- Top of page
- Materials and Methods
- Literature Cited
- Supporting Information
Umbilical cord blood has been used for a wide variety of immunologic investigations including assessments of developmental perturbations by antenatal exposures. Recent advances in multiparameter flow cytometry have allowed finer characterization of lymphocyte phenotype and function, revealing important differences between the fetal and adult immune systems. The degree of variability between human subjects confounds the ability to draw firm conclusions. Artifacts resulting from processing techniques exacerbate this variability. The unpredictable nature of deliveries, especially of premature infants, makes it difficult to control variables such as timing of umbilical cord mononuclear cell (UCMC) isolation and method of collection. Additionally, in multicenter studies dependent on central processing, delays are inevitable. However, little available literature describes systematic testing of the degree to which processing variations affect UCMC phenotype and function. Using multiparameter flow cytometry, we tested the effect of collection technique and length of time prior to UCMC isolation on T cell phenotype and function, with the goal of creating a standardized operating procedure for a multicenter investigation. The study also provides a benchmark data set including extensive surface and functional phenotyping of umbilical cord T cells. UCMC isolation delay of up to 24 h produced similar T cell phenotype and function as tested by in vitro SEB stimulation. There were few statistically significant differences between time points based on data medians. We conclude that, for the purpose of immunologic investigations, a 24-h time delay from sample collection to mononuclear cell isolation does not introduce a significant degree of variation in T cell phenotype and function when adhering to strict standard operating procedures. © 2012 International Society for Advancement of Cytometry
Flow cytometry is a powerful tool that can be used to comprehensively profile T cells. Use of higher order multiparameter fluorescence labeling allows selective gating for specific events in the presence of contaminating populations. As a sensitive tool, however, it is prone to high noise-signal ratio, which can be further confounded by subject-to-subject variation (1). It is critical, therefore, to avoid procedures that may introduce cellular changes with nonspecific staining or partial or complete loss of cell subpopulations during T cell phenotyping and functional analysis.
Umbilical cord blood samples are used for many purposes, including research on T cell development and responses in neonates, and cord blood banking for potential transplantation (2–5). This study focuses on optimizing collection and analysis procedures for developmental investigative purposes. One major challenge in analyzing umbilical cord T cells is the unpredictability of deliveries, especially in studies intending to include premature neonates. For most research studies, it is not feasible to maintain availability of technical assistance required for sample processing at all times. The majority of studies investigating umbilical T cell phenotype and function therefore collect only when assistance is available, or alternatively, limit to scheduled cesarean-section deliveries. To overcome subject-to-subject variability seen in human lymphocyte analysis, large numbers of subjects are required to achieve statistical significance, and scheduled processing may preclude efficient subject enrollment. Samples collected in large, multicenter studies are also subject to inevitable delays introduced by shipping for central processing.
There are few studies that address delays in peripheral blood mononuclear cell (PBMC) isolation as an independent variable affecting T cell phenotype or function. In one published study evaluating time delay effects on lymphocyte populations, the authors found more cellular debris after 48 h with both whole blood lysis and density gradient methods, and loss of T lymphocyte subsets with density gradient separation. The investigators attributed T cell subset changes to red blood cell (RBC) and cellular debris contamination (6). Another of the studies standardizing PBMC collection concluded that time between collection and isolation was the variable that contributed most significantly to diminished T cell function. In that study, time zero (T0) and 24-h (T24) isolations were performed in two separate centers, which in itself may have introduced variability (7). A subsequent study demonstrated that whole blood samples left at room temperature for 24 h allowed granulocyte activation, which suppressed T cell function (8). These studies did not include umbilical cord blood samples, which contain a higher fraction of naïve T cells. Naïve T cells are at baseline more quiescent and have higher threshold for activation. A recent publication proposing a protocol for UCB phenotyping by flow cytometry stated that UCMC should be isolated within 12 h, though a supporting reference was not provided. The antibody panel utilized was also not developed to study T cell subtype or function specifically (9). Finally, in a recent systematic review of established techniques aimed at measurement of T cell function, authors questioned the strength of evidence supporting time delay to processing as a critical variable (10).
The purpose of this pilot study was to optimize and standardize umbilical cord blood collection procedures to minimize variation introduced by technical approach. Specifically, we tested time delay to sample processing as a variable affecting T cell phenotype and function in flow cytometric analysis. Additionally, this study generated an extensive surface phenotype and functional analysis of full term umbilical cord T cells, which can be referenced in future studies investigating neonatal T cells.
