The ZeptoCTC workflow

The concept for setting up a true single cell protein analysis workflow was to combine and fully integrate already existing, well-validated sub-workflows of selected mature technology modules (schematically displayed in Fig. 1). This includes using the DLA and CellSearch® system to provide well-characterized CTC preparations (A), the CellCelector™ automated micro-manipulation system to isolate single cells by capillary picking from cell culture vessels, to prepare true single-cell lysates and to print in situ single cell lysates in array format (B), and the Zeptosens RPPA module for sensitive and robust downstream protein profiling, including the usage of the adapted ZeptoChip platform, matched lysis, and printing buffer reagents, upfront validated specific assay antibodies and customized single cell sample array and data analysis routines (C). The modules of the workflow concept were stepwise realized, adapted, and tested with single cells from tumor cell line cultures and then adjusted afterward to CTC.

Fig. 1

Overview of the established workflow. Depicted is the established workflow for isolating and analyzing single CTCs with Reverse Phase Protein Arrays (RPPA). The workflow consists of three primary steps: (A) Capture and verification of CTCs using Diagnostic Leukapheresis (DLA) combined with the CellSearch® system. (B) Single cell/CTC picking with a glass capillary, followed by in situ lysing and mixing on a Micropick 48 slide™ and subsequent printing onto ZeptoChip using a modified CellCelector™ instrument. (C) RPPA analysis of the chip using the ZeptoReader

In panel A, the CTCs are captured using the CellSearch® Profile Kit, and surface markers are stained for imaging. In panel B, a single cell with approximately 20 nL of DPBS is aspirated from a culture dish into the lysis buffer pre-filled glass capillary of the CellCelector™ instrument (total inner volume 40 nL). The CellCelector™ is programmed to deposit 2 nL volume of single cell lysate per sample spot in an array, therefore a total of 40 nL sample lysate volume is sufficient to produce up to 6 sample arrays per chip with 3 technical replicate spots per sample per array. Finally, in panel C, downstream RPPA analysis using the ZeptoReader: immunoassays for selected protein markers are performed on the printed and blocked ZeptoChip lysate arrays, fluorescence images of the arrays are recorded and mean spot assay signals quantified with ImageJ software.

It is worth noting that the miniaturization described in this workflow represents a significant improvement over existing standard RPPA techniques. While some RPPA methods, such as Zeptosens RPPA, have demonstrated single-cell sensitivity, they still rely on standard vial 96/384 microtiter plate formats and printing robot routines that require large volumes of buffers with many µL of dead volumes, typically necessitating high numbers of starting cancer cells (on the order of 104 to 106). The main challenge of miniaturizing this approach was to achieve comparably good protein readout performance from true single (or few) cell preparations while ensuring all key specifications for each of the involved technology modules, such as lysis and printing.

We performed multiple replicates of lysates obtained from individual cells during the experiments and deposited them onto a chip. This approach enabled us to employ both blank and target protein assays to measure and quantify the signals, as described in their respective sections or figures. The average fluorescence signals from the printed spots were analyzed using standard Image J software routines (version 1.50f, NIH, USA) and subjected to statistical analysis using a paired t-test performed with Excel software.

Optimizing single cell lysis

A critical step in the workflow was to compose a specific mixture of cell lysis buffer and print buffer to meet the key requirements of this miniaturized one step approach: high efficiency for full cell lysis and good printability from the same composition in parallel; procedures, which for regular RPPA are so far being executed as separate steps (see [31] for more details on buffer requirements). Based on the experience with Zeptosens RPPA lysis buffer CLB1 and print buffer CSBL1, we followed the approach of presenting a concentrated mixture of CLB1:CSBL1 pre-aspirated into the glass capillary and subsequently picking the single tumor cell accompanied with a controlled shot of PBS, which dilutes out the CLB1/CSBL1 mixture. CLB1 has a strong denaturing activity compared to CSBL1 due to its higher urea and thiourea concentrations (7 M urea, 2 M thiourea, and high detergent in CLB1, 3.89 M urea, 1.11 M thiourea in CSBL1, no detergent). On the other hand, CSBL1 containing 11.1% glycerol is beneficial for good printing behavior of the cell lysate and has a low tendency to crystallize. The final buffer composition in a total volume of 40 nL had to guarantee efficient and full lysis at a short duration with high protein yield, prevent crystallization in the glass capillary, and ensure good compatibility with the array printing and final RPPA analysis process. To this aim, both buffers as well as different mixtures of both were selected and tested.

