Improving Circulation Half-Life of Therapeutic Candidate N-TIMP2 by Unfolded Peptide Extension

Abstract

Matrix metalloproteinases (MMPs) are significant drivers of many diseases, including cancer, and are established targets for drug development. Tissue inhibitors of metalloproteinases (TIMPs) are endogenous MMP inhibitors and are being pursued for the development of anti-MMP therapeutics. TIMPs possess many attractive properties for drug candidates, such as complete MMP inhibition, low toxicity, low immunogenicity, and high tissue permeability. However, a major challenge with TIMPs is their rapid clearance from the bloodstream due to their small size. This study explores a method for extending the plasma half-life of the N-terminal domain of TIMP2 (N-TIMP2) by appending it with a long, intrinsically unfolded tail containing Pro, Ala, and Thr (PATylation). We designed and produced two PATylated N-TIMP2 constructs with tail lengths of 100 and 200 amino acids (N-TIMP2-PAT₁₀₀ and N-TIMP2-PAT₂₀₀). Both constructs demonstrated higher apparent molecular weights and retained high inhibitory activity against MMP-9. N-TIMP2-PAT₂₀₀ significantly increased plasma half-life in mice compared to the non-PATylated variant, enhancing its therapeutic potential. PATylation offers distinct advantages for half-life extension, such as fully genetic encoding, monodispersion, and biodegradability. It can be easily applied to N-TIMP2 variants engineered for high affinity and selectivity toward individual MMPs, creating promising candidates for drug development against MMP-related diseases.

1. Introduction

Matrix metalloproteinases (MMPs) are a family of 23 human enzymes that play a crucial role in degradation of the extracellular matrix (ECM), tissue remodeling, immunity, and wound healing [1,2]. Abnormal MMP activity has been implicated in the development and progression of several diseases, including different cancers, vascular diseases, arthritis, inflammatory bowel diseases, and others [3]. Several types of MMPs are upregulated in cancers and facilitate metastasis because they are directly involved in the degradation of the extracellular matrix (ECM). Additionally, they are indirectly involved in the regulation of cellular functions and signaling through proteolysis of inhibitors and signaling factors in the tumor and its microenvironment [4,5].

Much effort has been invested in the development of therapeutics to control improperly regulated MMP activity [6]. Early efforts for targeting MMPs concentrated on designing small-molecule MMP inhibitors that target highly conserved active sites centered on a catalytic zinc ion. These small molecules showed broad-spectrum activity, not differentiating between various MMP targets and frequently binding other homologous proteins in the body, resulting in high toxicity [7,8,9]. Recent efforts have shifted toward developing therapeutic antibodies [10,11], which can achieve better selectivity for targeting individual MMPs through a larger interaction interface, but these, too, have limitations, notably their difficulty in discriminating between active MMPs and their more abundant zymogen precursors [11,12]. While a number of such antibodies are presently undergoing clinical trials, none of them have been approved as drugs so far.

As an alternative to antibodies, the four tissue inhibitors of metalloproteinases (TIMP1–4) or their N-terminal domains (N-TIMPs) could be used as anti-MMP drug candidates. These inhibitors possess high affinity to all active MMPs and, as endogenous proteins, are nontoxic and nonimmunogenic in humans [13]. For example, TIMP2 has been demonstrated to effectively reduce MMP-related inflammation as well as to suppress tumor growth and metastasis in the models of triple negative breast cancer [14]. Furthermore, TIMPs can be engineered to bind with high affinity and specificity to specific disease-associated MMPs [15,16,17,18,19,20,21,22,23]. Therefore, TIMP-based therapeutics are promising tools to suppress and control cancers and other MMP-related diseases.

Yet, TIMP variants as small proteins (~15–20 kDa) possess short circulation half-lives. For example, full-length recombinantly produced human TIMP1 with a molecular weight of ~20 kDa is eliminated from mice with a half-life of only 1.1 h [24]. Rapid clearance from the bloodstream can limit the therapeutic effect of a TIMP, as it may not reach a sufficient concentration to exert the desired effects on MMPs. To facilitate the success of TIMPs as therapeutics, we set our goal to extend their half-life using a protein engineering approach.

