Protein Partitioning In A Droplet-Based Aqueous-Two Phase System Microuidic Device

Aqueous-Two Phase Systems (ATPS) is an important tool for the separation of biological entities as proteins, 33 membranes, enzymes, among others. On the other hand, microfluidics is an emerging technology that studies 34 and manipulates liquids either one single phase or dispersed fluids such as droplets at the micro or smaller 35 scales. Applications of microfluidics in different areas such as molecular biology, biochemical analysis and 36 bioprocess have increased in the last years. In this work, we proposed a droplet-based microfluidic approach to 37 generate ATPS systems and to observe how two model proteins, native ribonuclease A (RNase A) and its 38 PEGylated form (PEG-RNase A), behave and partition on these systems. Using polyethylenglycol (PEG) and 39 potasium phosphate salts as the phase-forming chemicals, we were able to form ATPS systems inside the 40 microfluidic device as commonly performed in conventional ATPS macrosystems. Even more, formation of 41 ATPS systems in which one of the fluids was present as a droplet was also achieved. As expected, model 42 proteins exhibited the same behavior as they do in a macrosystem, that is, they displaced to a particular phase 43 according to their affinity for them. When native RNase A was placed in the salt-rich phase, it remained there, 44 and migrated from the PEG-rich phase to the former. On its part, PEGylated RNase A remained in the PEG- 45 rich phase or migrated from salt-rich phase to the PEG-rich phase. These results open the possibility for a 46 prospect of micro bioprocess to separate interest biomolecules.


68
Aqueous-two phase system (ATPS) is a downstream processing technique used to fractionate proteins, and 69 other small biomolecules, based on their affinity to one of the two phases. Systems are formed usually of a 70 polymer, such as polyethylene glycol (PEG) or dextran, and a salt, but polymer-polymer, or even alcohol-salt 71 systems, have also been explored (Asenjo & Andrews, 2011;Hatti-Kaul, 2001). Beyond the traditional use of 72 ATPS, its implementation in microfluidic devices is a quite new and fertile field where important contributions 73 have been made. Lu et al. [3] experimentally studied the stability of the parallel laminar flow regime of ATPS 74 in microchannels (Lu et al., 2011). To do this, they used branched microchannels, and they were able to generate 75 a map to identify different regimes (i.e., bi-laminate flow or droplets). Furthermore, Meagher et al. [4] 76 demonstrated protein separation in ATPS via diffusion between streamlines, where PEG-rich and salt-rich 77 phases were used. They used FITC to track proteins moving towards the PEG-rich phase. On another 78 experiment they demonstrated the partitioning of bovine serum albumin and β-galactosidase between the two 79 streams (Meagher et al., 2008).

102
Here, we propose an ATPS droplet-based microfluidic device as a first approach to visualize the protein 103 partitioning behavior. The idea behind this is to emulate the traditional PEG-salt ATPS by a droplet-shape phase 104 and a flowing phase outside the droplet at a microscale level. Native ribonuclease A (RNase A) and its 105 PEGylated form were used as protein models as their partition behavior in traditional ATPS systems has been 106 well studied [8,9]. While the former has affinity for the salt-rich phase, the latter has affinity for the PEG rich 107 phase. So, it was hypothesized that both proteins would behave in the same manner inside the ATPS 108 microdroplet generator i.e., remining at or partitioning towards the phase for which they have more affinity.

111
An aqueous two-phases system (ATPS) forms when the components (polymers-water-salt) exceed the solubility 112 limits and generate new phases. A graphic to represent the solubility is the binodal curve (Fig. 1). Above the 113 curve, we found the biphasic systems, and under the curve the monobasic systems. The total concentration of 114 each component of the system is given by the values of x and y coordinates, being the bottom phase in the x 115 4 axis, and the top phase in the y axis. All the systems that share the same tie line length (TLL) have the same 116 physicochemical properties, and as well the same concentration of each component. A TLL (Eq. 1) is 117 constructed by cutting the binodal curve in two points, which represent the composition of each phase, each 118 TLL is at a specific distance from the critic point, just above this point the volume of both phases is theoretically 119 equal. The systems that share the same TLL have different volumes of phase (Fig. 1). This occurs because there 120 is a different solvent (water) migration between both phases. In this way, we can manage the relation between

161
The replica was fabricated by pouring a 10:1 previously degassed mixture of silicon elastomer and curing agent 162 (Sigma Aldrich, 761036) over the master mold, and cured at 120 ºC during 10 min. The inlets of the micro 163 channels were made by using a 2-mm diameter core puncher. Finally, sealing of the device was performed by 164 bonding the PDMS devices to a glass slide (1x3") using an air plasma cleaner (Harrick Plasma, PDC-001,

165
Ithaca, NY, USA), leaving both, replica and glass slide to be exposed to bright pink plasma for 3 minutes.

216
Both phases were introduced to the microfluidic flow-focusing devices to see droplet formation. c) Illustration of the third experiment.

