Flow cytometric measurement of intracellular pH in B16 tumors: Intercell variance and effects of pretreatment with glucose

Experimental

Flow Cytometric Tumors:

Cell Research 180 (1989) 106116

Measurement of Intracellular pH in B16 Intercell Variance and Effects of Pretreatment with Glucose

DAVID W. HEDLEY’

and HELLE

B. JORGENSEN

Ludwig Institute for Cancer Research {Sydney Branch}, Rlackbur~ Building, U~i~rsi~ Sydney, N.S. W. 2006, Australia

o~Sydn~,

Flow cytometry was used to measure cytoplasmic pH (pHJ of B16 melanoma cells taken from tumor-bearing animals. We used a ratiometric method to allow measurements on an individual cell basis which were independent of cellular content of the pH indicator BCECF. In order to “freeze” any intercell variance which may have existed within the tumor mass, tumors were mech~cally d&aggregated in bicarbonate-free medium containing 0.5 m&f amiloride at 4°C and loaded with BCECF in choline chloride-based Earle’s solution at 37°C. Studies using cells grown in vitro showed that this protocol prevented acid load recovery during the 30-min period typically required between tumor excision and pHi measurement. A calibration curve was obtained by resuspending BCECF-stained cells in a range of buffers containing the proton ionophore nigericin. The range of values for individual cells was estimated by comparing the coefficient of variation of the test sample with that obtained when nigericin was used to reduce all ceils to the PHi of the calibration buffer. The average value for mean tumor cell pH was 7.321-0.05 SD. Pretreatment of animals with intraperitoneal glucose for one hour resulted in an average for mean pHi of 7.17kO.17 SD. Mean coefficient of variation was 8.7%, and in the presence of nigericin, 8.1 4ro.These values indicate a variance in measured pHi of approximately kO.4 pH units, but most of this results from experimental error rather than true intercell pHi variance. The method used here is capable of detecting reduction in mean tumour pHi caused by ip glucose, but incapable of precise estimation of indi~du~ cell values. Despite these uncertainties, the results suggest that the range of pHi within B 16 tumors is small. @ 1989 Academic Press, Inc.

Solid tumors contain microregions where prol~er~tion is arrested either because of nutrient deprivation or because of failure of toxic catabolic products to diffuse out of the tumor mass via an inadequate neovasculature El]. An essential nutrient which has received particular attention is oxygen, and hypoxia is probably a significant cause of the increased lactic acid production observed in some tumors. There is an extensive literature reporting that solid tumors are in general more acidic than normal tissues [Z], while more recent studies using 31Ptopical NMR spectroscopy suggest that intracellular as well as extracellular pH may be reduced [3-51. Intracellular pH (pHi) can be regulated by HCO;/Cl- or Na+/H+ exchange across the plasma membrane via specific antiports [6, 71. The latter is better characterized, and possibly of greater significance in maintenance of pHi in solid tumors, since Na+/H+ antiport deficient mutants are nontumo~genic 181. The Naf/KC ATPase maintains a sodium ion gradient of approximately one log ’ To whom reprint requests should be addressed. Copyright @ 1989 by Academic Press, Inc. AU rights of reproduction in any form resened 0014-4827189s;o3.00

