- Email: [email protected]
Candida albicans adherence to endothelial cells
MICROVASCULAR
RESEARCH
43, 218-226 (1992)
TECHNICAL Candida CYNTHIA
albicans
L. MAYER,*
REPORT
Adherence
to Endothelial
SCOTT G. FILLER,*
‘Division of Adult Infectious Diseases, Department Torrance, California 90509, and -FlJCLA School Received
March
Cells
AND JOHN E. EDWARDS, JR.*‘-p’
of Medicine, of Medicine,
Harbor/UCLA Los Angeles,
Medical California
Center, 90024
28, 1991
Mechanisms of adherence to vascular endothelial cells by microorganisms on a molecular level can be elucidated by using monoclonal antibodies, purified cell wall constituents, and receptor analogues. Since these agents are expensive and available in limited quantities, a microsystem for probing adherence mechanisms to these cells has become essential. We studied techniques to accurately quantify the adherence of L-[35S]methionine-labeled Candida albicans to human umbilical vein endothelial cells in a 96-well microtiter plate system while avoiding specific problems related to Candida coadherence and avid binding to plastic. The endothelial cells were grown on a collagen matrix in individually detachable mircrowells enabling the determination of the number of adherent organisms from radioactive counts of the entire well. This procedure has the critically important advantage of obviating the need to remove adherent Candida from the wells. Expressing adherence to endothelial cell monolayers as the percentage of total organisms added to each well significantly decreases the variability of the assay. 0 19% Academic press, IW
INTRODUCTION Microbial pathogens seeding target organs by the hematogenous route must traverse the boundaries of the vascular compartment to enter the tissue parenchyma. It is highly probable that adherence of microorganisms to endothelial cells is the important intial step in their escape from the perfusing vasculature. To facilitate understanding the basic pathogenesis of this step in organ invasion, studies on the adherence of microbes to endothelial cells are being conducted (Cheung et al., 1991; Klotz et al., 1983; Rotrosen et al., 1986, 1985; Thomas et al., 1988). Such studies will likely become important for the development of therapeutic strategies focused on blocking adherence of the organisms to endothelium, thereby inhibiting their escape into parenchyma. By preventing the invading pathogens from leaving the circulation, defense mechanisms such as neutrophils and monocytes would have greater opportunity to eradicate them. Critical to the expansion of the field of microbial adherence is the development of mi’ To whom correspondence should be addressed at Division of Adult Infectious Diseases, E-5, Harbor/UCLA Medical Center, 1000 W. Carson St., Torrance, CA 90509. 218 0026.2862/92 $3.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved Printed in U.S.A.
TECHNICAL
REPORT
219
croassay systems to determine the ability of such reagents as monoclonal antibodies, cytokines, and microorganism cell wall fractions to modulate adherence. One of the most important organisms emerging as a frequent cause of blood borne infection resulting in widespread hematogenous dissemination is Candida albicans. In our studies on the pathogenesis of hematogenously disseminated candidiasis we have focused on the adherence and penetration of endothelial cells by this organism. Although much is known about the molecular basis of the interaction between endothelial cells and leukocytes (Springer, 1990) or bacteria (Cheung et ul., 1991), there is considerably less information concerning the endothelial cell adherence of fungi. For instance, the mechanism of Cundidu-endothelial cell interaction has not been elucidated and basic questions such as whether this interaction is specific remain to be answered. Additionally, there is no generally recognized “gold standard” assay to quantify Cundida adherence to biologic surfaces, nor has a standard inoculum been decided upon. Furthermore, studying Can&u-endothelial interactions is complicated by two of its biological properties. First, Cundidu strongly binds to the plastic containers that are commonly used in adherence assays (Rotrosen et al., 1986). Falsely high adherence values are obtained if monolayers are not confluent because the organism will bind to the plastic or endothelial cell submatrix between nonconfluent endothelial cells. Second, at concentrations greater than 1 x 10’ organisms/ml, the organisms avidly coadhere to form large clumps which may be counted as a single cell under common assay conditions (Kennedy et al., 1989). Both problems are magnified in microsystems where the relative surface area of the container is large and the organisms are suspended in a small volume of medium. In this paper, we describe a microassay for the accurate quantification of C. ulbicuns adherence to human endothelial cell monolayers. This assay circumvents specific difficulties that occur when working with the organisms. The assay also allows adherence to endothelial cells and other biological surfaces to be studied with monoclonal antibodies and purified, potential adherence modifiers which are limited in quantity and often very expensive.