217
PEGylated protein is initially in the salt phase to observe its movement towards the PEG phase. When droplet formation occurs, protein were added to the PEG phase. Thus, the salt + PEGylated protein was solubilized in the disperse phase, while 231 the PEG was present in the continuous phase (Fig. 3c). In the same way, for the fourth experimental setup, 232 native protein was placed on the PEG-rich phase to observe its displacement towards the salt-rich phase (Fig.   233 3d). To do this, we followed the same procedure as in the previous experiment, but in this case, the native

254
In the second scenario, droplets were formed because of the balance between inertial forces, viscosity ( ) and 255 velocity ( ), and surface tension ( ), that is, the capillary number (Eq. 2) In this part, we reduced the flow rate 256 value from 7 μl/min to ~0.7 μl/min. When the system was set to 0.7 μl/min, droplets started to form near the 257 junction (Fig. 4b). As time passed, plug flow, observed from the elongated droplets, started to appear replacing 258 the droplet formation (Fig. 4c). We attributed this to the change of the relation between inertial and surface 259 forces, in which inertial forces decreased as the velocity decreased, giving the opportunity to generate a droplet

264
The approximate value of droplet length obtained with the COMSOL simulations was of 220 μm (Fig. 4d),

265
which is comparable to that obtained experimentally, which was 270 μm. This little difference can be attributed In the second experiment, the salt-rich phase containing native protein was used as the disperse phase (main 276 channel), while PEG-rich phase was used as continuous phase (cross channel). Droplet formation did not occur 277 at the junction of the continuous and disperse channel, instead, two phases flow was formed first (Fig. 5a), and 278 the droplet formation occurred upstream in the main channel (Fig 5b). The fact that droplets did not form is a

302
The third experiment tested how PEGylated protein molecules displace outside the salt phase into the PEG 303 phase. In this scenario (Fig. 6), when droplet formation was achieved, three main stages were identified: 1) 304 protein partition at the interface between both liquids before droplet formation, 2) droplet formation, in which 305 droplets are deformed, apparently due to protein migration from the salt-rich phase to the rich-PEG phase, and

306
3) an apparent droplet stabilization. Droplet length value obtained here was of approximately 370 μm.

307
Presumably, due to the displacement of proteins from one phase to another, the formation of microdroplets was 308 affected, which is observed in the monodispersivity. When introducing PEGylated RNase A in the salt-rich phase to study its partition behavior towards the PEG-315 phase (Fig. 7a), a two-phase flow streamline was achieved. In these streamlines, the outer phases correspond to 316 the PEG-phase, while the non-fluorescent stream in the center represents the salt-rich phase. This stream-like 317 pattern occurs when inertial forces are higher as compared to surface forces, which in turn is traduced in a high 318 velocity ( > 1); here, it occurred at a flowrate of 7 µl/min. In this scenario, it was seen how protein started 319 to migrate from one phase to the other starting at the interface, i.e. the contact region between both phases. This 320 can be observed either in the intersection of the channel (Fig. 7a) or downstream (Fig. 7b). As both phases move 321 downstream, a major fluorescence in the PEG phase is clearly seen, which corresponds to the outer streamlines,

322
while the streamline in the middle, the salt phase, has lost part of the visible fluorescence that it presented at 323 the beginning, indicating that protein moved to the higher affinity phase as hypothesized.

324
On the other hand, lowering the flow velocity favored the formation of droplets. Likewise, the partition of the 325 PEGylated protein from the salt rich-phase to the PEG phase was observed at the interface (Fig. 7c). However,

326
as the droplet forms, at the cross-section area, it can be observed how most fluoresce remains outside the 327 droplets (Fig. 7d). Before this occurs, droplets suffer deformation, which presumable could be attributed to the In the fourth experiment, native FITC-labeled RNase A was introduced into the device by the inlet carrying the 339 PEG-rich phase, that is, the continuous phase. A two-phase streamline was produced again as shown in Fig.  12 (8). It is observed how proteins disrupted the salt streamline due to the protein partition from PEG rich-phase 341 to the salt-rich phase. This happened because inertial forces overcame surface tension ( > 1). In this 342 experiment, three stages were identified. 1) The salt streamline forms (Fig. 8a), followed by 2) a displacement 343 of protein from the PEG phase to the salt phase (Fig. 8b), which caused a discontinuity of the salt phase, and 344 finally 3) a total disruption of the phase, in which protein has already disrupted the salt-rich phase (Fig. 8c).

345
Unlike the other experiments, in this experiment we can highlight that the streamlines were formed followed 346 by their disruption, this can be attribute to the move of the protein from PEG phase (less affinity) to the salt 347 phase (more affinity). We were able to observe a kind of curvature in the streamline (Fig. 8a), which we 348 attributed to the effect of protein displacement from one phase to the other. Experiments for droplet formation 349 were also carried out when studying migration of native protein towards the salt-rich phase. To produce 350 droplets, the flow rate was lowered 10 times, approximately 0.7 μl/min. In this scenario, it was found that three 351 different stages occurred as shown in Fig. (8). In the same way as the previously analyzed scenarios, droplet 352 formation started to occur at the interface. 1) A dispersion of the salt phase in the PEG phase by an apparently 353 droplet detachment is observed. 2) Here, it was found that the protein enters to the droplet since the moment of 354 detachment (Fig. 8d), which in the second stage resulted in an incomplete droplet formation and salt phase 355 deformation (Fig. 8e). And finally, 3) a reshape of the salt droplets with fluorescent protein encapsulation was 356 observed. Although, this is not always the case, since usually the salt phase mix with the other phase (Fig. 8f).

357
This can be attributed to the higher protein concentration in the medium, as well as to the change in the surface

376
Protein partition and separation in microfluidic devices has the potential to become a powerful tool to carry out 377 downstream microbioprocess. In our work, we tested how model proteins, native and PEGylated RNase A, 378 fractionate in a microfluidic system using droplet-based aqueous two-phases systems (ATPS). So far, we have

388
Overall, this first approach can be a part of a total process on a chip. For example, the disperse phase can be the 389 products of a reaction, in which the undesired products can be separated to the other phase according to their 390 affinity, or even droplets acting as microreactors that separate undesired residues at the same time. Today, there 391 are few reports, which work using PEG and salt systems inside a microfluidic device, and even less attempts to