We have previously reported the synthesis of SMA41, a unimolecular combination of an epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor (TKI) of the quinazoline class and a DNA-damaging monomethyltriazene termed “combimolecule”. Hydrolysis of 1-[4-(m-tolylamino)-6-quinazolinyl]-3-methyltriazene (SMA41) gives rise to an intact TKI [6-amino-4-(3-methylanilino)quinazoline; SMA52] capable of inhibiting epidermal growth factor (EGF)-induced EGFR autophosphorylation and a DNA-targeting methyldiazonium species. Herein, we showed that SMA41 blocked EGF-induced EGFR autophosphorylation by an irreversible mechanism, suggesting that it may covalently damage the receptor in these cells. More importantly, this was associated with significant inhibition of mitogen-activated protein kinase activation in A431 cells. In cells treated with [14C]SMA41, radio-high-performance liquid chromatography detection of both N7- and O6-methylguanine revealed an almost complete repair of the O6-methylguanine lesions and a greater tolerance of the N7-methylguanine adducts 24 h post-treatment. In contrast to temozolomide (a cyclic triazene used in the clinic) and the reversible inhibitor SMA52, SMA41 induced significant cell cycle arrest in S, G2, and M phases 24 h after a 2-h drug exposure. Furthermore, in vivo studies demonstrated that SMA41 was well tolerated. At 200 mg/kg, it showed approximately 2-fold greater antiproliferative activity than SMA52 in A431 cells implanted in immunocompromised SCID mice. These results suggest that the binary targeting properties of SMA41 are associated with a binary cascade of events in the cells that seem to culminate into significant growth inhibition in vitro and in vivo.
The overexpression and dysregulation of tyrosine kinase is commonly observed in a large number of cancers. For example, the epidermal growth factor receptor (EGFR) is often overexpressed in breast, ovarian, and prostate tumors, and this can result in the formation of autocrine loops associated with poor patient prognosis and aggressive tumor growth (Moscatello et al., 1995; Wosikowski et al., 1997; Sherwood et al., 1998; Suo et al., 2002). Also, 40% of glioblastomas with EGFR amplification express EGFRvIII, a mutant form of EGFR that lacks a portion of the extracellular ligand-binding domain (Lammering et al., 2003; Charkravarti et al., 2004). Drugs have been designed to target EGFR and to block the signaling cascades mediated by this receptor. Of these compounds, two belonging to the quinazoline class are currently in phase III clinical trials (Allen et al., 2002; Baselga et al., 2002; Hirata et al., 2002; Wakeling et al., 2002). Iressa (AstraZeneca Pharmaceuticals LP, Wilmington, DE) (1), a water-soluble quinazoline identified as a highly effective competitive inhibitor of EGFR tyrosine kinase, has recently been approved for the therapy of lung cancers in Japan.
The inhibitors of EGFR tyrosine kinase act by competitively binding to its ATP site, thereby blocking subsequent activation of downstream signaling cascades, including mitogen-activated protein (MAP) kinase and the transcription of genes such as c-fos that are associated with cell proliferation (Albanell et al., 2001; Mitsui et al., 2001; Nelson and Fry, 2001; Magné et al., 2002). Due to the marked reversibility of the action of Iressa and related quinazolines, prolonged and repeated doses are required for the induction of sustained antitumour activity, and now the combination of this drug with classical cytotoxic agents has become a common approach to increase the potency of EGFR-directed therapy (Ciardiello et al., 2000; Sirotnak et al., 2000). Recent studies demonstrated significant synergy between Iressa and the anti-metabolite 5-fluorouracil (Magné et al., 2002). Similar results have been reported with DNA-damaging agents such as cisplatin, doxorubicin, and cyclophosphamide (Ciardiello et al., 2000; Sirotnak et al., 2000).
Another approach to overcome the reversible effects of EGFR TK inhibitors was the development of novel compounds such as 2, containing an α,β-unsaturated acrylamide designed to react with Cys773 of the ATP site of the receptor, thus inducing irreversible inhibition of EGFR tyrosine kinase (Fry et al., 1998). Although these compounds were shown to be more potent than Iressa in vivo, inhibition of a single target may not suffice to reduce the risk of relapse after chemotherapy, given common presence of receptor heterogeneity within solid tumor cell populations.
