simmons - 2016

SC I ENCE S I GNAL ING | R E S EARCH RE SOURCE

TECHN IQUES

1Epithelial Biology Center, Vanderbilt University Medical Center, Nashville, TN 37232,USA. 2Department of Cell and Developmental Biology, Vanderbilt University School ofMedicine, Nashville, TN 37232, USA. 3Department of Medicine, Vanderbilt UniversityMedical Center, Nashville, TN 37232, USA. 4Chemical and Physical Biology Program,Vanderbilt University Medical Center, Nashville, TN 37232, USA. 5Department ofCancer Biology, Vanderbilt University School of Medicine, Nashville, TN 37232, USA.6Department of Pathology, Microbiology, and Immunology, Vanderbilt UniversityMedical Center, Nashville, TN 37232, USA. 7Veterans Affairs Medical Center, Tennes-see Valley Healthcare System, Nashville, TN 37232, USA.*Corresponding author. Email: [email protected]

Simmons et al., Sci. Signal. 9, rs11 (2016) 11 October 2016

2016 © The Authors,

some rights reserved;

exclusive licensee

American Association

for the Advancement

of Science.

http://stD

ownloaded from

Impaired coordination between signaling pathways isrevealed in human colorectal cancer using single-cellmass cytometry of archival tissue blocksAlan J. Simmons,1,2 Cherié R. Scurrah,1,2 Eliot T. McKinley,1,3 Charles A. Herring,1,4

Jonathan M. Irish,5,6 M. Kay Washington,6 Robert J. Coffey,1,2,3,7 Ken S. Lau1,2,4*

Cellular heterogeneity poses a substantial challenge to understanding tissue-level phenotypes and confoundsconventional bulk analyses. To analyze signaling at the single-cell level in human tissues, we applied mass cytom-etry using cytometry time of flight to formalin-fixed, paraffin-embedded (FFPE) normal and diseased intestinalspecimens. This technique, called FFPE-DISSECT (disaggregation for intracellular signaling in single epithelial cellsfrom tissue), is a single-cell approach to characterizing signaling states in embedded tissue samples. We appliedFFPE-DISSECT coupled tomass cytometry and found differential signaling by tumor necrosis factor–a in intestinalenterocytes, goblet cells, and enteroendocrine cells, implicating the downstream RAS-RAF-MEK pathway indetermining goblet cell identity. Application of this technique and computational analyses to human colon speci-mens confirmed the reduced differentiation in colorectal cancer (CRC) compared to normal colon and revealedincreased intratissue and intertissue heterogeneity in CRC with quantitative changes in the regulation of signalingpathways. Specifically, coregulationof the kinases p38 andERK, the translation regulator 4EBP1, and the transcriptionfactor CREB in proliferating normal colon cells was lost in CRC. Our data suggest that this single-cell approach,applied in conjunction with genomic annotation, enables the rapid and detailed characterization of cellular heter-ogeneity from clinical repositories of embedded human tissues. This technique can be used to derive cellular land-scapes from archived patient samples (beyond CRC) and as a high-resolution tool for disease characterization andsubtyping.

ke.s
on O
ctober 14, 2016ciencem

ag.org/

INTRODUCTIONA distinguishing feature of cancer and other diseases of dysregulatedhomeostasis is the expanded degree of intratissue cellular heterogeneity(1–4). Heterogeneous cell populations arise from an aberrant differen-tiation process where cells adopt semimature or new progenitor stateson theWaddington landscape (5). Cellular heterogeneity has been dem-onstrated to present a significant challenge for treating these diseases, astherapies targeting one cell typemay not be effective in another (6). Fur-thermore, rare cell populations, such as cancer stem cells (7, 8), canadopt specialized, deleterious functions, including therapeuticresistance and metastatic ability (9–13). The phenotypic state of a cellis governed by its genetics and environment; information from thesesources is integrated by signaling and transcriptional networks intocellular behaviors. Investigations of cellular heterogeneity immenselybenefit from single-cell analysis (14, 15). However, it is not trivial tointerrogatemultipathway signaling activities at single-cell resolution be-cause cellular signaling states can be destabilized outside the nativetissue context (16–18).

A tried and true approach for preserving tissue morphology, andeven cellular signaling states, is the procedure of formalin fixationcoupled to paraffin embedding (FFPE). FFPE has been a standard prac-tice in clinical analysis of tissues for nearly a century, and its ability to

preserve tissues at ambient temperatures has been widely demonstrated(19). Because of the effectiveness of FFPE for preserving tissue, largerepositories of clinically annotated patient samples have been collectedover the years. These banks are valuable resources for scientific insightwhen coupled to next-generation analytical approaches (20, 21). Specif-ically, one of our goals is to conduct a single-cell signaling analysis onFFPE tissues to address cellular heterogeneity. To achieve this, we musttake careful measures to undo the effects of formalin cross-linking toaccess cells, proteins, and nucleic acids for sophisticated analyses.

To comprehensively assess the phenotypic state of cells, evaluatingthe activity of a single pathway is not sufficient. Recently, severalapproaches have been described formeasuring protein parameters fromFFPE tissue in a multiplex fashion (22). Most of these advances havebeen microscopy-based approaches for imaging tissue sections that are~5 mm in thickness. Approaches that enable multiplexing protein mea-surements include iterative rounds of fluorescence imaging (23–26)or metal-based detection (27, 28). To achieve single-cell resolution,single or multiple cell border markers are used in conjunction withsophisticated image processing algorithms to extract single-cellobjects from images (29). Oblique sectioning and imperfect segmen-tation of partial cells can lead to inaccurate quantification, makingthese approaches semiquantitative at best. Furthermore, because ofeither the iterative nature of cyclic immunofluorescence (IF) or ras-tering of samples for imaging mass spectrometry, these approachesare low throughput and require multiple days or weeks of analysis tofully sample a given specimen. Given their space-resolving capabil-ities, we surmise that these techniques will be very powerful whencombined with a primary strategy that confers feasibility to analyzea large number of samples with higher quantitative accuracy.

Our laboratoryhas recently reported a relatively rapidmass cytometry–based strategy for profiling signaling protein modifications at the

1 of 12

http://stke.sciencemag.org/


Dow

nloaded from

single-cell level from solid tissues (16). This strategy, namedDISSECT(disaggregation for intracellular signaling in single epithelial cells fromtissue), involves rapid, short fixation of freshly isolated tissue to main-tain native signaling in intact epithelia and then a series of coupledprocedures for staining and dissociation before mass cytometry anal-ysis. The present study examines whether the same approach can beapplied to FFPE-preserved tissues, given that FFPE preservation alsoinvolves the use of a formaldehyde fixative. Here, we present an opti-mized procedure for dissociating single cells fromFFPE-preserved sol-id tissues while maintaining their intact signaling states for masscytometry analysis.We conducted a proof-of-concept study on a smallof cohort of human normal colon and colorectal cancer (CRC) FFPEspecimens to sample signaling pathway heterogeneity at the single-celllevel. Our results indicate that in normal colonic tissues, signalingpathways are organized into modules according to surface-to-cryptdifferentiation status. This modular organization is undermined inCRC. In addition, examining tumor samples in combination with ge-nomic markers, such as microsatellite instability and mutational status,reveals distinct single-cell cancer phenotypes. This hypothesis-generatingstudy demonstrates FFPE-DISSECT coupled to mass cytometry analysison archival tissues, with the aimof extending to large cohort studies fromsolid tumor repositories that classify tumors in conjunction with ge-nomic, transcriptomic, epigenomic, and proteomic data.

on October 14, 2016


RESULTSThe FFPE-DISSECT method disaggregates single epithelialcells from archived tissue blocks while preserving cell-typeand signaling markersWe established a single-cell disaggregation approach for FFPE tissues(FFPE-DISSECT) combining heat-induced antigen retrieval with thewhole-mount staining and dissociation steps of DISSECT (Fig. 1)(16, 30). The steps of DISSECT were incorporated to enable epithe-lial signaling state preservation during the disaggregation process.We confirmed single-cell retrieval from FFPE tissues by bothbright-field and autofluorescence imaging (fig. S1, A and B). As withDISSECT, because tissue was kept intact until the end of the protocol,cell loss due to preanalytical processing was minimized. Thus, we rou-tinely yielded 5000 to 10,000 cells (7503 ± 2830 cells) per square milli-meter of tissue from a single 50-mm section. From the approximate areaoccupied per sample, we estimated that we yielded routinely on theorder of a million cells per 50-mm tissue section.

We first determined the preservation of cell identity markers forclassifying epithelial cell types using our approach on murine intestinaltissue embedded by FFPE. Up until tissue dissociation, chloride channelaccessory 1 (CLCA1) and cytokeratin 18 (CK18), markers for gobletand secretory cells, displayed substantial colocalization in whole-mountimmunofluorescent staining, as expected (fig. S2). Doublecortin-like ki-nase 1 (DCLK1), a marker of tuft cells, labeled a separate populationof CK18−/CLCA1− cells. Upon dissociation, these relationships re-mained intact in single epithelial cells (Fig. 2A). Furthermore, thecorrect subcellular localization of proteins within cells can be visua-lized in the absence of scattered light or convolution from neighbor-ing cells, namely, CK18 staining of cytoskeletal structures and CLCA1staining of mucous granules (Fig. 2B). We then quantitatively verifiedmarker coexpression using multiparameter flow cytometry. CK18+ andCLCA1+ cells were independently gated. Back-gating of CK18+ andCLCA1+ cells revealed that they largely fell within an overlapping pop-ulation, with CK18 marking a wider population of cells because of its


ability to label other cells in the secretory lineage (Fig. 2C and fig. S3A).These results demonstrated that cell types can be discerned indissociated epithelial cells after FFPE-DISSECT.

We determined that single cells retained their native signaling statesafter dissociation using FFPE-DISSECT. To activate signaling pathwaysin vivo, we exposed the murine intestinal epithelium to tumor necrosisfactor–a (TNF-a) via intravenous administration, as we have done pre-viously (31, 32). Duodenal tissues from the same animal were assessedas FFPE tissue sections or as single cells generated by FFPE-DISSECT.IF imaging of tissue sections revealed that the abundance of phosphoryl-ated (p-) cJUN (an early TNF-a–induced signal) was up-regulated at0.5 hours after TNF-a exposure, and p-STAT3 (signal transducer andactivator of transcription 3) (a late signal) was up-regulated at 2 hoursafter TNF-a exposure (Fig. 2D). Imaging of single-cell suspensionsprepared by FFPE-DISSECT from serial sectioning of the same tissueblock also revealed activation of the two signaling pathways at the ap-propriate time points compared to vehicle control (Fig. 2D). We quan-titatively compared signaling data from single-cell suspensionsprepared by the validated DISSECT approach from freshly isolated tis-sues (16) with those prepared by FFPE-DISSECT from embedded tis-sues. Using the median intensity calculated from single-celldistributions evaluated by flow cytometry (fig. S3B), we confirmed thatboth DISSECT and FFPE-DISSECT generated comparable signalingdata for both p-cJUN and p-STAT3 with similar dynamics (Fig. 2E).These results demonstrated the ability of FFPE-DISSECT in preservingsignaling states of p-cJUN and p-STAT3 in single epithelial cells disag-gregated from FFPE tissues.

