that featured a pyramidal tip. Triangular apertures on the front pyramid side close to the apex were milled using a focused ion beam (Figure 1B). We chose an opening size of 400 nm to allow most cellular components, such as RNA, peptides, and proteins, to pass through, while still ensuring complete aperture insertion inside the cell. We first established a coating protocol based on surface siliconization to prevent adsorption of cellular compounds in the microchannel surface and to allow liquid flow within the channel (see Figure S1). Considering the low amount of given analytes within a single cell, it was also highly critical to avoid dilution of the extracted material. To prevent rarefaction, the FluidFM probe was prefilled with mineral oil, which served as an immiscible phase to confine the extract to the front of the probe. Cultures of HeLa cells expressing green fluorescent protein (GFP) were used to demonstrate cellular extraction in a first set of experiments. Because intracellular GFP diffuses through the nuclear pores and distributes relatively homogenously in the cell’s cytoplasm and nucleus (Wachsmuth et al., 2003), it represents a convenient marker to follow the withdrawal of intracellular content by fluorescence microscopy from both compartments in real time. After aligning the pyramidal tip of the FluidFM probe on top of the desired cell compartment under optical control, force spectroscopy was initiated to drive the probe toward the cell, through its membrane, and further inside the cell, with continuous force monitoring. The probe was then stabilized inside the targeted cell compartment, resulting in a tight seal between the inserted pyramidal tip and the cell membrane (Guillaume-Gentil et al., 2013). Once the tip was inserted into the cell, underpressure was applied through the microchannel to extract the cellular content. The semitransparency of the cantilevers allowed us to visualize the extracted volume in the FluidFM Cell 166, 506–516, July 14, 2016 507

probe, with a clear boundary between the aqueous cellular extract and the pre-filled oil in transmitted light mode (see Figures 2A and 2B; Movie S1). After extraction, the observation of the cell by light microscopy provided a first assessment on the impact of the manipulation on the cell (Figure 2A and 2B). Following extraction both from the cytoplasm and from the nucleus, a significant decrease in GFP intensity was detected, indicating the extraction of the reporter protein (Figure 2C). As expected, we did not observe significant loss in fluorescence after extraction from the cytoplasm of cells containing labeled mitochondria or from the nucleus of cells expressing mCherry-tagged histones (Kuipers et al., 2011) (H2B) or after DNA staining (Figure 2C). Altogether, the results indicated that soluble proteins were readily extracted in contrast to molecules assembled into large macromolecular structures and small organelles. These results are in agreement with the size of the FluidFM probe aperture acting as a selective molecular sieve during the suction of the intracellular milieu. Cell Compartment Selective Extraction As FluidFM probes can be inserted at choice either into the cytoplasm or into the nucleus of a targeted cell (Guillaume-Gentil et al., 2013; Meister et al., 2009), we then assessed whether cytoplasmic and nuclear content can be selectively retrieved. Cell nuclei were labeled with two different fluorescent markers, and the fluorescence loss following extraction from either the nucleus or the cytoplasm was quantified (Figure 2D). The first nuclear marker consisted of a fluorescent protein (mRuby) tagged with a nuclear localization sequence (NLS). While mRuby has a size similar to the GFP protein, the NLS-tag leads to the active import of the reporter protein into the cell nucleus. Following extraction directly from the nucleus, a decrease in fluorescence was observed, whereas no decrease in fluorescence was detected when extracting from the cytoplasm (Figure 2D). These results indicated that the NLS-tagged protein was extracted from the nucleus, but not from the cytoplasm. As a second marker, we chose a 70 kDa dextran-conjugated fluorophore (fluorescein isothiocyanate [FITC]-dextran), which surpasses the molecular weight described for crossing nuclear pores without active transport (Weis, 2003) and which we injected into cell nuclei by FluidFM as described previously (GuillaumeGentil et al., 2013). A decrease in FITC fluorescence was detected following extraction from the nucleus, whereas no decrease in fluorescence was observed after extraction from the cytoplasm (Figure 2D). These results indicated that the two fluorescent markers remained confined and were selectively extracted from the nucleus, i.e., were not forced through the nuclear pores during extraction from the cytoplasm. Quantification of the Extracted Volumes Next, the fraction of cellular content retrieved in the microfluidic probe was examined by optical microscopy. The use of GFP-expressing cells enabled the observation of the extract in fluorescence mode (Figure 2E). Combined with the known geometrical design of the FluidFM probe (see Figure S2), the visibility of the extract further enabled the quantification of the extracted volumes based on their area in the cantilever. Throughout this study, the volumes harvested ranged between 508 Cell 166, 506–516, July 14, 2016