Recently, we designed a novel tumor targeting strategy termed “combi-targeting” that sought to confer a second target (e.g., DNA) to EGFR inhibitors with the purpose of adding an alternative cytotoxic property to the often cytostatic activity associated with inhibition of tyrosine kinase-mediated signaling (Matheson et al., 2001, 2003a; Brahimi et al., 2002; Qiu et al., 2003). This led to the design of small molecular system termed “combi-molecules” capable of mimicking classical combination therapy by mixing multiple mechanisms of antitumour action. Specifically, the combi-targeting concept postulates that a combi-molecule, TZ-I (Scheme 1), designed to behave not only as an inhibitor of a tyrosine kinase but also to be hydrolyzed to another inhibitor and a DNA-damaging species, TZ, should lead to more sustained antitumour activity than the reversible receptor inhibitor alone. If TZ-I is stable enough to penetrate intact cells, it may not only bind to the ATP site (see TZ-I-EGFR) but also react with amino acid residues of the latter, leading to a covalently damaged receptor (see TZ-EGFR). Thus the TZ-I was designed to be a single molecule encompassing all currently investigated approaches to effective EGFR-based chemotherapy with small molecules: 1) inhibition of EGFR TK mediated signaling, 2) combination of EGFR TKI with cytotoxic agents, and 3) irreversible inhibition of EGFR tyrosine kinase activity.
We recently reported the synthesis of our first combi-molecule, SMA41, containing a triazene tail and a quinazoline head (Matheson et al., 2001, 2003a). As outlined in Scheme 1, SMA41 (TZ-I) was found to be hydrolyzed under physiological conditions to generate the methyldiazonium ion that can damage DNA as well as an intact TKI SMA52 (I) (Matheson et al., 2001, 2003a). Additionally, the molecule was shown to block EGFR autophosphorylation and to induce significant levels of DNA damage in EGFR-expressing cells. Moreover, it demonstrated more than 8-fold greater antiproliferative activity than a combination of temozolomide, a clinical prodrug of the methyldiazonium (TZ) plus SMA52 (I) in vitro (Matheson et al., 2001). More recently, using a 14C-labeled SMA41 and the fluorescence properties of SMA52, we demonstrated that TZ was indeed bound to DNA and SMA52 was distributed in the perinuclear region of the cells (Matheson et al., 2003b, 2004). Here, we analyzed the biochemical effects of each element of the dual targeting principles of the TZ-I and study the feasibility of the combi-targeting concept in vivo.
Materials and Methods
Drug Treatment. SMA41 and SMA52 were synthesized in our laboratories according to published procedures (Matheson et al., 2001, 2003b). Temozolomide was provided by Schering Plough (Kenilworth, NJ). In all assays, drugs were dissolved in dimethyl sulfoxide and subsequently diluted in sterile RPMI 1640 medium containing 10% fetal bovine serum (Wisent, St. Bruno, Quebec, Canada) immediately before the treatment of cell cultures. In all assays, the concentration of dimethyl sulfoxide never exceeded 0.2% (v/v).
Cell Culture. The human tumor cell line A431 (American Type Culture Collection, Manassas, VA) was maintained in RPMI 1640 medium supplemented with fetal bovine serum (10%), gentamycin (50 mg/ml), and HEPES (12.5 mM) (Wisent). Cells were maintained in a monolayer culture at 37°C in a humidified environment of 5% CO2, 95% air and were kept in logarithmic growth by harvesting with a trypsin-EDTA solution containing 0.5 mg/ml trypsin and 0.2 mg/ml EDTA and replating before confluence. In all assays, the cells were plated at least 24 h before drug administration.
Flow Cytometry for Cell Cycle Analysis. Cells were grown in six-well plates in a monolayer until confluence and subsequently exposed to each compound for 2 h at 37°C. Thereafter, they were allowed to recover in drug-free media for either 24 or 48 h, harvested in trypsin-EDTA, collected by centrifugation, and resuspended in PBS. Samples were treated with Vindelov's solution (Vindelov, 1977) (0.01 M Tris-base, 10 mM NaCl, 700 U of RNase, 7.5 × 10-5 M propidium iodide, and 0.1% Nonidet P-40) for 10 min at 37°C, vortexed, and fluorescence was quantitated using FACScan (BD Biosciences, San Jose, CA).
