Site-specific PEGylation crosslinking of L-asparaginase subunits to improve its therapeutic efficiency

L-Asparaginase is an enzyme successfully being used in the treatment of acute lymphoblastic leukemia, acute myeloid leukemia, and non-Hodgkin’s lymphoma. However, some disadvantages still limit its full application potential, e.g., allergic reactions, pancreatitis, and blood clotting impairment. Therefore, much effort has been directed at improving its performance. A popular strategy is to randomly conjugate L-asparaginase with mono-methoxy polyethylene glycol, which became a commercial FDA approved formulation widely used in recent years. To improve this formulation by PEGylation, herein we performed cysteine-directed site-specific conjugation of the four L-asparaginase subunits to prevent dissociation-induced loss of activity. The conjugation sites were selected at surface-exposed positions on the protein to avoid affecting the catalytic activity. Three conjugates were obtained using different linear PEGs of 1000, 2000, and 5000 g/mol, with physical properties ranging from a semi-solid gel to a fully soluble state. The soluble-conjugate exhibited higher catalytic activity than the non-conjugated mutant, and the same activity than the native enzyme. Site-specific crosslinking of the L-asparaginase subunits produced a higher molecular weight conjugate compared to the native tetrameric enzyme. This strategy might improve L-asparaginase efficiency for leukemia treatment by reducing glomerular filtration due to the increase in hydrodynamic size thus extending half-live, while at the same time retaining full catalytic activity.


Introduction
While it is indisputably necessary to investigate other types of polymers and methods to improve the 83 pharmacodynamics of protein-based drugs, over two decades of clinical use has stablished PEGylation as the 84 most efficient method to extent biopharmaceutics half-life, besides being an affordable technology still evolving 85 [27]. As for the future of L-asparaginase, new formulations are constantly entering clinical trials [2,3], and there 86 is convincing evidence of its efficiency for more than 40 years [28,29]. For example, a recombinant form from 87 E. coli recently entered the European market [2]. An attractive way to modify L-asparaginase pharmacokinetics 88 while at the same time maintaining its pharmacodynamics is by site-specific PEGylation, that is the attachment 89 of PEG polymers at pre-selected sites on the protein surface [30][31][32][33]. By attaching PEGs at specific positions, 90 one can avoid obstructing the active site and limit the degree of PEGylation to maintain protein structural 91 dynamics thus to retain full catalytic activity [14,15]. This strategy has been partially validated for L-92 asparaginase through PEGylation at its natural disulfide bond without losing catalytic activity, independently of 93 the PEGs length [34,35]. 94 In this work we report for the first time the simultaneous site-specific PEGylation and intramolecular 95 crosslinking of L-asparaginase subunits at pre-selected canonical cysteines introduced by mutagenesis. The 96 advantage of this approach is that not only the degree of modification is kept to a minimum in order to retain 97 catalytic activity, but careful selection of the PEGylation positions also offers the possibility to target potential 98 proteolytic and immunogenic epitopes. Moreover, the molecular weight was increased at least four-fold 5 99 compared to the native L-asparaginase with as low as one PEG molecule per protein subunit and this 100 significantly reduces the chance for glomerular filtration and anti-PEG binding. PEGylation of this 101 therapeutically important enzyme at canonical amino acids is not reported, even though the conjugation 102 chemistry at cysteine residues is well known and highly specific [36]. The most convenient way to express 103 recombinant L-asparaginase is by secretion into the periplasmic space or culture medium [37][38][39][40], since 104 cytoplasmic expression leads to formation of inclusion bodies making the purification tedious [41]. However, 105 expression of canonical cysteines affects L-asparaginase secretory expression [42]. Herein we were able to 106 express a double-Cys mutation of L-asparaginase as a secreted product, to later perform the site-specific 107 crosslinking of the subunits. Our findings will benefit the evolving technological improvement of L-108 asparaginase as therapeutic agent by setting the prove-of-concept of this alternative PEGylation strategy. The plasmid pET-22b(+) was purchased from Novagen (Darmstadt, Germany). The vector pET22b-  periplasmic fraction was defined as the total activity found in the supernatants after centrifugation. The specific 154 asparaginase productivity (U/g) was calculated as the ratio of catalytic activity (U) per biomass weight (g).

155
Measurements were performed in triplicate.   The site-specific PEGylation crosslinking was carried out by reacting the terminal maleimide groups of 208 the PEG polymers with the surface-exposed cysteines on L-asparaginase using the well-known thiol-Michael were recovered by gel filtration chromatography or by repeated ultrafiltration steps in a 100 kDa cut-off filter.
Since L-asparaginase subunits have a mass of ~35 kDa, non-conjugated subunits should pass through this 220 membrane while tetrameric (or higher) conjugates are retained. caused the neo-conjugate rigidity to decrease until it became completely soluble. The 1kDa-PEG-conjugate was 261 a semi-solid gel (Fig 2A), while the 2kDa-PEG-conjugate was physically a soft gel in which a fraction of L-262 asparaginase (catalytic activity) remained in solution. The 5kDa-PEG-conjugate was a fully soluble construct. 263 We think these results can be explained by the degree of dynamic freedom of the PEG molecule and kinetics of which ran equal to the native L-asparaginase tetramer on a Native-PAGE ( Fig 2B) and did not dissociate during 279 denaturing SDS-PAGE electrophoresis (Fig 2C) proving that the subunits were indeed covalently crosslinked.

283
Note that a substantial amount of the 2kDa-PEG-conjugate did not enter the gel because it was too large.

285
Site-specific PEGylation with the 5000 g/mol PEG was characterized by size exclusion chromatography 286 to qualitatively follow the covalent crosslinking of the subunits. The mutant L-asparaginase (A38C-T263C) was 287 reduced with TCEP prior to and after PEGylation and then analyzed using gel filtration chromatography. Due to 288 the quite low purity of the starting mutant solution, eluted fractions were also evaluated for catalytic activity to 289 corroborate the presence L-asparaginase. Prior to PEGylation, it was observed that the mutant (A38C-T263C) 290 eluted at several fractions in which more than 50% of the catalytic activity was found from 8-13 ml, while after 291 PEGylation most of the activity was found at one fraction which eluted at 8-9 ml (Fig 3). The elution peak of 292 the 5kDa-PEG-conjugated is approximately related to a relative molecular weight of 600-900 kDa, double of 13 293 that reported for L-asparaginase randomly PEGylated using the lysine-directed approach [16,18]  The product recovery after the site-specific crosslinking reaction by gel filtration was modest. The 311 fraction where most of the 5kDa-PEG-conjugate eluted represents less than 20% of the initial catalytic activity 312 exhibited by the starting mutant enzyme solution. For that reason, we also tested ultrafiltration for product 313 purification using 10 and 100 kDa cut-off filters. The washing of the starting mutant enzyme solution after 314 reduction with TCEP was done in a 10 kDa cut-off filter, immediately followed by the crosslinking reaction.

315
The 5kDa-PEG-conjugate was pooled by performing repeated washing/concentration steps with a 100 kDa cut-316 off filter, since the non-conjugated L-asparaginase should flow-through these pores. The recovery by this 14 317 method was about 48% of the initial catalytic activity (S1 Table). This method is simpler and more efficient showed that this approach is linked to a cost in catalytic activity due to the Lys-directed conjugation employed. non-modified L-asparaginase (P<0.05), 116 ± 6 vs. 161 ± 9 U/mg, respectively (Table 1). This was unexpected 368 since the mutated cysteines are relatively far away from the active site (Fig 1) Randomly-PEGylated b 125 ± 10 77 ± 8 391 a Specific activity of native L-asparaginase was defined as 100%.

392
b Commercial randomly-PEGylated L-asparaginase formulation (Millipore Sigma, USA) was used for comparison.

393
Reported values are the average with the standard deviation calculated by error propagation.

395
In summary, a new site-specific PEGylation strategy to crosslink L-asparaginase subunits generated 396 higher molecular weight neo-conjugates and improved the average catalytic activity likely involving an 397 intrasubunit proximity-stabilization mechanism. Compared to previous reported PEGylation strategies, our 398 PEGylation method is the only one that improved the catalytic potential of L-asparaginase (Table 2). A similar 399 method has not been proposed before possibly due to the difficulty to express and purify recombinant L-400 asparaginase containing non-natural cysteines [42]. We are the first to report this kind of mutation in L-401 asparaginase designed for site-specific conjugation, which is beneficial to improve this therapeutic enzyme   increasing degree of modification (Fig 4), which can be explained by the restriction of protein structural  PEG-conjugate could be beneficial for industrial applications that use L-asparaginase. After treating raw 424 products with L-asparaginase to hydrolyze the acrylamide precursor asparagine, the enzyme is difficult to 425 recover due to its soluble state. Our neo-conjugate is a semi-solid gel that can be recovered from solution and 426 reutilized. We tested the catalytic activity of this neo-conjugate within a temperature range from 20℃ to 80℃ to 427 evaluate its potential for being used in industrial applications. The same 1kDa-PEG-conjugate sample was used 19 428 throughout the whole experiment by washing the gel with distilled water after each catalytic cycle. Our neo-429 conjugate proved to be highly reusable along the 27 cycles tested, with a maximum catalytic temperature up to 430 60℃ (Fig 5). Although L-asparaginase is a multimeric protein with the catalytic core in the middle of two 431 subunits [1,56], our discreet conjugation didn't affect its enzymatic function even in an insoluble state. This 432 conjugation strategy could be extended to other industrially important enzymes to improve cost-efficiency.

