Pharm Res (2018) 35:150 https://doi.org/10.1007/s11095-018-2422-5 RESEARCH PAPER Label-Free, Flow-Imaging Methods for Determination of Cell Concentration and Viability 1 1 1 2 1 A. S. Sediq & R. Klem & M. R. Nejadnik & P. Meij & Wim Jiskoot Received: 28 January 2018 /Accepted: 1 May 2018 # The Author(s) 2018 . . KEY WORDS cell based medicinal product cell viability ABSTRACT . . . Purpose To investigate the potential of two flow imaging micros- FlowCAM hemocytometry micro-flow imaging particle copy (FIM) techniques (Micro-Flow Imaging (MFI) and analysis FlowCAM) to determine total cell concentration and cell viability. Methods B-lineage acute lymphoblastic leukemia (B-ALL) cells of 2 different donors were exposed to ambient conditions. ABBREVIATIONS Samples were taken at different days and measured with MFI, ABD Area based diameter FlowCAM, hemocytometry and automated cell counting. B-ALL B-lineage acute lymphoblastic leukemia Dead and live cells from a fresh B-ALL cell suspension were CBMPs Cell-based medicinal products fractionated by flow cytometry in order to derive software filters DMSO Dimethylsulfoxide based on morphological parameters of separate cell populations K HPO Dipotassium phosphate 2 4 with MFI and FlowCAM. The filter sets were used to assess cell ECD Equivalent circular diameter viability in the measured samples. FACS Fluorescence-assisted cell sorting Results All techniques gave fairly similar cell concentration values FIM Flow imaging microscopy over the whole incubation period. MFI showed to be superior FSC Forward scatter with respect to precision, whereas FlowCAM provided particle HSA Human serum albumin images with a higher resolution. Moreover, both FIM methods MFI Micro-flow imaging were able to provide similar results for cell viability as the conven- MVAS MFI View analysis suite tional methods (hemocytometry and automated cell counting). MVSS MFI View system software Conclusion FIM-based methods may be advantageous over NaCl Sodium chloride conventional cell methods for determining total cell concen- SD Standard deviation tration and cell viability, as FIM measures much larger sample SSC Side scatter volumes, does not require labeling, is less laborious and pro- vides images of individual cells. INTRODUCTION Electronic supplementary material The online version of this article Cell-based medicinal products (CBMPs) are receiving (https://doi.org/10.1007/s11095-018-2422-5) contains supplementary increasing attention by the pharmaceutical industry be- material, which is available to authorized users. cause of their potential in treatment of a variety of Guest Editors: Karin Hoogendoorn and Christopher A. Bravery diseases,suchascancers,viral infections,and autoim- mune disorders (1,2). Like for any other pharmaceutical * Wim Jiskoot drug product, the quality of CBMPs highly determines email@example.com their safety and efficacy (3). The safety of a CBMP depends, amongst others, on the Division of BioTherapeutics, Leiden Academic Centre for Drug Research (LACDR), Leiden University, Leiden, The Netherlands presence of cellular and non-cellular impurities. When a spe- cific cell type is required for the therapy, unwanted cell pop- Department of Clinical Pharmacy and Toxicology, Leiden University Medical Center, Leiden, The Netherlands ulations are considered impurities and should be tested and 150 Page 2 of 10 Sediq et al. (2018) 35:150 controlled. Other impurities can be related to the process of morphological parameters of the particles can be extracted manufacturing, for instance, traces of raw materials used, like from the images (13). FIM can give valuable information micro carrier beads. about cells, without the need for labeling, and detect small Efficacy of the CBMPs is in mainly determined by cell- changes in cell size and morphology which have been shown specific functional properties; however the cell viability and total to be related to cell viability (14,15). In addition, FIM tech- concentration of the cells is also important in this. During pro- niques are generally easy and fast to perform. duction, starting cell materials undergo different manipulations, One of these FIM techniques is Micro-Flow Imaging such as harvesting, purification, genetic manipulation, expan- (MFI), which owes it popularity in this field mainly to sion, freezing and thawing (4). These process steps, as well as its user-friendliness and robust operation. The applica- other manipulations, such as storage, transport (5,6), and prep- tion of MFI for cells is limited but not unexplored. For aration for administration to the patient (7), can induce different instance, Martin et al. have used MFI to study aggrega- types of stresses to the cells and potentially affect cell viability. tion tendency of thawed hematopoietic stem cells (16). A Therefore, viability and total cell concentration are considered recent study of Farrell and coworkers used MFI to de- important quality attributes and should be measured as in- termine cell confluency on micro-carriers used in process control, at release and during stability testing. culture-derived bioreactors (17). Besides MFI, There are several methods to determine cell viability and FlowCAM has been explored for its potential in drug they all have their pros and cons. Despite continuous devel- product development (18–20) and drug delivery systems opment of new methods for assessing cell viability, (21). Even though MFI and FlowCAM are based on the hemocytometry (8), automatic cell counting