Alkaline Comet Assay for Quantitation of DNA Damage. A modified alkaline comet assay technique was used to quantitate DNA damage induced by SMA41, SMA52, and temozolomide (Yang et al., 1999; McNamee et al., 2000). A431 cells were exposed to drugs for 2 h and either harvested immediately with trypsin-EDTA or allowed to recover in drug-free media for 24 h. The cells were subsequently collected by centrifugation and resuspended in PBS. The resulting cell suspension was diluted to approximately 106 cells and mixed with agarose (1%) in PBS at 37°C in a 1:10 dilution. The gels were cast on Gelbond strips (Mandel Scientific, Guelph, ON, Canada) using gel casting chambers, as described previously (Matheson et al., 2001), and then immediately placed into a lysis buffer [2.5 M NaCl, 0.1 M tetra-sodium EDTA, 10 mM Tris-base, 1% (w/v) N-lauryl sarcosine, 10% (v/v) dimethyl sulfoxide, and 1% (v/v) Triton X-100]. After being kept on ice for 30 min, the gels were gently rinsed with distilled water and then immersed in a second lysis buffer (2.5 M NaCl, 0.1 M tetra-sodium EDTA, and 10 mM Tris-base), containing 1 mg/ml proteinase K for 60 min at 37°C. Thereafter, they were rinsed with distilled water, incubated in alkaline electrophoresis buffer for 30 min at 37°C, and electrophoresed at 300 mA for 60 min. The gels were subsequently rinsed with distilled water and placed into 1 M ammonium acetate for 30 min. They were further soaked in 100% ethanol for 2 h, dried overnight, and subsequently stained with SYBR Gold (1/10,000 dilution of stock; Molecular Probes, Eugene, OR) for 20 min. For evaluation of comets, DNA damage was assessed using the Tail Moment parameter (e.g., the product of the distance between the barycenters of the head and the tail of the comet). A minimum of 50 cell comets were analyzed for each sample, using ALKOMET version 3.1 software, and values are an average of tail moments for the entire cell population.
Detection of DNA Adducts Using HPLC. A431 cells were grown to confluence in T75 flasks and subsequently exposed to 15.4 μCi of [14C]SMA41 for either 2 h or 2-h exposure followed by 24-h recovery in drug-free media. DNA was isolated using the Wizard genomic DNA purification kit (Promega) as per the manufacturer's directions. After hydrolysis with 0.1 N HCl at 80°C for 30 min, samples were neutralized using 0.1 N NaOH and run on a C18 HPLC column (250 × 4.6 mm) at pH 3 (sodium citrate, sodium acetate, pH adjusted with HCl) in a 9:1 methanol/water ratio (Kaur and Halliwell, 1996). Fractions were collected from the column every 0.5 min and counted using liquid scintillation in a Wallac 1219 Rackbeta counter in Universol liquid scintillation cocktail (ICN, Montreal, Canada). Standards of N7-methylguanine (Sigma, Oakville, ON, Canada) and O6-methylguanine (Sigma) were used to calibrate the column.
Western Blotting for Irreversibility. Cells were grown in sixwell plates (1 × 106) in media containing 10% serum until confluence, washed three times with sterile PBS, and grown in phenol red-free RMPI 1640 medium without serum for 24 h. SMA41 or SMA52 was given for 90 min, washed with PBS, and grown in serum-free media for 2 h, washed again, and grown in serum-free media for another 4 h. Cells were subsequently stimulated with 100 ng/ml epidermal growth factor for 10 min, and protein lysates were prepared as described previously (Matheson et al., 2001). In parallel, another set of samples was prepared without the 6 h growth in serum-free media before stimulation. Samples were subjected to 10% SDS-polyacrylamide gel electrophoresis, transferred to a polyvinylidene difluoride membrane, and probed with PY20 anti-phosphotyrosine antibodies (Neomarkers, Fremont, CA) and anti-α-tubulin (Neomarkers, Fremont, CA). Proteins were detected using goat antimouse HRP-conjugated antibodies and enhanced chemiluminescence (Amersham Biosciences UK, Ltd., Little Chalfont, Buckinghamshire, UK). The membrane was subsequently stripped of antibody, and reprobed using anti-EGFR antibodies (Neomarkers) and detected using enhanced chemiluminescence. Protein bands were scanned using Syngene software imaging system (Syngene, Cambridge, UK).