0.1 and 7.0 pl. A volume of 0.1 pl corresponded to the filling of the pyramidal tip (see Figure S2); while smaller volumes were extractable, their visualization and quantification was challenging. Optical monitoring of the cantilever during suction of the cellular content also allowed us to estimate the volumetric flow rate of extraction. At maximal underpressure, the measured flow rates were 0.4 ± 0.1 pl/min for both the extraction from the cytoplasm and the extraction from the nucleus. The similarity in the obtained volumetric flow rates is consistent with the reported similarities in the viscosity of both compartments (Wachsmuth et al., 2003). The ability to visualize the extract and rapidly determine its volume in situ offered us the possibility to harvest defined volumes of cytoplasm or nucleoplasm by interrupting the extraction whenever the desired amount was collected in the probe. Post-extraction Cellular Viability Because the insertion of the FluidFM tip into the cell with or without injection did not adversely affect cell survival (Guillaume-Gentil et al., 2013), we next examined the cell’s survival as a function of the extracted cell volume and the cellular compartment targeted for extraction (Figure 3). After extraction of volumes up to 4.0 pl from the cytoplasm, 82% of the cells remained viable indicating that the cells have the ability to withstand the loss of a large proportion of the cytoplasm (Figure 3B). Extracted volumes larger than 4.5 pl all resulted in cell death. Considering the native cytoplasmic volumes measured in the pool of HeLa cells (maximum 4.4 pl; median 1.6 ± 0.7 pl; Figure 3A), the observed loss in viability after extraction was most likely related to the complete removal of the cytoplasmic content. At this point, we cannot exclude the possibility that continued aspiration after extraction of the entire cytoplasmic content forced the extra-cytoplasmic fluid through the cell membrane into the probe, likely leading to a loss of the membrane integrity and subsequent cell death. The non-viable cells all appeared flat and emptied in phase-contrast images and showed a dramatic, nearly complete loss of their GFP content (Figure 3D). Next, cell viability was assessed after the extraction of sub-picoliter volumes from the nucleus, ranging from 0.1 and 0.8 pl (Figure 3B). The cells survived the extraction of up to 0.6 pl, slightly below the median nucleus volume of 0.7 pl (Figure 3A). Extracted volumes of 0.7 pl and above compromised cell viability, and one cell did not survive the loss of 0.2 pl of nucleoplasm. The results indicate that extraction from the cell nucleus was more critical than extraction from the cytoplasm. However, extraction volumes of up to 0.6 pl had a high probability (86%) of preserving cellular viability. While physical damage, such as the loss of membrane integrity, can lead to immediate cell death, and is thus detectable by the classical cell viability assays that we performed in the few hours following extraction, deficiency in particular cell components could have an irreversible impact on cell physiology. Therefore, we also performed time-lapse experiments to monitor the extracted cells over several days using optical microscopy. The results confirmed the cell survival observed in the shortterm assessment, with the extracted cells remaining viable over the observation time (up to 5 days) and behaving similarly

Figure 2. Extraction from Fluorescently Labeled HeLa Cells (A) Phase-contrast (PhC) and fluorescent images (GFP) of a representative extraction from the nucleus. In the example shown, the impact of the removal of 0.3 pl of the nucleoplasm content did not induce noticeable morphological changes, and the decrease in the GFP content was minimal. The contour of the targeted cell is marked with a white dashed line. (B) Phase-contrast (PhC) and fluorescent images (GFP) of an example of cytoplasmic extraction. In this example, 1.5 pl were extracted from the cell cytoplasm; following extraction, the cell showed a flattened morphology, and its GFP content was dramatically decreased. The contour of the targeted cell is marked with a white dashed line. (C) Extraction of intracellular fluorescent markers examined by quantitative fluorescence microscopy. Plots show the loss of fluorescence in the cell after extraction. E, extracted cells; C, control cells. n = 5 extracted cells and n > 50 control cells per conditions. Dashed lines represent the means. *** denotes p < 0.001. (D) Compartment selective extraction of nuclear markers examined by quantitative fluorescence microscopy. Plots show the loss of fluorescence in the cell after extraction. E stands for extracted cells, C for control cells. For each conditions, n = 5 extracted cells for mRuby-NLS, 4 extracted cells for Dextran-FITC, and N > 50 control cells. Dashed lines represent the means. *** denotes p < 0.001. (E) Different volumes of GFP-containing extracts collected in the FluidFM probes. See also Figure S1 and Movie S1.

Cell 166, 506–516, July 14, 2016 509

Figure 3. Post-extraction Cell Viability (A) Distribution plots for the native whole-cell, cytoplasm, and nucleus volumes. The dashed lines indicate the median values, and the dotted lines show the minimal and maximal values. n = 132. (B) Cell viability as a function of the volumes extracted following cytoplasmic (top) and nuclear (bottom) extraction, determined using calcein AM stain. Each bar represents one count. The dashed and dotted lines indicate the median, minimal, and maximal values determined for native volumes, respectively. n = 13 for cytoplasmic extraction; n = 10 for nuclear extraction. (C and D) Representative examples of a stained viable cell (C) and an unstained non-viable cell (D). The phase-contrast (PhC) and fluorescent images (GFP) on the left show the extraction from the nucleus (0.5 pl and 0.7 pl for C and D, respectively). The fluorescent images on the right (CellTrace) show the calcein AM red-orange stain of the cells after extraction. The contour of the targeted cell is marked with a white dashed line. (E) Fluorescent images (GFP) from a 48-hr timelapse sequence following extraction. A volume of 2.9 pl was extracted from the cytoplasm of the lower cell, marked with a white dashed line; the upper cell was not manipulated. Both cells round up (29 hr) and divide (31 hr) at the same time, and the daughter cells finally respread on the substrate (40 hr).