106

Flow cytometric measurement of intracellular p

107

across the plasma membrane, and this potential can be used to extrude proto by one-to-one exchange with Naf ions. It should be noted that when the inward H’ gradient exceeds the outward Na+ gradient the antiport can reverse direction, leading to cytoplasmic acidification 191.Previous work from this laboratory shown that cell cycle progression can be regulated by extracellular pH [ During growth to plateau phase, fed, static cultures of PMC 22 human melanoma cells entered a distinct pH-sensitive quiescent state with arrest in CJ,. Entry into this state was determined by the fall in extracellular pH which occurs natura with increasing cell density, and cells reentered the cell cycle ap~roximateI~ after resuspension at lower concentrations in medium adjusted to pH=7.2. the period prior to entry into S-phase RNA synthesis continued, suggestin e presence of a specific pH-sensitive restriction point in G1, rather than nonspecific arrest of growth. A flow cytometric method was then developed measuring intracellular pH on an individual cell basis 11I], and it was shown t cytoplasmic acidification occurred at the external pH values associated with transition into a pa-dependent quiescent state 1121. Because microregions of markedly reduced tissue pH exist within some solid tumors, the evidence discussed above suggests that they might contai which were quiescent as a direct consequence of inability to maintain nor against an inwardly directed [H+] gradient. We therefore adapted t cytometric method developed for use with tissue culture lines to the i~vestigat~~~ of tumors grown in ho. The tumor selected was B16 mouse me1 earlier work had shown that these cells have an extracellular cycle in vitro [lo], and grow as a classical corded structure in UEUO, w distant from capillaries show failure to incorporate the NA precursor ~r~rn~deoxyuridine [ 131.

MATERIALS

AND METHODS

Tumors. B16 melanoma cells were injected intradermally into the flanks of C57Bl mice, and tumors were used after approximately 3 weeks growth. At this stage their greatest diameter was about 1 cm, and routine histological sections showed the presence of classical vascular cords without gross areas of necrosis. Previous work in this laboratory showed that a 30-min pretreatment of tumor-bearing animals with the thymidine analog bromodeoxyuridine (BrUdR) selectively labeled nuclei of cells lying adjacent to blood capillaries, with reduced labeling of cells lying three to four diameters further out 1131.In order to test for the presence of quiescent or slowly cycling cells, BrUdR was constantly infused into tumor-bearing animals using subcutaneously implanted miniosmotic pumps (Alza Corp, Palo Alto, CA) for 48 h and the tumors were then excised, disaggregated, fixed, and stained for DNA content and incorporated BrUdR, as previously described 1131.Simultaneous Bow cytomtric measurement of DNA versus BrUdR was then performed using a FACS Analyzer (Becton Dickinson, Mountain View, CA). Sample preparation. Animals were killed by cervical dislocation, and the tumor was excised, disaggregated, and stained using the pa-sensitive fluorochrome 2’,7’-bis(carboxyethyl)-5,6-carboxyfluorescein acetoxymethyl ester (BCECF-AM, Molecular Probes, Eugene, OR). Because acidified cells are capable of rapid pHi recovery via Na+/H+ or HCO;/Cll exchange, a variety of protocols was tested, using procedures designed to limit the acid load recovery of any acidotic cells present within the tumor. These procedures comprised the use of nominally bicarbonate-free medium to prevent HCO;/Cll exchange and the use of the Earle’s solution where choline ions had been substituted for sodium, to prevent Na+/H+s exchange. In addition, the Na+/H+ antiport blocker amiloride (Sigma, St. Louis, MO) was added to the medium, and the effect of temperature was investigated. All of these