Western Blotting for MAP Kinase. A431 cells were grown in six-well plates until confluence, washed three times with PBS, and grown in serum-free media for 24 h. The next day, SMA41 was administered for 2 h, after which cells were stimulated with 100 ng/ml epidermal growth factor for 20 min. Cells were collected, and protein lysates prepared as described previously (Matheson et al., 2001). Samples were separated on SDS-polyacrylamide gel, and transferred to a polyvinylidene difluoride membrane. The membrane was incubated with primary antibodies anti-phospho-MAP kinase (Cell Signaling Technology Inc., Beverly, MA) for determination of phosphorylated MAP kinase. Thereafter, blots were incubated with HRP anti-rabbit antibody (1:1000 dilution; Cell Signaling Technology Inc.), and the bands were visualized with an enhanced chemiluminescence system (Kirkegard & Perry Laboratories, Gaithersburg, MD). Membranes were stripped and reprobed with anti-MAP kinase (Cell Signaling Technology Inc.) to detect corresponding protein levels and subsequently with HRP anti-rabbit antibody (1:1000 dilution; Cell Signaling Technology Inc.).
Mouse Xenograft Studies (A431 Cells). SCID mice were maintained as per McGill animal safety protocols. Mice were subcutaneously injected with 1.5 × 106 cells suspended in 0.2 ml of PBS into the flank. Treatments began when tumors became palpable. The animals were placed into five treatment groups of six mice each, plus one control group given the vehicle (10% dimethyl sulfoxide, 90% H2O). The treatment groups were given SMA52 or SMA41 (100 and 200 mg/kg) every 5 days for 25 days, for a total of five injections. Tumor burden was measured twice a week, and tumor volume was calculated using the formula V = (tumor width)2 × tumor length × 0.5. Statistical analysis (e.g., unpaired t test) was performed using the GraphPad software package (GraphPad Software Inc., San Diego, CA).
Characterization and Repair of DNA Adducts Induced by the Methyldiazonium (TZ). Previous studies using the comet assay demonstrated that SMA41 induces dose-dependent DNA damage in A431 cells (Matheson et al., 2001). Here, we studied the repair of these DNA breaks by comparing DNA damage induced shortly after exposure with those observed 24 h post-treatment (Fig. 1, a–c). DNA damage induced by both SMA41 and temozolomide were repaired; however, levels of DNA damage remained high 24 h after SMA41 treatment. From the single-cell microelectrophoresis (comet) assay that is designed to detect alkali-labile lesions, we surmise that DNA strand breaks may originate from N7-alkylguanine adducts (Bignami et al., 2000). Nonetheless, O6-alkylguanine is the most cytotoxic DNA adduct induced by alkyltriazenes of the same class as temozolomide (Baer et al., 1993; Kokkinakis et al., 1997; Lage et al., 1999). Thus, to have a clearer insight into the mechanism of action of SMA41, we attempted to detect and quantitate O6- and N7-alkylguanine adducts in the A431 cell line.
Using SMA41 14C-labeled at the 3-methyl group (Matheson et al., 2003b, 2004), DNA adducts were characterized under a drug treatment schedule similar to that of the comet DNA repair assay. Two hours after treatment (Fig. 2a), HPLC analysis showed a major peak at 4.6 min that corresponded to N7-methylguanine and a minor peak at 7.6 min, indicating the presence of O6-methylguanine. The assignment was confirmed by peak matching with our standards. As shown in Fig. 2b, whereas the N7-methylguanine was still observed (although at a lower intensity) after 24 h, the O6-methylguanine peak disappeared, confirming the repair of the latter adduct and persistence of the N7-methylguanine lesions. These results suggest that, in contrast to the repair of O6-methylguanine adducts, the repair of the N7-methylguanine lesions is not completed even 24 h after drug treatment.
Effects of SMA41 on the Cell Cycle. In cell lines exposed to SMA41, the marked DNA repair activity observed 24 h post-treatment was accompanied by a significant dose-dependent increase in the number of cells accumulated at various phases of the cell cycle. As shown in Table 1, when cells were given SMA41 for 2 h and allowed to recover for 24 or 48 h, significant cell cycle arrest was observed in S and G2M phases at all doses. In contrast, no significant cell cycle perturbations were observed for temozolomide or SMA52 over the whole dose range. Cell cycle arrests induced by SMA41 were partially reversed 24 h later.
