Establishing the flow cytometric assessment of myeloid cells in kidney ischemia/reperfusion injury



Polychromatic flow cytometry is a powerful tool for assessing populations of cells in the kidney through times of homeostasis, disease and tissue remodeling. In particular, macrophages have been identified as having central roles in these three settings. However, because of the plasticity of myeloid cells it has been difficult to define a specific immunophenotype for these cells in the kidney. This study developed a gating strategy for identifying and assessing monocyte and macrophage subpopulations, along with neutrophils and epithelial cells in the healthy kidney and following ischemia/reperfusion (IR) injury in mice, using antibodies against CD45, CD11b, CD11c, Ly6C, Ly6G, F4/80, CSF-1R (CD115), MHC class II, mannose receptor (MR or CD206), an alternatively activated macrophage marker, and the epithelial cell adhesion marker (EpCAM or CD326). Backgating analysis and assessment of autofluorescence was used to extend the knowledge of various cell types and the changes that occur in the kidney at various time-points post-IR injury. In addition, the impact of enzymatic digestion of kidneys on cell surface markers and cell viability was assessed. Comparisons of kidney myeloid populations were also made with those in the spleen. These results provide a useful reference for future analyses of therapies aimed at modulating inflammation and enhancing endogenous remodeling following kidney injury. © 2013 International Society for Advancement of Cytometry


A common feature of the progression of immune and nonimmune kidney disease of diverse aetiology is the infiltration of inflammatory macrophages [1]. Macrophage numbers have shown to correlate with disease progression, making them a useful tool in predicting disease outcome [1-3]. More recently, macrophage heterogeneity has been shown to correspond to the diverse roles that these cells play in both the initiation of tissue fibrosis and the positive role in wound healing and tissue remodeling [4, 5]. Monocytes recruited in response to inflammatory cues can undergo differentiation into two broad macrophage subsets based on phenotype, function, and polarization state. The classically activated or M1 macrophage is the pro-inflammatory cell type closely associated with the innate immune response, whereas the alternatively activated or M2 macrophage possesses a range of anti-inflammatory and wound healing capabilities [6-8]. In part, achieving wound repair and tissue remodeling requires an appropriate balance between the M1 and M2 polarization states.

Traditionally, studies investigating the number of infiltrating macrophages in damaged kidneys have relied on immunohistochemistry (IHC) and immunofluorescence (IF) techniques to assess kidney histopathology, cell morphology, and receptor expression. However, flow cytometry is becoming an increasingly important tool, particularly because of the ability to evaluate a panel of cell surface and intracellular markers on individual cells at a rate of over 10,000 cells/second. Eight-color polychromatic flow cytometry in conjunction with two nonfluorescent parameters, forward and side light scattering, is now common and with the latest flow cytometers measuring up to 20 parameters, the information obtainable from each experiment is destined to grow, and with it the need for more rigorous methods of data analysis [9]. However, even with improving technology, there remain a number of key challenges related to the preparation of kidney samples for flow cytometry, the selection of appropriate target markers and the informative analysis of the resulting data, which need to be addressed.

The aim of this study was to assess the impact of enzymes (used to produce a kidney single cell suspension) and ischemia/reperfusion (IR) injury on cell yield, viability, surface marker expression, and autofluorescence. Gating strategies were created that best characterize various myeloid cell types, especially where particular receptors were expressed at low levels. The panel of monocyte-, macrophage-, dendritic cell (DC)- and granulocyte-associated markers used included CD11b, CD11c, Ly6C, Ly6G, major histocompatibility complex class II (MHCII), colony stimulating factor-1 receptor (CSF-1R or CD115), mannose receptor (MR or CD206), and F4/80. Particular emphasis of the study was on the assessment of kidney myeloid cell analysis in the inflammatory phase of IR injury, which is characterized by widespread epithelial cell death, an influx of pro-inflammatory cells and heightened inflammatory cytokine production.

In addition, the apoptotic and necrotic epithelial cells of the damaged kidney tubular epithelium, related to the reduced glomerular filtration that follows injury, leads to the accumulation of tubular casts [10]. This hallmark of acute kidney injury results in autofluorescence and nonspecific background signals, which leads to difficulties in interpretation of flow cytometric data that is unique to the kidney. Unless addressed, this can lead to erroneous analysis. The intrinsic autofluorescent properties of kidney cells also apply to macrophages because of their propensity to phagocytose cellular debris.

Finally, backgating analysis was used to define and extend the knowledge of myeloid subpopulations in terms of their co-expression of multiple markers and for their spatial location on parent dot plots. This study clarifies and addresses the anomalies encountered when assessing myeloid cells in the kidney, as compared to the more commonly assessed primary and secondary lymphoid organs, while forming a comparative base for which various therapies aimed at manipulating cell numbers and function can be referenced.

