Aqueous two-phase systems as multipurpose tools to improve biomarker analysis

Biomarker analysis in biological samples can boost risk profiling, diagnosis, prognosis and clinical decision-making, overall improving patient care. However, due to the complexity of most biological samples, biomarker analysis is a time, cost, labor-and resource-intensive task, commonly struggling with the lack of accurate results. To improve biomarker analysis, it is critical to develop efficient strategies to remove main contaminant/interfering molecules and to enrich/concentrate intact biomarkers prior to detection/quantification. Also, the development of more reliant and cost-effective quantification techniques and improved point-of-care (PoC) approaches are on demand. To bridge these gaps, properly designed aqueous two-phase systems (ATPSs) are potential candidates due to their high-water content, structural versatility and high selectivity. Over the past years, ATPSs have been shown to improve biomarker analysis, including proteins, nucleic acids, extracellular vesicles, bacteria, and viruses, in the following areas: (i) proteomic studies to expand proteome coverage; (ii) extraction of biomarkers to improve laboratory analysis; (iii) confinement strategies to improve enzyme-linked immunosorbent assays; and (iv) PoC applications, including the development of lateral flow immunoassays and one-pot reactions, and the miniaturization of extraction and purification techniques. This review describes and critically addresses the main applications of ATPSs in each of the mentioned areas by focusing on the types of ATPSs


Introduction
Scientific progress has significantly contributed to the marked increase in life expectancy experienced in the past century [1].On the other hand, both the risk of developing diseases and disease prevalence have increased, calling for the need of improved tools for the prevention, diagnosis, monitoring and treatment of various pathologies [1].Regarding diagnosis and monitoring, the identification and detection of disease biomarkers in biological samples represents an emerging and promising tool in clinical practice [2].Through a more accurate risk level assessment, a timelier diagnosis, a more reliable prognosis and by assisting in the choice of adequate therapies, biomarkers have improved patient care over the years [3].
In conformity with the Biomarker Qualification Program by the FDA (U.S. Food and Drug Administration), a biomarker is "a defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or interventions, including therapeutic interventions" [4].Accordingly, biomarkers can be classified into molecular, histologic, radiographic, and physiologic elements, and categorized according to their alleged applications (monitoring, safety, diagnostic, susceptibility/risk, prognostic, predictive and response biomarkers) [4].Major advances in genomics and proteomics, but also in metabolomics, lipidomics and glycomics, have unlocked the potentialities of many biomarkers as promising tools for the early detection of several diseases [5][6][7][8][9].As a result, the objective analysis of molecules (e.g., protein, nucleic acids) in bodily fluids (e.g., serum, plasma, urine, sputum, cerebrospinal fluids) or tissues can be useful to portray normal or abnormal biological processes, a condition or disease [2].Of all bodily fluids, those derived from blood are perhaps the most widely adopted in biomarker discovery studies as well as in laboratory testing.
Depending on the type and source of the biomarker, several techniques can be used to detect and quantify biomarkers in human fluids.However, due in large part to the complexity of most biological samples and to the fact that biomarkers are low abundance and labile molecules, detection constitutes a technically burdensome task [10].Due to the low limits of detection (LODs) often required, current techniques are commonly operated at centralized laboratories using time-, cost-and resource-intensive approaches [11].Moreover, non-reliable results are frequently obtained, with the analysis of biomarkers being prone to induce false positive or false negative results [10].To cope with the technological complexity and non-reliability of biomarker analysis, the extraction, purification and concentration of target biomarkers from their biological sources is an essential task [12].
Examples of traditional techniques used to extract, purify and/or concentrate biomarkers entail liquid-liquid extraction (LLE) [13,14], solid phase extraction (SPE) [15,16], protein precipitation (PP) [17,18] and sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) [19].LLE involves the extraction of analytes of interest from an aqueous matrix using a water-immiscible solvent [20], such as toluene, chloroform, hexane, and dichloromethane [21].However, LLE requires large amounts of volatile organic solvents, high sample volumes and long extraction times [22].In turn, SPE allows the purification and enrichment of analytes from a sample by their adsorption onto a solid phase (adsorbent) followed by desorption using an eluting solvent [23].Nonpolar, polar, ion exchange, immunoaffinity and mixed mode chemistries have inspired the development of multiple sorbents for SPE [23].Immunodepletion is a popular mode of SPE for the removal of protein contaminants, which takes advantage of the specific binding of antibodies to target proteins [24,25].Most immunodepletion kits commercially available are chromatographic matrices (columns) or other resins with immobilized antibodies used to specifically capture target proteins [26,27].Despite their high specificity, immunodepletion kits lack efficiency in protein removal due to the limited number of binding sites as well as reusability, since most resins are of single use [28].PP involves the denaturation of biomolecules present in the biological sample upon addition of an organic solvent, acid, metal, or salt [20].In addition to the simplicity and cost-effectiveness of the process [29], PP may lead to low sensitivity and possible biomarker losses [30].SDS-PAGE is commonly used before mass spectrometric analysis with the aim of fractionating proteins, therefore allowing the separation and identification of the low abundance proteins [19].In SDS-PAGE, proteins are separated based on their molecular weight, with smaller proteins migrating faster [31].However, this technique promotes protein denaturation, hindering further determination of enzymatic activity, protein binding interactions, and other properties [31].Moreover, it is a time-consuming technique and uses toxic reagents that may pose risk to the user [32].When considering complex samples analysis, SDS-PAGE struggles with masking effects of high abundance over low abundance proteins, allowing only for the identification of a limited portion of proteins [33].
For long, ATPSs have been used mainly to extract, purify, and/or concentrate biomolecules prior analysis, including disease biomarkers from biological samples, or to confine detection molecules (i.e., antibodies) during the analysis step.The use of ATPSs favorably affects the reliability of detection and quantification and, consequently, of diagnosis/prognosis.Mostly focusing on protein biomarkers, but also on nucleic acids, extracellular vesicles (EVs), bacteria and viruses, the related open literature suggests the following as the most investigated and successful applications of ATPSs: (i) in proteomic studies to improve proteome mapping; (ii) in the extraction of biomarkers for laboratory analysis; (iii) as confinement strategies to improve enzyme-linked immunosorbent assays (ELISA); and (iv) in the development of pointof-care (PoC) devices.
There is currently available a multitude of review articles that provide a general perspective on the fundamental and/or applied aspects of ATPSs, shedding light on how to design and develop ATPSs for a given application (e.g., [34][35][36][37].In addition to these, there are some reviews specifically addressing the application of ATPSs in proteomic studies [38], ELISA [39], Solvent Interaction Analysis (SIA) technology [40], bioanalytical measurements [41] and microscale assays [41].More recently, other articles have identified and overviewed the disruptive applications of ATPSs and microfluidic ATPSs, among which the extraction and separation of biomolecules (including possible biomarkers) is highlighted [42][43][44].Contrarily to these published reviews in the field of ATPSs, the current review offers a detailed and specific overview of the potential of ATPSs within biomarker analysis and diagnosis/prognosis by focusing in the four above mentioned areas and underlining the societal impact of the field.It covers available literature published over the last two decades and discusses aspects related to clinical validation, commercialization, and contributions to achieve the United Nations Sustainable Development Goals (SDGs).By describing the progress achieved within each of the described areas, the present review provides brief notions on the design and development of ATPSs and critically overviews the current applications, challenges, and prospects of using ATPSs in the field of biomarker analysis.Fig. 1 M.S.M. Mendes et al. summarizes the information included in the present review, showcasing the types of biomarkers, diseases and samples covered in the literature, as well as the contributions for the SDGs.