First, we determined completion of cell lysis indicated by changes in the cell’s and nucleus’ morphologies by microscopy in a time-lapse experiment with single cells of cultured MCF-7 breast cancer cell line, which were prelabeled with Celltracker Green CMFDA (Life Technologies, #C2925); final concentration 5 µM (1:2000) and NucBlue™ Live ReadyProbes™ Reagent (Hoechst 33342) (Invitrogen, R37605) in FITC and DAPI and incubated with different CLB1:CSBL1 ratios (1:0, 1:2, 1:5 and 1:10) (Fig. 2A). All CLB1:CSBL1 ratios lysed the cells completely in between 1 and 16 min. The fastest cell lysis with approx. 1 min was observed with CLB1-only buffer, followed by the 1:1 mix of CLB1 and CSBL1 with approx. 3 min. Lysis buffer ratios with higher contents of CSBL1 buffer resulted in lysis durations between 7 and 16 min (Fig. 2B). After careful consideration, we selected the 1:1 mix of CLB1 and CSBL1, which contained less (50%) detergent than CLB1-only buffer and caused minimal crystallization in the micromanipulator’s capillary. Although CLB1 buffer alone lysed the cells faster than the 1:1 mix of CLB1 and CSBL1, crystallization in the capillary made it unsuitable for our purposes.

Fig. 2

Optimization of conditions and hardware for single cell lysis and manipulation. (A) Monitoring of single cell lysis efficiency upon variation of lysis buffer composition. For better visualization, MCF-7 cells were prelabeled with Celltracker Green and NucBlue™ Live ReadyProbes™, and fluorescence images under the microscope were taken at indicated time points during lysis; Scale bar: 25 μm, Magnification:10x; DAPI (nucleus) and FITC (cytoplasm); (B) Lysis efficiency: comparison of complete lysis durations (min) for different CLB1:CSBL1 ratios; (C) Design of the holder for the ZeptoChip (left) and the adapter for the Micropick 48® slide, the Magnetpick® slide, and for the ZeptoChip (plus its holder) to position them precisely onto the CellCelector’s microscope stage (right)

In order to optimize single cell lysis and manipulation, additional hardware and software settings were implemented and adjusted. An adaptor was designed and implemented to position the ZeptoChip on the CellCelector™ microscope stage for fast and precise spotting of cell lysates. This setting was highly important to avoid drying artifacts due to the handling of the miniature volumes of single cell lysate (Fig. 2C).

Consequently, we finally applied this setting together with CellCelector™ glass capillaries pre-filled with 20 nL of a 1:1 mixture CLB1:CSBL1 and single cells picked with another 20 nL excess volume of DPBS under CellCelector™ control for all subsequent experiments.

Single cell lysate printing

To reproducibly print well-mixed single cell lysates with the CellCelector™, a series of steps were taken. Pre-labeled MCF-7 cells were scanned and lysed as described above. Although the lysis procedure immediately started in the micromanipulator’s glass capillary indicating that cell lysis buffer and DPBS were mixed, complete cell lysis and homogenous mixing of all components must be guaranteed. To this aim a total 40 nL of whole single cell lysate was released from the glass capillary into a circled hydrophilic area of a Micropick 48™ slide. The release of the single cell lysate mixture forced instantaneous good mixing of the glass capillary’s content; the completion of cell lysis was video-tracked under the microscope. After a final, efficient mixing by 5 times aspiration-and-release with the glass capillary on the Micropick 48™ slide, the printing procedure of the cell lysate sample started immediately using the same glass capillary and a ZeptoChip adjacently mounted on the CellCelector™ deck (Fig. 2C). Technical replicates of single droplet spots (typically n = 3) for each sample were printed by aspirating and release of approximately 2 nL volume portions of the complete 40 nL single cell lysate on the ZeptoChip surface, moving the glass capillary to pre-defined spot positions of the 6 pre-defined array fields of the ZeptoChip, one after the other. The lysate volume portions were deposited in close contact with the solution and the ZeptoChip surface. Up to three technical replicate spots per single cell sample were printed as inline series onto each of the 6 array replicate fields per ZeptoChip to examine the reproducibility of printing (see additional file 1).

