This kit was found to be highly sensitive for both rhTIMP-1 and PEG20K-TIMP-1, and to have minimal cross-reactivity with mouse TIMP-1. The dynamic range for detection of rhTIMP-1 and PEG20K-TIMP-1 was 39?500 pg/ mL; these concentrations were achieved in plasma samples after an additional 50?00-fold dilution in the assay (after the initial 10fold dilution). Standard curves were fit to data for 8 rhTIMP-1 concentrations run in duplicate, using a four-parameter sigmoidal (logistic) model (R2.0.99), and concentrations in plasma were determined from duplicate assays by nonlinear interpolation from the standard curve, using Prism version 4.0 (GraphPad Software, La Jolla, CA, USA). The data were displayed as a semilogarithmic plot and fitted to the equation for two phase exponential decay using Prism version 4.0 to obtain the best fit values for distribution and elimination; values were excluded for rhTIMP-1 for the 30 min and 45 min time points, where the measured values exceeded the quantifiable range of the assay.active site cleft of the protease and directly coordinates to the catalytic zinc (Fig. 1); this interaction is the basis for inhibition of MMPs by TIMPs. Any modification of the natural TIMP Nterminus, even extension by a single Ala residue, will destroy antiprotease activity [46], and mutations that disrupt the Nterminal disulfide bond between Cys-1 and Cys-70 are similarly destructive [47,48]. To avoid the risk of N-terminal PEGylation, and to direct PEG attachment to a site distant from the TIMPMMP interface, our first approach was to introduce a single unpaired Cys residue near the flexible C-terminus of TIMP-1, opposite the MMP-binding face, for reaction with a thiol-reactive PEG.
Conjugation of PEG to rhTIMP-1 via an Introduced Cys Residue
To facilitate site-specific PEGylation of rhTIMP-1, we made four mutants each introducing a Cys residue near the C-terminus (Fig. 1) and tested expression levels in the HEK 293E mammalian expression system, finding highest expression for the R180C mutant (Fig. S1A). The rhTIMP-1-R180C was produced in larger scale, purified, and analyzed under nonreducing conditions, which showed that while some disulfide dimers were present, the major species was monomeric (Fig. S1B), with the introduced Cys presumed to be available for modification. An mPEG-maleimide reagent of molecular weight 30K was selected with the rationale
Mammary Orthotopic Xenograft Study
A mixture of 16105 MDA-MB-231-luc2 human breast cancer cells and 16105 human mammary fibroblasts in 25 mL serum-free Eagle’s MEM medium plus 25 mL growth factor-reduced phenol red-free Matrigel was injected into the inguinal mammary gland of 6?0-week-old female Nod/LtSz-prkds(scid) (NOD/SCID) mice (Jackson Laboratory, Bar Harbor, ME, USA). Tumors were allowed to grow over the course of 11 weeks, monitored weekly by bioluminescence imaging using an IVIS Spectrum 3-dimensional imaging system (Caliper Life Sciences). PEG20K-TIMP-1 was dialyzed into PBS, filter sterilized, and protein concentration was determined by UV absorbance at 280 nm using a Nanodrop spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA). At 11 weeks post-implantation, two tumor-bearing mice were injected intraperitoneally with 2 mg/kg PEG20K-TIMP-1 and two tumor-bearing mice were injected with 0.9% saline only. This dose was selected because it is a dose that for rhTIMP-1 has been shown to have an antitumorigenic effect in a xenograft model of colon cancer [19]. Twenty four hours later, all four mice were euthanized by CO2 inhalation, and tumors were resected and flash frozen in liquid N2 and stored at 280uC until analysis. MMP activity was assessed in fresh-frozen tumors by in situ zymography as described above.
Results
In considering how best to approach PEGylation of rhTIMP-1, we analyzed the structure of the protein and its mode of interaction with an MMP catalytic domain (Fig. 1) [31,44]. TIMP-1 is a globular protein of 184 amino acids and is glycosylated on two Asn residues. It possesses 12 Cys residues, paired as 6 disulfide bonds, and 8 Lys residues. The most common approach to protein PEGylation involves coupling of activated PEGs to primary amines found on Lys side chains and at the protein N-terminus [45]. Figure 1. TIMP-1 sites of inhibitory interaction and PEGylation. TIMP-1 (blue) is shown binding to the MMP-3 catalytic domain (gray); the N-terminal Cys-1 backbone (magenta and blue spheres) coordinates directly to the MMP catalytic Zn (green). The C-terminal residues Arg180, Ser-181, Gln-182, and Ala-184 (red sticks; residues 182?84 unstructured in the crystal structure) were initially mutated to Cys and targeted for site-specific PEGylation. In a subsequent approach, the 8 natural Lys residues (yellow spheres) were targeted for PEGylation. ?Lys-138, Lys-157 and Lys-88 all lie within 10 A of the MMP-3 cd, while other Lys residues are distant from the binding interface. Structure coordinates are from PDB ID 1UEA [44]; figure was created using Figure 2. PEGylation on rhTIMP-1 Lys residues with mPEG-SCM reagents. (A) Silver stained gel shows unmodified rhTIMP-1 and PEGylation reactions carried out in the presence of 50?006 molar excess of mPEG-SCM-5K or 1?6 molar excess of mPEG-SCM-20K, as indicated beneath gel. (B) The same gel stained with barium iodide shows the electrophoretic migration of mPEG-5K and mPEG-20K hydrolysis products and PEGylated rhTIMP-1 species. (C) Silver stained gel shows a concentration-dependent increase in molecular weight upon PEGylation with mPEG-5K-SCM in increasing molar excess. (D) Graph shows retention of MMP-3cd inhibitory activity relative to rhTIMP-1 by PEGylated rhTIMP-1 species from the reactions shown in panels A-C. The molar excess over rhTIMP-1 and molecular weight of the activated PEG in each reaction is indicated below the gel. that attachment of a single very large PEG substituent would likely be sufficient to prevent rapid renal elimination. Reaction conditions were tested using a variety of PEG:protein molar ratios, pH ranges, temperatures, and reaction times. Surprisingly, we failed to detect rhTIMP-1-R180C PEGylation under any conditions. We verified 100% reactivity of the purchased mPEG-maleimide by titration against cysteine using Ellman’s reagent (5,59-dithiobis-(2-nitrobenzoic acid) to quantify residual free cysteine. We next hypothesized that failure of rhTIMP-1-R180C to react could be attributable either to (a) inaccessibility of Cys-180 caused by the fold of the protein, or (b) covalent blockage of Cys-180 by disulfide formation with small sulfhydryl compounds. To distinguish between these possibilities, we carried out test reactions in the presence of urea to partially denature the protein, alleviating steric hindrance, or TCEP, a disulfide reductant unreactive with mPEG-maleimide. Inclusion of TCEP resulted in complete PEGylation while urea denaturation had little effect, suggesting that the difficulties in PEGylating rhTIMP-1-R180C were caused solely by Cys oxidation. In an attempt to identify regioselective conditions for partial reduction that would allow deprotection of Cys-180 without disrupting the native disulphide bonds essential for TIMP-1 activity [47,48], we tested a wide variety of reducing agents and conditions, assessing both the effectiveness of PEGylation and the
retention of MMP inhibitory activity following reduction. As summarized in Table S1, regardless of the reducing agent employed, conditions resulting in efficient PEGylation of rhTIMP1-R180C also severely compromised MMP inhibitory activity. We concluded that PEGylation on Cys-180 of the mutant rhTIMP-1 was incompatible with retention of activity.