Inhibition of EGF-Induced Signaling. The combi-targeting concept postulates that the generated TZ (e.g., methyldiazonium) may react with the Cys773 to give a covalently modified and inactivated receptor. Thus, we hypothesized that an at least partially irreversible inhibition of EGFR autophosphorylation should be observed. Autophosphorylation activity was measured in A431 cells after multiple washouts. As shown in Fig. 3a, at a 25 μM dose, SMA52 lost almost 90% of its ability to block EGF-induced EGFR autophosphorylation. In contrast, the combi-molecule SMA41 retained approximately 90% of its initial EGFR inhibitory effect. Moreover, analysis of the effects of SMA41 on downstream signaling showed that blockade of EGFR was accompanied by a dose-dependent inhibition of MAP kinase (extracellular signal-regulated kinase 1,2) phosphorylation (Fig. 3b).
In Vivo Studies. Having analyzed the biological effects of DNA damage and blockade of EGFR autophosphorylation induced by the combi-molecule in vitro, we determined its potency in a mouse xenograft model using A431 in SCID mice. A tolerance study was first performed by weight loss monitoring that showed no significant changes in the weight of the animals over the whole dose range (25–100 mg/kg) when administered once every 3 days over a period of more than 20 days (Fig. 4). Because of the poor solubility of SMA41 and SMA52, we performed the efficacy study at 100- and 200-mg/kg doses. As shown in Fig. 5, a and b, a dose-dependent increase in tumor growth inhibition was induced by SMA41 but not by SMA52. Temozolomide, being water-soluble, was extremely cytotoxic at these doses (data not shown) and was therefore excluded from the comparison. Tumor growth was significantly delayed in mice given 200 mg/kg SMA41 compared with those given SMA52 or vehicle only (p < 0.05). Indeed, at 21 days post-treatment SMA41 showed approximately 2-fold greater growth inhibition than SMA52 at 200 mg/kg (p < 0.05) (Fig. 6), indicating that the triazene tail is a strong contributor to the potency of the combimolecule.
Previous studies demonstrated the mixed EGFR/DNA targeting properties of SMA41, the first molecular probe for the combi-targeting postulates. In addition to being able to bind to the EGFR ATP binding site on its own, this combi-molecule was designed to generate another inhibitor (SMA52) and a DNA-damaging methyldiazonium species (Brahimi et al., 2002; Matheson et al., 2001, 2003a). Considerable evidence has emerged to confirm the ability of SMA41 to damage DNA in whole cells. However, the exact nature of this damage remained to be determined. Methyltriazenes of the same class as SMA41 are known to induce N7- and O6-methylguanine DNA lesions (Bignami et al., 2000). Although the N7-methylguanine adducts are the most abundant (at least 70% of total DNA adducts), the O6-methylguanine that accounts for 5% of the total damage induced by methyltriazenes is the most cytotoxic (Baer et al., 1993; Kokkinakis et al., 1997; Tentori et al., 1999; Bignami et al., 2000). Tumor cells harboring AGT are resistant to methylating agents of the triazene class (Lage et al., 1999; Bignami et al., 2000; Kokkinakis et al., 2001). In this study, the characterization of DNA adducts induced by 14C-labeled SMA41 revealed the formation of both N7- and O6-alkylguanine lesions. The complete disappearance of the peak corresponding to O6-alkylguanine 24 h post-treatment indicated that this lesion was repaired, an observation that is consistent with the AGT-positive status of A431. In contrast, significant levels of N7-alkylguanine adducts were apparent 24 h post-treatment. This is consistent with the alkaline comet assay data that showed only a partial repair of alkali-labile lesions 24 h after treatment. Thus, the cytotoxic contribution of DNA damage induced by SMA41 may not depend on the O6-alkylguanine adducts, but rather on other types of lesions such as the persistent N7-methylguanine or perhaps N3-methyladenine. The levels of the latter adduct were too low to be detected by our radio-HPLC methods. More importantly, DNA damage and repair activities induced by SMA41 caused significant cell cycle arrest in S and G2M phases 24 h post-treatment. By contrast, temozolomide, which is known to inflict similar types of DNA lesions in tumor cells (Lage et al., 1999), induced only minor cell cycle perturbations at equimolar doses. Given that SMA41 (as demonstrated in earlier studies) is hydrolyzed within 1 h into an almost stoichiometric quantity of SMA52 and a methyldiazonium species at the same proportion as temozolomide (Matheson et al., 2001), one would expect it to induce as little cell cycle perturbation as temozolomide. Thus, the markedly distinct cell cycle response profile induced by SMA41 may be due to its mixed EGFR/DNA targeting properties. Blockade of EGFR-mediated downstream signaling may deprive the cells from the expression of critical genes required to rescue them from the cytotoxic effects of DNA lesions, leading to arrests at multiple phases of the cell cycle. This argument can be further corroborated by the ability of SMA41 to block EGF-induced MAPK activation in a dose-dependent manner.