Materials and Methods

Animals and Surgery

Male 6–8 week old (20–25 g) C57BL/6J mice obtained from Monash Animal Services (Melbourne, Australia) were used. All studies were approved by the Monash University Animal Ethics Committee and were performed in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes. Mice were anesthetized with 2.0% inhaled isofluorane (Abbott Australasia, Sydney, Australia) before the left renal pedicle was occluded using a vascular clamp (0.4–1.0 mm; Fine Science Tools, Heidelberg, Germany) for 40 min via a flank incision to induce unilateral IR injury (n = 5 mice/group/time-point). Following removal of the clamp, reperfusion was visually confirmed prior to wound closure using silk suture (size 5-0, Ethicon, NJ, USA). An additional group of mice served as a sham-operated control group where the animals were anaesthetized and a flank incision was performed without renal pedicle clamping.

Digestion and Preparation of the Kidney and Spleen for Flow Cytometry

Mice were culled using a CO2 cull chamber at 6 hours, 1 day or 7 days after IR injury. The spleen and left kidney were removed and placed in cold FACS buffer (PBS supplemented with 0.2% BSA, 0.02% NaN3 and 5 mM EDTA).

Spleens were cleaned of any connective tissue and mechanically digested in cold FACS buffer to produce a single cell suspension. Mechanical digestion (MD) was achieved by making small incisions in the side of the spleen before gently pressing the organ between two frosted glass slides.

Kidneys were decapsulated and finely chopped with surgical scissors before enzymatic digestion (ED) in 1 mL of dissociation media consisting of HBSS (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 3 mg/mL collagenase/dispase (Roche Applied Science, Penzberg, Germany), 0.2 mg/mL DNase type 1 (Roche Applied Science), 50 μM CaCl2, preheated to 37°C. The samples were mixed on a rotary tube suspension mixer (20 RPM; Ratek Instruments, Melbourne, Australia) at 37°C for 20 min and then mechanically digested using a 1000 μL pipette tip. The samples were mixed for two further 5 min periods (20 RPM) with mechanical dissociation in between. After 30 min, mechanical dissociation with an 18-gauge needle resulted in a single cell suspension. Nine mL of cold FACS buffer was added in order to inhibit enzyme activity.

ED for kidneys and MD for spleens were used for all aspects of this study except for the comparison between ED and MD (Section “Using Enzymes to Aid in the Digestion of the Kidney is more Suitable than MD Alone” and Fig. 1) where both ED and MD were performed on each of the organs.

Figure 1.

Using enzymes to aid in the digestion of the kidney is more suitable than mechanical digestion alone. To compare the effects of two different organ digestion methods, spleens and kidneys from mice at 24 hours post-IR injury were subjected to either mechanical digestion (MD) or enzymatic digestion (ED). For both organs, ED yielded a higher cell count (a). ED also resulted in a greater proportion of viable cells as assessed using propidium iodide (b). The gating hierarchy used to assess viable cells, CD45+ cells, Ly6G+ granulocytes, and subpopulations of CD11b+Ly6G- cells is shown (c). There was no difference in the proportion of CD45+ cells in the spleen between ED and MD. However, ED yielded a greater proportion of CD45+ cells in the kidney compared to MD (d). The digestion method had no impact on the proportion of CD11b+Ly6G+ granulocytes in the kidney but significantly reduced the proportion of granulocytes in the spleen (e). In the spleen ED resulted in a greater proportion of CD11b+CD11chigh and CD11b+CD11clow cells compared to MD (f). However, ED increased CD11b expression and resulted in less well-defined CD11c populations (f). The CD11b+CD11chigh group was largely absent in the kidney, while ED greatly increased the proportion of the CD11b+CD11clow population (f). There was no significant difference in the proportions between the CD11b+CD11c- populations in either organ with regards to the digestion method (f). The proportion of F4/80+ cells was significantly greater in the kidney following ED compared to MD (g). No difference was observed in this population in the spleen between MD and ED (g). The MFI of the CD11b+F4/80low/- population in the kidney (depicted graphically) was significantly increased following ED compared to MD (g). CSF-1R expression was dramatically reduced in the spleen following ED compared to MD for both Ly6Chigh and Ly6C- populations (h). Numbers on dot plots represent proportions of parent populations. Statistical analysis was performed using a Student's t-test (unpaired, two-tailed, with Welch's correction); **P < 0.01, ***P < 0.001, ****P < 0.0001. Data are displayed as means ± SEM (n = 5/group).

All single cell suspensions were incubated for 1 min with 1 mL of red blood cell lysis buffer (8.3 g/L Na4Cl, 10 mM Tris-HCl, pH7.5) to remove red blood cells. All samples were filtered with a 40 μm nylon cell strainer (BD Bioscience, San Jose, USA) prior to antibody labeling.

Cell Counts and Viability

For flow cytometry cell preparation, cell counts and viability determination were performed using a Z2 Coulter Counter (Beckman Coulter, USA). In addition, for the ED versus MD study, propidium iodide (PI) was also used to determine cell viability.