Quantitative single-cell–level data are obtained throughmass cytometry signaling analysis on FFPE tissueIn clinical practice, excised tissues requiring gross pathological exam-ination may not be immediately fixed. Reports have documented theeffects of ischemia and other factors on the degradation of proteinsignals in other tissues such as the breast (33). To examine the effects

Fig. 1. Schematic of the FFPE-DISSECT process for preserving native epithelialsignaling. Thick (50-mm) tissue slices were sectioned from FFPE blocks, antigen-retrieved, and then processed by following the steps of the DISSECT procedure. RT,room temperature; PFA, paraformaldehyde.

2 of 12



on October 14, 2016


Dow

nloaded from

of post-excision time outside the body on signaling in the intestinalepithelium, we harvested intestinal tissues from mice and fixed thetissues either immediately, 30 min after excision, or 1 hour after ex-cision. After standard FFPE processing, we examined changes inconstitutively active signaling pathways at homeostasis, for example,the abundance of p-ERK (extracellular signal–regulated kinase) inthe crypt and p-S6 at the tip of the villus. We performed such anal-ysis for markers across a wide breadth of signaling pathways (fig. S4)that we then examined in human patients (figs. S10 to S14). For theintestinal epithelium, there was minimal degradation of these signals


for up to 1 hour from the time ofharvesting the tissue. We further veri-fied that the length of fixation time, upto 72 hours, has minimal effect on thedetection of representative signalingmarkers in FFPE (fig. S5).

Having successfully assessed thevalidity of FFPE-DISSECT on selected sig-naling markers, we next sought to sys-tematically and quantitatively validateour approach over a broad range of sig-naling pathways. We used mass cytom-etry as a multiplex technique to quantifya broad range of signaling markers fromsingle-cell suspensions, comparing be-tween FFPE-DISSECT preparations fromembedded tissues and DISSECT prepara-tions from fresh tissues. Single-cell sig-naling data obtained by DISSECT havepreviously been rigorously validatedagainst those generated by conventionalbulk approaches such as immunoblotting(16). Mice were stimulated with TNF-a,and duodenal tissues were harvested overa time course to sample a quantitativerange of signaling activities as a functionof phosphorylated protein abundance.Harvested tissues were then divided, ei-ther to be freshly processed by DISSECTor to be embedded and then processedby FFPE-DISSECT. Mass cytometryanalysis was performed on both sets oftissues (isolated from the same animal)using the same panel of metal-conjugatedreagents for signaling markers (table S1).The normalized median intensities ofdistributions of signaling markers wereused as a direct comparison betweenDISSECT and FFPE-DISSECT prepara-tions (Fig. 3). The DISSECT approachwas optimized for scraped mucosa, andthus, the data generated were enrichedfor villus signals. In contrast, tissuesectioning enabled sampling of the entireepithelium for FFPE-DISSECT. Thus,crypt-enriched signals, such as p-RB (ret-inoblastoma protein), p-4EBP1, and p-p38(fig. S6A), did not show good concor-dance between the two methods be-

cause of the de-emphasis of crypt signals in DISSECT preps (Fig. 3A).Crypt proliferative signals (p-4EBP1 and p-RB) generated by FFPE-DISSECT showed an initial dip and a subsequent increase afterTNF-a exposure, mirroring the proliferative response of the intestinalepithelium to TNF-a (32). Examining villus-enriched signals, a strongcorrelation between the data generated by DISSECT and FFPE-DISSECT was observed (Fig. 3B). Quantitative correlation analysesusing villus-enriched signals resulted in a highly significant correlation(R = 0.85, P < 0.0001) of mass cytometry data generated by DISSECTagainst FFPE-DISSECT (Fig. 3C). Including crypt-enriched signals

Fig. 2. FFPE-DISSECT enables the identification of cell types and quantification of phosphoprotein signalingactivities. (A) IF imaging of dissociated cells from FFPE murine intestinal tissues prepared by FFPE-DISSECT, stainedfor cell-type markers CK18, CLCA1, and DCLK1. Scale bar, 50 mm. (B) IF imaging of a single epithelial cell stained fornucleic acid, CK18, and CLCA1. Scale bar, 10 mm. (C) Flow cytometry biplots of the mouse ileum prepared by FFPE-DISSECT. Manual gating of goblet cells by CK18 and CLCA1 and of tuft cells by DCLK1. CK18 and CLCA1 singularpositive cells are back-gated to a biaxial plot not used for the original gating to demonstrate that the cells comprisean overlapping goblet cell population. (D) IF imaging of intact FFPE intestinal tissues as 5-mm sections, compared tosingle cells prepared by FFPE-DISSECT, stained for p-cJUN (early signal) and p-STAT3 (late signal) in response to TNF-aat the indicated time points. Scale bars, 20 mm. (E) Quantification of p-cJUN and p-STAT3 from single-cell suspensionsgenerated from murine duodenal tissues, prepared immediately by DISSECT (green), or FFPE-embedded and then byFFPE-DISSECT (magenta), followed by flow cytometry. Median intensities calculated from single-cell distributions aredisplayed for comparisons. Tissues were harvested at specified time points after TNF-a administration. Data aremeans ± SEM from n = 3 animals. Data scales are Z score values derived from mean centering and variance scalingof each set of time course experiment. Data are representative of n = 3 animals.

3 of 12



on October 14, 2016


Dow

nloaded from

resulted in a slightly lower correlation (R = 0.76, P < 0.0001) (fig. S6B).We further verified FFPE-DISSECT and compared median signals ob-tained to those obtained by IF imaging (fig. S6, C and D) and quan-titative immunoblotting (fig. S6, E and F), comparing across differentcohorts of mice similarly stimulated with TNF-a as a time course.Again, FFPE-DISSECT compared favorably. Using FFPE-DISSECTin conjunction with mass cytometry, valid, single-cell level signalingdata can be obtained from embedded epithelial tissues.

Cell type–specific signaling reveals increased secretory cellsensitivity to basal and TNF-a–induced signalingIn addition to examining the average over epithelial distributions, wesought to determine howdifferent cell populations in the small intestinerespond to TNF-a using our single-cell approach. TNF-a triggers apo-ptosis and extrusionof duodenal epithelial cells uponhours of induction(31, 32, 34, 35). Our previous study demonstrated that the onset ofapoptosis occurs 1 hour after the intravenous administration of exog-enous TNF-a in mice, and thus, mass cytometry data enabled byFFPE-DISSECT were obtained from murine duodenal tissues at thistime point. t-Distributed stochastic neighbor embedding [t-SNE (36)]analysis on 15-channel signaling and cell identity data revealed a CC3+

population of apoptotic epithelial cells (Fig. 4A; for markers, see table


S1). This dying cell population has adistinct signaling signature, includingthe down-regulation of p-ERK and up-regulation of p-p38 (Fig. 4, A and B, andfig. S7), as reported previously.We previ-ously showed that p-ERK up-regulationin neighboring cells surrounding the ap-optotic cell is a contact-dependent sur-vival mechanism preventing large-scalebarrier defects in the gut (16). We thenfurther evaluated cell type–specific sig-naling by integrating signals from the en-tire TNF-a time course in murine cellpopulations expressing cell type–specificmarkers [CLCA1+ goblet cells, CHGA+

(chromogranin A) enteroendocrine cells,and CK+/CLCA1−/CHGA− enterocytes].Goblet cells generally have increasedsignaling across most pathways assayed,whereas enteroendocrine cells selectivelyup-regulate certain pathways when com-pared to enterocytes (Fig. 4C and fig.S8A). The relative differences in signalingbetween cell types can be reproduced byDISSECT on freshly isolated tissue, againconfirming the validity of our new ap-proach (Fig. 4Cand fig. S8A).Furthermore,the up-regulation of p-ERK, p-ATF2, andp-4EBP1 in goblet cells, and of only p-ATF2 in enteroendocrine cells, was cor-roborated by IF imaging (Fig. 4D andfig. S8B). These differences were also ob-served at the basal level without TNF-astimulation, perhaps demonstrating theimportance of these signaling pathwaysin the identity of these cells (fig. S8C).Hereafter, we focused on the role of p-

ERK in goblet cell identity.MEK [mitogen-activated protein kinase (MAPK) kinase]-ERK

signaling is canonically activated by upstream RAS activation. Themembers of the RAS family of small GTPases (KRAS4A, KRAS4B,NRAS, and HRAS) share N-terminal sequence identity and in vitro ef-fector binding but have distinct subcellular membrane distribution dueto differences in posttranscriptional modifications in their C-terminalhypervariable regions (37). Thus, different RAS isoforms can engagein different signaling effectors, such as RAF (rapidly accelerated fibro-sarcoma), PI3K (phosphatidylinositol 4,5-bisphosphate 3-kinase), andRAL (RAS-related protein), which can lead to different phenotypicmanifestations. Mutationally activated KRAS in the intestinal epitheli-um induces hyperproliferation, whereas activated NRAS does not (38).Given that both activated KRAS (38) andNRAS (34, 39) in the intestinecan sensitize downstreamMEK-ERK toward activation in different cir-cumstances, we surmise thatMAPK-induced goblet cell identitymay bea common feature of RAS activation. Villin-Cre driving an activatedKRAS (KRasLSL-G12D/+) allele in the murine intestinal epithelium notonly increased the number of goblet cells (fig. S9) but also induced hy-perplasia as documented previously (38, 40). The same inductionscheme with activated NRAS (NRasLSL-G12D/+) did not result in hyper-plastic growth. NRAS activation led to a similar increase in goblet cells

Fig. 3. Comparison between mass cytometry data generated by FFPE-DISSECT and the validated DISSECTmethod on the same intestinal tissue. (A and B) Dynamic signals of TNF-a stimulation time courses, and enrichedin either crypts (A) or villus (B) from murine duodenal harvested from specified time points after TNF-a administration.Tissues were split in two, with one set processed immediately by DISSECT (green), and the other set FFPE-embedded andthen processed by FFPE-DISSECT (magenta). Both sets of tissues were analyzed by mass cytometry with the same cross-reacting signaling antibody panel. CC3, cleaved caspase-3; RSK, ribosomal protein S6 kinase; ATF2, activating transcriptionfactor 2; CREB, adenosine 3′,5′-monophosphate response element–binding protein. (C) Correlation analysis combining allvillus signaling markers, comparing mass cytometry data generated by DISSECT against FFPE-DISSECT. Quantitative datafrom different time points were used to generate a range of variation for correlation analysis. Data are means ± SEM fromn = 3 animals. Data scales are Z score values derived from mean centering and variance scaling of each set of time courseexperiment. ****P ≤ 0.0001, by t test.