and flow same measurement principle, they differ in several as- cytometry (9) are still the most frequently applied techniques pects. The FlowCAM has a higher resolution and pro- for this purpose. The flow cytometer provides information vides more particle parameters, whereas the MFI tends about cell size and granularity, respectively. Granularity level to provide a more accurate determination of particle and size has been shown to be inversely related with cell via- concentration (20,22). bility (10). The use of fluorescent dyes or fluorescently labeled The aim of this study was to evaluate the feasibility of two antibodies directed against cell surface markers can aid in the FIM techniques, i.e., MFI and FlowCAM, to determine the evaluation of cell type and viability. Flow cytometry is consid- total cell concentration and cell viability. In the presented ered to be a very accurate and reproducible technique for cell study, the two techniques are used for the characterization viability tests, but is also considered very laborious. The sam- of two cell lines, stored for up to 8 days under ambient con- ple preparation including, e.g., labeling of the cells, can be ditions, and compared with hemocytometry and automated time-consuming and expensive, and validation of flow cyto- cell counting as well as with each other. metric methods has been reported to be challenging (11). The determination of cell viability with a hemocytometer is based on staining of the dead cells using dyes like eosin and MATERIALS AND METHODS trypan blue (8). Hemocytometry is currently the gold standard in clinical practice for cell counting and viability determina- Cell Materials tion. Whilst being a fast method, the method can be laborious and has certain weak points (e.g. errors originating from sam- B-lineage acute lymphoblastic leukemia (B-ALL) cells were ple preparation; low analysis volume (0.1 μL)) that potentially used as model cells in this study. The cells were cultured from compromise the accuracy of the method. With respect to the two different donors (ALL-CR and ALL-CM (23); further quantity of cell suspension measured and the laborious nature referred to as cell line 1 and cell line 2) and provided by the of the method, automated cell counters could provide im- Department of Hematology, Leiden University Medical provement, especially for routine measurements, however Center (LUMC, Leiden, the Netherlands). Cells were frozen the analysis volume is still low (0.4 μL). Despite the higher in 60% v/v wash medium (BioWhittaker® Iscove’s Modified throughput of the automated cell counter compared to the Dulbecco’s Medium (IMDM) 98.5% v/v, penicillin/ hemocytometer (12). streptomycin (Lonza) 1% v/v, HSA-20% 0.5% v/v) and a In this regard, less laborious, inexpensive techniques that final concentration of 10% v/v HSA-20%, 10% v/v DMSO allow for rapid and reliable counting of CBMPs would be (LUMC, Leiden, the Netherlands) at a concentration of about beneficial for improving the quality and success of CBMPs 1×10 cells/mL and stored at −80°C until the start of incu- in clinical practice. Emerging flow imaging microscopy bation experiments. After controlled thawing, washing and (FIM) techniques may fulfill these needs. In these systems the counting (using a hemocytometer), cells were suspended in sample flows through a flow cell where images are taken with a NaCl 0.9% m/v + HSA-20% 2% v/v at a concentration of high-magnification digital camera. With the help of the ded- 10 total cells/mL and were exposed to the stress condition icated instrument software, the quantity and several described below. Label-free, flow-imaging methods for determination of cell (2018) 35:150 Page 3 of 10 150 Stress Condition Hercules, California, USA). This procedure was repeated three times for each sample. The chosen stress condition was storage at ambient conditions (on the lab bench) for up to 8 days, during which the cells Micro-Flow Imaging (MFI) showed a gradual and reproducible decrease in viability, as demonstrated in a pilot experiment. The total cell concentra- In order to reduce the probability of detection of optically tion and morphological parameters (see below) of each sample overlapping particles, cell-containing samples were first dilut- were analyzed at different days by using MFI and FlowCAM. ed 4-fold with particle free NaCl 0.9% m/v + HSA-20% 2% For cell line 1 the measurements were performed on day 0, 1, v/v. The diluted samples were analyzed by using a Micro- 2, 4, 6 and 8. The study performed with the 2nd cell line Flow Imaging 5200 (Protein Simple, Santa Clara, CA, served as a confirmative study and therefore the measure- USA), with MFI View System Software (MVSS) Version 2. ments on day 4 and 6 were not performed. In parallel, the No software filters were applied during the runs. The 100-μm total cell concentration and number of viable cells in the same silane coated flow cell was rinsed with flow of ultrapure water samples were determined with a hemocytometer and an au- (18.2 MΩ.cm; dispensed by using a Purelab Ultra water pu- tomated cell counter. For the data analysis of both flow imag- rification system, ELGA LabWater, Marlow, UK) and there- ing microscopy techniques we only included particles with a after a background measurement was taken with particle free size ≥4 μm, which is the lowest detectable particle size for the NaCl 0.9% m/v + HSA-20% 2% v/v. For the analysis, automated cell counter, according to the manufacturer. 