MATERIALS Radioactive
Labeling
and Germination
AND METHODS of Cundidu
Blastospores of Cundidu ulbicuns (ATCC 36082) were prepared using a modification of the procedure of Rotrosen et al. (1986). The organisms were grown in yeast nitrogen base broth (YNB; Difco Laboratories, Detroit, MI) supplemented with 1% dextrose and 0.15% t,-asparagine at 27°C overnight on a rotating drum. The next day, the organisms were inoculated into fresh YNB broth and grown overnight. After three consecutive passages, the Cundidu were washed twice in saline and sonicated for 4 set (Branson Sonic Power Co., Danbury, CT). The cells were then labeled according to a modification of Sundstrom et al. (1987) as follows: the singlet blastospores were resuspended at 1 x lo7 organisms/ml in 4.5ml YNB broth without amino acids containing 70 $Zi of L-[35S]methionine (ICN, Irvine, CA). This mixture was incubated for 2 hr at 30°C on a rotary shaker at 250 rpm. The labeled cells were washed in saline seven times to remove the unincorporated radioactivity, sonicated, and resuspended in germination medium
220
TECHNICAL
REPORT
(0.15% sucrose and 0.075% gelatin in HzO) and incubated at 37°C for 1 hr. As soon as visible germination occurred, the singlet organisms were harvested and adjusted to 3 x lO’/ml in Dulbecco’s phosphate-buffered saline (PBS). For a typical experiment, 3 x lo4 organisms yielded 99,751 t 8867 cpm and > 90% of the radioactivity was cell-associated (the small percentage of non-cell-associated L-[35S]methionine released during germination was not rinsed out). To determine if there was a linear relationship between the radioactivity of a sample and the number of organisms contained within it, the organisms were labeled with L-[35S]methionine in the above manner and the radioactivity of various dilutions of the organisms was compared with the colony counts obtained by growing samples of the different dilutions on yeast-dextrose-potassium agar. Endothelial
Cell Monolayers
Human umbilical vein endothelial cells (HUVE) were prepared by a modification of the method of Jaffe et al. (1973). Second passage cells in Ml99 with 10% fetal bovine serum (GIBCO, Grand Island, NY) and 10% defined bovine calf serum (Hyclone, Logan, UT) and supplemented with endothelial cell growth factor (Bionetics Research Institute, Rockville, MD) were grown to confluency in Costar (Van Nuys, CA) 96-well, microtiter plates that were coated with fibronectin (1 pg/ml; Collaborative Research, Bedford, MA). In other experiments, we used RIA strip plates (Costar) which were made of rows of wells that are linked by easily breakable plastic bridges and fit into a holder. These rows of wells formed a 96-well plate identical to a 96-well, tissue culture, microtiter plate, except that the wells could be manually snapped apart. The strip plate wells were coated with either fibronectin or a collagen matrix (Vitrogen; Collagen Biomedical, Palo Alto, CA) prepared following the manufacturer’s instructions prior to addition of HUVE. The HUVE were grown to confluency as in the Costar tissue culture plates. Just before use in the adherence assay (see below), the medium was aspirated from the wells of the microtiter or strip plate and the endothelial cell monolayers were rinsed twice with warm PBS. Adherence
Assay
One hundred microliters of labeled C. al&cans suspension was added to each well of either the microtiter or strip plate. For each experiment, the plate was incubated at 37°C for 30 min. After incubation, the wells were rinsed four times with 150 ~1 of saline to remove all nonadherent Candida. All rinses were saved for scintillation counting. Preliminary experiments were performed in bare plastic wells to select methods for further evaluation on endothelial cell monolayers. In subsequent experiments, all wells contained confluent monolayers of HUVE. Methods
to Remove Candida from 96Well,
Tissue Culture Plates
We evaluated the ability of different treatments to remove the Candida remaining in the wells and adherent to the monolayer, after rinsing for scintillation counting (designated as adherent counts). In all cases, 100 ~1 of the various solutions was applied to the wells. One-step methods to remove adherent organisms from the wells included incubation in saline, NaOH (0.5, 1, or 5 N), or 0.25% trypsin (Sigma, St. Louis, MO). In other experiments, a two-step procedure was used. The wells were incubated in 5 N NaOH followed by Multiterge (50%
TECHNICAL
REPORT
221
in H,O; Diagnostic Systems, Inc., Gibbstown, NJ), or 5 N NaOH followed by Radiacwash (50% in H,O; Atomic Products, Shirley, NY) or 0.25% trypsin followed by 5 N NaOH. Both Multiterge and Radiacwash are nonionic detergents formulated to remove radioactivity from contaminated surfaces. After adding the first treatment, (either NaOH or trypsin) the wells were incubated for 15, 30, or 60 min at 37°C. Next, the second treatment of either NaOH or detergent was added and the wells were incubated for an additional 5 min. After each treatment, the contents of the wells were transferred to scintillation vials. Following both the one-step and two-step methods, the wells were rinsed twice with saline and these rinses were also added to the scintillation vials. To determine if any adherent Candida remained in the wells, the individual wells were separated with a hand saw, and the residual radioactivity was determined for each well by scintillation counting. Use of Snap-Apart
When adherent manually the total
Strip Plates
the strip plates were used, no treatments were employed to remove the Candida. After rinsing off the nonadherent organisms, the wells were snapped apart and added to scintillation vials, and the cpm representing adherent organisms obtained for each well directly.
Scintillation
Counting
All samples were placed in plastic scintillation vials containing 5 ml Ready Safe Scintillation Fluid (Beckman, Fullerton, CA) and the P-emissions were counted for 2 min in a Packard 2200CA Liquid Scintillation Counter (Packard Instrument Co., Downers Grove, IL). Data Analysis
Since greater than 90% of the radioactivity was cell-associated and all of the non-cell-associated radioactivity was rinsed out, the total adherent cpm were proportional to the number of adherent organisms. For the 96-well microtiter tissue culture plates, the number of adherent organisms in each well was determined by adding the cpm of the individually sawed apart wells to those of the organisms removed by the treatments and subsequent rinses. When the adherence assay was performed using the strip plates, the total number of adherent organisms was determined by simply measuring the cpm of the snapped-apart wells. The total number of organisms added to each well was proportional to the sum of adherent cpm plus the cpm of nonadherent organisms removed in the rinses. The percentage adherence for each well of either type of plate was equal to the total adherent cpm/total cpm added to that well. Comparison of the cpm of different treatment groups was performed with Student’s t test, using the Bonferroni correction for multiple comparisons. P values < 0.05 were considered significant. All conditions were tested in replicates of at least four. All results are expressed as mean +- 1 SD. RESULTS There was a linear relationship between the radioactivity of each sample and the number of colony-forming units (r = 0.99), as shown in Fig. 1. Candida adherence ranged from 43 to 57% of the total inoculum. As can be seen in Fig.
222
TECHNICAL
REPORT
CFU (x 1,000)
FIG. 1. Relationship between the colony counts and radioactivity of various numbers of radiolabeled C. nlbicans. Error bars represent 1 SD. At the lower inocula, the error bars are smaller than the symbol size.