4 of 12



on October 14, 2016


Dow

nloaded from

(fig. S9) in the murine intestinal epithelium, a phenotype that has notpreviously been associated with NRAS activation. Furthermore, in ac-cordance with the role of p-ERK in promoting enterocyte survival, gob-let cells have been shown to be resistant to TNF-a–induced apoptosis(16). We have demonstrated here that mass cytometry results fromFFPE-DISSECT corroborate with conclusions drawn from fresh tissueassays and produce biological insights, supporting its feasibility for gen-erating meaningful single-cell signaling data from embedded tissues.

Human CRCs present with dysregulated signalingand differentiationOne of the goals for FFPE-DISSECT application to embedded tissue isto enable single-cell signaling analysis on human patient tissue repos-itories stored as FFPE blocks. To that end, we procured a cohort ofclinically annotated colonic tissue samples from the Western Divisionof the Cooperative Human Tissue Network (CHTN), situated at theVanderbilt University Medical Center. After discarding samples withlow cellularity (<10%), our cohort included 7 normal colon controlsamples and 13 [6 microsatellite instable (MSI) and 7 microsatellitestable (MSS)] primary CRC samples. Control colon samples werecollected from a variety of conditions unrelated to CRC (such as adja-cent normal tissue from diverticulitis samples). According to our time-to-fixation optimization, we only selected samples with a post-excisiontime of <1 hour, a parameter tracked by the CHTN. Clinical and path-ological attributes of the CRCs, including microsatellite instability andKRAS/BRAF mutational status, were summarized (table S2). A board-certified pathologist further examined the hematoxylin and eosin stainsof these samples to confirm tumor histology (fig. S10). A panel ofcross-reacting antibodies against signaling proteins and cell-type


markers were prepared for mass cytometry analysis (table S3). Thesereagents were verified to stain human tissues by IF imaging (fig. S11)and to be on-target in a previous report (16). Mass cytometry afterFFPE-DISSECT was performed on this cohort of human colon andCRC samples. Because tumor tissue comprised only a minor propor-tion of each tissue section (fig. S10), we decided to focus specificallyon epithelial cells that are marked and can be gated by pan-CK (PCK)abundance (fig. S12, A and B). From mass cytometry data, we quan-titatively assessed the percentage of different epithelial cell types invarious differentiation states from normal colon versus CRC tissueswithin the epithelial compartment. As expected, terminally differ-entiated cells (CK20+) were significantly decreased in CRC comparedto normal colon (Fig. 5A and fig. S12A). Furthermore, goblet cells(CLCA1+) and enteroendocrine (CHGA+) cells were also significantlydecreased (Fig. 5, B and C, and fig. S12C). However, we discoveredthat a portion of protein markers representing signaling pathway ac-tivation were down-regulated in CRC (Fig. 5D and fig. S11). This re-sult was paradoxical given that cancer is often driven by mutationsthat ultimately activate signaling pathways. However, there is evidencefrom in vivo studies that demonstrate the up-regulation of negativefeedback mechanisms when MAPK signaling pathways are mutation-ally activated, only in the context of CRC (41). For instance, mutationalactivation of KRAS in CRC paradoxically results in the down-regulationof p-ERK due to the up-regulation of MAPK phosphatase 3 ERKphosphatase (38). Furthermore, as shown in our mouse studies, thereare substantial signaling activities in differentiated cells, and these cellsare largely absent in CRC (Fig. 4, C and D). To verify that the reduc-tion in signaling of these pathways did not result from poor penetra-tion of the fixative, we were able to detect similar stain intensities of

Fig. 4. Cell-specific signaling in the murine duodenal epithelium. (A) t-SNE analysis of mass cytometry data from the mouse duodenum exposed to TNF-a for1 hour, prepared by FFPE-DISSECT. Color overlaid represents the relative quantification of CC3, p-ERK, and p-p38 events, respectively. Labeled cells are as follows:apoptotic, CC3+ (3.07%); transit-amplifying (TA) cells, p-4EBP1+ (4.13%); goblet, CLCA1+ (5.44%); enterocytes, CKAE+, CLCA1−, and CHGA−. Numbers on the right axisrepresent minimum and maximum values of the color scale. (B) Biplots of CC3 with p-ERK or p-p38, demonstrating negative correlation in the former and positivecorrelation in the latter. (C) Signaling specific to epithelial cell types (enterocyte, CKAE+, CLCA1−, and CHGA−; goblet, CLCA1+; enteroendocrine, CHGA+) calculated byintegrating signal values over the entire TNF-a time course, comparing mass cytometry data generated by DISSECT against FFPE-DISSECT. Data are means ± SEM from n = 3animals. Data scales are Z score values derived from mean centering and variance scaling over data values for the three cell types for each method. (D) IF imaging to confirmcell type–specific signals (p-ERK, p-ATF2, and p-4EBP1) at baseline (unstimulated). Scale bar, 20 mm. Data are representative of n = 3 animals.

5 of 12



on October 14, 2016


Dow

nloaded from

multiple signaling markers in the peripheral and central regions of thesame tumors, for tumors displaying positive signals (fig. S13). Theseresults, all obtained from one sampling of tissue, suggested that differ-entiation is impaired in CRC, and these changes are associated withreduced signaling through certain pathways.

Modular organization of signaling pathways is disrupted inhuman CRCUsing t-SNE analysis to visualize multidimensional single-cell datafrom normal and CRC tissues, we observed defined organization ofsignaling pathways in normal colon tissues at the single-cell level.The abundances of phosphorylated signaling proteins in differentpathways formed distinct patterns on t-SNE maps (Fig. 5E; formarkers, see table S3); in one specimen, signaling markers formeda counterclockwise arrangement in association with surface-to-crypt status marked by CKs. These patterns can be broken downinto a modular architecture: p-cJUN correlated with CK20+ differ-entiated cells; PCK, p-S6, and p-RSK shared similar expressionpatterns; and p-ERK, p-p38, p-4EBP1, and p-CREB formed anoth-er module correlating to less differentiated crypt cells. These mod-ules can also be revealed by calculating the pairwise correlationbetween signaling markers over individual cells and by using cor-relative distances for hierarchical clustering per sample (Fig. 6A).Qualitatively, the components within each module were consistentbetween normal colon samples, signifying robust organization ofsignaling pathways between cellular populations. The correlationbetween signaling pathways over single cells was reduced in both


MSS and MSI CRC samples, signifying the usage of heterogeneousmodes of signaling pathway regulation between individual cells in atumor (Figs. 5E and 6A). Furthermore, modular organization ofsignaling pathways from hierarchical clustering was not preservedfrom sample to sample in CRC, implying significant intertumoralheterogeneity in signaling pathway regulation (fig. S14A). In placeof qualitative assessment of signaling heterogeneity on a sample-per-sample basis, we quantitatively assessed intratumoral and intertumoralheterogeneity over all the samples. We evaluated intratumoralsignaling regulation by quantifying the magnitude of correlation be-tween all signaling markers measured in a pairwise fashion, withthe notion that high correlations signify similar regulatory mecha-nisms between any pair of pathways used by all cells. On a per-samplebasis, this metric can be represented by the total intensity on a correl-ative distance heat map (Fig. 6A). Normal colon samples had signif-icantly greater total correlation between signaling markers than didMSS and MSI CRC samples (Fig. 6B), denoting the loss of signalingregulation homogeneity between cells in CRC. We quantitatively eval-uated intertumoral signaling heterogeneity by assessing the degree bywhich signaling modules are similar between samples. For this, wetook advantage of tools built previously to assess the similarity betweendendrograms to evaluate the similarity between the structures of hi-erarchical clustering trees (42). We used Baker’s g correlation coeffi-cient (43), a metric that is insensitive to the height of the branchesbut is sensitive to the position of each branch, to calculate pairwisesimilarities between hierarchical trees generated for each sample(fig. S14A). The mean g correlation coefficient showed a significant

Fig. 5. Mass cytometry analysis of human CRC specimens prepared by FFPE-DISSECT. (A to C) Percentage of CK20+ fully differentiated epithelial cells (A), CLCA+ goblet cells(B), and CHGA+ enteroendocrine cells (C) in samples of normal human colon compared to samples of human CRC. Data aremeans ± SEM from n > 7 different patient specimens.Inset depicts manual gating of differentiated cells by CK20. ***P ≤ 0.001, ****P ≤ 0.0001, by t test. (D) IF imaging of signaling markers (p-CJUN, p-S6, p-ERK, p-p38, and CC3)comparing normal colon andCRC. Scale bar, 100 mm. (E) t-SNEmapping ofmass cytometry data generated fromhuman colon,MSS, orMSI CRC specimens, overlaidwith signalingand selected differentiation markers. Numbers on the right axis represent minimum and maximum values of the color scale. The same scales were used between all samples.Proportional downsampling to 20,000 cells was performed for more equivalent representation because some samples have a small representation of actual tumor cells. Onaverage, 60,000 cell events were collected per sample. Data are representative of n > 6 human patient samples for each group (colon, MSS-CRC, and MSI-CRC).

6 of 12



on October 14, 2016


Dow

nloaded from

decrease in value in MSS and MSI CRC samples compared to normalcolon samples, suggesting that similar signaling modules recurrentlyexist across different normal colon samples, but less so in CRCs (Fig.6C). Using tissue-level data to perform the same analysis resulted indifferent interpretations, again, with normal colon samples havinghigh correlation between sets of signaling markers, MSS sampleshaving less correlation than normal colon tissue, and MSI sampleshaving a correlation between different sets of markers (fig. S14B).This difference may be due to the loss of single-cell resolution wheremarkers expressed in different cells are considered to be in the samecompartment as a sample average. These results demonstrated, in aquantitative fashion, that (i) cells within normal colon have sharedregulatory mechanisms between pathways but cells in CRC samplesdo not (intratumoral heterogeneity), and (ii) organized signaling mod-ules recurrently exist between normal colon samples but not betweenCRC samples (intertumoral heterogeneity).