0.50 ml of each sample was run at a flow rate of 0.17 mL/ min. The data analysis was performed with MFI View Analysis Suite (MVAS) Version 1.2. For the data analysis, Fluorescence-Activated Cell Sorting (FACS) the lower size limit was set at 2 μm in order to avoid analysis on the edge of the detectable particle size range (i.e., 1 μm). Fluorescence-activated cell sorting (FACS) on a FACSAria III The upper size limit was set at 20 μm because particles larger (BD Biosciences, New Jersey, USA) was used to separate dead than that were most likely contaminants (e.g., dust) and con- and/or dying cells from living cells based on the forward scat- tributed to less than 0.1% of the total particle concentration. ter (FSC) and side scatter (SSC) ‘live gate’ as shown in Table I summarizes the main morphological parameters pro- Supplementary Fig. S1. The presumably dead and live cells vided by the MVAS and their descriptions. The size distribu- that were collected with the FACS were then measured with tion of each sample was presented in equivalent circular di- MFI and FlowCAM. These FIM data were used to develop ameter (ECD). Each sample was measured three times with software filters for dead and live cells for each FIM technique MFI. (see results section). FlowCAM Hemocytometry The second flow imaging technique used in this study was a Cell suspensions were diluted twofold with a trypan blue so- FlowCAM VS1 (Fluid Imaging Technologies, Yarmouth, lution (0.4% (w/v)in0.81% m/vNaCland 0.06% m/v ME, USA). After rinsing the FC50 flow cell with ultrapure K HPO (Bio-Rad, Hercules, California, USA)). Ten μLof 2 4 water, 100 μL of each 4-fold diluted sample was run at a flow the mixture was placed on a Bright-Line hemocytometer glass rate of 0.030 ml/min controlled by a C70 syringe pump. (Sigma-Aldrich, Steinheim, Germany), and analyzed by using Images were taken with a Sony XCD-SX90 camera at 22 a light microscope (Zeiss Axiostar Plus, Carl Zeiss Light fps (shutter: 8, gain: 224, 20× lens). The data were analyzed Microscopy, Göttingen, Germany) with 10× magnification. by Visual SpreadSheet Version 3. For reasons described in the Both viable (not stained) and non-viable (stained) cells were MFI section, only particles between 2 and 20 μmwereinclud- counted in 25 frames of the hemocytometer according to the ed in the data analysis. In order to remove edge particles manufacturer’s recommendations and the percentage of via- (particles that were detected at the borders of the camera field, ble cells to the total was calculated (24). For each sample hence imaged partially), the acceptable detection field was triplicate measurements were conducted. reduced to 95–1183 and 6–952, respectively, for left-right and top-bottom orientations. The edge gradient parameter Automated Cell Counting provided by FlowCAM was used to exclude out-of-focus par- ticles. The acceptable range for edge gradient was determined Ten μL of the mixture prepared for the hemocytometry was in a preliminary study. In Table I, descriptions of the main introduced into the counting slide. Subsequently, the total cell morphological parameters provided by the Visual concentration and percentage of viable cells were measured SpreadSheet are given. It is worth mentioning that the FlowCAM can calculate the particle size through two different by using a Bio-Rad TC20 Automated Cell Counter (Bio Rad, 150 Page 4 of 10 Sediq et al. (2018) 35:150 Table I Morphological parameters used in this study and their descriptions as provided by MVAS (MFI) and Visual SpreadSheet (FlowCAM) Parameter Unit Description Micro-Flow Imaging Equivalent circular diameter (ECD) Microns The diameter of a circle occupying the same area as the particle Intensity mean Intensity (0–1023) The average intensity of all image pixels representing the particle Intensity standard Deviation Intensity (0–1023) The standard deviation of the intensity of all pixels representing the particle Circularity No units (0–1) The circumference of a circle with an equivalent area divided by the actual perimeter of the particle Aspect ratio No units (0–1) The ratio of the minor axis length over the major axis length of an ellipse that has the same second-moment-area as the particle FlowCAM Area based diameter (ABD) Microns The diameter based on a circle with an area that is equal to that of the particle Equivalent spherical diameter (ESD) Microns The mean value of 36 feret measurements (the perpendicular distance between parallel tangents touching opposite sides of the particle; VisualSpreadsheet makes 36 feret measurements for each particle, one each 5 degrees between −90 degrees and + 90 degrees) Symmetry No units (0–1) A measure of the symmetry of the particle around its center; if a particle is symmetric, then the value is one Aspect ratio No units (0–1) The ratio of the width (the shortest axis of the particle) and length (the longest axis of the particle) Circle fit No units (0–1) Deviation of the particle edge from a best-fit circle, normalized to the zero to one range where a perfect fit has a value of one Circularity No units (0–1) A shape parameter computed from the perimeter and the area; a circle has a value of one (formula: (4 x π x Area) / Perimeter ) algorithms (described in Table I). In our study we chose to predetermined time points, as described in materials and proceed with the area based diameter (ABD), because the methods. principles of ABD and ECD are similar. Monitoring Total Cell Concentration Over the Eight-Days Study Period Definition