Antibody Labeling

Three million cells from kidney or spleen single cell suspensions were incubated for 20 min at 4°C in the dark with the following fluorochrome-conjugated antimouse antibodies: anti-CD45 APC-Cy7 (clone 30-F11; Biolegend, San Diego, USA) and PE-Cy5 (clone 30-F11; BD Biosciences), anti-CD11b PE-Cy7 (clone M1/70; BD Biosciences), anti-CD11c Pacific Blue (clone N418; Biolegend), anti-I-A/I-E (MHCII) PE-Cy5 (clone M5/114.15.2; Biolegend), anti-CSF-1R (CD115) PE (clone AFS98; eBioscience, San Diego, USA), anti-F4/80 APC (clone BM8; eBioscience), anti-Ly6G Alexa Fluor 647 (clone 1A8; Biolegend), anti-Ly6C FITC (clone HK1.4; Biolegend), anti-CD206 (mannose receptor) Alexa Fluor 488 (clone C068C2; Biolegend), and anti-EpCAM (CD326) PE-Cy7 (clone G8.8; Biolegend). Fc receptor block (anti-CD16/32 antibody) was added to all antibody cocktails. Intracellular MR labeling involved the use of a CytoFix/CytoPerm kit (BD Biosciences). Following surface receptor labeling, cells were permeabilized and incubated with antibody for 30 min at 4°C in the dark before being washed twice in 1× Perm/Wash buffer (BD Biosciences) and resuspend in FACS buffer. Isotype matched controls were used for each antibody in a fluorescence minus one (FMO) manner.

Flow Cytometric Acquisition and Analysis

Data were acquired on a BD FACS Canto II flow cytometer (BD Biosciences) equipped with 405, 488, and 633 nm excitation lasers in conjunction with FACS Diva acquisition software (BD Biosciences). Compensation was performed with single color controls for each organ using the same conjugated antibodies used in the study. Data analysis was performed using FlowLogic FCS analysis software (Inivai Technologies, Melbourne, Australia).

Statistical Analysis

Statistical analysis was performed using GraphPad Prism software version 6.0c (GraphPad Software, San Diego, USA). A Student's t-test (unpaired, two-tailed, with Welch's correction) was used to analyze data between two groups. A one-way analysis of variance with a Tukey's multiple comparisons test was used to analyze data contained in three groups. Data are displayed as means ± SEM. P < 0.05 was considered statistically significant.


Using Enzymes to Aid in the Digestion of the Kidney is more Suitable than MD Alone

Using enzymes (collagenase/dispase, DNase type I) to aid in the digestion of kidney tissue risks cleaving particular cell surface receptors. In addition, optimal primary and secondary lymphoid organ cell preparations are often achieved with MD alone. It was therefore necessary to test whether ED is indeed required for kidney dissociation. Ten mice received 40 min unilateral IR injury and 24 hours later the spleen and injured kidney were removed. One group of 5 mice had both organs digested with the aid of enzymes, whereas the remaining mice had their organs digested purely by mechanical means. Once digested, cells from each organ were labeled with antibodies against CD45, CD11b, CD11c, Ly6C, Ly6G, MHCII, F4/80, and CSF-1R, and assessed using flow cytometry. Our data show that in both the spleen and the kidney, ED yielded a higher viable cell count compared to MD (spleen MD: 5.54 × 107, spleen ED: 1.49 × 108, P < 0.0001; kidney MD: 2.56 × 107, kidney ED: 4.40 × 107, P = 0.0025) (Fig. 1a). Furthermore, propidium iodide staining revealed that ED yielded greater viability for both spleen and kidney cells (spleen MD: 74.2%, spleen ED: 86.1%, P < 0.0001; kidney MD: 67.7%, kidney ED: 77.3%, P = 0.0153) (Fig. 1b). In assessing hematopoietic cells (as per the gating hierarchy described in Fig. 1c), we found no difference in the proportion of CD45+ leukocytes in the spleen with the different digestion methods (MD: 98.8%, ED: 98.3%). However, ED of the kidney resulted in a significantly greater proportion of CD45+ cells compared to MD (MD: 6.1%, ED: 13.9%, P = 0.0003) (Fig. 1d). Within the CD45+ cell pool in the kidney, the digestion method caused no difference in the proportion of Ly6G+ granulocytes (MD: 28.6%, ED: 26.0%). However, in the spleen, ED significantly reduced the proportion of this cell type (MD: 4.8%, ED: 1.8%, P = 0.0241) (Fig. 1e). In both organs, ED significantly increased the expression (mean fluorescence intensity) of both Ly6G and CD11b on this population, as seen in the dot plots (Fig. 1e) (data not shown).