We next examined whether single-cell signaling properties of tu-mors are associated with molecular characteristics (table S2). None ofthe four MSI tumors in our set with a BRAFV600E mutation presentedwith CK20+ differentiated cells, whereas all other nine tumors (MSS orMSIwithwild-typeBRAF) presentedwith somedegree of differentiation


(Fig. 6, D and E). Furthermore, all MSI-BRAF mutant tumors (four offour) had somedegree of CC3+ apoptotic cells, whereas only a small pro-portion (three of nine) of other tumors exhibited this phenotype (Fig. 6,D andE). ForMSS tumors specifically, aG12mutation (G12VorG12C)in KRAS down-regulated the abundance of p-ERK, increased cell prolif-eration (Ki67), and up-regulated CLCA1+ goblet cell specification com-pared to tumors with wild-type KRAS (Fig. 6F). These results providedevidence that genetic properties, such as microsatellite instability andmutations, but not pathologic details, such as grade and stage of the tu-mor, correlate with single-cell signaling phenotypes in CRC.

DISCUSSIONThere is an ongoing effort to use next-generation genomic, epigenomic,transcriptomic, and proteomic data to predict tumor outcomes andresponses to therapy (44). However, the degree of behavioral diversitywithin a tumor may be just as important, because different cellularpopulations may respond to drugs differently and cooperate to produceemergent behaviors. FFPE-DISSECT enables the analysis of single-cellsignaling activities in archival human tissues. Whereas large academiccenters have access to various methods for human tissue preservation,

Fig. 6. Insights into the heterogeneous organization of signaling pathways in CRC from single-cell data. (A) Heat map and hierarchical clustering derived from pairwisecorrelative distances between signalingmarkers calculated over all single cells in a sample. A highpairwise correlation signifies that twopathways are regulated in the sameway inall cells. Data are representative of n > 6 human patient samples for each group (colon, MSS-CRC, andMSI-CRC). (B) Themean value of all pairwise correlations between signalingmarkers calculated per sample, comparing between normal colon,MSS, andMSI. Data aremeans ± SEM from n>6different patient specimens. (C) Baker’s g correlation coefficientcomparing the similarity between hierarchical clustering trees computedbetween all sampleswithin each group (colon,MSS, andMSI). Data aremeans ± SEM from n>6differentpatient specimens. (D) t-SNE maps of mass cytometry data generated from an MSI-BRAFV600E mutant tumor compared to MSS or MSI-BRAF wild-type (WT) tumors overlaid withthe abundance of CK20 andCC3. Numbers on the right axis representminimumandmaximumvalues of the color scale. The same scales were used between all samples. Numberof cells was noted from each single tumor. Data are representative of n = 4MSI-BRAFV600E mutant tumors, n = 7MSS tumors, and n = 2MSI-BRAFWT tumors. (E) Percent CK20+ orCC3+ cells comparing MSI-BRAFV600E mutant tumors (n = 4) compared to tumors of other genotypes (n = 9). The dashed line represents the 2% threshold, and the inset is thenumber of samples passing the threshold. (F) t-SNEmaps of mass cytometry data generated from an MSS-KRASG12C mutant tumor compared to an MSS-KRASWT tumor overlaidwith abundances of p-ERK, Ki67, and CLCA1. Data are representative of n = 2 KRAS mutant and n = 2 KRASWT tumors. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, by analysis of variance(ANOVA) followed by Tukey post test (B and C) or t test (E).

7 of 12



on October 14, 2016


Dow

nloaded from

such as flash freezing, most community hospital settings only haveaccess to FFPE. These untapped, large human tissue resources cannow bemined at the single-cell level for building appropriately poweredmodels to inform how heterogeneity contributes to tumor behavior andhow cellular diversity changes in response to treatment.

There are several caveats to using FFPE-DISSECT, which fall underthe same limitations as other FFPE applications. First, the range of anti-bodies that actually work in FFPE tissues is reduced compared to freshlyisolated tissues, because not all conformationally blocked antigens canbe retrieved. We somewhat alleviated this problem by only using anti-bodies that arewell validated (such as by knockdown in human cell linesormouse tissues) and are widely used in the field for FFPE applications.This problem can perhaps be further addressed in the future by betterantibody generation practices. For example, a higher success rate for thegeneration of antibodies for FFPE applications may be achieved byusing fixed proteins as immunogens instead of native peptides. Second,the veracity of a stain of human FFPE section due to tissue degradationcomes into question. The preservation of signals, specifically ofsignaling proteins, is sensitive to the amount of time the tissue has beenoutside the body (33). Furthermore, storage conditions of FFPE block,such as temperature and humidity, may introduce variability in theresults (45). Standardized operating procedures regarding post-excisiontime, fixation, and storage, such as those adopted by the CHTN, arerequired to decrease the variability introduced during the tissue prepa-ration step. Third, mass cytometry, although multiplexed, remains acandidate-based method, and the biological insights derived are onlyas informative as the biomarker panel allows. A well-known short-coming of immunohistochemistry techniques is the reliance on celltype–specific markers and morphology to identify cell types, whereasthese properties may be altered by concomitant loss of architecture, in-filtration of host cells, and dedifferentiation in dysplastic tissue. Al-though we appreciate that cell identities in cancer may not reflectthose of normal tissue, the use of multiplex marker panels, specificallythose of signaling that represent the functional state of a cell, can allowfor the inference of the lineage of origin of cancer cells with unknownidentities. Usingmultiplex single-cell data with comparative algorithmssuch as Citrus (46), one can determine the similarity of cancer cells toreference signatures of normal cell types in marker space. Furthermore,candidate-based single-cell approaches can be coupled with single-cellRNA sequencing (47) and even other unbiased bulk-based methods tobecome a powerful discovery tool. There is high potential impact forcharacterizing unidentified transitional cells in cancer, because theymay have altered properties that contribute to malignancy, and, moreimportantly, may be targetable by therapy. All of the above limitationsare inherent to FFPE applications in general and should be consideredand be controlled for at the study design phase.

The ability to query signal transduction in a cell type–specific or evenat a single-cell level is a defined strength of our approach. The prevalentmethods to detect and quantify signaling proteins remain to be bulkapproaches such as Western blots and enzyme-linked immunosorbentassay (32, 48), which assume cellular homogeneity and are not alwayssuitable for tissue analysis. With these approaches, positive signals insmall subsets of cells are washed out by larger populations, and thecellular sources of positive signals cannot be determined. For imagingapplications in tissues, cell type–specific signaling is usually evaluatedwith lowmultiplexity, for instance, looking at one signalingmarkerwithone cell-typemarker.More recent advances, asmentioned above (22–28),allow for higher multiplexity but at the expense of feasible application onlarge sample sets. FFPE-DISSECT coupled to mass cytometry is a rela-


tively rapid method for performing multiplex single-cell signaling anal-ysis. It can be used for proposing interesting signaling markers that canbe followed up by imaging, as we have done in this study.

Themajor assumption of FFPE-DISSECT, which begins with archi-val tissue blocks, is that the tissues are handled properly during the pre-analytical fixation steps. This assumption is widely made in thehistopathology field, especially in tissue microarray or cohort studieswhere hundreds of samples are collected from different sources (49).Improperly fixed tissue will inevitably lead to invalid downstreamanalyses. To mitigate artifacts arising from this source, we adopted astandardized operating procedure for processing all tissues in this study.First, tissue thickness was limited to 5 mm, which, according to com-mon references (50, 51), should allow efficient penetration of fixationwithin 1 hour. Second, the fixative was incubatedwith amagnetic stirrerto maximize diffusion. Third, the fixation time was standardized at24 hours. Last, and most importantly, a board-certified pathologist hasreviewed the histology of tissue alongwith quality assurance and controldata. Histological characteristics indicating poor fixation quality or in-adequate fixative penetration include (i) processing observations basedon nuclear staining and appearance of cytoplasm, (ii) scratches orhatching of the specimen during microtomy, (iii) section disintegratingor pulling apart, (iv) smudging or unusual staining, (v) other unusualartifacts, (vi) stutter, (vii) degree of autolysis, and (viii) cells showingcrenation. Samples indicative of fixation problems were not includedin this study. Aside from preanalytical evaluation, the following addi-tional steps can be taken to identify potential artifacts after data collec-tion: (i) imaging tissue section from the same tissue block to ensureconcordance (% of host cell infiltrating, relative intensity of markers)with single-cell data, (ii) imaging single-cell suspensions to ensure dis-aggregation into single cells, (iii) evaluating proper conjugation of anti-bodies by stainingwith both the conjugated and the unconjugated clonecoupled to a secondary detection system, and (iv) assessing detectionspecificity by identifying cytometry time-of-flight (CyTOF) events thatare positive for all markers. Many of these artifacts arise from the FFPEprocess, and we remain hopeful that the widespread adoption of stan-dardized procedures and additional technological advances will mini-mize these issues in the future.

Our approach illuminated differential signaling patterns in differentcell types (enterocytes, goblet cells, and enteroendocrine cells), with theconclusion that secretory cells in general aremore sensitive to basal andTNF-a–induced signaling. Goblet cells have the highest signaling pro-pensity, with up-regulation ofmany pathways compared to enterocytes.Specifically, goblet cells up-regulate the phosphorylation of ERK, whichwe identified as a survival mechanism against TNF-a–induced apopto-sis; accordingly, goblet cells are resistant to TNF-a–induced apoptosis(16). Furthermore, epidermal growth factor receptor (EGFR) is a recep-tor upstream of ERK that plays critical roles in growth, survival, anddifferentiation in the stem cell niche (52). Following this line of logic,we established a link between RAS activation and goblet cell metaplasiain the intestinal epithelium. To our knowledge, this is the first time thatNRAS activation has been connected to this phenotype. Demonstratingthe casual effect of this pathway, ERK signaling down-regulation hasbeen documented to suppress goblet cell specification. Heuberger et al.have shown that epithelial-specific knockout of Src homology phos-phatase 2 suppresses p-ERK signaling and goblet cell differentiationby modulating transcription factor 4 isoform switching and WNT-dependent transcription (53). This effect on goblet cells can be rescuedby gain of function inMEK1. Amore recent report by de Jong et al. hasalso shown that the knockout of both ERK1 and ERK2 impairs goblet

8 of 12



on October 14, 2016


Dow

nloaded from

cell differentiation (54). Goblet cell numbers were reduced but not com-pletely ablated in that study, suggesting that there are compensatorymechanisms to maintain goblet cell number. Using a multiplex celltype–specific approach, we propose other candidate signaling pathways,such as through the phosphorylation of ATF2, that may act in synergywith p-ERK to control goblet cell specification.