Several approaches have been developed to prolong the half-life of therapeutic proteins and improve their delivery methods. PEGylation, which involves covalent attachment of polyethylene glycol (PEG) polymers, is currently the most commonly used method for increasing the half-life of biologics [25,26] and has been successfully applied to extend TIMP1 circulation half-life in mice [24]. However, PEGylation, when performed on cysteine or lysine residues, can lead to a reduction in protein activity. Moreover, if the conjugation site is not unique, PEGylation produces polydispersed conjugation of the sample, an undesirable property for therapeutic proteins. To address the issue of polydispersion, a noncanonical amino acid was introduced into N-TIMP2, and PEG polymer was conjugated to the protein using click chemistry [27]. This approach resulted in an approximately eight-fold extension of the N-TIMP2 elimination half-life. Nevertheless, the costly nature of protein expression with noncanonical amino acids poses a challenge to large-scale protein production using this method.

Alternative strategies for extending therapeutic protein half-life can avoid the complexities of chemical modification by relying instead on genetic fusions to create larger composite proteins with the desired pharmacokinetic properties. Examples include fusion to serum proteins with long half-lives such as albumin [28,29]. Previously, TIMP2 has been fused to human serum albumin, resulting in improved protein yield and an extended elimination half-life [30]. However, a potential drawback of this approach is that fusion with albumin or other human proteins could lead to interactions with other biological components, causing the protein to localize to unintended targets, potentially resulting in undesired effects.

Another type of genetic fusion aims to enlarge the protein via the addition of a long, intrinsically unfolded tail composed of natural amino acids. “PASylation” achieves this aim by appending a long amino acid stretch that contains a random combination of proline (P), alanine (A), and serine (S) residues [31]. The flexible tail adopts multiple conformations and effectively increases the hydrodynamic radius of the protein without affecting protein activity. This strategy is advantageous since the tail is chemically monodispersed (of uniform length and tethered at a single attachment point), applicable over a wide range of expression hosts, biodegradable, and economical since the protein is expressed with the tail already incorporated [31]. PASylation has been explored for at least 15 biological molecules, where it successfully improved protein elimination half-life and in some cases also enhanced therapeutic protein stability and activity [31,32]. The broader strategy of modulating protein behavior using intrinsically unfolded extensions is still in its infancy, and studies that further explore the sequence requirements for such unfolded tails and their applicability to different therapeutic proteins can help to advance the strategy toward clinical translation.

In this work, we decided to explore a new strategy for extending N-TIMP2 half-life by appending unstructured, genetically encoded tails. We generated and characterized constructs of N-TIMP2 with intrinsically unfolded tails containing a random combination of proline, alanine, and threonine (PATylation). The goals of the study were twofold: first, to determine whether an unfolded tail on N-TIMP2 can extend plasma half-life while retaining protein function, and, secondly, to demonstrate whether PATylation (comprising random sequences of proline, alanine, and threonine) can confer such an unfolded tail. The extension of the established method by substituting threonine for serine could offer increased versatility for protein engineering, by broadening the scope of possible modifications for extending protein half-life. For some applications and biomanufacturing processes, the PAT tail could potentially offer advantages, since threonine, while possessing similar biophysical properties to serine, is less reactive and less susceptible to chemical or biological phosphorylation [32,33]. Our results show that PATylated N-TIMP2 constructs exhibit an increased apparent molecular weight and retain full activity of WT N-TIMP2 toward its target MMP-9. Animal experiments reveal that our construct with a 200 amino-acid-long extension (N-TIMP2-PAT₂₀₀) demonstrates a ~four-fold increase in elimination half-life, presenting an attractive candidate for further therapeutic development.

2. Materials and Methods

2.1. Design of PATylation Sequences

We devised an algorithm for PAT sequence design that randomly chooses codons for alanine, threonine, and proline from a set of codons which are common in Pichia pastoris [34]. A total of 1000 candidate sequences for a PAT tail were computationally produced and the number of repeats was calculated for fragments of length of 4 bp to the maximal length of the sequence. The respective numbers of repeats for each fragment length were then summed to obtain a repeat score. The final sequences were chosen that possessed the smallest maximum DNA fragment repeat length and the smallest repeat score. The source code of the program for producing such repeat-minimized sequences has been deposited in GitHub and is publicly available (https://github.com/jshir1 (accessed on 15 September 2024)).