2A (representing the preliminary studies), 47 + 15% of the total adherent counts remained in the microtiter wells following efforts to remove Can&u with trypsin. This value was almost 25% of the total inoculum. Since it was clear that incubation with trypsin failed to remove a large percentage of the organisms attached to the plastic, stronger methods to remove the adherent Candida from the wells were evaluated. Figure 2B shows representative results of incubating the wells containing HUVE and adherent Can&u with saline, trypsin, and increasing con-
B
A 1ooT
1ooT
90
90 --
60
60 --
70
70 -60--
z= $ m 60 50
0 m
t
Treatment Wells
50 -40 -30-20 --
; Ttypsin
Saline
Trypsin & NaOH
0.5N NaOH
5N NaOH
5N NaOH Raiiac
FIG. 2. Removal of adherent C. albicans from microtiter plate wells with various treatment. All treatment incubations were 30 min except for the 5 N NaOH + Radiacwash which was 60 min in NaOH followed by 5 min in Radiacwash. (A) Cundida on bare plastic; (B) Candida on HUVE.
TECHNICAL
REPORT
223
centrations of NaOH. After treatment with saline, 94% of the Cundidu still remained in the wells. Higher concentrations of NaOH alone removed increasing numbers of organisms. Incubation with trypsin followed by a 5-min exposure to 5 N NaOH failed to remove 29% of the adherent organisms. However, even a 60-min incubation with 5 N NaOH followed by a 5-min incubation with Radiacwash left 18% of the adherent organisms in the wells. We determined that virtually all of the residual radioactivity remaining in the wells was Can&da-associated because, when L[35S]methionine alone was added to the wells, 195% of the radioactivity could be rinsed from the wells with saline (data not shown). Since we were unable to remove such a large percentage of the organisms from the wells of the solid Costar microtiter plate with chemicals, we investigated the use of RIA strip plates to avoid the time-consuming task of cutting up a radioactive, 96-well plate by hand. Because the plastic was not tissue culture treated, endothelial cells did not grow satisfactorily in these plates, even when the wells were coated with fibronectin. The weak adherence of the monolayer to this type of plate caused unacceptably high losses of endothelial cells when the wells were rinsed. However, coating the wells with a collagen matrix provided satisfactory endothelial cell attachment even with the extensive rinsing necesitated by the assay. Using this type of plate, the number of adherent organisms could be directly determined by counting the total radioactivity of each well after rinsing off the nonadherent organisms. To verify that a full range of values could be detected using the strip plates, adherence of Cundidu to endothelial cell monolayers was measured over time (Fig. 3). The percentage adherence increased over time and reached a plateau at 45 min. Expressing adherence values as the percentage of total counts added to each well resulted in low standard deviations, enabling the detection of small differences in adherence. DISCUSSION The first microassay described to measure Cundidu adherence to HUVE monolayers was the enzyme-linked immunosorbent assay (ELISA) developed by Filler et al. (1987). These authors used polyclonal anti-Cundidu antibodies to estimate the number of organisms adherent to the luminal surface of endothelial cells in a 96-well microtiter plate. Using this method, high numbers of organisms (e.g., lo4 to 105) could be tested and evaluation of small volumes of adherence modifiers was feasible. However, we found that some potential blockers of adherence were antigenically related to the Cundidu cell wall. They cross-reacted with the antiCundidu antibodies of the ELISA, and produced falsely elevated adherence values (unpublished data). In this paper we present a method which obviates the disadvantage of antigenic cross-reactivity in the ELISA while preserving the ability to test microquantities of potential adherence modifying substances; the method utilizes C. ulbicuns labeled with Q3*S]methionine so that the number of organisms adherent to the endothelial cell monolayer can be determined by scintillation counting. Conceptually, it is tempting to believe that the adherent organisms can be recovered from the wells by removing the endothelial cell monolayer. However, by using radiolabeled organisms we discovered that trypsinization of the monolayer
224
TECHNICAL
70 60--
:= E
REPORT
T .yT
50--
6 g‘; 30-40-:b 20CL
y-A
1
0/
lo-:
/ i
0-t 0
15
30
45
60
Time (minutes) FIG. 3. Adherence of C. nlbicans to HUVE as a function of incubation time. At 15 and 30 min, the error bars are smaller than the symbol size.
(an often-used procedure that removes the monolayer entirely) leaves approximately one-half of the adherent organisms in the wells. Probably, during the process of removing the monolayer, the Cundida readhere to the newly exposed subendothelial surface and bind avidly to it. Klotz and Maca (1988) have found that C. al&cans adheres more strongly to plastic and extracellular matrix than it does to endothelial cells. Additionally, when C. albicans are allowed to germinate on plastic, they produce extracellular mannoproteins that remain adherent to the plastic, even when the organisms have been physically removed with a rubber policeman (Tronchin et al., 1988). Thus, some of the residual radioactivity left in the microtiter wells may have been associated with exrtracellular mannoproteins synthesized by the Cundida when germinating through the monolayer rather than with whole organisms. In our study, even the use of such treatments as NaOH and detergents left 18% of the adherent radiolabel in the wells. Therefore, unless the microtiter wells are physically separated, or the radioactivity of the contents of the entire incubation container/vessel is somehow measured, the number of organisms adherent to the endothelial cell surface cannot be accurately determined. Thus, all adherent organisms must be recovered and counted to draw meaningful conclusions about adherence. We have been unable to devise ways to circumvent this necessity of counting residual Cundida adherent to the wells after the monolayer has been removed. An alternative might be to estimate the number of adherent Cundidu by subtracting the cpm of the nonadherent organisms from the total cpm added to each well. The average cpm per well could be determined by measuring the radioactivity of an aliquot of the organisms. However, large errors are likely to occur for two reasons. First, organisms may be lost in collecting the multiple rinses required to remove the nonadherent Candidu. Second, and more importantly, a random variation occurs in the number of organisms added to each well when pipetting small volumes of a particulate suspension. If all adherence values are based on a constant inoculum, the actual well-to-well variation in inoculum size results in significant