Cellular heterogeneity is an important topic in cancer biology fromboth genetic and cell biology perspectives. Differential signaling be-tween cells is a form of heterogeneity that controls cellular behaviorsbut is relatively unexplored. Gerdes et al. reported that pathway rela-tionships between signaling components identified from cell linesmay not hold true in human tissues when observed at the single-celllevel (23). Here, we identified signaling pathways that organize into amodular architecture associated with surface-to-crypt identity in thenormal colonic epithelium. Consistent correlation between pathwaysover single cells represents regulatory mechanisms that are recurrentlyused by all cells in the tissue. Maintenance of modular architecture be-tween samples reflects homogenous organization of signaling pathways.Quantifying these two properties using mass cytometry single-cell datasuggest that both intratissue and intertissue heterogeneity are increasedin CRC regarding signaling regulation. Heterogeneity in cancersignaling reflects the relaxation of constraints that allows a cancer cellto sample a wider state space. These constraints can be physical or bio-chemical, from disorganization of tissue architecture to rewiring ofsignaling networks. In turn, a cell can adopt novel behaviors andfunctions outside of normal cellular behaviors such as epithelial-to-mesenchymal transition (55).

The RAS-RAF-MEK-ERK kinase cascade plays a major role in thepathogenesis of CRC.Activating KRAS andNRASmutations are foundin ~50% of all CRC, and activating BRAFmutations are found in ~10%of CRC (56–58). Mutations in KRAS and downstream BRAF are a bio-marker for resistance to upstreamEGFR-targeted therapies, as expected(59, 60). However, downstream MEK inhibition has limited efficacy inCRC with KRAS and BRAF mutations (61, 62). Although acquiredresistancemechanisms, such as up-regulation of EGFR familymembersand BRAF gene amplification (63, 64), are seen in cell lines, analternative explanation for the lack of efficacy of MEK inhibitors inthe clinic may simply be that MAPK signaling downstream of mutantKRAS or BRAF is not up-regulated in CRC tissue, as shown in our datahere. Unlike cell lines (65), activating KRAS and BRAF in vivo results innegative feedback that up-regulates the expression of ERK-targetingphosphatases (41). Work from the Channing Der laboratory has ob-served that nuclear ERK phosphorylation in human CRC is notcorrelated to the mutational status of KRAS or, to a lesser extent, BRAF(66).Mousemodels also revealed that KRAS activation in normal intes-tinal epithelium activates ERK, but this effect is inhibited in the contextofmousemodels of CRC (38). These results support the idea that tissue,cell, and disease contexts strongly govern the influence of genetics onthe output of signaling pathways.

Although some argue for redundancy between KRAS and BRAFmutations inMAPK signaling by theirmutual exclusion (67), the cancerphenotypes induced by these mutations are vastly different. KRASmu-tationsmostly occur in common sporadicCRCs that are classified as thechromosomal instability phenotype, whereas BRAF mutations mostlyoccur in CRCs that are classified as the CpG island methylator pheno-type (68, 69). Hypermethylation of the MLH1 gene results indiminished DNA repair and induces an MSI-high phenotype distinctfrom that caused by mismatch repair gene mutations (such as inLynch’s syndrome) (70, 71). Thus, most BRAF mutations are seen in


MSI tumors and rarely in MSS tumors (72). BRAF mutant pathologyis also distinct from traditional adenocarcinomas, adopting a serratedmorphology (58). Within MSI tumors that often have good prognoses(4, 73), patients with BRAF mutant tumors have relatively poor overalloutcomes compared to thosewithBRAFwild-type tumors (71,74). Theseproperties may be due to the lack of differentiation in MSI-BRAF mu-tant tumors, resulting in a large number of “stem-like cells,” as seen inthis study and reported in others (5, 75), which confers resistance toconventional therapies. Perhaps, increased sensitivity to apoptosis inthese tumors, as marked by CC3 positivity, can be exploited as a ther-apeutic option. In our hands, KRAS mutation in MSS tumors is sug-gested to result in the down-regulation of ERK, an increase in thenumber of goblet cells, and an increase in cell proliferation in distinctpopulations thatmay identify with their relative differentiation states. Aweakness in our study is the low number of samples in each grouping,especially if we further partition samples by theirmolecular details. Ourintent here is to provide a proof-of-concept applicationof FFPE-DISSECTon human CRCs, and the hypotheses generated with this small cohortwill need to be confirmed in a larger set of tumors. However, giventhat our approach can be applied to FFPE tissue blocks, one can haveaccess tomuch larger repositories of retroactively collected samples thatcan power any study. FFPE-DISSECT coupled to mass cytometry ap-plied to archival samples is a powerful tool to generate large amounts ofsingle-cell data with acceptable throughput. These data are com-plementary to other precision medicine efforts to molecularly character-ize solid tumors for arriving at subtypes that can predict prognosis andtherapeutic response.

MATERIALS AND METHODSMouse experimentsAll animal experiments were performed under protocols approvedby the Vanderbilt University Animal Care and Use Committee andin accordance with the National Institutes of Health guidelines. Micewere stimulated with TNF-a as a time course, and their duodena(proximal small intestine) were collected for analysis as previouslydescribed (31, 32). For DISSECT, a previously published protocol wasused (16). For FFPE, tissues were previously fixed in formalin for 24 hoursand were then subjected to standardized embedding procedures. Tissueswere incubated in RPMI when outside of the body for extended time.

Human tissue acquisitionHuman normal colon and CRC tissues were obtained under protocolsapproved by theVanderbilt University through theCHTN.Clinical andpathology reports were attached to each sample before deidentificationof patient information. An optimized CHTN collection standardizedoperating procedure was used. Briefly, specimen sizes were limited to5mm in diameter and fixed for 24 hours inmagnetically stirred forma-lin (to facilitate diffusion), after which the specimenwas embeddedwitha standardized FFPE protocol. The time from which the specimen wasexcised from the patient to the fixative (during which the tissue wasexamined by the pathologists or their assistants) was recorded as thepost-excision time. Specimens with substantial tumor cell content de-termined by hematoxylin and eosin staining were selected for analysis.Overall, 7 normal colon samples and 13 CRC samples were selected.

DISSECT disaggregation on FFPE tissuesSections (50 mm) were freshly cut from each block and placed in 1.5-mlmicrocentrifuge tubes (Fisher). Samples were heated to 65°C for 25min

9 of 12



on October 14, 2016


Dow

nloaded from

tomelt wax and then washed three times with 1ml of Histo-Clear for8 min each. Tissues were then rehydrated in two washes each of 100,70, and 50% ethanol and then three washes of phosphate-bufferedsaline (PBS). Samples were washed for 10 min in PBS with 0.3% TritonX-100 and then washed for a final time in PBS before incubation inthe HIAR buffer (DAKO). Samples were incubated in the buffer underhigh heat and pressure for 20 min (actively heating for the first 4 min),followed by 20-min cooling of the bench. Samples were then washedthree additional times in PBS and stored at 4°C until staining.

Tissues were blocked at room temperature for 30 min in 2.5% don-key serum (Jackson ImmunoResearch) in PBS and stained overnight atroom temperature with antibodies diluted in the same buffer. Addition-al blocking with carefully chosen serum combinations were applied ifsecondary antibodies were used. After appropriate washing, sampleswere incubated for 30 min in 4% PFA to cross-link antibodies to theirtargets. Samples were washed and then incubated for 25 min at 37°C in200ml of PBSwith 1mg/ml each of collagenase (Calbiotech) and dispase(Life Technologies). Tissues were passaged 5 to 10 times through a 27-gauge needle tomechanically dissociate them into single cells. Cellswereincubated with a nuclear intercalating agent before analysis.

Cytometry analysesFor both fluorescence and mass cytometry, cells were initially gatedusing DNA content [Hoechst (fluorescence cytometry)] or intercalator[Iridium (mass cytometry)] following the established procedures toidentify intact single cells and eliminate cell doublets and clusters fromanalysis (16, 76, 77). Single cells were then analyzed for intensity of an-tibody conjugates. Fluorescence cytometry was performed on a BDLSRII with five lasers, and mass cytometry was performed on a Fluidigm-DVS CyTOF 1 instrument. Epithelial-specific analysis was achieved bygating cells positive for PCK.

IF imagingFFPE tissues were sectioned at 5 mm and processed using standardimmunohistological techniques, stained with appropriate primaryor primary and secondary antibodies. DISSECT-processed tissues werealso imaged before and after disaggregation in a whole-mount format.Slides were imaged using a Zeiss Axiophot fluorescence or bright-fieldmicroscope with a Zeiss Axiocam with five-channel imaging capabil-ities. Quantitative analysis of goblet cells in the villus was performed onImageJ using the particle analysis module. Ratios of areas occupied be-tween the CLCA1 channel and nuclear channel were calculated with acorrection factor for the typical size of a goblet granule against the typ-ical size of a nucleus.

Quantitative IF imaging and immunoblotting ofsignaling proteinsThe same antibody clones were used for FFPE-DISSECT-CyTOF.Quantifications were performed as described previously (16).

Antibody reagentsAntibodies used in this study are listed in tables S1 and S3. Allsignaling antibodies were previously validated and used in mass cy-tometry applications (16).

Data and statistical analysist-SNE analysis was performed using the viSNE implementation onCytobank.org following established single-cell analysis workflows(78–80). Gating for cell types was performed by considering a


first-decade (101) threshold for cell type–specific markers. Un-paired t tests and correlation analyses were performed using Prism(GraphPad). Multiple comparison tests were performed with ANOVAwith Tukey post hoc test (GraphPad). Correlative distances and heatmaps were generated using MATLAB (MathWorks). Hierarchicalclustering and dendrogram analysis were performed using the dendextendpackage in R (42).