The entire printing process was executed in a controlled atmosphere at 65% humidity. The printed ZeptoChip was kept at RT for 24 h followed by a drying incubation for 3.5 h at 37 °C. Finally, ZeptoChip was stored in the dark at 4 °C until further use.

Single cell protein analysis

The main difference to conventional RPPA is the way of producing the arrays, i.e., we printed them directly in situ after the single cell lysate preparation avoiding dead volumes. The following steps, especially performing the protein assays on the blocked and printed chips were conducted similarly as previously reported [31].

The available array replicates per chip were used to measure in parallel the expression and phosphorylation state of typically 2–4 different protein markers such as the cell surface protein Epithelial Cell Adhesion Molecule (EpCAM) and total or phosphorylated Erk1/2 and Akt proteins—the latter representing the key protein of the PI3K/Akt/mTOR pathway—in a direct fluorescence immunoassay with the proven specific RPPA-tested primary antibodies. The remaining arrays were used as control arrays to address potential auto-fluorescence and blank signal contributions in the absence of specific primary antibodies.

Data analysis of single cell lysate spot signals

The mean signals of the printed spots were analyzed using standard Image J software routines (version 1.50f, NIH, USA) and a flexible array of analysis spot circles. To account for variations in spot size due to capillary-to-capillary or single cell-to-single cell lysate variations, spot signals were analyzed as a mean signal density, with the spot area set constant for all printed spots at a diameter of approximately 150 μm to exclude background pixels. This was well comparable to the 160 μm analysis spot diameter applied in standard RPPA [31] and ensured reproducibility of the quantified assay signals from the technical spot replicates with mean coefficients of variation (CVs) well below 10% with a range between 4% and 7%. No significant variation of the mean signals from different arrays was observed. These observations were indicative of the robust printing and complete single-cell lysis processes. Unless otherwise stated, mean spot signals of the primary antibody and the blank assays were averaged to the mean values of the technical replicates (n = 2–3). Primary assay means were then corrected for the respective mean blank assay contributions in the absence of primary antibody, but under otherwise comparable conditions and given as Mean Fluorescence Intensity (MFI) values.

Protein analysis in single cellsEpCAM expression

The cell surface protein EpCAM represents the main target to detect and isolate CTCs in MBC. We, therefore, decided to measure EpCAM expression levels first on single BC cell line cells to verify the workflow’s performance and to prove the sensitivity of ZeptoChip’s readout. For this purpose, single EpCAM-positive MCF-7 and EpCAM-low/negative MDA-MB-231 cells were isolated and processed as described above.

The lysates of two separately processed pairs of single MCF-7 and MDA-MB-231 cultured cells (cell1, cell2) were printed in technical replicates (2 spots; n = 2) onto ZeptoChip replica arrays. On a separate replica array, only a secondary antibody was applied as a blank assay control.

After measuring the signal intensities with the ZeptoReader, the fluorescence images with the longest exposure time (4s) below pixel saturation (16 bit) were analyzed and single cell lysate spot signals were quantified as blank-corrected mean spot signals averaged over the technical replicates. As outlined in Fig. 3, the EpCAM MFI signals for both MCF-7 single cell preparations were detected well above blank, and were—as expected—significantly higher than for the EpCAM-low expressing MDA-MB-231 cells. The MDA-MB-231 cell 1 revealed very low, but still detectable residual EpCAM expression levels, for cell 2 no signal was detected. Moreover, these first workflow experiments yielded < 10% mean signal CVs ranging from 3 to 8% of the printed technical spot replicates, and hence indicated a good spot-to-spot reproducibility of cellular EpCAM protein detection (see also low error bars in Fig. 3 indicating standard deviations). Additionally, the EpCAM signals of two cells randomly picked from the same cell line (MCF-7, MDA-MB-231) were considerably different pointing towards well-known cell-to-cell heterogeneity. Nonetheless, such single cell protein variability and environmental changes seem now addressable with the presented method.