- Top of page
- Materials and Methods
- Literature Cited
- Supporting Information
Logistics for immunologic studies using umbilical cord blood T cells are complicated to manage, due to the uncontrolled timing of deliveries. Often with large neonatal studies, multicenter recruitment of subjects is necessary, requiring shipping of unprocessed cord blood samples. Such unpredictability in processing time is less problematic for studies including only full-term, elective cesarean section deliveries. However, in studies requiring samples from a more heterogeneous population of newborns, including prematurely born infants, quantification of variation introduced by time delay to processing is essential. This is particularly important when analyzing rare T cell populations using sensitive cytometric assays. Intra-assay variation and changes in cell markers based on collection or processing techniques may be attributed erroneously to biologic differences between study subjects.
This study performed multiple statistical comparisons of the effect of time of processing on the identification of T cell populations within a specimen incorporating an assessment of assay variability. Replicates had remarkable overlap within each subject (ICC >0.85 for cell type, >0.75 for CD4 subtype, >0.74 for CD4 cytokines and >0.71 for CD8 cytokines), reflecting the reproducibility of the standardized procedures. For most cell subsets and cytokine-positive frequencies, the order of variables contributing to overall variation was subject > time point > replicate. Markers with higher stain indices and event numbers allow for better separation between positive and negative populations. In newborn samples, chemokine receptor and cytokine staining are more difficult to gate accurately and consistently due to lower receptor expression and lower event numbers, which contribute to increased variance and lower sensitivity of statistical analyses for these populations. Nonetheless, we demonstrated stability of populations at 24 h, with only minor degradation (0.2–5% for chemokine receptor median expression, <0.1% for cytokine medians) or increased variance by 48 h.
Though not tested systematically, our team found several techniques that increased cell recovery and reproducibility of the flow results. Collection technique was designed to maximize cell recovery and minimize maternal cell contamination. Our initial, and common, collection technique of dripping blood from cut end of cord into a standard 150 USP heparin “green top” tube resulted in frequent clotting. Venipuncture of a clamped cord segment, rather than the dripping or stripping techniques, reduced this complication. The cord surface was also rubbed with alcohol prior to venipuncture to eliminate coating maternal cells. We achieved minimal coagulation by using venipuncture and increasing the heparin content per tube to 300 USP based on prior recommendations (11).
For flow analysis, RBC contamination remained a concern. Reduced separation of isolated UCMC from RBC was visibly noted after ficoll centrifugation of samples processed after a 48–72 h delay when compared to a 24-h delay. At all three time points, the RBC were not completely removed by using ammonium chloride lysis buffer. These contaminating RBC, that were not well lysed or removed by ficoll gradient separation, were likely nucleated RBC that also tend to survive cryopreservation. While published protocols suggest repeating the ficoll-paque gradient separation step for umbilical cord samples, we found that this resulted in an unacceptable loss in total cell recovery without adequate removal of the nucleated RBC. Unfortunately, the volume of cord blood available from premature deliveries is commonly <5 mL, which greatly limits the number of mononuclear cells that can be obtained. With an ultimate interest in analyzing rare events (such as polarized CD4+ T cells) in prematurely delivered infants, a repeat ficoll step reduces the recovery to an unacceptable degree. Flow analysis using a CD235 RBC marker showed RBC contamination that was stable over time, and present in overlapping regions with CD4 and CD8 positive events. Backgating analysis by flow cytometry showed that without excluding CD235+ populations, CD3+ (CD4+ and CD8+) events were contaminated with nonspecific staining of RBCs. Because some samples approached 50% contamination of UCMCs by RBCs, inclusion of these events in the final analysis will produce markedly erroneous results in population frequencies if not controlled.
Our study found few differences in median frequencies over time in many cell populations identified by surface markers. Surprisingly, precryopreservation viability and cell counts, and post-thaw viability and cell recovery were not affected significantly by time delay to processing. Among CD14+, CD56+, CD3+, CD4+ and CD8+, CD3+ CD56+(hi) showed significant variation between T0 and T48, but not for T24. There were no differences in the proportions of CD3+ T cell subsets, however, suggesting that changes are distributed equally across cell types. Stability of T cell markers through cryopreservation has been addressed in a previously published study on adult PBMC, which showed poor concordance between fresh and frozen CD4+CD62L+CD45RA+, CD4+CD45RO+, CD8+CD28+CD95−, and CD8+CD28+ CD95+ subpopulations. Other authors have shown loss of CD56 expression after cryopreservation, which may be compounded if there is loss of marker prior to cryopreservation (17, 18). One published study examined the effect of cryopreservation on umbilical cord mononuclear cell cytokine secretion (19). Authors found significant blunting of IFN-γ, IL-10, IL-12, and TNF-α to mitogen stimulation, with an unpredictable pattern of cytokine balance (cytokine concentrations neither changed equally nor consistently). Because this study measured only secreted cytokine, we cannot conclude that the changes observed in their study were consistent with our findings that measured negative/positive-cytokine T cell frequency generated from flow cytometry. A separate study comprehensively characterizing the effects of cryopreservation on umbilical cord mononuclear cells is necessary to adequately address this question.