108 Hedley and Jorgensen procedures were tested by first loading cells grown in tissue culture with BCECF, washing and resuspending in Earle’s solution containing 10 m&4 NH&l for 30 min at 37°C. The ammonium ion enters the cytoplasm via monovalent cation channels, and is in equilibrium with NH,+H+. Resuspension in ammonium-free medium allows rapid diffusion of ammonia through the membrane, trapping H+ in the cells and resulting in a fall in pHi of up to 1 pH unit [14]. The rate of pHi recovery was monitored using flow cytometry as described below. Measurement of intracellular pH. This was done using a previously described method which measures the pH-dependent emission of the fluorescein derivative BCECF [ll]. The green fluorescence of this stain is highly pH-dependent, while the red fluorescence is only weakly so. The ratio of two fluorescences therefore gives an estimate of pHi which is independent of cell dye content. Cells were suspended in a suitable buffer containing a 10 p&f concentration of the membrane permeable acetoxymethyl ester, which is hydrolyzed by intracellular esterases. A fluorescence-activated cell sorter (FACS III, Becton Dickinson) was used to measure fluorescence of BCECF-loaded cells. Excitation was at 488 nm, 400 mW using an argon ion laser, and the ratio of fluorescences at 520 and 620 nm was determined as previously described. The machine was triggered using forward light scatter, which is a measure of particle size. Because disaggregated tumors contain a substantial amount of debris, some of which is weakly fluorescent, 8 urn diameter calibration beads were run prior to the tumor sample, and a threshold was set so that only particles larger than this were measured by the fluorescence detectors. Mean Coulter volume of B16 cells grown in vitro was determined as 2900 urn’ using the FACS Analyser. Approximately 20,000 cells were measured, and the results were displayed as a fluorescence ratio histogram. Mean fluorescence ratio and coefficient of variation were determined using a computer program. A calibration curve was then prepared by suspending cells in high [K+] buffers (pH 6.4-7.6) containing a 10 u&f final concentration of the proton ionophore nigericin (Sigma). When extra- and intracellular concentrations of K+ are approximately equal, nigericin allows equilibration of intracellular pH to the external pH.

RESULTS Demonstration of Slowly Cycling Cells in B16 Tumors Tumor-bearing animals were infused with BrUdR for 48 h and the tumors were then excised, disaggregated, fixed, and stained using a monoclonal antibody to BrUdR, incorporated into DNA, and propidium iodide to measure total cell DNA content. Figure 1 shows a single-parameter flow cytometric analysis of cellular DNA content, and a two-parameter contour plot of log BrUdR content versus DNA content. Single-parameter DNA analysis showed a minor G1 peak corresponding to normal diploid host cells and a larger G1 peak of near tetraploid tumor cells. Computed cell cycle phase distribution of this population is shown. By measuring BrUdR content simultaneously it was shown that the majority of diploid cells had failed to incorporate DNA precursor, while the majority of tumor cells showed positive BrUdR staining. There was however a population of tumor cells which were BrUdR negative or only weakly stained, and this population showed a predominantly G1 distribution. Inhibition of Acid Load Recovery Before examining pHi of cells from B16 tumors, a sample preparation protocol was developed to minimize the acid load recovery of any acidotic cells which might be present within the tumors. BCECF-stained cells were acid loaded by NH&l pretreatment and resuspended in bicarbonate-free Earle’s solution, and pHi was measured at intervals using flow cytometry. The following observations were made:

Flow cytometric measurement of i~tracel~~~a~p

DNA content 2000

-c

1

0

%G 1 %s %G2+M C.V.

I

50

100

150

200

: = =

:g 714 4.4%

250

DNA content Fig. I. FIow cytometric analysis of DNA content of B 16 tumors following 48 h of infusion with BrUdR. Single parameter DNA analysis in lower half shows a small peak (left-most) corresponding to normal diploid Gi cells, while the major peak, which is approximately tetraploid, comprises Gi B16 melanoma cells. Cell cycle phase distribution of these cells is shown. Simultaneous measurement of BrUdR (log scale) shows that most diploid cells are negative while most tumor cells have incorporated BrUdR. Note that some, predominantly G,, tumor cells remain BrUdR negative.

(a) Eflect of temperature. Acid-loaded cells kept at room temperature sho a steady increase in pHi to a plateau value after 20-30 min. Tbe rate of p recovery was considerably reduced at 4°C (Fig. 2A), whereas cells incubated 37°C had returned to control pHi by the time the l- min observation was made min is the minimum time required to make a recording using flow cytometry). (b) Effect of amiloride. The Na+/H+ antiport blocking the drug amiloride (0.5 or 2.0 rnM) was added to the Earie’s solution, and the effects on pHi recovery at room temperature were monitored. As shown in Fig. 28, amiloride reduced but did not prevent the rate of acid load recovery. At 37°C acid load recovery was rapid, with the higher concentration of amiloride then leading to a s fall at a in IpHi. The combination of 0.5 mM amiloride and 4°C was able to plateau for at least 50 min (Fig. 2 C), and this was considered satisfactory. Hn a

110 Hedlq and Jorgensen llO-

llO-

B

A

loo-

loo-

0 0

0

QO-

F

2

0

5

.

