lable at ScienceDirect

Analytica Chimica Acta 1062 (2019) 147e155

Contents lists avai

Analytica Chimica Acta

journal homepage: www.elsevier .com/locate/aca

Towards a high peak capacity of 130 using nanoflow hydrophilicinteraction liquid chromatography

Ya Liu, Xiaofei Wang, Zhangqian Chen, Derek Hecheng Liang, Kaiyue Sun,Shanqing Huang, Jue Zhu, Xiaohui Shi, Juxing Zeng, Qiuquan Wang, Bo Zhang*

Department of Chemistry and The MOE Key Laboratory of Spectrochemical Analysis and Instrumentation, College of Chemistry and Chemical Engineering,Xiamen University, Xiamen, 361005, China

h i g h l i g h t s

* Corresponding author.E-mail address: bozhang@xmu.edu.cn (B. Zhang).

https://doi.org/10.1016/j.aca.2019.01.0600003-2670/© 2019 Elsevier B.V. All rights reserved.

g r a p h i c a l a b s t r a c t

� The multiplex retention mechanismplays a key role in dynamics ofnanoHILIC.

� Extreme peak capacity of 130 can bereached using meter long nanoHILICcolumns.

� Low backpressure of nanoHILIC sug-gests long column length for extremeresolution.

� NanoHILIC-MS and nanoRPLC-MShave complementary protein identi-fication capability.

� The meter long nanoHILIC columnidentified 3380 HeLa proteins in asingle run.

a r t i c l e i n f o

Article history:Received 7 February 2018Received in revised form28 January 2019Accepted 30 January 2019Available online 11 February 2019

Keywords:Hydrophilic interaction liquidchromatographyNanoflow liquid chromatographyCapillary column technologyPacked columnSingle particle fritProteomics

a b s t r a c t

Hydrophilic interaction liquid chromatography, HILIC, is a relatively new HPLC mode. Compared withother HPLC modes, HILIC is a high resolution chromatographic mode with high peak capacity for sep-arations of complex mixtures. Although the separation mechanism is still not completely clear, HILIC hasbeen widely used for analysis of hydrophilic compounds which are difficult for reversed phase chro-matography to retain and separate. In this study, we fabricated and investigated nanoHILIC columns interms of separation efficiency, van Deemter curves and more importantly, we focused on long packedcapillary columns, and studied their extreme resolution for protein digests. Using meter long nanoHILICcolumns packed with 5 mm particles, we realized a high peak capacity of 130. Based on nanoLC-MS, wecompared the resolution and protein identification capabilities of nanoHILIC and nanoRPLC. The resultsindicate both nanoHILIC and nanoRPLC can provide high resolution for protein sequencing but neithermode is significantly better than the other. Among the 99 digest peptides identified, 17 were uniquelyidentified by nanoHILIC-MS and 20 were uniquely identified by nanoRPLC-MS and 62 were identified byboth methods. Although at this moment in time, nanoRPLC is the most popular microseparation tool inproteomics, the excellent complementarity of nanoHILIC and nanoRPLC suggests their combined use inachieving deep-coverage in MS-based proteomics.

© 2019 Elsevier B.V. All rights reserved.

mailto:bozhang@xmu.edu.cnhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.aca.2019.01.060&domain=pdfwww.sciencedirect.com/science/journal/00032670www.elsevier.com/locate/acahttps://doi.org/10.1016/j.aca.2019.01.060https://doi.org/10.1016/j.aca.2019.01.060https://doi.org/10.1016/j.aca.2019.01.060

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155148

1. Introduction

Modern HPLC has provided various modes for liquid phaseseparation, such as reversed phase, normal phase, size exclusionand ion exchange etc. Among these, reversed phase liquid chro-matography (RPLC) is the most widely used chromatographicmode, due to its high resolution and wide applicability [1]. A goodexample is proteomics, an area where extremely high resolution isdemanded as there are tens of thousands of proteins/peptides needto be separated in order for them to be identified [2]. Nowadays, forcharacterization of complex biosystems, the hyphenation of liquidchromatography-mass spectrometry (LC-MS) is usually the bestchoice. To fit for MS detection, the front-end LC mode needs to useaqueous/organic effluent, in other words, nonvolatile effluent isharmful to MS. For this reason, HILIC is becoming an importantsupplement to RPLC in hyphenating with MS.

Modern HILIC was introduced by Alpert [3]. The separationmechanism underneath has been attributed to hydrophilic inter-action, ion exchange, and hydrogen bond [4e7]. Although themechanism is still unclear, the effectiveness of HILIC in resolvinghydrophilic compounds, which are usually difficult for RPLC toseparate, has been well demonstrated. Recently, many chroma-tography vendors have introduced their HILIC phases, such as BEHAmide fromWaters, ZIC-HILIC fromMerck and TSK Amide-80 fromTOSOH etc.

Due to its excellent hydrophilic selectivity for peptides, HILICphases have been adopted inmodern proteomics. For example, ZIC-HILIC phase has been combined with RPLC to build two dimen-sional HPLC (2DLC) systems for high resolution analysis of humanproteome, although a long pre-fractionation step was needed andquantitative accuracy was reduced in such 2D systems [8e11].Recently, Horie et al. developed meter long monolithic HILIC col-umns and used for single-shot proteomics [10]. Due to the longcapillary columns used, a long and shallow single dimensionalnanoLC-MS was conducted, instead of off-line 2D LC. They realizedidentification of 2605 HeLa cell proteins within 8 h on a 2m longmonolithic capillary column.

According to chromatographic theory, the longer the chro-matographic column the more theoretical plates, and consequentlythe higher the separation power. Therefore, in pursuing an extremeHILIC resolution, long columns are needed. Currently, commercialHILIC columns are particle-packed and normally supplied in thelength of 10e25 cm. In single-shot proteomics, there is a demandfor long capillary HILIC columns. Particulate HILIC phases are easilyaccessible, well quality-controlled and have a wide choice of sta-tionary phase chemistry [12].

In this study, we chose a particulate HILIC material, TSK Amide-80, to fabricate capillary columns for nanoflow liquid chromatog-raphy use. We evaluated the nanoHILIC columns’ peak efficiencyand van Deemter curve through comparison with C18 particle-packed nanoRPLC columns. We also investigated their separationpower for complex mixtures and suggested extreme peak capac-ities one can achieve from these long columns. To explore the po-tential of nanoHILIC in bioanalysis, we also studied its applicabilityin MS-based proteomics.

2. Experimental

2.1. Materials and apparatus

Fused silica capillaries, 100 mm i.d., 365 mm o.d., were purchasedfrom Yongnian Reafine Chromatography (Hebei, China). Perfusivesilica beads, 110 mm in diameter, pore size ~1 mm, to be used as fritsof capillary columns, were obtained from X-tec (Bromborough, UK).The HILIC phase, TSK Amide-80, 5 mm, particulate material was

provided by TOSOH Bioscience (Tokyo, Japan). The reversed phasematerial, Ultimate XB-C18 (5 mm, 300 Å) was obtained from WelchMaterials Inc. (Shanghai, China). Acenaphthene, uracil, adenosine,cytosine and guanine of analytical grade, dithiothreitol (DTT),iodoacetamide (IAA), trifluoroacetic acid (TFA), trypsin ofsequencing grade, standard proteins cytochrome C, bovine serumalbumin (BSA), transferrin, a-casein andmyoglobinwere purchasedfrom Sigma-Aldrich (St. Louis, MO). A standard HeLa protein digestpurchased from Thermo Scientific (San Jose, CA) was used forproteomics evaluation. Acetonitrile (ACN) and acetone of HPLCgrade were provided by Merck (Darmstadt, Germany). Ultrapurewater (18.2MU) was prepared in a Milli-Q system (Millipore FilterCo., Bedford, MA) and used throughout this study. All chemicals andreagents were at least of analytical or high grade and used withoutfurther purification. An Elite P230 high pressure pump from DalianElite Analytical Instruments (Dalian, China) was used for columnpacking.