to the neighboring, non-extracted cells. As expected, the cells that did not survive the extraction remained non-motile over the time course of the experiments. Figure 3E shows an example of a cell following the extraction of 2.9 pl from its cytoplasm (lower cell). Notably, the extracted cell divided after 30 hr, which was the same time as the neighboring unperturbed cell. The evaluation of the post-extraction cellular viability indicated that cytoplasm samples up to 4.0 pl and nucleoplasm samples up to 0.6 pl can be collected with a low risk of compromising cellular viability. Based on the largest measured native volumes for both cell compartments, these represent at least 90% and 20% of the native cytoplasmic and nuclear volumes, respectively. Molecular Analyses of the Collected Extracts In subsequent steps, we analyzed the subcellular levels of the endogenous molecules retrieved in the FluidFM probe. To this end, we transferred the extracts onto analytical substrates under optical inspection. To address a panel of analytical platforms, each with specific requirements and sample formats, we developed three different approaches that cover a wide range of applications. For imaging using transmission electron microscopy, the extracts were dispensed locally, in air, using the FluidFM force-control feedback to gently contact the grid. To perform biochemical assays, we dispensed the extracts into picoliter-containers containing the biochemical reagents. The third method consisted of the release of the extracts into a microliter water droplet that was subsequently manually transferred to a tube for transcript analyses. 510 Cell 166, 506–516, July 14, 2016

Extract Examination by Negative-Stain Transmission Electron Microscopy We imaged macromolecules from single-cell extracts by using negative-stain transmission electron microscopy (EM) (Figure 4). We spotted picoliter samples directly from the FluidFM cantilever onto a defined location on the EM grid using force-feedback and overpressure (Figure 4B). The extract, which was easily visible under a light microscope, stayed confined on the support film before rapidly drying. A drop of staining solution was then applied followed by blotting. The deposited cytoplasmic extracts exhibited a variety of structures (Figures 4C and S3). These differed in size and shape, including large vesicular structures (100 to 350 nm diameter), smaller globular structures (10 to 100 nm diameter), patches and filaments (5 to 20 nm diameter, up to 650 nm in length) indicative of membrane vesicles, macromolecular protein complexes, and cytoskeletal filaments. As a control, we imaged a HeLa cell lysate generated by freeze/thaw cycles, which showed a similar variety of structures. Because the lysate sample was not extracted and delivered onto the grid by FluidFM, structures larger than the aperture size were observed (see Figure S4). The examination of a water sample and a staining solution without sample did not reveal any of the cellular components observed in the cytoplasmic extract or cell lysate (data not shown). In contrast to the images generated from cytoplasmic extracts, those from nuclear extracts (Figures 4D and S5) showed mostly globular structures (sizes ranging from 5 to 25 nm in diameter) and short (20 to 100 nm) filaments (diameter <5 nm). These

Figure 4. Extract Imaging by Negative-Stain Transmission Electron Microscopy (A) Schematic representation of the strategy for the molecular imaging of cytoplasmic and nuclear extracts. (B) Scaled images of the electron microscopy grid and the FluidFM cantilever (left) and zoom-in on a selected location of the EM grid for local dispensing of the extract (right). The phasecontrast (PhC) and fluorescent (GFP) image series show the local dispensing of an extract (arrow) on the EM grid. (C) Representative electron micrographs of distinctive cellular structures observed in negatively stained cytoplasmic extracts, including large vesicular structures (top left), smaller globular structures (top right), and micro-sized filaments (bottom). (D) Representative electron micrographs of negatively stained cellular material from nuclear extracts, including globular structures and short filaments. See also Figures S3, S4, and S5.

nanoparticle, the presented approach is not limited to extracted cytoplasm or nucleoplasm samples and is applicable to any precious biological samples. Moreover, the deposited cell material is accessible for various labeling methods, e.g., immunogold staining.

observations were consistent with the expected nucleoplasm content, which should mainly consist of ribosomal proteins and nuclear lamina. Structures too large to pass through the 400 nm triangular aperture were not observed. The clear differences in the cellular elements noticed in the cytoplasmic extracts compared to those in the nuclear extracts substantiate the data obtained with nuclear specific fluorescent markers (Figure 2D), validating the developed cell extraction method and its ability to separately address the cytoplasm and the nucleus of a selected cell. Deposition by FluidFM onto the EM grid enabled the confinement of a low amount of cellular molecules within a small designated area of the grid, with the possibility to spot multiple samples onto the same specimen and to find these spots readily with TEM. As FluidFM can be used to dispense solutions or suspensions of virtually any molecule or

Detection of Enzymatic Activities Enzyme activity assays at the single-cell level are critical for elucidating the biochemical origin of cellular heterogeneity. To test the functional integrity of the harvested proteins, we performed enzymatic assays on the cytoplasmic extracts (Figure 5). First, we selected a beta-galactosidase (b-gal) assay using the fluorogenic substrate fluorescein di-b-D-galactopyranoside (FDG). b-gal is widely applied as reporter enzyme to assess the transcriptional activity of promoters due to relatively low endogenous enzyme levels in mammalian cells (approximately 40,000 copies or 33.5 fmol in a HeLa cell [Nagaraj et al., 2011]). Using photolithography with the negative photoresist SU-8, we produced picowells to assay the activity of endogenous b-gal retrieved from HeLa cell cytoplasm by confining the enzyme, its substrate, and its product into small observation volumes. The wells were pre-filled with the fluorogenic substrate and covered with oil to prevent evaporation while allowing for subsequent access with the FluidFM probe. The tip of the cantilever containing a cellular extract was positioned into the well, and the extract was released on application of overpressure (Figure 5B). The well was then monitored by fluorescence microscopy to detect the formation of fluorescein by enzymatic hydrolysis of FDG (Figures 5C and Cell 166, 506–516, July 14, 2016 511