0

QO-

80-

0

0

20

30

0

I 10

I 40

1 50

0

l

QO-

i

80-Ol o

0

0

0

o

I 20

, 30

4'0

I 30

l oorn

0

I 40

4 50

l

80- * 0

I 10

I 20

D

QO-

0 0

I 10

loo-

l

l

’ 0

0

0

llO-

C l

0 0

0 0

80-E

O

llO-

.

l 0

5: loo-

3

0

0

o"

E

'

0

0

, 50

0

O '0

O 0 20

0

0

I 30

40

I 50

MINUTES Fig. 2. Acid load recovery of NH&l-treated cells. Controls (0) were resuspended in HCO;-free Earle’s solution at room temperature. Test samples (0) were resuspended in Earle’s solution at 4°C (A), in Earle’s solution containing 0.5 m&f amiloride at room temperature (B), in Earle’s solution containing 0.5 m&I amiloride at 4°C (0, and in choline chloride-based Earle’s solution at 37°C (0). Fluorescence ratio is an arbitrary scale (channel numbers) which is directly proportional to pHi over the range of values shown here, 10 channel numbers corresponding to a difference of approximately 0.5 pH units.

parallel experiment using non-acid-loaded cells, fluorescence ratio was not affected by 0.5 rnM amiloride for up to 60 min. (c) Effect ofsodium substitution. Hydrolysis of BCECF acetoxymethyl ester is temperature dependent and occurs extremely slowly at the temperature range where amiloride is capable of clamping pHi. Substitution of choline chloride for sodium chloride prevents H+ extrusion via the Na+/H+ antiport, and was effective in preventing acid load recovery at 37°C (Fig. 20). Two potential problems using a sodium-free buffer are first the possibility that the small amount of sodium ion in mammalian cytoplasm might allow the Na+/H+ antiport to run in reverse, leading to cytoplasmic acidification, and second that a number of other Na+dependent membrane transport systems would be affected. Although it proved impossible to dispense with the use of choline chloride-based Earle’s solution during BCECF-loading, it was found that using 10 uM BCECF the period of incubation at 37°C could be limited to 5 min without loss of resolution. With incubation periods of less than 5 min the fluorescence ratio was unstable,

Flow cytometric measurement of intra~el~~l~r p

111

possibly because of a relative abundance of nonhydrolyzed BCECF-A ich has different spectral properties. d) Effect of bicarbonate. As expected, the addition of 10 solution allowed rapid pHi recovery at room temperature despite the amiloride or choline chloride (not shown). This was presumably via exchange, because it was inhibited by the specific blocker DIDS at 0.25 (e) Definitive protocol for sample preparation. Based on the above Q tions a standard protocol was used for handling all tumors. The tumor was ved from the mouse and immersed in Na+-base &&e’s solution wit amiloride in 5 cm diameter petri dishes placed o ice. A single-cell suspension was made with as little time delay as possible using crossed scalpel and filtered through gauze, and the cells were then counted, centrifuged, resusviable cells in 1 ml choline chloride-based Earle’s solution containCECF acetoxymethyl ester, and incubated at 37°C for 5 min. 0 ul were then spun down, aspirated, and he1 est sample was resuspended in Na’-based Earle’s soluti run through the flow cytometer. The instrum using forward light scatter, the threshold was set at the value calibration beads. Data were recorded as a ratio of 5201620n versus foward light scatter two-parameter contour plots. IJsing time interval between tumor excision and fluorescence ratio proximately 30 min. ibration curve was then prepared by resuspending aliquots of ells in 1 ml K+-based Earle’s solution at pH 6.4, 6.6, 6.8, 7.0, and 7.6, with nigericin at 4°C. The pH of these calibration buffers was checke daily. easurement of Intracellular