Aqueous two-phase systems
Due to simple preparation, structural versatility and water-rich environment, aqueous two-phase systems (ATPSs, also called aqueous biphasic systems -ABSs) appear as promising alternatives to most common liquid-liquid techniques [36].ATPSs have water as the main constituent and, if properly designed, may integrate extraction, purification and concentration of target molecules, including biomarkers, with no significant structural, stability or activity losses [36,45].Fitting within the class of LLE techniques, ATPSs are commonly generated by mixing two solutesconventionally, polymer pairs or a polymer and a saltin water under predefined conditions of concentration, temperature and pH; after allowing for the thermodynamic equilibrium to be reached, a clear interfacial boundary is formed dividing the two immiscible aqueous phases (each one richer in one of the solutes) [46].The most commonly used polymers are polyethylene glycol (PEG), dextran (DEX), and polypropylene glycol (PPG), whereas phosphates, sulfates, carbonates and citrates are some examples of typically employed salts [47].In systems formed by polymer-polymer combinations, the phase formation results from their incompatibility in aqueous solutions, yielding two coexisting phases in mutual equilibrium [48].In turn, when a polymer and a salt are used, the two phases formation occurs due to the "salting-out" effect of salt over the polymer, leading to the formation of a salt-rich and a polymer-rich phase [47,49].If on one hand the replacement of one of the polymers by a salt in ATPSs allows surpassing viscosity and slow phase separation obstacles, on the other hand, it imposes a higher ionicity to the medium, which may be detrimental to molecules with low tolerance to ionic strength [36].Despite their advantages, conventional polymer-polymer and polymer-salt ATPSs may provide insufficient polarity differences among the coexisting phases, thus affecting the efficiency and selectivity of the extraction process [50].To enhance operability and performance, while upgrading their degree of structural versatility, other typologies of ATPSs have been proposed along the past years.Polar organic solvents (e.g., shortchain alcohols, acetonitrile), sugars (e.g., monosaccharides, disaccharides, polysaccharides and polyols), amino acids (e.g., alanine, glycine, lysine, proline), surfactants (e.g., Triton X-100, sodium dodecyl sulfate, SDS), and mainly ionic liquids (ILs), have been proposed as alternative ATPS constituents or additives [51][52][53][54][55][56][57][58][59][60][61][62][63].
Fig. 2 provides a schematization of a theoretical phase diagram including the binodal curve and the tie-lines (TLs).Although ATPSs are generally composed of at least three compounds (ternary systems), their phase diagrams are usually provided in an orthogonal representation due to its higher simplicity to identify target mixture compositions than in a ternary phase diagram.In these, usually the water content/axis is not presented, corresponding to the amount required to reach 100% (w/ w) for a given mixture composition.The characterization of phase diagrams is critical from an application standpoint, since the binodal curve stipulates the frontier between the compositions that form two immiscible aqueous phases, i.e., ATPSs (above the binodal curve) from those that form a homogeneous single phase (below the binodal curve).Moreover, TLs are straight lines whose intersection with the binodal curve provides the compositions of the top and bottom phases for a given mixture composition, with their elongation (longer TL lengths, TLLs) intensifying the differences between the nature of the phases [34,64].By tuning the ATPS components and their initial composition as well as operating conditions, systems bearing two phases of different nature are originated, therefore allowing to carefully adjust the affinity between the ATPS phases and the target molecules.Using mixture points placed along a fixed TL allow projecting and tuning concentration factors (CFs) by creating systems with the same phases' composition, but distinct phases' volume/mass ratios [65].Moreover, longer TLLs provide more flexibility and conditions to generate longer TLLs that can lead to higher CFs.
A specific variant of ATPSs is represented by micellar ATPSs (also known as aqueous micellar two-phase systems, AMTPSs), which are constituted by surfactants [66].Being amphiphiles, that is molecules bearing a hydrophilic (polar) head and a hydrophobic (nonpolar) tail, surfactants form micelles in aqueous solutions when dissolved at concentrations higher than the critical micelle concentration (CMC) [67].Typically, in aqueous milieu, micelles are formed, wherein hydrophilic "head" groups lie contiguous to the adjacent solvent and the hydrophobic "tail" groups are accommodated in the core of the aggregate [67].Micellar ATPSs formation occurs when homogeneous micellar isotropic solutions of certain surfactants split into two immiscible aqueous phases, one micelle-rich and another (somewhat diluted) micellar-poor [68,69].Depending on the surfactant, the micelle-rich and the micelle-poor phase may represent either the top or bottom phase [70,71].Micellar ATPSs can be formulated by using single or mixtures of surfactants of non-ionic (e.g., polyoxyethylene alkyl ethers including Triton® X, poloxamers), zwitterionic (e.g., dioctanoyl phosphatidylcholine) or ionic [e.g., SDS, decyl-or dodecyltrimethylammonium bromide (C 10 TAB or C 12 TAB), ILs] nature [72][73][74].Moreover, ionic surfactants lead to the formation of mixed micelles in the presence of their non-ionic congeners [36], whose structure can be tailored to adjust cloud points and to maximize the extraction efficiency and selectivity for target molecules [72,[75][76][77].It should be however remarked that AMTPSs can be formed solely by one surfactant and water, commonly consisting of binary systems, whereas ATPSs require the presence of water and at least two solutes (at least are ternary systems).
Fig. 3 shows a phase diagram of an AMTPS comprising the coexistence curve and a TL, highlighting the dependence of these systems with both temperature and surfactant concentration.The coexistence curve is translated into a bell-shaped curve, whose concavity indicates the temperature dependency of micellar ATPSs formation [68].More commonly, the two-phase regime is placed above the cloud point (T cloud ), meaning that the coexistence curve allows ascertaining the conditions under which aqueous solutions of surfactants are biphasic, thus suitable to conduct extraction and purification processes [68].
Being defined as the large-scale study of proteins, proteomics uses spectroscopic, chromatographic and electrophoretic techniques to study the proteome of complex biological samples, such as serum, urine or even cell lines [94].However, proteome analysis is a complex task due to the unequal concentrations of distinct proteins in biological samples.As a consequence, it is necessary to deal with high abundance proteins that interfere in the analysis and limit the detection of lower abundance proteins, such as disease biomarkers [95,96].To this aim, a biological sample pretreatment is commonly employed involving depletion, fractionation, and/or enrichment procedures, which, if possible, should be conducted in one step [97].Depletion strategies consist in the removal of proteins and/or other contaminating high abundance molecules from the sample, leading to more complete and sensitive analyses of specific molecules including biomarkers [97].Fractionation and enrichment procedures simplify the complex nature of biological samples.The usual goal is to separate sample components into multiple fractions according to their physicochemical characteristics and to obtain some fractions enriched in the target analytes [97].
In the domain of proteins fractionation, da Silva and Arruda (2009) applied a micellar ATPS at 21 • C and composed of 4.0% (v/v) of Triton® X-114 and 0.8% (w/v) of SDS in 0.2 mM of phosphate buffer (pH 5.0) for plasma albumin removal.Firstly, pH and salt composition were optimized to promote the two-phase separation at room temperature and the extraction of albumin to the surfactant-rich phase.95% of extraction efficiency was reported in only 15 min.Gel electrophoresis was then performed to benchmark the proteomic profile of human plasma pretreated with micellar ATPSs against that of non-pretreated plasma.A more detailed electrophoretic profile of 18 protein bands instead of 10 was acquired after the pretreatment, and thus removal of albumin to the opposite phase, with micellar ATPSs [78].Although propitious, the use of micellar ATPSs for proteins separation in proteomics remains underexplored and limited.
Still in the domain of fractionation and enrichment procedures for proteins, ATPSs composed of two polymers [79,[84][85][86][87][88][89][90], polymers and salts [91,92], polymers and detergents [80,81,93] as well as alcohols and salts [82,83] have been applied.As the most used pair of polymers, PEG and DEX were used in several works [79,[86][87][88]90] in the enrichment of plasma membrane proteins from different samples (e.g., minute brain, rat dorsal root ganglions, rat liver, human breast cancer cell line, pancreatic cancer cell line).Schindler et al. [79] presented a polymerbased ATPS composed of 40% (w/w) of PEG with a molecular weight of 3350 g.mol − 1 (PEG 3350), 20% (w/w) of DEX with a molecular weight of 500000 g.mol − 1 (DEX 500000) and 200 mM Tris (pH = 7.8) for the enrichment of plasma membranes from small brain regions.These systems allowed enriching plasma membranes from all parts of the cells, except for mitochondria and endoplasmic reticulum in the PEG-rich phase.Mass spectrometry (MS) allowed to identify 586 proteins, with ca.26.1-36.5% of which falling within the plasma membrane proteins classification [79].Xiong et al. [86] used an ATPS composed of the same polymers, namely 20% (w/w) of DEX 500000 and 40% (w/w) of PEG 3350, for the purification of plasma membrane proteins but now from rat ganglions cells.The results from capillary liquid chromatography (capLC) coupled with tandem mass spectrometry (MS/MS) analysis after protein separation by SDS-PAGE revealed the presence of 954 proteins, with 205 being membrane proteins [86].Ziegler et al. [87] used ATPSs to isolate plasma membrane proteins from human breast cancer cells before MS analysis.Again, using the same type of polymers, each at 6.6% (w/w) in 200 mM of phosphate buffer (pH = 7.2), the separation of plasma membrane proteins from other intracellular membranes was studied.Ziegler et al. [87] identified a total of 36,907 proteins, of which 13,650 were plasma membrane proteins.Still with the same ATPS type, Zhong et al. [88] proposed the combination of PEG 3350 and DEX 500000, both at 6.4% (w/w), to improve the detection of prohibitin-1.Prohibitin-1 is a mitochondria membrane protein, whose levels correlate with pancreatic carcinoma differentiation.Zhong et al. [88] used ATPS combined with two-dimensional (2D) matrix-assisted laser desorption ionization (MALDI) time of flight (TOF) mass spectrometry (MS, 2D-MALDI-TOF-TOFMS/MS) analysis.This system allowed the isolation of 55 proteins, of which 31 were membrane proteins, including prohibitin-1 [88].Fig. 4 schematizes two representative approaches based on ATPSs composed of PEG and DEX to enrich target proteins.Fig. 4A shows an enrichment strategy for plasma membranes from human peripheral blood mononuclear cells, where the interface of the ATPS played a role.According to the technology proposed by Everberg et al. [89], plasma membranes were desired to be enriched at the interface, while other membrane proteins were expected to migrate to the bottom DEX-rich phase.Everberg et al. [89] formulated the ATPS using 4.8% (w/w) of PEG with a molecular weight of 8000 g.mol − 1 (PEG 8000), 3.8% (w/w) of DEX 500000, 90 mM of sodium phosphate buffer (disodium hydrogen phosphate/sodium dihydrogen phosphate) at pH 6.5 and 0.1 mM of sodium chloride (NaCl).After protein size fractionation by SDS-PAGE and liquid chromatography with tandem mass spectrometry (LC-MS/ MS), 80 proteins were identified, with 72% being plasma membrane proteins [89].Everberg et al. [89] concluded that ATPSs are a quicker strategy (approximately 1 h) for the enrichment of proteins when compared with other methods currently performed, such as sucrose gradient centrifugation that usually takes ca.8 h.In the approach represented in Fig. 4B, Cao et al. [90] showed, for the first time, the possibility to integrate ATPSs with sucrose density centrifugation. Co et al. [90] addressed the possibility to enrich plasma membrane proteins from rat liver for SDS-PAGE, mass spectrometry and bioinformatics analyses.20% (w/w) of DEX 500000 and 40% (w/w) of PEG 3350 were used to form the ATPS, allowing for the detection of 428 membrane proteins by nano Liquid Chromatography tandem mass spectrometry (nano-ESI-LC MS/MS) after separation by SDS-PAGE.Of these proteins, 67.1% are plasma membrane proteins [90].Moreover, results showed a 25-fold enrichment of plasma membrane proteins in the top PEG-rich phase, compared with total tissue lysate [90].
In addition to polymer-polymer ATPSs, polymer-salt ATPSs have been studied as well.Qu et al. [91] studied ATPSs formed by 15% (w/w) of PEG with a molecular weight of 4000 g.mol − 1 (PEG 4000) and variable compositions, from 6 to 12 % (w/w), of a phosphate salt for the selective separation and enrichment of proteins from human fluids (plasma and saliva).The probe molecules used were bovine serum albumin (BSA), cytochrome c, lysozyme, myoglobin, and trypsin.Results from reversed-phase high performance liquid chromatography (RP-HPLC) proved the selective separation and enrichment of proteins in min.Moving beyond the widely employed PEG and phosphate-based salts, Salabat et al. [92] selected PPG with a molecular weight of g.mol − 1 (PPG 425) in addition to PEG with a molecular weight of g.mol − 1 (PEG 6000) as the polymeric constituents of ATPSs.These polymers were combined with three sulfate salts, namely magnesium sulfate, ammonium sulphate and sodium sulfate.The model proteins studied by Salabat et al. [92] were BSA, lactoglobulin and zein (in synthetic mixtures).SDS-PAGE results showed the prefractionation of proteins into multigroups: BSA and lactoglobulin exhibited a preferential partition to the salt-rich phase whereas zein majorly partitioned to the polymer-rich phase.The proper tailoring of the monomeric structure of the polymers was key for the protein fractionation to succeed.PPG was more adequate for the extraction of zein than PEG due to the absence of protein precipitation events.The most promising results were obtained with the system composed of PPG 425 and sodium sulfate with a tie-line length (TLL) of 33.3% (w/w), with a recovery efficiency of 93.7% of zein in the PEG-rich phase [92].
Overall, high abundance water-soluble proteins and insoluble materials partitioned to the polymer-rich phase, while membrane proteins were enriched in the detergent-rich phase [80,81].Parameters optimization such as pH increase, addition of the anionic detergent SDS, or addition of chaotropic salts enhanced the ATPS separation performance [80,81].On the one hand, polymer-detergent-based ATPS combined with ion-exchange chromatography and one-dimensional electrophoresis surpass the need for a isoelectric focusing step of membrane proteins that causes aggregation in two-dimensional electrophoresis, providing a better protein coverage [80].On the other hand, detergentmediated solubilization followed by ATPS is a non-denaturing approach that safeguards the integrity of protein complexes; therefore, a more comprehensive proteomic analysis is granted, where not only integral membrane proteins, but also complexes can be identified [81].Among the 186 proteins in the detergent-rich phase uncovered by liquid chromatography-mass spectrometry (LC-MS/MS) with high confidence level (>99.95%),integral and peripheral membrane proteins as well as subunits of membrane protein complexes, chaperones and ribosomal proteins were found [81].
Further widening the types of ATPSs available for the fractionation of proteins, Bai et al. [82,83] focused on short-chain (n ≤ 4) alcohols and salts as the ATPS constituents.Bai et al. [82,83] studied the proteins human serum albumin (HSA), zein, and ç-globulin into multigroups.In a first attempt, Bai et al. [82] used n-butanol, ammonium sulfate, and water in the system formulation and further optimized the following operation parameters: pH, alcohol volume, protein and salt concentration.Firstly, the model proteins were separated using the ATPS and then gel electrophoresis analyses were carried out.The results showed that the top and bottom phases presented different proteins, indicating that ATPSs are capable of separating proteins in multigroups [82].In a second approach, Bai et al. [83] used systems bearing isopropanol, ammonium sulfate and water.Gel electrophoresis analysis revealed the ability of the system for prefractionation, since top and bottom phases as well as the interface extracted different proteins based on their solubility [83].However, n-butanol displays a limited solubility in water, restricting its application in ATPSs.Isopropanol and other shorter size alcohols (e.g., methanol, ethanol, propanol) are more adequate to form ATPSs with salts and could be more suitable for future proteomic works [98].
In case ATPSs fail to meet the selectivity required, the incorporation of affinity components is a propitious solution [99].In this ambit, Everberg et al. [84] developed a method for the enrichment of Escherichia coli (E.coli) inner membranes expressing a His-tagged integral membrane l-fucoseproton symporter (FucP).Nickel-nitrilotriacetic acid (Ni-NTA) immobilized on agarose beads was the affinity element introduced in the ATPS.The composition of the system was 6.45% (w/ w) of PEG 3350, 8.45% (w/w) of DEX with a molecular weight of g.mol − 1 (DEX 40000), 10 mM borate-Tris (pH = 7.8), 320 µL of Ni-N-TA-agarose slurry, and 20 mM of imidazole for the total weight of 1.00 g.The enrichment of the inner membranes occurred to the DEX-rich phase due to the interaction of the beads with FucP.After SDS-PAGE, LC-MS/MS data enabled the identification of 106 proteins.Of these, 36 inner membrane proteins were enriched using the ATPS procedure, compared to 29 using the sucrose gradient centrifugation method [84].Taking advantage of the affinity between microsomes and plasma membranes, Schindler et al. [85] addressed the potential of affinity ATPSs for separating and enriching brain plasma membrane proteins from Sprague-Dawley rats.Microsomes were added to 6.3% (w/w) of PEG 3350 and 6.3% (w/w) of DEX 500000 at pH 7.8 controlled with mM of Tris/Sulfuric acid.As revealed by protein assays using marker enzymes, 12.3-fold enrichment in the PEG-rich phase compared with the initial homogenate was obtained.Moreover, the membrane-protein enriched fraction was subjected to SDS-PAGE and LC-MS/MS, allowing the identification of 506 proteins; of these, 197 were plasma membrane proteins [85].
Despite the improved proteome coverage afforded by ATPSs (in both the presence and absence of affinity elements as well as combined with detergent-mediated solubilization or with gradient centrifugation techniques), the works reviewed mostly focused on ATPSs comprising two polymers, mainly the PEG-DEX pair and most of the times with the same molecular weight [79,[86][87][88][89][90].Although there is a large array of compounds that can be investigated in the field, which can lead to different fractionation patterns in the proteomics field, the ATPS constituents studied so far are quite limited, thus restricting the potentialities of ATPSs in proteomic studies.Furthermore, considerable attention is given to the development of techniques that can expedite proteome analysis, especially considering membrane proteins; yet, only few works report the role of identified proteins as potential biomarkers, such as prohibitin-1 [88].By allying ATPSs structural versatility with the need to establishing new disease biomarkers, it is expected that future proteomic studies will highly benefit from the application of ATPSs in fractionation and enrichment procedures.

