Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

It is widely accepted that the inefficient microvascular system in solid tumors limits delivery of some anticancer drugs to their target cells, and that this may contribute to treatment failure in cancer chemotherapy (Jain, 1998; Tannock, 1998). This problem is likely to be particularly severe for basic DNA binding drugs such as doxorubicin and mitoxantrone since, in addition to reversible DNA binding, many of these compounds are trapped by pH-dependent partitioning into acidic vesicles (Denny and Wilson, 1986). These reversible binding/uptake processes have a high capacity for sequestering drug in cells; the difficulty of saturating these sites can be expected to retard delivery of drug to cells distal from blood vessels (Durand, 1990; Wilson and Denny, 1992).

Fluorescence intensity gradients in multicellular spheroids and tumors confirm that doxorubicin diffuses poorly in tissue (Ozols et al., 1979; Durand, 1989), but the difficulty in determining drug concentrations from fluorescence in tissue has limited quantitative investigation. A recently developed three-dimensional cell culture system (Cowan et al., 1996; Hicks et al., 1997; Minchinton et al., 1997; Topp et al., 1998) provides an in vitro model in which drug transport can be measured more directly. In this model, tumor cells grown as multicellular layers (MCLs) are used to separate two compartments containing culture medium, which makes it possible to quantify drug transport by measuring flux of compound between the two compartments. A major advantage of this approach is that compound-specific analytical methods (e.g., HPLC) can be used to distinguish parent drug and metabolites.

MCLs have been used to investigate tissue diffusion properties of radiosensitizers (Cowan et al., 1996), bioreductive drugs (Hicks et al., 1998; Phillips et al., 1998; Kyle and Minchinton, 1999), antibody fragments (Topp et al., 1998), hypoxia markers (Zhang et al., 1998), and DNA intercalators (Hicks et al., 1997; Tunggal et al., 1999). These studies provide direct evidence for the slow diffusion of doxorubicin, mitoxantrone, and a dibasic acridine derivative, 9-[3-(-dimethylamino)propylamino]acridine (DAPA) (Fig. 1). Slow diffusion of the latter was shown to be largely due to pH-dependent sequestration in lysosomes rather than DNA binding (Hicks et al., 1997).

View larger version (9K): [in this window] [in a new window]   Fig. 1.   Structures of compounds investigated in their predominant ionization state at neutral pH. DACA-2H is the acridan metabolite of DACA, identified in this study.

The acridine DNA intercalator N-[2-(dimethylamino)ethyl]acridine-4-carboxamide (DACA, NSC 601 316, Fig. 1) is a dual inhibitor of topoisomerases I and II (Finlay et al., 1996), is insensitive to multidrug resistance (Finlay et al., 1993; Davey et al., 1997), and is currently in phase II clinical trial. DACA and other acridine-4-carboxamide analogs have greater antitumor activity than related aminoacridine derivatives (Atwell et al., 1987; Denny et al., 1987). The acridine-4-carboxamide chromophore in DACA (ring pK_a = 3.5) is much less basic than the 9-aminoacridine chromophore in DAPA (ring pK_a = 8.8) so that at physiological pH the former carries a +1 charge, and the latter a +2 charge. DACA is also a weaker DNA binder than related aminoacridines, with a 6-fold lower association constant than the corresponding 9-amino compound (Crenshaw et al., 1995). Literature values for binding of DACA and DAPA to calf thymus DNA under equivalent conditions (0.1 M ionic strength, pH 7.0, 37°C) show a similar differential in association constant, with values of 5 × 10⁴ M¹ for DACA (Crenshaw et al., 1995) and 2 × 10⁵ M¹ for DAPA (Siim et al., 2000).

The superior antitumor activity of DACA was suggested to be related to its relatively weak basicity and low DNA binding affinity, features considered likely to facilitate relatively efficient tissue distribution (Denny et al., 1987). DACA thus represents one of the earliest examples in which extravascular transport characteristics were considered explicitly in the optimization of a DNA-binding antitumor drug. However, there is no direct evidence to support this rationale. In fact the converse argument has also been proposed as a basis for the tumor selectivity of DACA. Thus, slow clearance of drug from Lewis Lung tumors, relative to plasma, in mice (Paxton et al., 1993) has been interpreted as reflecting slow extravascular transport (efflux), leading to the suggestion that prolonged exposure of cells distant from functional blood vessels could contribute to solid tumor selectivity (Paxton et al., 1993; Baguley and Finlay, 1995).