2.2. Protein digestion

Complex peptide mixtures were prepared by tryptic digestion ofstandard proteins in solution. In brief, proteins were solubilized in8M urea, 50mM NH4HCO3. Then, the sample was reduced by DTTand alkylated by IAA. Finally, trypsin was added at a protein-to-enzyme ratio of 50:1, the digestion was incubated at 37 �C overnight.

2.3. Column preparation

The capillary columns used in this study were prepared ac-cording to a single particle fritting technology as we previouslyreported [13e16]. The detailed information of the perfusive silicabeads can be found in Ref. [13]. In brief, a single perfusive particle,~110 mm in diameter, was tapped in from one end of the capillary tobe packed. This single particle was used as the outlet end frit of thecolumn. Then, from the other end of the capillary, a slurry of thepacking material was introduced and packed under a high pressureup to 6000 psi. When the column bed was packed to the length,another single particle was tapped in as the inlet end frit.

Capillary columns of 100 mm i.d., 365 mm o.d., with packedlength of 15, 25, 50 and 100 cm, were fabricated based on themethod described above. To investigate the column-to-columnreproducibility, three nanoHILIC columns of each length werefabricated and used for performance evaluation.

2.4. Nanoflow liquid chromatography

NanoLC experiments were carried out on an Ultimate 3000nanoLC system (Thermo-Dionex, Amsterdam, The Netherlands),equipped with an autosampler and a variable wavelength UVeVisdetector with a 3 nL flow cell. A 4 nL Valco nanovolume injector(VICI AG, Schenkon, Switzerland) was used for columnperformanceevaluation under isocratic condition. For large volume injectionsunder gradient elution, the autosampler with a 1 mL loop wasadopted. In performance evaluation under isocratic condition, amobile phase of ACN/10mM ammonium acetate, 88.6%:11.4% (v/v)was used for nanoHILIC columns and amobile phase of ACN/10mMammonium acetate, 4.5%:95.5% (v/v) was used for nanoRPLCcolumns.

2.5. NanoLC-MS

For analysis of the standard protein digest, nanoLC/MS/MSanalysis was performed with a Dionex Ultimate 3000 nano-LCsystem (Thermo Scientific, USA) coupled to a Q-TOF mass

Fig. 1. Comparison of chromatographic performance of nanoHILIC and nanoRPLC atsimilar retention factors. A: Isocratic separation of a mixture of acenaphthene, uracil,adenosine and guanine (in order of elution) on a nanoHILIC column (top) and amixture of thiourea, thymine, guanosine and adenosine (in order of elution) on ananoRPLC column (bottom). B: van Deemter curves of nanoHILIC and nanoRPLC usingguanine and adenosine as the standard, respectively. Experimental conditions: nano-HILIC column: TSK Amide-80, 5 mm, 150 mm � 100 mm i.d.; mobile phase: 88.6%ACNþ10 mM ammonium acetate; nanoRPLC column: Ultimate XB-C18, 5 mm,150 mm � 100 mm i.d.; mobile phase: 4.5%ACNþ10 mM ammonium acetate; UVdetection: 210 nm.

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155 149

spectrometer (Impact II, Bruker Daltonics, USA) with a Captive-Spray ion source. For nanoHILIC-MS experiments on 15 cm longcolumns: mobile phase A, H2Oþ0.1%FA, B, ACNþ0.1%FA, gradientprogram: 0e10min, 97%B; 10e70min, 97-65%B; 71e81min, 55%B;82e92min, 97%B. For nanoRPLC-MS experiments on 15 cm longcolumns: mobile phase A, H2Oþ0.1%FA, B, ACNþ0.1%FA, gradientprogram: 0e10min, 5%B; 10e70min, 5e35%B; 71e81min, 80%B;82e92min, 5%B. During peptide sequencing analysis, the massspectrometer was operated in positive ion mode and with a scanrange of m/z 50e2200. The capillary voltage was maintained at1400 V. The gas flow to the nebuliser was set at a pressure of 0.2 bar,while the drying temperature and the drying gas flow rate were150 �C and 3.0 L/min, respectively.

For analysis of the HeLa cell digest, nanoLC/MS/MS analysis wasperformed on aWaters nanoACQUITY system (Waters, Milford, MA,USA) coupled to a Q Exactive mass spectrometer (Thermo Scientific,USA). For nanoHILIC-MS experiments on 15 cm long columns:mobile phase A, (ACN:H2O ¼ 2:98)þ0.1%FA, B, (ACN:H2O ¼ 98:2)þ0.1%FA, gradient program: 0e5min, 100-90%B; 5e105min, 90-75%B; 105e125min, 75-65%B; 125e135min, 65-50%B; 135e145min,50%B. For nanoHILIC-MS experiments on 1m long columns: mobilephase A, (ACN:H2O ¼ 2:98)þ0.1%FA, B, (ACN:H2O ¼ 98:2)þ0.1%FA,gradient program: 0e20min, 100-90%B; 20e400min, 90-75%B;400e470min, 75-65%B; 470e520min, 65-50%B; 520e550min, 50%B. For nanoRPLC-MS experiments on 15 cm long columns: mobilephase A, (ACN:H2O ¼ 2:98)þ0.1%FA, B, (ACN:H2O ¼ 98:2)þ0.1%FA,gradient program: 0e5min, 2e10%B; 5e105min, 10e35%B;105e125min, 35e65%B; 125e135min, 65e95%B; 135e145min, 95%B. For each run, 1 mg HeLa protein digest was injected, and flow rateat 300 nL min�1. Electrospray of the elution was achieved with theEasy Spray ion source (Thermo Scientific, USA). Full MS spectrawere recorded at a resolution of 70,000 over a mass range of350e1800m/z. The automatic gain control target was set to 3� 106and a maximum injection time of 50ms was allowed. The top 20most intense peptide ions were chosen for fragmentation. The MS/MS spectra were recorded at a resolution of 17,500 with the auto-matic gain control target set to 1� 105 and a maximum injectiontime of 60ms. A mass window of 1.2m/z was applied to precursorselection. Normalized collisional energy for the higher-energycollision-induced dissociation fragmentation was set to 30%. Adynamic exclusion with a time window of 30 s was applied. Mol-ecules charged as 1, 7, 8 and> 8were not selected for fragmentationand the monoisotopic precursor selection was enabled.

The peptides and proteins were identified by ProteomeDiscoverer 1.4 (Thermo Scientific, USA) with Swiss-Prot completehuman proteome protein sequence database (version: 2017-09-28;number of sequences, 20,162). Parameters were specified as fol-lows: Peptides were generated from a tryptic digestion allowing forup to two missed cleavages; carbamidomethylation (C) was set asfixed modification; variable modification, oxidation (M); min andmax peptides length were set as 4 and 30, respectively. Precursormass tolerance was 10 ppm and product ions were searched at0.8 Da tolerances. Only peptides that were filtered with a confi-dence level of 95% were accepted.

3. Results and discussion

3.1. Separation efficiency

HILIC based analytical columns, i.e. with a bore size of 4.6-2mm,have been widely reported and evaluated [17e19]. In contrast,nanoHILIC columns (i.d. at 100 mm or smaller) are only supplied bya few vendors and their performance data are very limited [8e10].Therefore, we first evaluated the efficiency performance of theparticle-packed nanoHILIC columns, using a standard mixture of

acenaphthene, uracil, adenosine, cytosine and guanine. Since thestandard's retention factor may influence peak efficiency compar-ison, we intentionally adjusted the mobile phase conditions so thatthe two standards have similar retention factors (guanine fornanoHILIC with k¼ 3.4; adenosine for nanoRPLC with k¼ 3.1), asshown in Fig. 1A. Then the van Deemter curves were compared, asshown in Fig. 1B and the fitting listed in Table 1. Under the similarretention factor conditions, the nanoHILIC column's A and C termsare significantly higher than that of the nanoRPLC column. Whilefor the B term, nanoHILIC's is lower than that of nanoRPLC.

