Comparison of the cytotoxic activity of melphalan with L-prolyl-m-L-sarcolysyl-L-p-fluorophenylalanine in human tumour cell lines and primary cultures of tumour cells from patients.

m-L-sarcolysin (m-L-SL) is an isomer of melphalan (Mel) with the di(2-chloroethyl) amino group being substituted in the meta position of phenylalanine. By covalent conjugation of amino acids to m-L-SL, a peptide complex consisting of six m-L-SL-based oligopeptides known as peptichemio (PTC) was developed previously. In the present study, the cytotoxic activity pattern of the different oligopeptides of PTC was investigated in ten human tumour cell lines representing different mechanisms of cytotoxic drug resistance using the fluorometric microculture cytotoxicity assay (FMCA). In the cell line panel, L-prolyl-m-L-sarcolysyl-L-p-fluorophenylalanine (P2) was the most active oligopeptide, showing slightly lower mean IC50 values (2.6 vs 3.9 and 4.1 microg ml(-1)) than Mel and m-L-SL. The other five oligopeptides were less active than Mel. All active oligopeptides showed mechanistic similarity to Mel as judged by the correlation analysis of the cell line panel log IC50 values (R > or = 0.90), although P2 appeared to be less sensitive to GSH-mediated drug resistance. The relative activity of Mel and P2 was found to be related to degree of proliferation, P2 being more active towards low-proliferating cell lines. P2 and Mel were then further characterized in 49 fresh human tumour samples. In these samples P2 was considerably more active than Mel and showed a higher relative solid tumour activity (2.7 to 4.5-fold). However, the correlation of log IC50s between P2 and Mel in patient cells was high (R = 0.79), indicating a similar mechanism of action in this tumour model too. Cross-resistance with other standard drugs was lower for P2 than Mel. The results show that P2 is the most potent component of PTC and demonstrates a favourable activity profile compared with Mel. These data suggest that further investigation of P2 as a potential anti-tumour agent is warranted.

The 8'26/Dox, x was selected for doxorubicin resistance and shows the classical MDR phenotype with oxverexpression of Pglxcoprotein 170 (Dalton et al. 1986). The 82"6/LR-5 w-as selected for NMel resistance. proposed to be associated w-ith increased lexels of glutathione (Bellamr et al. 1991: Mulcahv et al. 1994. The t-937-Vcr was selected for -incristine resistance. proposed to be tubulin associated (Botling et al. 1994). The H69AR. selected for doxorubicin resistance. expresses a multidruc-resistant (MDR phenotype proposed to be mediated by a multidrug resistanceassociated protein (MRP Mirski et al. 1987: Cole et al. 1992). The CENIVM-1. selected for teniposide resistance. expresses an atypical MDR. which is proposed to be topoisomerase LI (topoll) associated (Danks et al. 1987(Danks et al. . 1988. The exact mechanism of resistance for the primary resistant ACHN cell line is not know-n and may be multifactorial (Nv-gren and Larsson. 1990).
The cell lines w ere grow-n in complete culture medium described belowx at 37^C in humidified atmosphere containing 5%' carbon dioxide. The 8226/Dox, vk-as treated once a month w-ith doxorubicin at 0.24 igc ml-' and the 8226/LR-5 at each change of medium w-ith Mel at 1.53 .gg ml-. The U-937-Vcr was continuously cultured in the presence of 10 ng ml-' vincristine and the H69AR >-as alternately fed w-ith drug-free medium and medium containincg 0.46 zg ml-' doxorubicin. The CEMI/VM-1 cell line was cultured in drug-free medium without anv loss of resistance for a period of 6-8 months. The resistance patterns of the cell lines were routinely confirmed in control experiments.

Patient samples
A total of 49 patient tumour samples from the different dia2nostic group w-as used to determine the actixitx of P2. MIel. and. for comparison. fix-e other cy totoxic drugs w-ere chosen to represent different mechanistic classes. Hoxxever. because of a limited number of cells. all drugs could not be tested in all samples.

