Use of Time-Resolved Fluorescence To Improve Sensitivity and

Subscriber access provided by GAZI UNIV

Article

Use of time-resolved fluorescence improves sensitivity and dynamic range of gel-based proteomics AnnSofi Sandberg, Volker Buschmann, Peter Kapusta, Rainer Erdmann, and Asa M Wheelock Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.5b03805 • Publication Date (Web): 08 Feb 2016 Downloaded from http://pubs.acs.org on February 12, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Use of time-resolved fluorescence to improve sensitivity and dynamic range of gel-based proteomics AnnSofi Sandberg1, Volker Buschmann2, Peter Kapusta2, Rainer Erdmann2 and Åsa M. Wheelock1,3,*

1

Respiratory Medicine Unit, Department of Medicine, Center for Molecular Medicine,

Karolinska Institutet, 171 76 Stockholm, Sweden 2

PicoQuant GmbH, Rudower Chaussee 29, 124 89 Berlin, Germany

3

Hikari Bio AB, Franzéngatan 39, 112 16 Stockholm, Sweden

Current author-affiliations: AS: Cancer Proteomics Mass Spectrometry, Department of Oncology-Pathology, Science for Life Laboratory, Karolinska Institutet, Stockholm, Sweden PK: J. Heyrovsky Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, Prague, Czech Republic

*Correspondence to be addressed to: Åsa M. Wheelock Pulmonomics group, Building L4:01 Respiratory Medicine Unit, Department of Medicine Karolinska Institutet SE-171 76 Stockholm Running title (50 char): Time-resolved fluorescence in gel-based proteomics

Page count: 5419 words + (3 Tables = 1 page) + (4 figures = [2 x 250] + [2 x 500] words) = 8 pages

1 Environment ACS Paragon Plus


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT Limitations in the sensitivity and dynamic range of two-dimensional gel electrophoresis (2DE) are currently hampering its utility in global proteomics and biomarker discovery applications. In the current study, we present proof-of-concept analyses showing that introducing time-resolved fluorescence in the image acquisition step of in-gel protein quantification provides a sensitive and accurate method for subtracting confounding background fluorescence at the photon level. In-gel protein detection using the minimal DIGE workflow showed improvements in lowest limit of quantification (LLOQ) in terms of CyDye molecules/pixel of 330-fold in the blue-green area (Cy2), and 8000-fold in the red region (Cy5) over conventional state-of-the-art image acquisition instrumentation, here represented by the Typhoon 9400 instrument. These improvements make possible the detection of low abundance proteins present at sub-attomolar levels, thereby representing a quantum leap for the use of gel-based proteomics in biomarker discovery. These improvements were achieved using significantly lower laser powers and overall excitation times, thereby drastically decreasing photo-bleaching during repeated scanning. The single-fluorochrome detection limits achieved by the CuTEDGE technology facilitates in-depth proteomics characterization of very scarce samples, e.g. primary human tissue materials collected through in clinical studies. The unique information provided by high sensitivity 2-DE, including positional shifts due to post-translational modifications, may increase the chance to detect biomarker signatures of relevance for identification of disease sub-phenotypes.


Page 2 of 28

Page 3 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


INTRODUCTION Almost three decades after its introduction, two-dimensional gel electrophoresis (2-DE) remains the paramount method for separation of undigested (intact) proteins, and represents a cornerstone in the field of quantitative intact proteomics (QIP). The launch of the Difference Gel Electrophoresis (DIGE) technology featuring mass- and charge- matched cyanine dyes (Cy2, Cy3 and Cy5) to facilitate multiplexing with a maintained superimposable protein separation pattern, thereby enabling correction for experimental variation by means of a pooled internal standard, represented a key step in improving the quantitative aspects and overall statistical power of the 2-DE technique. However, the lack of sensitivity and dynamic range is still hampering the utility of gel-based proteomics in e.g. biomarker discovery: Physiological protein abundances may span as much as 10-12 orders of magnitude1. In contrast minimal DIGE, the most sensitive and accurate global 2-DE workflow currently available on the market, is generally limited to a dynamic range of ~104 with sensitivities of 0.5 ng (~10 fmol) protein2,3. Conventional fluorescence imaging approaches utilized in current state-of-the-art gel image acquisition instruments (e.g. the Typhoon laser scanners, GE Healthcare) do not exploit the full potential of the CyDye fluorochromes. The use of a constant laser source contributes to auto-fluorescence and other non-specific fluorescent components that mask signals from low abundant proteins and thereby limit both sensitivity and dynamic range. Current solutions to reduce background noise, such as in silico background subtraction applied in the downstream image analysis steps, are prone to error and tend to introduce variance4,5,6 particularly for faint features such as low abundance proteins, with shallow grayscale gradients between protein spot/band and background6. To address these issues, our patented CuTEDGE (Cumulative Time-resolved Emission 2Dimensional Gel Electrophoresis) approach introduces time-resolved fluorescence (TRF) in the image acquisition step of gel-based proteomics7. CyDyes have previously been



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

implemented in TRF-mediated investigations of protein characteristics, including localization, folding dynamics, and intracellular movement8,9,10. In CuTEDGE, TRF is utilized to facilitate background subtraction at the photon level: Non-specific confounding fluorescence as well as Raman and Raleigh scattering contributions and detector dark counts (background noise) is quantified and subtracted based on its fluorescence decay signature, which is distinct from fluorescence originating from CyDye-conjugated proteins (signal)7.