of Software Filters to Discriminate Live and Dead Cell Populations Figure 1 shows the total (live and dead) cell concentrations as measured with all four techniques over the 8-days study peri- Based on measurements of the FACS-sorted live and dead od, for both investigated cell lines. The results indicate that all particle poplulations (see above), the parameter that showed the techniques gave fairly similar total cell concentrations. the largest difference (ECD for MFI and ABD for FlowCAM) FlowCAM appeared to have the lowest precision, followed between live and dead cells, was used as a primary filter. After by hemocytometry, as judged by the standard deviations. applying this primary filter, the changes of all the other pa- The total cell concentration of cell line 1 showed a slowly rameters were evaluated and their threshold values could sys- decreasing trend over time, whereas for cell line 2 the cell tematically be fine-tuned. At the end, the established set of counts remained fairly stable. morphological filters was tested on the analyzed sorted frac- tions and FIM-derived viability was compared to the trypan Sorting and FIM Derived Morphology of Dead and Live blue-assisted values found for each cell sample (see results Cells section for details). The fresh cell suspension of cell line 1 was analyzed with a flow cytometer to derive an appropriate gate RESULTS for sorting dead and live cells (see Supplementary Fig. S1). After the fractions were collected, a control trypan The two B-ALL cell lines were thawed, analyzed and then left blue assisted viability test of each fraction was per- at ambient temperature for 8 days and analyzed at formed on the automated cell counter. From these Label-free, flow-imaging methods for determination of cell (2018) 35:150 Page 5 of 10 150 Fig. 1 Total cell concentration of (a) cell line 1 and (b) cell line 2 during 8 days of storage at ambient conditions measured by hemocytometry (black), automated cell counting (red), MFI (yellow) and FlowCAM (blue). Error bars represent the standard deviation of triplicate measurements with each technique. control measurements, it was found that the live popu- Comparing Cell Viability Determination by FIM lation contained almost 90% viable cells, whereas the Techniques, Hemocytometry and Automated Cell dead population contained no more than 20% viable Counting cells. For comparison, the viability of the unfractionated andunstressedcellpopulation was about 75%. With the help of the established software filters, the percentage The sorted populations were measured with both of viable cells, i.e., cell viability, was calculated for both cell lines FIM techniques. Although it was difficult to see the at different storage time points (Fig. 4). These percentages ob- differences visually (Figs. 2 and 3), evaluation of differ- tained from different techniques showed similar trends for the ent morphological parameters for live and dead cells viability of both cell lines, i.e., a gradually decreasing viability showed a statistically significant difference between the over incubation time. In addition, cell line 2 showed a stronger values of each listed parameter for live and dead cells, survival at the studied incubation conditions (75% decrease in except for the circularity values derived from MFI and viability after 8 days) than cell line 1 (50% decrease). FlowCAM (Table II). The filters were constructed by setting one value at a time starting from the most dis- Morphological Parameters Obtained by FIM tinctive parameter until no change was observed in the Techniques populations. Using this approach we defined software filters based on the monitored morphological parameters Both FIM techniques provide morphological parameters of the for dead and live cells, as shown in Table III. detected particles (including cells) obtained from the individual MFI morphology ﬁlter FACS sorted Live Dead Live Dead 2 – 4 µm 4 – 11 µm 11 – 20 µm Fig. 2 Representative images of particles detected by MFI in a B-ALL cell population. The left column shows the particles that were identified as dead or live cells based on the morphological filters. The right column shows images of particles that were found in FACS-assisted sorted samples of live and dead cells. The MFI morphological filter (for dead and live cells) uses only the size in the range between 4 and 11 μm. 150 Page 6 of 10 Sediq et al. (2018) 35:150 FlowCAM morphology ﬁlter FACS sorted Live Dead Live Dead 2 – 4 µm 4 – 12 µm 12 – 20 µm Fig. 3 Representative images of particles detected by FlowCAM in a B-ALL cell population. The left column shows the particles that were identified as dead or live cells based on the morphological filters. The right column shows images of particles that were found in FACS-assisted sorted samples of live and dead cells. The FlowCAM morphological filter (for dead and live cells) uses only the size in the range between 4 and 12 μm. images. Representative images of individual particles detected compared to MFI. Nevertheless, some of the parameters by the two FIM techniques are shown in Figs. 2 and 3.When allowed a comparison of FlowCAM with MFI. In our 8-days comparing these cell images derived from MFI and FlowCAM, study we monitored changes in all the 5 parameters listed in it is obvious that FlowCAM has a much higher lens magnifica- Table I for the studied cells. Most of these parameters showed a tion. The high-resolution images of FlowCAM may result in trend of a gradual change in the course of the 