Figure 5. Enzyme Activity Assays (A) Schematic representation of the extract dispensing in picoliter containers for biochemical analysis. (B) Microscopy image series showing the release of a cytoplasmic extract (3.0 pl) into a 3.5 pl well pre-filled with the fluorogenic substrate FDG. (C and D) Detection of endogenous b-gal activity. (C) Fluorescent images of the well with added cytoplasmic extract (top) and a control well prefilled with the enzyme substrate but without extract (bottom); the formation of fluorescein in the presence of b-gal is visible after 1 and 1.5 hr. (D) Graphic representation of the fluorescence intensity over time; an increase in fluorescence is detected in the well containing the cytoplasmic extract, but not in the control well containing only FDG. (E) b-gal activity assay comparing non-transfected (LacZ ) and transfected (LacZ +) HeLa cells. n = 5 cytoplasmic extracts per conditions. (F) Casp3 activity assay. HeLa cells were treated with staurosporin (Stau) and pre-infected with vaccinia virus (VACV) as indicated. n = 4 cytoplasmic extracts per conditions. Dashed lines represent the means. *p < 0.05, **p < 0.01.

5D). Next, we adapted the assay to differentiate cells with endogenous activity from those transfected with lacZ. Indeed, the measured fluorescein intensity was significantly higher in transfected compared with non-transfected cells (Figure 5E). To demonstrate that single-cell enzymatic assays can be used to record the status of a cell, we developed a second enzymatic assay. Caspase 3 (Casp3) is a critical enzyme in apoptosis and can be measured using the fluorogenic substrate Ac-DEVD-AMC (Figure 5F). While apoptosis is not expected to be initiated in unperturbed HeLa cells, the intrinsic apoptotic pathway can be triggered by treatment with the protein kinase inhibitor staurosporine (Bertrand et al., 1994). We 512 Cell 166, 506–516, July 14, 2016

thus first compared the Casp3 activity in untreated HeLa cells and in cells incubated with the apoptosis inducer. The cytoplasmic extracts from staurosporine-treated HeLa cells showed a significantly higher level of Casp3 activity compared to extracts from untreated cells at two time points after treatment. We then performed the assays with extracts from HeLa cells treated with staurosporine, but pre-infected with vaccinia virus. Viral pathogens have evolved multiple strategies to evade the apoptosis of their host cell and promote infection (Amara and Mercer, 2015). Vaccinia virus, for instance, can interfere with both the intrinsic and the extrinsic apoptotic pathways, with at least six known viral proteins (Veyer et al., 2014; Wasilenko et al., 2003). In the vaccinia-infected cells, the treatment with staurosporine did not induce detectable level of Casp3 activity (Figure 5F), as expected. This second enzyme activity assay enabled to expose the effects of both an apoptosis initiator (e.g., Staurosporine) and an apoptosis inhibitor (e.g., vaccinia virus). The successful detection of b-gal and Casp3 activities in the cytoplasmic extracts showed that the functional integrity of the collected molecules is preserved and that biochemical analyses on subcellular samples can be performed in picoliter reaction chambers. Furthermore, the performed assays showed the possibility to distinguish differently treated cells by quantifying enzyme activities from single cells.

Figure 6. Gene Expression Analysis (A) Schematic illustrating the dispensing of cytoplasmic and nuclear extracts into a microliter drop, followed by manual transfer into PCR tubes for gene expression analysis. (B) Scaled images of a 1.5-ml drop and the FluidFM cantilever (left) and zoom-in images showing the extract release in the drop. The phase contrast image series (up) shows the introduction of the cantilever into the drop, the fluorescent image series (GFP, bottom) shows the extract release following application of overpressure. (C) Graphs of threshold cycle (Ct) values of ERCC spike in controls, GFP, B2M, and ACTB transcripts obtained using cytoplasmic extracts from GFPtransfected cells (n = 21), cytoplasmic extracts from non-transfected cells (n = 3), and nuclear extracts from transfected cells (n = 8). Error bars are the SD of technical triplicates. The volumes of each extract analyzed are indicated in the x axis. (D) Threshold cycle (Ct) values of ERCC spike in controls, and GFP, B2M, and ACTB transcripts obtained using a cytoplasmic (1.7 pl) and a nuclear (1.3 pl) extract from the same cell. Three technical replicates are shown for each transcript. See also Experimental Procedures and Figure S6.

Detection of Transcripts Gene expression studies for single cells usually rely on reverse transcription and PCR to amplify transcripts, followed by qPCR. The quantification of mRNAs from single eukaryotic cells is generally conducted after physical separation of the cells, followed by cell lysis (Hashimshony et al., 2012; Picelli et al., 2014; Ramsko¨ld et al., 2012; Tang et al., 2010). Moreover, the separation of nuclei, followed by transcriptional analysis, has also been achieved (Grindberg et al., 2013). Here, we first validated the single-cell transcript analysis using dilutions of bulk and single-cell samples and showed that we were able to detect the transcripts from samples corresponding to approximately 0.01 pg of total RNA (see Figure S6; Supplemental Experimental Procedures). We examined the cellular extracts obtained by FluidFM in a subsequent step using standard PCR tubes for cDNA synthesis and qPCR. Because these tubes cannot be directly accessed by the FluidFM probe, the extract was first released into a microliter drop of RNase-free water, followed by the addition of a drop of oil to prevent evaporation; the diluted extract was then manually pipetted into a PCR tube (Figures 6A and 6B). cDNA synthesis fol-