Page 4 of 28

Page 5 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


EXPERIMENTAL PROCEDURES Protein separation using 1D-SDS-PAGE BSA protein standard (Albumin monomer bovine A1900, Sigma) was solubilized in lysis buffer (30mM Tris, 2M thiourea, 7M urea, 4% CHAPS, pH 8.5) to a concentration of 5 mg/ml and labeled with DIGE fluorochromes in the dark according to the manufacturer’s instructions (Mw and pI matched N-hydroxy succinimidyl esters of Cy2, and Cy5 respectively; GE Healthcare, Uppsala, Sweden). Minimal DIGE labeling strategies were utilized (400pg CyDye/50µg protein), resulting in labeling of 2-3% of total protein. Labeled samples were diluted in sample loading buffer (63mM Tris, 10% glycerol, 2.0% SDS, pH 6.8) to a final concentration corresponding to 0.1 µg to 1.0 fg protein/well. The specifics of the prepared dilution series, including the theoretical number of CyDye fluorochromes loaded/well, are presented in Table 1.

For the purpose of in-gel measurements, the prepared dilution series were loaded onto small format (8x8 cm), 15 well 1.0 mm thick Novex Tris-Glycine 4-20% gradient polyacrylamide gels (Invitrogen, Carlsbad, CA, USA) under reducing conditions. Proteins were separated using the Mighty Small II electrophoresis unit (GE Healthcare) at 15ºC, 15 mA/gel until the dye front reached the edge of the gel. Due to impurities present in the BSA reagent, the resulting electrophoresis separation pattern consisted of a multitude of bands, thereby resembling the separation pattern of a more complex protein sample. An estimated 80% of loaded protein appeared in the main BSA band of 50 kDa. Following electrophoresis, gels were removed from the plastic backing and placed directly on the glass platen for scans using the Typhoon 9400 instrument. For scans using the MicroTime 100 custom setup (PicoQuant GmbH, Berlin, Germany), gels were sandwiched between 10x10.5 cm low fluorescence glass



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

plates (Hoefer Pharmacia Biotech low fluorescence glass plates; No. SE262GN, Fisher Scientific, Gothenburg, Sweden) prior to scanning.

Protein separation of complex cellular samples using 2-DE PAGE The human monocytic cell line U-937 was stimulated with PMA, thereby mimicking differentiation into macrophages, and solubilized in lysis buffer (30mM Tris, 2M thiourea, 7M urea, 4% CHAPS, 0.5% Triton, 2% (v/v) protease inhibitor and DTT 1mg/ml, pH 8.5) to a concentration of 5 mg/ml and labeled with DIGE fluorochromes (Mw and pI matched Nhydroxy succinimidyl esters of Cy2, Cy3 and Cy5 respectively; GE Healthcare) in the dark according to the manufacturer’s instructions. Minimal DIGE labeling strategies were utilized (400pg CyDye/50µg protein), resulting in labeling of 2-3% of total proteins.

For 2-DE DIGE, 50 µg total protein of each labeled cell lysate sample was rehydrated onto 18 cm IPG strips, pH gradient 4-7 (GE Healthcare) and separated by isoelectric focusing for 40.7 kVhrs, then stored in -80°C until use. Prior to SDS-PAGE, the IPG strips were cut into 8 cm sections to accommodate the 10 x 10 cm scanning board of the custom MicroTime 100 scanner setup (MicroTime100, see Image acquisition below), equilibrated in DTT and IAA, and separated on lab cast 10x10 cm 12% polyacrylamide gels as previously described11.

Image acquisition All samples were prepared in duplicates to allow comparison of conventional (Typhoon) and TRF (MicroTime 100) scanning of fresh, identical gels, and thereby avoiding bias through photo-bleaching or diffusion caused by sequential scanning of the same gel sample. In addition, all experiments were repeated at two independent occasions, at two different sites (Karolinska Institute, Stockholm and PicoQuant, Berlin, respectively). A custom script