Proteins
Due to the structural and metabolic roles played by proteins in humans, abnormal expression of proteins is often associated with a certain pathology; therefore, these can be useful biomarkers in medical diagnostics, prognostics and therapeutics [114].Protein biomarkers are low abundance molecules found in body fluids [10,115], whose detection is frequently based on gel electrophoresis [116,117], enzyme-linked immunosorbent assay (ELISA) [118], surface plasmon resonance (SPR) [119,120], surface enhanced Raman spectroscopy (SERS) [121,122], mass-sensing BioCD protein array [123], fluorescence methods [124,125], colorimetric and chemical assays [126,127].ATPSs have been reported to extract proteins from biological samples, such as serum [104,111] and plasma [105], envisioning a more reliable detection of pathologies including bone [111], liver [104], and nutritional disorders [105].
Assisting the diagnosis of bone diseases, Raymond et al. [111] showed that an ATPS composed of PEG 6000 and DEX 500000 can separate the bone disease biomarker alkaline phosphatase (ALP) from its isoforms.ALP is anchored to cell membranes by a glycanphosphatidylinositol (GPI) moiety and occurs in different isoforms in body fluids, such as human serum or bile, or supernatants after extraction from membrane fraction from tissues or cells.In serum, a GPI phospholipase can degrade the GPI anchor of ALP reducing its lipophilicity, whereas in isoforms found in other body fluids anchor remains unchanged.In this work [111], the partition behavior of various isoforms of ALP was analyzed, namely ALP in human bile (B-ALP), human serum (S-ALP) and human placenta (P-ALP).ATPS formulations consisting of variable concentrations of PEG -from 4.5 to 7% (w/w)and DEX -from 5 to 7% (w/w) -were investigated.The results obtained showed that the P-ALP was the isoform with higher affinity for the top PEG-rich phase.For example, in a system containing 5% (w/w) of PEG and 4.5% (w/w) of DEX, ca.70% of P-ALP partitioned into the PEG-rich phase, while ca.50% of B-ALP and S-ALP also partitioned into the PEGrich phase.By increasing the concentration of polymer, the partition of all isoforms into the PEG-rich phase decreased, thus increasing selectivity.S-ALP was the isoform with higher affinity to the DEX-rich phase, with 78.6% of S-ALP partitioning into such phase in the system formed by 7% (w/w) of DEX and 7% (w/w) of PEG.The B-ALP isoform precipitated in the interface (73.7%) in the system formed by 5% (w/w) of DEX and 5% (w/w) of PEG.Furthermore, the presence or absence of intact GPI-anchors allowed the selective separation of alkaline phosphatase isoforms, which in turn can be manipulated by the DEX and PEG compositions [111].Showing potential for the detection of liver disorders at early stages, Garza-Madrid et al. [104] studied the partition of purified HSA as a model protein followed by a proof of principle using human serum.It has been claimed that ATPS were prepared in flexible disposable devices based on clinical blood bags due to economic benefits and ease of validation.The ATPSs investigated resorted to PEG and a phosphate buffer (potassium dihydrogen phosphate/dipotassium hydrogen phosphate), with several molecular weights of PEG and different TLL being evaluated.Overall, the partition behavior of proteins depends on all these parameters.Gel electrophoresis showed that the largest partition of albumin for the top PEG-rich occurred in systems composed of PEG with a molecular weight of 400 g.mol − 1 (PEG 400) (TLL of 25% (w/w)) and PEG with a molecular weight of 1000 g.mol − 1 (PEG 1000) (TLL of 45% (w/w)), with recovery yields of 80% and 85%, respectively.Remarkably, the highest HSA recovery was 92% in the saltrich phase using an ATPS formed by PEG 3350 with a TLL of 25% (w/w) [104].Although not directly applied in cancer diagnosis, these reported results provide insights on this sense.Regarding the diagnosis of nutritional disorders diseases, Mahn et al. [105] focused on the quantification transthyretina biomarker of selenium consumptionin plasma.ATPSs were investigated to selectively separate albumin and transthyretin from plasma, outperforming currently commercialized immunoaffinity kits in albumin removal.It was possible to separate transthyretin from albumin (as shown by SDS-PAGE) after using an ATPS consisting of 15% (w/w) of PEG 1000 and 20% (w/w) of potassium dihydrogen phosphate/dipotassium hydrogen phosphate.The results showed that albumin majorly partitioned to the PEG-rich phase, reducing its interference in transthyretin quantification that is comparatively present in higher concentration in the salt-rich phase.It should be noted that ATPSs were considered as a first step in the pretreatment of plasma, preceding separation by affinity chromatography.Thyroxine was used as the affinity ligand due to the high binding affinity for transthyretin.However, other, higher abundance proteins (e.g., thyroxine-binding globulin and albumin) also bound to thyroxine, limiting their removal.Even so, the albumin content was reduced at some extent, meaning that the two-step protocol proposed by Mahn et al. [105] can be only foreseen as a tool in nutritional diagnosis or as a preliminary separation technique in its current stage.Further optimization or the introduction of extra purification steps is still required [105].
Based on the set of results described so far [104,105,111], ATPSs composed of polymers and salts are mild approaches for the extraction of protein biomarkers with clinical significance.Although restricted to conventional polymer-polymer and polymer-salt systems and to the detection of single proteins, studies reported made clear that detailed process optimizations are vital to maximize the extraction performance and selectivity of ATPSs.Even so, ATPSs still remain a better option to conduct an initial pretreatment of biological samples, with the need for conjugation with other (higher resolution) separation techniques, such as chromatography.Regardless of the promising selectivity and recovery obtained, there is a lack of studies focusing on more than one biomarker as well as of detailed process optimizations and investigation of alternative ATPSs phase-forming agents and their combinations.
As a strategy to expand the ATPSs phase-forming agents available, ILs stand out due to their "designer solvent" nature [128].Neat ILs can be formed by over a million cation-anion arrangements; by considering binary or even ternary mixtures of ILs, such a value can be increased by up to six orders of magnitude [129].A much wider degree of polarities and affinities in ATPSs is thus granted by ILs than any other type of phase-forming agents [50].ILs used in the formulation of ATPSs are hydrophilic and frequently contain nitrogen-or phosphor-based cations, including imidazolium, pyridinium (both nitrogen-based cyclic aromatic), pyrrolidinium, piperidinium (both nitrogen-based cyclic nonaromatic), quaternary ammonium and phosphonium (both acyclic), which can be additionally tailored with alkyl chains of variable lengths and/or different functional groups [34].These are commonly conjugated with anions comprising halides [e.g., chloride, Cl -, and bromide, Br -), fluorinated compounds (e.g., tetrafluoroborate, [BF 4 ] -), carboxylates (e.g., acetate, [C 1 CO 2 ] -) and amino acids, just to name a few [34].Properly selecting the IL structure and the other phase splitting agent, highly efficient and selective extraction techniques can be developed, particularly showing potential in the extraction/concentration of biomarkers [34].
Du et al. [103] pioneered the application of IL-based ATPSs for the extraction of proteins from biological samples.ATPSs composed of 1butyl-3-methylimidazolium chloride ([C 4 C 1 im]Cl) and dipotassium hydrogen phosphate were studied regarding the extraction of the model protein BSA from spiked artificial urine.The concentration of total protein in each phase was quantified by the Bradford method.Results showed that proteins partitioned to the top [C 4 C 1 im]Cl-rich phase with a CF of 5, while metal species and some other contaminants partitioned into the salt-rich phase.Furthermore, by adding salt to the separated [C 4 C 1 im] Cl-rich phase, an improved CF of 20 was obtained.Remarkably, Du et al. [103] reported extraction efficiencies of almost 100% when the salt amount was 0.75 g.Although a successful proof of concept was delivered [103], the CFs achieved were rather limited.To improve the potential of ATPSs in clinical settings, an IL-based ATPS was later developed to concurrently extract and concentrate the prostate specific antigen (PSA) from urine [65].Pereira et al. [65] not only used the gold standard for prostate cancer diagnosis -PSA -but did also a thorough characterization of the ATPSs phase diagrams to achieve CFs as large as needed.A set of ILs was investigated by fixing the common cation tetrabutylphosphonium [CHES] were selected to further perform concentration studies with both aqueous solutions of PSA and spiked human urine.In concentration studies, LODs together with the cut-off of PSA in human urine enabled the determination of the CFs required to achieve a reliable analysis by size exclusion high performance liquid chromatography (SE-HPLC).Varying the ATPS mixture composition along a unique TL, it was possible to estimate CFs matching the levels of PSA concentration quantifiable by SE-HPLC using the lever-arm rule.Fig. 6 provides a schematic representation of the lever-arm rule applied by Pereira et al. [65].CFs were determined as the ratio between the volume of urine and that of the IL-rich phase (i.e., the phase in which PSA was enriched).After attesting the consistency between theoretically and experimentally determined CFs, CFs of PSA as high as 250-fold were achieved with ATPSs comprising Good Buffers-ILs.Studies involving non-spiked human urine samples [CHES] and contrarily to non-treated samples, it was possible to quantify PSA as well as PSA isoforms, offering new alternatives for the differential diagnosis of prostate cancer using non-invasive samples [65].
Compared to conventional polymeric systems [104,105,111], ILs contribute to develop more efficient extraction and concentration platforms for biomarkers, but only if well designed.Due to possible biomarker losses through precipitation/denaturation/saturation and/or low CFs achieved [65,103], a proper IL design and a careful ATPS characterization is vital for a reliable quantification.Achieving high CFs is particularly relevant if more expedite, yet less sensitive and specific, quantification techniques such as HPLC are used instead of traditional immunoassays.Overall, to reach higher CFs, phase components chosen should bear certain properties (e.g., enhanced hydrophobicity, strong "salting-out" effect) so that enlarged biphasic areas and, thus, wider ranges of workable TLs are yielded [65].For example, it was predicted that ATPS composed of a quaternary-ammonium-based IL and tripotassium citrate could potentially attain maximum CFs as high as 28595fold (it should be noted that this value was reported for pollution tracers from wastewaters) [130].

