TFA was purchased from Merck (Darmstadt, Germany). ACN was HPLC grade from Fisher Scientific (Fairlawn, NJ, USA). Bovine serum albumin (BSA, fraction V) was obtained from Bio Basic Inc (Toronto, Canada). (TPCK)- treated trypsins, cytochrome c (EC 232-700-9), myglobin were purchased from Sigma Chemical (St. Louis, MO). Sinapinic acid and -cyano-4-hydroxycinnamic 6 acid (CHCA) were purchased from Sigma (St. Louis, MO, USA). Water was purified using a Milli Q system (Millipore,Molsheim, France). All of the other chemicals were of analytical grade and used as received.

According to our previous method (Deng YH et al, 2008), the synthesis ofFe₃O₄@nSiO₂@mSiO₂ magnetic mesoporous SiO₂ microspheres were performed. At first, the Fe₃O₄@nSiO₂ microspheres were redispersed in a mixed solution containing cetyl trimethyl ammonium bromide (CTAB) (0.30 g, 0.823 mmol) deioned water (80 mL), concentrated ammonia aqueous solution (1.00 g, 28 wt %) and ethanol (60 mL). The mixed solution was homogenized for 0.5 h to form a uniform dispersion. 0.40 g of TEOS (1.90 mmol) was added dropwise to the dispersion with continuous stirring. After the reaction for 6 h, the product was collected with a magnet and washed repeatedly with ethanol and water to remove nonmagnetic by-products. Finally, the purified microspheres were redispersed in 60 mL of acetone and refluxed at 80°C for 48 h to remove the template CTAB. The extraction was repeated for three times, and the microspheres were washed with deioned water, and Fe₃O₄@nSiO₂@mSiO₂ microspheres were finally produced.

80 mg of Fe₃O₄@nSiO₂@mSiO₂ microspheres were redispersed in 20 ml methylbenzene containing 0.35 ml GLYMO with the help of ultrasonication. Subsequently, the suspension was refluxed at 80oC for 12 h. Finally, the microspheres were washed with ethanol three times, and then vacuum dried at 60°C for 24 h.

For enzyme immobilization (Scheme 1), 1.0 mg of GLYMO-functionalized Fe₃O₄@nSiO₂@mSiO₂ microspheres was incubated with TPCK-treated trypsin (0.1 mL; 5 mg/mL) for 1 h under gentle rotation. After removal of the excess trypsin solution, the trypsin immobilized GLYMO-functionalized Fe₃O₄@nSiO₂@mSiO₂ microspheres were washed with 25 mM NH₄HCO₃ (4 × 200 µ L). The final product was stored in 25 mM NH₄HCO₃ at 4oC before use.

After the trypsin immobilization, the microspheres were retained by a magnet, and the UV absorption value of the supernatant solution was measured at l = 280 nm to calculate the amount of trypsin immobilized on the GLYMOfunctionalized Fe₃O₄@nSiO₂@mSiO₂microspheres.

The procedure of tryptic digestion using trypsin-immobilized Fe₃O₄@nSiO₂@mSiO₂ magnetic microspheres is shown in Scheme S2. Three standard proteins, cytochrome c (Cyt-c), myglobin (MYG) and bovine serum albumin (BSA), in 25 mM NH₄HCO₃ buffer solution (pH 7.7), were used as model substrate to evaluate the digestion performance. The trypsin-immobilized microspheres were transferred into 40 µL protein solution (0.20 µg/µL) in a 0.6mL Eppendorf tube. A domestic microwave oven (output power 700 W) was used to conduct the microwave-assisted protein digestion process. After microwave irradiation, using an external magnet to retain the magnetic microspheres, the supernatant was deposited onto a MALDI plate directly.

For comparison, the digestions of Cyt-C, MYG and BSA were also performed by free trypsin in solution according to the conventional procedure. The standard proteins were firstly denatured in 25 mM NH₄HCO₃ buffer containing 8 M urea for 1 h at 37°C, followed by dilution with 25 mM NH₄HCO₃ (pH 7.7) buffer to the concentration of urea below 1 M. The in-solution digestion was performed by adding trypsin into the protein solution at a substrate-toenzyme ratio of 40:1, and the solution was incubated at 37°C for 12 h. After digestion, 1.0 µ L of formic acid was added into the solution to stop the reaction.