The present study uses the MCL model to measure the parameters controlling extravascular transport of DACA. Cell uptake and metabolism is first assessed in single cell suspensions to guide selection of transport models (Wilson and Hicks, 1999), and comparison is made with the more strongly basic acridine DAPA. The derived transport parameters are used to model pharmacokinetics and pharmacodynamics (cytotoxicity) as a function of distance from blood vessels in tumors.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals and Radiochemicals. [¹⁴C]Urea (185 MBq/mmol) was purchased from PerkinElmer Life Science Products, Boston, MA. [³H]DAPA (14.9 GBq/mmol) was synthesized and purified as previously described (Hicks et al., 1997). DACA and known DACA metabolites [the corresponding 9(10H)-acridone, N-monomethyl derivative, side chain N-oxide, and 4-carboxylic acid hydrolysis product (Robertson et al., 1993; Schofield et al., 1999)] were supplied by Professor W. A. Denny of the Auckland Cancer Society Research Center. [³H]DACA was prepared from nonradioactive DACA by Sibtech Corporation, Elmsford, NY (1.75 TBq/mmol, 97.5% purity by HPLC with radiochemical detection) and was stored in methanol at 20°C and repurified by HPLC before use as described for DAPA (Hicks et al., 1997). The acridan derivative of DACA, DACA-2H {N-[2-(dimethylamino)ethyl]-9,10-dihydroacridine-4-carboxamide; Fig. 1}, was prepared by reduction of an ethanolic solution of the 9(10H)-acridone derivative of DACA with mercury-aluminum amalgam.

Cell Lines and Growth of Multicellular Layers. V79-171b Chinese hamster fibroblasts and EMT6/Ak mouse mammary tumor cells were passaged as monolayer cultures in alpha minimal essential medium containing 5% fetal calf serum, without antibiotics, by weekly trypsinization. Multicellular layers were grown on collagen-coated Teflon support membranes in Millicell-CM cell culture inserts (Millipore Corporation, Bedford, MA) in alpha minimal essential medium containing 10% fetal calf serum with penicillin (100 units/ml) and streptomycin (100 µg/ml) as previously described (Hicks et al., 1997) by seeding at 2 × 10⁵ cells/insert and growing submerged in stirred medium at 4 days. MCLs were enzymatically dissociated to single cell suspensions by incubation in 5 ml of 0.07% trypsin in saline containing trisodium citrate (14 mM, pH 7.6) for 10 min. All experiments were performed at 37°C under aerobic (95% O₂) conditions unless otherwise stated.

Diffusion Chamber Apparatus. Diffusion chambers (Fig. 2), based on a design described by Kyle and Minchinton (1999) were constructed of Perspex (Lucite), and placed on custom-made variable speed magnetic stirrers in a 37°C water bath. Gas mixtures (5% CO₂ in 95% O₂ or 95% N₂) were supplied continuously at a flow rate of 5 ml min¹ through polyetheretherketone (PEEK) tubing (0.18 mm i.d.; Alltech Associates, Deerfield, IL) connected to a stainless steel manifold. The equipment was sterilized in 70% ethanol.

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Diffusion chamber apparatus. Shown is a six-chamber apparatus allowing three diffusion measurements to be made simultaneously. A cross-section is shown through the middle of the end pair of chambers. A Millicell-CM insert, with or without MCL, is held between two compartments containing culture medium (5-10 ml) in a 37°C water bath. The compartments are magnetically stirred and gassed at 5 ml/min via PEEK tubing connected to a manifold maintained at 100 kPa. Drug is added to the donor compartment, and timed samples are taken from the donor and receiver compartments to determine drug concentrations.