Nucleic acids
Nucleic acids, namely deoxyribonucleic acid (DNA) and ribonucleic acid (RNA), are used as biomarkers in different clinical assessments including diagnosis, monitoring and therapy of diverse pathologies [131,132].The detection of nucleic acids biomarkers in cells and biological fluids is usually performed using nucleic-acid binding dyes [133][134][135][136], Northern and Southern blots [137], microarrays [138], in situ hybridization (ISH) [139], and quantitative polymerase chain reaction (qPCR) or real-time polymerase chain reaction (RT-PCR) [140,141].ATPSs have been proposed as efficient techniques for extracting and concentrating DNA in cell cultures [108,109] and in plasma samples [106] for diagnosis.
Mashayekhi et al. [108] published a seminal work in the domain of oncologic diseases diagnosis, in which the DNA concentration from mammalian cell cultures using ATPSs was investigated.Micellar systems formed by the nonionic surfactant Triton X-114 at 7.83% (w/w) in phosphate-saline buffer (PBS) at 32 • C were studied.DNA fragments were quantified using the QuantiT dsDNA High Sensitivity Assay Kit (Invitrogen).It was shown that DNA fragments partitioned to the micelle-poor phase, this behavior being driven by steric, repulsive and excluded-volume interactions between the micelles and the DNA fragments.After varying the volume ratio of the phases from 1 down to 0.1, an increase in DNA concentration between 2 and 9 times in the micellepoor phase was achieved [108].More recently, Janku et al. [106] developed two new DNA isolation kits, namely the PHASIFY MAX kit and the PHASIFY ENRICH kit.PHASIFY kits workflows are shown in Fig. 7, consisting in a sequence of two ATPS extraction steps to isolate and purify cell-free DNA (cfDNA) from other contaminants; by the manipulation of the phases' volume ratio, cfDNA is concentrated in the desired phase [106].The PHASIFY MAX kit differs from the PHASIFY ENRICH kit as the latter entails an extra step of fractionation with a sizeselection solution after the second ATPS.Janku et al. [106] Fig. 6.Application of the lever-arm rule to design concentration platforms for biomarkers using ATPSs based on ILs. Figure adapted from information reported on reference [65].
benchmarked the PHASIFY kits against a conventional SPE kitthe QIAamp Circulating Nucleic Acid (QCNA) kit.Results from a validation study using 91 clinical plasma samples showed an improvement of 60% in cfDNA yield and 171% increase in mutant copy recovery with the PHASIFY MAX kit and a 35% decrease in genomic DNA yield with a 153% increase in mutant copy recovery with the PHASIFY ENRICH kit.Furthermore, kits were applied to isolate cfDNA in plasma samples from oncologic patients that yielded negative digital droplet PCR (ddPCR) cfDNA mutation results by conventional SPE methods.From the 47 samples analyzed, 9 tested positive for the cfDNA mutation and 71.6 mutant copies per milliliter of plasma among the 9 samples were recovered [106].Compared with conventional SPE, PHASIFY kits exhibit higher cfDNA recoveries and improved sensitivities in the detection of mutations.These kits are currently commercialized and associated with the biotech company Phase Scientific [142,143].
For the inherited genetic condition cystic fibrosis, Ribeiro et al. [109] studied the partition of a plasmid vector containing a cystic fibrosis gene in polymer-salt systems.PEGs with different molecular weights (200, 300, 600, 1000 and 8000 g.mol − 1 ) and the salt dipotassium hydrogen phosphate were used in the preparation of the ATPSs.Depending on the PEG molecular weight, concentrations ranging from 10 to 20% (w/w) of each component were used to formulate the systems.Plasmid DNA partition patterns were highly contingent on the PEG molecular weight.In ATPS bearing PEGs with lower molecular weights (<400 g.mol − 1 ), partition occurred to the PEG-rich phase, whereas with PEGs with higher molecular weights the partition occurred to the salt-rich phase [109].ATPSs composed of PEG 300, 600 and 1000 were then selected to appraise the partition of contaminant molecules (proteins, genomic DNA and RNA) by agarose gel analysis.The system containing PEG 1000 led to the best plasmid DNA yields (≥67%) [109]; however, copartition of RNA was observed despite genomic DNA and proteins partitioned to the opposite phase.In ATPS formed by PEG 300, copartition of proteins was observed.PEG 600 led to the best purification of plasmid DNA at the salt-rich phase with a reduced content of genomic DNA present and no proteins detected; regarding contaminating RNA, precipitation at the interface was observed.Moreover, preliminary studies with a real sample (i.e., a plasmid DNA-containing lysate) suggested that the ATPS composed of PEG 600 achieved yields of approximately 100% [109].
Circulating tumor DNA (ctDNA) is liberated from cancerous cells into the blood stream, representing important biomarkers within the context of the liquid biopsy approach [144].The analysis of ctDNA grants potential in cancer early detection and monitoring because ctDNA conserves alterations that are tumor-specific [144].Since ctDNA only accounts for less than 1% of cfDNA, amplification techniques are often required before detection [110].PCR amplification is the most commonly used approach; however, environmental crosscontaminations and poor amplification efficiency for short nucleic acids such as ctDNA restrain further development and clinical utility [110].To encourage the clinical application of ctDNA as biomarkers, Li et al. [110] proposed the use of ATPSs to enhance the sensitivity of amplification-free detection resorting to surface-based methods, particularly the equilibrium Poisson sampling (SiMREPS) assay.In their approach [110], ATPSs acted as a ctDNA concentration approach prior detection, for which a biphasic mixture composition of 37.7% (w/w) of PEG 3350, 2% (w/w) of sodium citrate, and 2.8% (w/w) of sodium chloride was used.As a result of the 20-fold enrichment of ctDNA from plasma into the ATPS salt-rich phase, the capture efficiency was improved from 1% to 18% within single molecule recognition through SiMREPS assays.By this approach, ctDNA detection sensitivity was increased by 300-fold compared to previously reported SiMREPS assays, while using small volumes of the biological sample (10 µL) [110].
The works discussed display the suitability of ATPSs, particularly those composed of surfactants, polymers, and salts, for the purification and concentration of DNA-based biomarkers, with some preliminary results proving the suitability to address the requirements of liquid biopsy based on ctDNA.Studies comprising RNAs, which are also considered promising candidates as biomarkers for multiple diseases including cancer, are currently lacking.Although limited in number, available works delivered promising outputs regarding CFs and selectivity.Still, as discussed with proteins, significant improvements can be achieved in these parameters using IL-based ATPSs.It should be highlighted that bio-based ILs and their derived ATPSs are suitable stabilizing media and extraction platforms for nucleic acids [145][146][147], making them worth studying in the analysis of nucleic acid biomarkers.