lowed by qPCR was then performed to detect the mRNA levels of three different genes: one encoding GFP, which is produced in the extracted cells, and two housekeeping genes, beta-actin (ACTB) and beta-2-microglobulin (B2M). In addition, spike-in controls (External RNA Controls Consortium [ERCC]) were implemented (Jiang et al., 2011). The spike-in controls were detected in all samples, indicating a successful cDNA synthesis reaction for each sample (Figure 6C). Cytoplasmic extracts with volumes ranging between 0.6 and 7.0 pl were analyzed for the expression of the above-mentioned genes. 90% of the analyzed samples showed expression of at least one of the three genes assayed, and all three transcripts were detected in two-thirds of the cytoplasmic extracts. Two samples were negative for all three biological markers, indicating that the concentrations of the transcripts for each gene were below the detection limit of our assay (less than five copies; Supplemental Experimental Procedures). The variations of the transcript levels in all positive samples (log10 range of 4.8 to 5.7) largely surpassed the differences in the analyzed volumes (approximately 12-fold). Although technical variability cannot be excluded, the relatively high variation observed here was not unexpected because the mRNA levels in the cytoplasm are determined by a complex interplay of transcription, processing, and degradation and because brief episodes of mRNA synthesis are followed by periods of transcriptional silence. In fact, it is well established that the transcript levels greatly vary from cell to cell, even for housekeeping genes (Bengtsson et al., 2005; Raj et al., 2006; Taniguchi et al., 2009). Cytoplasmic extracts Cell 166, 506–516, July 14, 2016 513

retrieved from HeLa cells not transfected with pmaxGFP were also analyzed for comparison (Figure 6C). In those extracts, transcripts from the two housekeeping genes were successfully detected, but not GFP gene expression. The analysis of the nuclear extracts revealed that transcripts of the marker genes were not detected in the smallest volumes (0.2, 0.3, and 0.5 pl, respectively), while detection of transcripts from at least one gene was possible in all cases with volumes of 0.7 pl or more, which, therefore, defines our detection limit for nuclear extract. Collectively, the nuclear extracts were more challenging than those from the cytoplasm, due to their smaller volumes and the likely lower amount of total mRNAs in the nucleus relative to the cytoplasm. Notably, we also examined the nucleus and cytoplasm from the same cell. Figure 6D shows an example of a differential analysis in which the mRNAs from all three genes were detected in both compartments. Approximately one order of magnitude higher amounts of each of the transcripts were detected in the cytoplasm compared to the nucleus. (GFP: 73; ACTB: 153; B2M: 83). Taken together, we were able to conduct a transcriptional analysis from the subcellular compartments, providing a unique foundation to investigate the spatial resolution of the transcript levels and potentially also transcript modifications. DISCUSSION The growing interest in studying biological processes at the single-cell level has fostered abundant methodological and technological developments to address the challenges associated with drastically reduced quantities of analytes. However, the technologies that enable molecular analyses of spread live cells are still sparse. The existing approaches focus on only one component, do not preserve the physiological and spatial context, and usually sacrifice the cell to access the intracellular molecules. Here, we present a generic and versatile platform to address individual living cells in a physiological context and gain direct access to a cell’s endogenous molecules. The developed extraction method allowed us to quantify and control the volumes collected by simultaneous real-time monitoring using optical microscopy. Furthermore, the intracellular fluid was selectively harvested from either the cytoplasm or the nucleus, enabling differential analyses on both cell compartments. Experiments with fluorescently labeled cellular components, as well as high-resolution imaging, showed the retrieval of soluble proteins and smaller molecules, while further biochemical and transcriptional assays demonstrated the integrity of the recovered cellular components. Analyses on subcellular concentrations are still challenging for most current analytical technologies, and, as such, the possibility to harvest the nearly entire cell compartment may facilitate downstream analysis. While transcripts are inherently highly variable due to rapid turnover, proteins present another level of examining the state of the cell. As we show, enzymes can be measured after extraction from single cells and used to show the activation of a regulatory pathway, shown exemplarily with apoptosis. Here, we have used picoliter volume reactors; however, it is easily conceivable to even tune down to ever smaller reactors when combining readouts with methods 514 Cell 166, 506–516, July 14, 2016