Diffusion Experiments. After loading with an MCL, the diffusion chamber was equilibrated for 1 h with magnetic stirring at 150 rpm using the same medium as described above, and was maintained at pH 7.4 ± 0.1 throughout the experiment by gassing. Prior to addition of drug, 1 ml of medium was sampled from each chamber and the number of cells determined with an electronic particle counter; MCLs were not used if there were >10³ floating cells/ml. A further sample (0.5 ml) was taken for determination of background radioactivity and HPLC baseline. Flux was initiated by adding 50 µl of medium containing internal standard ([¹⁴C]urea) and drug to the donor compartment using a Hamilton syringe via the stainless steel sampling (gas outflow) port, and samples (0.5 ml) were taken from both compartments at intervals thereafter. [¹⁴C]Urea was determined by off-line scintillation counting (50 µl, in 5 ml of Emulsifier-Safe water-accepting scintillant, Packard Tricarb 1500 liquid scintillation analyser; Canberra Packard, Meriden, CT) and the balance of the sample was stored at 80°C for HPLC analysis. MCL were frozen after completion of some flux experiments for determination of thickness from frozen sections (Hicks et al., 1997).

HPLC and LC/MS. Medium was analyzed by HPLC by direct injection of up to 200 µl, using a WISP 712 refrigerated autoinjector and WISP 600 pump (Waters, Milford, MA) and Alltima C8 column (150 × 4.6 mm, 5 µm; Alltech Associates), with monitoring of absorbance at 254 nm for DACA, 266 nm for DAPA, and 296 nm for DACA-2H using a diode array detector (Hewlett Packard 1040A). On-line radiochemical analysis was performed by a Radiomatic 150TR detector (Canberra Packard) using Utima-Flo AP scintillant (Canberra Packard) at 2.0 ml min¹ and a flow cell volume of 500 µl. The mobile phase (flow rate 1.0 ml min¹) comprised a linear gradient of 25 to 65% acetonitrile in 0.45 M ammonium formate, pH 4.5, for 20 min, returning to 25% acetonitrile between 25 and 30 min. On-line mass spectrometry was performed using a Hewlett-Packard LC/MS system comprising a HP1100 HPLC with a single quadrupole mass spectrometer using positive-mode electrospray ionization (N₂ drying gas flow rate 10 l min¹ at 350°C, nebulizer pressure 25 psi gauge (psig), capillary voltage 4000 V) in the scan-mode (m/z 50-800, stepsize 0.05) using variable fragmentor voltages. The column was an Alltima C8 (150 × 3.2 mm, 5 µm), with a flow rate of 0.5 ml min¹ and an injection volume of 10 µl.

Mathematical Modeling of Drug Flux and Parameter Estimation. Transport in MCL was modeled as one-dimensional diffusion with reaction in a series of homogeneous phases (donor chamber, MCL, Teflon support membrane, and receiving chamber), as described previously (Hicks et al., 1997), with appropriate modification for the new diffusion chamber apparatus (both chambers stirred). For each compartment the diffusion equation is written with reaction terms to represent spontaneous hydrolysis, metabolism, and loss of free drug due to intracellular sequestration (modeled as binding to two classes of sites, one saturable and one nonsaturable):

(1)

where c_f, c_b1, and c_b2 are the concentrations at position x and time t, of free, nonsaturably bound, and saturably bound drug, respectively. D_f is the diffusion coefficient of the free drug; k₁ and k₁ are the forward (association) and reverse (dissociation) rate constants for nonsaturable binding; k₂ and k₂ are the corresponding rate constants for saturable binding; B_max is the total concentration of saturable binding sites; k_met is the first order metabolic rate constant; and k_hydr is the rate constant for hydrolysis. The boundary conditions were for zero flux into the donor and receiver compartments and the initial conditions were c_f(x,0) = c_1,0, the initial concentration in the donor compartment, c_f(x,0) = 0 otherwise, and c_b1(x,0) = c_b2(x,0) = 0 in all compartments. The free fraction of DACA in culture medium containing 10% fetal calf serum has been estimated as 97% (Finlay and Baguley, 2000); binding to serum proteins in medium was therefore ignored in the flux modeling, and in investigation of uptake and metabolism in cells (see below). The MCL transport model was solved numerically for c_f, and c_b1 + c_b2 in each compartment at appropriate times using the routine DO3PBF from the NAG Fortran library (Numerical Algorithms Group, Oxford, UK) with compensation for volume removed during sampling as previously described (Hicks et al., 1997). The predicted concentrations in the donor (c_f, donor) and receiver (c_f, receiver) compartments were fitted simultaneously to the measured concentrations, by minimization of the residual sum of squares.

