Introduction

The resemblance of DESI spectra to typical ESI spectra, alongside little to no sample preparation prior to analysis, gives DESI an advantage over many other ionization techniques [1]. Indeed, since its development just over a decade ago, DESI has shown great versatility for investigating an assortment of analytes [2], such as intact bacteria in vitro and in vivo [3,4,5], secondary metabolites [6, 7], diverse compounds in pharmaceutical industry [8,9,10,11,12], thin-layer chromatography [13,14,15,16], and imaging a wide variety of analytes from biological tissues [17,18,19,20], recently including imaging of low molecular weight proteins [21, 22]. However, DESI suffers a significant mass-dependent loss in sensitivity. As the mass of the protein increases, the limit of detection increases exponentially [23, 24]. Although it is commonly believed ESI and DESI are similar in ionization mechanism, there is evidence that suggests differences between the two [25]. Ionization in DESI is commonly believed to involve the “droplet pickup” mechanism, i.e., extraction of the analyte into the solvent surface layer, followed by liberation of secondary solvent droplets, and finally electrospray ionization mechanisms [26]. Our group has previously developed methods that enable investigating desorption and ionization steps of DESI independently [27, 28]. Those results suggested that the loss in protein signal intensity was not due to problems with physical desorption or ionization, but rather due to incomplete protein dissolution during the desorption step which results in distribution of protein signal across nonspecific protein adducts.

A simple method for improving protein solubility and long-term stability, especially in a concentrated solution of proteins, is addition of amino acids [29, 30]. Such amino acid stabilizers are routinely added to protein solutions during biochemical processes and are favorable additives due to their low cost and safety. Arginine (Arg) and proline (Pro) stabilizers have shown to suppress protein aggregation during refolding [31,32,33,34,35] presumably by increasing the solubility of aggregated proteins [36, 37]. Histidine (His) [38] and alanine (Ala) [39] have demonstrated stabilizing capabilities by suppressing heat-induced denaturation. The stabilizing effect of amino acids against thermal denaturation of proteins and noncovalent protein complexes has been confirmed in ESI-MS [40].

Adducts caused by non-volatile salts such as alkali metal ions Na+ and K+ can cause salt-induced “signal suppression” [41,42,43,44,45] and deteriorate signal to noise ratio (S/N) [46, 47] even at micromolar concentrations [48]. Several methods have been developed to address this problem in ESI-MS, such as buffer loading [41, 47, 49,50,51], supercharging reagents [52], organic vapors [53], and additives such as volatile buffers [54,55,56,57] or salts such as ammonium acetate [49], ammonium bicarbonate, and formic acid [41]. A recent addition to the list of additives is free amino acids which at low millimolar concentration showed removal of sodium adducts during native nESI-MS of large proteins, increased S/N ~ 4 fold, and caused peak narrowing by 10 fold [58]. In this study by Clarke et al., serine was the most successful amino acid in reducing sodium adduction to native-state proteins during ESI-MS, and removal of up to 1 mM NaCl was demonstrated.

DESI, much like ESI, also suffers from well-known interferences caused by non-volatile salts [59]. Similar approaches regarding additives have been carried out for DESI-MS spray solvent composition and some of these additives were successful at improving DESI-MS sensitivity, selectivity, and limit of detection for smaller analytes [60,61,62,63]. Data have shown that addition of ammonium bicarbonate to the DESI solvent system can improve S/N for some proteins between two and three fold relative to the same solvent system containing 0.1% (v/v) formic acid, and more than seven times relative to 50% MeOH:H2O [64].

In this study, we explored the effect of serine as an additive on the analysis of proteins by DESI-MS with different solvent systems. Different proteins with high and low isoelectric point (pI) and molecular weights ranging from 12 to 66 kDa were studied to assess the efficacy of serine in adduct removal and enhancing protein signal. Data show that sodium adducts could be significantly reduced from spiked protein, and signal intensity improvement with co-additives was observed which can be attributed to improvement in dissolution and desorption during the droplet pickup process in DESI.