According to van Deemter equation, H ¼ A þ B/u þ Cu, A term iseddy diffusion, B term is molecular diffusion and C term is masstransfer. In general, A term reflects the fineness of the packingmaterial and the uniformity of the packed bed. To investigate this,we compared the SEMs of the two phases used for nanoHILIC andnanoRPLC columns, respectively, as shown in Fig. 2. The twonominal 5 mm particulate phases have very similar actual size:average particle size of 5.18 mm for Amide-80 vs. 5.24 mm for C18

Table 1Van Deemter coefficients from curve fitting to H ¼ A þ B/u þ Cu, (R2>0.99).

Column Mobile phase dp (mm) A(mm) B(� 10�5 cm2/s) C(ms) uopt(mm/s) Hmin(mm)Ultimate C18 4.5%ACNþ10 mM ammonium acetate 5 7.65 0.91 11.52 0.36 14.34Amide-80 HILIC 88.6%ACNþ10 mM ammonium acetate 5 13.56 0.61 32.89 0.13 22.56

Fig. 2. Comparison of SEM images and size distribution analysis of the HILIC phase TSK Amide-80, 5 mm and the RPLC phase Ultimate XB-C18, 5 mm.

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155150

particles. In terms of particle size distribution, the C18 phase has aslightly wider distribution but the majority lies in 4e6 mm range,similar to that of the Amide-80 HILIC phase.

While for the C term, we also investigated the pore size of thetwo phases through BET method (Figs. S1eS2 in Supplementarydata I). The Amide-80 phase has a pore size of 97.7 Å and the C18phase has a pore size of 311.8 Å. This means the nanoRPLC columnshould have a slightly larger C term due to its wider pore size andtherefore longer mass transfer path. However, the experimentalvan Deemter curves show the opposite result. Regarding mobilephase conditions, the solvent for nanoHILIC is 88.6% ACN, while fornanoRPLC, 4.5% ACN was adopted. The lower viscosity solvent(88.6% vs. 4.5% ACN) should also result in a better C term fornanoHILIC than for nanoRPLC. In practice, however, the nanoHILICcolumn presented a C term around 3 times of that of the nanoRPLCcolumn (Table 1), implying a significantly higher mass transferresistance.

To explain this discrepant dynamics between nanoHILIC andnanoRPLC, the retention mechanism has to be taken into account.In RPLC, the retention and separation of analytes is based on theirdifferential partition between stationary and mobile phases, whichis a relatively simple and pure process. While for HILIC, the reten-tion and resolution is due to multi-factors, including partition,adsorption, ion exchange/dipole-dipole and hydrogen bond,depending on the concrete stationary phase chemistry [3,11,20,21].It is commonly believed, the mobile phase forms a water-rich layeron the HILIC stationary phase surface vs. thewater-deficient mobilephase, creating a liquid/liquid extraction system [3,11]. McCalleyand Neue [22] concluded that about 4e13% of the pore volume ofthe silica-based HILIC phase is occupied by a water-rich layer whenthere is 75e90% acetonitrile in the mobile phase.

Guiochon et al. conducted a series of theoretical and experi-mental work on the retention mechanism of HILIC [23e27].

Depending on the water/acetonitrile concentrations, the analytesmay have adsorption onto silanol or siloxane groups and also dis-tribution into preferential water-rich and acetonitrile-rich layers[27]. In our case, apart from the silica support, the amide functionalgroup should also be taken into account, due to its hydrogen-bonding and dipole-dipole interactions with the analytes, whichfurther add to the multiplicity of the retention process. This mul-tiplicity of retention process leads to multiple mass transfer pro-cesses, both on the stationary phase surface and in the multipleliquid layers dispersed preferentially to the bulk mobile phase [12].As a result, the superimposing of multiple mass transfer processescontributed to band-broadening, leading to the rapid increase ofplate heights in nanoHILIC at the high flow rate range (C term,Fig. 1).

As Giddings pointed out [28,29], the A term is a multiple pro-cess, apart from eddy diffusion, it couples with mobile phase masstransfer. As evidenced in the SEM analysis (Fig. 2), both the nano-HILIC and nanoRPLC columns were randomly packed and quitesimilar to each other. The significant difference of A term (13.56 vs.7.65 mm) shouldmainly stem from the heterogeneous mobile phasemass transfer in HILIC in comparisonwith that of RPLC. In RPLC, themobile phase mass transfer takes place in a homogeneous phase.Whereas in HILIC, the preferential distribution of liquid layers fromthe surface of stationary phase towards the bulk mobile phasebroke the homogeneity of the mobile phase and led to a hetero-geneous mass transfer.

Recently, Desmet's group also investigated and compared dy-namics of RPLC and HILIC columns (Waters XBridge C18 vs. WatersXBridge Amide HILIC, both 2.1� 150mm, 3.5 mm) [30]. Their resultsalso highlighted higher A and C terms achieved in HILIC mode thanin RPLCmode, althoughwider bore analytical columns (2.1mm i.d.)were used in their work.

As abovementioned, the B term of the HILIC phase is slightly

Fig. 3. Peak capacity as a function of gradient time on nanoHILIC columns of differentlengths. From bottom to top: columns of 15, 25, 50 and 100 cm long, 100 mm i.d., allpacked with TSK Amide-80, 5 mm. Gradient separations of tryptic digest of cytochromeC were used for peak capacity evaluation. Gradient conditions: mobile phase A,H2Oþ0.1%TFA, B, ACNþ0.1%TFA, 97-75%B in various gradient time; injection volume:0.2 mL; flow rate: 400 nL min�1; UV detection: 210 nm. For each column length, threecolumns were prepared and used for evaluation. Each data point is the average peakcapacity obtained on the three columns of the same length.

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155 151

lower than that on the nanoRPLC column. This seemed opposite toexpectation, as the high ACN (low viscosity) mobile phase used inHILIC should result in a higher diffusivity. In fact, however, it is thelocalization effect led to the opposite result [23,31]. As discussedabove, on the HILIC stationary phase surface, there is adsorbed(high viscosity) water-rich layer, especially in the silica pores. Thisin effect formed a high viscosity liquid layer during the surfacediffusion process on the stationary phase, although the bulk mobilephase is organic-rich (low viscosity). This is in contrast to the casein classical RPLC, where the organic-rich layer on the C18 phasefacilitates a smooth surface diffusion due to its low viscosity [24].The viscosity of pure water is ~2e3 times greater than that of pureACN [32]. As a result, surface diffusion in the water-rich layer inHILIC is considerably less than that in the surface layer of ACNassociatedwith a C18 phase in RPLC. In fact, more than 30 years ago,Snyder and co-workers have demonstrated localization is thereason for small B terms measured in adsorption chromatography(normal phase) using bare silica columns [31,33].

It needs to be mentioned that Gritti and Guiochon have pub-lished an excellent review [34] on adjusting van Deemter curveaccording to experimental data obtained on RPLC and HILIC col-umns on modern HPLC instruments. They pointed out thatalthough van Deemter equation still fits very well to accurate HETPdata, the individual dispersion terms, i.e. longitudinal diffusion,eddy diffusion andmass transfer resistance in themobile phase andstationary phase, need to be adjusted to fit to the experimentaldata. They highlighted the impact of the retention factor on masstransfer in the inter-particle mobile phase, which is especially thecase in HILIC as we discussed here, although all their work wereperformed on normal scale analytical columns (4.6mm i.d.).