for detection at the single-molecule level (Liebherr et al., 2015; Rissin et al., 2010; Rondelez et al., 2005). Our study also assessed the boundaries at which cell volumes can be extracted to preserve cell viability. Using the rather gentle and controlled force offered by FluidFM, we have found that cells have the remarkable ability to survive and undergo cell division, even after the extraction of a large fraction of the cytoplasm. As anticipated, extraction from the nucleus is more critical for the cell; however, withdrawal for molecular analyses is also possible from this compartment. The determination of the lower and upper limits for extraction and subsequent analyses should allow the study of the dynamics of processes in single cells. Sampling small volumes may also enable studies on the endogenous molecules localized at specific cell locations, e.g., specific mRNAs localized in a neuron’s dendrites (Pfeiffer-Guglielmi et al., 2014). Moreover, the FluidFM-based extraction technology presented here offers perspectives for third-generation sequencing technologies that allow single-molecule sequencing from undiluted samples, the determination of epigenetic changes, and novel synthetic biology studies, e.g., to build and analyze artificial cells. EXPERIMENTAL PROCEDURES Cells HeLa cells were maintained in DMEM with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin at 37 C in a 5% CO2 humidified incubator. For all extraction experiments, the cells were seeded on 50-mm tissue culturetreated low m-dishes (Vitaris), and growth medium was replaced with CO2-independent growth medium supplemented with 2 mM L-glutamine, 10% FBS, and 1% penicillin-streptomycin. For information on cell transfection and staining, see the Supplemental Experimental Procedures. FluidFM Setup A FluidFM system composed of a FlexAFM-NIR scan head, a C3000 controller, a digital pressure controller, microfluidic probes (Cytosurge), and EasyScan2 software (Nanosurf) was used. The scan head was mounted on an inverted microscope equipped with a temperature-controlled incubation chamber. A syringe pressure kit with a three-way valve (Cytosurge) was used in addition to the digital pressure controller to apply under- and overpressure differences larger than 800 and 1,000 mbar. FluidFM probes were prepared as described in the Supplemental Experimental Procedures. Microscopy Phase-contrast and fluorescence imaging were performed using a Zeiss Axio Observer Z1 microscope equipped with 103 and 403 (0.6 na) objectives and Colibri 365, 470, and 556 LED modules. Microscopy images were captured with a Zeiss AxioCam MRm R3 camera and analyzed using the AxioVision and ImageJ software. For the time-lapse microscopy of cells post-extraction, images were acquired every 30 or 60 min in fluorescence and phase-contrast modes for up to 5 days. Viable (n = 4 for cytoplasm and n = 1 for nucleoplasm extraction) and non-viable (n = 11) cells were monitored. For details on the determination of the HeLa cell volume, see the Supplemental Experimental Procedures. Extraction Experiments The cell to be extracted was visualized by light microscopy, and the FluidFM probe was placed above the desired point of insertion, i.e., on top of the nucleus or the cytoplasm. The probe was then inserted into the cell compartment through a forward force spectroscopy routine driven by the Z-piezo. The probe was then maintained inside the cell at constant force (550 nN). Underpressure larger than 800 mbar was applied to aspirate the cellular content. The

pressure-assisted flow of the intracellular content into the FluidFM probe was interrupted by switching the pressure back to zero. The probe was then retracted through backward force spectroscopy. For details on the experimental procedures, see the Supplemental Experimental Procedures. Determination of the Extracted Volumes and Volumetric Flow Rate The extracted volumes were obtained by measuring the area occupied by the extract confined in the cantilever on micrographs, multiplying by the channel height of 1 mm, and adding the volume of the hollow pyramidal tip (90 fl) (see Figure S2 for the details on the tip volume). To determine the volumetric flow rate, the extracted volumes measured at different time points during the extraction were plotted against time and a linear regression fit (R2 R 0.99) was used to determine the volumetric flow rate. n = 12 for the extraction from the cytoplasm, and n = 9 for the extraction from the nucleus. Examination by Negative-Stain Electron Microscopy The extract-containing probe was positioned onto a chosen location of a FCF200-Cu specimen grid (Electron Microscopy Sciences) by a force-controlled approach with a set point of 20 nN. Overpressure (>1000 mbar) was then applied to flow the extract out of the probe. The specimen grid was then kept on ice until it was stained. For negative staining, a 4-ml drop of 1% phosphotungstic acid (PTA) was applied to the grid for 20 s and carefully blotted away by applying a piece of Whatman filter paper on the edge of the grid. The grid was then viewed with an FEI Morgagni 268 electron microscope operated at 100 kV (at SCOPEM, ETHZ). See the Supplemental Experimental Procedures for the preparation of the water and cell lysates controls. Enzyme Assays Beta-galactosidase activity and caspase activity were determined in picoliterwell arrays, the fabrication of which is described in the Supplemental Experimental Procedures. Following single-cell extraction, the cytoplasmic samples were dispensed in wells prefilled with the fluorogenic substrate, and the wells were monitored over time by fluorescence microscopy. For details on fluorogenic substrates and procedures, see the Supplemental Experimental Procedures. mRNA Amplification and Detection A 1.5-ml drop of RNase-free water, supplemented with 2 U ml 1 RNase inhibitors (Life Technologies), was deposited onto an AG480F AmpliGrid (LTF Labortechnik). The cantilever was introduced into the drop using the micrometer screws to displace either the AFM or the sample stage. Once the cantilever was located inside the drop, overpressure (> 1,000 mbar) was applied to release the extract. The cantilever was then withdrawn from the drop using the micrometer screws. A 4-ml sealing solution (Beckman Coulter) was deposited on top of the 1.5-ml drop before the entire 5.5-ml solution was pipetted into a PCR tube. The solution was then briefly centrifuged and stored at 80 C until further use. For cDNA synthesis and qPCR protocols and their validation, see Figure S6, Tables S1 and S2, and Supplemental Experimental Procedures. Statistical Analysis Significant differences between groups were determined by comparing means using an unpaired one-tailed, two-sample t test. All statistical analyses were performed with the Origin Software. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, six figures, two tables, and one movie and can be found with this article online at http://dx.doi.org/10.1016/j.cell.2016.06.025. AUTHOR CONTRIBUTIONS O.G.-G., T.Z., and J.A.V. designed the study. O.G.-G. developed the FluidFM extraction protocol, conducted the extraction experiments, and carried out the biochemical assays. R.V.G. performed the qPCR assays. R.K. and M.P. performed the EM experiments. L.D.-C. and D.O. conducted FIB. V.M. produced