Reagents and drugs
Fluorescein diacetate (FDA: Sigma) was dissolxed in DMSO and kept frozen (-20-C) as a stock solution protected from liaht. A complete medium consistine of culture medium RPMI-1640 (Hy Clone. Cramlington. UK) supplemented w-ith 10% inactix-ated FCS. 2 nmI glutamine. 50 .tg ml-' streptomwcin and 60 !gL ml-' penicillin xxas used throuthout for both cell lines and patient samples. Mel x-as obtained from the Wellcome Foundation. London. UK. The drug was receixed as a sterile powxder. 2 mg of wxhich wxere dissolxed in 0.5-1 ml of 92% ethanol xwith 2% hydrogen chloride and further diluted in cell culture medium to the desired drug concentrations. The components of PTC and m-L-m-L-SL_ere obtained from Istituto Sieroterapico. Milanese. S. Belfanti. Milan. Italy. The peptides 1-5 wxere obtained as ethyl esters and peptide 6 as methyl ester (Table 1). An aliquot of 2 mg of each >-as dissolxed in 0.5-1 ml of 92% ethanol wxith hvdrochloric acid and further diluted in cell culture medium to the desired drug concentrations. Cisplatin. cvtarabine. doxorubicin.
etoposide and x incristine xxere obtained from commercial sources and wxere dissolxved accordin2 to cuidelines from the manufacturer and further diluted in phosphate-buffered saline (PBS: HvClone) or sterile wxater.
In the cell line panel all drugs xxere tested at four different drug concentrations. obtained by fixvefold serial dilution from the maximum 10 [ag ml-'. On a molar basis the concentration of the different oligopetides are 39-43% of that of Mel and m-L-SL. To determine the dose-response relationship for Mel and P2 in patient samples. fi e different drug concentrations x ere used. obtained bx a fixefold serial dilution of the drugs from 50 jg ml-l. In the patient samples. the concentrations chosen for comparison wxith standard drugs xxere the empirically derixed cut-off concentrations EDCCs). defined as the concentration that produces a significant scatter of sun-ixal index (SI) -alues among haematological tumours. This concentration wxas used to optimize the conditions for exaluating cross-resistance. The concentrations 2 and 0.08 jg ml-' xere chosen for Mel and P2. respectixely. and the EDCCs for the other drugs hax-e been described prexviously ) Larsson et al. 1992 were stored frozen at -70°C for up to 2 months until further use.
Under these conditions. no apparent change in drug activity was observed ).
The fluorometric microculture cytotoxicity assay procedure The fluorometfic microculture cytotoxicity assay (FMCA) is based on measurement of fluorescence generated from hydrolysis of FDA to fluorescein by cells with intact plasma membranes and has been described in detail previously .
Briefly. the cells were resuspended in complete medium. and 180 jl of cell suspension was seeded into the wells of 96-well experimental microtitre plates prepared with drugs as described.
Cell densities were 5-20 x 10cells per well for the cell lines. 10-20 x 103 cells per well for the solid tumour cells and 50-100 x 103 cells per well for the haematological tumour cells. Each drug and concentration was tested in triplicate. Six wells with cells but without drugs served as control and six wells with only culture medium as blank.
The plates were incubated for 72 h at 37°C in humidified conditions containing 5% carbon dioxide. At the end of the incubation period the plates were centrifuged (200g, 5 min) and the medium was removed by aspiration. After one wash in PBS. 100 gl per well of FDA dissolved in PBS (10 jg ml-') was added. The plates were incubated for 45 min and the generated fluorescence (excitation 480 nm) from each well was measured at 538 nm in a 96well scanning fluorometer (Fluoroscan II. Labsystems Oy, Helsinki. Finland). The fluorescence is proportional to the number of intact cells in the well.
To evaluate the schedule dependency of drug activity. CCRF-CEM cells and ACHN cells were used and were exposed to the drug for 2. 4 or 72 h followed by washing with PBS, addition of new culture medium and analysis at 72 h. Stability of P2 and Mel under assay conditions was investigated by a bioassay. Plates prepared with Mel and P2 were preincubated with 100 Rl medium per well for different time periods. ranging from 0 to 72 h. at 37°C before cell suspension (U-937-GTB) was added. The activity of the drugs after different preincubation times was evaluated by comparing the SI values obtained after a further 72 h incubation with FMCA. as described above.