TECHNICAL
REPORT
225
variability in adherence values within an experiment (data not shown). This variability obscures small differences in adherence. To account for this variation in inoculum size, we found it necessary to express the adherence values as the percentage of total cpm added to each well. Since the total number of organisms added to each well is equal to the sum of the adherent and nonadherent organisms added to that well, it is not possible to quantify the total cpm added to each well unless the radioactivity of the adherent organisms in that well is determined directly. From our data, we conclude that the only accurate method of directly measuring the number of Candida that adhere to the endothelial cells in a microsystem is to count the radioactivity of the entire well. By coating the RIA strip wells with a collagen matrix, we have successfully grown confluent endothelial cell monolayers. Growing endothelial cells on this substrate in these plates eliminates two problems. First, the monolayers remain intact through the rinsing procedure due to enhanced adherence of the endothelial cells to the collagen matrix. Second, since individual wells can be easily separated for scintillation counting, methods to remove adherent Candida from the wells are not required. The deletion of this extra step from the procedure substantially decreases well-to-well variability in adherence values. An added advantage of measuring the total cpm per well occurs when comparing the adherence of different strains or species of Cundidu in the same assay. This procedure allows the adherence values to be corrected for the varying rates of radiolabel incorporation that may occur between different organisms. The importance of accounting for residual Cundidu in the microtiter wells of an adherence assay can be extrapolated to other Cundidu assay systems performed in any plasticware. If a small volume of a Cundidu suspension is placed in a plastic container, a significant fraction of the total inoculum may be unrecoverable due to adherence of the organisms to the plastic. For instance, we found that 34 2 5% of the Cundidu inoculum remained bound to the sides of microfuge tubes, even after rinsing the tubes four times with saline (data not shown). Accounting for this factor is critical in assays such as those designed to measure the killing of Cundidu by neutrophils, where organisms are incubated with leukocytes in either plastic vials or microtiter plates and then an aliquot of the mixture is withdrawn for analysis of Cundidu viability (Husseini et uE., 1985). Not only does Cundidu adherence to the sides of the container cause a falsely elevated estimate of the number of organisms killed, but the organisms remaining in suspension may represent a subset of Cundidu with characteristics significantly different from those of the total population. The assay we have developed for measuring adherence of Cundidu to endothelial cells obviates a source of significant error. The use of this system will facilitate future studies directed at elucidating the endothelial cell adherence mechanisms of Cundidu on a molecular level; studies which are critical to fully understanding the mechanisms of escape of the organism from the vascular compartment. ACKNOWLEDGMENTS This work was supported by a USPH Service Grant AI19990 and in part by Grant 927Fl-1 from the American Heart Association (Greater Los Angeles Affiliate). S. G. Filler was the Gilbert Dalldorf
226
TECHNICAL REPORT
Fellow in Medical Mycology. The IMT-2 Olympus phase-contrast microscope used in these studies was graciously donated by Toyota U.S.A. We thank M. Ghannoum for his invaluable assistance in preparing this manuscript.
REFERENCES A. L., KRISHNAN, M., JAFFE, E. A., AND RSCHETTI, V. A. (1991). Fibrinogen acts as a bridging molecule in the adherence of Staphylococcus aureus to cultured human endothelial cells.
CHEUNG,
.I. Clin. Invest. 87, 2236-2245. FILLER, S. G., DER, L. C., MAYER,
C. L., CHIRSTENSON, P. D., AND EDWARDS, J. E., JR. (1987). An enzyme-linked immunosorbent assay for quantifying adherence of Candida to human vascular endothelium. J. Infect. Dis. 156, 561-566. HUSSEINI, R. H., HOADLEY, M. E., HUTCHINSON, J. J. P., PENN, C. W., AND SMITH, H. (1985). Intracellular killing of Cundidu abicans by human polymorphonuclear leucocytes: Comparison of three methods of assessment. J. Zmmunol. Methods 81, 215-221. JAFFE, E. A., NACHMAN, R. L., BECKER, C. G., AND MINICK, C. R. (1973). Culture of human endothelial cells derived from umbilical veins: Identification by morphologic and immunologic criteria. J. C/in. Invest. 52, 2745-2756. KENNEDY, M. J., ROGERS,
A. L., AND YANCEY, R. J., JR. (1989). Environmental alterations and phenotyoic regulation of Candida albicans adhesion to plastic. Infect. Immun. 57, 3876-3881. KLOTZ, S. A., DRUTZ, D. J., HARRISON, J. L., AND HUPPERT, M. (1983). Adherence and penetration of vascular endothelium by Candida yeasts. infect. Immun. 42, 374-384. KLOTZ, S. A., AND MACA, R. D. (1988). Endothelial cell contraction increases Candida adherence to exposed extracellular matrix. Infect. Immun. 56, 2495-2498. ROTROSEN, D., CALDERONE, R. A., AND EDWARDS, J. E., JR. (1986). Adherence of Cundida species to host tissues and plastic surfaces. Rev. Infect. Dis. 8, 73-85. ROTROSEN, D., EDWARDS, J. E., JR., GIBSON, T. R., MOORE, J. C., COHEN, A. H., AND GREEN, I. (1985). Adherence of Cundida to cultured vascular endothelial cells: Mechanisms of attachment and endothelial cell penetration. J. Infect. Dis. 152, 1264-1274. SPRINGER, T. A. (1990). Adhesion receptors of the immune system. Nature 346, 425-434. SUNDSTROM, P. M., NICHOLS, E. J., AND KENNY, G. E. (1987). Antigenic differences between mannoproteins of germ tubes and blastospores of Candida albicans. Infect. Immun. 55, 616-620. THOMAS, P. D., HAMPSON, F. W., AND HUNNINGHAKE, G. W. (1988). Bacterial adherence to human endothelial cells. J. Appl. Physiol. 65, 1372-1376. TRONCHIN, G., BOUCHARA, J. P., ROBERT, R., AND SENET, J. M. (1988). Adherence of Candida albicans germ tubes to plastic: Ultrastructural and molecular studies of fibrillar adhesins. Infect. Immun. 56, 1987-1993.