8-day study. the ability to derive more morphological parameters as Figure 5 shows the size distribution of particles derived from the MFI and FlowCAM analysis for both cell lines dur- ing the 8-days incubation study. All the size distribution Table II Derived Morphological Parameters (mean ± standard deviation) graphs show that there was a decrease in the number of the Provided by MFI and FlowCAM for the Two Cell Fractions of Cell Line 1 larger particles and an increase in the number of smaller par- Sorted Using FACS ticles over time. When focusing on cell line 1 (Fig. 5a, b), it is Flow imaging microscopy Live cell Dead cell R * obvious that the peak around 8 μminthe MFI-derived distri- morphological population population bution on day 0 slowly descended, while a new peak around parameters* Micro-Flow Imaging Table III Specification of the Software Based Morphological Filters Used to ECD 7.6 ± 2.2 μm 5.8 ± 1.8 μm0.171 Identify Dead And Live Cells from Analysis Results of the FIM Methods IntMean 546 ± 87 573 ± 81 0.026 Flow imaging microscopy Filter for live cell Filter for dead IntSD 179 ± 53 173 ± 55 0.003 morphological parameters population cell population Cir 0.88 ± 0.06 0.88 ± 0.05 0 AR 0.85 ± 0.12 0.87 ± 0.10 0.008 Micro-Flow Imaging FlowCAM ECD 7.25–11 μm4–7.25 μm ABD 7.7 ± 3.0 μm 6.4 ± 2.6 μm0.055 IntMean ≤ 550 ≥ 550 Sym 0.74 ± 0.19 0.69 ± 0.17 0.019 IntSD ≤ 170 ≥ 170 AR 0.82 ± 0.16 0.80 ± 0.13 0.005 FlowCAM CF 0.77 ± 0.17 0.75 ± 0.13 0.004 ABD 7–11 μm4–9 μm Cir 0.78 ± 0.17 0.78 ± 0.12 0 Sym ≥ 0.7 ≤ 0.8 AR ≥ 0.8 0.6–0.8 *Statistical comparison of the morphological parameters of the dead and live CF ≥ 0.7 ≤ 0.8 cells. The comparison is derived after applying t-test with GraphPad Prism 5®. R quantifies the fraction of all the variations in the samples that is accounted Cir ≥ 0.8 0.3–0.8 for by a difference between the group means Label-free, flow-imaging methods for determination of cell (2018) 35:150 Page 7 of 10 150 Cell line 1 Cell line 2 100 100 Hemocytometry 80 80 Automated cell counting 60 60 FIM (MFI) 40 40 FIM (FlowCAM) 20 20 0 0 0 2 4 6 8 0 2 4 6 8 Time (days) Time (days) Fig. 4 Cell viability determined with different analytical methods for cell line 1 at different time points during 8 days of storage at ambient conditions: hemocytometry (black), automated cell counting (red), MFI (yellow) and FlowCAM (blue). The error bars represent standard deviations of triplicate measure- ments with each method. 6 μm arose and became apparent on day 8. For FlowCAM, vertical lines in Fig. 5) were used with MFI and FlowCAM, the change in the ABD size distribution was fairly similar to respectively. The lower size limit was fixed for both methods, the changes seen in the ECD size distribution with MFI. Fresh because the lowest detectable particle size of the automated cell sample showed a distinct peak at about 12 μm and over cell counter is 4 μm. Particles below 4 μm may consist of time this peak disappeared and was replaced by relatively fragmented cells or cell debris. Particles above 11 or 12 μm broad peak at a smaller size range (around 6 μm). However, include clustered cells or aggregates. From the size distribu- there was also a third transient peak at about 8 μmseen, which tion graphs in Fig. 5 it is seen that the concentrations of pre- was already present in the fresh sample and increased after sumably cell debris (particle size 2–4 μm) and clustered cells or 24 h, and then diminished during the following days. MFI and aggregates (particle size 11–20 μm for MFI and 12–20 μmfor FlowCAM also showed similar trends for cell line 2, which FlowCAM) change over time. The concentration of the 2– however showed a somewhat smaller cell size for the main 4 μm particles steadily increased as measured with MFI. population on day 0. With FlowCAM the increase was only apparent during the For the determination of viability only the size range be- first few days. With respect to the aggregates, both FIM tech- tween 4 and 11 μmand 4–12 μm (indicated by the dotted niques showed decreasing concentrations. ab day 0 10 10 day 1 8 15 8 day 2 6 6 day 4 4 4 day 6 2 2 day 8 0 0 5 10 15 20 5 10 15 20 Equivalent circular diameter ( m) Area based diameter ( m) cd 10 10 8 8 6 6 4 4 2 2 0 0 5 10 15 20 5 10 15 20 Equivalent circular diameter ( m) Area based diameter ( m) Fig. 5 Size frequency distribution of the particles encountered in samples of cell line 1 (a, b) and cell line 2 (c, d) measured with MFI (a, c) and FlowCAM (c, d) during 8 days of storage at ambient conditions. The size range actually used for the determination of total cell concentration and cell viability by using morphological filters is indicated with the vertical dotted lines. Viability (%) Frequency (% ) Frequency (% ) Viability (%) Frequency (%) Frequency (%) 150 Page 8 of 10 Sediq et al. (2018) 35:150 Changes were also observed there during the 8-day study sample) and thus count considerably more cells than the con- in the other morphological parameters of both FIM tech- ventional methods such as hemocytometry. This ability is a niques. The details are shown in the Supplementary great asset for the accuracy and precision of both total cell Information (see Supplementary Fig. S2-S5 for both cell lines). concentration and cell viability determinations. Another ad- vantage of FIM lies in its ability to image individual cells and obtain morphological characteristics of each detected cell. DISCUSSION Moreover, non-cellular materials (e.g., beads) can be manual- ly removed from