Forward scatter pulse width signals resolve multiple populations of endosomes

Development of a flow cytometry-based pulse-width assay for detection of aggregates in cellular therapeutics to be infused by catheter

BACKGROUND AIMS

Delivery of cell-based therapies through the carotid artery with the use of an intra-arterial catheter could introduce aggregates and cause focal ischemia in the brain. We developed a pulse-width flow cytometry method for aggregate detection and quantification. The assay was designed to be used as a cell product release assay in a clinical trial seeking to treat ischemic stroke with sorted cells brightly expressing aldehyde dehydrogenase (ALDH(br) cells) delivered through intra-arterial catheters.

METHODS

The forward light scatter pulse-width axis of a flow cytometer was calibrated for particle diameter measurements through the use of traceable standard microspheres and linear regression. As a positive control, Concanavalin A-aggregated cells were counted manually and sorted onto slides to compare with pulse width-determined values. Known numbers of aggregates were spiked into purified singlet cells for quantification. A clinical standard for aggregate count and diameter was determined. The assay was used to qualify catheters with the use of ALDH(br) cells.

RESULTS

The pulse-width axis was highly linear for microsphere diameter (r(2) > 0.99), which allowed for size calibration. Microscopically determined counts and diameters corresponded to pulse width-determined values. Known aggregate counts were linear with pulse width-determined aggregate counts (r(2) = 0.98). The limit of detection was determined to be 0.004%. Flow of ALDH(br) cells through catheters did not generate aggregates. The final method to be used as a release assay for the stroke clinical trial was tested successfully on samples from volunteer donors.

CONCLUSIONS

The pulse-width aggregate detection assay provides a reliable, reproducible, accurate and rapid means of detection, classification and quantification of aggregates in cell therapy products.

Substance P dependence of endosomal fusion during bladder inflammation

Urinary bladder instillation of ovalbumin into presensitized guinea pigs stimulates rapid development of local bladder inflammation. Substance P is an important mediator of this inflammatory response, as substance P antagonists largely reverse the process. Vacuolization of the subapical endosomal compartment of the transitional epithelial cells lining the bladder suggests that changes in endosomal trafficking and fusion are also part of the inflammatory response. To test directly for substance P mediation of changes in endosomal fusion, we reconstituted fusion of transitional cell endosomes in vitro using both cuvette-based and flow cytometry energy transfer assays. Bladders were loaded with fluorescent dyes by a hypotonic withdrawal protocol before endosomal isolation by gradient centrifugation. Endosomal fusion assayed by energy transfer during in vitro reconstitution was both cytosol and ATP dependent. Fusion was confirmed by the increase in vesicle size on electron micrographs of fused endosomal preparations compared with controls. In inflamed bladders, dye uptake was inhibited 20% and endosomal fusion was inhibited 50%. These changes are partly mediated by the neurokinin-1 (NK1) receptor (NK1R), as 4 mg/kg of CP-96,345, a highly selective NK1 antagonist, increased fusion in inflamed bladders but had no effect on control bladders. The receptor-mediated nature of this effect was demonstrated by the expression of substance P receptor mRNA in rat bladder lumen scrapings and by the detection of the NK1R message in guinea pig subapical endosomes by Western blot analysis. The NK1Rs were significantly upregulated following induction of an inflammatory response in the bladder. These results demonstrate that 1) in ovalbumin-induced inflammation in the guinea pig bladder, in vitro fusion of apical endosomes is inhibited, showing endocytotic processes are altered in inflammation; 2) pretreatment in vivo with an NK1R antagonist blocks this inhibition of in vitro fusion, demonstrating a role for NK1R in this process; and 3) the NK1R is present in higher amounts in apical endosomes of inflamed bladder, suggesting changes in translation or trafficking of the NK1R during the inflammatory process. This suggests that NK1R can change the fusion properties of membranes in which it resides.