465
Reported values are the average with standard deviation.

467
The results from secretory expression suggest that the presence of cysteine residues in L-asparaginase is 488 Therefore, elimination of the natural disulfide bond should not significantly affect the catalytic activity. We 489 replaced the cysteine amino acids (C77 and C105) by two serine residues with the intention of minimizing any 490 alteration of the surrounding environment of these residues. We found no significant difference in the specific 491 activity exhibited by the native L-asparaginase and C77-105S mutant (Table 4). This implies that removal of 492 the natural disulfide bond should not destabilize the tetrameric-structure of L-asparaginase.   as starting coordinates for the natural cysteines (C77 and C105) along with four residues (D76, D78, K104, and 501 D106) contained in the model [1]. After this model was geometrically optimized, the sulfur atoms in the 502 cysteines were replaced by hydroxyl groups to generate the serine residues, and the resulting structure was 503 optimized. Interestingly, the C77-105S mutation does not cause significant structural disturbance, as revealed 504 by the superposition of the two optimized structures (Fig 6). It is highly possible that one hydrogen bond (1.828 505 Å O···HO) is formed between the serine residues. More details are presented in the supporting information (S6 506 Fig and S7 Table). These calculations agree with the catalytic activity observed for the C77-105S mutant, 507 showing that no structural disturbance takes place hence the active site should not be affected. These findings 508 support the hypothesis that formation of the tetrameric-structure of L-asparaginase should be driven by other  The final overall yield for this construct was 29 mg of pure L-asparaginase per liter of culture, which is an 521 excellent although not as good as in previous reports using affinity purification methods [37,38,40].

522
Unfortunately, a His-tag affinity purification has been reported to interfere with the catalytic activity of L-  can only be assumed that the non-natural cysteines in the mutant may interfere in some way with the column 528 binding. By last, the mutant C77-105S was purified with a very low yield, although this was expected because 529 of the poor secretory expression observed (Table 3), the purity was also inferior compared to the other  The relative molecular weight of the recombinant L-asparaginase constructs was calculated by mass 532 spectroscopy. The main m/z peak was observed at 34605 g/mol for the native L-asparaginase, while 34634 533 g/mol for the mutant A38C-T263C (Fig 7). Unfortunately, it was not possible to obtain detailed information for Herein we report an alternative site-specific conjugation strategy to modify the therapeutic enzyme L-553 asparaginase. Our strategy is based on three main pillars: firstly, that the molecular weight of L-asparaginase 554 can be increased by crosslinking the subunits without the need to modify this protein excessively; secondly, that 555 the polymers can be linked to the enzyme at pre-selected non-natural residues along the surface, to hide 556 potential proteolytic and immunogenic epitopes; and thirdly, that intramolecular crosslinking of the subunits 557 stabilizes the active tetrameric-structure to enhance catalytic activity. We demonstrated that site-specific 558 PEGylation can be used to conjugate L-asparaginase at two different positions on its surface, while at the same 559 time maintaining its full enzymatic potential. We found that introduction or removal of cysteine residues seems 560 important for the proper extracellular expression of L-asparaginase in E. coli and achieved to purify this 561 recombinant protein using a single anion exchange step.

562
Our 5kDa-PEG-conjugate was fully soluble with a molecular weight about double of randomly-

563
PEGylated formulations, which is beneficial to delay glomerular filtration during the in vivo therapeutic 564 application of this enzyme. Even more important, this neo-conjugate retained full catalytic activity, and we 565 proposed that this is possible due to a tetrameric-structure stabilization effect that accelerates the formation of 566 the active quaternary structure of L-asparaginase. In addition, the 1kDa-PEG-conjugate was a highly reusable 567 semi-solid gel that exhibited catalytic activity and can be used for applications in the food industry. The 568 reported site-specific conjugation approach provides the possibility to modify L-asparaginase pharmacokinetics 569 while maintaining its pharmacodynamics properties, which is beneficial for the application of this drug in 570 biological conditions.