pH in B16 Tumors

Figure 3 shows a typical series of measurements of a tumor sample. Fluorescence ratio is recorded as channel numbers which are arbitrary units. Results are displayed as a correlated two-parameter plot of forward light scatter versus fluorescence ratio (top left-hand panel) and as a series of single-parameter orescence ratio histograms, starting with the test sample and then progres through the calibration series. Mean pHi of the test sample was determined usi a calculator to plot a linear regression through the calibration points. In all ca this yielded a correlation coefficient rM.95. Average value for mean nHi was 7.3250.05 SD, n=5. The histograms illustrated in Fig. 3 are made up of fluorescence ratios obtai for approximately 20,000 cells and show a range of values. Some “cells” a lowest limit of forward light scatter gave extremely low ratio values, but probably debris. Otherwise there was little relationship between fluorescent ratio size. The range of values for fluorescent ratio was calculated as toe cient of variation (cv): cv = Half height peak width x 100 Mean channel number 2.354’ 8-898331

112 Hedley and Jorgensen *-0 N CI I

64

“I

fg?$;:i::. ::: ::.

L: 7.6

scatter

CHANNEL

NUMBER

Fig. 3. Typical series of measmements for a Bl6 tumor. Top left panel shows a correlated twoparameter plot of pHi (ratio) versus forward light scatter, threshold set at light scatter of 8 urn calibration beads, Middle top panel shows single-parameter fluorescence ratio histogram of BI6 tumor, and the remaining panels show the series of calibration samples, each labeled with pH of suspension buffer, with mean channel number shown at the top right of the peaks. Note that the B 16 tumor test sample shows a slightly broader peak than the calibration series, indicating a true variance in pHi.

Mean cv was 9.56 %, a value approximately twice that previously observed by us using cell lines. This implies a spread of measured pHi values of the order of mean kO.4 pH units, but the addition of nigericin to reduce all cells to a pHi of the calibration buffer produced a mean cv of 9.20 %, which was almost as large as the test sample. This indicates that most of the range of pHi values observed within tumor cell populations was artifactual, and that any true intercell variance in pHi is small. Effects O~GEUCOSE on pHi

~$I316

TI.LMO~T

We were concerned that our inability to detect any marked inter-cell variation in pH< was due to its loss during sample preparation. In an attempt to test this, tumor-bearing animals were pretreated with 0.5 ml 50 % glucose intraperitoneally

6.9

Fig. 4. Mean tumor pHi values for five B I6 tumors (left) and for five tumors taken from animals pretreated with 50% glucose, 0.5 ml intraperitoneally (right).

rior to sacrifice. It has previously been shown that glucose lowers tumor by increasing lactic acid production [ 151, while recent 31PN a indicated a broadening in pHi values within RIF-1 tumors [I&]. Effects Qse on mean -@Ii values are shown in Fig. 4. The average for five tumors was 7.1723.17 the variation between tumors being broader than that seen with animals not g glucose. The mean cv for glucose-pretreated tumors was 7.73 and 6.92 % wit ition of nigericin. Although this suggests that the true intercell variance i was greater following glucose pretreatment, this was still small when compared to the variance due to observational error. DISCUSSION ith its potential for detecting cellular heterogeneity and for viable cell sorting tumors own in flow cytometric analysis of cytoplasmic pH in cells taken uivo could be an extremely useful approach to studying ts of sol tumor biology. Our results using B16 tumors suggest however that the range Q-Sp within tumors is small, possibly no more than zk0.1 pH unit. The fact that we were able to show a fall in mean pHi following ip glucose indicates that the e preparation protocol was at least partially successful in ecwery by cells with low in situ pHi. As pointed out by Shapiro, apart from DNA content PIIi appears to be t cell property currently measured using flow cytometry which gives a very narr frequency-distribution histogram [ 171. Considering the importance of [II+