Renal endosomes contain angiotensin peptides, converting enzyme, and AT(1A) receptors

Kidney cortex and proximal tubular angiotensin II (ANG II) levels are greater than can be explained on the basis of circulating ANG II, suggesting intrarenal compartmentalization of these peptides. One possible site of intracellular accumulation is the endosomes. In the present study, we tested for endosomal ANG I, ANG II, angiotensin type 1A receptor (AT(1A)), and angiotensin converting enzyme (ACE) activity and determined whether these levels are regulated by salt intake. Male Sprague-Dawley rats were fed chow containing either high or low dietary sodium for 10-14 days. Blood and kidneys were harvested and processed for measurement of plasma, kidney, and renal intermicrovillar cleft and endosomal angiotensin levels. Kidney ANG I averaged 179 +/- 20 fmol/g and ANG II averaged 258 +/- 36 fmol/g in rats fed a high-sodium diet and were significantly higher, averaging 347 +/- 58 fmol/g and 386 +/- 55 fmol/g, respectively, in rats fed a low-salt diet. Renal intermicrovillar clefts and endosomes contained ANG I and ANG II. Intermicrovillar cleft ANG I and ANG II levels averaged 8.4 +/- 2.6 and 74 +/- 26 fmol/mg, respectively, in rats fed a high-salt diet and 7.6 +/- 1.7 and 70 +/- 25 fmol/mg in rats fed a low-salt diet. Endosomal ANG I and ANG II levels averaged 12.3 +/- 4.4 and 43 +/- 19 fmol/mg, respectively, in rats fed a high-salt diet, and these levels were similar to those observed in rats fed a low-salt diet. Renal endosomes from rats fed a low-salt diet demonstrated significantly more AT(1A) receptor binding compared with rats fed a high-salt diet. ACE activity was detectable in renal intermicrovillar clefts and was 2.5-fold higher than the levels observed in renal endosomes. Acute enalaprilat treatment decreased ACE activity in renal intermicrovillar clefts by 90% and in renal endosomes by 84%. Likewise, intermicrovillar cleft and endosomal ANG II levels decreased by 61% and 52%, respectively, in enalaprilat-treated animals. These data demonstrate the presence of intact angiotensin peptides and ACE activity in renal intermicrovillar clefts and endosomes, indicating that intact angiotensin peptides are formed and/or trafficked through intracellular endosomal compartments and are dependent on ACE activity.

Alveolar macrophage apoptosis and TNF-alpha, but not p53, expression correlate with murine response to bleomycin

Apoptosis is considered to be a protective mechanism that limits lung injury. However, apoptosis might contribute to the inflammatory burden present in the injured lung. The exposure of mice to bleomycin (BLM) is a well-established model for the study of lung injury. BLM exposure induces DNA damage and enhances tumor necrosis factor (TNF)-alpha expression in the lung. To evaluate the importance of alveolar macrophage (AM) apoptosis in the pathogenesis of lung injury, we exposed BLM-sensitive (C57BL/6) and BLM-resistant (BALB/c) mice to BLM (120 mg/kg) and studied the induction of apoptosis [by light-microscopy changes (2, 8, 12, 24, 48, and 72 h) and annexin V uptake by flow cytometry (24 h)], the secretion of TNF-alpha (measured by ELISA), and the expression of p53 (by immunoblotting) in AM retrieved from these mice. BLM, but not vehicle, induced apoptosis in AM from both murine strains. The numbers of apoptotic AM were significantly greater (P < 0.001) in C57BL/6 mice (52.9%) compared with BALB/c mice (40.8%) as demonstrated by annexin V uptake. BLM induction of apoptosis in AM was preceded by an increased secretion of TNF-alpha in C57BL/6 but not in BALB/c mice. Furthermore, double TNF-alpha receptor-deficient mice, developed on a C57BL/6 background, demonstrated significantly (P < 0.001) lower numbers of apoptotic AM compared with C57BL/6 and BALB/c mice. BLM also enhanced p53 expression in AM from both murine strains. However, p53-deficient mice developed BLM-induced lung injury, exhibited similar lung cell proliferation (measured as proliferating cell nuclear antigen immunostaining), and accumulated similar amounts of lung hydroxyproline (65 +/- 6.9 microgram/lung) as did C57BL/6 (62 +/- 6.5 microgram/lung) mice. Therefore, AM apoptosis is occurring during BLM-induced lung injury in a manner that correlates with murine strain sensitivity to BLM. Furthermore, TNF-alpha secretion rather than p53 expression contributes to the difference in murine strain response to BLM.tumor necrosis factor; strain susceptibility