Extracellular vesicles
EVs consist of nano-sized lipid bilayer-enclosed nanoparticles released from all cells types [148,149].Depending on their biosynthesis, physical properties and release routes, EVs can be grouped into exosomes, microvesicles or apoptotic bodies [148][149][150].The possibility of isolating EVs from most biological samples and their presence in increasing levels due to pathological conditions make them promising biomarkers for disease detection, especially cancer [151][152][153].Presently, EVs isolation and detection resort to ultra-centrifugation (U/C), immunoaffinity and SE-HPLC [154].However, the complexity of the methods together with the low yields and purities as well as compromised EVs integrity is precluding further development [155].Given the absence of an universal method for EVs isolation from their biological sources, ATPSs have been proposed as a viable candidate to make clinical application a reality [101,102,107,112,113].Particularly, ATPSs bearing PEG and DEX have shown efficacy in extracting and concentrating EVs from plasma [102,112], serum [107], urine [101], prostate tissue [112], and other samples such as plants, cell cultures, and parasite cultures [113].
In a triad of works, the Park's group has investigated the extraction of EVs from human fluids, namely blood-derived and urine samples [101,102,107].For blood-based diagnostics, the separation of EVs, particularly melanoma-derived, from proteins using ATPSs was evaluated [102,107].EVs recovery efficiencies using ATPSs were compared with those of conventional methods, such as U/C and ExoQuick ® (Polymer that induces EVs precipitation) [102,107].EVs partitioned to the DEX-rich phase with recovery efficiencies ranging from 70% to 75% and CFs of ca. 5 in approximately 15 min, outperforming both U/C (recovery efficiencies between 8% and 16% and CFs of 0.56) and Exo-Quick ® (recovery efficiencies 40% and CFs not reported) [102,107].Moreover, by combining ATPSs with a batch procedure through the replacement of the PEG-rich phase by fresh PEG-rich phase, ca.95% of the contaminant proteins were removed, thus increasing the purity of the isolated EVs [107].To confirm the diagnostic applicability of ATPSs, Park and collaborators [102,107] performed RT-PCR and western blotting, verifying that mRNA from melanoma cells and CD81 were present in the isolated EVs.Although more investigation is needed, the clinical usefulness of this technology [102] was further supported in a preliminary work studying the diagnostic and prognostic potential of EVs for prostate cancer [112].Using specimens (plasma and prostate tissue) from the Korea Prostate Bank, Park et al. [112] have been able to perform the differential diagnosis of prostate cancer and benign prostatic hyperplasia and to establish a good correlation between the EVs levels in plasma and both the severity and recurrence of prostate cancer.Using urine from cancer patients, Shin et al. [101] further expanded the types of biological fluids and diseases covered with EVs.Fig. 8 shows the applicability of ATPS in assisting prostate cancer diagnosis.As with the blood-derived samples [102,107], EVs partitioned to the more hydrophilic DEX-rich phase, with recovery efficiency of ca.97% within 30 min, and outperformed conventional U/C (recovery efficiencies ranging from 6.85% to 21% depending if one or two U/C cycles were performed) [101].Furthermore, the detection of EVs from prostate cancer cells by ELISA and PCR was enhanced by ATPS [101].Shin et al. [101] evaluated sensitivity and specificity of analysis of EVs for the differential diagnosis of prostate cancer using receiver operating characteristic (ROC) curves.
Results indicated a larger area under the curve (AUC) than other diagnostic methods, such as the measurement of total PSA (tPSA) in serum, demonstrating the better diagnostic performance of EVs isolation with ATPS.The sensitivity and specificity obtained for the EVs isolation using ATPS were 0.8 and 1, whereas for the tPSA test were 0.7 and 0.7, respectively [101].
In spite of the good performance and promising diagnostic outputs reported [101,102,107], there is still room for further recovery efficiency and CFs improvements.This can be achieved by employing other ATPS typologies.In addition, extraction times between 15 and 30 min could be reduced if at least one of the polymers is substituted by a less viscous compound.Furthermore, it is important to highlight that the ATPS-mediated EV isolation approach is highly flexible as it was proven to be easily transposed to other biological samples (e.g., plant, cell culture, and parasite culture sources) [113].In this way, several possibilities in applications requiring the analysis of EVs as biomarkers can be created.

Confinement strategies to improve enzyme-linked immunosorbent assays
This section highlights the advances achieved by using ATPSs as confinement strategies to overcome the limitations of traditional ELISA in diagnostics.ATPSs composed of PEG and DEX have been proposed as a strategy to prevent the cross-reactions through the confinement of antibodies at specific sites of the assay plate [156][157][158][159][160]. Table 3 provides a summary of the works performed in this context.
Although allowing for the detection and quantification of a specific protein in a complex mixture [161], conventional singleplex ELISAs are limited in diagnosing complex pathologies such as oncologic, immunological and neurodegenerative disorders [162].The detection and quantification of panels of biomarkers instead of single molecules allows achieving a more reliable and differential diagnosis and prognosis for specific diseases.When the analysis of panels of biomarkers is required, multiplexing ELISA allows for the expedite quantification of multiple biomarkers using smaller sample volumes [163].Multiplex ELISA formats require however several washing steps and are prone to crossreactions among detection antibodies, and thus to yield false positive results [164].An expert note by Frampton [39] has outlined the potential of ATPS-ELISA to enhance multiplex diagnostics when compared to its conventional counterparts.Accordingly, not only cross-reactions are prevented, but also several technicalities are improved, such as: (i) no need for validation when foreseeing the analysis of additional biomarkers; (ii) universal application as cheap and widely accessible polymers are used instead of trademarked reagents or equipment; (iii) compatibility with currently existing ELISA kits; and (iv) reduction of reagents, samples and costs [39].
Fig. 9 offers a general overview of the ATPS-ELISA strategy and its technological developments, further providing a comparison with a conventional assay and by evidencing the benefits in terms of crossreactions.The common version of the ATPS-ELISA method is based on the following concepts: (i) the higher density of DEX droplets compared to PEG make DEX drops drop down and lie contiguous to the assay plate throughout incubation; (ii) the interfacial tensions between PEG and DEX, and DEX and the assay plate lead to the formation of dome-shaped and stable DEX droplets; (iii) the partition of detection antibodies to the DEX-rich phase allows them to be confined in the DEX-rich phase [156][157][158][159].The multiplex detection of protein biomarkers with potential use in the diagnosis of Graft versus host disease (GVHD) -an immune-mediated condition that is one of the largest causes of mortality after bone marrow transplants [165] was studied [156][157][158][159].The capacity to partition antibody/bead reagents into one phase, namely the DEX-rich phase, was addressed using ATPSs composed of 20% (w/w) of PEG with a molecular weight of 35000 g.mol − 1 (PEG 35000) and 20% (w/w) of DEX 500000 [156,158], and composed of 18% (w/w) of PEG 35000 and 18% (w/w) of DEX with a molecular weight of 10000 g. mol − 1 (DEX 10000) or DEX 500000 [157].Antibodies partition took place to the DEX-rich phase, due to its lower molecular weight and more hydrophilic nature than the PEG [156][157][158].In the multiplex assay, plasma samples were fed in the PEG-rich phase and the proteins constituting the different panels of GVHD biomarkers studied partitioned to the DEX-rich phase (together with the antibodies) [156][157][158].The composition of the protein panels investigated was as follows: (i) hepatocyte growth factor (HGF), elafine, soluble interleukin-1 receptorlike 1 (ST2) and tumor necrosis factor receptor 1 (TNFR-1) [156]; (ii) C-X-C motif chemokine 10 (CXCL10) and C-X-C motif chemokine 9 (CXCL9) [157]; and (iii) C-reactive protein, transforming growth factor beta 1 (TGF-β1), and CXCL10 [158].With antibodies being confined in DEX-rich phase droplets, nonspecific interactions between mismatched antibody pairs were avoided [156][157][158].
Overall, it has been found that the ATPS-ELISA technology provides superior assay performance when compared to both conventional singleplex and multiplex ELISAs [156][157][158].Contrasted to conventional  singleplex ELISA, reductions in the amount of patient samples are enabled due to the different antibodies being co-localized in the same assay well instead of being placed in separate wells.For example, a 4 times reduction in sample volume was achieved with ATPS-ELISA in the analysis of a panel of four biomarkers [156].Furthermore, a 1.4 to 4.2 times decrease of the LODs of most proteins analyzed was observed as compared to singleplex ELISA [156].As compared to conventional multiplex ELISA, the use of ATPS-mediated confinement approaches can reduce the amount of antibodies and reagents spent [156][157][158].
Technological improvements in the original ATPS-ELISA approach were further attempted.Tongdee et al. [159] used ATPS to reduce the incubation time of multiplex ATPS-ELISA.PEG 35000 and DEX 500000 were used in the preparation of the system and the performance of ATPS-ELISA was evaluated using five cytokines, namely interleukin 6 (IL-6), interleukin 10 (IL-10), tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β) and C-C motif chemokine ligand 18 (CCL18).Tongdee et al. [159] included the horseradish peroxidase (HRP) substrate in the PEG solution, which allowed the reduction of wash steps and the incubation time decreased from 2 h to only 1 h.Later, Eiden et al. [160] proposed a simplified version of the ATPS-ELISA by placing all reagents dehydrated in the ELISA plate, as shown in Fig. 9.By doing so, Eiden et al. [160] alleged that the difficult step of adding small droplets (ca. 1 µL) of reagents required in previous ATPS-ELISA versions [156,157] is circumvented, while eliminating this possible source of error.The quantification performance of this new ATPS-ELISA version was evaluated towards five cytokines, viz.IL-1β, IL-6, IL-8, IL-10 and TNF-α [160], and further benchmarked against the results previously obtained by Frampton et al. [156].Although the LODs for all proteins analyzed were 1.2 to 3.2 times higher than in the previous multiplex ATPS-ELISA version, the prespotted version allows for a more cost-effective analysis due to a reduction in the amount of reagents used [160].
Considerable improvements have been achieved in the diagnosis performance and cost-efficiency of conventional ELISA using the ATPS-ELISA strategy.Up to date, however, reported applications rely on PEG-DEX ATPS-ELISA to diagnose GVHD [156][157][158][159], leaving space to fully cover the potentialities of these systems using other phase-forming components and addressing other diseases.Although the execution of the original ATPS-ELISA version resembles that of conventional ELISA, technological developments through the inclusion of HRP [159] or the prespotted ATPS-ELISA [160] have reduced time, materials and/or technological complexity.Yet, the search for cost-efficient, fast and user-friendlier ATPS-ELISA remains in progress and, as already stressed by Frampton [39], the need for automatized and miniaturized ATPS-ELISA formats is still unsettled.By addressing this need, the accessibility of ATPS-ELISA will be generalized, eventually finding application as a PoC test.