DACA Uptake by Single Cells. Cellular uptake of [³H]DACA was investigated in stirred single cell suspensions (1-2.2 × 10⁶ cells/ml of V79 and EMT6 cells, obtained by dissociation of MCL) at pH 7.4 by the spin-through-oil technique as for [³H]DAPA (Hicks et al., 1997). Suspensions were equilibrated under 20%O₂, 5%CO₂ at 37°C for 1 h before the addition of drug. Concentrations were determined by off-line scintillation counting of extracellular medium and cell pellets as previously described (Hicks et al., 1997). In contrast to DAPA, the time course of DACA uptake into cells was too rapid to measure initial rates by this technique. Thus, only steady-state values for the extracellular and intracellular concentrations were determined. The binding model fitted to the data is the steady-state solution of the kinetic model described for DAPA (Hicks et al., 1997).

(2)

and

(3)

where

(4)

and

(5)

Assuming that the intracellular free drug (c_f) is equal to extracellular drug concentration (c_e)

(6)

and the intracellular concentration

(7)

The model predictions for c_i (eq. 7) were fitted to the measured values from the uptake experiments using nonlinear regression (Sigmaplot scientific graphing software, version 4.01; SPSS Inc, San Rafael, CA) with K₁, K₂, and B_max as the fitted parameters.

DACA Metabolism by Single Cells. Metabolism of [³H]DACA by V79 cells was examined under the same conditions as the cell uptake experiments, by adding DACA to 10 µM and the extracellular medium was assayed for DACA by HPLC. The rate constant (k_met) for metabolic loss of DACA was estimated by fitting a monoexponential decay function to the extracellular concentration data (c_e):

(8)

where c₀ is the extracellular concentration after cellular uptake is complete (5 min).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Uptake of DACA in Single Cell Suspensions. The cellular pharmacology of DACA was first investigated in stirred single cell suspensions to identify processes likely to influence extravascular transport (accumulation in cells and drug metabolism). [³H]DACA was rapidly taken up by EMT6 and V79 cells under aerobic conditions at 37°C; the ratio c_i/c_e was constant with time (5-15 min for V79, 5-60 min for EMT6; 1 µl/10⁶ cells) within experimental error (data not shown). The steady-state value of c_i/c_e was ca. 420 to 450 at low DACA concentrations, with partial saturation of uptake at higher concentration (50% saturation at 2-3 µM DACA; Fig. 3). These values were based on total radioactivity, and do not take into account the slight (<10%) metabolic conversion to DACA-2H by V79 cells over this time (see below). Similar values have recently been reported for DACA uptake in other cell lines (e.g., c_i/c_e of 550 for the Lewis Lung carcinoma line LLTC; Haldane et al., 1999).

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Single cell uptake of [3H]DACA. The ratio of intracellular (ci) to extracellular (ce) DACA concentration is plotted against the extracellular concentration in V79 (A) and EMT6 (B) cells in single cell suspensions (). A separate experiment in which DACA uptake was determined in V79 cells with () and without () 50 mM NH4Cl is also shown. Each point is the steady-state value (mean and S.E.M.) determined by averaging concentrations at 5, 10, and 15 min (V79) or 10, 20, 30, 40, and 60 min (EMT6). The values are based on total radioactivity, uncorrected for metabolism, which was negligible under these conditions.

3

View this table: [in this window] [in a new window]   TABLE 1 Intracellular binding and metabolism parameters for DACA and DAPA estimated from data in aerobic stirred single cell suspensions at 37°C (ca. 106 cells/ml) Values are mean ± S.E. for fit determined by nonlinear regression (eqs. 2-7) for binding, and linear regression for metabolism (eq. 8). The parameter values are scaled to give the intracellular values using the intracellular volume fraction (i) of the cell suspensions assuming a cell volume of 1 µl/106 cells.

Metabolism of DACA in V79 Cells. In the above-mentioned experiments, a radiolabeled product with a retention time of 10.1 min (cf. 6.4 min for DACA) accumulated in the V79 cell suspensions, with a minor second metabolite (R_t = 12.8 min, "metabolite 2") at later incubation times (Fig. 4). Loss of DACA in V79 cell suspensions was first order (data not shown) with a rate constant of (2.5 ± 0.5) × 10³ min¹ (mean ± S.E. for three experiments) at 10⁶ cells/ml. The major metabolite was partially converted back to DACA when samples were deproteinized using perchloric acid, or frozen and thawed, with the recovered DACA having the same specific activity as the starting [³H]DACA as determined by comparison of absorbance and radioactivity signals in the chromatogram (data not shown).

View larger version (23K): [in this window] [in a new window]   Fig. 4.   HPLC analysis of DACA and its metabolites in the supernatant of aerobic V79 cell suspensions incubated at 37°C for 12 h, using on-line diode-array detection (left) and electrospray mass spectroscopy (right). The mass spectra of DACA and DACA-2H were acquired using a fragmentor voltage of 160 V. Unlabeled peaks were also present in control medium.

Internal Standard ([¹⁴C]Urea) Flux in Diffusion Chamber. DACA transport through MCL was investigated using a symmetrically stirred diffusion apparatus (Fig. 2), allowing measurement of drug concentrations on both sides of the MCL. Our previous experimental system maintained strictly unstirred conditions in the donor compartment (by adding agar at the same time as drug), permitting measurement in the receiver compartment only. The new apparatus was first characterized by investigating the flux of [¹⁴C]urea and [³H]DAPA through V79 MCL, previously investigated using the agar method (Hicks et al., 1997). Representative data for a single MCL are shown in Fig. 5; the lines are fits to the diffusion-reaction model. Flux of [¹⁴C]urea through collagen-coated Teflon support membranes (thickness 30 µm) without MCL (Fig. 5, A-C) was fitted to give a urea diffusion coefficient in the collagen-coated Teflon support, D_s, of (1.84 ± 0.06) × 10⁶ cm² s¹ (mean ± S.E., n = 19). This is 11.0 ± 0.4% of the value of D for urea in culture medium (Hicks et al., 1997), which is similar to the effective porosity (D_s/D) determined previously using unstirred donor compartments (9.8 ± 0.7%; Hicks et al., 1998). The transport of urea across V79 MCL (Fig. 5, A-C), was also well modeled as simple diffusion through the support and MCL in series. The thickness of each MCL was determined from frozen sections after the flux experiments (thickness 202 ± 20 µm, mean ± S.D. for eight MCL). Fitting the diffusion coefficient of urea in the V79 multilayers (D_{m, urea}) gave (1.72 ± 0.07) × 10⁶ cm² s¹ (n = 8), in good agreement with the value of (1.5 ± 0.1) × 10⁶ cm² s¹ found using the unstirred donor compartment system (Hicks et al., 1997).

View larger version (26K): [in this window] [in a new window]   Fig. 5.   Simultaneous transport of internal standard ([14C]urea, left) and [3H]DAPA (right) through collagen-coated Teflon membranes (no MCL; open symbols) and V79 multicellular layers (closed symbols) under aerobic conditions at 37°C. Single representative curves are shown. c is the measured concentration and c1,0 the initial concentration in the donor compartment. The volume in the donor and receiver compartments at time zero was 7 ml. A and D, donor compartment. B and E, mean concentration in both compartments. C and F, receiver compartment. Solid lines are model fits for diffusion with reaction (eq. 1). The dashed curves in D and F are the DAPA flux curves expected using the transport parameters fitted in the previous study (Hicks et al., 1997).

DAPA Transport through V79 Multilayers. Flux of DAPA through the support membrane without MCL (Fig. 5, D-F) was accompanied by slow hydrolysis to 9(10H)-acridone (k_hydr = 1.15 × 10⁴ min¹). This reaction term was included in the diffusion model (eq. 1), giving an estimate of D_s for DAPA of (0.940 ± 0.002) × 10⁶ cm² s¹ (Table 2). When a V79 MCL was present, DAPA flux into the receiver compartment was much slower than for urea (Fig. 5F) and was accompanied by considerable uptake into the MCL as demonstrated by the lowering of the concentration averaged over donor and receiver compartments (Fig. 5E). The concentration-time curves were modeled, with D_m as the sole fitted parameter, as diffusion with reversible binding and hydrolysis using the flux of the urea internal standard to estimate thickness of each MCL and using the binding parameters determined from single cell uptake studies (Table 1) after scaling to the cell density in the MCL. When DAPA transport parameters determined previously (Hicks et al., 1997) by the agar method (as in Table 2, except _i = 0.29) were used, the model predicted the concentration-time curve for the receiver compartment well (Fig. 5F, dashed line) but substantially underpredicted the rate of removal of DAPA from the donor compartment (Fig. 5D). The combined data set could not be fitted simultaneously except by increasing the intracellular volume fraction _i in the MCL (i.e., the scaling factor for K₁ and B_max between intracellular parameters derived from single cell suspension and MCL) from 0.29 to 0.75, with a compensating 3-fold increase in D_m to (1.45 ± 0.12) × 10⁶ cm² s¹ (n = 4). With these modifications, the model provided an excellent fit to the concentration-time profiles in both chambers (Fig. 5, D-F, solid lines). This value for _i is consistent with estimates for V79 spheroids (Durand, 1980; Freyer and Sutherland, 1983).