114 Hedley and Jorgensen conformation of many charged macromolecules this is hardly surprising. Using cell lines grown in tissue culture it was observed that fluorescence ratio peaks were narrower in the presence of proton ionophore, implying a range of pHi values in vitro [ 111. Subsequent sorting experiments indicated that pHi was cell cycle related, being somewhat higher in later phases compared to cells in Gi [12]. As cells in suspension culture grew to an environmental pH-determined quiescent state mean pHi fell, but contrary to our expectation the range of values did not widen, and individual quiescent cells could not be distinguished from those which continued to synthesize DNA on the basis of reduced pHi. Many studies of solid tumors using pa-sensitive microelectrodes have reported reduced values compared to normal subcutaneous tissues (see Ref. [2] for review). It therefore seemed plausible that in some areas tissue pH might fall to levels where cells producing excess hydrogen ions would no longer be able to extrude them via the Na+/H+ antiport, with consequent cytoplasmic acidification. Because of the rapidity with which cells recover from an acid load in vitro, a great deal of preliminary work was done studying methods for preventing this form occurring in any acidotic cells recovered from solid tumor masses during preparation for flow cytometry. Although removal of bicarbonate from solutions effectively prevents acid load recovery with HCO;/Cl- exchange without disturbing other cell processes, inhibition of Na+/H+ exchange by the use of choline chloride-based medium is likely to influence a number of other sodium iondependent membrane transport processes, while intercell variation in Naf concentration could influence the distribution of measured pHi by reversal of the Na+/H+ antiport in Naf-free medium. Amiloride is known to affect Na+/Ca’+ exchange and probably has other unwanted actions [ 181.Highly potent amiloride analogs have been described which are more specific to the Na+/H’ antiport, and if available would probably be well suited to sample preparation [18]. The major problem was, however, the fact that either staining or machine artifacts gave a spurious variance in intercell pHi measurements which was of sufficient magnitude to obscure any true variance in pHi. In common with all measurements, flow cytometry introduces a degree of uncertainty which, when the distribution of measurements is roughly gaussian, can be expressed as a coefficient of variation (cv). Our previously reported studies of pHi of exponentially growing cells in vitro showed a cv of approximately 4.5% failing by 0.5 % on addition of proton ionophore [12]. In the present series using B16 tumors, prepared so as to prevent acid load recovery and stained and run at pH 7.2, the mean cv of the fluorescence ratio was 8.7%, while for the calibration sample at the same pH but with nigericin added it was 8.1%. The observation that the fluorescence ratio was unrelated to forward light scatter suggests that the variance in measured pHi was not related to the range of cell size distribution. A cytochemical stain for melanin however showed that a variable proportion of B 16 cells contained this ligh,t-absorbing pigment, which might have adversely affected the spectral performance of BCECF. Finally, it should perhaps be noted that we were working close to the limits of machine performance, and that even when measuring an invariant cell property such as Gi DNA content, cv’s of 3-4% are