Point-of-Care applications
In this section, focus is on the improvement of biomarker analysis at the PoC, for which ATPSs have been used with the following purposes: (i) to extract and concentrate biomarkers prior detection with lateral flow immunoassays (LFA) [166][167][168][169][170][171][172][173]; (ii) to streamline sample processing by one-pot reactions [174]; and (iii) as miniaturized extraction techniques within microfluidic devices [175,176].Table 4 provides a summary of the works discussed.
Gold standard diagnostic technologies are most often conducted in well-equipped centralized laboratories and involve heavy sample processing steps.Given the complexity of the procedures and the high cost of the analysis, laboratory-equipment reliant techniques are not convenient in resource-limited scenarios, i.e., where access to electricity, laboratory equipment, trained personnel, etc., is insufficient [177].The development of PoC devices is thus necessary, especially for use in low-income countries and to provide better care to a more generalized population [178].The development of appropriate PoC diagnostic tests should comply with the "ASSURED" guidelines proposed by The World Health Organization (WHO) [179]."ASSURED" stands for "Affordable", "Sensitive", "Specific", "User-friendly", "Rapid" and "Robust", "Equipment-free", and "Deliverable to end-users" -the key features that a PoC device should display [179].However, accuracy, reproducibility and cost-efficiency issues remain difficult to solve at the PoC, for which ATPSs can be the solution.

Lateral flow immunoassays
Today, due to its user-friendliness, portability, and limited need for power, LFA is the preferred format of commercially available PoC devices [180,181].However, when compared to laboratory-based approaches including ELISA and PCR, LFA provides lower sensitivity of disease detection [182].To enhance the sensitivity of LFA, ATPS were firstly tested as a concentration technique assisting the detection of the model protein transferrin [166][167][168].In these proofs of principle, polymer-salt ATPSs (PEG + potassium dihydrogen phosphate/dipotassium hydrogen phosphate) [166,167,183] and micellar ATPSs (Triton X-114 in PBS at 26.1 • C) [168] were used.Dextran-coated gold nanoparticles (DGNPs) or colloidal gold particles both anchored with specific transferrin antibodies were used for detection purposes [166][167][168]183].
After observing a preferential partition of the nanoparticles for the saltrich [166,167,183] or micelle-poor phase [168] depending on the ATPS type, transferrin was quantified by LFA.Through the correct choice of the volume ratio of the phases (i.e., top to bottom phase from 9:1, 1:1, or 1:9) and of the ATPS type, improvements between 10 and 100 times in the LOD of transferrin were achieved [166][167][168]183].In the case of the PEG-potassium phosphate systems, the LOD was decreased from 1.0 ng.µ L − 1 to a minimum of 0.01 ng.µL − 1 , whereas for micellar ATPS the LOD was lowered from 0.5 to 0.05 µg.mL − 1 [166][167][168]183].Following these preliminary studies with the model protein transferrin [166][167][168], the feasibility of ATPSs for the diagnosis of malaria, especially in low-income nations, was validated [169].To do so, Pereira et al. [169] took advantage of the fact that a micellar ATPS was able to separate on paper to integrate the concentration step by ATPS and detection step by LFA into a single device.ATPS were formulated at 25 • C using a 1:9 volume ratio (top to bottom phase) and Triton X-114 with PBS as the constituents.This system was applied to concentrate the malaria protein biomarker plasmodium lactate dehydrogenase (pLDH) from undiluted bovine serum.To detect pLDH, gold nanoprobes (GNPs) with specific pLDH antibodies were employed.GNPs complexed with pLDH and partitioned to the micelle-poor phase, where the successful detection of pLDH was allowed at 1.0 ng.mL − 1 .This value represents a 10-fold decrease in the LOD compared to conventional LFA (10 ng.mL − 1 ).The integration of micellar ATPS in a 3-D paper-based strip allowed the reduction of the macroscopic phase separation time from at least 8 h to 3 min [169].In addition to protein biomarkers, this type of integrated paper-based devices was also reported in the detection of bacteria [170].ATPS formed by 12.9% (w/w) of the copolymer poly (ethylene glycol-ran-propylene glycol) (EOPO) with 12000 g.mol − 1 and 10% (w/w) of sodium citrate buffer (2.6:1 trisodium citrate:citric acid at pH 5) were applied in the concentration of Escherichia coli (E.coli) from cell cultures. Snce further quantification by LFA with nanozyme signal enhancement in a single step is intended, LFA signal enhancement reagents (such as anti-E.coli platinum-coated gold nanozyme probes, PtGNPs) were also introduced in the ATPS.Results showed that E. coli partitioned to the copolymer-poor phase, being detected at a concentration of 3.3 × 10 4 CFU.mL − 1 with the ATPS-assisted LFA method with signal enhancement.This concentration reflects an improvement of about 30-fold in the LOD when compared to traditional LFA (10 6 CFU.mL − 1 ).A schematic representation of the paper-based device proposed along with its mode of functioning is shown in Fig. 10 [170].The mode of functioning of the device is as follows: (i) after emerging the test strip into the ATPS solution, the ATPS is formed on paper; (ii) E. coli concentration occurs toward the leading (copolymer-rich) phase at the same time that the 3,3 ′ ,5,5 ′ -tetramethylbenzidine (TMB) (one of the enhancement reagents) partitions to the lagging (salt-rich phase); (iii) together with E. coli, anti-E.coli PtGNPs reach the detection module as the leading phase flows; (iv) at the detection module, the capture of E. coli occurs in a sandwich-type assay due to the presence of antibodies at the test line and the PtGNPs; (v) the signal enhancement reaction occurs as the lagging phase reaches the detection module, carrying the enhancement reagent.
To address the possibility for viral detection with ATPS assistance, two works focused on the bacteriophage M13 as a model structure.While Mashayekhi et al. [171] used a micellar ATPS composed of 9.50% (w/w) in PBS at 26 • C [171], Jue et al. [172] applied an ATPS composed of PEG 8000 and potassium dihydrogen phosphate/dipotassium hydrogen phosphate.As with proteins and regardless of the ATPS typology, the manipulation of the volume ratio of the two coexisting phases allowed improving the LOD up to 10-fold [171,172].
The typologies of ATPSs used to assist LFA were further expanded by Yee et al. [173] by moving beyond the more commonly studied polymersalt and micellar ATPSs.Inspired by the remarkable structural diversity of ILs, Yee et al. [173] pioneered the use of IL-based ATPSs, in particular formed by an IL and a salt, for the enhanced detection of biomarkers by LFA.A schematic overview of the IL-based ATPS-LFA technology, highlighting both the working principle of the method and possible LFA formats, is given in Fig. 11.Using E. coli and transferrin as model biomarkers, two LFA formats were investigated: a sandwich LFA format for E. coli detection and a competitive LFA format for transferrin detection.In the former, the reagents involved were GNPs with E. coli antibodies, while in the latter DGNPs carrying specific transferrin antibodies were utilized.In the sandwich LFA format, when E. coli binds to the GNPs antibodies, a complex is formed, consequently binding to the immobilized antibodies on the test line, where the appearance of a red band indicates a positive result.If no E. coli is detected, the complex formation does not occur, and no test line appears (negative result).For the competitive LFA format, the absence of transferrin in the sample leads the antibodies on the DGNPs to bind to the immobilized transferrin at the test line.Under these circumstances, the appearance of two red bands is noticed, which is a sign of a negative result; instead, a positive result is indicated by a single red line at the control zone, which emerges due to the inability of the formed DGNPs-transferrin complex to bind to the transferrin at the test line.Comparing with LFA only, the LODs for both E. coli and transferrin at a volume ratio of 1:9 (top to bottom phase) were decreased 8 (from 3.6 × 10 5 to 4.5 × 10 4 CFU.mL − 1 ) and 20 times (from 5 to 0.25 ng.μg − 1 ), respectively, after the ATPS concentration step [173].Nevertheless, the IL used was the 1-butyl-3-methylimidazolium tetrafluoroborate ([C 4 C 1 im][BF 4 ]) which, despite the good results, is easily hydrolyzed forming hydrofluoridric acid [184], being thus not appropriate for ATPS formulation.As shown in the work by Pereira et al. [65] on PSA quantification, other bio-friendlier ILs can be used to form ATPS and to extract biomarkers.
So far, significant reductions, i.e., up to 100-fold [166,167], in the LODs as compared with traditional LFA formats can be achieved.Also, the possibility of integrating the ATPS and LFA in a single (paper-based) device represents a steppingstone towards enhanced portability.
However, most studies are more centered on model rather than clinically relevant biomarkers and samples, as done with malaria [169].Such a transition allied with the high-performance and structural versatility of ATPSs as well as their capacity to be formed on paper is crucial at this point to expedite the development of novel LFA diagnostic tests with important impact on resource-limited scenarios.