After excluding the granulocytes, three populations of CD11b+ cells were assessed in conjunction with CD11c expression (Fig. 1f). In the spleen, ED resulted in a greater proportion of CD11b+CD11chigh DCs (MD: 0.9%, ED: 1.4%, P = 0.0034), although the populations were less well defined compared to those acquired following MD (Fig. 1f). There were very few cells that shared this phenotype in the kidney, regardless of the digestion method.

The proportion of a second population, which expressed low levels of CD11c, was statistically higher following ED in both the kidney (MD: 4.2%, ED: 13.5%, P = 0.0003) and spleen (MD: 1.6%, ED: 2.1%, P = 0.0084) (Fig. 1f).

There was no significant difference in the proportion of CD11b+CD11c-(Ly6G-) cells between the two groups in either the kidney (MD: 50.7%, ED: 56.4%) or spleen (MD: 4.0%, ED: 3.2%) (Fig. 1f).

F4/80 expression was assessed on the same CD11b+Ly6G- population with a notable difference identified between the two digestion methods in the kidney. With MD, the F4/80+ cells were barely detectable but made up over 9% of CD11b+Ly6G- cells following ED (MD: 2.8%, ED: 9.1%, P<0.0001) (Fig. 1g). In the spleen there was no difference in the proportion of F4/80+ cells (MD: 1.4%, ED: 1.3%) although the population appeared more dispersed following ED (Fig. 1g). In addition, the two digestion methods resulted in substantial differences in the CD11b+F4/80low/- populations in the kidney. Once gated, the MFI for the F4/80-APC parameter was assessed and shown to be significantly greater following ED (MD: 505 MFI, ED: 1070 MFI, P < 0.0001) (Fig. 1g).

In the spleen, ED reduced the expression of two CSF-1R+ populations: a Ly6ChighCSF-1R+ (MD: 1.38%, ED: 0.03%, P = 0.0073) and a Ly6C-CSF-1R+ population (MD: 1.0%, ED: 0.2%, P < 0.0001) (Fig. 1h). Very few CSF-1R+ cells were detected in either group in the kidney (data not shown). It must be noted that a change in the proportion of one population can affect the proportion of other populations. However, ED does appear important for assessing F4/80 expression in the kidney, while dramatically reducing surface CSF-1R expression, as demonstrated in the spleen. With this knowledge, a suitable gating strategy was created to clearly identify subpopulations of CD11b+ cells in the kidney, both in the steady state and in the inflammatory phase following IR injury.

Gating Strategy for Myeloid Cells in the Kidney

With up to eight-color flow cytometry commonly employed to assess cell phenotypes, there are inevitably many different theoretical subsets that can be defined in any experiment. Here we describe a gating procedure designed to clearly identify important myeloid cell populations in the kidney, accounting for the high potential for autofluorescence, particularly following injury. Figure 2a outlines the population hierarchy used to distinguish between CD11b+Ly6G+ granulocytes and CD11b+Ly6G- nongranulocytes. Initially, a polygon gate was created on the FSC-A vs. FSC-H plot to select the ‘Single’ cells that passed by the lasers individually (Fig. 2b). CD45+ cells from the resulting daughter population were subsequently viewed against FSC-A. These cells represent a viable CD45+ population as compared with a similar population identified using propidium iodide to exclude dead cells (data not shown). These CD45+ cells were colored red and viewed on a FSC-A vs. SCA-A plot. The coloring of this population enabled a ‘Live’ gate to be drawn on the FSC-A vs. SCA-A plot, to select viable hematopoietic cells and exclude debris (Fig. 2b). This is otherwise difficult to achieve when assessing cells in the kidney as compared to those from lymphoid organs because of the low proportion of CD45+ cells. This same technique can also be employed to aid in the creation of the initial ‘Single’ cells gate. A population of CD45+CD11b+ cells (encompassing resident and infiltrating myeloid cells) was selected from the ‘Live’ cell pool (Fig. 2b). The plots in Figure 2b represent the cells in the kidney 6 hours post-IR surgery, which is characterized by an influx of CD45+ cells. Granulocytes were identified in the resulting daughter population based on the positive expression of Ly6G (also Ly6Clow) (Fig. 2c) with their proportion being significantly higher at 6 hrs post-IR injury (sham-IR: 23.0%, IR: 28.3%, P = 0.0222). An inverse gate effectively excluded the granulocytes for further analysis of myeloid cell subpopulations. Examples from sham-IR and IR kidneys at 6 hours post-surgery are shown (Fig. 2c).

Figure 2.