View this table: [in this window] [in a new window]   TABLE 2 Transport parameters for the acridines DACA and DAPA in MCL Transport was determined using the diffusion apparatus illustrated in Fig. 2, with an initial drug concentration c1,0 of 10 µM. Diffusion coefficients in the support membrane (Ds) and MCL (DM) were estimated by fitting the flux data to eq. 1, using the intracellular binding and metabolism parameters (Table 1) scaled by the intracellular volume fraction (i) for the MCL.

DACA Transport Characteristics. [³H]DACA was stable in culture medium under the conditions of the flux experiments (radiochemical purity >99% after 6 h), and no loss was detected during flux through collagen-coated support membranes (Fig. 6, A-F). Flux was well modeled as simple diffusion with a fitted D_s of (1.04 ± 0.10) × 10⁶ cm² s¹ (n = 13). There was no observable concentration dependence in D_s over the range 0.004 to 10 µM (data not shown).

View larger version (27K): [in this window] [in a new window]   Fig. 6.   Transport of [3H]DACA (10 µM) through collagen-coated Teflon membranes without multilayers (open symbols) and with EMT6 and V79 multilayers present (closed symbols) under aerobic conditions at 37°C. The volume in the donor and receiver compartments at time zero was 9 ml. A, D, and G, donor compartment. B, E, and H, mean concentration in both compartments. C, F, and I, receiver compartment. , EMT6 MCL; , V79 MCL without NH4Cl; , V79 MCL with 50 mM NH4Cl. Lines are fits to the diffusion-reaction model, except for G-I. Dashed line, urea flux in the same EMT6 MCL as for DACA. G-I, accumulation of acridan metabolite ([3H]DACA-2H) during the DACA flux experiment shown in D-F (lines are not fitted curves).

6

1

6

1

6

1

6

1

Simulation of DACA Pharmacokinetics (PK) and Pharmacodynamics. The reaction-diffusion parameters controlling transport of DACA in MCL, as determined above, were used to simulate PK and pharmacodynamics as a function of distance from blood vessels in tumors. The input to the extravascular compartment was provided by fitting a two-compartment open model to the reported concentration-time data for mouse plasma (Paxton et al., 1993):

(9)

where the rate constant for absorption from the peritoneum K_a is 0.12 min¹ and the fitted coefficients for the biexponential decay are A = 86.5 µM, B = 0.85 µM,  = 0.052 min¹, and  = 0.004 min¹. The data and fitted curve are shown in Fig. 7A. We then investigated whether the extravascular transport properties of DACA, as measured in MCL, could be used to predict the average concentrations measured in Lewis Lung tumors in the above-mentioned study (Paxton et al., 1993). We have not been able to grow the LLTC cell line (derived from Lewis Lung tumors) as MCL, but like EMT6 it does not metabolize DACA to DACA-2H. Given that the transport parameters in both V79 and EMT6 cells are similar (with the exception of k_met), the latter would appear to provide a reasonable approximation for Lewis Lung tumors. In addition, cellular uptake of DACA is similar in all cell lines examined so far, including LLTC cells (Haldane et al., 1999).

View larger version (17K): [in this window] [in a new window]   Fig. 7.   A, pharmacokinetics of total DACA (free plus bound) in plasma () and subcutaneous LL tumors () following a single i.p. dose (410 µmol/kg) to BDF1 mice (data from Paxton et al., 1993). The curve for plasma is an empirical PK model (eq. 9). The curve for tumor is the predicted average DACA concentration calculated using the DACA transport parameters in EMT6 multilayers (Tables 1 and 2). B, free DACA concentration in plasma () and 125 µm from the plasma interface (  ), calculated using the EMT6 MCL transport parameters for DACA. The slower penetration predicted if the transport parameters were the same as DAPA is also shown (· ·  · ·).