According to the references (Li Y et al, 2007; Che FY et al, 2005; Zhan, XQ et al, 2006; Liu F et al, 2006), the extraction of proteins in human pituitary was performed as the following protocol. The human pituitary tissue was cleaned with Milli-Q water to remove some possible contaminants, cut into small pieces, and homogenized in water containing 9.0 M urea, 2% 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS), 50 mM dithiothreitol (DTT), and 1.0 mM phenylmethylsulfonyl fluoride (PMSF) using a glass vessel in an ice bath. The resulting homogenate was swirled for 30 min and centrifuged for 20 min at 18000 g. The supernatant was collected, fractionated in aliquots, and stored at -80°C till further analysis. Protein concentration was measured using the Bradford assay using BSA as standard, 20 µg/µL for human pituitary tissue.

Sample solutions were deposited on the MALDI target using the dried droplet method. An amount of 1 µL of sample solution was spotted onto the MALDI plate, and then another 0.5 µL of CHCA matrix solution (5 mg/mL, 0.1% TFA in 50% ACN/H2O solution) was introduced. Positive ion MALDI-TOF-MS spectra were acquired on a 4700 Proteomics Analyzer (Applied Biosystems). The sample was excited using an Nd:YAG laser (355 nm) operated at a repetition rate of 200 Hz and acceleration voltage of 20 kV. Before identifying the samples, the MS instrument was calibrated by an internal calibration with tryptic peptides of myoglobin. The MASCOT server was used to interpret the MALDI-TOF MS data by searching the species of Mammals from sprot-horse for identification of three standard proteins with peptide fingerprint mass spectra.

The elution gradient for the RPLC column was from 5 to 90% buffer B (0.1% formic acid, 95% ACN). Eluted peptides were detected in a survey scan from 400 to 1800 amu (1 microscan) followed by 8 data-dependant MS/MS scans in a completely automated fashion on an LTQ-Orbitrap ESI mass spectrometer. According to Washburn's method (Washburn M et al, 2001), the filtering criteria was calculated through a reverse database searching and the Xcorr value vs charges was obtained as following: p< 0.01, >2.78(+3), >2.10(+2), >2.0(+1);  Cn> 0.1 and peptide length > 7 were also applied.

The Fe₃O₄@nSiO₂@mSiO₂ microspheres were synthesized according to previously reported approach (Scheme 1) which involves first coating Fe3O4 paritcles (~ 300 nm in diameter) with nonporous silica layer and then with mesoporous silica layer by using organic surfactant as the templates. Transmission electron microscopy (TEM) image shows that the obtained Fe₃O₄@nSiO₂@mSiO₂ microspheres have a mean diameter of about 500 nm and possess well-defined silica-coated magnetite core and mesoporous silica shell (Figure S1a). The nonporous silica layer is 20 nm in thickness, which can serve as protective coating for magnetite, and the mesoporous silica layer is 70 nm in thickness, which can provide the microspheres with high surface area for derivation of numerous functional groups. Notably, the mesopores (~ 2.0 nm in diameter) in the shell were found to be cylindrical and perpendicular to the microsphere surface, which provide good accessibility for reactants. Scanning electron microscophy (SEM) images show that the microspheres are very uniform both in size and shape (Figure S1b). The unique microstructure of the obtained microspheres would be very useful for many applications. According to our method (Lin S et al, 2007), GLYMO-functionalized Fe₃O₄@nSiO₂@mSiO₂ magnetic mesoporous silica microspheres were prepared by modifying their surface with a silane coupling agent GLYMO.