Gating strategy for assessing myeloid cells in the kidney. The population hierarchy shows the CD11b+ gating strategy (a). ‘Single’ cells (excluding doublets and triplets) were selected with a polygon gate on a FSC-A vs. FSC-H dot plot (b). CD45+ cells were gated on the resulting daughter population on a FSC-A vs. CD45 dot plot. These cells were colored (red) and viewed on a FSC-A vs. SCA-A dot plot (b). A ‘Live’ cell gate (which excludes debris) was created with the aid of the colored CD45+ cells (b). CD45+CD11b+ cells were selected with a polygon gate (b). Granulocytes were selected by gating on Ly6ClowLy6G+ cells (c). An inverse gate to select CD11b+Ly6C+/-Ly6G- cells (pink) was used to gate out granulocytes (black) for further analysis of myeloid cell subsets (c). Plots in b are from a kidney taken 6 hours post-IR injury. Plots in c are from kidneys taken 6 hours post-IR surgery from IR and sham-IR animals. Numbers on dot plots represent proportions of parent populations.

Gating Strategy for Myeloid Cell Subpopulations in the Kidney

The gating strategy used to interrogate CD11b+Ly6G- subsets shown in Figure 3a extends from the gating procedure described in Section “Gating Strategy for Myeloid Cells in the Kidney. Expression of the antigen-presenting molecule MHCII was compared to other markers to identify subpopulations of monocytes and macrophages. An intracellular antibody against MR was used to identify M2 macrophages (Fig. 3b). A quadrant gate was used to identify two MR+ populations based on a combination of MR and MHCII expression. While most mature M2 macrophages co-express MHCII (16.9% of CD45+CD11b+Ly6G- cells at 24 hrs post-IR injury), there was a population of MHCII- cells in which MR was detected (8.5%). The example shown is from a kidney assessed 24 hours following IR injury, prior to the recognized tissue remodeling phase, where M2 macrophages are the predominant macrophage population [11].

Figure 3.

Gating strategy for CD11b+ cell subpopulations in the kidney. The gating hierarchy (continued from Figure 2) shows the procedure used to assess CD11b+ cells following the exclusion of Ly6G+ granulocytes (a). M2 macrophages, defined as being MR+, were assessed in conjunction with the expression of MHCII (b). Subsequent monocyte/macrophage subsets were defined based on the cellular expression of MHCII, Ly6C, F4/80 and CD11c (c–e). Ly6C was used to distinguish monocytes at various maturation stages. Ly6Chigh cells (MHCII-) are immature monocytes. The marker is down regulated as the cells mature. A prominent Ly6Chigh (MHCII-) population is present at 6hrs post-IR injury (green) (c), along with a smaller Ly6ClowMHCII- population (c). A maturing or transitioning population of MHCIIlowLy6C+ cells exist, particularly following IR-injury (c). A prominent Ly6C-MHCIIhigh population exists in kidneys following both sham-IR and IR-surgery (c). MHCII can also be used to distinguish between three F4/80+ populations. A prominent F4/80+MHCIIhigh population was identified (gated cells) (d). The dot plot overlay shows this population (pink) compared with an isotype control antibody (light blue) (d). The overlay also helped identify populations of F4/80+MHCIIlow and F4/80+MHCII- cells. The latter corresponds to the Ly6Chigh population (green) (d). Low levels of CD11c expression can make it difficult to distinctly categorize CD11c+ cells in the kidney, as opposed to its expression when examined in lymphoid organs or the blood. Here the CD11c labeled cells (orange) were overlayed with an isotype control (light blue) (e). In addition, the MFI of the CD11c-Pacific Blue antibody was assessed for the MHCIIhigh population. These data are displayed graphically with the MFI for the isotype controls indicated using a broken line (e). Appropriate isotype controls (iso) are displayed. Numbers on dot plots represent proportions of parent populations. Statistical analysis was performed using a Student's t-test (unpaired, two-tailed, with Welch's correction); *P < 0.05. Data are displayed as means ± SEM (n = 5/group).

CD11b+Ly6G- cells were also examined for their expression of the monocyte-associated marker Ly6C (Fig. 3c), the historical mature macrophage marker F4/80 (Fig. 3d) and the DC-associated marker CD11c (Fig. 3e). These markers were all compared to the expression of MHCII. Both a Ly6Chigh (sham-IR: 3.8%, IR: 27.9%, P = 0.0004) and a Ly6Clow (sham-IR: 1.4%, IR: 4.5%, P=0.0257) population not expressing MHCII were identified with a much greater proportion in the IR injured kidney. Ly6C is a marker of monocyte immaturity and expression is lost as monocytes transition into macrophages. A general population of Ly6C+ cells expressing MHCII was seen (sham-IR: 3.3%, IR: 11.0%, P = 0.0066), indicating a population of maturing monocytes, which appear to down-regulate their expression of Ly6C and up-regulate MHCII concomitantly (Fig. 3c). F4/80 has historically been regarded as a mature macrophage marker [12]. However, more recent reports have shown that it is not expressed on all macrophage populations and has been identified on some Ly6C+ monocytes along with a range of other myeloid cells, revoking its status as a sole identifier of macrophages [13-16]. When viewed against MHCII, three F4/80+ populations were identified (Fig. 3d). The classical F4/80+MHCIIhigh mature macrophage was prominent in both sham-IR and IR groups (gated population) (sham-IR: 59.0%, IR: 30.5%, P = 0.0001). When viewed as an overlay containing F4/80 stained cells and an isotype control antibody, an F4/80+MHCIIlow and an F4/80+MHCII- population were made evident, particularly following IR-injury (Fig. 3d). The latter population also corresponded with the Ly6Chigh monocyte population when these cells were gated on a MHCII vs. Ly6C plot and colored (green) (Fig. 3d).