Fig. 3

EpCAM protein expression signals measured from single cell preparations using ZeptoChip. Quantification of fluorescence signals depicted from two MCF-7 and two MDA-MB-231 single cell lysate samples, respectively. The mean fluorescence signals (MFI) of both MCF-7 single cells were significantly higher than those of both MDA-MB-231 cells (p-value = 0.0435). Error bars indicate the standard deviations of the mean technical replicate spot signals (n = 3)

Measuring phosphorylated Akt in single tumor cell line cells upon Capivasertib treatment

In the next step, we tested if our workflow can be used to measure in parallel expression level and its phosphorylation status of a signal protein of interest. Since the PI3K/Akt signaling pathway represents a key pathway not only in BC development but also in its therapy resistance, we selected the Akt protein and its functional form phosphorylated at serine 473 (Akt-P-Ser473), in short pAkt, as key analytes. Phosphorylated Akt was also good to proof for sensitivity, since especially this phosphorylated analyte (and also others) is known for its relative low abundance and hence a challenge of detection. MCF-7 cells were treated with 5 µM capivasertib for 24 h. Capivasertib is an Akt1-specific inhibitor and is known to lead to elevated pAkt levels, due to its ATP-competitive mechanism of action and accumulation of inactivated hyper-phosphorylated Akt [36]. The increased pAkt levels were verified by Western Blot analysis of lysates from bulk MCF-7 cells (Fig. 4B). From the same cell culture used for Western Blot analysis, capivasertib-treated single MCF-7 cells (MCF-7/treated, 3 single cells) and untreated control cells (MCF-7/untreated, 2 single cells) were micromanipulated and processed according to the established workflow. Four technical replicates per single cell lysate were printed for each replica array. The arrays were incubated with primary antibodies against Akt and pAkt, followed by the anti-species Alexa Fluor™ 647 conjugated secondary antibody and fluorescence detection with the ZeptoReader (Fig. 4). One array was used as blank control.

Fig. 4

Expression signals of Akt and pAkt (Ser473) in treated single MCF-7 cells (description see main text)

Blank-corrected MFI fluorescence signals for pAkt in single MCF-7 cells were significantly higher compared to the total Akt protein (see Fig. 4). Treatment-to-Control Ratio (TCR) for pAkt as calculated from the respective single cell MFI mean values showed clear and significant up-regulation (2.4 fold, p = 0.0047) in MCF-7-treated over MCF-7-non-treated single cells (Fig. 4A left). In contrast, the TCR of total Akt protein remained almost unchanged (1.1), as expected from capivasertib’s exclusive effect only on the functional phosphorylation of Akt (see Fig. 4A left). Consequently, the mean TCR of the Akt-normalized pAkt/Akt ratios was then also more than 2-fold significantly higher (2.8 fold, p = 0.0012) in the MCF-7/treated over the MCF-7/non-treated cells (see Fig. 4A right). The observed single cell results have been verified by co-performed Western Blot analysis with bulk cell pellet lysate preparations (> 2 × 106 cells starting material) from respective same treated and non-treated MCF-7 cell line cultures, (with about 3-fold TCR of pAkt/Akt (see Fig. 4B).

In summary, the data obtained with ZeptoChip detection and true single cell lysates were well in accordance with western blot analysis measured from more than 2 × 106-fold starting cell material and supported that the established workflow is highly sensitive and able to measure both the expression and functional (phosphorylation) status of signaling protein markers in parallel from single cells and in a multiplex format.

Application of the single cell protein analysis workflow to CTCs

In order to adapt the workflow to CTCs, we decided to use fresh frozen aliquots of DLA-products for mainly two reasons [34]. First, the cells are not treated with a fixative such as paraformaldehyde, keeping the CTCs viable and avoiding negative effects hampering cell processing and the measurement of proteins and their phosphorylation levels. Second, since one aliquot is always inspected with the CellSearch® System, both CTC numbers and their expected quality were known.

Detection and processing of single CTCs

DLA aliquots were processed with the CellSearch® profile kit and the CellSearch® system to capture CTCs. Since it is essential to prevent the permeabilization of the cell membrane to avoid the loss of intracellular proteins, the CTCs were identified with a mixture of antibodies targeting only surface proteins [37]. Selection of the used antibodies, their type as well as applied detection labels were carefully coordinated with the RPPA detection routines. The labeled cells were scanned with the CellCelector™ automated micromanipulator (Sartorius, Jena), and single CTCs were isolated and lysed, as described before for the single cancer cell preparations.

The isolated single CTC lysates were printed onto the ZeptoChip slide in the array formats as described before.

Measuring phosphorylated Akt and Erk1/2 in CTCs derived from a MBC patient

Since we aimed to measure protein expression level and in parallel its functional phosphorylation status in single CTCs, we selected a cryopreserved and CTC-positive DLA product obtained from an MBC patient (patient 1, see additional file 2). Prior DNA sequencing analysis had confirmed that approx. 80% of her CTCs harbored the activating Akt1(E17K) mutation (see additional file 3) [38].