the microwell arrays. O.G.-G. and J.A.V. wrote the manuscript with support from R.V.G., R.K., M.P., and T.Z. ACKNOWLEDGMENTS We thank Ja´nos Vo¨ro¨s (LBB, ETHZ), Michael Gabi, Pascal Behr, Pablo Do¨rig (Cytosurge), Patrick Frederix (Nanosurf), and Edin Sarajlic (SmartTip) for their constant support, Jason Mercer (UCL) for his help with vaccinia virus experiments, Miriam Bortfeld-Miller (IMB, ETHZ), Stephen Wheeler, and Martin Lanz (ETHZ, LBB workshop) for technical assistance, Ikonaut GmbH for the scientific cartoons (Figures 1A, 4A, 5A, and 6A). TEM, SEM, and FIB data were collected at the Scientific Center for Optical and Electron Microscopy facility (ScopeM, ETH Zurich). This project was supported by the Swiss Innovation Promotion Agency CTI-KTI (CTI no. 14336.1 PFNM-NM) (to T.Z. and J.A.V.), the Swiss National Science Foundation (CR23I2_135535) (to T.Z.), and the Promedica Foundation (to J.A.V.). Received: October 5, 2015 Revised: April 5, 2016 Accepted: June 1, 2016 Published: July 14, 2016 REFERENCES Actis, P., Maalouf, M.M., Kim, H.J., Lohith, A., Vilozny, B., Seger, R.A., and Pourmand, N. (2014). Compartmental genomics in living cells revealed by single-cell nanobiopsy. ACS Nano 8, 546–553. Amara, A., and Mercer, J. (2015). Viral apoptotic mimicry. Nat. Rev. Microbiol. 13, 461–469. Bengtsson, M., Sta˚hlberg, A., Rorsman, P., and Kubista, M. (2005). Gene expression profiling in single cells from the pancreatic islets of Langerhans reveals lognormal distribution of mRNA levels. Genome Res. 15, 1388–1392. Bertrand, R., Solary, E., O’Connor, P., Kohn, K.W., and Pommier, Y. (1994). Induction of a common pathway of apoptosis by staurosporine. Exp. Cell Res. 211, 314–321. Cai, X., Evrony, G.D., Lehmann, H.S., Elhosary, P.C., Mehta, B.K., Poduri, A., and Walsh, C.A. (2014). Single-cell, genome-wide sequencing identifies clonal somatic copy-number variation in the human brain. Cell Rep. 8, 1280–1289. Grindberg, R.V., Yee-Greenbaum, J.L., McConnell, M.J., Novotny, M., O’Shaughnessy, A.L., Lambert, G.M., Arau´zo-Bravo, M.J., Lee, J., Fishman, M., Robbins, G.E., et al. (2013). RNA-sequencing from single nuclei. Proc. Natl. Acad. Sci. USA 110, 19802–19807. Guillaume-Gentil, O., Potthoff, E., Ossola, D., Do¨rig, P., Zambelli, T., and Vorholt, J.A. (2013). Force-controlled fluidic injection into single cell nuclei. Small 9, 1904–1907. Guillaume-Gentil, O., Potthoff, E., Ossola, D., Franz, C.M., Zambelli, T., and Vorholt, J.A. (2014). Force-controlled manipulation of single cells: from AFM to FluidFM. Trends Biotechnol. 32, 381–388. Hashimshony, T., Wagner, F., Sher, N., and Yanai, I. (2012). CEL-Seq: singlecell RNA-Seq by multiplexed linear amplification. Cell Rep. 2, 666–673. Jiang, L., Schlesinger, F., Davis, C.A., Zhang, Y., Li, R., Salit, M., Gingeras, T.R., and Oliver, B. (2011). Synthetic spike-in standards for RNA-seq experiments. Genome Res. 21, 1543–1551. Kovarik, M.L., and Allbritton, N.L. (2011). Measuring enzyme activity in single cells. Trends Biotechnol. 29, 222–230. Kuipers, M.A., Stasevich, T.J., Sasaki, T., Wilson, K.A., Hazelwood, K.L., McNally, J.G., Davidson, M.W., and Gilbert, D.M. (2011). Highly stable loading of Mcm proteins onto chromatin in living cells requires replication to unload. J. Cell Biol. 192, 29–41. Liebherr, R.B., Hutterer, A., Mickert, M.J., Vogl, F.C., Beutner, A., Lechner, A., Hummel, H., and Gorris, H.H. (2015). Three-in-one enzyme assay based on single molecule detection in femtoliter arrays. Anal. Bioanal. Chem. 407, 7443–7452.

Cell 166, 506–516, July 14, 2016 515

Lo, S.J., and Yao, D.J. (2015). Get to understand more from single-cells: current studies of microfluidic-based techniques for single-cell analysis. Int. J. Mol. Sci. 16, 16763–16777.

Saha-Shah, A., Weber, A.E., Karty, J.A., Ray, S.J., Hieftje, G.M., and Baker, L.A. (2015). Nanopipettes: probes for local sample analysis. Chem. Sci. 6, 3334–3341.

Meister, A., Gabi, M., Behr, P., Studer, P., Vo¨ro¨s, J., Niedermann, P., Bitterli, J., Polesel-Maris, J., Liley, M., Heinzelmann, H., and Zambelli, T. (2009). FluidFM: combining atomic force microscopy and nanofluidics in a universal liquid delivery system for single cell applications and beyond. Nano Lett. 9, 2501–2507.