There is much discussion surrounding the similarities and differences between macrophages and DCs. In this model, a clearly defined CD11b+CD11chigh population, generally recognized as DCs, was not observed in the kidney (Fig. 3e). There were cells that expressed a low level of CD11c but this population differs from the distinct CD11chigh DCs seen in other organs, such as the spleen. For this reason, CD11c expression was viewed on the CD45+CD11b+ population, rather than as an initial differentiating marker for macrophages and DCs. To further investigate the changes to these cells following IR injury, the entire MHCIIhigh population was gated and the change in the MFI for the anti-CD11c antibody analyzed. As seen in the overlay plots for both the sham-IR and the IR groups at 6 hours postsurgery, antibody labeling exists at levels above the isotype control. There is also a significant increase in the MFI of this parameter following IR injury (sham-IR: 570 MFI, IR: 770 MFI, P = 0.031) (Fig. 3e).

Kidney and Spleen Ly6C+, Ly6G+, and MHCII+ Cell Population Comparison

Backgating analysis was used to further characterize various myeloid subpopulations in the kidney. Comparisons were also made between these cells and their counterparts in the spleen. Figure 4a shows Ly6G+(Ly6Clow) granulocytes (dark blue). These cells are also displayed on the grandparent FSC-A vs. SSC-A plot (Figure 4b). Granulocytes in the spleen appear similar to those in the sham-IR and IR kidneys 6 hours postsurgery. However, they compose a greater proportion of the CD45+CD11b+ pool (spleen: 72.5%, sham-IR kidney: 23.0%, IR kidney: 28.3%).

Figure 4.

Kidney and spleen Ly6C+, Ly6G+ and MHCII+ cell population comparison. Backgating analysis of flow cytometry data was used to compare the relative positioning of Ly6G+ granulocytes (dark blue), MHCII+Ly6C- cells (red), MHCII-Ly6Chigh cells (green), MHCII-Ly6Clow cells (purple) and maturing or transitioning monocytes (MHCIIlowLy6C+) (light blue) (a). Backgating analysis of these populations shows their profiles on FSC-A vs. SSC-A dot plots (b). Examples from spleen and kidneys at 6 hours post-sham-IR and IR surgery. Numbers on dot plots represent proportions of parent populations.

In a similar fashion, MHCII+Ly6C- cells (red) (Fig. 4a) were backgated and overlayed onto the same FSC-A vs. SSC-A plots (Fig. 4b) (spleen: 34.2%, sham-IR kidney: 89.0%, IR kidney: 52.2% of the CD45+CD11b+Ly6G- pool). A far greater proportion and number of Ly6Chigh cells (green) were present in the IR kidney compared to the sham-IR kidneys (spleen: 31.2%, sham-IR kidney: 3.8%, IR kidney: 27.9% of the CD45+CD11b+Ly6G- pool) (Fig. 4a). There were distinctly fewer MHCII-Ly6Clow cells (purple) compared to the Ly6Chigh population (spleen: 9.2%, sham-IR kidney: 1.4%, IR kidney: 4.5% of the CD45+CD11b+Ly6G- pool) (Fig. 4a). The maturing or transitioning monocytes (MHCIIlowLy6C+, light blue) are also most prevalent in the IR compared to the sham-IR kidneys (spleen: 12.4%, sham-IR kidney: 3.3%, IR kidney: 11.0% of the CD45+CD11b+Ly6G- pool) (Fig. 4a). All of the Ly6C expressing cells from both organs present in a similar fashion on the FSC-A vs. SCA-A plots, as do the MHCII+ populations. The granulocyte population in the spleen appears to be composed of cells with a greater range of size and granularity compared to that in the kidney (Fig. 4b).