Page 6 of 28

Page 7 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


facilitating an expanded capacity of the photon counting software was implemented in the 2nd setup. Due to the improved statistical reliability provided by the broader dynamic range of the photon counting software, only the latter are presented in the results section below. However, the overall sensitivities of detection and lifetime components were comparable for the two setups. Instrumentation for time-resolved fluorescence image acquisition. A custom set-up of MicroTime 100 (PicoQuant GmbH, Berlin, Germany) featuring an BX41 upright microscope (Olympus) equipped with a10x10 cm x,y-cross stage with linear piezo motors (Figure 1) was utilized for time-resolved fluorescence measurements by means of Time-Correlated Single Photon Counting (TCSPC)12. Specifically, the external TCSPC equipment consisted of picosecond pulsed laser diodes of wavelength of 485 nm for excitation of Cy2 or 640 nm for excitation of Cy5 (PDL 800-D with LDH-P-C-485 or LDH-P-C-640B, PicoQuant GmbH) with a pulse rate of 40 MHz, and the TimeHarp 200 TCSPC board (PicoQuant GmbH). Synchronization of the data acquisition with the scanner movement was controlled by SymPhoTime software using the Time-Tagged Time-Resolved (TTTR) measurement mode in which each photon is stored separately12,13,14,15. A PMA-182-N-M photomultiplier tube (PMT) detector (PicoQuant GmbH) with a low photocathode quantum efficiency (15% at 450nm and 9% at 550 nm) along with bandpass emission filters HQ520/35 for Cy2, HQ580/70 for Cy3 or HQ685/70 for Cy5 (AHF Analysentechnik, Tübingen, Germany) were used for measuring the fluorescent emission. The instrument response function (IRF) of the whole setup was measured to be approximately 250ps. A PlanN objective with 10x magnification (Olympus) and a numerical aperture (NA) of 0.25 was used. In order to maximize the axial depth of the measurements, a fiber optic cable with a diameter of 600µm was utilized, thereby essentially eliminating the confocal detection restriction of the instrument, with a resulting axial depth of



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

28 µm. Images were recorded with 100 µm per pixel resolution. Fluorescence decay curves were recorded with a time bin width of 32 ps. Data processing for time resolved measurements. Multiexponential lifetime fitting of the TCSPC data was performed using a modified version of SymPhoTime vs 5.1.2 (PicoQuant). Reconvolution fitting using reconstructed instrument response function (IRF) was used to calculate the contributions of the individual lifetime components. Evaluation of the optimal fit was performed both on a global image scale as well as on selected regions of interest in 1Dand 2DE gel images. Due to low light intensities (photons per pixel) Maximum likelihood estimator (MLE)16 instead of common least squares criterion was applied for improved fitting. The quality of the fit was judged by the reduced Chi-squared (χ2) value and visual inspection of the weighted residuals plot as a measure of discrepancy between the model and the actual data. Pixel-by-pixel fitting of the images was performed using the optimized in-gel lifetime settings for the respective fluorochrome. A script using the Stupslang programming language (integral part of SymPhoTime vs 5) was generated to allow export of the intensity images of the individual lifetime components as .dat file in plain ASCII format for further quantitative and qualitative analysis using ImageJ. The scripts created for the purpose of this project for export of single-lifetime-component images are made available in the current version of SymPhoTime; SymPhoTime64. Conventional image acquisition. Fresh replicate gels were scanned using a Typhoon 9400 laser scanner (GE Healthcare), featuring an axial depth of 300 µm, using the following filter and laser settings: Cy2; 488nm excitation (Argon ion laser, 20 mW) with emission filter bp520/40; Cy3: 488 nm excitation with emission filter bp580/30 (comparable to TRF measurements) and 532 nm excitation (solid-state doubled frequency Nd:YAG laser, 20 mW) with emission filter 580/30 (optimal for fluorochrome); Cy5: 633 nm excitation (HeliumNeon laser, 10 mW) with emission filter bp 670/30. PMT voltage was optimized for each scan


Page 8 of 28

Page 9 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


to achieve maximum dynamic range but avoid saturation. The images were exported as .gel data files, a proprietary exponential conversion of the linear 16-bit .tif file format with a dynamic range of 6.5 *104 into a dynamic range of 105. Measurements of the respective dilutions series pipetted into low fluorescent microtiter plates were also performed in order to estimate the overall fluorescent contribution of the acrylamide gel matrix in conventional fluorescence measurements.

Image and data analysis Fluorescence images from MicroTime 100 were exported as ASCII files and plotted as 32 bit grey scale images using ImageJ freeware. Fluorescence images acquired using the Typhoon 9400 were exported both as the standard 16-bit .tif file format, as well as the GEHC’s proprietary .gel data file format, representing an exponential conversion of the linear 16-bit .tif file format with a dynamic range of 6.5*104 into a dynamic range of 1.0*105. The .gel image transformation format reduces the numerical reporting of relative fluorescence in the lower detection range, and may thereby give the impression of a decreased background fluorescence, as confirmed by corresponding analyses performed using ImageJ. Image data acquired by the Typhoon 9400 were thus analyzed in .gel format using ImageQuant 5.2 (GE Healthcare) in order to report the best possible results for the instrument. Dynamic range calculations were performed applying a snug fit with the square area selection tools of ImageQuant/ImageJ to the individual protein bands of the 1D-gel dilution series, and calculating the intensity density (equivalent to protein spot volume in 2-DE) of each protein band. Signal-to-Noise (S/N) calculations were performed by generating an intensity profile across the center of the protein bands/protein spots. S/N ratios were calculated as ([SignalBackground]/Noise). Signal: five-point-average in the most stable area of each protein band peak; Background: Average amplitude in region free of fluorochrome; Noise: SD of a