RESEARCH
43, 218-226 (1992)
TECHNICAL Candida CYNTHIA
albicans
L. MAYER,*
REPORT
Adherence
to Endothelial
SCOTT G. FILLER,*
‘Division of Adult Infectious Diseases, Department Torrance, California 90509, and -FlJCLA School Received
March
Cells
AND JOHN E. EDWARDS, JR.*‘-p’
of Medicine, of Medicine,
Harbor/UCLA Los Angeles,
Medical California
Center, 90024
28, 1991
Mechanisms of adherence to vascular endothelial cells by microorganisms on a molecular level can be elucidated by using monoclonal antibodies, purified cell wall constituents, and receptor analogues. Since these agents are expensive and available in limited quantities, a microsystem for probing adherence mechanisms to these cells has become essential. We studied techniques to accurately quantify the adherence of L-[35S]methionine-labeled Candida albicans to human umbilical vein endothelial cells in a 96-well microtiter plate system while avoiding specific problems related to Candida coadherence and avid binding to plastic. The endothelial cells were grown on a collagen matrix in individually detachable mircrowells enabling the determination of the number of adherent organisms from radioactive counts of the entire well. This procedure has the critically important advantage of obviating the need to remove adherent Candida from the wells. Expressing adherence to endothelial cell monolayers as the percentage of total organisms added to each well significantly decreases the variability of the assay. 0 19% Academic press, IW
INTRODUCTION Microbial pathogens seeding target organs by the hematogenous route must traverse the boundaries of the vascular compartment to enter the tissue parenchyma. It is highly probable that adherence of microorganisms to endothelial cells is the important intial step in their escape from the perfusing vasculature. To facilitate understanding the basic pathogenesis of this step in organ invasion, studies on the adherence of microbes to endothelial cells are being conducted (Cheung et al., 1991; Klotz et al., 1983; Rotrosen et al., 1986, 1985; Thomas et al., 1988). Such studies will likely become important for the development of therapeutic strategies focused on blocking adherence of the organisms to endothelium, thereby inhibiting their escape into parenchyma. By preventing the invading pathogens from leaving the circulation, defense mechanisms such as neutrophils and monocytes would have greater opportunity to eradicate them. Critical to the expansion of the field of microbial adherence is the development of mi’ To whom correspondence should be addressed at Division of Adult Infectious Diseases, E-5, Harbor/UCLA Medical Center, 1000 W. Carson St., Torrance, CA 90509. 218 0026.2862/92 $3.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved Printed in U.S.A.
TECHNICAL
REPORT
219
croassay systems to determine the ability of such reagents as monoclonal antibodies, cytokines, and microorganism cell wall fractions to modulate adherence. One of the most important organisms emerging as a frequent cause of blood borne infection resulting in widespread hematogenous dissemination is Candida albicans. In our studies on the pathogenesis of hematogenously disseminated candidiasis we have focused on the adherence and penetration of endothelial cells by this organism. Although much is known about the molecular basis of the interaction between endothelial cells and leukocytes (Springer, 1990) or bacteria (Cheung et ul., 1991), there is considerably less information concerning the endothelial cell adherence of fungi. For instance, the mechanism of Cundidu-endothelial cell interaction has not been elucidated and basic questions such as whether this interaction is specific remain to be answered. Additionally, there is no generally recognized “gold standard” assay to quantify Cundida adherence to biologic surfaces, nor has a standard inoculum been decided upon. Furthermore, studying Can&u-endothelial interactions is complicated by two of its biological properties. First, Cundidu strongly binds to the plastic containers that are commonly used in adherence assays (Rotrosen et al., 1986). Falsely high adherence values are obtained if monolayers are not confluent because the organism will bind to the plastic or endothelial cell submatrix between nonconfluent endothelial cells. Second, at concentrations greater than 1 x 10’ organisms/ml, the organisms avidly coadhere to form large clumps which may be counted as a single cell under common assay conditions (Kennedy et al., 1989). Both problems are magnified in microsystems where the relative surface area of the container is large and the organisms are suspended in a small volume of medium. In this paper, we describe a microassay for the accurate quantification of C. ulbicuns adherence to human endothelial cell monolayers. This assay circumvents specific difficulties that occur when working with the organisms. The assay also allows adherence to endothelial cells and other biological surfaces to be studied with monoclonal antibodies and purified, potential adherence modifiers which are limited in quantity and often very expensive.