(10)

View larger version (19K): [in this window] [in a new window]   Fig. 8.   A, pharmacodynamic model (fitted lines, eq. 10) for DACA cytotoxicity in LLTC cell cultures (105 cells/ml; Haldane et al., 1992). DACA exposure times were 1 (), 3 (), 6 (), 24 (), and () 72 h. B, predicted pharmacokinetics and pharmacodynamics in LL tumors as a function of distance into the extravascular compartment. Free drug AUC was calculated by integration of the predicted concentration-time curves (as illustrated in Fig. 7B) to 72 h (   ). Surviving fraction was calculated from the pharmacodynamic model for DACA (eq. 10) using the concentration-time profile (predicted by eq. 1) at each distance using the DACA and DAPA transport parameters.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

All intercalators studied to date with the MCL system have shown very slow penetration kinetics, including DAPA (Hicks et al., 1997), mitoxantrone (Tunggal et al., 1999), and doxorubicin (Phillips et al., 1998; Tunggal et al., 1999). DACA (mol. wt. 294) appears to be dramatically different, with trans-MCL flux even faster than the internal standard urea (mol. wt. 60) in EMT6 MCL (Fig. 6C). Rapid flux of DACA was also seen in preliminary experiments with MCL grown from the human colon carcinoma cell line WiDr, with normalized DACA concentrations in the receiver compartment 3-fold higher than urea at 6 h (data not shown). Even in V79 MCL, which rapidly metabolize DACA, its net transport was efficient (and much greater than the nonmetabolized acridine DAPA). No direct comparison has been made between DACA and doxorubicin in the same experimental system, but Tunggal et al. (1999) have demonstrated very slow flux of doxorubicin through EMT6 multilayers, with a concentration in the receiver compartment only 3 to 4% of the equilibrium value by 6 h. In contrast, the measured EMT6 multilayer transport parameters for DACA (Table 2) lead to a predicted receiver concentration that is 30% of the equilibrium value by 6 h under these same conditions (unstirred donor compartment of 0.5 ml with an 8-ml stirred receiver).

The finding that DACA is reduced by some mammalian cell lines (V79, AA8, SiHa) to the corresponding acridan (DACA-2H, Fig. 1) was unexpected. This is, to our knowledge, the first demonstration of metabolic hydrogenation of an acridine ring in mammalian cells. The failure to detect this metabolite in earlier investigations of the biotransformation of DACA (Robertson et al., 1993; Schofield et al., 1999) may relate to the potential for reoxidation of the acridan to DACA during sample preparation. Preliminary data (data not shown) indicate that DACA-2H is less cytotoxic than DACA, as expected given the anticipated lowering of DNA binding affinity with loss of aromaticity of the acridine ring.

The rapid diffusivity of DACA through V79 MCL initially prompted us to consider whether the DACA-2H metabolite might act as a less DNA-affinic carrier form of DACA with rapid metabolic interconversion (futile cycling) between the two species. This would have parallels with the reported activity of the acridan of acriflavin as a carrier of the acridine across microbial cell walls (Adamus et al., 1998). Investigation of diffusion of DACA-2H (c_1,0 13 µM) through V79 MCL (data not shown) demonstrated that DACA-2H does indeed diffuse even more readily than DACA (primarily because of its greater metabolic stability in this line). However, we were not able to demonstrate reoxidation to DACA in MCL or cells. In addition, comparison of DACA flux through V79 MCL (Fig. 6F) and EMT6 MCL (Fig. 6C), which do not metabolize DACA to DACA-2H, showed that this metabolism impedes rather than assists net DACA transport.

An important objective of the present study was to make a quantitative comparison between DACA and its more basic (and more DNA affinic) analog DAPA. The published MCL penetration data for DAPA were obtained using an experimental system in which only the receiver compartment was stirred, with an agar gel in the donor compartment to prevent mixing (Hicks et al., 1997). When we reevaluated DAPA flux through V79 MCL using the new dual stirred chamber system (Fig. 2), the kinetics with which the compound appeared in the receiver compartment was indistinguishable from that predicted (for the new conditions) using the transport parameters determined in the previous system (Fig. 5F). Thus, the MCL flux kinetics is equivalent in the two experimental systems (as also shown by the similar bare support membrane porosity estimates, and similar urea diffusion coefficients in V79 MCL). However, with the ability to measure the concentration profile in the donor as well as receiver compartment, it became evident that the previous mathematical model was not optimal. Fitting both compartments simultaneously required an increase in D_m (i.e., faster diffusion of free DAPA) and _i (i.e., increased overall binding in the MCL). This demonstrates the value of measuring concentrations in both compartments to constrain the mathematical modeling.