SUPPLEMENTARY MATERIALSwww.sciencesignaling.org/cgi/content/full/9/449/rs11/DC1Fig. S1. Imaging of single-cell suspensions prepared by FFPE-DISSECT from embedded mouseintestinal tissue by bright-field and autofluorescence imaging.Fig. S2. Whole-mount staining of embedded mouse intestinal tissue prepared by FFPE-DISSECTbefore single-cell disaggregation.Fig. S3. Fluorescence cytometry of samples prepared by FFPE-DISSECT.Fig. S4. Signals are preserved for up to 1 hour after excision.Fig. S5. Fixation time does not affect signaling marker detection in FFPE.Fig. S6. Villus signal comparison between FFPE-DISSECT, quantitative imaging, andquantitative immunoblotting.Fig. S7. Replicates over multiple animals depicting relationships between apoptosis andsignaling pathways using t-SNE and biplots.Fig. S8. Cell type–specific signaling in the murine duodenal epithelium.Fig. S9. RAS induction of MEK-ERK signaling induces goblet cell identity.Fig. S10. Hematoxylin and eosin of MSS and MSI CRCs showing tissue areas occupied bytumors.Fig. S11. Signaling in human normal colon and CRC.Fig. S12. Differentiation in human normal colon and CRC.Fig. S13. Evidence of fixative penetration.Fig. S14. Analysis of the organization of signaling pathways in human specimens.Table S1. Mouse antibody reagent panel.Table S2. Summary of pathological characteristics of human colon cancer samples.Table S3. Human antibody reagent panel.

REFERENCES AND NOTES1. A. S. Cleary, T. L. Leonard, S. A. Gestl, E. J. Gunther, Tumour cell heterogeneity maintained

by cooperating subclones in Wnt-driven mammary cancers. Nature 508, 113–117 (2014).2. A. Marusyk, D. P. Tabassum, P. M. Altrock, V. Almendro, F. Michor, K. Polyak, Non-cell-

autonomous driving of tumour growth supports sub-clonal heterogeneity. Nature 514,54–58 (2014).

3. E. Wagenblast, M. Soto, S. Gutiérrez-Ángel, C. A. Hartl, A. L. Gable, A. R. Maceli, N. Erard,A. M. Williams, S. Y. Kim, S. Dickopf, J. C. Harrell, A. D. Smith, C. M. Perou, J. E. Wilkinson,G. J. Hannon, S. R. V. Knott, A model of breast cancer heterogeneity reveals vascularmimicry as a driver of metastasis. Nature 520, 358–362 (2015).

4. L. De Smedt, J. Lemahieu, S. Palmans, O. Govaere, T. Tousseyn, E. Van Cutsem, H. Prenen,S. Tejpar, M. Spaepen, G. Matthijs, C. Decaestecker, X. Moles Lopez, P. Demetter,I. Salmon, X. Sagaert, Microsatellite instable vs stable colon carcinomas: Analysis oftumour heterogeneity, inflammation and angiogenesis. Br. J. Cancer 113, 500–509 (2015).

5. P. Dalerba, T. Kalisky, D. Sahoo, P. S. Rajendran, M. E. Rothenberg, A. A. Leyrat, S. Sim,J. Okamoto, D. M. Johnston, D. Qian, M. Zabala, J. Bueno, N. F. Neff, J. Wang, A. A. Shelton,B. Visser, S. Hisamori, Y. Shimono, M. van de Wetering, H. Clevers, M. F. Clarke, S. R. Quake,Single-cell dissection of transcriptional heterogeneity in human colon tumors. Nat.Biotechnol. 29, 1120–1127 (2011).

6. A. Kreso, C. A. O’Brien, P. van Galen, O. I. Gan, F. Notta, A. M. K. Brown, K. Ng, J. Ma,E. Wienholds, C. Dunant, A. Pollett, S. Gallinger, J. McPherson, C. G. Mullighan, D. Shibata,J. E. Dick, Variable clonal repopulation dynamics influence chemotherapy response incolorectal cancer. Science 339, 543–548 (2013).

7. P. Dalerba, S. J. Dylla, I.-K. Park, R. Liu, X. Wang, R. W. Cho, T. Hoey, A. Gurney, E. H. Huang,D. M. Simeone, A. A. Shelton, G. Parmiani, C. Castelli, M. F. Clarke, Phenotypiccharacterization of human colorectal cancer stem cells. Proc. Natl. Acad. Sci. U.S.A. 104,10158–10163 (2007).

8. S. Kozar, E. Morrissey, A. M. Nicholson, M. van der Heijden, H. I. Zecchini, R. Kemp,S. Tavaré, L. Vermeulen, D. J. Winton, Continuous clonal labeling reveals small numbers offunctional stem cells in intestinal crypts and adenomas. Cell Stem Cell 13, 626–633 (2013).

9. C. Scheel, R. A. Weinberg, Cancer stem cells and epithelial–mesenchymal transition:Concepts and molecular links. Semin. Cancer Biol. 22, 396–403 (2012).

10. E. Fabrizi, S. di Martino, F. Pelacchi, L. Ricci-Vitiani, Therapeutic implications of coloncancer stem cells. World J. Gastroenterol. 16, 3871–3877 (2010).

10 of 12

http://www.sciencesignaling.org/cgi/content/full/9/449/rs11/DC1



on October 14, 2016


Dow

nloaded from

11. S. Colak, J. P. Medema, Cancer stem cells—Important players in tumor therapy resistance.FEBS J. 281, 4779–4791 (2014).

12. M. Dean, T. Fojo, S. Bates, Tumour stem cells and drug resistance. Nat. Rev. Cancer 5,275–284 (2005).

13. C. A. Del Vecchio, Y. Feng, E. S. Sokol, E. J. Tillman, S. Sanduja, F. Reinhardt, P. B. Gupta,De-differentiation confers multidrug resistance via noncanonical PERK-Nrf2 signaling.PLOS Biol. 12, e1001945 (2014).

14. J. M. Irish, D. B. Doxie, High-dimensional single-cell cancer biology. Curr. Top. Microbiol.Immunol. 377, 1–21 (2014).

15. J. M. Irish, N. Kotecha, G. P. Nolan, Mapping normal and cancer cell signalling networks:Towards single-cell proteomics. Nat. Rev. Cancer 6, 146–155 (2006).

16. A. J. Simmons, A. Banerjee, E. T. McKinley, C. R. Scurrah, C. A. Herring, L. S. Gewin,R. Masuzaki, S. J. Karp, J. L. Franklin, M. J. Gerdes, J. M. Irish, R. J. Coffey, K. S. Lau,Cytometry-based single-cell analysis of intact epithelial signaling reveals MAPK activationdivergent from TNF-a-induced apoptosis in vivo. Mol. Syst. Biol. 11, 835 (2015).

17. I. Abrahamsen, J. B. Lorens, Evaluatingextracellularmatrix influenceon adherent cell signalingbycold trypsin phosphorylation-specific flow cytometry. BMC Cell Biol. 14, 36 (2013).

18. J. Zeng, A. Mohammadreza, W. Gao, S. Merza, D. Smith, L. Kelbauskas, D. R. Meldrum,A minimally invasive method for retrieving single adherent cells of different typesfrom cultures. Sci. Rep. 4, 5424 (2014).

19. T. J. Kokkat, M. S. Patel, D. McGarvey, V. A. LiVolsi, Z. W. Baloch, Archived formalin-fixedparaffin-embedded (FFPE) blocks: A valuable underexploited resource for extraction ofDNA, RNA, and protein. Biopreserv. Biobank. 11, 101–106 (2013).

20. P. Li, A. Conley, H. Zhang, H. L. Kim, Whole-Transcriptome profiling of formalin-fixed,paraffin-embedded renal cell carcinoma by RNA-seq. BMC Genomics 15, 1087 (2014).

21. J. Hedegaard, K. Thorsen, M. K. Lund, A.-M. K. Hein, S. J. Hamilton-Dutoit, S. Vang,I. Nordentoft, K. Birkenkamp-Demtröder, M. Kruhøffer, H. Hager, B. Knudsen,C. L. Andersen, K. D. Sørensen, J. S. Pedersen, T. F. Ørntoft, L. Dyrskjøt, Next-generationsequencing of RNA and DNA isolated from paired fresh-frozen and formalin-fixedparaffin-embedded samples of human cancer and normal tissue. PLOS ONE 9, e98187(2014).

22. J. Lyons, C. A. Herring, A. Banerjee, A. J. Simmons, K. S. Lau, Multiscale analysis of themurine intestine for modeling human diseases. Integr. Biol. 7, 740–757 (2015).

23. M. J. Gerdes, C. J. Sevinsky, A. Sood, S. Adak, M. O. Bello, A. Bordwell, A. Can, A. Corwin,S. Dinn, R. J. Filkins, D. Hollman, V. Kamath, S. Kaanumalle, K. Kenny, M. Larsen, M. Lazare,Q. Li, C. Lowes, C. C. McCulloch, E. McDonough, M. C. Montalto, Z. Pang, J. Rittscher,A. Santamaria-Pang, B. D. Sarachan, M. L. Seel, A. Seppo, K. Shaikh, Y. Sui, J. Zhang,F. Ginty, Highly multiplexed single-cell analysis of formalin-fixed, paraffin-embeddedcancer tissue. Proc. Natl. Acad. Sci. U.S.A. 110, 11982–11987 (2013).

24. J.-R. Lin, M. Fallahi-Sichani, P. K. Sorger, Highly multiplexed imaging of single cellsusing a high-throughput cyclic immunofluorescence method. Nat. Commun. 6, 8390(2015).

25. P. Zrazhevskiy, X. Gao, Quantum dot imaging platform for single-cell molecular profiling.Nat. Commun. 4, 1619 (2013).

26. D. P. Riordan, S. Varma, R. B. West, P. O. Brown, Automated analysis and classification ofhistological tissue features by multi-dimensional microscopic molecular profiling. PLOSONE 10, e0128975 (2015).

27. M. Angelo, S. C. Bendall, R. Finck, M. B. Hale, C. Hitzman, A. D. Borowsky, R. M. Levenson,J. B. Lowe, S. D. Liu, S. Zhao, Y. Natkunam, G. P. Nolan, Multiplexed ion beam imaging ofhuman breast tumors. Nat. Med. 20, 436–442 (2014).

28. C. Giesen, H. A. O. Wang, D. Schapiro, N. Zivanovic, A. Jacobs, B. Hattendorf, P. J. Schüffler,D. Grolimund, J. M. Buhmann, S. Brandt, Z. Varga, P. J. Wild, D. Günther, B. Bodenmiller,Highly multiplexed imaging of tumor tissues with subcellular resolution by masscytometry. Nat. Methods 11, 417–422 (2014).

29. P. J. Schüffler, D. Schapiro, C. Giesen, H. A. O. Wang, B. Bodenmiller, J. M. Buhmann,Automatic single cell segmentation on highly multiplexed tissue images. Cytometry A 87,936–942 (2015).

30. W. E. Corver, N. T. ter Haar, High-resolution multiparameter DNA flow cytometry for thedetection and sorting of tumor and stromal subpopulations from paraffin-embeddedtissues. Curr. Protoc. Cytom. Chapter 7, Unit 7.37 (2011).