Quality control
Quality criteria for a successful analysis included a fluorescence signal in the control wells of more than five times mean blank value. a mean coefficient of variation (CV) in the control wells of less than 30% and more than 70% tumor cells in the cell preparation before incubation.

Quantification of FMCA results
Cell survival is presented as survival index (SI). defined as the fluorescence in experimental wells in per cent of that in control wells. with blank values subtracted. The IC., was defined as the concentration giving a SI of 50%.
For both cell lines and primary cultures. the ICsos were evaluated for each individual cell line and drug with custom-made computer software . A delta value was calculated as the logarithm of the IC5O of the individual cell line minus the mean of all ten log IC<os (Fridborg et al. 1996). The resistance factors (RFs) in each subline were defined as the IC., of the resistant subline divided by the IC., of its sensitive parental cell line.
The pairs of parental/resistant cell lines used for RF calculations of P-glycoprotein (P-gp). MRP. topo II. glutathione (GSH) and tubulin-associated resistance were RPMI 8226S/8226Dox40. NCI-H69/H69AR. CCRF-CEM/CEM-VM-1. RPMI 8226S/8226LR-5 and U-937-GTB/U-937-Vcr respectively. Correlation coefficients were determined using Pearson's correlation coefficient. Response rate was defined as the fraction of samples having a SI below 50% at 0.5 jg ml-' for all samples investigated. In vitro therapeutic index was calculated as median IC., of CLL samples/median IC., of normal PBMCs.

Measurement of DNA synthesis
In some experiments bromodeoxyuridine (BrdU) incorporation into cellular DNA was determined with an enzyme-linked immunosorbent assay (ELISA) kit from Boehringer Mannheim (Mannheim, Germany) essentially according to the protocol provided by the manufacturer. Briefly. cells were incubated in 96-well plates for 72 h in the presence of BrdU. The cells were then fixed and an antibody directed to BrdU was added. The formed immune complex was detected by a substrate reaction using tetramethylbenzidine and measured in a spectrophotometric microplate reader (Dynatech. Billingshurst. UK). P2 is more potent than the other PTC oligopeptides P2 was the most active m-L-SL oligopeptide. which showed a slightly lower mean IC5, (2.6 jg ml-') compared with Mel (3.9 jg ml-') and m-L-SL (4.1 jg ml-'). However, on a molar basis. the IC5, value for P2 was 3.3 times lower than m-L-SL. P1 showed an IC5o of 4.1 jg ml-' whereas the remaining m-L-SL oligopeptides had IC_ s between 5.8 and 9.1 (Table 2). P2 was the only oligopeptide producing a SI below 50% in all the tested cell lines (Table 2).

RESULTS
P2 appears not to be affected by GSH-associated resistance Although. the overall activity profile resembled that of Mel. P2 appeared not to be affected by GSH-associated resistance as determined by the low resistance factor obtained using the LR5-parental IC< ratio (RF 1.05. Table 3). Mel and m-L-SL, on the other hand.
showed RFs of 3.1 and 3.8 respectively. P2 also appeared less sensitive to MRP-associated resistance than Mel and m-L-SL with RFs of 1.55. 4.0 and 4.17 respectively ( Table 3). Neither of the drugs was affected by the remaining resistance mechanisms.  Mel (2.0 g.l ml-) an in xitro response rate (percentage of samples x-ith > 50%7decrease in SI) of 67-i and 0%`7 x-as observed for haematological and solid tumour samples respectix ely. The corresponding response rates for P2 wvas 100%;and 43%e (Table 4).
P2 is more active than Mel on low-proliferating tumour cell systems To investioate w hether the increased actixvitx of P2 could be related to the lowx proliferatixe rate of the primarv cultures. the ratio of Mel vs P2 IC__s in the cell lines >-as plotted against the rate of proliferation under assay conditions in V-shaped plates (Figure 3 . An inxerse relationship w-as observed (R = 0.70). P2 beina more actixe against the low--proliferating cell lines. The next series of experiments aimed to determine w-hether this relationship w-as causall1 related to proliferation rather than being cell-type specific. ACHN. w-hich shows a lowx grow-th rate in V-shaped plates but proliferates rapidlv w hen seeded into flat-bottomed plates. was used for this purpose. VWhen tested in flat-bottomed plates P2. Mel and m-L-SL showed similar IC.,s (not shown). In  (Figure 4). was the correlation between P2 and Mel relatively high, indicating a similar mode of action. However, crossresistance to standard drugs determined using the haematological samples was generally low (0.14-0.43. Table 5). The correlation with doxorubicin. etoposide and cisplatin was much lower for P2 than Mel. whereas the correlations with cytarabine and vincristine were of similar magnitude (Table 5) P2 shows a similar in vitro therapeutic index to Mel in low-prolifeaing cell systems P2 also showed lower IC 0s than Mel in PBMCs with median values of 0.27 and 4.0 respectively (n = 5). However. when compared with median IC-, values of malignant CLL samples. 0.07 and 1.4 gg ml-' respectively (n = 5) the in vitro therapeutic index (IC_, PBMCs/IC_0 CLL) was 3.9 for P2 and 2.8 for Mel (Table 6).