MATERIALS Radioactive
Labeling
and Germination
AND METHODS of Cundidu
Blastospores of Cundidu ulbicuns (ATCC 36082) were prepared using a modification of the procedure of Rotrosen et al. (1986). The organisms were grown in yeast nitrogen base broth (YNB; Difco Laboratories, Detroit, MI) supplemented with 1% dextrose and 0.15% t,-asparagine at 27°C overnight on a rotating drum. The next day, the organisms were inoculated into fresh YNB broth and grown overnight. After three consecutive passages, the Cundidu were washed twice in saline and sonicated for 4 set (Branson Sonic Power Co., Danbury, CT). The cells were then labeled according to a modification of Sundstrom et al. (1987) as follows: the singlet blastospores were resuspended at 1 x lo7 organisms/ml in 4.5ml YNB broth without amino acids containing 70 $Zi of L-[35S]methionine (ICN, Irvine, CA). This mixture was incubated for 2 hr at 30°C on a rotary shaker at 250 rpm. The labeled cells were washed in saline seven times to remove the unincorporated radioactivity, sonicated, and resuspended in germination medium
220
TECHNICAL
REPORT
(0.15% sucrose and 0.075% gelatin in HzO) and incubated at 37°C for 1 hr. As soon as visible germination occurred, the singlet organisms were harvested and adjusted to 3 x lO’/ml in Dulbecco’s phosphate-buffered saline (PBS). For a typical experiment, 3 x lo4 organisms yielded 99,751 t 8867 cpm and > 90% of the radioactivity was cell-associated (the small percentage of non-cell-associated L-[35S]methionine released during germination was not rinsed out). To determine if there was a linear relationship between the radioactivity of a sample and the number of organisms contained within it, the organisms were labeled with L-[35S]methionine in the above manner and the radioactivity of various dilutions of the organisms was compared with the colony counts obtained by growing samples of the different dilutions on yeast-dextrose-potassium agar. Endothelial
Cell Monolayers
Human umbilical vein endothelial cells (HUVE) were prepared by a modification of the method of Jaffe et al. (1973). Second passage cells in Ml99 with 10% fetal bovine serum (GIBCO, Grand Island, NY) and 10% defined bovine calf serum (Hyclone, Logan, UT) and supplemented with endothelial cell growth factor (Bionetics Research Institute, Rockville, MD) were grown to confluency in Costar (Van Nuys, CA) 96-well, microtiter plates that were coated with fibronectin (1 pg/ml; Collaborative Research, Bedford, MA). In other experiments, we used RIA strip plates (Costar) which were made of rows of wells that are linked by easily breakable plastic bridges and fit into a holder. These rows of wells formed a 96-well plate identical to a 96-well, tissue culture, microtiter plate, except that the wells could be manually snapped apart. The strip plate wells were coated with either fibronectin or a collagen matrix (Vitrogen; Collagen Biomedical, Palo Alto, CA) prepared following the manufacturer’s instructions prior to addition of HUVE. The HUVE were grown to confluency as in the Costar tissue culture plates. Just before use in the adherence assay (see below), the medium was aspirated from the wells of the microtiter or strip plate and the endothelial cell monolayers were rinsed twice with warm PBS. Adherence
Assay
One hundred microliters of labeled C. al&cans suspension was added to each well of either the microtiter or strip plate. For each experiment, the plate was incubated at 37°C for 30 min. After incubation, the wells were rinsed four times with 150 ~1 of saline to remove all nonadherent Candida. All rinses were saved for scintillation counting. Preliminary experiments were performed in bare plastic wells to select methods for further evaluation on endothelial cell monolayers. In subsequent experiments, all wells contained confluent monolayers of HUVE. Methods
to Remove Candida from 96Well,
Tissue Culture Plates
We evaluated the ability of different treatments to remove the Candida remaining in the wells and adherent to the monolayer, after rinsing for scintillation counting (designated as adherent counts). In all cases, 100 ~1 of the various solutions was applied to the wells. One-step methods to remove adherent organisms from the wells included incubation in saline, NaOH (0.5, 1, or 5 N), or 0.25% trypsin (Sigma, St. Louis, MO). In other experiments, a two-step procedure was used. The wells were incubated in 5 N NaOH followed by Multiterge (50%
TECHNICAL
REPORT
221
in H,O; Diagnostic Systems, Inc., Gibbstown, NJ), or 5 N NaOH followed by Radiacwash (50% in H,O; Atomic Products, Shirley, NY) or 0.25% trypsin followed by 5 N NaOH. Both Multiterge and Radiacwash are nonionic detergents formulated to remove radioactivity from contaminated surfaces. After adding the first treatment, (either NaOH or trypsin) the wells were incubated for 15, 30, or 60 min at 37°C. Next, the second treatment of either NaOH or detergent was added and the wells were incubated for an additional 5 min. After each treatment, the contents of the wells were transferred to scintillation vials. Following both the one-step and two-step methods, the wells were rinsed twice with saline and these rinses were also added to the scintillation vials. To determine if any adherent Candida remained in the wells, the individual wells were separated with a hand saw, and the residual radioactivity was determined for each well by scintillation counting. Use of Snap-Apart
When adherent manually the total
Strip Plates
the strip plates were used, no treatments were employed to remove the Candida. After rinsing off the nonadherent organisms, the wells were snapped apart and added to scintillation vials, and the cpm representing adherent organisms obtained for each well directly.
Scintillation
Counting
All samples were placed in plastic scintillation vials containing 5 ml Ready Safe Scintillation Fluid (Beckman, Fullerton, CA) and the P-emissions were counted for 2 min in a Packard 2200CA Liquid Scintillation Counter (Packard Instrument Co., Downers Grove, IL). Data Analysis
Since greater than 90% of the radioactivity was cell-associated and all of the non-cell-associated radioactivity was rinsed out, the total adherent cpm were proportional to the number of adherent organisms. For the 96-well microtiter tissue culture plates, the number of adherent organisms in each well was determined by adding the cpm of the individually sawed apart wells to those of the organisms removed by the treatments and subsequent rinses. When the adherence assay was performed using the strip plates, the total number of adherent organisms was determined by simply measuring the cpm of the snapped-apart wells. The total number of organisms added to each well was proportional to the sum of adherent cpm plus the cpm of nonadherent organisms removed in the rinses. The percentage adherence for each well of either type of plate was equal to the total adherent cpm/total cpm added to that well. Comparison of the cpm of different treatment groups was performed with Student’s t test, using the Bonferroni correction for multiple comparisons. P values < 0.05 were considered significant. All conditions were tested in replicates of at least four. All results are expressed as mean +- 1 SD. RESULTS There was a linear relationship between the radioactivity of each sample and the number of colony-forming units (r = 0.99), as shown in Fig. 1. Candida adherence ranged from 43 to 57% of the total inoculum. As can be seen in Fig.
222
TECHNICAL
REPORT
CFU (x 1,000)
FIG. 1. Relationship between the colony counts and radioactivity of various numbers of radiolabeled C. nlbicans. Error bars represent 1 SD. At the lower inocula, the error bars are smaller than the symbol size.