31. K. S. Lau, V. Cortez-Retamozo, S. R. Philips, M. J. Pittet, D. A. Lauffenburger, K. M. Haigis,Multi-scale in vivo systems analysis reveals the influence of immune cells on TNF-a-induced apoptosis in the intestinal epithelium. PLOS Biol. 10, e1001393 (2012).

32. K. S. Lau, A. M. Juchheim, K. R. Cavaliere, S. R. Philips, D. A. Lauffenburger, K. M. Haigis, In vivosystems analysis identifies spatial and temporal aspects of the modulation of TNF-a-inducedapoptosis and proliferation by MAPKs. Sci. Signal. 4, ra16 (2011).

33. I. F. Pinhel, F. A. Macneill, M. J. Hills, J. Salter, S. Detre, R. A’hern, A. Nerurkar, P. Osin,I. E. Smith, M. Dowsett, Extreme loss of immunoreactive p-Akt and p-Erk1/2 duringroutine fixation of primary breast cancer. Breast Cancer Res. 12, R76 (2010).

34. K. S. Lau, S. B. Schrier, J. Gierut, J. Lyons, D. A. Lauffenburger, K. M. Haigis, Networkanalysis of differential Ras isoform mutation effects on intestinal epithelial responses toTNF-a. Integr. Biol. 5, 1355–1365 (2013).


35. J. J. Gierut, L. B. Wood, K. S. Lau, Y.-J. Lin, C. Genetti, A. A. Samatar, D. A. Lauffenburger,K. M. Haigis, Network-level effects of kinase inhibitors modulate TNF-a-induced apoptosisin the intestinal epithelium. Sci. Signal. 8, ra129 (2015).

36. L. van der Maaten, G. Hinton, Visualizing data using t-SNE. J. Mach. Learn. Res. 9,2579–2605 (2008).

37. K. S. Lau, K. M. Haigis, Non-redundancy within the RAS oncogene family: Insights intomutational disparities in cancer. Mol. Cells 28, 315–320 (2009).

38. K. M. Haigis, K. R. Kendall, Y. Wang, A. Cheung, M. C. Haigis, J. N. Glickman,M. Niwa-Kawakita, A. Sweet-Cordero, J. Sebolt-Leopold, K. M. Shannon, J. Settleman,M. Giovannini, T. Jacks, Differential effects of oncogenic K-Ras and N-Ras on proliferation,differentiation and tumor progression in the colon. Nat. Genet. 40, 600–608 (2008).

39. Y. Wang, S. Velho, E. Vakiani, S. Peng, A. J. Bass, G. C. Chu, J. Gierut, J. M. Bugni, C. J. Der,M. Philips, D. B. Solit, K. M. Haigis, Mutant N-RAS protects colorectal cancer cells fromstress-induced apoptosis and contributes to cancer development and progression.Cancer Discov. 3, 294–307 (2013).

40. Y. Feng, G. T. Bommer, J. Zhao, M. Green, E. Sands, Y. Zhai, K. Brown, A. Burberry, K. R. Cho,E. R. Fearon, Mutant kras promotes hyperplasia and alters differentiation in the colonepithelium but does not expand the presumptive stem cell pool. Gastroenterology 141,1003–1013.e10 (2011).

41. S. Cagnol, N. Rivard, Oncogenic KRAS and BRAF activation of the MEK/ERK signalingpathway promotes expression of dual-specificity phosphatase 4 (DUSP4/MKP2) resultingin nuclear ERK1/2 inhibition. Oncogene 32, 564–576 (2013).

42. T. Galili, dendextend: An R package for visualizing, adjusting and comparing trees ofhierarchical clustering. Bioinformatics 31, 3718–3720 (2015).

43. F. B. Baker, Stability of two hierarchical grouping techniques case 1: Sensitivity to dataerrors. J. Am. Stat. Assoc. 69, 440–445 (1974).

44. B. Zhang, J. Wang, X. Wang, J. Zhu, Q. Liu, Z. Shi, M. C. Chambers, L. J. Zimmerman,K. F. Shaddox, S. Kim, S. R. Davies, S. Wang, P. Wang, C. R. Kinsinger, R. C. Rivers,H. Rodriguez, R. R. Townsend, M. J. C. Ellis, S. A. Carr, D. L. Tabb, R. J. Coffey, R. J. C. Slebos,D. C. Liebler; NCI CPTAC, Proteogenomic characterization of human colon and rectalcancer. Nature 513, 382–387 (2014).

45. R. Xie, J.-Y. Chung, K. Ylaya, R. L. Williams, N. Guerrero, N. Nakatsuka, C. Badie, S. M. Hewitt,Factors influencing the degradation of archival formalin-fixed paraffin-embedded tissuesections. J. Histochem. Cytochem. 59, 356–365 (2011).

46. R. V. Bruggner, B. Bodenmiller, D. L. Dill, R. J. Tibshirani, G. P. Nolan, Automatedidentification of stratifying signatures in cellular subpopulations. Proc. Natl. Acad. Sci. U.S.A.111, E2770–E2777 (2014).

47. E. Z. Macosko, A. Basu, R. Satija, J. Nemesh, K. Shekhar, M. Goldman, I. Tirosh, A. R. Bialas,N. Kamitaki, E. M. Martersteck, J. J. Trombetta, D. A. Weitz, J. R. Sanes, A. K. Shalek,A. Regev, S. A. McCarroll, Highly parallel genome-wide expression profiling of individualcells using nanoliter droplets. Cell 161, 1202–1214 (2015).

48. M. F. Ciaccio, J. P. Wagner, C.-P. Chuu, D. A. Lauffenburger, R. B. Jones, Systems analysis ofEGF receptor signaling dynamics with microwestern arrays. Nat. Methods 7, 148–155 (2010).

49. D. L. Rimm, R. L. Camp, L. A. Charette, D. A. Olsen, E. Provost, Amplification of tissue byconstruction of tissue microarrays. Exp. Mol. Pathol. 70, 255–264 (2001).

50. D. R. Bauer, B. Stevens, D. Chafin, A. P. Theiss, M. Otter, Active monitoring offormaldehyde diffusion into histological tissues with digital acoustic interferometry.J. Med. Imaging 3, 017002 (2016).

51. J. R. Baker, Principles of Biological Microtechnique: A Study of Fixation and Dyeing(John Wiley & Sons, 1958).

52. T. Sato, J. H. van Es, H. J. Snippert, D. E. Stange, R. G. Vries, M. van den Born, N. Barker,N. F. Shroyer, M. van de Wetering, H. Clevers, Paneth cells constitute the niche for Lgr5stem cells in intestinal crypts. Nature 469, 415–418 (2011).

53. J. Heuberger, F. Kosel, J. Qi, K. S. Grossmann, K. Rajewsky, W. Birchmeier, Shp2/MAPKsignaling controls goblet/paneth cell fate decisions in the intestine. Proc. Natl. Acad. Sci.U.S.A. 111, 3472–3477 (2014).

54. P. R. de Jong, K. Taniguchi, A. R. Harris, S. Bertin, N. Takahashi, J. Duong, A. D. Campos,G. Powis, M. Corr, M. Karin, E. Raz, ERK5 signalling rescues intestinal epithelial turnoverand tumour cell proliferation upon ERK1/2 abrogation. Nat. Commun. 7, 11551 (2016).

55. R. Kalluri, R. A. Weinberg, The basics of epithelial-mesenchymal transition. J. Clin. Invest.119, 1420–1428 (2009).

56. B. Vogelstein, E. R. Fearon, S. R. Hamilton, S. E. Kern, A. C. Preisinger, M. Leppert,Y. Nakamura, R. White, A. M. M. Smits, J. L. Bos, Genetic alterations during colorectal-tumor development. N. Engl. J. Med. 319, 525–532 (1988).

57. Cancer Genome Atlas Network, Comprehensive molecular characterization of humancolon and rectal cancer. Nature 487, 330–337 (2012).

58. D. Barras, BRAF mutation in colorectal cancer: An update. Biomark. Cancer 7 (Suppl. 1),9–12 (2015).

59. C. S. Karapetis, S. Khambata-Ford, D. J. Jonker, C. J. O’Callaghan, D. Tu, N. C. Tebbutt,R. J. Simes, H. Chalchal, J. D. Shapiro, S. Robitaille, T. J. Price, L. Shepherd, H.-J. Au,C. Langer, M. J. Moore, J. R. Zalcberg, K-ras mutations and benefit from cetuximab inadvanced colorectal cancer. N. Engl. J. Med. 359, 1757–1765 (2008).

11 of 12



on October 14, 2016


Dow

nloaded from

60. F. Pietrantonio, F. Petrelli, A. Coinu, M. Di Bartolomeo, K. Borgonovo, C. Maggi,M. Cabiddu, R. Iacovelli, I. Bossi, V. Lonati, M. Ghilardi, F. de Braud, S. Barni, Predictive roleof BRAF mutations in patients with advanced colorectal cancer receiving cetuximab andpanitumumab: A meta-analysis. Eur. J. Cancer 51, 587–594 (2015).

61. B. B. Friday, A. A. Adjei, Advances in targeting the Ras/Raf/MEK/Erk mitogen-activatedprotein kinase cascade with MEK inhibitors for cancer therapy. Clin. Cancer Res. 14,342–346 (2008).

62. A. A. Adjei, R. B. Cohen, W. Franklin, C. Morris, D. Wilson, J. R. Molina, L. J. Hanson, L. Gore,L. Chow, S. Leong, L. Maloney, G. Gordon, H. Simmons, A. Marlow, K. Litwiler, S. Brown,G. Poch, K. Kane, J. Haney, S. G. Eckhardt, Phase I pharmacokinetic andpharmacodynamic study of the oral, small-molecule mitogen-activated proteinkinase kinase 1/2 inhibitor AZD6244 (ARRY-142886) in patients with advanced cancers.J. Clin. Oncol. 26, 2139–2146 (2008).

63. C. Sun, S. Hobor, A. Bertotti, D. Zecchin, S. Huang, F. Galimi, F. Cottino, A. Prahallad,W. Grernrum, A. Tzani, A. Schlicker, L. F. A. Wessels, E. F. Smit, E. Thunnissen, P. Halonen,C. Lieftink, R. L. Beijersbergen, F. Di Nicolantonio, A. Bardelli, L. Trusolino, R. Bernards,Intrinsic resistance to MEK inhibition in KRAS mutant lung and colon cancer throughtranscriptional induction of ERBB3. Cell Rep. 7, 86–93 (2014).