The most important factor of conventional enzymatic proteolysis is the enzyme-substrate ratio, and the digestion efficiency increases observably with high enzyme-substrate ratio. Nevertheless, if a high concentration of free enzyme is used, the digestion proceeds quicker but autolysis products become more abundant, and enzyme autoproteolysis would impair signal interpretation, especially for low amounts of analytes (Bonneil E et al, 2000). However, the immobilization can improve the enzyme stability, and retain its activity. The mesoporous SiO₂ with high specific surface can provide sufficient functional groups such as hydroxyl group for further modification, so more enzyme can be immobilized onto the channels of the microspheres. The elevated enzymesubstrate ratio on the surface leads to enhanced digestion efficiency. In the work, trypsin can be immobilized onto the functionalized magnetic microspheres only through a onestep reaction of its amine group with GLYMO group. As the procedures that mentioned above the amount of trypsin immobilized on the magnetic microspheres was about 188 µg/mg, which was much more than that on the previous magnetic materials including commercial magnetic materials (Lin S et al, 2008; Lin S et al, 2008; Li Y et al, 2007; Li Y et al, 2007; Li Y et al, 2007; Li Y et al, 2007). The protein to enzyme ratio is about 1:5 as compared to 40:1 in in-solution digestion procedure.

The procedure of tryptic digestion using trypsin-immobilized Fe₃O₄@nSiO₂@mSiO₂magnetic microspheres is shown in Scheme S2. As we know, in many references [Juan HF et al, 2005; Chen WY et al, 2007; Fan J et al, 2005; Qiao L et al, 2008], MYG (MW 16 700) or Cyt-C (MW 12 384) and BSA (MW 66 000) was used as model proteins to test the digestion efficiency. In our work, the three proteins were also used. For comparison, the digestions were also performed by free trypsin in solution for 12 h. Cyt-C and MYG is small molecule proteins which would be proteolysis easily, and were used to investigate the feasibility of our new method. The peptide mass mapping of Cyt-C and MYG from microwave-assisted digestion were displayed in Figure 1. Many digest fragments were observed from the MS spectra suggesting a highly proteolytic efficiency for the Fe₃O₄@nSiO₂@mSiO₂ microspheres. Notably, there are no distinct peaks with m/z > 2500, indicating virtually complete digestion. The proteolytic results were listed in Table 1 in detail. The observation corresponded to the detection of fragments containing 81 out of 104 possible amino acids of Cyt-C, 137 out of 153 possible amino acids of MYG. The sequence coverage obtained from the database is 77% for Cyt-C, 89% for MYG. Therefore, we can confirm that our new approach is feasible for fast protein digestion. Then BSA with larger molecule weight was used to do further investigation of our method. The observation corresponded to the detection of fragments containing 263 out of 607 possible amino acids of BSA, and the sequence coverage is 45% for BSA. The results indicate that BSA has been digested completely, which means that the new method is suitable for large protein molecule digestion. The identification results are comparative with those by in solution digestion that required a reaction time of 12 h (Table 1). Meanwhile, the sample volume is only 40 µ L per analysis. Moreover, no trypsin autolysis peaks were observed from mass spectra in the microwave-assisted digestion method, which demonstrated that enzyme immobilization technique can overcome the trypsin autolysis (Dogruel D et al, 1995; Nelson R, 1997; Gobom J et al, 1997; Jiang HH et al, 2000; Ekstrom S et al, 2000; Peterson D et al, 2002; Licklider L et al, 1995; Ma JF et al, 2007;Svec F, 2006). Since the magnetic microspheres are excellent microwave absorbers; and thus greatly improve the efficiency of protein digestion. We compared the digestion efficiency of our protocol with other reported methods (Chen WY et al, 2007; Fan J et al, 2005; Zhang YH et al, 2006; Guo Z et al, 2003), and the results were listed in Table S2 (see supporting information). The digestion time speared from 20 s to 15 min with different digestion method, and the sequence coverage ranges from 77% to 89% of cytochrome c and 44% to 89% of myglobin. This demonstrates that our method shows comparable digestion efficiency with other methods, and our method takes short digestion time (only 20s).