Experimental

Samples and Reagents

Equine cytochrome c (Cyt c, 12.3 kDa, pI = 10.5), bovine hemoglobin alpha subunit (Hb, 15.1 kDa, pI = 8.0), bovine myoglobin (Myo, 16.7 kDa, pI = 6.8), bovine erythrocyte carbonic anhydrase isozyme II (CAII, 30.0 kDa, pI = 4.7), and bovine serum albumin (BSA, 66 kDa, pI = 5.8) were purchased from Sigma-Aldrich (St. Louis, MO). Proteins were used without further purification unless stated otherwise. Ammonium bicarbonate and L-serine were obtained from Sigma-Aldrich (St. Louis, MO). HPLC-MS-grade methanol and LC-MS-grade formic acid were purchased from Fluka Analytical (Morris Plains, NJ). Ultrapure water was supplied from Thermo-Barnstead Water Polisher. Porous-polyethylene surfaces (PE) with average pore size of 15–45 μm (POREX-4900) were purchased from Interstate Specialty Products (Sutton, MA).

Protein Solutions and Solvent Systems

Stock solutions of each individual protein were made by dissolving lyophilized protein powder in ultrapure water to a final concentration of 400 μM. Serial dilution from 100 mM NaCl solution was used to spike Na+ ions into protein solutions prior to spraying the sample on PE surface. To create homogenous lines of protein, a pneumatically assisted nebulizer made of two coaxial fused silica capillaries [65] was used to spray 80 μM cyt c, 80 μM Mb, 160 μM CAII and 160 μM Hb, and 80 μM BSA separately on PE surfaces. The height of sample sprayer from the surface was ~ 2 mm. Nebulizing gas pressure and flow rate were optimized around 100 psi and 3 μL/min. The result was protein bands with an average band width of 1 mm, which gave approximate surface concentration of 25 pmol/mm2 for cytochrome c, myoglobin, and BSA and about 50 pmol/mm2 for Hb and CAII.

All solvent systems were made in 50% MeOH:H2O. Aqueous stock solutions of 2.0 M ammonium bicarbonate and 2.0 M ammonium acetate were used to prepare 200 mM dilutions in 50%MeOH. LC-MS-grade formic acid was used to prepare 0.1% (v/v) formic acid in 50% MeOH. Serial dilutions from aqueous 1.0 M serine stock solution were used to make different concentrations of L-serine. All the solvent systems and the stock solutions were prepared daily before the analysis.

DESI Source and Mass Spectrometry

A linear ion trap mass spectrometer, LTQ (Thermo Scientific, Waltham, MA, USA), was combined with a 3-dimensional translational stage (Purdue University, West Lafayette, IN, USA) for DESI analysis. An electrospray emitter was prepared from a Swagelok T-piece and two coaxial fused silica capillary tubing [65]. The outer capillary (for sheath gas) was approximately 20 mm in length with an outer diameter of 430 μm and inner diameter of 320 μm. The internal capillary (for solvent) had an outer diameter of 220 μm, and inner diameter of 50 μm. The solvent capillary extended through the T-piece and was connected to a syringe pump which delivered the solvent, and extended 0.5 mm beyond the outer gas capillary. A spray potential of + 4.0 kV potential was applied to the liquid junction of a stainless-steel syringe needle which delivered solvent at flow rate 5 μL/min with N2 as nebulizing gas at 100 psi. The distances between sprayer tip and LTQ heated extended capillary were approximately 4 mm and 1 mm from sprayer to the surface, while the incident spray angle was 55°. Capillary temperature was set at 250 °C. Transfer capillary voltage and tube lens voltage were 30 V and 130 V, respectively. For native-state conserving conditions, DESI solvents were 50% MeOH:H2O or 200 mM ammonium acetate in 50% MeOH:H2O. For denaturing conditions, the solvent was 200 mM ammonium bicarbonate in 50% MeOH:H2O or 0.1% v/v formic acid in 50% MeOH:H2O. ESI experiments were performed with similar conditions, except instead of desorption of proteins from the PE surface, 10 μM protein in each solvent system was directly sprayed into the mass spectrometer inlet.

Data Analysis

Mass spectra were collected by Xcalibur software and viewed in Qual Browser (2.0.7). Four independent trials were conducted for each solvent system. In each independent trial, 4 lines were perpendicularly scanned and averaged. Signal intensity and S/N for each trial were calculated based on the averaged spectra of scanned lines. MagTran software (1.03) was used for charge-state deconvolution to give the “zero-charge” spectra and integrated protein signal intensity as described by Zhang and Marshall using the Zscore algorithm [66]. Error bars represent ± mean standard deviation.