2A (representing the preliminary studies), 47 + 15% of the total adherent counts remained in the microtiter wells following efforts to remove Can&u with trypsin. This value was almost 25% of the total inoculum. Since it was clear that incubation with trypsin failed to remove a large percentage of the organisms attached to the plastic, stronger methods to remove the adherent Candida from the wells were evaluated. Figure 2B shows representative results of incubating the wells containing HUVE and adherent Can&u with saline, trypsin, and increasing con-
B
A 1ooT
1ooT
90
90 --
60
60 --
70
70 -60--
z= $ m 60 50
0 m
t
Treatment Wells
50 -40 -30-20 --
; Ttypsin
Saline
Trypsin & NaOH
0.5N NaOH
5N NaOH
5N NaOH Raiiac
FIG. 2. Removal of adherent C. albicans from microtiter plate wells with various treatment. All treatment incubations were 30 min except for the 5 N NaOH + Radiacwash which was 60 min in NaOH followed by 5 min in Radiacwash. (A) Cundida on bare plastic; (B) Candida on HUVE.
TECHNICAL
REPORT
223
centrations of NaOH. After treatment with saline, 94% of the Cundidu still remained in the wells. Higher concentrations of NaOH alone removed increasing numbers of organisms. Incubation with trypsin followed by a 5-min exposure to 5 N NaOH failed to remove 29% of the adherent organisms. However, even a 60-min incubation with 5 N NaOH followed by a 5-min incubation with Radiacwash left 18% of the adherent organisms in the wells. We determined that virtually all of the residual radioactivity remaining in the wells was Can&da-associated because, when L[35S]methionine alone was added to the wells, 195% of the radioactivity could be rinsed from the wells with saline (data not shown). Since we were unable to remove such a large percentage of the organisms from the wells of the solid Costar microtiter plate with chemicals, we investigated the use of RIA strip plates to avoid the time-consuming task of cutting up a radioactive, 96-well plate by hand. Because the plastic was not tissue culture treated, endothelial cells did not grow satisfactorily in these plates, even when the wells were coated with fibronectin. The weak adherence of the monolayer to this type of plate caused unacceptably high losses of endothelial cells when the wells were rinsed. However, coating the wells with a collagen matrix provided satisfactory endothelial cell attachment even with the extensive rinsing necesitated by the assay. Using this type of plate, the number of adherent organisms could be directly determined by counting the total radioactivity of each well after rinsing off the nonadherent organisms. To verify that a full range of values could be detected using the strip plates, adherence of Cundidu to endothelial cell monolayers was measured over time (Fig. 3). The percentage adherence increased over time and reached a plateau at 45 min. Expressing adherence values as the percentage of total counts added to each well resulted in low standard deviations, enabling the detection of small differences in adherence. DISCUSSION The first microassay described to measure Cundidu adherence to HUVE monolayers was the enzyme-linked immunosorbent assay (ELISA) developed by Filler et al. (1987). These authors used polyclonal anti-Cundidu antibodies to estimate the number of organisms adherent to the luminal surface of endothelial cells in a 96-well microtiter plate. Using this method, high numbers of organisms (e.g., lo4 to 105) could be tested and evaluation of small volumes of adherence modifiers was feasible. However, we found that some potential blockers of adherence were antigenically related to the Cundidu cell wall. They cross-reacted with the antiCundidu antibodies of the ELISA, and produced falsely elevated adherence values (unpublished data). In this paper we present a method which obviates the disadvantage of antigenic cross-reactivity in the ELISA while preserving the ability to test microquantities of potential adherence modifying substances; the method utilizes C. ulbicuns labeled with Q3*S]methionine so that the number of organisms adherent to the endothelial cell monolayer can be determined by scintillation counting. Conceptually, it is tempting to believe that the adherent organisms can be recovered from the wells by removing the endothelial cell monolayer. However, by using radiolabeled organisms we discovered that trypsinization of the monolayer
224
TECHNICAL
70 60--
:= E
REPORT
T .yT
50--
6 g‘; 30-40-:b 20CL
y-A
1
0/
lo-:
/ i
0-t 0
15
30
45
60
Time (minutes) FIG. 3. Adherence of C. nlbicans to HUVE as a function of incubation time. At 15 and 30 min, the error bars are smaller than the symbol size.
(an often-used procedure that removes the monolayer entirely) leaves approximately one-half of the adherent organisms in the wells. Probably, during the process of removing the monolayer, the Cundida readhere to the newly exposed subendothelial surface and bind avidly to it. Klotz and Maca (1988) have found that C. al&cans adheres more strongly to plastic and extracellular matrix than it does to endothelial cells. Additionally, when C. albicans are allowed to germinate on plastic, they produce extracellular mannoproteins that remain adherent to the plastic, even when the organisms have been physically removed with a rubber policeman (Tronchin et al., 1988). Thus, some of the residual radioactivity left in the microtiter wells may have been associated with exrtracellular mannoproteins synthesized by the Cundida when germinating through the monolayer rather than with whole organisms. In our study, even the use of such treatments as NaOH and detergents left 18% of the adherent radiolabel in the wells. Therefore, unless the microtiter wells are physically separated, or the radioactivity of the contents of the entire incubation container/vessel is somehow measured, the number of organisms adherent to the endothelial cell surface cannot be accurately determined. Thus, all adherent organisms must be recovered and counted to draw meaningful conclusions about adherence. We have been unable to devise ways to circumvent this necessity of counting residual Cundida adherent to the wells after the monolayer has been removed. An alternative might be to estimate the number of adherent Cundidu by subtracting the cpm of the nonadherent organisms from the total cpm added to each well. The average cpm per well could be determined by measuring the radioactivity of an aliquot of the organisms. However, large errors are likely to occur for two reasons. First, organisms may be lost in collecting the multiple rinses required to remove the nonadherent Candidu. Second, and more importantly, a random variation occurs in the number of organisms added to each well when pipetting small volumes of a particulate suspension. If all adherence values are based on a constant inoculum, the actual well-to-well variation in inoculum size results in significant