2.2. Design of Internal FLAG-Tag in the N-TIMP2 Sequence

To facilitate N-TIMP2 detection in plasma, a construct was designed to incorporate a FLAG-tag into the N-TIMP2 sequences. The FLAG tag could not be incorporated at the N-terminus of N-TIMP2, since this terminus is used for MMP recognition, and could not be incorporated at the C-terminus due to proximity to the c-myc tag, used for protein immobilization, and possible steric interference between binding of the two antibodies to these respective tags. Hence, it was introduced internally in one of the N-TIMP2 loops. For this purpose, we searched for a stretch of the protein that does not possess any secondary structure, has low interconnectivity with the rest of the protein, and is located on the surface of the protein far from the MMP binding site. We identified such a site in the loop that contains residues 54–59 on N-TIMP2 and inserted a FLAG-tag (DYKDDDDK) preceded by a short flexible linker (SSG) between residues 57 and 58 of N-TIMP2. The correct folding of the protein was predicted by AlphaFold [35]. The C-terminus of the constructs was supplemented by a myc-tag and a His-tag. Thus, the sequences of 3 constructs were designed with the FLAG-tag and PAT underlined:

• N-TIMP2:

CSCSPVHPQQAFCNADVVIRAKAVSEKEVDSGNDIYGNPIKRIQYEIKQIKMFKGPESSGDYKDDDDKDIEFIYTAPSSAVCGVSLDVGGKKEYLIAGKAEGDGKMHITLCDFIVPWDTLSTTQKKSLNHRYQMGCEAAASFLEQKLISEEDLNSAVDHHHHHH

• N-TIMP2-PAT₁₀₀:

CSCSPVHPQQAFCNADVVIRAKAVSEKEVDSGNDIYGNPIKRIQYEIKQIKMFKGPESSGDYKDDDDKDIEFIYTAPSSAVCGVSLDVGGKKEYLIAGKAEGDGKMHITLCDFIVPWDTLSTTQKKSLNHRYQMGCETAPAAPATPAPTAPTPTPAAPAPTTPATPAPTPAAPAPAPTAPTAPAAPTAPATPAAPTTPTPTTPTPATPTPTAPAPATPTTPTPTAPAAPTPAPTTPAAAASFLEQKLISEEDLNSAVDSSGHHHHHH

• N-TIMP2-PAT₂₀₀:

CSCSPVHPQQAFCNADVVIRAKAVSEKEVDSGNDIYGNPIKRIQYEIKQIKMFKGPESSGDYKDDDDKDIEFIYTAPSSAVCGVSLDVGGKKEYLIAGKAEGDGKMHITLCDFIVPWDTLSTTQKKSLNHRYQMGCEATPTPAPTAPTTPTPTPTTPAPTTPTTPATPAPTTPATPAPTPATPTPAAPAPTTPTPAAPAPTPATPTAPAPTTPTAPTAPTPATPATPAAPTPTPTTPTTPTAPAAPTPTAPTAPAPTAPTAPTAPATPAAPATPTPAPTAPATPTPAAPTTPTAPTPTPTPAPATPAAPAPTPATPAAPTTPAAPTAPAAPTTPTPTAAASFLEQKLISEEDLNSAVDSSGHHHHHH

2.3. N-TIMP2 Protein Production and Purification

Initially, the three proteins were expressed in P. pastoris as previously described [18]. Briefly, P. pastoris clones containing our mutants or WT N-TIMP2 genes were grown in 50 mL BMGY medium until an OD₆₀₀ of 10, transferred into 500 mL inductive BMMY medium, and induction was continued for 72 h at 30 °C, with the addition of 0.5% methanol every 22–26 h before pelleting the cells and collecting the supernatant. Subsequently, to obtain sufficient amounts of the proteins, N-TIMP2-PAT₁₀₀ and N-TIMP2-PAT₂₀₀ were purchased from GenScript Biotech (Singapore), where they were expressed in HEK293 cells in the pcDNA3.1 vector and purified by Ni-affinity chromatography. The PATylated proteins were further purified on a Hi load 16/60 Superdex 200 pregraded column (C17-1069-01 Cytiva, Marlborough, MA, USA) in 1X PBS pH 7.2. Pure fractions were collected, concentrated using a 10 kDa MWCO Amicon concentrator, and filter sterilized prior to injection into mice. Concentrations of the purified protein samples were determined by absorbance on a NanoDrop spectrophotometer using calculated extinction coefficients.

2.4. MMP-9 Expression

A high-stability mutant of the MMP-9 catalytic domain (MMP9_CAT*) lacking the fibronectin-like domain (residues 107–215, 391–443) was expressed in BL21 E. coli cells, refolded, and purified as described in our previous publication [36].

2.5. Circular Dichroism

Circular dichroism (CD) spectra measurements were performed on the purified N-TIMP2 constructs. Size exclusion purified monomers were dialyzed against the following buffer: 100 mM sodium phosphate, 100 mM sodium fluoride, pH = 7.2 in Slide-a-lyzer mini dialysis devices with a 3.5 kDa molecular weight cutoff (Thermo Scientific (Waltham, MA, USA) cat. No. TS-88400). Samples were concentrated to 10 µM by centrifugal concentrators with a molecular weight cutoff of 3 kDa (Vivaspin (Göttingen, Germany),cat. No. VS0691). CD spectra were measured at 25 °C in a 1 mm path length quartz cuvette (Hellma Analytics (Müllheim, Germany), cat. No. 110-1-40) on a Jasco J-1100 CD spectrophotometer (Jasco, Tokyo, Japan) in the range of 190–260 nm.

