Naturally Occurring Epsilon Gamma Glutamyl Lysine Isopeptide Crosslinks in Human Neuroblastoma SH-SY5Y Cells.

Zero-length isopeptide crosslinks between the side chains of glutamine and lysine are the product of transglutaminase activity. It is generally accepted that transglutaminase activity is dormant under physiological conditions because the calcium concentration inside cells is too low to activate transglutaminase to an open conformation with access to the catalytic triad. Traditional assays for transglutaminase activity measure incorporation of biotin pentylamine or of radiolabeled putrescine in the presence of added calcium. In this report, we identified naturally occurring isopeptide crosslinked proteins using the following steps: immunopurification of tryptic peptides by binding to anti-isopeptide antibody 81D1C2, separation of immunopurified peptides by liquid chromatography-tandem mass spectrometry, Protein Prospector database searches of mass spectrometry data for isopeptide crosslinked peptides, and manual evaluation of candidate crosslinked peptide pairs. The most labor intense step was manual evaluation. We developed criteria for accepting and rejecting candidate crosslinked peptides and showed examples of MS/MS spectra that confirm or invalidate a possible crosslink. The SH-SY5Y cells that we examined for crosslinked proteins had not been exposed to calcium and had been lysed in the presence of ethylenediaminetetraacetic acid. This precaution allows us to claim that the crosslinks we found inside the cells occurred naturally under physiological conditions. The quantity of crosslinks was very low, and the crosslinked proteins were mostly low abundance proteins. In conclusion, intracellular transglutaminase crosslinking/transamidase activity is very low but detectable. The low level of intracellular crosslinked proteins is consistent with tight regulation of transglutaminase activity.

Introduction

Zero-length isopeptide crosslinks between the side chains of lysine and glutamine are the product of transglutaminase activity.[1−3] Transglutaminase (TG2) activity is essential for neurite outgrowth in human neuroblastoma cells.[4] Knockdown of transglutaminase 2 in primary cortical neurons decreased the viability of neurons.[5] TG2-deficient mice have a normal phenotype,[6,7] though they are susceptible to apoptotic stress. Isopeptide bonds stabilize blood clots, skin, hair, and the protective mucus network lining the gastrointestinal tract.[8]

Excessive or inappropriate isopeptide bond formation promotes polymerization of proteins to high molecular weight aggregates implicated in Alzheimer’s disease (AD), Huntington’s chorea, Parkinson’s disease, and Lewy body disease.[9−11] Transglutaminase-mediated crosslinking of proteins in the eye results in cataracts.[12] Deamidation of gluten peptides by transglutaminase leads to celiac disease.[13,14] Cancer stem cells that overexpress transglutaminase are associated with metastatic spread and drug resistance.[15,16] Transglutaminase has been proposed as a target for treatment of mesothelioma, renal cell carcinoma, and gastric cancers.[17−19]

Intracellular transglutaminase adopts a closed conformation that blocks access to the catalytic triad Cys277, His335, and Asp358.[20] Therefore, intracellular TG2 is generally assumed to have minimal or no crosslinking activity. (See the Discussion section.) Our finding of KQ crosslinked peptides inside cells supports the conclusion that intracellular TG2 crosslinking activity is not zero.

Established methods to identify protein targets of transglutaminase crosslinking activity include labeling with fluorescent dansyl or biotinylated probes[21] followed by mass spectrometry analysis[22,23] or by incorporation of radiolabeled putrescine followed by immunoblotting.[24,25] Our mass spectrometry method builds on the work of Nemes et al. who identified crosslinked proteins in the brains of AD and Lewy body disease patients by immunopurifying peptides containing an isopeptide bond, followed by mass spectrometry.[9,10]

In this report, we used mass spectrometry, Protein Prospector database searches, and manual evaluation to identify naturally occurring crosslinked peptides in human neuroblastoma SH-SY5Y cells. Manual evaluation was a critical step for identifying crosslinked peptides. Our goal in this report is to establish criteria for accepting and rejecting candidate crosslinked peptide pairs in a complex protein mixture.

Materials

Human neuroblastoma SH-SY5Y cells (ATCC CRL-2266) were grown in DMEM/F12 GlutaMAX (Gibco 10565-018) supplemented with fetal bovine serum (Life Tech 16000044) and penicillin & streptomycin (Gibco 15140-122). Cells were differentiated in a serum-free medium containing trans-retinoic acid (Sigma-Aldrich 554720). Cells were lysed in Pierce IP lysis buffer (Thermo Scientific 87787) (25 mM TrisCl pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM ethylenediaminetetraacetic acid (EDTA), 5% glycerol) supplemented with Halt protease inhibitor cocktail (100×) (Thermo Scientific 87786). Protein concentration was determined with a bicinchoninic acid protein assay kit (Thermo Scientific 23228). Proteins in the cell lysate were digested with trypsin (Promega V5113). Tryptic peptides were complexed with mouse anti-isopeptide monoclonal 81D1C2 (LS Bio LS-C153331 reconstituted with water to 1 mg/mL). Protein G agarose (Protein Mods LLC PGGH) beads captured the antibody–peptide complexes. Beads were washed with RIPA buffer (Pierce 89900) (25 mM Tris–HCl pH 7.6, 1% NP-40, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 140 mM NaCl).

Cell Culture

SH-SY5Y cells (ATCC CRL-2266) in T75 flasks were grown in DMEM/F12 GlutaMAX supplemented with 10% fetal bovine serum, penicillin, and streptomycin, in a humidified 5% carbon dioxide incubator at 37 °C. After 5 days, when cells were 70–80% confluent, cells were washed with phosphate buffered saline (PBS) and harvested.

Another set of 70–80% confluent SH-SY5Y cells was incubated in DMEM/F12 GlutaMAX (no serum) supplemented with 10 μM trans-retinoic acid and 10 μM dichlorvos. After 2 days, cells were harvested from seven T75 flasks. Cytoplasmic transglutaminase protein is upregulated in the serum-free medium supplemented with retinoic acid.[4]

Cell Lysis and Protein Concentration

Cells were washed with PBS and lysed with 100 to 500 μL of IP lysis buffer (25 mM TrisCl pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM EDTA, 5% glycerol, Pierce 87787) containing Halt protease inhibitor cocktail. Cell debris was removed by centrifugation at 14,000 × g for 20 min at 4 °C. The protein concentration in the supernatant was 13.2 and 8.59 mg/mL as determined by the bicinchoninic acid protein assay.

Six cell lysate samples were digested with trypsin, immunopurified with anti-isopeptide antibody 81D1C2, and subjected to liquid chromatography tandem mass spectrometry.

Trypsin Digestion

The cell lysate supernatant containing 200 μg of protein was diluted with 20 mM ammonium bicarbonate pH 8 to 200 μL. Proteins were denatured in a boiling water bath for 3 min. The denatured proteins were digested with 4 μg of trypsin (8 μL) at 37 °C for 16 h. Trypsin was inactivated by heating the digest in a boiling water bath for 3 min.

Immunopurification of Tryptic Peptides

The heat-treated digest was incubated with 8 μg (8 μL) of anti-isopeptide monoclonal 81D1C2 at room temperature for 8 h. The antibody–peptide complexes were immobilized by adding 0.1 mL of a 1:1 suspension of Protein G agarose beads in PBS. The sample was rotated overnight at room temperature.

The beads and liquid were transferred to a 0.45 μm Durapore spin filter (Millipore UFC30HV00). Use of the spin filter maximized recovery because beads were not lost in the wash steps. Beads were washed 5 times with 0.4 mL of RIPA buffer (25 mM Tris–HCl pH 7.6, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, 140 mM NaCl) followed by 5 washes with water. Salts and detergents were washed off with water. The flow through in each wash step was discarded.

The basket of washed beads was transferred to a new microfuge tube. Bound peptides were released from the washed beads by incubating the basket of beads with 0.1 mL of 50% acetonitrile and 1% formic acid for 0.5 to 1 h at room temperature. The released peptides were collected in the flow through by brief centrifugation. The extraction step was repeated twice. The combined flow through was dried by vacuum centrifugation.

Sample Preparation for Mass Spectrometry

The dry sample was dissolved in 20 μL of water. The sample was centrifuged for 30 min at 14,000 × g and 4 °C. The top ten microliters were transferred to an autosampler vial.

Liquid Chromatography–Tandem Mass Spectrometry

Peptide separation was performed with a Thermo RSLC Ultimate 3000 ultrahigh pressure liquid chromatography system (Thermo Scientific) at 36 °C. Solvent A was 0.1% formic acid in water, and solvent B was 0.1% formic acid in 80% acetonitrile. Peptides were loaded onto an Acclaim PepMap 100 C18 trap column (75 μm × 2 cm; Thermo Scientific cat# 165535) at a flow rate of 4 μL/min and washed with 98% solvent A/2% solvent B for 10 min. Then, they were transferred to a Thermo Easy-Spray PepMap RSLC C18 column (75 μm × 50 cm with 2 μm particles, Thermo Scientific cat# ES803) and separated at a flow rate of 300 nL/min using a gradient of 9 to 50% solvent B in 30 min, 50 to 99% solvent B in 40 min, hold at 99% solvent B for 10 min, 99 to 9% solvent B in 4 min, and hold at 9% solvent B for 16 min.

Eluted peptides were sprayed directly into a Thermo Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Scientific). Data were collected using data dependent acquisition. A survey full scan MS (from 350 to 1800 m/z) was acquired in the Orbitrap with a resolution of 120,000. The AGC target (Automatic Gain Control for setting the ion population in the Orbitrap before collecting the MS) was set at 4 × 105, and the ion filling time was set at 50 ms. The 25 most intense ions with a charge state of 2–6 were isolated in a 3 s cycle and fragmented using high-energy collision-induced dissociation with 35% normalized collision energy. Fragment ions were detected in the Orbitrap with a mass resolution of 30,000 at 200 m/z. The AGC target for MS/MS was set at 5 × 104, and dynamic exclusion was set at 30 s with a 10 ppm mass window. Data were reported in *.raw format. The *.raw data files were converted to *.mgf files using MSConvert (ProteoWizard Tools from SourceForge).

Protein Prospector Search for Crosslinked Peptides

The search parameters on the Batch-Tag Web page in Protein Prospector (prospector.ucsf.edu/prospector/mshome.htm) were as follows. (1) Database: SwissProt.2020.09.02. (2) Taxonomy: Homo sapiens. (3) Precursor charge range: 2, 3, 4, and 5. (4) Parent Tol 20 ppm, Frag Tol 30 ppm. (5) Digest: trypsin. (6) Max missed cleavages: 3. (7) Constant mods: none selected because the proteins were not reduced and alkylated. (8) Expectation calc method: none. (9) Variable mods: oxidation (M). (10) User-defined variable modifications were left blank. (11) Mass modifications left at default setting −18 to 3883 Da. (12) Checkmark in the boxes for K and Q. (13) Checkmark in the box Uncleaved avoids reporting peptides cleaved at a modified lysine. (14) Crosslinking: user defined link. (15) User defined link parameters: link AAs K,Protein N-term > Q. (16) Bridge Elem Comp N-1 H-3. (17) Instrument: ESI-Q-high-res.

Selection of High Scoring Crosslinked Peptides

A typical Search Compare Search Results page in Protein Prospector reported 13,645 data spectra for 1513 proteins and 8586 peptides. The report included the sequence of each crosslinked peptide pair, their accession numbers, protein name, the crosslinked amino acid numbers, M + H, m/z, z, ppm, retention time, and Score and Score difference. A link to the MS/MS spectrum showed the % match between ions assigned to the crosslinked peptide and total ions in the spectrum. A Table of Peak Matches linked to each MS/MS spectrum listed the mass of each peak, the ion type (y, b, a), blank for peaks that did not fit the crosslinked peptide, the charge of each ion, and mass error. We used the Table of Peak Matches to calculate the number of crosslink-specific ions for each candidate crosslinked peptide. A crosslink-specific ion contains a fragment from one peptide plus the entire second peptide. We selected a preliminary set of crosslinked peptides for further evaluation based on a Score + Score difference greater than 30, matched intensity greater than 40%, and a minimum of two crosslink specific ions in a series that defined an amino acid.

Manual Evaluation

Manual evaluation of each MS/MS spectrum started with assigning the charge state of each peak in the MS/MS spectrum retrieved from Thermo Scientific Xcalibur/Qual Browser. Protein Prospector did not give the charge state of peaks that did not fit the candidate crosslinked peptide. Therefore, the charge state was obtained from Qual Browser files. The monoisotopic mass of each peak was determined. The interval between peaks, grouped by the charge state, was calculated and correlated with the mass of an amino acid.

Results

Transglutaminase makes a covalent bond between the side chains of glutamine and lysine to create an epsilon (gamma-glutamyl) lysine isopeptide bond with release of ammonia. See Figure .

Figure 1

Transglutaminase (TG) catalyzes the formation of a covalent bond between the side chains of glutamine and lysine in proteins. The isopeptide bond is resistant to trypsin.

The MS/MS spectrum in Figure is a typical fragmentation pattern for the crosslinked peptides listed in Table . The peptides crosslinked in Figure were identified in SH-SY5Y cells that had been treated with retinoic acid and lysed in the presence of EDTA without added calcium chloride. The absence of calcium chloride is emphasized because transglutaminase crosslinking activity is expected to occur only in the presence of calcium chloride concentrations high enough to change the TG2 protein conformation from closed to open.[20] The intracellular calcium chloride concentration is estimated to be too low at 100 nM to activate transglutaminase to the open conformation that has crosslinking activity.[26] Despite the low intracellular calcium concentration, TG2-catalyzed crosslinking activity is detectable by mass spectrometry of immunopurified crosslinked peptides.