The more rapid transport of DACA than DAPA suggests physicochemical features that may be important to the penetration kinetics of basic DNA intercalators. The proportion of unionized (lipophilic) free base at pH 7.4 is higher for DACA than DAPA (0.5 versus 0.02%), which may contribute to the higher D_m for free DACA than DAPA, and thus to the improved penetration kinetics of the former. The second advantage of DACA lies in a reduced binding site barrier relative to DAPA, as shown by lesser sequestration in cells. This may reflect the higher DNA affinity of DAPA, but is probably more dependent on differing propensities of the two compounds for entrapment in acidic vesicles. The latter are clearly major sites of loss for both compounds, as demonstrated by effects of NH₄Cl on cellular uptake (Fig. 3) and MCL flux (Fig. 6F), but the effect of NH₄Cl on flux in V79 MCL is greater for DAPA (Hicks et al., 1997) than DACA (this study). This indicates that entrapment of the dication at low pH in endosomes is the major impediment to penetration, and that the lower binding site barrier of DACA is related to decreased endosomal entrapment because of the lower basicity of its chromophore. These factors (higher D_m for free drug and lower sequestration) combine to allow DACA to penetrate much more rapidly than DAPA. Although this analysis is qualitatively reasonable, some caution is warranted in that cellular uptake of weak bases is a function of extracellular pH, and the effects of pH gradients in MCL have not been incorporated in the current model, which treats the MCL as a homogeneous phase with binding sites. The justification for this homogeneous model, and limitations of assuming a constant mean pH, have been previously discussed (Hicks et al., 1997).

The implications of the measured transport parameters for DACA were explored by simulating transport in the extravascular compartment of a tumor. This required knowledge of the plasma pharmacokinetics of DACA, determined previously for BDF₁ mice with Lewis Lung tumors (Paxton et al., 1993). It also required specification of the geometry and diffusion distances involved, which are not known. To overcome this problem we assumed a simplified, uniform, planar geometry, and solved the reaction-diffusion equation (eq. 1) using the measured MCL transport parameters to predict the average tumor concentration of DACA. This provided a good fit to the measured average concentrations in Lewis Lung tumors when we assumed the maximum diffusion distance (to the center of the planar slab) to be 250 µm. This gives confidence that the parameters measured in the MCL model are physiologically relevant. An even better fit would be obtained by assuming an ensemble of diffusion distances (giving faster efflux at short times followed by slower efflux from more distant regions), which would also be more physiologically appropriate.

This micropharmacokinetic model for the extravascular compartment of tumors demonstrates that DACA distributes relatively well, although the maximum concentration 125 µm from a vessel is 2.5-fold lower, and occurs 15 min later, than in plasma (Fig. 7B). The implications of this delayed and "damped" pharmacokinetics depend on the pharmacodynamic model assumed. If killing by DACA were linearly proportional to AUC, which is essentially independent of diffusion distance (at large distances, slow efflux compensates for slow influx), we would predict no impediment to cytotoxic activity in cells distant from blood vessels. However, the model that best describes DACA cytotoxicity in low cell density cultures (Fig. 8A; Results) leads to the prediction of a small decrease in killing efficiency with distance as a result of impeded drug supply (Fig. 8B). Given that the fraction of cycling cells generally decreases with distance into the extravascular compartment in tumors (Tannock, 1968; Kennedy et al., 1997) an additional loss of sensitivity due to resistance of noncycling cells to DACA (Finlay and Baguley, 1989) is likely to be superimposed on this micropharmacokinetic effect.

In conclusion, this study demonstrates that relatively small structural differences between analogs can lead to substantial differences in extravascular transport characteristics, and that basic DNA intercalators are not necessarily poor diffusers in tumor tissue. The demonstration of efficient transport of DACA in this study supports the earlier hypothesis (Denny et al., 1987) that one of the reasons for its superior antitumor activity relative to more basic acridines is that it distributes relatively well in tumors. It also points to the potential for using DACA-like chromophores as DNA targeting moieties for delivering other agents to cells distant from blood vessels in solid tumors.