2.6. Enzymatic Activity Inhibition Assay

Varying concentrations of N-TIMP2 were incubated with 0.26 nM MMP-9_CAT* for 60 min at 37 °C in 50 mM Tris-HCl, 0.15 M NaCl, 10 mM CaCl₂, and 0.02% Brij-35 pH 7.5. After incubation, the fluorogenic MMP substrate (MCA-Lys-Pro-Leu-Gly-Leu-DNP-Dpa-Ala-Arg-NH₂ [37] (where MCA is (7-methoxycoumarin-4-yl)acetyl; DNP-Dpa is N-3-(2,4-dinitrophenyl)-L-2,3 diaminopropionic acid)) (Sigma Aldrich Cat. no. SCP0193-1MG) at a final concentration of 7.5 µM was added to the preincubated N-TIMP2:MMP solution. Fluorescence at 395 nm was measured at least every ten seconds for at least 6 min, immediately after addition of fluorogenic substrate with excitation at 325 nm on a BioTek Synergy H1 plate reader (BioTek, Winooski, VT, USA) preincubated to 37 °C. Controls were also performed to determine uninhibited enzyme activity as well as basal fluorescence levels (without N-TIMP2 or MMP). At least three assays were performed for each N-TIMP2 protein.

2.7. K_i^app Determination

Initial velocities of the enzymatic (MMP) cleavage of the fluorogenic substrate were derived from the fluorescence generated by this cleavage. The fraction of MMP activity was determined from the enzymatic assays, expressed as the inhibited velocity (V_i) divided by the uninhibited velocity (V₀), and was plotted against the corresponding concentration of N-TIMP2. From this plot, the apparent K_i (K_i^app) value was determined by fitting to an equation derived from the Morrison equation (Equation (1)) [38] using MATLAB:

$$f={\displaystyle \frac{\left[\left(\left[MMP\right]-\left[TIMP\right]-K_i^{app}\right)+\sqrt{{(\left[MMP\right]+\left[TIMP\right]+K_i^{app})}^2-4[MMP][TIMP]}\right]}{2\left[MMP\right]}}$$

(1)

where $f$ is fraction of enzyme activity (V_i/V₀) and K_i^app values are an average of the K_i^app values determined from single inhibition assay experiments for each N-TIMP2 variant.

2.8. Murine Pharmacokinetics Study

Ten-week old female NOD scid gamma (NSG) mice were injected intraperitoneally with 20 mg/kg N-TIMP2, N-TIMP2-PAT₁₀₀, or N-TIMP2-PAT₂₀₀. The selected dose was shown in pilot studies to yield levels detectable by ELISA in mouse plasma. Treatment groups of 6 mice were divided into 2 subgroups (A and B) of 3 mice each to stagger bleeding times. Mice were then weighed and dosage was calculated. Prior to injection of N-TIMP2 proteins and at regular intervals thereafter, approximately 20 µL of blood was collected from the tail vein according to our previously described procedure [24]. In brief, the mice were immobilized in a specialized tail vein illuminator apparatus (Braintree Scientific Inc. TV-150, Braintree, MA, USA), the tail was cleaned with an alcohol wipe, the tail was coated with petroleum jelly, then a 22.5 Ga needle (BD 3015156) was used to make an incision. A preweighted heparinized capillary tube (Sarstedt 16.444.100, Newton, NC, USA) was used to collect the blood sample (volume inferred by mass change), which was then transferred to an EDTA-coated Microvette tube (Sarstedt 20.1345.100) and mixed with 9 parts 0.1 M trisodium citrate buffer. The samples were then centrifuged to separate cellular components from plasma, and the plasma supernatant was transferred to new protein LoBind tubes. Blood was collected from the A group mice at preinjection, 30 min, 1 h, 2 h, 3 h, 6 h, 12 h, and 24 h, and from B group mice at preinjection, 45 min, 1.5 h, 2.5 h, 4 h, 8 h, 48 h, and 72 h. After the final collection, mice were sacrificed via CO₂ inhalation (primary) and terminal bleeding via cardiac puncture (secondary).