Apart from the in-column effect on peak efficiency, the frit effectshould also be taken into account. Since the single particle frit itselfis silica-based, in HILIC mode, hydrophilic compounds may haveretention on the frit segment, causing extra-column band broad-ening. We intentionally fabricated a 45 cm long nanoHILIC column,and tested the plate height of cytosine, at its original column length(45 cm) and after cut it to 25 and 15 cm, respectively. As shown inFig. S3 in Supplementary data I, the plate height increased from49 mmat the column length of 45 cm, to 64 mmat the column lengthof 25 cm, and finally to 73 mmat the column length of 15 cm. Thisclearly indicates the existence of frit-induced extra-column effectas it decreased as the column length increased.

3.2. Resolution and extreme peak capacity

The most attractive property of HILIC is its resolution for com-plex hydrophilic mixtures. Previously, we studied long nanoRPLCcolumns for their extreme separation power [16]. In this study, wefabricated nanoHILIC columns of difference length and investigatedtheir peak capacity. To also demonstrate the column-to-columnreproducibility, three nanoHILIC columns of each length wereused for peak capacity evaluation. According to the previous re-ports [35], peak capacities were calculated based on the average 4speak width using tryptic digest of cytochrome C as the probe. Asshown in Fig. 3, peak capacities were evaluated as a function ofcolumn lengths and gradient times, typical chromatograms areshown in Fig. 4.

It clearly shows, all the nanoHILIC columns have reached theirextreme peak capacities under the experimental conditions. For allthe four column lengths investigated, peak capacities increased asgradient times were extended, and reached a relatively constantlevel at certain stages. The trend coordinates well with gradientelution theories and our previous work on nanoRPLC columns[16,36,37]. Practically, double the column length resulted in30e50% increase of extreme peak capacities. For nanoHILIC

columns of 15, 25, 50 and 100 cm long, extreme peak capacities of60, 70, 100 and 130 have been achieved, respectively. For the meterlong nanoHILIC column, it took over 600min gradient time torealize a peak capacity of 130.

Previously, we have evaluated long nanoRPLC columns’ extremepeak capacities [16]. A meter long nanoRPLC column, packed alsowith 5 mm particles, can realize an extreme peak capacity of 800, ina 600min gradient separation of protein digests, which is ~6 timesof the extreme peak capacity of the meter long nanoHILIC column.This is mainly due to the less efficient peaks generated in nanoHILICcaused by the multiplex dynamics, as discussed in Session 3.1.Nevertheless, nanoHILIC is still a high resolution liquid chromato-graphic mode and as demonstrated in Fig. 4, can act as an excellentnanoflow separation tool for analysis of complex mixtures of bio-molecules. In terms of column-to-column reproducibility, threecolumns of each length have been evaluated and presented goodreproducibility of extreme peak capacity at RSD¼ 4.0, 2.6, 1.9 and1.7% for 15, 25, 50 and 100 cm long nanoHILIC columns, respec-tively, as shown in Fig. 3.

On the other hand, the longer the column length the higher thecolumn backpressure. We compared the column backpressurecurves of a nanoHILIC column and a nanoRPLC column of the samelength (15 cm), during a gradient separation of the Cytochrome Cdigest (Fig. S4 in Supplementary data I). For nanoHILIC, the gradientprogram is as: 0e5min, 97% ACN; 5e35min, 97-75% ACN;36e46min, 65% ACN; 47e60min, 97% ACN. For nanoRPLC, thegradient program is as: 0e5min 5% ACN; 5e35min, 5%e45% ACN;36e46min, 80% ACN; 47e60min, 5% ACN. The nanoHILIC column'sbackpressure is comprehensively lower than that of the nanoRPLCcolumn. The nanoRPLC column's minimum backpressure, 47 bar, issimilar to the maximum backpressure, 48 bar, of the nanoHILICcolumn. Meanwhile, on the nanoHILIC column, the high back-pressure of 45e48 bar only lasted for ~10min in the total run time,while the nanoRPLC column's high backpressure of ~70 bar lastedfor almost 45min of the 60min long program. Clearly, the opera-tion of nanoHILIC can greatly lower the challenge of pumpingpressures of the nanoLC instrumentation. More importantly, longcapillary columns, e.g. meter long, can be comfortably used innanoHILIC. In practice, long nanoHILIC columns are recommendedto operate under a relatively low flow rate to achieve the highestefficiency and extreme peak capacity.

Fig. 4. Long gradient separations of cytochrome C digest to reach the maximum peak capacity on nanoHILIC columns of different lengths. The three marked peaks in the bottomframe are used for average 4s peak width calculation for peak capacity evaluation. Gradient conditions: 97-75%B in 300, 400, 500 and 650min on 15, 25, 50 and 100 cm longcolumns, respectively. Other conditions as in Fig. 3.

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155152

3.3. Applicability in protein analysis and complementarity tonanoRPLC

In MS-based proteomics, proteins are sequenced in the form oftheir digested peptides. Due to the increased complexity afterdigestion, performance of MS-based proteomics depends stronglyon the resolution of the front-end nanoLC [2,38]. Among all the LCmodes, nanoRPLC is so far the most widely usedmethod for its highresolution. However, hydrophilic peptides cannot be effectivelyretained and resolved in nanoRPLC. For these species, nanoHILICcan efficiently resolve and enhance the peptide/proteinidentification.

Based on the nanoHILIC and nanoRPLC columns we fabricated,we first investigated their protein identification capability using adigest mixture of five proteins (cytochrome C, bovine serum albu-min, transferrin, a-casein and myoglobin). The total ion chro-matograms (TICs) are shown in Fig. 5 and the identified peptidesare listed in Table 2. Both nanoHILIC- and nanoRPLC-MS basedmethods have identified all the five proteins. With nanoHILIC-MS, asequence coverage of 28e63% has been realized for the five

proteins respectively; while using nanoRPLC-MS, a sequencecoverage of 24e64% has been achieved for the five proteinsrespectively. The detailed MS-based protein sequencing protocoland coverage data are provided in Supplementary data I.

As shown in Table 2, of all the 99 peptide sequences identified,62 were identified by both nanoHILIC and nanoRPLC, 20 were onlyidentified by nanoRPLC and 17 only identified by nanoHILIC. Ingeneral, GRAVY (grand average of hydropathy) value is used as ameasure of hydrophilicity, it is defined by the sum of hydropathyvalues of all amino acids divided by the sequence length, and hasbeen used for evaluation of hydrophilicity in HILIC [10]. As shown inFig. 6, the GRAVY distributions of the peptide sequences eluted inthe nanoHILIC and nanoRPLC modes, aligned with the retentiontimes, were evaluated. The GRAVY values of the peptides identifiedin nanoHILIC were negatively correlated with the retention times,and hydrophilic peptides tended to be more strongly retained thanhydrophobic peptides as expected, whereas a positive correlationwas inversely observed in the nanoRPLC mode. This clearly showsnanoHILIC is favorable for hydrophilic peptides’ retention, whilenanoRPLC is good for retention of hydrophobic peptides.

Fig. 5. Total ion chromatograms of the five protein digest performed with nanoHILIC-MS (top) and nanoRPLC-MS (bottom). NanoHILIC-MS: column: TSK Amide-80, 5 mm,150 mm � 100 mm i.d.; mobile phase A, H2Oþ0.1%FA, B, ACNþ0.1%FA, gradient pro-gram: 0e10min, 97%B; 10e70min, 97-65%B; 71e81min, 55%B; 82e92min, 97%B.NanoRPLC-MS conditions: column: Ultimate XB-C18, 5 mm, 150 mm � 100 mm i.d.;mobile phase A, H2Oþ0.1%FA, B, ACNþ0.1%FA, gradient program: 0e10min, 5%B;10e70min, 5e35%B; 71e81min, 80%B; 82e92min, 5%B. Flow rate: 400 nL min�1. Massspectrometer: Impact II Q-TOF. For other details, see Experimental Section.