DISCUSSION
Mel and m-L-SL are closely related aromatic nitrogen mustard derivatives. The two molecules differ only in the position of the di-(2-chloroethyl) amino-group, which is in the para position in Mel and in the meta position in m-L-SL. By conjugation of additional amino acids to the carboxyl and amino groups of m-L-SL. a complex consisting of six different peptides has been developed. This mixture of peptides. PTC. has shown clinical activity in several human malignancies (Hug et al, 1980;Paccagnella et al. 1986;Zaniboni et al, 1988). In previous investigations with the human melanoma cell line RPMI 8322. PTC as well as some of its individual peptides were more effective than Mel and m-L-SL (Lewensohn et al, 1991a: Hansson et al. 1991. Myeloma cells isolated from bone marrow of patients with primary myelomas were more sensitive to PTC than to Mel (Paccagnella et al, 1985).
In the human melanoma cell line RPMI 8322, used in the abovementioned investigation, we found that one of the peptides in PTC.
In the present study we show that P2 was the most active component of PTC when tested in a panel of human tumour cell lines. Moreover, P2 was also more active than Mel and m-L-SL against several of the cell lines. Correlation analysis of cell line panel activity pattems demonstrated a close relationship between P2 and Mel, suggesting a similar mode of action. However. unlike Mel and m-L-SL, P2 appeared not to be affected by GSHand MRP-associated resistance to any greater extent. We have previously found, using a clonogenic assay, that buthionine sulphoximine, which depletes GSH, sensitizes a melanoma cell line to Mel but to a lesser extent to P2 (Hansson et al, 1991 Figure 3 Relationship between MeUP2 ICratios and number of doubling times 72 h-determined by haemocytometer counts in the cell line panel (n = 10). R = Pearson's correlation coefficent   In contrast to proliferating cell lines. human tumour biopsy cells R 9 were as a group significantly more sensitix-e to the cxtotoxic * activitx of P2 than Mel. The reason for this mav be related to the loxw proliferatix-e activitx of the primarx cultures in the present assax system (Weisenthal et al. 1991W as low-proliferating cell lines also showed higher relatix-e P2 sensitivitv. Furthermnore. direct manipulation of the proliferative rate of the ACHN cell line produced the correspondinc alterations of Mel vs P2 sensitivitt-. From a clinical point of view. the demonstrated abilitv of P2 to retain activity against non-cycling cells may be a distinct advantaae as the lox-grow-th fraction of manv solid tumours is a limiting * factor for therapeutic responses of most currently used antineoplastic arents. The indications of a w ider spectrum of anti-tumour X r l activity and a favourable therapeutic index in vitro as ",ell as the -1.5 -1 -0.5 0 0.5 1 1.5 2 low cross-resistance with standard agents clearly adds to the potential of P2 being a clinically useful anti-tumour agent. WNhat Log lC5 P2 then is the mechanism for increased toxicityof P2 against primary cultures and other non-proliferating cell systems? Although. the Correlation between log IC5 values for Mel and P2 in 49 primary druc appears to act mechanistically similar to Mel both in the cell mour samples. R = Pearson's correlation coefficient lines and the primarv cultures. one may speculate on. at least. t o possible explanations. On one hand the effect of a bifunctional alkvlating agent is related to the frequency of DNA damage such GSH lexels for P2 than Mel sensitivity. The explanation as DNA cross-links (Lew-ensohn et al. 1991a). The frequency of lack of GSH-mediated resistance in reponse to P2 does not DNA cross-links may. hoxxever. be regulated by-DNA repair to involhe intracellular liberation of m-L-SL as this mechanisms. xhich at least in some cell lines is correlated