Sarkar, A., Kolitz, S., Lauffenburger, D.A., and Han, J. (2014). Microfluidic probe for single-cell analysis in adherent tissue culture. Nat. Commun. 5, 3421.

Nagaraj, N., Wisniewski, J.R., Geiger, T., Cox, J., Kircher, M., Kelso, J., Pa¨a¨bo, S., and Mann, M. (2011). Deep proteome and transcriptome mapping of a human cancer cell line. Mol. Syst. Biol. 7, 548. Nawarathna, D., Turan, T., and Wickramasinghe, H.K. (2009). Selective probing of mRNA expression levels within a living cell. Appl. Phys. Lett. 95, 83117. O’Huallachain, M., Karczewski, K.J., Weissman, S.M., Urban, A.E., and Snyder, M.P. (2012). Extensive genetic variation in somatic human tissues. Proc. Natl. Acad. Sci. USA 109, 18018–18023. Osada, T., Uehara, H., Kim, H., and Ikai, A. (2003). mRNA analysis of single living cells. J. Nanobiotechnology 1, 2. Pfeiffer-Guglielmi, B., Dombert, B., Jablonka, S., Hausherr, V., van Thriel, C., Scho¨bel, N., and Jansen, R.P. (2014). Axonal and dendritic localization of mRNAs for glycogen-metabolizing enzymes in cultured rodent neurons. BMC Neurosci. 15, 70.

Schmid, A., Kortmann, H., Dittrich, P.S., and Blank, L.M. (2010). Chemical and biological single cell analysis. Curr. Opin. Biotechnol. 21, 12–20. Tang, F., Barbacioru, C., Nordman, E., Li, B., Xu, N., Bashkirov, V.I., Lao, K., and Surani, M.A. (2010). RNA-Seq analysis to capture the transcriptome landscape of a single cell. Nat. Protoc. 5, 516–535. Taniguchi, K., Kajiyama, T., and Kambara, H. (2009). Quantitative analysis of gene expression in a single cell by qPCR. Nat. Methods 6, 503–506. Van Gelder, R.N., von Zastrow, M.E., Yool, A., Dement, W.C., Barchas, J.D., and Eberwine, J.H. (1990). Amplified RNA synthesized from limited quantities of heterogeneous cDNA. Proc. Natl. Acad. Sci. USA 87, 1663–1667. Veyer, D.L., Maluquer de Motes, C., Sumner, R.P., Ludwig, L., Johnson, B.F., and Smith, G.L. (2014). Analysis of the anti-apoptotic activity of four vaccinia virus proteins demonstrates that B13 is the most potent inhibitor in isolation and during viral infection. J. Gen. Virol. 95, 2757–2768.

Picelli, S., Faridani, O.R., Bjo¨rklund, A.K., Winberg, G., Sagasser, S., and Sandberg, R. (2014). Full-length RNA-seq from single cells using Smartseq2. Nat. Protoc. 9, 171–181.

Wachsmuth, M., Weidemann, T., Mu¨ller, G., Hoffmann-Rohrer, U.W., Knoch, T.A., Waldeck, W., and Langowski, J. (2003). Analyzing intracellular binding and diffusion with continuous fluorescence photobleaching. Biophys. J. 84, 3353–3363.

Raj, A., Peskin, C.S., Tranchina, D., Vargas, D.Y., and Tyagi, S. (2006). Stochastic mRNA synthesis in mammalian cells. PLoS Biol. 4, e309.

Wang, D., and Bodovitz, S. (2010). Single cell analysis: the new frontier in ‘omics’. Trends Biotechnol. 28, 281–290.

Ramsko¨ld, D., Luo, S., Wang, Y.C., Li, R., Deng, Q., Faridani, O.R., Daniels, G.A., Khrebtukova, I., Loring, J.F., Laurent, L.C., et al. (2012). Full-length mRNA-Seq from single-cell levels of RNA and individual circulating tumor cells. Nat. Biotechnol. 30, 777–782.

Wasilenko, S.T., Stewart, T.L., Meyers, A.F., and Barry, M. (2003). Vaccinia virus encodes a previously uncharacterized mitochondrial-associated inhibitor of apoptosis. Proc. Natl. Acad. Sci. USA 100, 14345–14350.

Rissin, D.M., Kan, C.W., Campbell, T.G., Howes, S.C., Fournier, D.R., Song, L., Piech, T., Patel, P.P., Chang, L., Rivnak, A.J., et al. (2010). Single-molecule enzyme-linked immunosorbent assay detects serum proteins at subfemtomolar concentrations. Nat. Biotechnol. 28, 595–599. Rondelez, Y., Tresset, G., Tabata, K.V., Arata, H., Fujita, H., Takeuchi, S., and Noji, H. (2005). Microfabricated arrays of femtoliter chambers allow single molecule enzymology. Nat. Biotechnol. 23, 361–365.

516 Cell 166, 506–516, July 14, 2016

Weis, K. (2003). Regulating access to the genome: nucleocytoplasmic transport throughout the cell cycle. Cell 112, 441–451. Wu, M., and Singh, A.K. (2012). Single-cell protein analysis. Curr. Opin. Biotechnol. 23, 83–88. Zhao, L., Kroenke, C.D., Song, J., Piwnica-Worms, D., Ackerman, J.J., and Neil, J.J. (2008). Intracellular water-specific MR of microbead-adherent cells: the HeLa cell intracellular water exchange lifetime. NMR Biomed. 21, 159–164.