the data, to avoid inaccurate counting. Viability and total cell concentrations are important quality Furthermore, the imaging capability of especially FlowCAM attributes of CBMPs. The availability of a rapid, easy and may allow the discrimination of different types of cells in a heterogeneous cell population. reliable method is highly beneficial for viability measurement of cells from starting material procurement throughout the Although hemocytometry, automated cell counting, MFI drug product manufacturing process including CBMP release and FlowCAM are all able to provide an estimation of the testing, storage, shipment and administration to the patient. total and viable cell concentration, standard deviations pre- In this study we have investigated the potential of two dif- sented in Fig. 1 indicate that the precision of methods with ferent FIM techniques to assess the total cell concentration respect to the cell concentration measurement follows this and cell viability. For this purpose, the cells were placed at pattern: MFI > automated cell counter > hemocytometry > ambient conditions for up to 8 days, because pilot studies FlowCAM. For FlowCAM, the combination of the type of had shown that this is a convenient way to reproducibly obtain flow cell and image frequencies used in our settings resulted changes in viability as function of time. Moreover, the chosen in a theoretical analysis efficiency of around 20%, meaning conditions are relevant from a clinical perspective, as cells are that only 20% of the dispensed cell suspension was actually commonly kept for some time at room temperature before imaged. This limitation in combination with the inability of administration. Our data show that both techniques can be the analysis package in exclusion of stuck particles (i.e., parti- used to measure the concentration and the viability of cells, cles adhered to the measurement cell) that appear in several yielding comparable results to those obtained with conven- images from the analysis may be the most important contrib- tional cell counting methods. uting factors to the relatively low precision. Despite the larger The FIM methods, once developed, are easy to perform amount of suspension volume measured with the automated and do not require labeling of the cells before the measure- cell counter (cf. Table IV), automated cell counting did not ment (see comparison made in Table IV). These techniques offer a much better precision over hemocytometry with re- measure a relatively large sample volume (of 4-fold diluted spect to total cell concentration. This may be caused by the Table IV Comparison of the Characteristics of the Techniques Evaluated in this Study Characteristics Hemocytometry Automated cell FIM (MFI) FIM (FlowCAM) counting General Analyzed sample volume 0.1 μL0.4 μL260 μL20 μL Sample pretreatment Labeling; dilution if needed Labeling; dilution if needed Dilution Dilution Analysis time per sample 5 min 1 min 15 min 10 min (depends on (measurement + data analysis) measurement settings) Cell counting and viability determination Accuracy* Moderate Moderate High Moderate Precision Moderate Moderate High Low Additional features Non-cellular particles Discarded visually from May interfere with the Can be removed Can be removed from the cell counts cell counting from the data afterwards the data afterwards Detection of cell debris Depends on magnification Not possible Possible Possible lens used Cell identification Visual identification Not possible Probable, when using Highly probable, when using morphology morphology based software filters based software filters *Determined based on the effectively measured sample volume, extrapolation factor to final cell concentration unit (e.g., cells/mL) Determined based on the outcome of our study Label-free, flow-imaging methods for determination of cell (2018) 35:150 Page 9 of 10 150 interference of non-cellular material (e.g., contaminants such Moreover, cells at later time points (as of day 1) appeared to as dust) with the cell counting. In contrast, MFI resulted in the have a higher intensity, a lower intensity SD and smaller aspect highest precision. The relatively large volume of imaged sam- ratios (see Supplementary Fig. S2 and S3). The latter, together ple (more than 65 μL vs. 0.1 μL in hemocytometry), high with lower symmetry and circle fit values, indicate that the cells analysis efficiency (about 85%) and ability to remove stuck became less symmetric and more elongated in shape. All these particles (see above) may explain the pronounced perfor- observations point towards a decrease in the population of live mance of MFI with respect to the total cell concentration cells and/or changes in the quality of the live cell population, determinations. Moreover, the precision of hemocytometry e.g., because of apoptosis. Shrinkage of cell size and changes in may be affected by the operator, since the method requires cell shape are observed for dead and dying cells whose concen- visual counting and viability assessment based on visual dis- trations are expected to increase under stress (14). crimination of the color of the cells. Furthermore, a decrease in the intensity SD can be a sign of Stability studies mimicking the clinical conditions of cell disappearance of the cell organelles that contributed to varia- preparation for administration and storage at ambient condi- tions in intensity of the image of a cell. Further investigation tions (i.e., in-use stability) are very important