64. R. B. Corcoran, D. Dias-Santagata, K. Bergethon, A. J. Iafrate, J. Settleman, J. A. Engelman,BRAF gene amplification can promote acquired resistance to MEK inhibitors in cancercells harboring the BRAF V600E mutation. Sci. Signal. 3, ra84 (2010).

65. C. A. Pratilas, B. S. Taylor, Q. Ye, A. Viale, C. Sander, D. B. Solit, N. Rosen, V600EBRAF isassociated with disabled feedback inhibition of RAF-MEK signaling and elevatedtranscriptional output of the pathway. Proc. Natl. Acad. Sci. U.S.A. 106, 4519–4524 (2009).

66. J. J. Yeh, E. D. Routh, T. Rubinas, J. Peacock, T. D. Martin, X. J. Shen, R. S. Sandler, H. J. Kim,T. O. Keku, C. J. Der, KRAS/BRAF mutation status and ERK1/2 activation as biomarkers forMEK1/2 inhibitor therapy in colorectal cancer. Mol. Cancer Ther. 8, 834–843 (2009).

67. I. H. Sahin, S. M. A. Kazmi, J. T. Yorio, N. A. Bhadkamkar, B. K. Kee, C. R. Garrett, Rare thoughnot mutually exclusive: A report of three cases of concomitant KRAS and BRAF mutationand a review of the literature. J. Cancer 4, 320–322 (2013).

68. D. J. Weisenberger, K. D. Siegmund, M. Campan, J. Young, T. I. Long, M. A. Faasse,G. H. Kang, M. Widschwendter, D. Weener, D. Buchanan, H. Koh, L. Simms, M. Barker,B. Leggett, J. Levine, M. Kim, A. J. French, S. N. Thibodeau, J. Jass, R. Haile, P. W. Laird,CpG island methylator phenotype underlies sporadic microsatellite instability and istightly associated with BRAF mutation in colorectal cancer. Nat. Genet. 38, 787–793(2006).

69. M. Morkel, P. Riemer, H. Bläker, C. Sers, Similar but different: Distinct roles for KRAS andBRAF oncogenes in colorectal cancer development and therapy resistance. Oncotarget 6,20785–20800 (2015).

70. A. J. French, D. J. Sargent, L. J. Burgart, N. R. Foster, B. F. Kabat, R. Goldberg, L. Shepherd,H. E. Windschitl, S. N. Thibodeau, Prognostic significance of defective mismatchrepair and BRAF V600E in patients with colon cancer. Clin. Cancer Res. 14, 3408–3415 (2008).

71. P. Lochhead, A. Kuchiba, Y. Imamura, X. Liao, M. Yamauchi, R. Nishihara, Z. R. Qian,T. Morikawa, J. Shen, J. A. Meyerhardt, C. S. Fuchs, S. Ogino, Microsatellite instability andBRAF mutation testing in colorectal cancer prognostication. J. Natl. Cancer Inst. 105,1151–1156 (2013).

72. K. Fransén, M. Klintenäs, A. Österström, J. Dimberg, H.-J. Monstein, P. Söderkvist, Mutationanalysis of the BRAF, ARAF and RAF-1 genes in human colorectal adenocarcinomas.Carcinogenesis 25, 527–533 (2004).

73. Y. Xiao, G. J. Freeman, The microsatellite instable subset of colorectal cancer is aparticularly good candidate for checkpoint blockade immunotherapy. Cancer Discov. 5,16–18 (2015).

74. B. Tran, S. Kopetz, J. Tie, P. Gibbs, Z.-Q. Jiang, C. H. Lieu, A. Agarwal, D. M. Maru, O. Sieber,J. Desai, Impact of BRAF mutation and microsatellite instability on the pattern ofmetastatic spread and prognosis in metastatic colorectal cancer. Cancer 117, 4623–4632(2011).


75. J. H. Kim, Y.-Y. Rhee, J. M. Bae, N.-Y. Cho, G. H. Kang, Loss of CDX2/CK20 expression isassociated with poorly differentiated carcinoma, the CpG island methylator phenotype,and adverse prognosis in microsatellite-unstable colorectal cancer. Am. J. Surg. Pathol. 37,1532–1541 (2013).

76. K. J. Nicholas, A. R. Greenplate, D. K. Flaherty, B. K. Matlock, J. S. Juan, R. M. Smith,J. M. Irish, S. A. Kalams, Multiparameter analysis of stimulated human peripheral bloodmononuclear cells: A comparison of mass and fluorescence cytometry. Cytometry A 89,271–280 (2016).

77. N. Leelatian, K. E. Diggins, J. M. Irish, Characterizing phenotypes and signaling networksof single human cells by mass cytometry. Methods Mol. Biol. 1346, 99–113 (2015).

78. E.-a. D. Amir, K. L. Davis, M. D. Tadmor, E. F. Simonds, J. H. Levine, S. C. Bendall,D. K. Shenfeld, S. Krishnaswamy, G. P. Nolan, D. Pe’er, viSNE enables visualization of highdimensional single-cell data and reveals phenotypic heterogeneity of leukemia. Nat.Biotechnol. 31, 545–552 (2013).

79. K. E. Diggins, P. B. Ferrell Jr., J. M. Irish, Methods for discovery and characterization of cellsubsets in high dimensional mass cytometry data. Methods 82, 55–63 (2015).

80. Y. Saeys, S. Van Gassen, B. N. Lambrecht, Computational flow cytometry: Helping to makesense of high-dimensional immunology data. Nat. Rev. Immunol. 16, 449–462 (2016).

Acknowledgments: We thank K. Weller and D. Flaherty at the Vanderbilt Flow CytometryCore, J. Roland at the Vanderbilt Digital Histology Shared Resource, and R. Carnahan at theVanderbilt Antibody and Protein Resource for their technical assistance; R. Mernaugh at theDepartment of Biochemistry, Vanderbilt University School of Medicine, K. Wiles at the CHTN,G. Ayers at the Vanderbilt Center for Quantitative Sciences, and A. Banerjee and S. W. Kim inthe Lau laboratory for their helpful advice; and K. Haigis at Beth Israel Deaconess MedicalCenter for his consultation on RAS mutations. Funding: K.S.L. was funded by R01DK103831, anInnovator Award from the American Association for Cancer Research–Landon Foundation(15-20-27-LAUK), a Crohn’s & Colitis Foundation of America Career Development Award(308221), the Vanderbilt Institute for Clinical and Translational Research (2UL1TR000445), andpilot project grants from P30CA068485, P30DK058404, P50CA095103, and U24CA159988.A.J.S. was funded by R01DK103831. C.A.H. was funded by a training grant from T32HD007502and F31GM120940. C.R.S. was funded by a training grant from T32AI007281. E.T.M. was fundedby a training grant from R25CA092043. R.J.C. was funded by R01CA174377, and R.J.C. andM.K.W. were funded by P50CA095103. J.M.I. was funded by R00CA143231. Authorcontributions: A.J.S. designed and performed all mouse, cytometry, and imagingexperiments. C.R.S. extracted and organized human patient data. E.T.M. assisted with imagingexperiments. C.A.H. assisted with computational data analysis. M.K.W. facilitated theacquisition of human tissues, performed pathological assessment, and intellectuallycontributed to the tissue preparation process. J.M.I. and R.J.C. intellectually contributed to thestudy and to the writing of the manuscript. K.S.L. conceived the study, performed cytometryand imaging analysis, performed computational data analysis, wrote the manuscript, andsupervised the research. Competing interests: The authors declare that they have nocompeting financial interests. Data and materials availability: Single-cell data for humantumor and mouse intestine generated by FFPE-DISSECT are shared on Cytobank.org (mousedata: https://community.cytobank.org/cytobank/experiments/57176; human data: https://community.cytobank.org/cytobank/experiments/57155).

Submitted 28 June 2016Accepted 21 September 2016Published 11 October 201610.1126/scisignal.aah4413

Citation: A. J. Simmons, C. R. Scurrah, E. T. McKinley, C. A. Herring, J. M. Irish, M. K. Washington,R. J. Coffey, K. S. Lau, Impaired coordination between signaling pathways is revealed in humancolorectal cancer using single-cell mass cytometry of archival tissue blocks. Sci. Signal. 9, rs11(2016).

12 of 12

https://community.cytobank.org/cytobank/experiments/57176




(449), rs11. [doi: 10.1126/scisignal.aah4413]9Science Signaling and Ken S. Lau (October 11, 2016) Herring, Jonathan M. Irish, M. Kay Washington, Robert J. Coffey Alan J. Simmons, Cherié R. Scurrah, Eliot T. McKinley, Charles A.archival tissue blocksin human colorectal cancer using single-cell mass cytometry of Impaired coordination between signaling pathways is revealed

This information is current as of October 14, 2016. The following resources related to this article are available online at http://stke.sciencemag.org.

Article Tools

http://stke.sciencemag.org/content/9/449/rs11article tools: Visit the online version of this article to access the personalization and

MaterialsSupplemental

http://stke.sciencemag.org/content/suppl/2016/10/06/9.449.rs11.DC1"Supplementary Materials"

Related Content

http://stm.sciencemag.org/content/scitransmed/6/219/219ra9.fullhttp://stke.sciencemag.org/content/sigtrans/8/359/ra4.fullhttp://stke.sciencemag.org/content/sigtrans/8/403/ra116.fullhttp://stke.sciencemag.org/content/sigtrans/8/401/pe3.full

's sites:ScienceThe editors suggest related resources on

Referenceshttp://stke.sciencemag.org/content/9/449/rs11#BIBLThis article cites 79 articles, 20 of which you can access for free at:

Permissionshttp://www.sciencemag.org/about/permissions.dtlObtain information about reproducing this article:

reserved. DC 20005. Copyright 2016 by the American Association for the Advancement of Science; all rightsAmerican Association for the Advancement of Science, 1200 New York Avenue, NW, Washington,

(ISSN 1937-9145) is published weekly, except the last December, by theScience Signaling

on October 14, 2016


Dow

nloaded from

http://stke.sciencemag.org/content/9/449/rs11

http://stke.sciencemag.org/content/suppl/2016/10/06/9.449.rs11.DC1

http://stke.sciencemag.org/content/sigtrans/8/401/pe3.full

http://stke.sciencemag.org/content/sigtrans/8/403/ra116.full

http://stke.sciencemag.org/content/sigtrans/8/359/ra4.full

http://stm.sciencemag.org/content/scitransmed/6/219/219ra9.full

http://stke.sciencemag.org/content/9/449/rs11#BIBL

http://www.sciencemag.org/about/permissions.dtl


simmons - 2016

Documents