2.9. ELISA Detection of Plasma N-TIMP2 Proteins

ELISA 96 well plates (Thermo Scientific 442404) were coated with 100 μL of 10 µg/mL anti-C-myc (Abcam ab18185, Waltham, MA, USA) in coating buffer (0.15 M sodium carbonate, 0.35 M sodium bicarbonate, pH 9.6) overnight at 4 °C. Plates were washed 6 times with 200 µL/well of wash buffer (0.05% Tween20 in PBS pH 7.4) and then blocked for 1 h with 200 µL/well of blocking buffer (1% BSA (Boston BioProducts, Milford, MA, USA- cat# C10132F) in PBS pH 7.4. After the blocking step, plates were washed 6 times with wash buffer. Concentrated working stocks of 1 µM were prepared for the standards (WT N-TIMP2, N-TIMP2-PAT₁₀₀, or N-TIMP2-PAT₂₀₀) in dilution buffer (0.1% BSA/PBS). From the 1 µM stock a series of serial dilutions were made, also in dilution buffer, to achieve 5× working stocks of 50 nM, 25 nM, 12.5 nM, 6.25 nM, 3.125 nM, 1.563 nM, 0.78 nM, 0.391 nM, 0.195 nM, and 0.0975 nM. A final 1× dilution series was made by mixing 1 part 5× working stock, 3 parts dilution buffer, and 1 part NSG mouse plasma diluted 10-fold in 0.1 M citrate buffer (to simulate in the standards the equivalent background signal of the experimental samples drawn from mouse blood). The final standard samples contained the following nM concentrations: 10 nM, 5 nM, 2.5 nM, 1.25 nM, 0.625 nM, 0.313 nM, 0.156 nM, 0.078 nM, 0.039 nM, and 0.0195 nM. For the experimental samples, plasma from the mouse bleed time course (already diluted 10-fold in citrate buffer) was diluted 1:5 in dilution buffer. In addition, negative controls containing no N-TIMP2 or no antibody were prepared. Standard and experimental samples were plated in triplicate (100 µL per well). After overnight incubation, the wells were washed 6 times with wash buffer and then incubated with 100 µL per well of mouse biotinylated anti-DDDDK (Abcam ab173832) at 1:3000 in dilution buffer overnight at 4 °C. The wells were then washed 6 times with wash buffer and subsequently incubated with 100 µL per well of 1:10,000 streptavidin-HRP (Thermo Scientific 21130) in dilution buffer for 1 h at room temperature. The plates were then washed 6 times with wash buffer, and 200 µL of TMB HRP (MP Biomedicals, Irvine, CA, USA, cat# 152346) substrate was added to each well. Absorbance at 655 nm was then measured vs. time with a BioTek Synergy HT plate reader. The slope of the linear portions of each curve was quantified and plotted in Prism (GraphPad Prism v9.2.0). A four-parameter logistic regression sigmoidal curve was fitted to the N-TIMP2, N-TIMP2-PAT₁₀₀, or N-TIMP2-PAT₂₀₀ standards with the Prism software to generate a standard curve, and the plasma concentration for each timepoint was determined via nonlinear interpolation from the standard curve data.

2.10. Analysis of ELISA

The interpolated concentrations vs. time were plotted and fitted with a logarithmic regression model for each mouse individually. Outlier data points were selected by the least-squares method and omitted to optimize the logarithmic correlation coefficient for each mouse. The data were then combined, averaging 3 mice per timepoint (error represented by SEM), and fitting the A and B group mice onto the same curve. To compare elimination half-lives, curves were then evaluated using a two-phase decay with least-squares nonlinear fit in the GraphPad software v9.2.0. To investigate statistical significance of the half-life, the null hypothesis was that the rate of biological elimination (K_slow) for N-TIMP2-PAT₁₀₀ (or N-TIMP2-PAT₂₀₀) was equal to that of N-TIMP2, with the alternative hypothesis being that they are not equal. Since elimination is the slow rate, and T^1/2_slow = ln(2)/K_slow, this is equivalent to comparing half-lives.

3. Results

3.1. PATylated N-TIMP2 Sequence Design

PATylation could be performed either at the N- or the C-terminus of a protein of interest. Since the N-terminus of N-TIMP2 is crucial for its binding to MMP enzymes [39], we decided to introduce the PAT tail at the C-terminus of the protein (Figure 1A). In this study, we explored two N-TIMP2 PATylated constructs containing 100- or 200-amino-acid unfolded extensions. While additions of up to 800 residues have been explored previously in PASylation, these longer tails frequently resulted in a reduction in protein activity in a manner proportional to the length of the PAS extension [31], along with reduced protein expression yield, whereas shorter tails have been sufficient to achieve pharmacokinetic improvements [40]. We thus decided to explore the shortest PAT extensions that might be anticipated to enhance plasma half-life without compromising other protein properties. The PAT tail sequences were computationally designed to minimize the number of DNA repetitive sequences, as such repetitions could interfere with gene cloning and protein expression [41,42].