with nd shows RFs of similar magnitude to Mel for GSH-asso-drug sensitivity (Batist et al. 1989). It would then seem possible resistance. Altered substrate recognition of m-L-SL that a bifunctional aly lating agent in the form of an oligopeptide ptides bx cellular GSH-dependent enzymes is one potential would not be recognized and excised from the DNA by the same tion for the phenomenon. In x itro sensitivity to Mel has repair mechanism as Mel. On the other hand another possible sly been noted to result in only a limited increase in toxi-explanation is that of a more effectixe cellular uptake of the rted by PTC as compared xx-ith Mel in freshly obtained bifunctional alkyvlator xxhen in the form of an oligopeptide as arrox-myeloma cells from untreated patients. compared xx-ith Mel only. In this context. it is interestinc to note stinglv. this findine xxas contrasted by the relatixvelv that another tripeptide of m-L-SL. 3-(p-fluorophenx1)-L-alany 1-3iced sensitixity to PTC in cell populations xxith in xitro [m-bis( 2-chloroethxl) aminopheny l]-L-alany-l-L-methionine ethyl ce to Mel (Lewxensohn et al. 199  found that this peptide used multiple transport pathways in L1210 cells (Yagi et al. 1988). Both the above alternatives are currently being explored.
In whole blood P2 is rapidly degraded to m-L-SL. a fact that may limit the activity of the drug in vivo (Ehrsson et al, 1993). This finding indicates that peptidase activity probably degrades the P2 compound intracellularly. Degradation of di-, tniand tetrapeptides has previously been observed in erythrocytes and leucocytes that have high peptidase activity (Stem et al, 1951). More attempts will be made to characterize exactly the intracellular degradation of P2 and test its efficacy in comparison with m-L-SL in vivo.
In the present study, we used a human cell line panel in combination with a panel of primary tumour cultures from patients for in vitro evaluation of differential drug responses of PTC oligopeptides. In a previous study  we showed that the present cell line panel is capable of detecting mechanisms of action of standard drugs in addition to its ability to evaluate sensitivity to drugs to defined types of mechanisms of resistance. Complementary to this, non-clonogenic assays used on fresh primary tumour cultures from patients have been shown to mimic the known clinical activity pattern of standard drugs. We have also previously shown that non-clonogenic cytotoxicity assays such as the FMCA can detect tumour type specific activity retrospectively for a series of standard drugs ) and prospectively for early phase I-H drugs such as vinorelbine. idarubicin. CdA. gemcitabine. taxol and topotecan Csoka et al. 1995: Nygren et al. 1995: Fridborg et al. 1996: Jonsson et al, 1997. Thus, experience gained so far suggests that these model systems may be valid tools for initial predictions of the activity and potential utility of novel anticancer drugs.
In summary, we have demonstrated high anti-tumour activity of a m-L-SL oligopeptide against cell lines and primary cultures of tumour cells from patients. The drug appears to show retained activity against non-proliferating cell systems, shows a positive therapeutic index and demonstrates low levels of cross-resistance with standard drugs. Formal testing of these in vitro predictions will require in vivo testing in relevant tumour models and these studies are currently under way. glutamylcysteine synthetase activity in melphalan-resistant human multiple myeloma cells expressing increased glutathione levels. Cancer Chemother Pharnacol 34: 67-71 Nygren P and Larsson R (1990) Verapamil and cyclosporin A sensitize human kidney tumor cells to vincristine in absence of membrane P-glycoprotein and without apparent changes in the cytoplasmic free Ca-concentration. Biosci Rep 10: 231-237