Comprehensive phenotyping on umbilical CD4+ T cells is underrepresented in the literature, although standards have been established for adult peripheral blood samples. While the subsetting of CD4+ T cell using chemokine receptors is a useful tool to aid in understanding broad concepts of T cell differentiation and function, nuances of these populations continue to be redefined. As our knowledge of the various marker combinations expands, the heterogeneity of such populations will be increasingly appreciated. As such, we acknowledge the limitations in assigning eponyms to cell events based on surface markers alone.
Of CD4+ subsets we analyzed, late activated events (CD27−CD28−) appeared to be affected more by time than naïve and early activated (CD27+CD28+). Significant differences in CD4+ subsets were found more frequently at 48 h (8 of 23 subsets identified) with fewer subsets affected at 24 h (4 of 23 subsets). Loss of CD28 expression on CD4+ T cells is associated with cells that have undergone multiple cycles of proliferation and terminally differentiated into IFNγ effectors or into a senescent phenotype (20). Although properties of umbilical cord cells may lend to more rapid cell cycling or down-regulation of CD28 expression, it is unlikely that increases in CD28− frequencies is a result of rapid ex vivo expansion given the short incubation. It is interesting that within the CD28+CD27+ subset, there was a relative loss of chemokine receptor CXCR3 and gain of CCR4+ frequencies at 48 h, with a corresponding gain of both CXCR3 and CCR4 in the CD27−CD28− subset at 48 h. It is possible that there is a fraction of umbilical cord CD4+ T cells that are activated perinatally and differentiate early in vitro.
Time did not have a significant effect on SEB-stimulated intracellular cytokine-positive frequencies for CD4 or CD8+ T cells, except for IL-2, which was affected in opposite directions for the two cell types (decreased from 2.02 to 1.43% CD4 and increased from 0.05 to 0.09% in CD8), and CD4+ MIP-1β (increased from 0.10% at T0 to 0.21% at T48). Umbilical cord cells are known to produce high amounts of IL-2, in both CD4+ and CD8+ T cell subsets (21), and constitutive production in CD4+ T cells may decrease over time as culture substrate is depleted. Brief antigen stimulation also results in increased IL-2 production in CD8+ T cells without concurrent increase in cytotoxic activity (22). It is possible that introduction of antigen through delivery, sample collection or processing could provide enough stimulus to increase background IL-2 production of CD8+ T cells in a the given period of time; this effect may be more evident in umbilical cord samples which produce high amounts of IL-2 at baseline.
While we have demonstrated small changes with delay in processing of UBMC, it is important to recall that our study included blood only from full term scheduled cesarean-section deliveries to control the variable of time delay to processing. It is possible that inflammatory signals present in cord blood collected from premature and/or infected, deliveries after labor or maternal preeclampsia could change the phenotype and function of T cells, changes that may be amplified by delay in processing. In fact, there are studies that support changes in T cells associated with the presence of superantigen (23, 24), through toll-like receptors and independent of TCR engagement (25) or physiologic stressors (26, 27). Stability of UCMC at 24 h after blood sample collection may be altered in infants whose peripheral T cells were exposed to inflammatory signals in utero and continued to be stimulated while at room temperature prior to processing. The slight decrease in naïve/early activated and increase in late activated CD4+ T cells demonstrated in our study may be more exaggerated following stressful delivery or delivery through infected membranes.
Notably, exclusion of dead cells (live/dead stain positive) did increase the specificity of staining and therefore reproducibility of results with cryopreserved specimens. This study was not designed to test the effects of cryopreservation on the various cell markers, but as all cells were frozen prior to analysis for similar periods of time, any effect is expected to be applied uniformly across time points. Our study did not test T cell functional changes that might affect cord blood cell engraftment, which would require much more comprehensive functional and phenotypic analysis and was not the purpose of the study.
This study does not evaluate the biologic significance of the small time-dependent changes measured. It is not known, for example, whether there is a loss/gain of cell type or if changes across time represent down/upregulation of surface proteins. Surface receptor changes are more likely than differences in cell populations surviving over time, given the absence of differences observed in viability and recovery. When using a combined statistical approach of pair-wise comparisons and measurement of interclass variation, a few cell types are affected at 48 h, and fewer are significantly changed after a 24-h time delay. On the basis of these results, we conclude that our standardized operating procedure for collection and processing of umbilical cord mononuclear cells can generate highly reproducible results, even after a time delay of up to 24 h prior to cell isolation. Of interest, the apparent alterations seen at 24 h, for example in IL-2, suggests that there are caveats associated with delayed processing and that it may indeed by important to optimize processing times to less than a day. For studies in which delay to processing of umbilical cord samples is inevitable however, our data quantifying expected variation and range of results can be applied to appropriately power or correct for changes that may occur secondary to time delay to processing. Our methods can be used by multicenter studies to reliably analyze T cell phenotype and function in umbilical cord blood.