Two additional tags were incorporated into the N-TIMP2 gene construct to facilitate capture and detection of N-TIMP2 variants in mouse plasma. A c-myc tag, used for protein capture, was introduced after the PAT tail before the His-tag (Figure 1A). For protein detection, we introduced an internal FLAG tag [43] in one of the N-TIMP2 loops, between residues 57 and 58 (see Methods for loop design). We, thus, obtained three protein constructs: WT N-TIMP2 containing internal FLAG-tag, c-myc-tag, and His-tag (referred to as simply N-TIMP2 throughout the manuscript), and the two PATylated constructs, N-TIMP2-PAT₁₀₀ and N-TIMP2-PAT₂₀₀, that contained the same tags as N-TIMP2 and a PATylation tail of 100 and 200 amino acids, respectively (Figure 1A and Methods).

To verify the correct fold of the PATylated N-TIMP2 constructs, we predicted their structures using AlphaFold [35]. The predictions showed that all constructs exhibit a well-folded core corresponding to the N-TIMP2 fold with the FLAG-tag slightly protruding away from the N-TIMP2 protein (Figure 1B). On the contrary, the PAT tails in both the 100- and 200-amino-acid constructs were completely unfolded and encircled the protein, increasing the effective radius of gyration.

3.2. Construct Production and Characterization

N-TIMP2, N-TIMP2-PAT₁₀₀, and N-TIMP2-PAT₂₀₀ were first expressed in P. pastoris and subsequently produced by GenScript using HEK293 cells, and were purified by Ni-affinity chromatography and size-exclusion chromatography (SEC) and analyzed by SDS-PAGE. Our results showed that N-TIMP2 appeared on the gel at the expected molecular weight of 17 kDa (Figure 2A). In contrast, N-TIMP2-PAT₁₀₀ and N-TIMP2-PAT₂₀₀ appeared as broad bands between 60 and 90 kDa and between 120 and 160 kDa, respectively. Both PATylated constructs exhibited much higher apparent molecular weight by SDS PAGE in comparison to their actual weights of 27.4 kDa and 36.7 kDa, respectively, which are well below the typical cutoff for renal clearance of ~66 kDa for a globular protein [44]. The increased apparent molecular weights of the PATylated constructs observed by SDS-PAGE are consistent with an expected increase in the radius of gyration upon PATylation. The structural models obtained by AlphaFold (Figure 1B) also support an increased size due to the unfolded tails, with calculated radii of gyration 2.10 nm, 3.89 nm, and 4.44 nm for N-TIMP2, N-TIMP2-PAT₁₀₀, and N-TIMP2-PAT₂₀₀, respectively. To experimentally measure the impact of PATylation on particle size, we conducted dynamic light scattering (DLS) of the three proteins. DLS revealed increasing hydrodynamic radii of 2.806 nm, 4.253 nm, and 5.612 nm for N-TIMP2, N-TIMP2-PAT₁₀₀, and N-TIMP2-PAT₂₀₀, respectively (Figure 2B). Similar impact of PASylation on protein size and behavior has been reported previously [31].

3.3. Structural Analysis of PATylated Sequences by Circular Dichroism

To further evaluate whether the designed PAT tails were, indeed, intrinsically unfolded, we next measured the circular dichroism (CD) spectra of the three constructs N-TIMP2, N-TIMP2-PAT₁₀₀, and N-TIMP2-PAT₂₀₀. The N-TIMP2 spectra showed all characteristics of a folded mostly β protein (Figure 2C). The spectra of N-TIMP2-PAT₁₀₀ and N-TIMP2-PAT₂₀₀, on the contrary, showed sharp negative peaks at 203–205 nm, indicative of the presence of significant unfolded structure. This negative peak is about twice as large for N-TIMP2-PAT₂₀₀ compared to N-TIMP2-PAT₁₀₀, consistent with its two-times longer unfolded tail. The spectra obtained for N-TIMP2-PAT₁₀₀ and N-TIMP2-PAT₂₀₀ are very similar to those observed previously for other PASylated proteins containing Pro/Ala/Ser tails of similar sizes [31]. These data further corroborate the hypothesis that the designed PAT sequences are, indeed, intrinsically unfolded and behave as random coil.