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

minimum of 10 consecutive data points in a region free of fluorochrome. The limit of detection (LOD) was defined as S/N ratio >3 * SD of background noise in accordance with IUPAC4, and the lowest limit of quantification (LLOQ) was defined as S/N ratio >5 * SD of the background noise in accordance with the U.S. Food and Drug Administration (FDA) guidelines17.


Page 10 of 28

Page 11 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


RESULTS The performance of the CuTEDGE methodology was compared to that of a current state-ofthe-art image acquisition instrumentation using conventional lasers, here represented by the Typhoon 9400 (GE Healthcare). A MicroTime 100 (PicoQuant) setup for Fluorescence Lifetime Imaging (FLIM) featuring an Olympus BX41 upright confocal microscope equipped with a piezo-motored scanning table coupled to TCSPC equipment was utilized for timeresolved fluorescence measurements (Figure 1)12. To assure that the observed improvements were not merely an effect of superior optics, conventional analysis as well as time-resolved analysis of the fluorescent output of the custom MicroTime 100 was performed. Evaluation of sensitivity and dynamic range was performed using a dilution series of CyDye-labeled BSA in 10-fold increments ranging from 0.1 µg to 1.0 fg total BSA protein/lane separated by 1dimensional SDS-PAGE. In addition, human macrophage samples separated by 2-DE DIGE were also analyzed to evaluate the performance of the method in proteomics applications of complex cellular samples.

Effect of TRF-mediated background subtraction in the blue-green region A tri-exponential decay model provides a very good description of the overall fluorescence decay of in-gel measurements of Cy2-labeled protein (Figure 2C; χ2= 1.3; residual trace displayed in panel insert). The three components consisted of a very short lifetime of 0.45ns, a longer component of 4.5ns co-localizing with empty background regions, as well as a 1.06ns component co-localizing with CyDye-labeled protein spots/bands. The lifetime distribution image for the fitted dilution curve is displayed in Figure 2A, and the corresponding reconvoluted decay curves are displayed in Figure 2D. The longer component of 4.5ns was highly correlated with the acrylamide concentration when measured in gradient acrylamide



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

gels (r=0.95; Insert, Figure 2D). As demonstrated both by a cross-sectional line tracing of the fluorescence intensity (Figure 2B) as well as the protein band density (Figure 2E), the MicroTime 100 conventional fluorescent output (blue boxes) mimics that of the conventional Typhoon imager (gray triangles), with the confounding background fluorescence restricting the lowest limit of quantification (LLOQ; defined as S/N ratio >5 * SD of the background noise17), to 10 pg (=0.2 fmol) total protein (Figure 2F; gray triangles). Following lifetime fitting and subsequent TRF-mediated background subtraction (Figure 2F; red diamonds), the LLOQ in terms of total protein is extended 30-fold to 0.3 pg (=6 amol) protein. It should be noted that the axial measurement depth differed 10-fold between the Typhoon 9400 (300 µm) and the MicroTime 100 setup (28µm). To evaluate the axial distribution of the CyDye-labeled proteins, 10 sequential scans were performed with the MicroTime 100, at 50 µm increments with the first section centered 50µm form the glass plate. The fluorescent output was highly reproducible across the measured 500 µm section at all three relevant lifetimes (1.06 ns: rsd=3.5%; 4.5ns: rsd=3.4%; 0.45ns: rsd= 6.4%; Supplemental Figure 1), thereby indicating that the CyDye-labelled proteins are evenly dispersed throughout the thickness of the gel. As such, the actual difference in the number of CyDye molecules detected per pixel was 330-fold (Table 1). Conventional measurements using the MicroTime 100 setup (Figure 2D; blue boxes) indicated that the MicroTime 100 performed worse than the Typhoon instrument, with 3-fold higher LLOQ for the MicroTime 100 conventional measurements (Table 2), resulting in part from a 2-fold higher variance of the background noise from the MicroTime 100 measurements (Table 3). The LODs and LLOQs for the respective instrument setups and CyDyes are summarized in Table 2, and the background amplitudes and associated variances are summarized in Table 3.


Page 12 of 28

Page 13 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Qualitative analyses using multi-exponential reconvolution of Cy3-labelled protein revealed characteristics very similar to those of the Cy2 fluorochrome, with two lifetimes of 0.45ns and 4.5ns associated with background, and a single lifetime of 1.0ns associated with Cy3-labeled proteins. While the 485 nm laser utilized in these measurements falls well within the excitation peak of the Cy3 fluorochrome, thereby providing reliable data for qualitative purposes, the positioning at the shoulder of the excitation peak does not make it optimal for quantitative purposes. Given that the more optimal 532 nm laser was not available in the utilized MicroTime 100 setup, quantitative analyses were not performed for this fluorochrome.