Results and Discussion

Effect of Serine on Sodium Removal in DESI

Cytochrome c Without Added Sodium Chloride

The effect of serine was first studied on mass spectrometry-grade cytochrome c without addition of salt. Multiple studies on sodium adduction have concluded that lower charge states are more susceptible to sodium adduction [48, 49, 67, 68] and as expected, cytochrome c charge states 7+ and 8+ were heavily adducted peaks in the spectra even without doping sodium chloride in the depositing solution. Sodium adduction of cytochrome c charge states 7+ and 8+ when sprayed with 50% MeOH:H2O and 200 mM ammonium bicarbonate (ABC) in 50% MeOH:H2O with and without 1 mM serine are compared in Figure 1. Mass spectra of complete charge-state distributions can be found in Figure S1. In Figure 1a, aqueous cytochrome c without any addition of sodium chloride was analyzed from PE surface with 50% MeOH:H2O, a standard solvent which has been shown to often produce “native-like” charge states of proteins in DESI [69], and was compared to spray containing no other additive but 1 mM serine added to 50% MeOH:H2O (Figure 1b). The result was considerable sharpening of both charge states by removing adducts that spread the signal over multiple peaks and a significant increase in protonated peak intensity especially for charge state 7+. Sodium removal from the same two charge states was also evaluated when the denaturing additive ammonium bicarbonate (ABC) was added into the desorbing spray. This additive was previously shown to increase signal to noise ratio of cytochrome c drastically in DESI [64]. Addition of ABC leads to an increase in the proportion of the protonated form to both charge states 8+ and 7+, compared to 50% MeOH:H2O, but still had multiple adduct peaks (Figure 1c). These adducts were significantly further removed with the addition of 1 mM serine (Figure 1d) together with ABC. Addition of 10 mM serine, however, did not yield better signal, and in fact, suppressed protein signals in both solvent systems, presumably due to high abundances of stable serine clusters, especially protonated serine octamer at m/z 840 and protonated serine dimer at m/z 211 (as discussed later). Formation and characteristics of serine [70], and other amino acid clusters, have been reported and extensively studied by electrospray mass spectrometry [71].

Figure 1
figure 1

Representative DESI-MS spectra of cytochrome c charge state 8+ and 7+ sprayed with (a) 50% MeOH:H2O and (b) 1 mM serine added to 50% MeOH:H2O, (c) denaturing solvent additive 200 mM ammonium bicarbonate (ABC) in 50% MeOH:H2O, and (d) 1 mM serine added to 200 mM ammonium bicarbonate (ABC) in 50% MeOH:H2O. The base peak in panel (a) was [M + 5H++3Na+]8+ and in other panels [M + 7H+]7+ or [M + 8H+]8+ as indicated

Full size image

Cytochrome c with Added Sodium Chloride

Aqueous solutions of cytochrome c (80 μM) containing 1 mM and 10 mM NaCl were spray-deposited on PE and the effects of analyzing these samples with 1 mM and 10 mM serine in 50% MeOH:H2O as desorption spray were investigated (Figure 2). With 1 mM NaCl (Figure 2a), the signal was considerably deteriorated compared to cytochrome c without added salt (Figure S1a); however, protein peaks could still be detected with S/N > 10. With 10 mM NaCl, protein peaks were hardly detectable (Figure 2b). By adding 1 mM serine to desorption spray, signal was significantly improved for both samples. The charge states also shifted from mostly native state like to higher values indicative of protein unfolding (Figure 2c, d) as was also previously reported for a variety of proteins when analyzed by ESI [40]. Unlike the results shown on the analysis of cytochrome c with no added NaCl, 10 mM serine in desorption spray did not suppress protein signal and in fact gave a signal intensity improvement close to the one obtained by addition of 1 mM serine (Figure 2e, f). This suggested that optimal concentration of serine and the tolerance for the amount of serine in desorption spray could also be dependent on the amount of sodium present in the sample. The exact ratio of serine to sodium concentration is more complicated to determine in DESI compared to ESI, as it is dependent on the size of the desorption footprint of the DESI spray, the exact composition of the primary solvent droplet as it reaches the surface and the final concentration of serine in the droplet, and surface concentration of sodium. A rough estimation of the ratios can however be attempted based on simple calculations: For a sample-stage scan speed of 150 μm/s and an estimated 200-μm-diameter DESI desorption footprint, approx. 1.80 mm2 of a sample surface is analyzed per minute. When 80 μM protein sample is spiked with 10 mM NaCl and sprayed onto a PE surface, samples with surface concentrations of 25 pmol/mm2 protein and 3300 pmol/mm2 NaCl were prepared. This leads to an estimated 6000 pmol salt present during the DESI analysis per minute, assuming complete removal from within the DESI footprint. For cytochrome c spiked with 1 mM NaCl, 600 pmol salt was analyzed under the exact same conditions per minute. The amount of serine delivered by DESI droplets when 1 mM or 10 mM serine was added into the 5 μL/min spray is estimated to be 5000 pmol and 50,000 pmol per minute respectively.

