JOURNAL OF CLINICAL MICROBIOLOGY, Nov. 1984, p. 877-8800095-1 137/84/110877-04$02.00/0Copyright © 1984. American Society for Microbiology

Analysis of Pneumocystis carinii Cysts with a Fluorescence-Activated Cell Sorter

CLAUDIA R. LIBERTIN,' GAYLE E. WOLOSCHAK,2 WALTER R. WILSON,1 AND THOMAS F. SMITH'*Division of Infectious Diseases1 and Departments of Cell Biology2 and Laboratory Medicine,3 Mayo Clinic and

Foundation, Rochester, Minnesota 55905

Received 19 April 1984/Accepted 27 July 1984

Human sera from Pneumocystis carinii-infected patients and specific rabbit antisera have antibodies againstthe cyst form of the organism. Lung tissue concentrations from cortisone-treated C3H/HeN mice and six openlung biopsy-positive patients were centrifuged and suspended, and immunofluorescent staining was done. Weutilized the fluorescence-activated cell sorter to analyze and sort P. carinii cysts from lung homogenates into a

morphologically distinct population. A quantitative basis was used for the definition of the cyst population bydisplaying the frequency of cells as a function of parameter (fluorescence intensity and light scatter) expression.In 14 of 15 histogram analyses, P. carinii-infected homogenates were differentiated from normal- and bacterial-control homogenates. The parameter range of light scatter (size) was 2 to 8 ,um, and the fluorescence intensitywas greater than a threshold based on the histogram profile.

The fluorescence-activated cell sorter (FACS) is a sophis-ticated, computerized laser illumination system that is usedboth for analysis and for separation of cells in suspension (1,4, 6). It rapidly measures the size and fluorescence intensityof individual cells as they flow in a stream past a laserillumination system coupled with a highly sensitive detector.The measurements, collected and displayed, create an ana-lytical profile of cell populations. Differential light scatterpatterns have been used to identify bacteria (17) and virus-infected cells (2). A wide range of applications includesmeasurement of the DNA content of individual chromo-somes for karyotyping (3) as well as the amount of surfaceimmunoglobulin on B cells (9).We utilized the FACS to analyze and sort P. carinii cysts

from lung homogenates into a morphologically distinct popu-lation. A quantitative basis was used for the definition of a

cyst population by displaying the frequency of cells as a

function of parameter (fluorescence intensity and light scat-ter) expression.

This technique may provide a method of determiningparasite counts in cell procedures as well as of isolating theorganism for life cycle studies.

(This work was presented at the 83rd Annual Meeting ofthe American Society of Microbiology, New Orleans, La.,1983, abstr. no. C95.)

MATERIALS AND METHODS

Experimental animals. C3H/HeN mice, weighing 90 to 120g each, were used in all experiments. The animals wererandomly assigned to control or treatment groups and main-

tained ad libitum on a low protein (8%) diet (ICN NutritionalBiochemical, Cleveland, Ohio). Cortisone acetate injectionswere given intraperitoneally (1 mg twice weekly) duringweeks 3 through 9. Tetracycline (1 mg/ml) in drinking waterwas given to all groups. This protocol has been shown byourselves and by others (15. 16) to induce P. ctriniii lunginfections. Mice were sacrificed by cervical dislocation atweek 9 and autopsied immediately. Each lung was removed,and impression smears were prepared on glass slides which

Corresponding author.

were stained with methenamine silver to assure the presenceof P. carinii cysts (5, 7).

Patients. Six patients at the Mayo Clinic during this timeperiod had P. carinii cysts on impression smears from openlung biopsy (OLB) specimens. Tissue samples and sera wereobtained from these individuals. In addition, six OLB pa-tients with only bacterial infections and six with negativemicrobiological tests (i.e., negative cultures for virus, fungi,Chlamydia spp., Mycoplasma spp., and other bacteria) were

used as control specimens.Preparation of lung specimens. Each lung tissue was ho-

mogenized in 2 ml of nutrient broth by blending for 2 min in a

stomacher (Tekmar Co., Cincinnati, Ohio) (12). The homog-enate was centrifuged for 20 min at 800 x g. The concentratewas then incubated for 45 min at room temperature with thepooled human sera from P. c ariniii-infected patients or

specific rabbit antisera against P. caciriniii obtained from

Linda Pifer, University of Tennessee, Memphis. Fluoresce-in-labeled goat anti-human or goat anti-rabbit immunoglob-ulin sera (Cappel Laboratories, Cochranville, Pa.) wereincubated with the concentrate for 45 min followed by twowashings with phosphate-buffered saline, pH 7.2. The sus-

pension was immediately analyzed with the FACS.Reagents. Sera were obtained from patients with OLB-

confirmed P. carinii infection and from normal controlhumans. To determine the presence and titer of antibody toP. carinii in sera, an indirect immunofluorescence test wasdone. Touch impression smears of P. carinii were overlaidwith dilutions of the sera to be tested. After incubation atroom temperature for 30 min in a humid chamber andwashings with phosphate buffered saline to remove non-

bound antibodies, the smears were stained with fluorescein-labeled anti-human immunoglobulin G globulin (Cappel Lab-oratories). The titer of antibody in the serum was establishedas the reciprocal of the highest dilution which resulted in atleast 2+ fluorescence of P. carinii cysts. The average titer ofsera from OLB patients with P. carinii was 1:32. and that ofsera from normal control humans was 1:8. All of the six

normal human control sera demonstrated anti-P. cariniiantibodies. Pifer et al. (11) and Norman and Kagan (10)similarly demonstrated the presence of anti-P. carinii anti-bodies in normal human sera.

Fluorescein-conjugated reagents were stored at 4°C under

877

Vol. 20, No. 5

on October 30, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

878 LIBERTIN ET AL.

sterile conditions to prevent formation of aggregated materi-als. Just before staining, all reagents were centrifuged at100,000 x g for 20 min to remove any remaining aggregatedmaterial that could interfere with analysis.

Instrument. All samples were analyzed under the same

conditions by a FACS IV cell sorter (Becton Dickinson andCo., Paramus, N.J.). The homogenate of cells was forcedunder pressure through a micronozzle, in which the cellswere centered in the effluent jet by an outer coaxial flow ofcell-free fluid. The stream, which was ca. 50 ,um in diameter,was illuminated by an argon-ion laser (model 164; Spectra-Physics, Palo Alto, Calif.) immediately after leaving thenozzle. The laser beam was focused on the stream with a

spherical lens having a focal length of 124 mm. The noise inthe fluorescence channel was decreased with a dual-channelmode of operation. The fluorescence channel was activatedonly when a scatter signal indicated that the cell was present(scatter-gated fluorescence). The reciprocal operation, fluo-rescence-gated scatter, was performed so that the scattersignal was not accepted unless the fluorescence pulse was

greater than a specified threshold of intensity.Light for the scattering channel was collected at a forward

angle of 0.5 to 12% from the incident beam. The fluorescencedetector was a photomultiplier tube. The filters used in thefluorescence channel for fluorescein emission included twocut-on filters (520.0 and 530.0 nm; series D; Ditric Optics,Inc., Marlboro, Mass.). Fluorescein was excited by the 488-nm line of the laser, which was operated at a 400-mW output.Cells were analyzed at a rate of 500 to 1,000/s.

Before each experiment, the FACS was adjusted to givethe same scatter and fluorescence profiles with a standardpreparation of glutaraldehyde-fixed chicken erythrocytes.Parallel analyses of normal, bacterial-infected, and P. car-

iiii-infected lung homogenates were done. For each homog-enate, dot plots and histograms were obtained and comparedwith one another. Sorted cells were placed on microscopeslides, fixed in methanol, and stained with methenaminesilver stain for cytological examination.

RESULTSVarious reagents were reacted and assayed by indirect

immunofluorescence techniques against five different lunghomogenates to confirm antibody specificity (Table 1). Serafrom P. carinii-infected patients and rabbit anti-P. cariniisera revealed significant immunofluorescence within thelung tissue. When stained with methenamine silver, theimmunofluorescent organisms were recognized as typical P.carinii cysts. Pooled normal human sera demonstrated anti-P. carinii antibodies but at lower titers (1:8) than theseantibodies were demonstrated in sera from P. (arinii-infect-ed patients (1:32) or in rabbit anti-P. ((crinii sera.

This serological reactivity in normal humans has beendemonstrated by others (10, 11, 13) and is consistent with theomnipresent nature of this opportunistic pathogen.The negative controls of normal mouse and normal human

lungs failed to show immunofluorescence of the lung tissue.Bacteria-infected human lung homogenates were also usedas a control to show that staining was specific and notsecondary to an inflammatory cell response. These bacteria-infected lungs failed to demonstrate immunofluorescence in

all experiments, indicating that the immunofluorescence inP. carinii-infected tissue was not due to other opportunisticpathogens, bacteria, or a nonspecific host reaction.Therefore, our ability to specifically tag P. cariniii cysts by

specific fluorescent antibodies enabled us to separate theseorganisms into a distinct population of cells by the FACS.

TABLE 1. Results of treatment of lung homogenates with variousreagents and assay of these treated homogenates by indirect

immunofluorescence

No. of positive samples/no. tested when treatedwith:

Homogenate Phosphate- Sera from Pooled Rabbitbuffered P. corinii- normal p. carinsalinere infected human antisera'

humans" sera a

Normal mouse 0/9 0/9 0/4 0/4lung

P. carinii-infected 0/9 9/9 4/4 4/4mouse lung

Normal human 0/6 0/6 0/6 0/6lung

Bacteria-infected 0/6 0/6 0/6 0/6human lung

P. carinii-infected 0/6 6/6 6/6 6/6human lung

" Sera obtained from six patients with P. carinii pneumonia diagnosed byOLB. Titer was 1:32.

Sera obtained from normal control patients. Titer was 1:8.Antisera obtained from L. L. Pifer, University of Tennessee. Memphis.

To show that the FACS was capable of identifying asubpopulation of cells representing P. (arinii in mouse andhuman lung homogenates, homogenates treated with thevarious reagents listed in Table 1 were analyzed, and datawere presented in two manners, the dot plot and compositehistograms.Dot plots, representative of the light scatter parameters

and fluorescence intensity measurement, were taken foreach cell in each homogenate after 106 cells were analyzed(Fig. 1). Each dot plot represents a cell at a given level offluorescence and light scatter (size). The upper left one-fourth of the dot plot represents an area of highly fluorescentsmall cells. The lower one-third of the dot plot representslow-fluorescent background materials, which are not P.(carinii cysts. The blanked area represents the location of ascatter (size) threshold (gate) which allowed for analysis ofcells larger than 2 VLm. The dot plots of normal mouse (datanot shown) and normal human lung homogenates (Fig. IA)showed no specific distinctive population of cells that couldbe identified. Similarly, the dot plots of bacteria-infectedlung homogenates did not reveal such a population of highlyfluorescent cells (Fig. IB). However, the dot plots of P.(arinii-infected human lung specimens showed a populationof cells that were brightly fluorescent and small (Fig. 1C). AllP. (curinii-infected mouse lung homogenates demonstrateddot plots with a detectable population of fluorescing cells.These dot plots provided a semiquantitative indicator of apopulation density present in P. (aruinii-infected homoge-nates.By combining dot plots from designated homogenates, we

obtained composite histograms. Histograms display the dis-tribution of cells as a function of linear increase in parameter(size and fluorescence) for each sample. A composite histo-gram is comprised of two linear graphs in each diagramrepresenting a given homogenate. In each histogram, thegraph line farthest to the left is the normal human homoge-nate. If no area between the two graph lines exists or thegraph lines are superimposed on one another, no distinctpopulation of cells exists between the compared homoge-nates. However, if a histogram reveals a significant areabetween the two graph lines, then a distinct subpopulation of

J. CLIN. MICROBIOL.

on October 30, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

ANALYSIS OF P. CARINII CYSTS 879

(0)

av0-

c0a)

(0

0

Light scatter (size) >00

FIG. 1. Representative dot plots of fluorescence intensity (verti-

cal axis) and light scatter or size (horizontal axis) from lung

homogenates after 106 cells were analyzed by a FACS IV. (A) Dot

plot of a normal human lung homogenate stained with sera from P.

carinii-infected patients and fluorescein-labeled goat anti-human

immunoglobulin. (B) Dot plot of a bacteria-infected human lung

homogenate stained as above. (C) Dot plot of P. c-arinii-infected

human lung homogenate stained as above. The blanked area repre-

sents the location of a scatter (size) threshold (gate) which isolates

the population of cells greater than 2 p.m in size. The upper left one-

fourth of the dot plots represents an area of highly fluorescent cellsthat were small. Only panel C demonstrates cells in this area.

cells exists in that homogenate which is not present in the

control. All fluorescence histograms with the FACS were

scatter gated. Fluorescence-gated scatter was performed so

that the scatter signal was not accepted unless the fluores-cence pulse was greater than a specified threshold of intensi-ty. Cells were analyzed at a rate of ca. 500 cells per s.

Composite fluorescence histograms of normal human ho-

mogenates were obtained and compared with one another.

No difference among the graphs, which would represent a

cell population unique to a given homogenate, was found

with each analysis (Fig. 2A). Likewise, normal and bacteria-

infected lung homogenates, when analyzed by compositehistograms, demonstrated no significant differences in

graphs (Fig. 2B). These data are consistent with their

respective dot plots and demonstrated a population of cellsdevoid of P. carinii cysts.Composite fluorescence histograms of P. carinii-infected

human homogenates revealed significant differences whencompared with those of normal human homogenates in allbut 1 of 15 instances (P < 0.01; range, 95%; confidence level,68.1 to 99.8%). This exception was a patient who hadundergone OLB in which routine methenamine silver stain-ing of touch preparations of the tissue failed to show thepresence of P. carinii cysts. However, utilization of a tissueconcentration method (15) revealed the presence of a feworganisms in 15 fields examined at x 200 magnification.Many other variables including the quantity of organisms perhomogenate could have contributed to failure to identify thissmall population of cysts in the homogenate. Nevertheless,composite histograms of lung homogenates showed a dis-tinct cell population of P. carinii cysts in infected samplesand not in normal or bacteria-infected homogenates. Acomparison of fluorescence histograms of a normal humanlung homogenate and of a P. carinii-infected human homog-enate is shown in Fig. 2C. The difference between the twocomposite graphs represents the population of fluorescentcells, P. carinii cysts, that are unique to the P. carinii-infected homogenate. With the scatter-gate set, this popula-tion of cells was reanalyzed to obtain a second histogram onthe vertical axis showing the fluorescence distribution ofcells within the size range of 2 to 8 p.m.

Consequently, these fluorescence histograms provided aquantitative basis for the definition of the population of cystsby displaying the frequency of cells in the population as afunction of parameter expression. Last, the histogramsallowed visualization of cell clustering within our restrictedparameter range and light-scattering size of 2 to 8 p.m, andfluorescence intensity was greater than a threshold band onthe histogram profile. When cells within this restrictedparameter range were sorted and stained with methenaminesilver, cytological findings were consistent for P. carindicysts.

DISCUSSIONThis investigation demonstrated that the FACS was capa-

ble of analyzing and sorting P. cariniii cysts from lunghomogenates of infected mice and humans. A quantitativebasis was used for the definition of the cyst population bydisplaying the frequency of cells as a function of parameter(fluorescence intensity and light scatter) expression. Theparameter range of light scatter (size) was 2 to 8 p.m, and thefluorescence intensity was greater than a threshold based ona histogram profile.

Application of FACS technology to P. carinii researchwould be beneficial. Analysis and sorting of cysts from lunghomogenates would provide a means of separation based onimmunofluorescent staining but would not interfere withtheir biological function. Homogenate samples in microlitervolumes potentially could be analyzed at rates to 5,000 cellsper s, at purities to 99%, and at recoveries to 90% (6). Thismethod of single-cell analysis offers an expedient, quantita-tive measurement of individual cells such as P. carinii cystswithin cell populations and has very high statistical preci-sion. For example, FACS methodology would provide ameans of determining parasite counts in cell culture proce-dures.With the FACS system, it is also possible to determine the

frequency distribution of cells of different sizes, i.e., tropho-zoites (2 to 4 p.m) and cysts (2 to 8 pLm). In addition, twofluorochromes, i.e., fluorescein and rhodamine, may be

VOL. 20, 1984

on October 30, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

880 LIBERTIN ET AL.

Ligh scte sze ---

c01)

oC

FIG. 2. Representative composite fluorescence histograms derived from dot plots in Fig. 1. Histograms display the distribution of cells asa function of linear increase in parameter (size and fluorescence). In each histogram, the graph line farthest to the left is the normal human ho-mogenate. (A) Comparison cof normal human sera versus normal human lung homogenates. (B) Comparison of normal human versus bacteria-infected human lung homogenates. (C) Comparison of normal human versus P. (carinii-infected human lung homogenates. The area betweenthe graphs represents the distinct cell population unique to P. lcarinii-infected homogenates. Fluorescein was excited by the 488-nm line of thelaser of FACS IV, which was operated at 400 mW. Scatter signal was not accepted unless fluorescence pulse was greater than a threshold ofintensity based on the histogram profile. Gain was 800.

detected simultaneously at different wavelengths with thesame cell. Since the trophozoites stain with auramine-rhodamine (14) and cysts stain with fluorescein, the FACScould potentially analyze and sort cysts containing thetrophozoites for life cycle studies.

In any field of science, a powerful new method permits anew series of questions to be asked and answered. Similarly,the powerful tool of the FACS system may assist in parasi-tology research.

ACKNOWLEDGMENTS

We thank L. L. Pifer for the generous gift of rabbit anti-P. cariniisera, James Tarara and Bridget Tierney for FACS technical assist-ance, and Roberta Kondert for aid in preparing the manuscript.

LITERATURE CITED

1. Bonner, W. A., H. R. Hulelt, R. G. Sweet, and L. A. Herzenberg.1972. Fluorescence activated cell sorting. Rev. Sci. Instrum.43:404-424.

2. Cram, L. S., and A. Brunsting. 1973. Fluorescence and lightscattering measurements on hog cholera-infected PK-15 cells.Exp. Cell Res. 78:209-213.

3. Gray, J. W., A. V. Carrano, L. L. Steinmetz, M. A. Dilla, D. H.Moore, B. H. Mayall, and M. L. Mandelsohn. 1975. Chromo-some measurement and sorting by flow systems. Proc. Natl.Acad. Sci. U.S.A. 72:1231-1240.

4. Herzenberg, L. A., and R. G. Sweet. 1976. Fluorescence activat-ed cell sorting: a new tool for isolating functional cell types. Sci.Are. 234:108-114.

5. Hughes, W. T. 1975. Current status of laboratory diagnosis ofPneumocystic carinii pneumonia. Crit. Rev. Clin. Lab. Sci.6:145-170.

6. Loken, M. R., and L. A. Herzenberg. 1975. Analysis of cell

7. Mahan, C. T., and G. E. Sale. 1978. Rapid methenamine silverstain for Penu,nocvstis and fungi. Arch. Pathol. Lab. Med.102:351-352.

8. Meuwissen, H. E. T., Jr., I. Tauber, A. D. E. M. Leeuwenberg,P. J. A. Beckers, and M. Sieben. 1977. Parasitologic andserologic observations of infection with Pneutnocshtis in hu-mans. J. Infect. Dis. 136:43-49.

9. Meyer, R. A., S. F. Haase, S. E. Podulso, and G. M. McKhann.1974. Light scattering patterns of isolated oligodendroglia. J.Histochem. Cytochem. 22:594-598.

10. Norman, L., and I. G. Kagan. 1973. Some observations on theserology of Pneutnocystis carinii infections in the United States.Infect. Immun. 8:317-321.

11. Pifer, L. L., W. T. Hughes, S. Stagno, and D. Woods. 1978.Pneumocvstis carinii infection: evidence for high prevalence innormal and immunosuppressed children. Pediatrics 61:35-41.

12. Sharpe, A. N., and A. K. Johnson. 1972. Stomaching: a newconcept in bacteriological sample preparation. Appl. Microbiol.24:175-178.

13. Shepherd, V., B. Jameson, and G. K. Knowles. 1979. Pneumo-cystis carinii pneumonitis: a serologic study. Clin. Pathol.32:773-777.

14. Thomson, R. B., Jr., and T. F. Smith. 1982. Acridine orangestaining of Pneumocystis carinii. J. Clin. Microbiol. 16:191-192.

15. Thomson, R. B., Jr., T. F. Smith, and W. R. Wilson. 1982.Comparison of two methods used to prepare smears of mouselung tissue for detection of Pneumocystis carinii. J. Clin.Microbiol. 16:303-306.

16. Walzer, P. D., R. D. Powell, Jr., and K. Yoneda. 1979. Experi-mental Pneumocystis carinii pneumonia in different strains ofcortisonized mice. Infect. Immun. 24:939-947.

17. Wyatt, P. J. 1968. Differential light scattering: a physicalmethod for identifying living bacterial cells. Appl. Opt. 7:1879-1883.

J. CLIN. MICROBIOL.

on October 30, 2020 by guest

http://jcm.asm

.org/D

ownloaded from