Myeloma light chains are ligands for cubilin (gp280)

Although myeloma light chains are known to undergo receptor-mediated endocytosis in the kidney, the molecular identity of the receptor has not been characterized. We examined the interaction between cubilin (gp280) and four species of light chains isolated from the urine of patients with multiple myeloma. Four lines of evidence identify cubilin, a giant glycoprotein receptor, which is restricted in distribution to endocytic scavenger pathways and which has potent effects on endosomal trafficking, as a potentially physiologically relevant binding site for light chains: 1) light chains coeluted during immunoaffinity purification of cubilin; 2) polyclonal antisera to cubilin but not control sera, displaced human light chain binding from rat renal brush-border membranes; 3) cubilin bound to multiple species of light chains during surface plasmon resonance; 4) anti-cubilin antiserum interfered with light chain endocytosis by visceral yolk sac epithelial cells. However, both binding of light chains to brush-border membranes and endocytosis of light chains by yolk sac epithelial cells were only partially inhibited by anticubilin antibodies, suggesting presence of additional or alternate binding sites for light chains. Excess light chain had a potent inhibitory effect on endosomal fusion in vitro. Binding showed dose and time-dependent saturability with low-affinity, high-capacity equilibrium binding parameters. These data demonstrate that cubilin plays a role in the endocytosis and trafficking of light chains in renal proximal tubule cells.

Separation of endosomes by aqueous two-phase partition and free-flow electrophoresis

We have developed two endosome models to evaluate the separation of endosome populations by aqueous two-phase partition. In the first model, bovine kidney endosomes were used. In the second model. HeLa endosomes were identified in homogenates by means of a latent drug-(capsaicin-)inhibited NADH oxidase (NOX). Endosomes were first isolated by aqueous two-phase partition. To separate early and late endosomes, the endosomes were incubated with ATP to acidify the endosome interiors by activating a proton-translocating ATPase. Thus far, we have been able to resolve the early and late endosomes from any source only by preparative free-flow electrophoresis and not by phase-partition. Previous studies have shown that gravitational forces may be important for separation of endosomes by phase partition. Low-speed centrifugation (< or =12.5 g) during phase resolution altered the activity of the latent NADH oxidase used as a marker for HeLa cell endosomes.

Regulation of expression of transcobalamin II receptor in the rat

Surface and intracellular membrane distribution and hormonal regulation of transcobalamin II receptor (TC II-R) activity and protein levels have been studied in an effort to understand its regulation of expression in the rat. TC II-R activity and the levels of the 62 kDa monomeric and 124 kDa dimeric forms of TC II-R were highest in the rat kidney and intestine, and in these tissues the receptor expression was not dependent upon the postnatal development of the rat. TC II-R expression was uniform in the various regions of the gut. Surface membrane distribution of TC II-R in the kidney revealed the expression of the 124 kDa dimer form of TC II-R in the apical and basolateral membranes in the ratio of 1:10. Further subcellular distribution of TC II-R in the kidney revealed the expression of the 124 kDa dimer in the intermicrovillar clefts and clathrin-coated vesicles and the 62 kDa monomer in the microsomes. Neither the monomer nor the dimer could be detected in the early endosomes or lysosomes. Membrane TC II-R activity and TC II-R protein levels and cobalamin (Cbl; vitamin B12) transport in vivo were inhibited by about 90% in adrenalectomized rats and all three returned to normal levels by oral treatment of these animals with cortisone acetate. In contrast, thyroidectomy or experimentally induced diabetes had no effect on TC II-R activity or Cbl transport. Based on these observations, we suggest that TC II-R expression is not developmentally or regionally regulated in rat renal and intestinal membranes and its expression in the kidney is asymmetrically distributed between the apical (10%) and basolateral (90%) membranes. In addition, our results also show that the dimerization of TC II-R is a post-microsomal event and that the expression of TC II-R and plasma Cbl transport is regulated by cortisone.

Flow cytometric measurement of NK cell immunoconjugates by pulse width processing

Pulse width analysis, in flow cytometry, has been widely used for optimal cell size resolution in cell kinetics analysis. Pulses, generated by scattered light or fluorescence of cells, are electronically analyzed for their height and width. The information generated from these two properties of the pulses is utilized to distinguish signals from single cells vs. signals from cell clumps or aggregates. Pulse width, unlike pulse height, is more sensitive to differences in cell diameter, and therefore can discriminate very small differences in it, which pulse height cannot. We have exploited this property of pulse widths to measure immunoconjugates between NK cells and their targets. Discrimination of the free target cells from the conjugated ones is possible by the pulse widths of only light scatter signals, both forward and/or orthogonal. This resolution was not obtained if pulse height of the same signals was visualized. Using this resolution it was possible to distinguish single cells from the aggregates between target and effector cells. We propose that this is a better method for distinguishing conjugates than the method in which prior vital staining of cells is used.

Permeability properties of the mammalian bladder apical membrane

The luminal surface of mammalian bladder is exposed to urine with a composition widely different from that of plasma that bathes the basolateral surface of epithelium. Therefore we predict that the bladder permeability barrier, which is likely located in the apical membrane (AM), will exhibit low permeabilities to water, urea, NH3, H+, and small nonelectrolytes. AM surface area increases as the bladder fills with urine and decreases during emptying, a process that involves cyclical endocytosis and reinsertion of membrane from a pool of AM endosomes (AME). Rigid-appearing plaques composed of three proteins, uroplakins, have been identified and occupy 70-90% of AM surface area. To determine permeability properties of the AM permeability barrier, we purified AME and measured their permeabilities. Rabbit urinary bladders were removed, and their apical surface was exposed to carboxyfluorescein (CF) or horseradish peroxidase (HRP). Exposure to hypotonic and then isotonic basolateral solutions induced endocytosis of luminal CF or HRP into AME. Electron microscopy of bladders after this treatment revealed HRP entrapped within AME bordered by plaques. AME were purified by differential and sucrose-gradient centrifugation, and CF-containing AME were purified 17.0 +/- 3-fold (SD) with respect to homogenate. Analysis of purified AME by flow cytometry showed that > 95% of vesicles contained CF entrapped from luminal solution and were selectively labeled with anti-uroplakin antibody. AME osmotic water permeability averaged 2.3 +/- 0.66 x 10(-4) cm/s and exhibited a high activation energy, indicating that AM contains no water channels. Permeability to urea and NH3 averaged 7.8 +/- 3.7 x 10(-7) and 1.5 +/- 0.3 x 10(-3) cm/s, respectively, which are exceptionally low and similar to permeabilities of other water-tight membranes, including toad urinary bladder and gastric mucosa. AME behaved as a single population in all permeability studies, which will permit future characterization of protein and lipid structure responsible for these unique permeability properties.

Heterogeneity of endosomal populations in the rat renal cortex: light endosomes

The endosomal pathway of the rat renal cortex was labeled by intravenous infusion of fluorescent dextran small enough to cross the glomerular ultrafiltration barrier and be taken up by luminal endocytosis in the proximal tubule. Using Percoll gradient centrifugation, we isolated and characterized a previously undetected renal cortical endosomal fraction slightly lighter than basolateral membranes. Assay of entrapped dextran on a vesicle-by-vesicle basis using small-particle flow cytometry techniques demonstrates homogeneity for entrapped dextran. Flow cytometry colocalization of entrapped markers with brush-border enzymes in > 99% of the vesicles and the absence of Na-K-adenosinetriphosphatase (ATPase) suggest both that these vesicles are of apical origin and that apical enzymes traffic into endosomal elements. Furthermore, two glycoproteins derived from intermicrovillar clefts are detectable in this fraction. Ultrastructurally the vesicles are heterogeneous, consisting of multivesicular bodies together with vesicles of diverse size and coating. Populations of vesicles can be cleanly separated from each other and basolateral membranes according to their surface charge by high-resolution free-flow electrophoresis. Multiparameter flow cytometry analysis demonstrates that a more abundant population of smaller vesicles has brisker H(+)-ATPase activity, whereas a less abundant population of larger vesicles has slower H(+)-ATPase activity. In contrast, brush-border membrane vesicles contained no entrapped markers and lacked H(+)-ATPase activity. Analysis of vesicles prepared after addition of dextran to the homogenate confirms that the vesicles form in vivo. Hence a heterogeneous renal cortical endosomal population is of apical origin.

Experimental basis for separation of membrane vesicles by preparative free-flow electrophoresis

In practice it has been possible to separate membrane particles of different origins but of similar chemical composition by preparative free-flow electrophoresis. Examples include the vacuolar (tonoplast) and plasma membranes of plants and membranes derived from the cis and trans regions of the rat liver Golgi apparatus. Yet, when analyzed for intrinsic molecules that might contribute to significant differences in surface charge, the separated membranes were surprisingly similar. As more information was generated, it became apparent that the membranes with greatest electrophoretic mobility (i.e. lysosomes, rightside-out tonoplast vesicles and membranes from the trans region of the Golgi apparatus), where those membranes with an inherent ability to acidify their interiors. By so doing, the vesicles generate a membrane potential, negative outside, which might serve as a basis for enhanced electrophoretic mobility. To test the hypothesis, tonoplast membranes were incubated with ATP to drive proton import or with monensin to dissipate the ATP-supported proton gradient. With ATP, mobility was enhanced. Also, when ATP-treated vesicles were analyzed in the presence of monensin, the ATP effect on mobility was reversed. Similarly with Golgi apparatus, mobility of the most electrophoretically mobile portions of the separation was enhanced by ATP and the ATP effect was reversed with monensin. A trans origin of the vesicles was verified by assay of the trans Golgi apparatus marker, thiamine pyrophosphatase. Finally, incubation with ATP (and reversal by monensin) was employed as an aid to the free-flow electrophoretic separation of kidney endosomes from complex mixtures. These lysosomal derivatives also are capable of acidification of their interiors in an ATP-dependent process and of generating, at the same time, a negative (outside) membrane potential. The findings provide both an experimental basis to enhance membrane separations by preparative free-flow electrophoresis and, at the same time, a theoretical basis to help explain why certain membranes of very similar overall chemical composition may be separated by electrophoretic methods.

Isolation of highly purified, functional endosomes from toad urinary bladder

Endosomes are difficult to isolate as they share size and density properties with much more abundant cellular organelles such as mitochondria. In cultured cell lines the tandem use of charge-dependent isolation techniques and differential centrifugation is necessary to isolate endosomes. Endosomal populations of the toad urinary bladder are of special interest because they are thought to contain a water channel. Understanding of the molecular structure of the water channel has been constrained, as there is currently no practical method to isolate functional water-channel-containing vesicles. This study reports the tandem use of charge-dependent techniques and centrifugation to isolate populations of endosomes from the toad urinary bladder. To purify water-channel-containing vesicles aqueous two-phase partition was utilized to fractionate a preparation partially purified by differential centrifugation. Populations of endosomes were analysed by small-particle flow cytometry techniques. A 5-fold enrichment in endosomes, achieved with aqueous two-phase partition, allowed us to identify two populations of endosomes of diverse size in a toad bladder endosomal fraction. Preenrichment also improved the efficiency of flow cytometry sorting, allowing isolation of the two endosomal populations in sufficient quantities for secondary analysis. A population of larger endosomes had very high water permeability, indicating the presence of water channels. The two populations had different SDS/PAGE fingerprints. Electron micrographs of the flow-sorted material shows a uniform population of membrane vesicles devoid of mitochondria and other identifiable cellular organelles. Hence, aqueous two-phase partition and flow cytometry allow identification of two populations of endosomes in the toad urinary bladder which have diverse structural and functional properties. Isolation of functional water-channel-containing vesicles allows co-localization of water-channel function with candidate water-channel proteins.