to establish revealed that upon storage at ambient temperature also within whether the cells are still viable upon administration. the defined populations (e.g., live and dead cells) changes in Monitoring of the total cell concentration of cell line 1 over morphology were observed, most obviously in the FlowCAM storage time revealed a decreasing trend in total cell concen- data. This suggests that in particular FlowCAM is able to pick tration for all the methods. It is known that dying and dead up early stages of loss in viability that may cause changes in the cells undergo fragmentation into smaller particles (25). These transparency and shape of the cells (14). cell debris particles are below the lower size limit of the auto- It has to be noted that the two FIM techniques described mated cell counting (4 μm) and the lower size limit chosen for herein offer several other morphological parameters that FIM techniques, while in hemocytometry non-cellular parti- could be used in analysis of the cells. However, the combina- cles are visually excluded and therefore not counted. FIM data tion of a high-magnification lens and a thin focus plane of the confirm this, as the ECD and ABD size distribution diagrams flow cell results in substantial numbers of imaged particles that (Fig. 5) show a clear increase in the number of particles below were out of focus. These particles affect the values of a few 4 μm over time, especially within the first 24 h of incubation. morphological parameters (e.g., intensity), and therefore were This observation suggests that the FIM techniques are pre- not included in our study. sumably able to detect and count cell fragments as well, which During the development of our FIM based methods, we may be useful for product characterization. However, it is not have used one type of cells, i.e., B-ALL cell lines. However, easy to reliably determine morphological parameters of these CBMPs may contain cells with different morphological prop- erties than B-ALL cells or contain multiple cell types (27). In small particles, which appear as blurry dots in MFI and show very limited morphological attributes in FlowCAM (Fig. 2 and principle, one can apply the same approach to other cell types Fig. 3,respectively). and heterogeneous cell populations. Therefore, one needs to Another interesting finding from our study is that the FIM develop specific software filters for each type of cells. techniques were able to detect clustered or aggregated cells. It Our investigation shows that FIM techniques have the po- may be expected that clustered cells will always be not picked tential to determine total cell concentration and cell viability. up with the conventional techniques, given the low concentra- This provides a strong basis for more detailed and specific tion of this population (Fig. 5) and low sample volume mea- studies in order to validate FIM-based methods for a broad sured by these techniques. It should be noted that one cannot range of CBMPs. The FIM techniques, just like the conven- exclude that some of the aggregates detected by FIM are tional techniques, were able to detect differences in cell viabil- artefacts created by the transition of the wide tubing diameter ity between the two cell lines. This may be caused by the to the narrow flow cell, which condenses the cells in response difference in genetic background or disease status of the do- to spatial shrinking. However, the aggregates were also detect- nors. Moreover, we have shown other potentials that these ed when more diluted samples (up to 16×) were measured techniques can offer in characterization of CBMPs. The ca- with FIM techniques. Furthermore it is shown that aggregates pability of detecting and imaging cell debris, cell aggregates are formed around decaying and dead cells (26). and potentially different cell types offers an excellent applica- Analysis of the FIM parameters revealed clear changes dur- tion in characterization of impurities in CBMPs. ing storage in the majority of the parameters highlighted in this study, namely ECD, intensity mean, intensity SD and aspect ratio for MFI; and ABD, symmetry, aspect ratio, and circle fit CONCLUSION for FlowCAM. Both MFI and FlowCAM revealed a decreasing trend in the size (ECD and ABD, respectively) of the cells dur- In this study we have developed label-free methods for cell ing the storage for up to eight days under ambient conditions. concentration and viability determination based on two 150 Page 10 of 10 Sediq et al. (2018) 35:150 during drug development: II assays. J Immunol Methods. different FIM techniques, MFI and FlowCAM. Our data sug- 2011;363(2):120–34. gests that both methods deliver fairly similar results for total 12. Cadena-Herrera D, Esparza-De Lara JE, Ramirez-Ibanez ND, cell concentration and cell viability as traditional methods, i.e., Lopez-Morales CA, Perez NO, Flores-Ortiz LF, et al.Validation hemocytometry and the automated cell counting. Whereas of three viable-cell counting methods: manual, semi-automated, and automated. Biotechnol Rep. 2015;7:9–16. the MFI based method showed a higher precision with respect 13. Sharma DK, King D, Oma P, Merchant C. Micro-flow imaging: to determination of the total cell concentration, the FlowCAM flow microscopy applied to sub-visible particulate analysis in pro- based method provides higher-resolution images. The latter tein formulations. Aaps Journal. 