The effect of different addition amounts of trypsinimmobilizedFe₃O₄@nSiO₂@mSiO₂ magnetic microspheres on the sequence coverage of MYG is shown in Figure 3. The sequence coverage of MYG increases slowly when the addition amounts are less than 200 µg. Increasing the amounts of microspheres does not change the sequence coverage distinctly. It suggests that the optimal addition amounts of the microspheres are 200 µg. Figure 4 shows the influence of incubation time on sequence coverage of MYG obtained from MALDI-TOF MS analysis. When the incubation time increases from 5 to 20 sec, the sequence coverage accordingly increases from 78 to 89%. With further increasement of the time, no significant change on sequence coverage is observed, suggesting that 20 s is enough for efficient microwave-assisted digestion. The digestion time by the microwave-assisted digestion was much less than that (5 min) in trypsin-immobilized magnetic microspheres without microwave assistance (Li Y et al, 2007). Figure 5 distinctly. Therefore, we concluded that 200 µg microspheres with microwave incubation at 700 W for 20 sec is the optimal condition for effective protein digestion. This shows that microwave-assisted digestion by using trypsin-immobilized Fe₃O₄@nSiO₂@mSiO₂ magnetic microspheres has very high digestion efficiency.

The high digestion efficiency of the procedure may mostly count on two reasons: firstly, the mesoporous SiO₂ with high specific surface can provide sufficient functional groups such as hydroxyl group for further modification, so more enzyme can be immobilized onto the channels of the microspheres. The elevated enzyme-substrate ratio on the surface leads to enhanced digestion efficiency. Secondly, the functionalized magnetic microspheres served not only as substrate for enzyme immobilization, but also as excellent microwave absorbers, thus greatly improved the efficiency of protein digestion.

To investigate dynamic range of protein by the proposed microwave-assisted digestion, 200 mg of trypsin-immobilized magnetic microspheres were added into 40 µl of MYG solution with the concentration of 100 ng/ µl, 50 ng/ µl, 20 ng/ µl, 10 ng/ µl respectively. The microwave-assisted digestion was conducted in the same conditions described as above. The protein sequence coverage of the four protein concentration (10 to 100 ng/µl) is 73%, 79%, 79%, 83%, respectively. The results indicated that the proposed approach can be used for fast digestion of low concentration of proteins.

To test the stability, seven consecutive operations for MYG with incubation for 20 sec using the trypsin-immobilized microspheres were conducted. As shown in Figure 6, no obvious decrease is observed in the runs, suggesting that the activity of the enzyme immobilized is not destroyed apparently.

Furthermore, we studied the longevity of the enzyme-immobilized microspheres, we used 200 mg of the enzyme-immobilized magnetic microspheres for microwaveassisted digestion in the same conditions described as above. Then, the microspheres were washed with 25 mM NH₄HCO₃ (4×200µl), and resuspended in 200µl 25mM NH₄HCO₃ for further use. The same digestion procedure was conducted every other 24 h with the same microreactor. After the experiment ran four times, the protein sequence coverage didn't change obviously. It means that the activity of the protein microreactor didn't minish after 96 h. This shows that the microreactor has the longevity of more than 4 days.

In recent decades, there has been an explosion of interest in the identification and characterization of proteins, using the techniques of mass spectrometry and database searching,with the aim of establishing links to pathological conditions. A formal step in identification of proteins is protein digestion prior to mass spectrometry analysis. Our approach provides a facile and low-cost way to produce protein fragmentations, which generate sequence information and ultimately identification.

Here, to further confirm the feasibility of microreactor for the analysis of complex protein mixtures, it was applied to human pituitary extract. Without any preparation and prefractionation procedure, the entire proteome was digested for only 1 min and went through LC-ESI-MS/MS directly. Figure 7A was the total ion chromatogram acquired from the microwave-assisted protein digestion. After a database search according to the SEQUEST criteria set above (Experimental Section), 951 peptides were identified, 589 proteins were identified with p < 0.01 (Table S1). Figure 7B is the precursor mass scan at 13.22 min and Figure 7C is the corresponding MS/MS spectrum of the m/z 1095.46 in Figure 7C. Most y-ions together with b-ions produced from the precursor ion matched together and resulted in the high reliability for peptide sequence of R.NMGGPYGGGNYGPGGSGGSGGYGGR.S. These results clearly show that this novel digestion approach can be used for large-scale proteomic analysis.