Brain tumor patients face a poor prognosis despite significant advances in tumor imaging, neurosurgery and radiation therapy. Potent chemotherapeutic drugs fail when used to treat brain tumors because biochemical and physiological barriers limit drug delivery into the brain.

A retro-convection enhanced delivery (R-CED) method has been developed to improve the entry of intravenously administered therapeutics within solid brain tumors. R-CED uses an osmotic gradient to withdraw brain interstitial fluid in a controlled manner via an implanted microdialysis catheter. The transmembrane osmotic gradient increased the local tissue specific gravity in normal brain and induced movement of small proteins and nanoparticulates from the blood into normal brain and an orthotopic 9L tumor. The magnitude of the R-CED effect decreased to control values within six hours and did not cause acute anatomical damage beyond that of probe insertion.

A one hour R-CED treatment applied immediately after intravenous injection of liposomal doxorubicin ten days after tumor implantation showed no therapeutic effect compared to animals treated with intravenous PBS alone. The result suggests that R-CED may be more efficacious by modifying the R-CED probe design, increasing the R-CED treatment time, and improving the chemotherapeutic type and dosing schedule.

A mathematical model has been developed to predict the size of the R-CED affected region while varying the interstitial and vascular hydraulic conductivity and the fluid removal rate. The model enables the prediction of non-spherical flow patterns during simultaneous R-CED and CED or dual-probe CED via two catheters, guiding probe placement to achieve drug patterning and enabling wide distribution of therapeutics while sparing normal tissues. The alteration of solute flow was confirmed for dual-probe CED in vitro, but was difficult to visualize in vivo, where tissue inhomogeneities altered the flow from the idealized model prediction.

These studies demonstrate that R-CED is a viable and well-tolerated technique to enhance the distribution of systemically administered drugs in both the normal tissue-tumor margin as well as the central tumor core. They also highlight the opportunities to modify the R-CED design and enable future drug distribution and antitumor efficacy studies in larger tumors and possibly humans.

The importance of therapeutic recombinant proteins in medicine has led to a variety of tactics to increase their circulation time or to enable routes of administration other than injection. One clinically successful tactic to improve protein circulation and delivery is to fuse the Fc-domain of immunoglobulin G (IgG) to therapeutic proteins so that the resulting fusion proteins interact with the neonatal Fc receptor (FcRn). Although successful, Fc-fusion proteins significantly increase molecular weight thereby restricting tissue penetration, decrease protein function, and are limited to mammalian expression systems. As an alternative to grafting the high molecular weight Fc-domain to therapeutic proteins, we have modified their N- and/or C-terminus with a short peptide sequence that interacts with FcRn. Our strategy was motivated by results from Mezo and coworkers [Mezo et al. (2008) PNAS 105:2337-42] who identified peptides that compete with human IgG for FcRn. The small size and simple structure of the FcRn binding peptide (FcBP) allows for expression of FcBP fusion proteins in E. coli and results in their pH-dependent binding to FcRn with an affinity comparable to that of hIgG1. The FcBP fusion proteins are internalized, recycled and transcytosed across cell monolayers that express FcRn. Although FcBP fusion proteins mimic the human IgG1 interaction with human FcRn in vitro, the half-life of FcBP fusion proteins in wild type C57BL/6J and human FcRn transgenic mice is independent of FcRn binding, whereas the half-life of human and mouse IgG1 correlate with their species matched FcRn-binding affinity. These results promoted a detailed investigation into potential factors that may contribute to the lack of correlation between the in vitro and in vivo results including: serum stability, serum competition, renal clearance, and the function of the hybrid human-mouse FcRn/B2m receptor; however, no single variable explains the FcRn-independent half-life of FcBP fusion proteins. We speculate that an additional component(s), in collaboration with FcRn, regulates IgG homeostasis in vivo and is not replicated by FcBP fusion. If such a novel FcRn regulatory component exists, it would have a significant impact on FcRn biology and provide new opportunities to engineer IgG or alternative FcRn-targeted molecules for enhanced serum persistence.

The hypothesis of my thesis project was that targeting the extracellular matrix of a solid tumor for a local generation of a cytotoxic drug could inhibit tumor growth and progression. To test this hypothesis of the matrix attachment therapy (MAT), a gene therapy and recombinant protein therapy were pursued. For the gene therapy, I cloned a mammalian expression vector containing the extracellular domain of CD44 (sCD44), yeast cytosine deaminase (CD), and E.coli uracil phosphoribosyltransferase (UPRT) genes to transfect tumor cells in vitro. The sCD44::CD::UPRT fusion gene construct was designed to secrete the expressed fusion protein and generate cytotoxic drugs to cause cell deaths in the presence of the prodrug 5-fluorocytosine (5-FC). This gene therapy strategy did not result in a potent cytotoxic effect on cells that expressed the fusion gene due to low expression of the fusion genes in transfected cells.

For the protein therapy, a recombinant fusion protein containing TSG-6 Link (Link) and the CD enzyme was expressed in E.coli and purified to characterize the functions of the LinkCD fusion protein. LinkCD exhibits Km of 0.33 mM and Vmax of 14 µM/min/µg for the conversion of 5-FC to 5-FU. LinkCD can bind to a hyaluronan oligomer (12-mer) at a KD of 55 µM at pH 7.4 and a KD of 5.32 µM at pH 6.0 measured using surface plasmon resonance (SPR). To evaluate the anti-tumor effect of LinkCD/5-FC combination therapy in vivo, mice received intratumoral injections of LinkCD on days 11 and 14 after C26 tumor implantation and drinking water containing 10 mg/mL of 5-FC starting on day 11. Animals that received LinkCD/5-FC treatment showed significant tumor size reduction and increased survival compared to the CD/5-FC treatment group. Other treatment groups that were unable to produce 5-FU had no effect on the tumor growth despite receiving the fusion protein that contained the Link domain. The results strongly suggest that a treatment regime consisting of a fusion protein containing the Link domain, the active CD enzyme, and the prodrug 5-FC are required to produce an anti-tumor effect. Intravenous administration of LinkCD to animals tumored with C26 that received 5-FC in the drinking water did not result in a therapeutic effect, possibly due to a short circulation time of LinkCD. I created chimeras of the CD44 Link domain in silico that could potentially be used to increase the circulation time of LinkCD in the blood. Thus, MAT is a promising alternative to antibody-directed prodrug enzyme therapy approach for cancer treatment.