Effect of TRF-mediated background subtraction in the red region A tri-exponential decay model provides an excellent description of the overall fluorescence decay also for in-gel Cy5 measurements (χ2=1.2; Figure 3C), with the optimal fit featuring two lifetimes of 1.08ns and 1.64ns associated with Cy5-conjugated proteins, and a 3rd lifetime of 0.11ns associated with background regions. The lifetime distribution image for the fitted dilution curve is displayed in Figure 3A, and the corresponding reconvoluted decay curves are displayed in Figure 3D. In terms of conventional fluorescent output, the MicroTime 100 (blue trace) resembled that of the conventional Typhoon imager (gray trace), as evidenced both by a cross-sectional line tracing of the fluorescence intensity (Figure 3B) as well as the protein band density (Figure 3E). Due to the lower level of confounding background fluorescence in the red region (Table 3), an LLOQ of 0.8 pg (=16 amol) protein was achieved with the conventional Typhoon system (Figure 3F; gray triangles). Following lifetime fitting and subsequent TRF-mediated background subtraction, the LLOQ was extended 800-fold, to 1 fg (=20 zmol) protein (Figure 3D; red diamonds). Taking into account the 10-fold difference in axial depth between the Typhoon 9400 (300 µm) and the MicroTime 100 setup (28µm), the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

actual difference in the number of CyDye molecules detected per pixel was 8000-fold (Table 2). However, part of this improvement may be due to instrument performance, as conventional measurements using the MicroTime 100 system (Figure 3F, blue boxes) reached a 20-fold better LLOQ than the corresponding measurements using the Typhoon (Figure 3F, blue boxes; Table 2).

The proximity of the two Cy5-associated lifetimes poses challenges in multi-exponential fitting. As such three alternate tri-exponential decay models, all consisting of two lifetimes of 1-2ns representing the fluorochrome (1.25ns + 1.88ns; or 1.30ns + 2.05ns) in addition to the short background-associated component, gave very similar results in terms of quantitation and goodness-of-fit. Singh previously reported a 2.0ns LT for BSA-embedded Cy5, in contrast to the well characterized LT of 1.0ns in an aqueous environment18. It is thus plausible that the two lifetimes observed here for in-gel measurements represents the protein-dense versus aqueous microenvironments reported by Singh18.

CuTEDGE analysis of complex cellular samples separated by 2-DE DIGE In order to evaluate the utility of the CuTEDGE technology in gel-based proteomics applications, reconvolution analyses and TRF-mediated background subtraction corresponding to those performed on the protein dilution series in 1D-gels were performed also for a human cell lysate separated by 2-DE DIGE. Multi-exponential fitting revealed three fluorescent lifetime components corresponding to those determined for the 1D dilution series for all three CyDyes. Visual inspection of the conventional intensity image (Figure 4A; Cy2) versus lifetime distribution images (Figure 4B; Cy2) revealed a superimposable spot pattern. In a similar fashion, image alignment analysis using the SameSpots software (TotalLab Ltd., Newcastle, U.K.) of fluorescence intensity images from conventional Typhoon image


Page 14 of 28

Page 15 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


acquisition, conventional MicroTime 100 image acquisition, and CuTEDGE images resulted in superimposable images with identical spot patterns, thereby further validating the accuracy of the TRF-mediated background subtraction also on complex cellular samples. The much higher residual background fluorescence present in the conventional images also after SameSpots background subtraction appears to have a confounding effect on the spot shape (Figure 4C). DISCUSSION The primary purpose of the current project was to perform proof-of-concept validations of the CuTEDGE technology by means of a customized MicroTime 100 instrument setup with FLIM capability (Figure 1), and compare the quantitative performance of the DIGE CyDyes in comparison to the current state-of-the-art image acquisition instrumentation. The overall performance of the CuTEDGE technology provided an extension of the sensitivity and dynamic range of detection of orders of magnitude for both Cy2 and Cy5 using the minimal DIGE labeling strategies. For Cy2 the quantitative dynamic range in terms of CyDye molecules/pixel was extended 330-fold as compared to the Typhoon 9400 reference instrument, reaching a LLOQ of 0.1 pg (=2 amol) protein/band. For Cy5, the corresponding LLOQ was extended 8000-fold to 1 fg (=20 zmol) protein.