Fig. 6. GRAVY distributions of peptide sequences eluted in nanoHILIC (A) andnanoRPLC (B) modes, aligned with the retention times.

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155 153

Based on the same 15 cm long nanoHILIC and nanoRPLC col-umns, we further investigated their protein identification capa-bility using a standard HeLa protein digest.

The nanoRPLC column identified 2916 proteins and the nano-HILIC column identified 1580 proteins, based on two replicates,respectively, as shown in Fig. 7, detailed protein id data were pro-vided in Supplementary data II. Combining the two sets of data(nanoRPLC and nanoHILIC), we found totally 3827 proteins wereidentified. Among all these identified proteins, only 669 wereidentified by both columns; while 2247 proteins were uniquelyidentified by nanoRPLC and 911 proteins were uniquely identifiedby nanoHILIC, respectively. In other words, only 17% of the identi-fied proteins were covered by both methods; although 59% of the

Table 2Identification of digest peptides using nanoHILIC- and nanoRPLC-MS.

Proteins Peptides identified by

RP only HILIC only

Bovine serumalbumin

DLGEEHFK、LKPDPNTLCDEFKADEK、AEFVEVTKLVTDLTK、LVTDLTK、ECCHGDLLECADDRADLAK、DAIPENLPPLTADFAEDKDVCK、DAFLGSFLYEYSR、CCAADDKEACFAVEGPK

DTHKSEIAHR、TCVADESHAGCEK、TVMENFVAFVDK、

Cytochrome C GITWGEETLMEYLENPK、GITWGEETLMEYLENPKK

TGPNLHGLFGR、EDLIAYLK

Transferrin HSTIFENLANK、DSAHGFLK、CDEWSVNSVGK、LCMGSGLNLCEPNNK、YLGEEYVK、

SCHTGLGR、WCALSHHER、SCHTAVGR、HQTVPQNTGAPNHAVVTR、CSTSSLLEACTFR

a-Casein YKVPQLEIVPNSAEER EKVNELSK、EDVPSER、EGIHAQQK、

Myoglobin LFTGHPETLEK、LFTGHPETLEKFDK、HLKTEAEMK、HPGDFGADAQGAMTK

GLSDGEWQQVLNVWGK、HGTVVLTALGGILK、YKELG

Total ID numbers 20 17

total proteins were uniquely identified by nanoRPLC indicating itsexcellent resolution power, a significant portion (24%) of the totalproteins were single-handedly identified by nanoHILIC, clearlyhighlighting the effectiveness of nanoHILIC.

Previously, Horie et al. reported the proteomics analysis per-formance of 2m long capillary monolithic columns, C18 function-alized for nanoRPLC and urea functionalized for nanoHILIC,respectively [10]. Also using HeLa cell lysate as the sample, theyidentified 2605 proteins on the monolithic nanoHILIC column and2529 proteins on the monolithic nanoRPLC column, respectively.

Both

FKDLGEEHFK、GLVLIAFSQYLQQCPFDEHVK、LVNELTEFAK、SLHTLFGDELCK、NECFLSHK、NECFLSHKDDSPDLPK、DDSPDLPK、LKPDPNTLCDEFK、YLYEIAR、YNGVFQECCQAEDK、GACLLPK、AEFVEVTK、ECCHGDLLECADDR、YICDNQDTISSK、LKECCDKPLLEK、ECCDKPLLEK、SHCIAEVEK、DAIPENLPPLTADFAEDK、RHPEYAVSVLLR、HPEYAVSVLLR、EYEATLEECCAK、DDPHACYSTVFDK、HLVDEPQNLIK、LGEYGFQNALIVR、KVPQVSTPTLVEVSR、VPQVSTPTLVEVSR、MPCTEDYLSLILNR、LCVLHEK、CCTESLVNR、RPCFSALTPDETYVPK、LFTFHADICTLPDTEK、KQTALVELLK、QTALVELLK、EACFAVEGPK、LVVSTQTALAKTGQAPGFSYTDANK、TGQAPGFSYTDANK、MIFAGIK

GK、WCAVSEHEATK、SVIPSDGPSVACVK、ASYLDCIR、DGAGDVAFVK、ADRDQYELLCLDNTR、DCHLAQVPSHTVVAR、EDLIWELLNQAQEHFGK、EGTCPEAPTDECKPVK、IECVSAETTEDCIAK、SASDLTWDNLK、FDEFFSEGCAPGSK、EGYYGYTGAFR、GDVAFVK、NPDPWAK、DYELLCLDGTR、KPVEEYANCHLAR、DDTVCLAKHQGLPQEVLNENLLR、FFVAPFPEVFGK、YLGYLEQLLR、

FQGVEADIAGHGQEVLIR、HGTVVLTALGGILK、YLEFISDAIIHVLHSK、ALELFRNDIAAK、

62

Fig. 7. Venn analysis of HeLa protein identification numbers using 15 cm long nano-HILIC and nanoRPLC, respectively.

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155154

Although both columns they used are silica monolith-based, themonolithic nanoRPLC column identified less proteins than theparticle-packed nanoRPLC column (2529 vs. 2916), while themonolithic nanoHILIC column identified more proteins than theparticle-packed nanoHILIC column (2605 vs. 1580). It needs to behighlighted that in Horie et al.'s work [10], both monolithic col-umns are 2m long, while in this work, both particle-packed col-umns are 15 cm long. In terms of stationary phase chemistry, bothparticle-packed and monolithic nanoRPLC columns used C18functional group to enable hydrophobic interaction. While fornanoHILIC separations, the particle-packed column used in thisstudy was amide-based, while the monolithic column used byHorie et al. was urea-functioned. The different HILIC functionalgroups adopted should be the main cause for the differential HILICselectivity and therefore the identified proteins.

We further investigated the proteomics performance of the 1mlong particle-packed nanoHILIC column, and totally 3380 proteinswere identified (detailed protein id data were provided in Sup-plementary data II), which is greater than the 2605 proteins iden-tified by the 2m long monolithic nanoHILIC column as Horie et al.reported [10]. Nevertheless, it is not fair to conclude the particle-packed column is better than the monolithic column. This isbecause, in terms of the MS instrumentation, a Q Exactive massspectrometer (Thermo Fisher) was used in this study, while a Tri-pleTOF 5600 mass spectrometer from AB was adopted in Horieet al.'s work. They use totally different types of mass analyzer, one istriple time-of-flight analyzer, and the other is orbitrap analyzer,and therefore different protein identification capability. The otherthing is the search engine and proteome database used for datasearching. Horie et al. used a Mascot v2.3 (Matrix Science) to searchthe 2014 version of the Swiss-Prot database, while we used ProteinDiscoverer 1.4 (Thermo Fisher) to search the 2017 version of thesame database, which is quite up-to-date.

Last but not the least is that, HILIC stationary phase is not justone type. Horie et al. [10] used urea group to function hydrophi-licity, while in this work, amide was used to function HILIC. Due tothe difference in stationary phase chemistry, it is not fair to do aquantitative comparison. On the other hand, so far, particle-packednanoHILIC columns seem to have more choices of the HILIC phase,such as silica, diol, cyano, amino, amide and even mixed-mode etc.,as summarized by Buszewski and Noga [12]. In this regard, differentHILIC phases can be adopted for HILIC selectivity tuning.

As a result, when extreme resolution and deep-coverage pro-teomics is demanded [38e40], the combined use of the two nanoLCmodes is highly recommended.