3.4. Activity of the PATylated N-TIMP2 Variants

To verify that the addition of the internal FLAG-tag and the PATylated tail does not alter N-TIMP2 folding and activity, we measured the ability of the three constructs (N-TIMP2, N-TIMP2-PAT₁₀₀, and N-TIMP2-PAT₂₀₀) to inhibit one representative MMP, MMP-9. In this experiment, the activity of MMP-9 was measured by monitoring the appearance of a fluorogenic product in the presence of varying concentrations of N-TIMP2-based constructs, and the inhibition constant was calculated for each construct according to Equation (1) (Figure 2D). K_i^app for N-TIMP2 was measured to be 0.32 ± 0.03 nM. This K_i^app was very similar to K_i^app of the N-TIMP2 protein that does not contain the internal FLAG-tag (Supplementary Figure S1) and to the previously published data on a similar construct [18], proving that the internal FLAG-tag does not affect N-TIMP2 anti-MMP activity. N-TIMP2-PAT₁₀₀ exhibited a slightly stronger K_i^app of 0.24 ± 0.03 nM, and N-TIMP2-PAT₂₀₀ exhibited a slightly weaker K_i^app of 0.59 ± 0.06 nM, compared to N-TIMP2; these modest differences were significant. Therefore, we can conclude that the addition of the internal FLAG tag and the 100- and 200-amino-acid PAT tails do not disrupt the N-TIMP2 fold and preserve high inhibitory activity of N-TIMP2 against MMP-9.

3.5. Optimization of ELISA for N-TIMP2 Detection in Plasma

To measure plasma half-life of the N-TIMP2 constructs in mice, we first developed an ELISA-based protocol for detection of blood plasma concentrations of the recombinant N-TIMP2 constructs. Here, the ability to detect low (pM) concentrations of N-TIMP2 constructs was important in order to compare the half-life of various constructs. We experimented with several setups where, in all cases, N-TIMP2 constructs were captured on the ELISA plate via an immobilized anti-c-myc antibody (Figure 3A). We first attempted to detect N-TIMP2 via polyclonal anti-TIMP2 antibodies raised against a full-length TIMP2 peptide, but none of these antibodies exhibited substantial binding to folded N-TIMP2. We then switched to detection with a biotinylated anti-FLAG antibody that binds to the internal FLAG-tag of the three constructs (Figure 1A,B). The biotinylated anti-FLAG antibody is further detected by streptavidin coupled to horseradish peroxidase (HRP) that produces blue color upon reaction with the substrate. We further optimized the protocol to reach the detection level of ~30 pM, which was sufficient for determining the half-life of the N-TIMP2 constructs in mice according to a previous study [24]. Since the ELISA response is not linear, a calibration curve was constructed with samples of known N-TIMP2 variant concentrations added to plasma and the ELISA signal recorded on the same plate where the mouse samples were evaluated (Supplementary Figure S2). This calibration curve was fitted to a sigmoidal curve and used to derive N-TIMP2 variant concentrations from ELISA signals generated by actual plasma samples taken from mice in the pharmacokinetic study described below.

3.6. Measurement of N-TIMP2 Construct Half-Life in Mice

We next carried out a pharmacokinetic study to compare the half-life of N-TIMP2 in circulation to those of the two PATylated constructs. We injected each of the three N-TIMP2 constructs at 20 mg/kg dose into mice and collected blood samples over a 72 h time period at various intervals (Figure 3B; see also Methods). The concentration of the N-TIMP2 constructs obtained from plasma at different time points were measured by ELISA and plotted vs. time (Figure 3C and Supplementary Figure S3).

We then analyzed the data, finding a best fit to a two-phase decay model of the plasma concentration vs. time data. The first phase is the initial fast rate where the drug is distributed from the plasma compartment to another compartment, e.g., tissues; the second phase is slower and is associated with the elimination of the drug [45]. The two-phase decay model is typical for certain types of drug administration, such as IP and IV [46,47,48], where the elimination half-life is given by the slow phase rather than the time it takes the plasma concentration to reach 50% of the administered dose as in a simpler one-phase decay model [45]. Our data show that the rate of biological elimination for N-TIMP2-PAT₂₀₀ is significantly slower than that of N-TIMP2 (p < 0.0001); the half-life was extended approximately 3.5-fold, from 2.5 h to 8.5 h for the N-TIMP2-PAT₂₀₀. Over the period of time from 5–20 h, the concentration of N-TIMP2-PAT₂₀₀ remained about one order of magnitude higher compared to the concentration of N-TIMP2, providing a much longer timeframe for the N-TIMP2-PAT₂₀₀ to exert activity against MMPs. In contrast, no significant difference was observed between the rate of elimination of N-TIMP2-PAT₁₀₀ and that of N-TIMP2, suggesting that this PAT tail is too short to exert an effect in vivo (Supplementary Figure S3). Thus, one of the evaluated PATylated constructs, N-TIMP2-PAT_200, demonstrated increased circulating longevity of N-TIMP2 in vivo without compromising its anti-MMP inhibitory activity.