typically obtained. By reference to calibration curves, the fal1 in mea that the ma~o~ty Of cells maintain pHi to within i-O.1 p units of the areas value, but does not exclude the possibility of a minor popul n of more a~~dQti~celEs. y ~ornpa~so~~ our previous study using PMC 22 cells showed that extern pendent cellular quiescence in vitro was associated with a fall in approximately 0.3 pH units [ 121. although the present results do not suggest that reduce PHi iS aI%im~Q~a~t cause of cellular quiescence in I316 tumors, they are ~~~o~~~~s~ve~ ~ute~~edto infuse animals with BrUdR prior to sorting tumor pE&. Rerunning sorted populations stained for incorporated have shown whe er reduced pHi was associated with quies ly, at feast for 16 tumors, the expe~rne~t~ error of p measnreme~t 95 exceeds the true variance in pHi that sorting w red~~ti~~ in mean PHi seen follows tn ho ~~rn~at~b~ewith previous observations using suggests that despite the problems encountere of at Least ~~t~a~~ypreventing acid load ret This is ~~~~er evidence that the true intercelf ~obab~y blows The fundamental question on a possible reason for failure of cancer treatments suing a~t~~ro~~~erat~v~ also because ahernative treatment strategies aimed s~~~i~~~~~~at a ave been proposed [13]. Future d~re~t~o~s s~o~~~ be to ~~~r~ase the g power of flow cytometric pHi dete~rn~~at~o~~either by the use of am~li~er pe~orrna~~~ so as to make full use of F. Meanwhile, it woukl be worth r~~~~t~~g the the dynamic range of present ~~~~rirne~t§ using alternative tumor models. The two pammeter plot of DNA content versus BrIJdR ~co~orat~on shown in Fig. 1 was prepared in this laboratory by Mary Seaman and Anna deFazio. We once again thank Judy IIood for her co~~~~~ngentbus~asm for preparing mauusc~pts.

1. TannO&,

4. 5. 6. 7.

8. 9.

1. F. (1968) Bit.

j. Cancer 22, 258.

. L., Raveman, J., and Reinhold, H. S. (1984) ~udi~t~~r. Oncol. 2, 343. T., Ng, T. C., Lilly, ,&I. B., Lawson, A. J., Corbett, ‘I. H., Durant, .I. R., and Olickson, J. D..(1983) Prac. Natl. Acad. Sci. USA 80, 334. Irving, M. G., Simpson, S. J., Field, J,, and Doddrell, D, I& (1985) Cancer Res. 45, 4gl. Gkunieff, P. G., Koutcher, J. A., Gerweck, L., Scotland, E., Witzig, B., Urano, M., Brady, T., Neuringer, L., and Suit, R. D. (19%) Inr. .I. Radiat. Oncol. Biol. I?&. 12, 793. Olsnes, S., Tonnessen, T. I., and Sandvig, K. (1986) J. Cell Bioi. 1@2,%7. Seifter, J. L., and Aronson, P. S. (1986) J. Clin. Invest. 78, X59. Lagarde, A. E., and Pouyssegur, J. M. (1986) Cancer Biuchem. Biophys. 9, 1. Pouyssegur, J., Sardet, C., Franchi, A., L’Allemain, G., and Paris, S. (1984) Proc. Natl, Aead. Sci. USA $1, 4833.

IQ. Taylor, I. W., and Hodson, P. J. (1984) .f. Ceil. Physiol. X21, 517. Il. Musgrove, E., Rugg, C., and Hedley, D. (1986) C~~~~ei9 7, 347. 12. ~US~QY~, E., Seaman, M., and Hedley, D. (1987) Exp. C&I Res. 172, 65.

116

Hedley and Jorgensen

13. deFazio, A., Lear-y, J. A., Hedley, D. W., and Tattersall, M. (1987) J. Hisiochem. Cytochem. 35, 571. 14. Moolenaar, W. H., Tertoolen, L. G. J., and delaat, S. W. (1984) J. Biol. Chem. 259, 7563. 15. Eden, M., Haines, B., and Kahler, H. (19.55)J. Natl. Cancer Inst. 16, 541. 16. Evelhoch, J. L., Sapareto, S. A., Jick, D. E. L., and Ackerman, J. J. H. (1984) Proc. Natl. Acad Ski. USA 81, 6496. 17. Shapiro, H. M. (1985) Practical Flow Cytometry, pp. 155-157, A. R. Liss, New York. 18. L’AIlemain, G. L., Franchi, A., Cragoe, E., and Pouyssegur, J. (1984) J. Biot. Chem. 259,4313. 19. Rotin, D., Wan, P., Ginstein, S., and Tar-mock, I. (1987) Cancer Res. 47, 1497. Received March 22, 1988 Revised version received August 9, 1988

Printed

in Sweden