Figure 2
figure 2

Representative DESI-MS spectra of cytochrome c spiked with 1 mM NaCl (left) and spiked with 10 mM NaCl (right) when sprayed with (a) and (b) 50% MeOH:H2O (no additive), (c) and (d) 1 mM Ser in 50% MeOH:H2O, (e) and (f) 10 mM Ser in 50% MeOH:H2O. (Int: absolute intensity of protein highest intensity charge state)

Full size image

It was previously shown that serine desalts native proteins during ESI-MS [58]. A decrease in the amount of sodium adduction with DESI-MS (as seen in Figures 1 and 2) was expected as it is widely believed DESI and ESI share a similar ionization mechanism. In the same study, the optimal ratio between serine and sodium was reported to be 10:1; however, their approach was described to be less effective for sodium concentrations above 2 mM. Although not directly comparable, when considering the sodium concentration in the original sample solution before deposition, serine in DESI is capable of removing higher concentrations of salt from proteins compared to ESI.

The same study also suggested that the sodium removal effect is due to direct binding of free amino acids to sodium ions. This conclusion was derived from comparing the sodium removal effect of five different amino acids with empirical findings on sodium affinity of amino acids. Amino acid sodium affinities increase in the following order: Gly, Ala, Cys, Val, (Leu, lle), Ser, Met, Thr, (Phe, Pro), Asp, Tyr, (Glu, Lys), Trp, Asn, Gln, His [72] where the amino acids studied by Clarke. et al. [58] are highlighted in bold. Both alanine and glycine in the study by Clarke et al. were less successful than histidine, lysine, and serine in sodium removal. During DESI experiments, sodiated serine ion was observed in the spectrum. Curiously, at higher serine concentrations, when serine dimers and octamers were also observed during experiments, only protonated clusters were present.

Another, important consideration of amino acid behavior in electrospray is proton affinity, as it can affect the ionization of proteins by competing for available protons in the electrospray with protein molecules during the ionization process. Proton affinity of 20 common α-amino acids has been computationally calculated [73, 74] and compared [75]. As reported by Clarke et al., the shift in charge-state distribution when histidine or lysine was added to the electrospray is evidence of the competition for charge between these amino acid additives and protein. With DESI, the data presented in Figure S1 and Figure 2 showed a shift to the higher charge states in the bimodal distribution of protein peaks with addition of serine to 50% MeOH:H2O in desorption spray. However, for already denatured protein envelopes, such as those obtained when ammonium bicarbonate was also present in the desorbing solvent system, a slight decrease in HICS (highest intensity charge state) or HOCS (highest observed charge state) was observed (Figure S1).

For protein standards deposited out of solutions containing high NaCl concentrations, when analyzed together with 200 mM ammonium bicarbonate as desorbing spray additive (Figure S2e, f), improvements relative to the spectra shown in Figure 2 a, b are already evident, since ammonium bicarbonate also aids in sodium adduct removal, as described in our earlier paper [64]. However, addition of 1 mM serine further improved the intensities of the protein HICS when 1 mM or 10 mM NaCl was present (Figure S2a–d). Since, with this denaturing desorption solvent composition, most of the signal is concentrated in higher, less adducted charge states, it appears that serine does so through a mechanism different from sodium adduction removal.