4. Discussion and Conclusions

Various MMPs are key targets for cancer and other diseases, and their natural inhibitors TIMPs are attractive candidates for anti-MMP drug development. Therapeutics based on human proteins offer advantages over small-molecule drugs, such as greater specificity and lower toxicity. Moreover, in comparison to antibodies, full-length TIMPs and their N-terminal domains possess certain advantages such as higher tissue permeability, minimal immunogenicity, and complete MMP inhibition. However, N-TIMPs as small proteins also pose challenges in formulation and delivery as their circulation half-life is naturally short. To overcome these challenges, in this study, we explored a PATylation strategy as an approach to extend the plasma half-life of N-TIMP2.

We designed, produced, and evaluated two N-TIMP2 constructs containing long intrinsically unfolded PAT tails of either 100 or 200 amino acids in length. Both constructs retained the high inhibitory activity toward MMP-9, but only the longer construct, N-TIMP2-PAT₂₀₀, resulted in a substantial increase in plasma half-life in mice. These results prove that the addition of these C-terminal unfolded extensions does not substantially affect the N-TIMP2 fold or activity, but only the 200 amino acid long tail was effective in preventing the fast protein clearance from plasma. This is likely explained by the differences in the apparent protein size of the N-TIMP2-PAT₁₀₀ and N-TIMP2-PAT₂₀₀ contracts, as documented in our SDS-PAGE and DLS experiments (Figure 2A,B), with the shorter construct not reaching the minimum apparent size required to prevent clearance [44]. N-TIMP2-PAT₂₀₀ demonstrated a plasma half-life in mice of ~8 h, which could allow adequate time for the therapeutic effects of N-TIMP2 and variants, and enable studies in animals with 1× per day dosing. These results highlight the effectiveness of PATylation in improving the pharmacokinetic properties of N-TIMP2.

In our study, we investigated relatively short unfolded PAT tails of 100 and 200 amino acids. For comparison, PASylation extensions typically consist of 200 or 600 residues [40]. While a 600-residue PAS extension generally results in longer circulation half-lives compared to the 200-residue extension, extending PAS tails beyond 200 residues often reduces target affinity, as observed in several proteins such as Fab fragment, interferon, and human growth hormone [31]. In agreement with these findings, we found that TIMP2-PAT₁₀₀ exhibited a modestly improved inhibition of MMP-9 relative to N-TIMP2, whereas TIMP2-PAT₂₀₀ showed slightly reduced inhibition. This suggests that extending the PAT tail beyond 200 residues may lead to further reduction in N-TIMP2 activity. Another challenge of using longer unfolded tails for half-life extension can be the technical difficulty in achieving high expression yields and pure preparations [40]. Therefore, when selecting the length of PATylation or other unfolded protein extension, it is crucial to strike a balance between prolonged circulation half-life and high protein activity and expression yield.

An additional area for future exploration could be more direct comparison of the functional and pharmacokinetic consequences of PATylation versus PASylation. While the approaches are similar, we anticipate that PATylation may offer advantages for some applications because, relative to serine, threonine is less reactive and less susceptible to chemical or biological phosphorylation [32,33]. Thus, a PATylated protein is at lower risk of carrying unintended phosphorylations, introduced during its biological production, which confer negative charge and may consequently impact affinity and specificity toward the intended molecular targets. An additional technical advantage of the PAT tail design approach introduced here is that it allows for the incorporation of any desired tail length. By contrast, the PASylation approach, as previously implemented, relies on the use of commercially available sequences that are multiples of predetermined and fixed peptide lengths. The design method introduced here offers increased flexibility and adaptability for end users with differing aims and requirements. We anticipate that this new approach will offer a valuable tool to the protein engineering community.

Several alternative techniques for half-life extension have already been applied to TIMPs. As opposed to other methods, our PATylation method presents a number of advantages over the previously applied techniques including fully genetic encoding of the gene construct with the PAT tail, monodispersion (of uniform defined chemical composition), and biodegradability. While our study focused on the wild-type N-TIMP2 protein only, PATylation could be easily applied to engineered N-TIMP2 variants with high affinity and selectivity toward individual MMP family members, thus creating attractive candidates for drug development against MMP-related diseases.