One-pot reactions
In the framework of infectious diseases diagnosis, nucleic acid amplification test (NAAT) is the preferred choice technique for the detection of infectious agents due to high specificity and sensibility [185].However, it comprises multiple steps: cell lysis, extraction of DNA, amplification and optical detection [186].Amplification is most often accomplished via PCR, which is an expensive and laboratoryreliant technique.
Fig. 12 shows a simplified NAAT developed through the joint use of a micellar ATPS with a more portable amplification approach, namely the thermophilic helicase-dependent amplification (tHDA).With this approach [174], it was aimed to overcome the number of steps, high costs and need for qualified staff of the conventional method.To accomplish the proposed goals, the micellar ATPS components, the tHDA reagents and the sample are mixed at 68 • C. The simplified NAAT performs by integrating cell lysis (through the action of heat and Triton X-114), extraction (through the concomitant partition of DNA and tHDA reagents to the micelle-poor phase) and improved amplification in a one-pot reaction.Lower LODs of 10 2 CFU.mL − 1 than those of the already existing tHDA (10 7 CFU.mL− 1 ) were obtained, significantly improving the performance of portable amplification approaches [174].A successful proof of concept was provided with the model pathogen E. coli O157:H7, although Cheung et al. [174] claimed an easy adjustment of the one-pot reaction to detect specific DNA from other pathogens.Even if more studies are needed, the pioneering micellar ATPS-tHDA test can be envisaged as a flexible and ease-of-use approach to detect infectious diseases at the PoC, particularly in resource limited scenarios.

Microfluidic devices
Following the progress of microfluidic technologies along the years, unmet needs in disease diagnostics at the PoC are now possible to be met Fig. 11.Combined use of ATPSs and lateral-flow immunoassays for the detection of biomarkers.Figure adapted from information reported on reference [173].[187].Microfluidics allow for the miniaturization, automatization, and integration of multiple tasks including sample preparation, detection, and analysis within portable, user-friendly, and cost-effective devices.Works hitherto reported focused on the miniaturization/automatization of the biomarker extraction and purification using ATPSs.While keeping or even improving the performance of common (batch, larger scale) ATPSs, ATPS and microfluidic technologies were coupled with the following main goals [175,176]: (i) to reduce the sample volume usually needed in common (batch) ATPSs (>100 µL), (ii) to avoid the manual collection of the phases that may limit the yield of the extraction; and/or (iii) to speed up the extraction process.In the microfluidics arena, polymer-detergent and polymer-polymer systems were applied with success in two seminal works for the extraction and purification of membrane proteins and EVs.
As previously discussed for proteomic studies resorting to larger scale ATPSs [93], polymer-detergent ATPS combine detergent-mediated protein solubilization and ATPS-mediated enrichment.This typology of ATPS combined with microfluidics can be applied to expand proteome coverage as shown with the purification of membrane proteins from cell cultures [175].To test this premise, Hu et al. [175] fabricated a microfluidic device using polydimethylsiloxane (PDMS) with serpentine microchannels, three inlets and three outlets.While the inlets served to feed the device with the PEG-rich phase (two inlets) and a crude membrane extract in detergent (one inlet), at the outlets two water soluble proteins streams and a fraction of purified membrane proteins were obtained [175].In these miniaturized systems, integral and peripheral membrane proteins partitioned to the detergent-rich phase, while the soluble proteins migrated to the PEG-rich phase [175]; this migration pattern is the same as that previously discussed for the batch mode [80,93].Remarkably, 90% of membrane proteins were purified within 5 to 7 s, outpacing the performance of batch ATPS, where 67% membrane proteins were purified from rat liver within at least 5 min [93].
Aiming to separate EVs from plasma, Han et al. [176] proposed a simple microfluidic ATPS using PEG 35000 and DEX 500000 at 3.5% (w/w) and 1.5 % (w/w), respectively.The microfluidic device developed, illustrated in Fig. 13, was made of PDMS, and comprised of three inlets and three outlets.This configuration allowed for the formation of two interfacial layers.PEG was injected at the top and bottom inlets, while DEX was mixed with the EV-containing samples and injected in the middle inlet; at the outlets, two waste streams and an EV-enriched fraction were collected.Prior to move to real plasma samples, Han et al. [176] appraised the performance of the microfluidic ATPS to recover EVs from an EV-protein mixture.By NTA, recovery efficiencies of 83.4% were obtained with a removal of 65.4% of the contaminant proteins (determined by the Bradford assay) from the EV-protein mixture.However, with real plasma samples, challenges related with the presence of other particles of similar size (including some proteins) and the high protein content were faced.These challenges precluded the quantification of EVs by NTA (western blotting was used instead) and imposed the need to dilute the plasma sample prior injection in the microfluidic ATPS.With real plasma, the removal of contaminant proteins reached 51.8%.Given the modest values of protein removal in either case, further optimizations could be performed in both microfluidic design/geometry (e.g., channel width, phase flow rate) and ATPS formulation; for example, Han et al. [176] suggested to optimize the phases' flow ratio or to adopt multiple separation cycles.From the results gathered, the microfluidic ATPS developed might be more appropriate for other human fluids with lower protein levels (e.g., saliva, urine, and cerebrospinal fluid) despite extra studies are required.Moreover, this work evidenced the importance of validating any strategy with real samples.In addition to the weaker efficiency of the extraction method than with synthetic mixtures, the EV quantification technique used was not appropriate for plasma.The transition from model to real matrices and from plasma to other bodily fluids may not be as straightforward as initially anticipated (mainly due to sample complexity), so that several process and analytical aspects may need to be reassessed.

From clinical validation to commercialization: IsoPSA TM as a case study
Even though promising results were obtained so far, the applicability of ATPS-mediated diagnosis in clinical care remains quite restricted.Most works previously discussed do not study samples from real patients or use very small donor populations.Generally, commercialization and application in a real context are lagging behind due to inadequate validation.However, there are successful cases in which ATPSs have moved beyond the laboratory and reached the market, as is the case of IsoPSA TM .
The "Solvent Interaction Analysis" (SIA) method initially proposed by Zaslavsky et al. [40,188] is based on protein partition in ATPSs composed of conventional phase-forming agents (polymers and salts).It can be used for the detection and monitoring of protein biomarkers in biological fluids.This is a straightforward and cost-effective technology that can be used for the characterization and analysis of single proteins and their interactions with other proteins [40,188].Successful applications allowed the discovery and monitoring of protein biomarkers of prostate, breast and ovarian cancers [40,188].
In a preliminary study aimed to determine different PSA isoforms by the SIA method in urine, samples from 222 men scheduled for prostate biopsy were collected for analysis [189].To perform the PSA/SIA test, the partition behavior of heterogeneous PSA isoform between the two phases of an ATPS was appraised.The different partition coefficients of PSA obtained were sufficient for the determination of its composite structural index (K).The K indicates the cut-off value and allows the categorization between malign and benign types of cancer regardless of the tPSA levels, which may not be cancer specific.The validity of the PSA/SIA test was attested by receiver operating characteristic (ROC) analysis and considering the biopsy results.ROC results indicated an area under the curve (AUC) of 0.90 for the PSA/SIA test versus an AUC of 0.58 obtained for the conventional tPSA test.Remarkably, the PSA/SIA test displayed a sensitivity of 100%, specificity of 80.3%, positive predictive value of 80.6%, and negative predictive value of 100%.Altogether, these parameters suggest a better diagnostic performance of PSA/SIA test than conventional methods, such as the tPSA test [189].
Fig. 14 schematizes the SIA technology, which is currently the basis for the company Cleveland Diagnostics, Inc and its main product -the IsoPSA TM test [190].The IsoPSA TM test relies on the partition of PSA isoforms from blood-based samples between the top, inter and bottom phases of an ATPS formed by polymers and phosphate buffer.The Iso-PSA TM test is expected to change the paradigm of the prostate cancer diagnosis by help tackling overdiagnosis and overtreatment of lowgrade disease, undiagnosed cancers and the occurrence of false  positive/negative results [191].In this domain, the IsoPSA TM test can be used to help predicting the risk of high-grade or low-grade prostate cancer through the analysis of the partition behavior of PSA isoforms in a ATPS [192].Considering the patient background and other clinical data, the IsoPSA TM test provides an auxiliary tool for decision-making in patients with alarming tPSA results.The output of the test is the ratiometric parameter K, which can be appraised by ELISA and serves as a clinical cut-off.So far, several ATPSs were screened to identify conditions providing different partition of PSA in samples from patients with prostate cancer and PSA from patients with benign prostate diseases [192].
A study analyzed the partition of PSA isoforms from serum samples of 261 men assigned to prostate biopsy.ROC analysis obtained for cancer versus no cancer endpoint showed that IsoPSA TM test has an AUC of 0.79, sensitivity of 90% and selectivity of 48%.IsoPSA TM test was able to outperform the tPSA test, which has a AUC of 0.61, a sensitivity of 87% and a selectivity of 15% [192].For high-grade cancer versus benign cancer, the AUC obtained for the IsoPSA TM and the tPSA test was 0.81 and 0.69, respectively [192].Furthermore, the clinical validation of IsoPSA TM test was later performed by Stovsky et al. [193].In this case, the test population consisted of 271 samples from men scheduled for prostate biopsy.Results showed that the IsoPSA TM test outperformed the gold standard tPSA test as well as the % free PSA (%fPSA) test, and the two tests combined.ROC analysis for the high grade against low grade cancer endpoint indicated an AUC of 0.784 for the IsoPSA TM test.This value is 0.114, 0.057 and 0.017 units higher than the AUCs obtained for the tPSA test, the %fPSA test and the tPSA test combined with the %fPSA test, respectively.Thus, the IsoPSA TM test based on the SIA technology represents a promising alternative for the identification of malign and benign phenotypes of cancer showing better performance than the conventional tests (either alone or combined).Overall, it contributes to reducing the need for biopsies in patients with the benign cancer, while allowing the identification of patients in need for treatment; yet, it is not able to provide a prostate cancer diagnosis [192,193].At the moment, the IsoPSA TM test is part of the Prostate Cancer Early Detection Guidelines of the National Comprehensive Cancer Network (NCCN®) and helps decreasing unnecessary prostate biopsies by up to 55% [191].