TECHNICAL
REPORT
225
variability in adherence values within an experiment (data not shown). This variability obscures small differences in adherence. To account for this variation in inoculum size, we found it necessary to express the adherence values as the percentage of total cpm added to each well. Since the total number of organisms added to each well is equal to the sum of the adherent and nonadherent organisms added to that well, it is not possible to quantify the total cpm added to each well unless the radioactivity of the adherent organisms in that well is determined directly. From our data, we conclude that the only accurate method of directly measuring the number of Candida that adhere to the endothelial cells in a microsystem is to count the radioactivity of the entire well. By coating the RIA strip wells with a collagen matrix, we have successfully grown confluent endothelial cell monolayers. Growing endothelial cells on this substrate in these plates eliminates two problems. First, the monolayers remain intact through the rinsing procedure due to enhanced adherence of the endothelial cells to the collagen matrix. Second, since individual wells can be easily separated for scintillation counting, methods to remove adherent Candida from the wells are not required. The deletion of this extra step from the procedure substantially decreases well-to-well variability in adherence values. An added advantage of measuring the total cpm per well occurs when comparing the adherence of different strains or species of Cundidu in the same assay. This procedure allows the adherence values to be corrected for the varying rates of radiolabel incorporation that may occur between different organisms. The importance of accounting for residual Cundidu in the microtiter wells of an adherence assay can be extrapolated to other Cundidu assay systems performed in any plasticware. If a small volume of a Cundidu suspension is placed in a plastic container, a significant fraction of the total inoculum may be unrecoverable due to adherence of the organisms to the plastic. For instance, we found that 34 2 5% of the Cundidu inoculum remained bound to the sides of microfuge tubes, even after rinsing the tubes four times with saline (data not shown). Accounting for this factor is critical in assays such as those designed to measure the killing of Cundidu by neutrophils, where organisms are incubated with leukocytes in either plastic vials or microtiter plates and then an aliquot of the mixture is withdrawn for analysis of Cundidu viability (Husseini et uE., 1985). Not only does Cundidu adherence to the sides of the container cause a falsely elevated estimate of the number of organisms killed, but the organisms remaining in suspension may represent a subset of Cundidu with characteristics significantly different from those of the total population. The assay we have developed for measuring adherence of Cundidu to endothelial cells obviates a source of significant error. The use of this system will facilitate future studies directed at elucidating the endothelial cell adherence mechanisms of Cundidu on a molecular level; studies which are critical to fully understanding the mechanisms of escape of the organism from the vascular compartment. ACKNOWLEDGMENTS This work was supported by a USPH Service Grant AI19990 and in part by Grant 927Fl-1 from the American Heart Association (Greater Los Angeles Affiliate). S. G. Filler was the Gilbert Dalldorf
226
TECHNICAL REPORT
Fellow in Medical Mycology. The IMT-2 Olympus phase-contrast microscope used in these studies was graciously donated by Toyota U.S.A. We thank M. Ghannoum for his invaluable assistance in preparing this manuscript.
REFERENCES A. L., KRISHNAN, M., JAFFE, E. A., AND RSCHETTI, V. A. (1991). Fibrinogen acts as a bridging molecule in the adherence of Staphylococcus aureus to cultured human endothelial cells.
CHEUNG,
.I. Clin. Invest. 87, 2236-2245. FILLER, S. G., DER, L. C., MAYER,
C. L., CHIRSTENSON, P. D., AND EDWARDS, J. E., JR. (1987). An enzyme-linked immunosorbent assay for quantifying adherence of Candida to human vascular endothelium. J. Infect. Dis. 156, 561-566. HUSSEINI, R. H., HOADLEY, M. E., HUTCHINSON, J. J. P., PENN, C. W., AND SMITH, H. (1985). Intracellular killing of Cundidu abicans by human polymorphonuclear leucocytes: Comparison of three methods of assessment. J. Zmmunol. Methods 81, 215-221. JAFFE, E. A., NACHMAN, R. L., BECKER, C. G., AND MINICK, C. R. (1973). Culture of human endothelial cells derived from umbilical veins: Identification by morphologic and immunologic criteria. J. C/in. Invest. 52, 2745-2756. KENNEDY, M. J., ROGERS,
A. L., AND YANCEY, R. J., JR. (1989). Environmental alterations and phenotyoic regulation of Candida albicans adhesion to plastic. Infect. Immun. 57, 3876-3881. KLOTZ, S. A., DRUTZ, D. J., HARRISON, J. L., AND HUPPERT, M. (1983). Adherence and penetration of vascular endothelium by Candida yeasts. infect. Immun. 42, 374-384. KLOTZ, S. A., AND MACA, R. D. (1988). Endothelial cell contraction increases Candida adherence to exposed extracellular matrix. Infect. Immun. 56, 2495-2498. ROTROSEN, D., CALDERONE, R. A., AND EDWARDS, J. E., JR. (1986). Adherence of Cundida species to host tissues and plastic surfaces. Rev. Infect. Dis. 8, 73-85. ROTROSEN, D., EDWARDS, J. E., JR., GIBSON, T. R., MOORE, J. C., COHEN, A. H., AND GREEN, I. (1985). Adherence of Cundida to cultured vascular endothelial cells: Mechanisms of attachment and endothelial cell penetration. J. Infect. Dis. 152, 1264-1274. SPRINGER, T. A. (1990). Adhesion receptors of the immune system. Nature 346, 425-434. SUNDSTROM, P. M., NICHOLS, E. J., AND KENNY, G. E. (1987). Antigenic differences between mannoproteins of germ tubes and blastospores of Candida albicans. Infect. Immun. 55, 616-620. THOMAS, P. D., HAMPSON, F. W., AND HUNNINGHAKE, G. W. (1988). Bacterial adherence to human endothelial cells. J. Appl. Physiol. 65, 1372-1376. TRONCHIN, G., BOUCHARA, J. P., ROBERT, R., AND SENET, J. M. (1988). Adherence of Candida albicans germ tubes to plastic: Ultrastructural and molecular studies of fibrillar adhesins. Infect. Immun. 56, 1987-1993.