Although this mutation should result in the phosphorylation of the Akt protein at serine 473 [39], we did not know at which level and frequency. This sparked us to also measure the phosphorylation level of Erk1/2 protein. This protein is a key regulator of the MAPKinase signaling pathway for cellular growth, often interacting directly with the PI3K/Akt pathway in a cross-compensatory manner. For total Erk1/2 and phosphorylated Erk1/2 (Erk1/2-P-Thr202/Tyr204) proteins, well validated and measurable assays were available (since these protein forms are often presented at higher abundance/active state than e.g., pAkt), as part of our established list of RPPA assays (700 + antibodies). For this experiment, we also included—apart from the single CTC samples - lysates of single patient-matched WBCs, since in contrast to treated or untreated cell lines we do not dispose of ‘unstimulated’ CTCs from the same patient. Furthermore, we included validated RPPA standard lysates, which we regularly apply in routine RPPA protein profiling studies, as quality control (QC) for array printing and assay performance. These standard QC lysates had been prepared upfront and characterized from large bulk amounts of MCF-7 tumor cell line cultures (> 10 × 106 cells). These standard samples came as aliquoted pairs of treated and control (non-treated) MCF-7 lysates (kindly provided by NMI TT, Reutlingen, Germany) with pre-confirmed Akt and pAkt (and other protein) levels, and were co-printed onto the replica arrays of this experiment, at a concentration representing also single cell equivalent total protein material per spot, together with the true single CTC and WBC sample lysates.

Two different CTCs were isolated from a thawed DLA product and processed as described above (Fig. 5A) (see additional file 4). All CTCs, WBCs, and standard cell lysates were printed in technical replicates into ZeptoChip replica arrays, with single cell lysate volumes sufficient to determine the multiple Akt-P-Ser473 (pAkt), Erk1/2-P-Thr202/Tyr204 (pErk) protein and blank assay signals, respectively. The quality of the printed arrays is illustrated in Fig. 5 (top) with clippings of the fluorescence image recordings. The lysate print achieved homogeneous spot morphology, uniform spot diameters and a good signal reproducibility of the printed spots, evident from a low mean coefficient of variation (CV = 6%) of the sample MFIs quantified by the ImageJ software and averaged overall the different samples printed on the chip.

Fig. 5

Evaluation of phosphorylated Akt and Erk1/2 proteins in single CTCs derived from a MBC patient. (A) False color clippings of RPPA Zeptosens fluorescence array images (exposure 4s) for analysis of pAkt and pErk1/2 protein levels in CTC and WBC single cells from a MBC patient, and comparison to bulk cell QC (MCF) controls (3 replicate spots printed per sample). (B) Mean fluorescence signals (MFI) as quantified by ImageJ: elevated pAkt and pErk1/2 levels in CTCs compared to WBCs in line with QC lysates; pErk1/2 abundancy signal 3–11 times higher than pAkt as evident from different MFI scales

The blank-corrected mean fluorescence intensities (MFIs) showed more than two-fold difference among the different CTCs and were clearly more pronounced for pAkt than for pErk1/2 levels (see Fig. 5B). One of the two printed single CTC lysates (CTC2) showed a clear and significantly higher pAkt level compared to the patient-matched WBC (2.4-fold, p = 0.02), whereas the pErk1/2 levels (CTC1 and CTC2) were almost comparable (1.2-fold, p = 0.02). Notably, the pErk1/2 abundancy signals were 3–11 times higher than for pAkt as evident from the different absolute MFI scales (Fig. 5B). Significantly elevated pAkt and pErk1/2 levels observed in the treated standard QC lysates were well in agreement with the measured enhanced signals in the CTCs. Also, the absolute signal levels of the co-printed standard QC controls and the single CTC lysate MFIs were well in a comparable order range, even though they were prepared according to completely different protocols (one prepared from bulk cells and the other from only single cell amounts). Besides, the TCRs of the standard QC lysates were at well comparable values and quality when printed and measured (i) in a routine RPPA setting (> 106 cells lysed at mL volumes) and (ii) in the new, true single cell lysate workflow (single cells lysed at few nL volumes).

These data demonstrate that the new workflow can measure multiple proteins expressed with low and high abundancies from true single cell sample preparations.

Measuring phosphorylated Akt protein in single CTCs derived from two MBC patients

As a next step, we aimed to investigate the potential of the workflow in measuring expression and activation signals in CTCs derived from patient samples, both with and without the Akt1(E17K) mutation. To achieve this purpose, cryopreserved and CTC-positive DLA products were obtained from patient 1 [Akt1(E17K)] and another MBC patient (patient 2) with CTCs of wild-type AKT1 genotype (Supplement Seq Data). Single CTCs and WBCs were picked with the CellCelector™ and processed as described above (see Fig. 6A and B for results). The printed single cell CTC lysates of the two patients, co-printed patient-matched WBC lysates, and standard QC lysates were investigated, this time with a focus on pAkt and total Akt protein levels. As indicated by the low standard deviations and the QC standard lysate signals (Fig. 6), the quality of the printed arrays and the reproducibility of the replica spot MFI assay signals were again good and successfully attained. The CVs of mean replica spot MFI assay signals averaged over all printed CTC and WBC single cell lysates of patient 1, were 3% for the pAkt and for the total Akt. For the CTC and WBC single cell lysates of patient 2, similar low mean CVs were achieved (5% for pAkt and 3% for total Akt, respectively). With these numbers, print and assay signal CVs of the application were well comparable with those reported for standard RPPA applications (see [31] for numbers). The MFI signals and TCR of the co-printed standard QC lysates for both pAkt and Akt confirmed the correct readout process for both pAkt and total Akt. In the CTC single cells from patient 2 with AKT1 wildtype CTCs, both the Akt and pAkt average absolute signals were slightly below than in the patient-matched WBC (0.9-fold for pAkt, 0.8-fold for Akt), see Fig. 6B left. This resulted in about comparable pAkt/Akt mean signal ratios (5.4 fold for CTC, 4.1 fold for WBC), see Fig. 5B right. The pAkt/Akt signal ratios indicate that also pAkt/Akt levels in these two cell types were almost comparable. In CTCs isolated from patient 1 with AKT1 mutated CTCs, slightly higher absolute pAkt (1.1-fold) and almost double-fold Akt signals (1.9-fold) were observed compared to the respective WBCs (Fig. 6A left). This resulted in reduced pAkt/Akt signal ratios in the CTCs (mean 2.4-fold) compared to the WBCs (mean 4.5-fold, comparable to the respective wild-type ratio), see Fig. 6A right, mainly due to increased relative total Akt in the mutated compared to wild-type CTCs.

Fig. 6

Measuring pAkt to Akt protein in single CTCs derived from two index MBC patients. (A) (left) RPPA Zeptosens analysis of pAkt and Akt protein signals (MFI) in single CTC and WBC samples derived from index patient 1 harboring an Akt1(E17K) mutation, in addition to co-printed quality control samples (QC treated, QC control); (right) pAkt/Akt mean signal ratios of respective patient 1 single CTC and WBC samples. (B) (left) RPPA Zeptosens analysis of pAkt and Akt protein signals (MFI) in single CTCs and WBCs from index patient 2 (wild type); (right) pAkt/Akt mean signal ratios of respective patient 2 single CTC and WBC samples

Assessing also the good reproducibility of the quality control signal intensities in both chips (mean CV < 10%), in the patient with harboring CTCs with Akt1(E17K) mutation, the absolute MFIs of the CTCs lysates in the pAkt assay were clearly higher (1.8-fold) than those of the CTC lysates from the patient with AKT1(WT) CTCs (670 ± 47 for mutated versus 382 ± 39 for wildtype), and were even higher (3.4 fold) in the total Akt assay (274 ± 39 for mutated versus 81 ± 26 for wild-type). Although we do not know the AKT1 genotype of each CTC investigated, this observation suggests that the workflow has the capacity to measure a higher signal intensity for pAkt in CTCs obtained from the MBC patient with Akt1(E17K) mutated CTCs compared CTCs from the MBC with AKT1(WT) CTCs.

In summary and in line with our quality control measures, these data demonstrate that the ZeptoCTC workflow is robust and able to measure target proteins in single CTCs. Furthermore, the possibility to process up to 6 replica arrays/assays in a single chip run with a number of different single cell lysate preparations, opens the window to analyse several key pathway protein markers in parallel in a flexible (RPPA like) multiple assay and sample setting format.