The minimal DIGE workflow utilizes a labeling strategy in which only 2-3% of the protein molecules are conjugated with CyDye. While the minimal labeling strategy efficiently prevents spot trains or smearing that would be associated with a variant number of CyDye moieties binding to each protein species, it also causes a corresponding decrease in per-pixel fluorochrome density. Given that the total number of protein copy numbers as well as the covalently attached CyDye fluorochromes are evenly dispersed throughout the z-planes of each detected protein band (Supplemental Figure 1), the axial measurement depth of the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

respective instrument setup will also influence the actual number of CyDye fluorochromes detected in each voxel. Taking into account both the 3% minimal DIGE labeling efficiency and the 28 µm axial depth of the MicroTime 100 setup, corresponding to 3% of the total 1Dgel thickness of 1 mm, the actual number of CyDye molecules being quantified in the respective LLOQ thus corresponds to on average 3.3 CyDye molecules/pixel for Cy2, and 0.012 CyDye molecules per pixel for Cy5 (Table 1). As such, it can be concluded that the CuTEDGE technology using the current setup has achieved quantification corresponding to single-fluorochrome sensitivities, and it is evident that the number of fluorescent molecules rather than the CuTEDGE approach becomes the limiting factor in these measurements. The corresponding number of CyDye fluorochromes required for detection by the Typhoon 9400 instrument, based on the 300µm z-resolution of the instrument, amounted to 1100/pixel for Cy2, and 100/pixel for Cy5 (Table 1). The CuTEDGE technology in the current setup thereby represents an actual improvement in sensitivity of 330-8000 over existing technology. The observed improvements were achieved using significantly lower laser powers, thereby decreasing the potential for photo-bleaching during repeated scanning.

Qualitative evaluations of TRF-mediated background subtraction on complex cellular samples separated by 2-DE confirmed that the CuTEDGE technology is applicable also to 2-DE DIGE images. Reconvolution using the respective fluorescent lifetime component determined in 1-D gel images resulted in an excellent fit also for the 2-DE images, with superimposable spot patterns of conventional fluorescent intensity images and lifetime images (Figure 4). The superimposable spot patterns were further validated by SameSpots image analysis of the fluorescent intensity images of the conventional image acquisition methods (Typhoon and MicroTime 100) versus the CuTEDGE fluorescence images, thereby validating the utility of the method also for 2-DE DIGE analysis of complex cellular samples. However, the


Page 16 of 28

Page 17 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


discrepancy in the achieved sensitivity between Cy2 and Cy5 may represent a limitation in a multiplexing setting, and one might argue that detection limit of the least sensitive CyDye will mandate the overall sensitivity of the method. On the other hand, given that the CuTEDGE technology facilitates LLOQs representing detection of single-CyDye-molecules per voxel also for Cy2, the discrepancy in sensitivity is likely to be of minor importance. Given the extremely high sensitivities for Cy5, the discrepancy could be counteracted by shorter scan times for Cy5, while still maintaining single-fluorochrome LLOQs also for Cy5.

The more modest improvements in Cy2 performance compared to Cy5 are in part due to the higher variance of the background in the blue region. The open microscope setup of the MicroTime 100 may cause more leakage of ambient light compared to the closed Typhoon system, thereby contributing to a higher variance of the background noise, with associated decreases in S/N performance of the MicroTime 100 measurements. The substantially higher background noise levels of the conventional Typhoon system (Table 3) may contribute to bleed-over between adjacent pixels, with an associated binning effect that would cause an artificial lowering of the variance of the background noise, thus producing an apparent increase of the S/N ratios for the Typhoon system compared to the much more specific measurements performed with the photon counting setup. The single-fluorochrome detection limits achieved by the CuTEDGE technology will give rise to new challenges to overcome in 2-DE workflow. While the challenges arising in downstream mass spectrometry-based protein identification workflows can be overcome by pooling of protein spots from multiple 2-DE gels, the need for improved resolving powers in the 2-DE separation technology are urgently needed in order to unleash the full potential of the CuTEDGE technology. For the time being the power of the CuTEDGE technology may be best utilized to facilitate in-depth proteomics characterization of very scarce samples, e.g.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

primary human tissue materials collected through in clinical studies. The unique information provided by high sensitivity 2-DE, including positional shifts due to post-translational modifications, may increase the chance to detect biomarker signatures of relevance for identification of disease sub-phenotypes. CONCLUSION In conclusion, we have demonstrated that the introduction of time-resolved fluorescence in the image acquisition step of in-gel protein detection provides a sensitive and accurate method for subtracting confounding background fluorescence at a photon level. Proof-of-concept by means of in-gel protein detection using the minimal DIGE workflow showed improvements in LLOQ in terms of CyDye molecules/pixel of 330-fold in the blue-green area (Cy2) and 8000fold in the red region (Cy5) over conventional state-of-the art image acquisition instrumentation, here represented by the Typhoon 9400 instrument. The CuTEDGE technology thus harbors the potential to enhance the performance of the DIGE proteomics workflow to a biologically relevant sensitivity and dynamic range also for very scarce samples, thereby making possible analysis of low abundance proteins which are of great interest in biomarker discovery as well as the search for novel therapeutic targets. The custom instrument setup utilized in the presented proof-of-principle measurements is available as an add-on module compatible with most confocal microscopes, now also including the 532 nm pulsed laser diode optimal for excitation of the Cy3 fluorochrome. In addition, the scripts required for export of single-lifetime-component intensity images is available in the updated version of SymPhoTime software, thereby making the CuTEDGE technology available to the proteomics community, albeit with limitations in gel size and scan speed.


Page 18 of 28

Page 19 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ACKNOWLEDGEMENTS The research was funded by grants from the Swedish Foundation for Strategic Research (SSF), The Innovation Bridge, the Swedish Governmental Agency for Innovation Systems (VINNOVA), the Swedish Heart-Lung Foundation, EU FP6 Marie Curie International Reintegration Grant, and the Swedish Research Council.

CONFLICT OF INTEREST No part of the research presented in this manuscript has been funded by industry sources. The involvement of PicoQuant GmbH was performed on a consultancy basis, funded by academic research grants as outlined above (PI: ÅMW). The involvement of Hikari Bio AB, (founded by ÅMW) is limited to holder of the CuTEDGE patent.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

REFERENCES (1) Anderson, N. L.; Anderson, N. G. Mol Cell Proteomics 2002, 1, 845-867. (2) Viswanathan, S.; Unlu, M.; Minden, J. S. Nat Protoc 2006, 1, 1351-1358. (3) Tonge, R.; Shaw, J.; Middleton, B.; Rowlinson, R.; Rayner, S.; Young, J.; Pognan, F.; Hawkins, E.; Currie, I.; Davison, M. Proteomics 2001, 1, 377-396. (4) IUPAC. Compendium of Chemical Terminology, 2nd ed. (the "Gold Book"). Compiled by A. D. McNaught and A. Wilkinson. Blackwell Scientific Publications, Oxford (1997). XML online corrected version: http://goldbook.iupac.org (2006-) created by M. Nic, J. Jirat, B. Kosata; updates compiled by A. Jenkins. ISBN 0-9678550-9-8. (5) Wheelock, A. M.; Goto, S. Expert Rev.Proteomics. 2006, 3, 129-142. (6) Silva, E.; O´Gorman, M.; Becker, S.; Auer, G.; Eklund, A.; Grunewald, J.; Wheelock, Å. M. J Proteome Res 2010, 9, 1522-1532. (7) Patent approved in the USA (US008218878 B2: 2012), Europe (EP 2283350: 2011), Japan (5068388: 2012) and. P.R. of China (ZL200980117868.7: 2015). 2011-2015. (8) Wells, N. P.; Lessard, G. A.; Goodwin, P. M.; Phipps, M. E.; Cutler, P. J.; Lidke, D. S.; Wilson, B. S.; Werner, J. H. Nano letters 2010, 10, 4732-4737. (9) Suhling, K.; French, P. M.; Phillips, D. Photochem Photobiol Sci 2005, 4, 13-22. (10) Wallrabe, H.; Periasamy, A. Current opinion in biotechnology 2005, 16, 19-27. (11) Kohler, M.; Sandberg, A.; Kjellqvist, S.; Thomas, A.; Karimi, R.; Nyren, S.; Eklund, A.; Thevis, M.; Skold, C. M.; Wheelock, A. M. The Journal of allergy and clinical immunology 2013, 131, 743-751. (12) Koberling, F.; Buschmann, V.; Hille, C.; Patting, M.; Dosche, C.; Sandberg, A.; Wheelock, A. M.; Erdmann, R. Proc. SPIE 2010, 7569, 31-38. (13) Ortmann, U.; Wahl, M.; Kapusta, P. In Standardization and Quality Assurance in Fluorescence Measurements I, Resch-Genger, U., Ed.; Springer, 2008, pp 259-275. (14) Böhmer, M.; Pampaloni, F.; Wahl, M.; Rahn, H. J.; Erdmann, R.; Enderlein, J. Rev. Sci. Instrum. 2001, 72, 4145. (15) Koberling, F.; Wahl, M.; Patting, M.; Rahn, H. J.; Kapusta, P.; Erdmann, R. Proc. SPIE 2003, 5143. (16) Enderlein, J.; Goodwin, P. M.; Van Orden, A.; Ambrose, W. P.; Erdmann, R.; Keller, R. A. Chemical Physics Letters 1997, 270, 464-470. (17) U.S.FDA, Guidance for Industry; Bioanalytical Method Validation; 2001. (18) Singh, M. K. Physical chemistry chemical physics : PCCP 2009, 11, 7225-7230.


Page 20 of 28

Page 21 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


FIGURE LEGENDS Figure 1: Schematic sketch of the utilized MicroTime 100 system, which is based on the upright microscope BX41 from Olympus, picosecond pulsed diode lasers, photon counting detectors as well as Time-Correlated Single Photon Counting (TCSPC) electronics. Pulsed excitation light was guided via an optical fiber into the confocal microscope. Fluorescence output was guided via a multi-mode fiber to the PMT detector. Laser reference signals, detector signals and line scan reference signals from the scanner board were guided into the TimeHarp 200 time-correlated photon counting unit, which was controlled by the SymPhoTime software.

Figure 2: A tri-exponential decay model provides a very good model of the overall fluorescence decay for Cy2-labeled protein separated in a polyacrylamide gel, as shown both by the lifetime distribution image (A) as well as the goodness-of-fit for the gel image (C;χ2= 1.3; residual trace displayed in panel insert). The 3 resulting lifetime (LT) components of 0.45ns, 1.06ns, and 4.5ns are displayed in (D). The intensity (photon count) of the longest lifetime is highly correlated with the percent acrylamide of a gradient SDS-PAGE gel (r=0.95; Insert, panel D). The dynamic range (E), calculated based on the LLOQ (i.e. S/N>5; F), achieved by TRF-mediated background subtraction (red diamonds) improved >30-fold when measured in total amount of protein/band, and >330-fold calculated as the number of fluorochromes per pixel, as compared to conventional fluorescence measurements using the MicroTime 100 (blue boxes) or Typhoon (gray triangles). The line trace of the respective image acquisition methods is displayed in panel (B).



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3: The 3-component model displayed a very good fit for the overall fluorescent decay of Cy5-labeled proteins separated in an acrylamide gel, as evidenced both by the lifetime distribution image (A) as well as the goodness-of-fit for the gel image (C; χ2= 1.19; residual trace displayed in panel insert). The 3 resulting lifetime (LT) components of 0.11ns, 1.08ns, and 1.64ns are displayed in (D), with the two longer LT components representing Cy5. The dynamic range (E) achieved by TRF-mediated background subtraction of Cy5 labeled BSA protein (red diamonds) improved ∼800-fold when measured in total amount of protein/band, and >8000-fold when measured in number of fluorochromes per pixel, compared to conventional fluorescence measurements using the Typhoon (gray triangles). Based on the results from the conventional measurements using the MicroTime 100 instrument (blue boxes), part of these improvements appear to be the due to the instrumentation rather than the TRF-mediated background subtraction. The line trace of the respective image acquisition methods is displayed in panel (B).

Figure 4: The superimposable protein spot patterns of the conventional fluorescence intensity image (A) and the lifetime distribution image (B) of CyDye-labeled cell lysate separated by 2DE DIGE shows that the CyDye-labeled proteins have a fluorescent lifetime that is distinctly different from that of the surrounding background (here shown for Cy2). Downstream analysis by SameSpots 2DE analysis software further confirms identical protein spot outlines from image acquisition with conventional Typhoon instrument and CuTEDGE for both Cy2 and Cy5 (C), thereby confirming the utility of the CuTEDGE technology also on complex cellular samples separated by 2DE DIGE. The much higher residual background fluorescence on the conventional images appears to have a confounding effect on the spot shape.


Page 22 of 28

Page 23 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


TABLES Table 1. Protein amounts and number of CyDye molecules of CyDye-BSA dilution series Well BSA/well BSA/well BSA/well CyDye/well* Band area CyDye/voxel** CyDye/voxel# no. (g) (mol) (molecules) (molecules) (pixels) (Typhoon) (MicroTime100) 12

3.6*10

10

3186

3.4*10

6

3.2*10

5

1.2*10

11

3.6*10

9

2205

4.9*10

5

4.6*10

4

20 fmol

1.2*10

10

3.6*10

8

1665

6.5*10

4

6.1*10

3

2.0 fmol

1.2*10

9

3.6*10

7

1584

6.8*10

3

640

1.2*10

8

3.6*10

6

968

1.1*10

3

100

1.2*10

7

3.6*10

5

924

120

11

1.2*10

6

3.6*10

4

924

12

1.1

1.2*10

5

3.6*10

3

820

1.3

0.12

0.13

0.012

2.0 pmol

1.2*10

2

0.10 µg 10 ng

0.20 pmol

3

1.0 ng

4

0.10 ng

1

5 6 7 8

10 pg 1.0 pg 0.10 pg 10 fg

0.20 fmol 20 amol 2.0 amol 0.20 amol

4

1.0 fg 20 zmol 1.2*10 360 817 9 * Based on 3% labeling efficiency of minimal DIGE ** Based on 300 µm axial measurement depth of Typhoon 9400 # Based on 28 µm axial measurement depth of MicroTime 100 setup

Table 2: Summary of LOD and LLOQ for detection of pre-labeled BSA in polyacrylamide matrix Typhoon: MT100: MT100: Fold improvement in LLOQ Conventional Conventional CuTEDGE Typhoon vs CuTEDGE CyDye LOD LLOQ LOD LLOQ LOD LLOQ g prot/band CyDye-molecules/pixel Cy2

5 pg

10 pg 40 pg 30 pg

Use of Time-Resolved Fluorescence To Improve Sensitivity and

Recommend Documents