Assessing Epithelial Cells and Autofluorescence in the Post-Ischemic Kidney

Epithelial proliferation leading to regeneration and repair is central to processes of healing following various forms of kidney disease, including IR injury [17]. As such, the pan epithelial marker EpCAM was used to assess the impact of IR injury on epithelial cell populations. To assess EpCAM+ cells, ‘Single’ cells were gated, followed by ‘Live’ cells (to exclude debris), as depicted in the population hierarchy (Fig. 5a). EpCAM expression was then compared to CD45 expression, with a gate placed around the CD45-EpCAM+ population (Fig. 5b). The proportion of EpCAM+ cells had already significantly decreased at 6 hours post-IR injury (sham-IR: 16.3%, IR: 9.6%, P=0.0001) and fell further as seen at day 7 postinjury (sham-IR: 14.3%, IR: 5.9%, P=0.0001). Autofluorescence can pose a problem, as is evident when the kidneys taken 6 hours post-IR are displayed alongside those taken 7 days postinjury, where prominent autofluorescence is visible in the IR anti-EpCAM antibody and isotype control groups (Fig. 5b). The autofluorescence was not present in the IR group 6 hours postinjury. For the day 7 time-point, a modified EpCAM+ gate was created in order to exclude the autofluorescence from the EpCAM+ population. This method can also be employed for clearly distinguishing CD45+ cells from the rest of the kidney cells. Backgating analysis of the EpCAM+, autofluorescent and CD45+ cells was performed to view their location on the parent FSC-A vs. SSC-A dot plot (Fig. 5c). The difference between the different cell types is clear, with the CD45+ cells forming a tighter group further along the forward scatter axis compared to the EpCAM+ cells and autofluorescent events.

Figure 5.

Assessing epithelial cells and autofluorescence in the post-ischemic kidney. The population hierarchy resulting from the EpCAM+ epithelial gating analysis is shown (a). Following the gating of ‘Single’ cells (FSC-A vs. FSC-H) and ‘Live’ cells (FSC-A vs. SSC-A) (data not shown), EpCAM+ epithelial cells were selected for their expression of EpCAM and for a lack of expression of the hematopoietic marker CD45 (b). With the progression of time in the IR model, autofluorescence becomes increasingly prominent. In this example, at 7 days post-IR, the EpCAM+ gate was altered so as not to include autofluorescent cells (b). Backgating analysis shows the difference in the FSC-A vs. SSC-A profile of CD45-EpCAM+, autofluorescent and CD45+ populations (c). An autofluorescent population appeared when examining the CD45+CD11b+ cell pool in the kidney following IR injury (d). On the MHCII vs. Ly6C dot plots, autofluorescence became more prominent with time after injury (d). This autofluorescent population was backgated and displayed in pink on the parent CD11b vs. CD45 plot (d). The increase in the proportion of this autofluorescent population with time (after injury) is shown graphically (e). Numbers on dot plots represent proportions of parent populations. Statistical analysis was performed using a one-way analysis of variance with a Tukey's multiple comparisons test; **P < 0.01, ***P < 0.001, ****P < 0.0001. Data are displayed as means ± SEM (n = 5/group).

Autofluorescence increases progressively with time after IR injury. Figure 5d shows autofluorescent cells, after gating on CD45+CD11b+ cells, on a Ly6C vs. MHCII plot from a sham-IR kidney along with injured kidneys at 24 hours and 7 days post-IR. The autofluorescent populations were backgated and shown in pink on the CD11b vs. CD45 parent plot. At 7 days post-IR the autofluorescence is very difficult to distinguish from nonautofluoroescent CD11b+ cells. Empty channels may be useful for gating out autofluorescence that is associated with IR-induced damage. The increase in the autofluorescence increased almost threefold between 24 hours and 7 days post-IR injury (sham-IR: 1.3%, 24 hrs post-IR: 7.8%, 7 days post-IR: 21.2%) (Fig. 5e).


Identifying and characterizing macrophage functional/polarization states is necessary to understand processes of disease progression and healing. Here, we have described a polychromatic flow cytometry analysis strategy, taking into account light scattering and autofluorescent characteristics, to assess infiltrating and resident cells in the uninjured kidney and in the inflammatory phase following IR injury. Performing backgating analysis along with coloring populations and viewing them against multiple parameters will lead to more detailed phenotypic and functional descriptions. This includes information regarding the maturation state of the cell, its autofluorescent properties and functional capacity, which can be linked to other data, such as cytokine production and enzyme activity. This is particularly relevant to tissue macrophages because of their heterogeneity, especially in the disease setting where they play central roles in inflammation and tissue remodeling.

One marker that we focused on was Ly6C, as its expression can be used to define monocyte maturation and function, with Ly6Chigh pro-inflammatory cells down regulating the marker as they mature into Ly6C- macrophages [18]. In addition, the activation of monocytes at various maturation stages leads to mature macrophages of distinct functional states [18]. Following unilateral ureteral obstruction, Ly6Chigh cells have been shown to home to kidneys where they differentiate into monocytes/macrophages of distinct functional states, indeed identified by the level of Ly6C expression [19]. Our data showed that the initial inflammatory phase of the IR model involves a dramatic increase in the proportion and number of Ly6Chigh monocytes. As such, assessing changes in this population with various treatments or in fact targeting this cell type directly may impact the degree of injury or provide increased potential for regeneration. A number of studies have used antibodies against Gr-1, a complex formed by both Ly6C and Ly6G, to separate monocytes from granulocytes [20]. Confirmed here in the kidney, using separate antibodies against Ly6C and Ly6G allows for an easier delineation of monocytes and granulocytes, and where applicable allows for further separation of the granulocyte pool into neutrophils and eosinophils [21]. Monocyte populations have also been defined by their expression of the chemokine receptors, CX3CR1 and CCR2 [22, 23]. CD11b+CCR2lowGR1-Ly6C-CX3CR1high monocytes migrate to normal tissues, whereas inflammatory monocytes with a CD11b+CCR2highGR1intLy6ChighCX3CR1low phenotype home to injured tissues [24].