Implications for the United Nations Sustainable development goals
Back in 2015, the United Nations (UN) Member States agreed to commit with the 2030 Agenda for Sustainable Development.This agenda represents "the blueprint to achieve a better and more sustainable future for all" and is based on 17 core Sustainable Development Goals (SDGs) [194].SDGs are designed in a global, cross-disciplinary, and multi-stakeholder manner and aim at coping with the most challenging societal concerns of our times, including poverty, inequality, peace, justice, health, and environment deterioration [194].These represent the priority issues for the 17 SDGs, each incorporating specific targets accompanied by indicators for evaluating the level of attainment [195].The UN SDGs are the following: no poverty (SDG 1), zero hunger (SDG 2), good health and wellbeing (SDG 3), quality education (SDG 4), gender equality (SDG 5, clean water and sanitation (SDG 6), affordable and clean energy (SDG 7), decent work and economic growth (SDG 8), industry innovation and infrastructure (SDG 9), reduced inequalities (SDG 10), sustainable cities and communities (SDG 11), responsible consumption and production (SDG 12), climate action (SDG 13), life below water (SDG 14), life on land (SDG 15), peace, justice and strong institutions (SDG 16) and partnerships for the goals (SDG 17).
Table 5 provides a simplified view of the expected contributions of ATPS-assisted analysis of biomarkers towards attaining SDGs.Based on the literature overviewed, it is our opinion that SDGs 3 and 12 are those more likely to be affected by the application of ATPSs in biomarker analysis.Regarding SDG 3, ATPSs were shown to provide more accurate analyses, which may help promoting the role of biomarkers in risk level assessment, diagnosis, reliable prognosis and clinical or PoC decisionmaking, particularly in resource-limited scenarios.Concerning SDG 12, the inherent ecological and cost-effective features of well-designed ATPSs as well as the possibility for time and resource savings in ELISA, miniaturization and one-pot reactions are particularly relevant.This is in line with the recommendations of Green Analytical Chemistry, where it is considered that every step of analysis (viz., sampling, sample pretreatment, and measurement) should be developed with small dependence on chemical substances such as organic solvents, reduced consumption of energy, good waste management practices and safety ensured for the staff [196].

Conclusions and future perspectives
Herein, a review presenting the main advances accomplished in the field of biomarker analysis by using ATPSs was provided.By reviewing the literature available, it was possible to highlight the critical benefits brought by ATPSs as extraction, purification, concentration and confinement strategies in various facets of biomarker analysis, including: (i) expanded proteome coverage in proteomic studies; (ii) improved analysis of biomarkers in laboratory analysis or consented analysis by more expedite laboratory techniques; (iii) cheaper, faster and simpler analysis of biomarkers by ELISA; and (iv) wider access to biomarker analysis at the PoC, especially in resource-limited settings, through miniaturization, automation, and simplification of the analytical process (ideally through the creation of portable and user-friendly devices).
In general, to develop efficient ATPSs for biomarker analysis, optimization of both extraction/purification/concentration and detection steps should be considered.During extraction/purification/concentration, physicochemical properties, capacity to retain biomarker structure/stability/activity and achievable CFs are at the forefront of the design of efficient ATPSs.At the detection level, interference, and compatibility of the ATPS should be appraised.While the interference of ATPS matrices within detection and quantification may preclude accurate and precise analysis leading to false positive or false negative results, compatibility with analytical equipment may ultimately preclude analysis.To tackle these technical limitations, steps involving the dilution or even the removal of the ATPS components can be introduced in the experimental procedure [197].
Conventional ATPSs mostly formed by two polymers, but also by polymers and salts, are the most investigated systems in each topic herein overviewed; in proteomic studies, however, high interest on systems bearing detergents was also demonstrated due to their solubilization properties.Despite the promising results reported, there is also evidence that other phase splitting agents can improve the extraction performance/selectivity and CFs as well as the system operability (fast separation and low viscosity of ATPSs).Showing remarkable results is the use of ILs, which due to their tunable properties, can be designed to develop efficient extraction and concentration techniques, both in laboratory equipment reliant and PoC applications.Yet, ILs have not been reported in the ambit of proteomic studies or confinement strategies to improve ELISA.In brief, high performance ILs to be applied in ATPSs assisting biomarker analysis should be: (i) able to maintain the target biomarker intact; (ii) water-stable as well as water-soluble, yet hydrophobic enough to form ATPS in combination with salts and to afford high CFs; (iii) with high affinity for the target biomarker, so that the biomarker partition is (ideally) complete to the IL-rich phase of the ATPS, while interfering molecules partition to the other phase; and (iv) compatible with the detection method either undiluted (ideally) or diluted.For laboratory analysis, properly designed ILs allowed not only high extraction/concentration ability for PSA but also its determination (and of an isoform) by HPLCa more expedite and cost-effective quantification technique than the commonly used immunoassays [65].At the PoC, evidence on the compatibility with LFA and improved accuracy was provided with IL-based ATPSs [173].In addition to the conventional polymers and salts or the less investigated surfactants/ detergents and ILs, there is still a variety of other phase-forming agents worth studying, such as alcohols, sugars, and amino acids.Furthermore, most works reported in the literature on polymeric ATPS resort to PEG and/or DEX, but there is room to improve the structural flexibility of polymers by studying other types (e.g., PPG, maltodextrins, pluronics, UCONs).The expansion of the types of ATPS constituents and their combinations will allow achieving superior analytical applications by providing a broader range of interactions and affinities to be explored in biomarker extraction/concentration.
In addition to the possibilities created by ATPSs in biomarker discovery by enhancing proteome mapping, their clinical potential is also underlined in the open literature.It has been shown that ATPSs positively assist the analysis of biomarkers, either single or multiplex, with potential in the detection of a myriad of disorders; indeed, some of these are included as priority diseases in the SDG 3, such as infectious (e.g., malaria) and oncologic diseases.Moreover, the analysis of single biomarkers might not be the most adequate since several biomarkers are not tumor specific.To promote the development of this field, it is expected that research on the development of multiplex approaches will be intensified, particularly resorting to ATPS-ELISA.Regarding biological samples, literature focus has been on blood-derived samples (plasma and serum) and cell cultures or tissues to analyze mostly proteins, but also DNA, bacteria, viruses (bacteriophages) and EVs.It should be however highlighted that the use of less invasive and simpler samples (e.g., urine, saliva) is gaining momentum [65,91,101,103] and is expected to add value to PoC applications, especially when considering more vulnerable populations or people with trypanophobia.
An important weakness identified in several works is the lack of proper clinical validation.While only a few used samples from oncologic patients [101,106,112,192,193], most of the works did not proceed with validation.Representative exceptions are the IsoPSA TM (>200 specimens) [192,193] and the PHASIFY kits (>90 specimens) [106], which already carried on with validations comprising considerable populations and are currently part of the product portfolio of the companies Cleveland Diagnostics, Inc [190] and Phase Scientific [142,143], respectively.Although useful for technology design and optimization, the use of model and/or single molecules with limited clinical value instead of clinically approved biomarkers, panels of biomarkers (e.g., [156][157][158][159][160]) or the identification of isoforms (in the case of protein biomarkers) (e.g., [192,193]) is a more common practice.In this regard, more efforts should be directed into proving the applicability of developed ATPS-based technologies when dealing with more meaningful biomarkers and/or real biological samples.
Although seldom investigated, another possibility of using ATPSs is the creation of all-aqueous droplets, which can be used, among other applications, for bioassays [198].Successful applications are mostly related with ATPS-ELISA [156][157][158][159][160]. Preliminary evidences on the formation of ATPS sessile droplets due to water evaporation for the detection of proteins were also provided using fluorescently labeled BSA as a model protein [199].Regardless of their disruptive nature, allaqueous droplets remain underexplored for biomarker analysis, being expected to become an emerging research topic in the next years.Expected contributions of all-aqueous droplets are mainly related with applications requiring multiplex analysis, small sample volumes and for PoC testing.
Despite the encouraging results, much remains to be done to leverage the application of ATPS-based technologies and ultimately of biomarker analysis in real world (clinical and/or PoC) scenarios.In addition to the questionable clinical validity of several biomarkers investigated and the lack of proper clinical validations, the economic feasibility in relation to the conditions of the recipient health system remain misunderstood.Studies on the environmental impact (e.g., through life cycle assessment) are also missing, despite their relevance within a Green Analytical Chemistry framework and for the SDG 12. Studies dedicated to the automatization, miniaturization, high-throughput analysis and, ultimately, prototypes are also on demand.All these aspects are primordial to support the transition of the most promising technologies to the market, for which the high capital investment required may be an additional limiting step.Despite the long journey ahead, now is the time to bring together multidisciplinary teams embodying medical doctors, engineers, biochemists, and analytical chemists to better understand how ATPS-based technologies compare or can improve the gold standard analytical tools.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Summary of the information covered in the present review.

Fig. 2 .Fig. 3 .
Fig. 2. Phase diagram of a theoretical ATPS including the binodal curve (red dashed line), the experimental percentage by weight data for the binodal curve (red circles) and the tie-lines (orange lines connecting orange circles) for a given extraction point (orange squares).Extraction points placed on the same tie-line yield biphasic systems of variable volume/mass ratios of the phases.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4 .
Fig. 4. Examples of protein fractionation and enrichment strategies resorting to ATPSs composed of two polymers.Figure adapted from information provided in references [89] (A) and [90] (B).

Fig. 5 .
Fig. 5. Examples of protein fractionation strategies based on detergent-mediated solubilization and ATPSs composed of polymers and detergents.Figure adapted from information provided in references [80] (A) and [81] (B).

Fig. 7 .
Fig. 7. Desoxyribonucleic acid isolation protocol using the PHASIFY MAX kit and the PHASIFY ENRICH kit. Figure adapted from information reported on reference [106].

Fig. 8 .
Fig. 8. Extracellular vesicles isolation by ATPSs for the diagnosis of prostate cancer.Figure adapted from information reported on reference [101].

160 ]Fig. 9 .
Fig. 9. Use of ATPSs to develop multiplex ELISAs and their comparison with conventional assays regarding the occurrence of cross-reactions.Figure adapted from information provided in references [156,160].

Fig. 10 .
Fig. 10.Paper-based device integrating preconcentration and capture by ATPSs and detection by lateral-flow immunoassays with signal enhancement.Figure adapted from information reported on reference [170].

Fig. 12 .
Fig. 12. One-pot reaction approach for the detection of desoxyribonucleic acid by combining ATPSs and thermophilic helicase-dependent amplification.Figure adapted from information reported on reference [174].

Fig. 13 .
Fig. 13.Design of a microfluidic ATPS used in the isolation of extracellular vesicles from human plasma.Figure adapted from information reported on reference [176].

Table 1
Summary of the ATPSs investigated and main results in proteomic studies.

Table 2
Summary of the ATPSs investigated and main results in biomarker extraction for laboratory analysis.

Table 3
Summary of the ATPSs investigated and main results as confinement strategies to improve ELISA.

Table 4
Summary of the ATPSs investigated and main results in PoC applications.

Table 5
Possible contributions to the UN SDGs.Details on the sustainable goals, targets and indicators were retrieved from the United Nation website https://sdgs.un.org/goals/ (accessed on 10 May 2022).