4. Conclusions

Nanoflow hydrophilic interaction liquid chromatography wasinvestigated in terms of separation efficiency, column length,extreme resolution and applicability in MS-based proteomics. Us-ing Amide-80 particle-packed capillary columns, nanoHILIC's per-formance was evaluated. In comparison with nanoRPLC, nanoHILIC

presented higher A and C terms and lower B term of the vanDeemter curve. The multiplex retention mechanism-induced masstransfer process and the localized poor diffusivity on the surface ofHILIC phase led to the less efficient dynamics of nanoHILIC. How-ever, the low backpressure of nanoHILIC columns allows the use oflong column lengths. The study has suggested extreme peak ca-pacities of 60, 70, 100 and 130 for 15, 25, 50 and 100 cm longnanoHILIC columns, respectively. Targeting at MS-based prote-omics, the results of nanoLC-MS of a HeLa protein digest indicateboth nanoHILIC and nanoRPLC can provide high resolution forprotein sequencing. In terms of sequencing performance andcoverage, only 17% of the identified proteins were covered by bothmethods; although 59% of the total proteins were uniquely iden-tified by nanoRPLC indicating its excellent resolution power, asignificant portion (24%) of the total proteins were single-handedlyidentified by nanoHILIC, clearly highlighting the complementarityof nanoHILIC and nanoRPLC. By using a meter long nanoHILICcolumn, 3380 HeLa proteins were identified in a single run, sug-gesting the long columns' excellent applicability in deep-coveragesingle shot proteomics.

Declaration of interests

The authors declare that they have no known competingfinancial interests or personal relationships that could haveappeared to influence the work reported in this paper.

Acknowledgements

This work was supported by National Natural Science Founda-tion of China (21475110, 21535007, 21521004), FundamentalResearch Funds for Central Universities of China (20720150161,20720160051, 20720172008), Xiamen Science and TechnologyProject (3502Z20173019), NFFTBS (J1310024) and PCSIRT(IRT13036). We are also grateful for TOSOH Bioscience kindlydonating Amide-80 HILIC material used in this work.

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.aca.2019.01.060.

References

[1] L.R. Snyder, J.J. Kirkland, J.W. Dolan, Introduction to Modern Liquid Chroma-tography, third ed., Wiley, Hoboken, 2010.

[2] F. Xie, R.D. Smith, Y. Shen, Advanced proteomic liquid chromatography,J. Chromatogr. A 1261 (2012) 78e90.

[3] A.J. Alpert, Hydrophilic-interaction chromatography for the separation ofpeptides, nucleic acids and other polar compounds, J. Chromatogr. A 499(1990) 177e196.

[4] S.M. Melnikov, A. H€oltzel, A. Seidel-Morgenstern, U. Tallerek, Adsorption ofwater-acetonitrile mixtures to model silica surfaces, J. Phys. Chem. C 117(2013) 6620e6631.

[5] N.P. Dinh, T. Jonsson, K. Irgum, Water uptake on polar stationary phases underconditions for hydrophilic interaction chromatography and its relation tosolute retention, J. Chromatogr. A 1320 (2013) 33e47.

[6] G. Greco, S. Grosse, T. Letzel, Study of the retention behavior in zwitterionichydrophilic interaction chromatography of isomeric hydroxy- and amino-benzoic acids, J. Chromatogr. A 1235 (2012) 60e67.

[7] R. Mountain, Molecular dynamics simulation of water-acetonitrile mixtures ina silica slit, J. Phys. Chem. C 117 (2013) 3923e3929.

[8] Z. Aturki, G. D'Orazio, A. Rocco, K. Si-Ahmed, S. Fanali, Investigation of polarstationary phases for the separation of sympathomimetic drugs with nano-liquid chromatography in hydrophilic interaction liquid chromatographymode, Anal. Chim. Acta 685 (2011) 103e110.

[9] S.D. Palma, S. Mohammed, A.J.R. Heck, ZIC-cHILIC as a fractionation methodfor sensitive and powerful shotgun proteomics, Nat. Protoc. 7 (2012)2041e2055.

[10] K. Horie, T. Kamakura, T. Ikegami, M. Wakabayashi, T. Kato, N. Tanaka,Y. Ishihama, Hydrophilic interaction chromatography using a meter-scale

https://doi.org/10.1016/j.aca.2019.01.060http://refhub.elsevier.com/S0003-2670(19)30161-8/sref1http://refhub.elsevier.com/S0003-2670(19)30161-8/sref1http://refhub.elsevier.com/S0003-2670(19)30161-8/sref2http://refhub.elsevier.com/S0003-2670(19)30161-8/sref2http://refhub.elsevier.com/S0003-2670(19)30161-8/sref2http://refhub.elsevier.com/S0003-2670(19)30161-8/sref3http://refhub.elsevier.com/S0003-2670(19)30161-8/sref3http://refhub.elsevier.com/S0003-2670(19)30161-8/sref3http://refhub.elsevier.com/S0003-2670(19)30161-8/sref3http://refhub.elsevier.com/S0003-2670(19)30161-8/sref4http://refhub.elsevier.com/S0003-2670(19)30161-8/sref4http://refhub.elsevier.com/S0003-2670(19)30161-8/sref4http://refhub.elsevier.com/S0003-2670(19)30161-8/sref4http://refhub.elsevier.com/S0003-2670(19)30161-8/sref4http://refhub.elsevier.com/S0003-2670(19)30161-8/sref5http://refhub.elsevier.com/S0003-2670(19)30161-8/sref5http://refhub.elsevier.com/S0003-2670(19)30161-8/sref5http://refhub.elsevier.com/S0003-2670(19)30161-8/sref5http://refhub.elsevier.com/S0003-2670(19)30161-8/sref6http://refhub.elsevier.com/S0003-2670(19)30161-8/sref6http://refhub.elsevier.com/S0003-2670(19)30161-8/sref6http://refhub.elsevier.com/S0003-2670(19)30161-8/sref6http://refhub.elsevier.com/S0003-2670(19)30161-8/sref7http://refhub.elsevier.com/S0003-2670(19)30161-8/sref7http://refhub.elsevier.com/S0003-2670(19)30161-8/sref7http://refhub.elsevier.com/S0003-2670(19)30161-8/sref8http://refhub.elsevier.com/S0003-2670(19)30161-8/sref8http://refhub.elsevier.com/S0003-2670(19)30161-8/sref8http://refhub.elsevier.com/S0003-2670(19)30161-8/sref8http://refhub.elsevier.com/S0003-2670(19)30161-8/sref8http://refhub.elsevier.com/S0003-2670(19)30161-8/sref9http://refhub.elsevier.com/S0003-2670(19)30161-8/sref9http://refhub.elsevier.com/S0003-2670(19)30161-8/sref9http://refhub.elsevier.com/S0003-2670(19)30161-8/sref9http://refhub.elsevier.com/S0003-2670(19)30161-8/sref10http://refhub.elsevier.com/S0003-2670(19)30161-8/sref10

Y. Liu et al. / Analytica Chimica Acta 1062 (2019) 147e155 155

monolithic silica capillary column for proteomics LC-MS, Anal. Chem. 86(2014) 3817e3824.

[11] P. Hemstrom, K. Irgum, Hydrophilic interaction chromatography, J. Separ. Sci.29 (2006) 1784e1821.

[12] B. Buszewski, S. Noga, Hydrophilic interaction liquid chromatography (HILIC)-a powerful separation technique, Anal. Bioanal. Chem. 402 (2012) 231e247.

[13] B. Zhang, Q. Liu, L.J. Yang, Q.Q. Wang, Performance of single particle frittedcapillary columns in electrochromatography, J. Chromatogr. A 1272 (2013)136e140.

[14] Q. Liu, L. Wang, Z.H. Zhou, Q.Q. Wang, L.J. Yan, B. Zhang, Toward rapidpreparation of capillary columns for electrochromatography use, Electro-phoresis 35 (2014) 836e839.

[15] Q. Liu, L.J. Yang, Q.Q. Wang, B. Zhang, Fabrication and investigation of elec-trochromatographic columns with a simplex configuration, J. Chromatogr. A1349 (2014) 90e95.

[16] J. Han, L.Q. Ye, L.J. Xu, Z.H. Zhou, F. Gao, Z.L. Xiao, Q.Q. Wang, B. Zhang, To-wards high peak capacity separations in normal pressure nanoflow liquidchromatography using meter long packed capillary columns, Anal. Chim. Acta852 (2014) 267e273.

[17] T. Yoshida, Peptide separation by hydrophilic-interaction chromatography: areview, J. Biochem. Biophys. Methods 60 (2004) 265e280.

[18] M. Gilar, P. Olivova, A.E. Daly, J.C. Gebler, Orthogonality of separation in two-dimensional liquid chromatography, Anal. Chem. 77 (2005) 6426e6434.

[19] P. Jandera, Stationary and mobile phases in hydrophilic interaction chroma-tography: a review, Anal. Chim. Acta 692 (2011) 1e25.

[20] W. Naidong, Bioanalytical liquid chromatography tandem mass spectrometrymethods on underivatized silica columns with aqueous/organic mobile pha-ses, J. Chromatogr. B 796 (2003) 209e224.

[21] E.S. Grumbach, D.M. Wagrowski-Diehl, J.R. Mazzeo, B. Alden, P.C. Iraneta,Hydrophilic interaction chromatography using silica columns for the reten-tion of polar analytes and enhanced ESI-MS sensitivity, LC-GC N. Am. 22(2004) 1010e1023.

[22] D.V. McCalley, U.D. Neue, Estimation of the extent of the water-rich layerassociated with the silica surface in hydrophilic interaction chromatography,J. Chromatogr. A 1192 (2008) 225e229.

[23] F. Gritti, G. Guiochon, Comparison between the intra-particle diffusivity in thehydrophilic interaction chromatography and reversed phase liquid chroma-tography modes. Impact on the column efficiency, J. Chromatogr. A 1297(2013) 85e95.

[24] F. Gritti, G. Guiochon, Influence of the degree of coverage of C-18-bondedstationary phases on the mass transfer mechanism and its kinetics,J. Chromatogr. A 1128 (2006) 45e60.

[25] F. Gritti, G. Guiochon, Mass transfer mechanism in hydrophilic interactionchromatography, J. Chromatogr. A 1302 (2013) 55e64.

[26] F. Gritti, Y.V. Kazakevich, G. Guiochon, Effect of the surface coverage of end-capped C-18-silica on the excess adsorption isotherms of commonly usedorganic solvents from water in reversed phase liquid chromatography,

J. Chromatogr. A 1169 (2007) 111e124.[27] F. Gritti, A.S. Pereira, P. Sandra, G. Guiochon, Comparison of the adsorption

mechanisms of pyridine in hydrophilic interaction chromatography and inreversed-phase aqueous liquid chromatography, J. Chromatogr. A 1216 (2009)8496e8504.

[28] J.C. Giddings, Dynamics of Chromatography, Marcel Dekker, New York, 1965.[29] J.C. Giddings, Unified Separation Science, John Wiley & Sons, New York, 1991.[30] H.Y. Song, E. Adams, G. Desmet, D. Cabooter, Evaluation and comparison of the

kinetic performance of ultra-high performance liquid chromatography andhigh-performance liquid chromatography columns in hydrophilic interactionand reversed-phase liquid chromatography conditions, J. Chromatogr. A 1369(2014) 83e91.

[31] J.C. Heaton, X. Wang, W.E. Barber, S.M.C. Buckenmaier, D.V. McCalley, Practicalobservations on the performance of bare silica in hydrophilic interactioncompared with C18 reversed-phase liquid chromatography, J. Chromatogr. A1328 (2014) 7e15.

[32] H. Colin, J.C. Diez-Masa, G. Guiochon, T. Czajkowska, I. Miedziak, The role ofthe temperature in reversed-phase high-performance liquid chromatographyusing pyrocarbon containing adsorbents, J. Chromatogr. A 167 (1978) 41e65.

[33] R.W. Stout, J.J. Destefano, L.R. Snyder, High-performance liquid chromato-graphic column efficiency as a function of particle composition and geometryand capacity factor, J. Chromatogr. A 282 (1983) 263e286.

[34] F. Gritti, G. Guiochon, The van Deemter equation: assumptions, limits, andadjustment to modern high performance liquid chromatography,J. Chromatogr. A 1302 (2013) 1e13.

[35] M. Iwasaki, N. Sugiyama, N. Tanaka, Y. Ishihama, Human proteome analysis byusing reversed phase monolithic silica capillary columns with enhancedsensitivity, J. Chromatogr. A 1228 (2012) 292e297.

[36] M. Gilar, A.E. Daly, M. Kele, U.D. Neue, J.C. Gebler, Implications of column peakcapacity on the separation of complex peptide mixtures in single- and two-dimensional high-performance liquid chromatography, J. Chromatogr. A 1061(2004) 183e192.

[37] H. Liu, J.W. Finch, M.J. Lavallee, R.A. Collamati, C.C. Benevides, J.C. Gebler, Ef-fects of column length, particle size, gradient length and flow rate on peakcapacity of nano-scale liquid chromatography for peptide separations,J. Chromatogr. A 1147 (2007) 30e36.

[38] T. Kocher, R. Swart, K. Mechtler, Ultra-high-pressure RPLC hyphenated to anLTQ-Orbitrap Velos reveals a linear relation between peak capacity andnumber of identified peptides, Anal. Chem. 83 (2011) 2699e2704.

[39] S.S. Thakur, T. Geiger, B. Chatterjee, P. Bandilla, F. Fr€ohlich, J. Cox, M. Mann,Deep and highly sensitive proteome coverage by LC-MS/MS without pre-fractionation, Mol. Cell. Proteomics 10 (2011). M110.003699.

[40] R. Yamana, M. Iwasaki, M. Wakabayashi, M. Nakagawa, S. Yamanaka,Y. Ishihama, Rapid and deep profiling of human induced pluripotent stem cellproteome by one-shot nanoLC-MS/MS analysis with meter-scale monolithicsilica columns, J. Proteome Res. 12 (2013) 214e221.

http://refhub.elsevier.com/S0003-2670(19)30161-8/sref10http://refhub.elsevier.com/S0003-2670(19)30161-8/sref10http://refhub.elsevier.com/S0003-2670(19)30161-8/sref10http://refhub.elsevier.com/S0003-2670(19)30161-8/sref11http://refhub.elsevier.com/S0003-2670(19)30161-8/sref11http://refhub.elsevier.com/S0003-2670(19)30161-8/sref11http://refhub.elsevier.com/S0003-2670(19)30161-8/sref12http://refhub.elsevier.com/S0003-2670(19)30161-8/sref12http://refhub.elsevier.com/S0003-2670(19)30161-8/sref12http://refhub.elsevier.com/S0003-2670(19)30161-8/sref13http://refhub.elsevier.com/S0003-2670(19)30161-8/sref13http://refhub.elsevier.com/S0003-2670(19)30161-8/sref13http://refhub.elsevier.com/S0003-2670(19)30161-8/sref13http://refhub.elsevier.com/S0003-2670(19)30161-8/sref14http://refhub.elsevier.com/S0003-2670(19)30161-8/sref14http://refhub.elsevier.com/S0003-2670(19)30161-8/sref14http://refhub.elsevier.com/S0003-2670(19)30161-8/sref14http://refhub.elsevier.com/S0003-2670(19)30161-8/sref15http://refhub.elsevier.com/S0003-2670(19)30161-8/sref15http://refhub.elsevier.com/S0003-2670(19)30161-8/sref15http://refhub.elsevier.com/S0003-2670(19)30161-8/sref15http://refhub.elsevier.com/S0003-2670(19)30161-8/sref16http://refhub.elsevier.com/S0003-2670(19)30161-8/sref16http://refhub.elsevier.com/S0003-2670(19)30161-8/sref16http://refhub.elsevier.com/S0003-2670(19)30161-8/sref16http://refhub.elsevier.com/S0003-2670(19)30161-8/sref16http://refhub.elsevier.com/S0003-2670(19)30161-8/sref17http://refhub.elsevier.com/S0003-2670(19)30161-8/sref17http://refhub.elsevier.com/S0003-2670(19)30161-8/sref17http://refhub.elsevier.com/S0003-2670(19)30161-8/sref18http://refhub.elsevier.com/S0003-2670(19)30161-8/sref18http://refhub.elsevier.com/S0003-2670(19)30161-8/sref18http://refhub.elsevier.com/S0003-2670(19)30161-8/sref19http://refhub.elsevier.com/S0003-2670(19)30161-8/sref19http://refhub.elsevier.com/S0003-2670(19)30161-8/sref19http://refhub.elsevier.com/S0003-2670(19)30161-8/sref20http://refhub.elsevier.com/S0003-2670(19)30161-8/sref20http://refhub.elsevier.com/S0003-2670(19)30161-8/sref20http://refhub.elsevier.com/S0003-2670(19)30161-8/sref20http://refhub.elsevier.com/S0003-2670(19)30161-8/sref21http://refhub.elsevier.com/S0003-2670(19)30161-8/sref21http://refhub.elsevier.com/S0003-2670(19)30161-8/sref21http://refhub.elsevier.com/S0003-2670(19)30161-8/sref21http://refhub.elsevier.com/S0003-2670(19)30161-8/sref21http://refhub.elsevier.com/S0003-2670(19)30161-8/sref22http://refhub.elsevier.com/S0003-2670(19)30161-8/sref22http://refhub.elsevier.com/S0003-2670(19)30161-8/sref22http://refhub.elsevier.com/S0003-2670(19)30161-8/sref22http://refhub.elsevier.com/S0003-2670(19)30161-8/sref23http://refhub.elsevier.com/S0003-2670(19)30161-8/sref23http://refhub.elsevier.com/S0003-2670(19)30161-8/sref23http://refhub.elsevier.com/S0003-2670(19)30161-8/sref23http://refhub.elsevier.com/S0003-2670(19)30161-8/sref23http://refhub.elsevier.com/S0003-2670(19)30161-8/sref24http://refhub.elsevier.com/S0003-2670(19)30161-8/sref24http://refhub.elsevier.com/S0003-2670(19)30161-8/sref24http://refhub.elsevier.com/S0003-2670(19)30161-8/sref24http://refhub.elsevier.com/S0003-2670(19)30161-8/sref25http://refhub.elsevier.com/S0003-2670(19)30161-8/sref25http://refhub.elsevier.com/S0003-2670(19)30161-8/sref25http://refhub.elsevier.com/S0003-2670(19)30161-8/sref26http://refhub.elsevier.com/S0003-2670(19)30161-8/sref26http://refhub.elsevier.com/S0003-2670(19)30161-8/sref26http://refhub.elsevier.com/S0003-2670(19)30161-8/sref26http://refhub.elsevier.com/S0003-2670(19)30161-8/sref26http://refhub.elsevier.com/S0003-2670(19)30161-8/sref27http://refhub.elsevier.com/S0003-2670(19)30161-8/sref27http://refhub.elsevier.com/S0003-2670(19)30161-8/sref27http://refhub.elsevier.com/S0003-2670(19)30161-8/sref27http://refhub.elsevier.com/S0003-2670(19)30161-8/sref27http://refhub.elsevier.com/S0003-2670(19)30161-8/sref28http://refhub.elsevier.com/S0003-2670(19)30161-8/sref29http://refhub.elsevier.com/S0003-2670(19)30161-8/sref29http://refhub.elsevier.com/S0003-2670(19)30161-8/sref30http://refhub.elsevier.com/S0003-2670(19)30161-8/sref30http://refhub.elsevier.com/S0003-2670(19)30161-8/sref30http://refhub.elsevier.com/S0003-2670(19)30161-8/sref30http://refhub.elsevier.com/S0003-2670(19)30161-8/sref30http://refhub.elsevier.com/S0003-2670(19)30161-8/sref30http://refhub.elsevier.com/S0003-2670(19)30161-8/sref31http://refhub.elsevier.com/S0003-2670(19)30161-8/sref31http://refhub.elsevier.com/S0003-2670(19)30161-8/sref31http://refhub.elsevier.com/S0003-2670(19)30161-8/sref31http://refhub.elsevier.com/S0003-2670(19)30161-8/sref31http://refhub.elsevier.com/S0003-2670(19)30161-8/sref32http://refhub.elsevier.com/S0003-2670(19)30161-8/sref32http://refhub.elsevier.com/S0003-2670(19)30161-8/sref32http://refhub.elsevier.com/S0003-2670(19)30161-8/sref32http://refhub.elsevier.com/S0003-2670(19)30161-8/sref33http://refhub.elsevier.com/S0003-2670(19)30161-8/sref33http://refhub.elsevier.com/S0003-2670(19)30161-8/sref33http://refhub.elsevier.com/S0003-2670(19)30161-8/sref33http://refhub.elsevier.com/S0003-2670(19)30161-8/sref34http://refhub.elsevier.com/S0003-2670(19)30161-8/sref34http://refhub.elsevier.com/S0003-2670(19)30161-8/sref34http://refhub.elsevier.com/S0003-2670(19)30161-8/sref34http://refhub.elsevier.com/S0003-2670(19)30161-8/sref35http://refhub.elsevier.com/S0003-2670(19)30161-8/sref35http://refhub.elsevier.com/S0003-2670(19)30161-8/sref35http://refhub.elsevier.com/S0003-2670(19)30161-8/sref35http://refhub.elsevier.com/S0003-2670(19)30161-8/sref36http://refhub.elsevier.com/S0003-2670(19)30161-8/sref36http://refhub.elsevier.com/S0003-2670(19)30161-8/sref36http://refhub.elsevier.com/S0003-2670(19)30161-8/sref36http://refhub.elsevier.com/S0003-2670(19)30161-8/sref36http://refhub.elsevier.com/S0003-2670(19)30161-8/sref37http://refhub.elsevier.com/S0003-2670(19)30161-8/sref37http://refhub.elsevier.com/S0003-2670(19)30161-8/sref37http://refhub.elsevier.com/S0003-2670(19)30161-8/sref37http://refhub.elsevier.com/S0003-2670(19)30161-8/sref37http://refhub.elsevier.com/S0003-2670(19)30161-8/sref38http://refhub.elsevier.com/S0003-2670(19)30161-8/sref38http://refhub.elsevier.com/S0003-2670(19)30161-8/sref38http://refhub.elsevier.com/S0003-2670(19)30161-8/sref38http://refhub.elsevier.com/S0003-2670(19)30161-8/sref39http://refhub.elsevier.com/S0003-2670(19)30161-8/sref39http://refhub.elsevier.com/S0003-2670(19)30161-8/sref39http://refhub.elsevier.com/S0003-2670(19)30161-8/sref39http://refhub.elsevier.com/S0003-2670(19)30161-8/sref40http://refhub.elsevier.com/S0003-2670(19)30161-8/sref40http://refhub.elsevier.com/S0003-2670(19)30161-8/sref40http://refhub.elsevier.com/S0003-2670(19)30161-8/sref40http://refhub.elsevier.com/S0003-2670(19)30161-8/sref40

Towards a high peak capacity of 130 using nanoflow hydrophilic interaction liquid chromatography1. Introduction2. Experimental2.1. Materials and apparatus2.2. Protein digestion2.3. Column preparation2.4. Nanoflow liquid chromatography2.5. NanoLC-MS

3. Results and discussion3.1. Separation efficiency3.2. Resolution and extreme peak capacity3.3. Applicability in protein analysis and complementarity to nanoRPLC

4. ConclusionsDeclaration of interestsAcknowledgementsAppendix A. Supplementary dataReferences