We also chose to assess MR expression as it is a useful identifier of M2 macrophages [4, 25]. Indeed, mannose receptor 2 has been shown to be upregulated on macrophages following unilateral ureteral obstruction and is believed to play a role in modulating fibrosis through binding and internalizing collagen via an extracellular fibronectin type II domain [26, 27]. Interestingly, this study showed that two populations of MR+ cells (MHCII- and MHCII+) exist in the kidney at 24 hours post-IR injury. Again, targeting or manipulating this cell type may help promote kidney remodeling and regeneration. When considering assessing MR expression with flow cytometry, it should be noted that MR is expressed weakly on the cell surface [28]. Membrane permeabilization may result in more effective labeling, although this does not allow for isolation of a potential viable M2 population via FACS.

Autofluorescence is another characteristic of kidney IR injury that needs to be considered carefully. Myeloid cells, particularly those expressing CD11b, CD11c and high levels of F4/80, exhibit autofluorescence at a range of excitation and emission wavelengths [14]. Certain myeloid populations can even be defined based on their autofluorescence signature. However, if a full panel of fluorochromes is being used then there is a risk of erroneous emission signals. Using an FMO approach for antibody controls is useful for identifying and minimizing the effects of autofluorescence [29]. This study showed that autofluorescence increases over time in kidney IR injury and can be potentially problematic when assessing both hematopoietic and nonhematopoietic populations. Measuring autofluorescence may also prove to be a useful indicator of injury and repair, especially if assessed over a longer time-course and correlated with other injury biomarkers.

The subtle differential expression of markers such as MHCII may also prove to be important in characterizing macrophage subsets and determining functional capabilities. Even the notion of a DC has been challenged in recent times with some evidence suggesting that they might be more closely associated with macrophages than previously thought. This study highlights the difference in the expression of the classical DC marker, CD11c, between the spleen and the kidney, and that the lack of a clear CD11c population may mean that examining CD11c on subpopulations may be more useful than trying to, for example, separate the CD45+ population into macrophages and DCs. The assessment of CD11c expression in this study also demonstrates the usefulness of measuring MFI for a particular antibody in lieu of, or in addition to, population proportions, especially when the expression is low or when shifts in expression levels are subtle.

Part of the challenge in using flow cytometry to assess subpopulations of cells in the kidney is choosing an appropriate panel of markers to investigate. This is further complicated knowing that different digestion methods may enhance detection of a particular cell type or negatively impact individual markers or receptors. The ED protocol described in this paper was optimized for the combination of enzymes used (collagenase/dispase, DNase type 1). The enzyme concentrations and incubation times, along with the method of mechanical dissociation (size of pipette tip and timing of the dissociations), were all methodically tested to achieve an optimal digestion as determined by cell counts, viability and flow cytometric profiles. This study demonstrated that ED is indeed required to achieve greater viable and CD45+ cells yields and to most effectively study cells expressing markers such as F4/80. However, variations in dissociation media may be required for different disease models, as some are characterized by inflammation, cell infiltrate, and cell death, whilst others may centre on fibrosis and collagen deposition. The combination of collagenase/dispase and DNase type 1 appeared to impact negatively on CSF-1R expression, as seen on Ly6Chigh and Ly6C- cells in the spleen, again highlighting the need to optimize digestion methods for each specific study.

Equally as rapid as the advancements in flow cytometer technology, is the development of new fluorochromes and viability dyes. These are providing narrower emission spectra allowing for greater clarity in population identification. There is also now a range of viability dyes available for a large variety of excitation and emission wavelengths. The interactive tools available online, such as spectra viewers and panel builders are also very useful in creating optimal antibody cocktails.


This study has highlighted some of the advantages and limitations associated with assessing kidney cells using flow cytometry, particularly in the IR injury model. This can be an incredibly powerful tool but requires a tested and systematic approach, including the method for organ digestion, antibody selection (target antigen and fluorochrome) and specific gating strategies. Other analytical techniques, including IHC, IF, and PCR should be used in conjunction with flow cytometry data to provide a complete depiction of cell types present together with localization in the tissue in which they reside. The obvious extension of the use of flow cytometry to analyze cell populations is the sorting of live populations for further investigations in vitro or in adoptive transfer experiments.


Timothy M. Williams and Daniel S. Layton have been involved in the development of FlowLogic FCS analysis software.