Effect of Serine on Increasing Signal Intensity

As seen in Figure S2, another interesting effect was observed when serine was added to a solution also containing the denaturing additive ammonium bicarbonate, where its presence caused significant improvement in signal intensity in DESI-MS. Contrary to native-state conserving spray where the signal stays constant or is even slightly reduced (Figure S1), a significant improvement in signal intensity of HICS and integrated signal intensity of deconvoluted protein peak was observed for proteins under denaturing conditions of 200 mM ammonium bicarbonate. To study this effect further, 0.1% formic acid was also used as a denaturing co-additive on multiple proteins and compared to when ammonium bicarbonate and serine were present (Figure 3). Under denaturing conditions, much less adduction is usually observed, since additives such as formic acid and ammonium bicarbonate both reduce adduction and due to the denatured state of the protein less adduction is typically observed for higher charge states. From the deconvoluted spectra in Figure 3, it can be observed that the overall extent of adduction did not change much with addition of 1 mM serine, relative to solutions that already contain formic acid or ammonium bicarbonate. This suggests that signal intensity improvement is not only related to adduct removal. The improvement was dependent on protein pI and solvent system composition. For high pI protein (cytochrome c), improvement was achieved with both formic acid and ammonium bicarbonate. On the other hand, low pI proteins (myoglobin and carbonic anhydrase) only showed an improvement with serine and formic acid, but when used with ammonium bicarbonate, a reduction in signal was observed. Representative spectra are presented in Figure S3. Similarly, bovine serum albumin (BSA) spectra were also only improved with serine and formic acid in the solvent system (Figure S4). The relationship between additives and protein pI has been reported and investigated before. Pan et.al [67] demonstrated that maximum signal in positive mode is obtained when solution pH is about 3 units below the protein pI. Moreover, our previous publication also showed that proteins with high pI yield more improvement with ammonium bicarbonate in DESI compared to low pI proteins [64]. Indeed, several studies show that solvent pH and protein pI influence protein ionization in charge residue model (CRM) of electrospray ionization process [46, 67]. The isoelectric point of carbonic anhydrase and myoglobin (pI = 4.7 and 6.8 respectively) are higher than the pH of 0.1% formic acid (pH = 2.5), but lower than pH of ammonium bicarbonate (pH = 6.7), while cytochrome c pI (10.8) is higher than the pH of the ammonium bicarbonate solution. While it can be argued that the improvement in integrated protein signal intensity from addition of serine to denaturing additives could be a result of adduct removal from lower charge states, it should be kept in mind that the contribution of these charge states to the overall signal intensity of protein (i.e., integrated intensity values shown in bar graphs of Figure 3) is minor. Preliminary data showed significant improvement in signal of myoglobin when analyzed from an untreated raw meat imprint, analyzed with 100 μM serine in 80:20 ACN:H2O and 0.2% formic acid (Figure S5), suggesting the potential of serine as an additive for improving protein detection from biological tissues. This effect was further explored by additional experiments discussed in the next section.

Figure 3
figure 3

Effect of 1 mM serine with different co-additives on integrated signal intensity in DESI-MS of proteins. Error bars represent ± mean standard deviation. Deconvoluted spectra compare the protein signal when serine is added to the solvent system (red) to when there is no serine (black). Solvent systems are 50% MeOH with 200 mM ammonium bicarbonate (ABC), or 0.1% formic acid (FA). The proteins are (a) cytochrome c, (b) myoglobin, and (c) carbonic anhydrase II

Full size image

Concentration Dependency of Signal Improvement in DESI

Different concentrations between 1 μM and 10 mM of serine in 50% MeOH:H2O with 200 mM ammonium bicarbonate for cytochrome c and 0.1% formic acid for carbonic anhydrase were used as DESI solvent. The protein samples were analyzed without the addition of NaCl to highlight the improvements in signal intensity through a mechanism believed to be distinct from adduct removal. As can be observed in Figure 4a, c, with increasing amounts of serine, there was an increase in protein peak intensity up to the point that non-volatile clusters and serine adducts induced ion suppression and decreased protein signal. With DESI, the signal improvement vs. serine concentration followed a similar trend for both cytochrome c and carbonic anhydrase (Figure 4a, c). Micromolar concentrations appeared more effective at signal improvement and high concentrations (5 mM and 10 mM) significantly decreased signal intensity due to ion suppression that is presumably caused by serine clusters. In concentrations above 1 mM, protonated serine dimer [Ser2 + H]+ at m/z 211 and protonated serine octamer [Ser8 + H]+ at m/z 840 became the most abundant species, strongly dominating the spectra (Figure 5c). Another interesting observation was formation of protein–serine adducts that spread protein signal into multiple peaks, thus decreasing signal intensity for carbonic anhydrase (Figure 5c) and for cytochrome c (Figure S6c). A concentration in the range of high micromolar up to 1 mM serine improved signal effectively without inducing adducts and suppressing protein ions. Looking at the spectra in Figure 5 and Figure S6, the intensity of many charge states, including HICS, was increased by addition of different amounts of serine to the DESI spray. Interestingly, contrary to native-like conditions where addition of serine caused protein unfolding (Figure S1), in denaturing solutions, increasing concentrations of serine caused a shift to lower charge states in both DESI and ESI, as observed in Figure 5 for carbonic anhydrase and Figure S6 for cytochrome c.

Figure 4
figure 4

Integrated signal intensity for deconvoluted spectra with various concentrations of serine. The intensities for (a) cyt c and (c) CAII by DESI improve with serine addition, while the intensities of (b) cyt c and (d) CAII by ESI do not improve

Full size image
Figure 5
figure 5

Comparison between representative spectra of DESI (panels (a)–(c)) and ESI (panels (d)–(f)) with different concentrations of serine additive in 0.1% formic acid for analyzing carbonic anhydrase II, (Int: absolute intensity of protein HICS or highest intensity charge state)

Full size image

Other amino acids previously have shown to have a stabilizing effect on protein in high concentrations [30]. Figure S7 shows the analysis of hemoglobin and cytochrome c with 0.1% formic acid in 50% MeOH:H2O and equimolar concentrations of arginine (Arg) and glutamic acid (Glu) in the desorption spray of DESI. This suggests that the signal intensity improvement is not specific to serine, and further supports the role of the solution-stabilizing effect in the observed signal improvements.

DESI vs. ESI and Signal Improvement

Interestingly, this improvement in integrated signal intensity was not detected with ESI. Figure 4b, d present the results of 10 μM cytochrome c with different concentrations of serine in 200 mM ammonium bicarbonate and 10 μM carbonic anhydrase with 0.1% formic acid when directly analyzed by micro-ESI using a similar emitter as used for DESI, but pointed directly at the inlet. In ESI, unlike DESI, no significant signal improvement was observed with the addition of serine. However, similar to DESI, at higher concentrations of 5 and 10 mM serine, signal suppression was an issue especially for carbonic anhydrase (Figure 4c, d). The signal improvement was not observed in ESI under similar conditions (with similar concentrations of ammonium bicarbonate for cytochrome c and formic acid for carbonic anhydrase II).

This also supports the hypothesis that serine could play a facilitating role in desorption or “droplet pickup” mechanism of DESI, rather than through adduct removal or some other process relating to ionization. This signal improvement by serine in DESI can originate from increasing either dissolution or solubility during the droplet pickup process, an effect that plays no role in ESI. Previous studies were able to show an improvement in DESI ion signal by adding very low concentrations of surfactants to standard 50% MeOH:H2O solvent spray [76]. As mentioned previously, multiple studies have highlighted the role of amino acids in improving protein solubility and keeping proteins in solution by inhibiting aggregation [30, 32, 36]. It seems likely that there is a link between inhibition of protein aggregation and improving protein signal in DESI. Based on the data, serine improves protein solubility in the micro-localized liquid layer formed on the surface during the desorption step of DESI. This effect could be caused by reducing denaturation-induced aggregation by inhibition of nonspecific interaction of exposed hydrophobic cores of unfolded proteins, based on a mechanism previously suggested for other amino acids and their role in improving protein solubility [77].

Conclusion

Similar to previous ESI results, serine is a successful additive in significantly reducing sodium adduction from natively analyzed protein in DESI. Interestingly, serine was successful in removal of 10 mM sodium from cytochrome c; whereas in ESI, only concentrations up to 1 mM seemed to benefit from addition of serine to the ESI working solution. Other than salt removal, significant signal improvement was achieved when a suitable denaturing co-additive was combined with serine in the desorption spray. The effect was dependent on matching protein pI and solvent system pH. The combination of micromolar concentrations of serine with formic acid seems to be most effective in improving protein signal for both low and high pI proteins. In cases where an acidic solution is not desirable, ammonium bicarbonate can also improve protein signal intensity of high pI proteins. Since this enhancement in signal intensity of denatured proteins was not observed by similar ESI experiments, we propose that serine improves dissolution of dried protein spots during formation of the micro-localized liquid layer in DESI by increasing protein solubility. A possible mechanism of this effect based on previous studies through inhibition of protein aggregation during denaturing conditions seems likely. Overall, serine was shown to be an effective additive for improving detection of proteins with DESI by enhancing signal intensity and S/N.