2010;12(3):455–64. may be useful to identify non-cellular particles and potentially 14. Matsubayashi M, Ando H, Kimata I, Nakagawa H, Furuya M, discriminate between different types of cells. Tani H, et al. Morphological changes and viability of Cryptosporidium parvum sporozoites after excystation in cell-free culture media. Parasitology. 2010;137(13):1861–6. Open Access This article is distributed under the terms of the 15. Tsaousis KT, Kopsachilis N, Tsinopoulos IT, Dimitrakos SA, Kruse FE, Welge-Luessen U. Time-dependent morphological al- Creative Commons Attribution 4.0 International License terations and viability of cultured human trabecular cells after ex- (http://creativecommons.org/licenses/by/4.0/), which per- posure to trypan blue. Clin Exp Ophthalmol. 2013;41(5):484–90. mits unrestricted use, distribution, and reproduction in any 16. Wu L, Martin T, Li Y, Yang L, Halpenny M, Giulivi A, et al.Cell medium, provided you give appropriate credit to the original aggregation in thawed haematopoietic stem cell products visualised using micro-flow imaging. Transfus Med. 2012;22(3):218–20. author(s) and the source, provide a link to the Creative 17. Farrell CJ, Cicalese SM, Davis HB, Dogdas B, Shah T, Culp T, Commons license, and indicate if changes were made. et al. Cell confluency analysis on microcarriers by micro-flow imag- ing. Cytotechnology. 2016;68(6):2469–78. 18. Werk T, Volkin DB, Mahler HC. Effect of solution properties on REFERENCES the counting and sizing of subvisible particle standards as measured by light obscuration and digital imaging methods. European J 1. Larijani B, Esfahani EN, Amini P, Nikbin B, Alimoghaddam K, Pharm Sci. 2014;53:95–108. Amiri S, et al. Stem cell therapy in treatment of different diseases. 19. Frahm GE, Pochopsky AW, Clarke TM, Johnston MJ. Evaluation Acta Med Iran. 2012;50(2):79–96. of microflow digital imaging particle analysis for sub-visible parti- 2. Mason C, Brindley DA, Cume-Seymour EL, Davie NL. Cell ther- cles formulated with an opaque vaccine adjuvant. PLoS One. apy industry: billion dollar global business with unlimited potential. 2016;11(2):e0150229. Regen Med. 2011;6(3):265–72. 20. Zölls S, Weinbuch D, Wiggenhorn M, Winter G, Friess W, Jiskoot 3. International Conference on Harmonisation (ICH). Specifications: W, et al. Flow imaging microscopy for protein particle analysis–a test procedures and acceptance criteria for biotechnological/ comparative evaluation of four different analytical instruments. biological products. In: ICH harmonised tripartite guideline Q 6 AAPS Journal. 2013;15(4):1200–11. B; 1999. 21. Sediq AS, Waasdorp SK, Nejadnik MR, van Beers MM, 4. Moviglia GA, Vina RF, Brizuela JA, Saslavsky J, Vraslovic F, Meulenaar J, Verrijk R, et al. Determination of the porosity of Varela G, et al. Combined protocol of cell therapy for chronic spinal PLGA microparticles by tracking their sedimentation velocity using cord injury. Report on the electrical and functional recovery of two a flow imaging microscope (FlowCAM). Pharm Res. 2017;34(5): patients. Cytotherapy. 2006;8(3):202–9. 1104–14. 5. Hickey MJ, Malone CC, Erickson KE, Lin A, Soto H, Ha ET, et al. 22. Weinbuch D, Zölls S, Wiggenhorn M, Friess W, Winter G, Jiskoot Implementing preclinical study findings to protocol design: transla- W, et al. Micro-flow imaging and resonant mass measurement tional studies with alloreactive CTL for gliomas. Am J Transl Res. (Archimedes)–complementary methods to quantitatively differenti- 2012;4(1):114–26. ate protein particles and silicone oil droplets. J Pharm Sci. 6. Gomez-Lechon MJ, Lahoz A, Jimenez N, Bonora A, Castell JV, 2013;102(7):2152–65. Donato MT. Evaluation of drug-metabolizing and functional com- 23. Nijmeijer BA, Szuhai K, Goselink HM, van Schie ML, van der petence of human hepatocytes incubated under hypothermia in Burg M, de Jong D, et al. Long-term culture of primary human different media for clinical infusion. Cell Transplant. 2008;17(8): lymphoblastic leukemia cells in the absence of serum or hematopoi- 887–97. etic growth factors. Exp Hematol. 2009;37(3):376–85. 7. Leverett LB, Hellums JD, Alfrey CP, Lynch EC. Red blood cell 24. Louis KS, Siegel AC. Cell viability analysis using trypan blue: man- damage by shear stress. Biophys J. 1972;12(3):257–73. ual and automated methods. Methods Mol Biol. 2011;740:7–12. 8. Strober W. Trypan blue exclusion test of cell viability. Current 25. Kerr JFR, Wyllie AH, Currie AR. Apoptosis - basic biological phe- protocols in immunology / edited by John E Coligan 2001; nomenon with wide-ranging implications in tissue kinetics. Brit J Appendix 3: Appendix 3B Cancer. 1972;26(4):239-&. 9. Panterne BR, Richard MJ, Sabatini C, Ardiot S, et al. Ten Years of 26. Renner WA, Jordan M, Eppenberger HM, Leist C. Cell-cell adhe- External Quality Control for Cellular Therapy Products in France: sion and aggregation: influence on the growth behavior of CHO InTech; 2011 cells. Biotechnol Bioeng. 1993;41(2):188–93. 10. Donner KJ, Becker KM, Hissong BD, Ahmed SA. Comparison of multiple assays for kinetic detection of apoptosis in thymocytes ex- 27. Services USDoHaH. Cellular & gene therapy products: marketed posed to dexamethasone or diethylstilbesterol. Cytometry. products. 2016. http://www.fda.gov/BiologicsBloodVaccines/ 1999;35(1):80–90. CellularGeneTherapyProducts/ApprovedProducts/ (Accessed 28 11. O'Hara DM, Xu Y, Liang Z, Reddy MP, Wu DY, Litwin V. Dec 2017) Recommendations for the validation of flow cytometric testing
Pharmaceutical Research – Springer Journals
Published: May 30, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera