A rapid in vivo assay system for analyzing the organogenetic capacity of human kidney cells

Abstract

Transplantation of human kidney-derived cells is a potential therapeutic modality for promoting regeneration of diseased renal tissue. However, assays that determine the ability of candidate populations for renal cell therapy to undergo appropriate differentiation and morphogenesis are limited. We report here a rapid and humane assay for characterizing tubulogenic potency utilizing the well-established chorioallantoic membrane (CAM) of the chick embryo.

Adult human kidney-derived cells expanded in monolayer were suspended in Matrigel and grafted onto the CAM. After a week, grafts were assessed histologically. Strikingly, many of the renal cells self-organized into tubular structures. Host blood vessels penetrated and presumably fed the grafts. Immuno- and histochemical staining revealed that tubular structures were epithelial, but not blood vessels. Some of the cells both within and outside the tubules were dividing. Analysis for markers of proximal and distal renal tubules revealed that grafts contained individual cells of a proximal tubular phenotype and many tubules of distal tubule character.

Our results demonstrate that the chick CAM is a useful xenograft system for screening for differentiation and morphogenesis in cells with potential use in renal regenerative medicine.

Introduction

The therapeutic options for renal disease are limited. Cell therapy based on infusion of cell suspensions is an exciting potential treatment that is currently being evaluated in experimental animal models of renal disease. The therapeutic action of grafted cells involves generation of specific cell types and/or paracrine or immunomodulatory effects that are beneficial to remaining host tissue. In the case of the kidney, grafted bone marrow cells apparently act by paracrine mechanisms, while they lack nephrogenic capacities.¹ Kidney-derived epithelial cells might be advantageous as a treatment modality, as they have intrinsic potential to reconstitute renal structures.

Critical to the development of cell-based therapies that involve reconstitution of renal structures is an assay system that demonstrates capacity for making epithelia with a nephric phenotype. In vitro models for morphogenesis of embryonic kidney progenitors have been developed with work focusing on rodent renal cells (reviewed in ref. 2). These studies have shown budding and branching tubulogenesis from the ureteric bud and have been recombined with metanephric mesenchyme and transplanted to adult rodent hosts. However, they are based on non-human embryonic tissue and are therefore not ideal for assays more directly applicable to regenerative medicine.

The most common in vivo model for testing nephrogenic ability of human cells is the xenografting into immunodeficient mice and histological assessment of morphogenesis of tubular epithelial structures appropriate for kidney repair. For example, human adult progenitor cell types were subcutaneously injected into SCID mice to assess their differentiation.^3,4

The chorioallantoic membrane (CAM) of bird embryos is a highly vascularized but non-innervated structure normally adherent to the egg shell that is responsible for gas exchange. This vascular plexus forms around day 4 of incubation and grows dramatically during the incubation period. The vessels readily fuse with grafted tissues providing a “normal” blood supply and retention of 3D organization. Grafting of human tissues to the CAM of the chick egg is a well-established xenograft system, i.e., for human skin.^5–7 Xenografting to the chick embryo was recently reviewed in reference ⁸.

This system has been used for numerous biomedical studies, especially those examining angiogenesis and anti-angiogenesis pioneered by Judah Folkman decades ago. Tumor ontogeny (reviewed in ref. 9), metastasis (reviewed in ref. 10) and chemotherapy (reviewed in ref. 11) assays have also been performed with the chick CAM system. The CAM has also been used in many studies to observe the development and regeneration of embryonic tissues such as limb buds¹² or liver.¹³ Of particular relevance to the proposed study, nephric anlage, such as the meso- and metanephros,¹⁴ have been grafted to the CAM.

Here we demonstrate that grafting to the CAM is an alternative humane and rapid ex vivo model for assessing the ability of expanded adult human kidney cells to undergo renal tubulogenesis. The present study is, to our knowledge, the first demonstration that the CAM system can be used for testing for the organogenetic potential of in vitro expanded human kidney-derived cells.

Results

hAK cells, but not a cell line derived from human kidney cells, undergo spontaneous tubulogenesis when grafted to the CAM.

Suspensions of hAK cells in Matrigel grafted to the CAM coalesced into solid structures that appeared to attract growth of host blood vessels (Fig. 1A and B). Removal and observation of the unfixed grafts from the “embryo” side of the CAM revealed the vascularization more clearly (Fig. 1B).

Figure 1.

Grafts of human adult kidney progenitor cells, cells from the HeK293 kidney-derived line suspended in Matrigel and grafts of Matrigel only to the chick chorioallantoic membrane (CAM). Suspensions of 2.5 × 10⁶ adult human kidney cells suspended in Matrigel formed compact structures (asterisks) one week after transplantation that apparently attracted host blood vessels (arrows). the graft is shown in situ in the egg in (A) surrounded by a plastic ring that was used to contain the grafted Matrigel. In (B), the graft is shown after removal from the egg and being flipped “upside down” to show the blood vessels beneath. Histological analysis of two different grafts of adult human kidney progenitors from a single donor. (C) is a low-power micrograph of a section through a graft containing Matrigel (M), large amounts of unorganized cells (asterisk) and distinct tubules. Arrows in (C) denote chick blood vessels. At higher magnification in (D), the cuboidal epithelium of the tubules (tub) is easily observed. Red arrowheads point to bright pink-stained erythrocytes in capillaries and small blood vessels that are present throughout the graft. Micrographs of another graft of hAK cells (E and F) show large numbers of tubules of various sizes. two blood vessels (BV) containing magenta-colored erythrocytes are shown at higher magnification in the inset in (F). Grafts of HeK293 cells and Matrigel without cells. (G) is a micrograph of a section through a mass resulting from the grafting of a suspension of cells from the line HeK293, originally derived from human fetal kidney. Large “tumors” (asterisks) formed from the HeK293 cells in the CAM, but no tubular morphogenesis was observed. Large blood sinuses/vessels containing variable amounts of leukocytes penetrated the graft (arrows). When Matrigel (M) only (not containing cells) was transplanted to the CAM, a mass was present containing many blood vessels (arrowheads) after seven days, but was not invaded by other CAM cells (H). Scale bars: (A and B) 1 mm, (C) 500 µm, (D and F) 50 µm, (E and G) 200 µm, (H) 100 µm.

Serial paraffin sections were made of the hAK grafts and samples throughout stained with hematoxylin and eosin. Strikingly, the dissociated hAK cells had organized into tubular structures (Fig. 1C–F). These tubules often were in contact with host capillaries (see below) containing nucleated avian erythrocytes (Fig. 1F, inset). No structures with obvious glomerular morphology were observed. Some of the hAK cells did not organize into tubular structures and remained as amorphous masses (Fig. 1C and D). The presence of tubules in the sections was variable, both in individual grafts and between different grafts made from cells from the same patient. Some sections showed almost exclusively tubules (i.e., Fig. 1E and F), some mostly unorganized cells and some sections mixtures of both (Fig. 1C and D). Distinct tubules were observed in 13 of 13 grafts of 2–2.5 − 10⁶ hAK cells in three separate experiments, each using cells from a different donor, and were absent from 6 of 6 grafts of 2.5 × 10⁶ HEK293 cells from three experiments (see below).

In contrast to grafts of hAK cells, grafts of suspensions of the HEK293 line (derived from human embryonic kidney) in Matrigel generated only large masses without tubular structures (Fig. 1G). Large blood vessels/sinuses as well as capillaries were observed among the grafted HEK293 cells. The massive proliferation of the HEK293 cells resulted in the near absence of “empty” regions of Matrigel that were observed in hAK grafts. Grafts of Matrigel without cells to the CAM contained small host blood vessels, presumably attracted by the growth factors Matrigel contains, but large blood sinuses or other obvious host CAM cells did not penetrate the Matrigel in Matrigel-only grafts (Fig. 1H).

Characterization of grafts of human renal cells by immuno- and lectin histochemistry.

The grafts were then characterized with a series of antibodies and lectins. Sections from 14 week-gestation human fetal kidneys were used as positive controls for the stainings. Both the grafted tissue and the fetal kidney exhibited strong autofluorescence. To overcome this, spectral unmixing was used to reveal specific antibody and lectin staining (see methods).

Staining with a pan-cytokeratin antibody labeled many of the tubular structures in the hAK grafts (Fig. 2A, B and D). The human identity of the cells in the hAK grafts was confirmed using an antibody that recognized a human lysosome protein that did not react with chick tissues (Fig. 2C and D). Virtually all cells in hAK grafts with large nuclei stained for this marker, as did the large cells in the “tumors” formed by grafted HEK293 cells (Fig. 2E). As noted by us (reviewed in ref. 15) and others, chick cells in general have smaller nuclei than mammalian cells. By contrast, HEK293 cells were never stained with antibodies to cytokeratin, although some chick epithelia were immunopositive (Fig. 2F). Next, the proliferation state of the transplanted cells was assessed using an antibody to Ki67. Individual cells, both part of tubules and outside them, were observed to be cycling in hAK grafts (Fig. 2G), in contrast to virtually all the cells in the 293 grafts (Fig. 2H). Collagen IV, a major component of basal lamina, was massively produced by cells in hAK grafts (Fig. 2I) but was present at low levels in small patches between cells in the HEK293 cell grafts (Fig. 2J).

Figure 2.

Immuno- and histochemical analysis of CAM grafts of human kidney-derived cells. In (A) a section through a graft of hAK cells stained with a pan-cytokeratin antibody (green) is shown. This section contains mostly cytokeratin-immunopositive tubules. The CAM mesenchyme is unstained (asterisk). (B–D) are micrographs of a section double immunostained with antibodies to cytokeratins (green) and specific for a human lysosome protein (red). Cells in the few tubules present in this section (arrowheads) are labeled with the cytokeratin antibody (B). The human lysosome staining is present in virtually all cells, confirming their derivation from the graft (C). (D) is a merged image showing the overlap of the cytokeratin and lysosome staining of the tubules and the many human lysosome⁺ cells that had not organized into epithelia. Sections through grafts of HeK293 kidney-derived cells immunostained for human lysosomes (E, red) and cytokeratins (F, green). All the grafted cells in (E), but not the cells of the host CAM (asterisk) are immunopositive for the human cell marker. By contrast, the HeK293 cells are not labeled with the cytokeratin antibody in (F). The immunopositive structure in (F) is the epithelial surface of the CAM (arrow); the antibody used bound avian cytokeratins. Staining of a section through a graft of hAK cells with an antibody to cell division marker Ki67 (G, red) shows cells in a tubule (arrowhead) and an individual cell that are cycling. By contrast, all the cells in a graft of HeK293 (H) cells are dividing. Immunostaining with an antibody to collagen IV, a major component of the basal lamina, shows extensive expression of this protein in a section through a CAM graft of hAK cells (I), while the chick mesenchyme (asterisk) is unstained. Small patches of collagen IV are present in a graft of HeK293 cells (J). Staining of an hAK graft with Dolichos lectin (K) labeled many tubules (i.e., arrowhead), indicating their distal/collecting tubule phenotype. The inset shows several Dolichos-positive (green) and one negative (arrow) tubules in a section through a human fetal kidney. Some AK cells, apparently not part of a tubule, bind Lotus lectin, indicating differentiation to a proximal tubule phenotype (L). In all images, erythrocytes and granulocytes were autofluorescent in both the green and red fluorescent channels and were therefore removed by the spectral unmixing process (see methods). Nuclei (blue) were stained with Hoechst or propidium iodide. The propidium iodide is shown as blue for consistency between the panels. Bar in (A) represents 50 µm in (A, E, F, H–K) and 25 µm in (B–D, G and L).

Lectin histochemistry revealed that all of the tubular structures formed were of distal/collecting tubule phenotype. Dolichos lectin (DBA) decorated many of the tubules of the hAK grafts (Fig. 2K) and control fetal kidney sections (Fig. 2K, inset). Only individual cells were observed to stain in hAK grafts with the Lotus lectin (LTA) that identifies proximal tubules (Fig. 2L), while LTA labeled many tubules in control fetal kidney section s(not shown). Neither of the lectins stained grafts of HEK293 cells (not shown).

Sections through hAK grafts stained with an antibody specific for mammalian CD31 did not contain immunopositive structures, while sections of fetal kidney contained multiple stained blood vessels (not shown), suggesting that the AK cell population grafted did not contain endothelial cells or progenitors. Antibody stainings were performed for several other renal markers, including Wilms' tumor (WT1), Tamm-Horsfall glycoprotein, epithelial cell adhesion molecule (EpCAM), vimentin and neural cell adhesion molecule (NCAM), but after spectral unmixing, no specific staining was obtained for any of these. This was apparently due to absence of cells with renal phenotypes that express these proteins (see discussion).

Discussion

An important component of the development of a cell therapy approach to treatment of renal degenerative disease is an assay system for appropriate organogenesis by the transplanted cells. The classic system for testing this potential of stem-progenitor cells is that of the immunodeficient mouse. We describe here the use of the chick CAM as a supplement/alternative to the use of nude/SCID mice for assay of neogenesis of renal structures from cells with therapeutic potential. We observed rapid (1 week) tubulogenesis by expanded hAK cells and expression of specific phenotypic markers.

The CAM system has several advantages over the better-known murine model. For example, fertile eggs are inexpensive, require no maintenance, rapid engraftment is obtained and eggs are naturally “immunodeficient” (full immunocompetence only develops weeks after hatching). The CAM is not innervated, and experiments are terminated before the development of centers in the brain associated with pain perception, making this a humane system not requiring animal experimentation permissions. In addition, the graft is continually visible on the CAM, and component cells can be easily visualized using fluorescently labeled cells (reviewed in ref. 15).

Although we observed extensive morphogenesis of tubules, many having a distal/collecting duct phenotype, we did not observe nephrons or proximal tubules. It is possible that hAK cells passaged in adherent culture lose the ability to differentiate into additional renal cell types and structures. However, lack of other renal structures/phenotypes could have been due to chick CAM grafts being limited in time to a week or so. This time limitation could potentially be overcome by performing serial transplantations from one egg to another, as is done in rodents. Alternatively, using eggs of an avian species with a longer incubation period, such as the turkey,¹⁵ would allow extension of the graft period and perhaps result in more extensive differentiation. Another potential approach to achieving the differentiation of additional renal phenotypes would be to alter the microenvironment of the cells. We used standard Matrigel, which includes a complex mixture of growth factors secreted by the sarcoma cells from which it was made. Use of growth factor-depleted Matrigel or other scaffold materials, with or without the addition of specific nephrogenic factors,¹⁶ could allow the generation of other phenotypes in the grafts.

The establishment of this CAM assay for morphogenesis and differentiation by normal human kidney-derived cells now allows rapid screening of cell populations with potential in cell therapy of renal pathologies. In addition, the CAM system, used for many years in the study of human cancers, is useful for studying malignant renal stem/progenitor cells. Accordingly, we recently analyzed the effects of antibody-mediated targeting of a Wilms tumor progenitor population on the engraftment to the CAM of proliferating human Wilms Tumor cells.¹⁷ In summary, the avian CAM system is a rapid and humane assay that with continued development could become a powerful tool for characterizing cells with potential in renal regenerative medicine.

Materials and Methods

Preparation of human kidney cells.

Normal adult kidney samples (hAK) were retrieved from borders of renal cell carcinoma tumors from partial and total nephrectomy patients. Tissues were obtained from both Sheba Medical Center and Wolfson hospital following informed consent and with institutional ethical permissions. The samples were minced in HBSS, incubated in collagenase for 2 hr and cultured in Iscoves modified MEM with 10% FBS, 1% L-glutamine, 1% penicillin-streptomycin and the following growth factors: 50 ng/ml bFGF, 50 ng/ml EGF and 5 ng/ ml SCF (R and D systems). Cells were split upon reaching 90% confluence and the medium changed every 3 days. hAK cells used in the grafting experiments described here were from the second or third passage. HEK293 cells were used as controls for the grafting experiments; they were maintained in DMEM with 10% fetal calf serum, penicillin-streptomycin and glutamine.

CAM grafting.

Fertile chicken eggs were obtained from a commercial supplier and incubated at 37°C at 60–70% humidity in a forced-draft incubator. At 3 days of incubation, an artificial air sac was established dropping the CAM. A window was opened in the shell and the CAM exposed on day 9 or 10 of incubation. Suspensions of cells in Matrigel (BD Biosciences, USA) were kept on ice to prevent gelation and 50 µl/egg pipetted into an approximately 0.8 cm plastic ring cut from a 200 µl pipette tip placed on the membrane. The eggs were then sealed with adhesive tape and returned to the incubator. After an additional 5–7 days of incubation, the tape was removed and grafts photographed both in situ and after removal and pinning in a silicone rubber-coated dish. The transplants were then fixed overnight in 4% buffered paraformaldehyde and embedded in paraffin.

Histology and microscopy.

Serial 6 µm sections were made of the grafts and samples at a distance of 50–100 µm stained with hematoxylin and eosin for histological assessment of renal morphogenesis. After the graft was identified and characterized histologically, additional sections were stained with antibodies or lectins (Table 1). Immunocytochemical staining was performed with microwave antigen retrieval. As positive controls, paraffin sections of human fetal kidneys from 14 week abortuses (obtained with the appropriate ethical permissions) were stained simultaneously, usually on the same slide. Counterstaining was performed with Hoechst nuclear stain or propidium iodide. Photomicrographs were taken with color and monochrome Scion digital cameras attached to Olympus SZX and BX51 microscopes and figures assembled using PaintShop Pro. Only changes in brightness, contrast and sharpness were made, and these changes were made to the entire image.

Table 1.

Antibodies and lectins used in this study

Antibody/lectin/avidin	Source	Dilution
Human lysosomal protein	Developmental Studies Hybridoma Bank H4A3	1:50
Ki67	Thermo-Scientific, #RM-9106-S0	1:200
Dolichos biflorus agglutinin	Vectorlabs #B-1035	1:1250
Lotus tetragonolobus agglutinin	Vectorlabs #B-1325	1:400
Pan-cytokeratin	Dako #ZO622	1:200
Collagen IV	Develop. Studies Hybridoma Bank M3F7	1:5
Strepavidin-Cy2	Jackson Immunoresearch. #016-220-084	1:500
Goat anti-mouse alexa 488	Invitrogen #A11001	1:300
Goat anti-rabbit alexa 488	Invitrogen #A11008	1:200
Goat anti-mouse alexa 594	Invitrogen #21203	1:300
Goat anti-rabbit alexa 594	Invitrogen #A11012	1:100

Spectral unmixing of fluorescence.

Strong autofluorescence was present in both the grafted tissue and the fetal kidney sections. To obtain a specific fluorescent signal, spectral unmixing was performed using a multispectral imaging system (NuanceFX camera and software, CRi, USA) attached to an Olympus AX70 motorized microscope. In brief, a series of spectra of autofluorescent areas in unstained tissue (both of erythrocytes and other cells/extracellular matrix) and pure solutions of the secondary antibodies were collected at a wide range of excitation and emission wavelengths. The spectra of the autofluorescence and dyes were then subtracted digitally to generate an image with only the specific fluorescence.

Abbreviations

CAM

chorioallantoic membrane

hAK

human adult kidney

SCID

severe combined immunodeficient