One of the premises of the combi-targeting concept was that SMA41 (TZ-I) can alkylate the receptor to form a TZ-EGFR adduct (Scheme 1). Covalent damage to the receptor may induce its irreversible inhibition. As demonstrated, unlike the reversible inhibitor SMA52 (I), SMA41 could induce irreversible inhibition of EGFR. These results support the hypothesis that it may covalently modify the receptor, perhaps through methylation of cysteine 773. A similar type of reaction has already been reported for irreversible inhibitors of the 6-acrylamidoquinazoline class (Fry et al., 1998).
We have now demonstrated that the mixed EGFR/DNA targeting properties of the TZ-I SMA41 triggers a binary cascade of events: those associated with DNA damage, and those related to EGF-dependent phosphorylation in A431 carcinoma cells, a cell line that expresses the DNA repair enzyme AGT and that aggressively proliferates via a transforming growth factor α/EGFR-mediated autocrine loop. Although these events translate into significant antiproliferative activity in vitro (Matheson et al., 2001), the demonstration of the relevance of the combi-targeting principles to the clinical therapy of EGF-dependent carcinomas required an in vivo model. The choice of controls for comparison was rather difficult because there is no known quinazoline-triazene in the clinic and the reversible EGFR inhibitor Iressa is structurally too different from SMA41 to be used as a proper reference drug. Temozolomide was also too different and too water soluble to be compared with SMA41. We therefore chose to compare the potency of SMA41 with that of SMA52 since its structure differs from that of SMA41 only by the NNMe moiety. SMA52 and SMA41 were well tolerated in the dose range used, with no significant weight loss over the test period. The fact that SMA41 showed significantly more potent antiproliferative activities than SMA52 (p < 0.05) suggests that the addition of a DNA damaging tail to the quinazoline backbone confers significant antiproliferative advantage. It is also important to note that SMA52, being more water-soluble than SMA41, may have achieved a higher plasma concentration, indicating that an even greater potency could have been observed for SMA41, if it had a similar degree of water solubility.
The results conclusively demonstrate that the in vivo activity of the TZ-I is significantly superior to that of the I alone. This study also permitted the verification of the primary postulate according to which a TZ-I (Scheme 1), designed to behave not only as an inhibitor of a tyrosine kinase but also to be hydrolyzed to another inhibitor I and a DNA-damaging species, TZ, should be more potent than the reversible inhibitor alone. Thus, we presented herein the first evidence that, in contrast to previous Michael acceptor-based approaches to the design of irreversible inhibitors of EGFR, increased potency is achievable by conferring an additional (non-EGFR) targeting property to the structure. Further studies are ongoing to increase the water solubility of SMA41 with the purpose of enhancing its bioavailability and its efficacy in vivo.
We thank Nicole Teoh for assistance with HPLC analysis.
This study was funded by National Cancer Institute of Canada Grant 013377.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitor; MAP, mitogen-activated protein; SMA41, 1-[4-(m-tolylamino)-6-quinazolinyl]-3-methyltriazene; SMA 52, 6-amino-4-(3-methylanilino)quinazoline; PBS, phosphate-buffered saline; HPLC, high-performance liquid chromatography; HRP, horseradish peroxidase; AGT, O6-alkylguanine transferase.
- The American Society for Pharmacology and Experimental Therapeutics