Figure 2

MS/MS spectrum showing lysine 109 in the green peptide (Q8WUU5. GATA zinc finger domain-containing protein 1. GATD1) crosslinked to glutamine 230 in the blue peptide (P35711. Transcription factor, SOX5) via a zero-length isopeptide bond. The parent ion MH+4 is at 586.32 m/z. Crosslink specific ions are green y7+3, y8+3, and y9+3, blue y7+3, y8+3, and y9+3, and blue y8+4 and y9+4. The structures of green crosslink specific ions y7+3, y8+3, and y9+3 give a visual definition of the term “crosslink specific ion” as ions that contain residues from both peptides. The structures of blue y8+4 and y9+4 ions demonstrate the meaning of the term “ladder ions” as ions that have the same charge as the parent ion, though they have lost residues from the N-terminal of one peptide.

Table 1

Naturally Occurring KQ Isopeptide Crosslinks in SH-SY5Y Cells

#	Z	Score + Score diff	% match	crosslinked peptides with linked amino acids numbered	UniProt	short name
1	+3	43.8 + 6.5	67.2	GKVRVEK2031EKa,b	Q01484	ANK2
				VGKQLASQ678K	Q8WYQ5	DGCR8
2	+3	40.2 + 5.1	54.3	EKK119LQGKGPGGPPPK.a	Q9NS71	GKN1
				ALRPGREPRQSEPPAQ38R	Q99933	BAG1
3	+3	30.9 + 7.9	60.5	SAEK67SPFPEEK	Q9NP80	PLPL8
				TYGQ3409TPR.	Q99698	LYST
4	+3	30.7 + 1.9	43.5	TTVKVPGK738RPPRa	Q99490	AGAP2
				KGVQ1354HIFR	Q15772	SPEG
5	+4	55.5 + 9.0	74.8	KK241KQEEEQEKa	Q92541	RTF1
				IQ511NLEQKLSGDSR	Q9Y2G4	ANKR6
6	+4	52.2 + 4.7	79.5	KVVRNLNYQK298Ka	Q9HCC0	MCCB
				VKVGSLQ710TTAK	Q96NW7	LRRC7
7	+4	43.7 + 15.1	50.0	YK338AYLYLDEAHSIGALGPTGR	O15270	SPTC2
				LVEYLQ721AMoxR	Q96RT8	GCP5
8	+4	42.1 + 10.9	73.5	K48SVAKLQDERb	O14818	PSA7
				AALERVLRQ1353K	A6NJZ7	RIM3C
9	+4	38.2 + 12.1	59.5	REHVAK349MoxKa	Q16181	SEPT7
				AGHTIPRIFQAVVQ87R	Q6PCB7	S27A1
10	+4	38.0 + 1.4	57.1	MVVSAIVDTLK689TAFPRa	Q92543	SNX19
				YNVIRIQ1151K	O95271	TNKS1
11	+4	37.9 + 10.0	61.3	SAPAAEKK109VSTKa	Q8WUU5	GATD1
				LAASQ230IEKQR	P35711	SOX5
12	+4	36.0 + 5.5	58.5	RGNDGRVSLIK743QRa	Q75QN2	INT8
				KIRITTNDGRQ786SMVTLK	Q96JH7	VCIP1
13	+4	32.5 + 9.2	70.3	REK117KAAELAK	P17568	NDUB7
				KELSRLAGQ148IRR	Q6PF06	TM10B
14	+5	64.2 + 5.5	70.8	QSK160PVTTPEEIAQVATISANGDK	P10809	CH60
				ELRNFSSLRAILSALQ351SNPIYR	Q3MIN7	RGL3
15	+5	56.0 + 17.0	61.6	SLNSLVLYGNK359ITELPK	O94813	SLIT2
				AIHKETGQ54IVAIK	Q13043	STK4
16	+5	48.7 + 12.2	75.2	K48SVAKLQDERb	O14818	PSA7
				AALERVLRQ1353K	A6NJZ7	RIM3C
17	+5	46.6 + 3.0	81.9	K48SVAKLQDERTVR	O14818	PSA7
				AALERVLRQ1353Kb	A6NJZ7	RIM3C
18	+5	45.9 + 14.0	81.1	KGQEVQK676GPAVEIAK	Q9UI46	DNAI1
				KEIKDILIQ114YDRTLLVADPR	P62249	RS16
19	+5	42.5 + 6.7	67.3	LLEAEK985RIKEK	Q9UIF8	BAZ2B
				LIQEEKENTEQ655R	Q13136	LIPA1
20	+5	39.9 + 6.7	48.0	DILKECANFIK119VLK	Q14563	SEM3A
				VLESVIEQ509EQKR	Q08999	RBL2
21	+5	38.4 + 7.6	48.9	ISAFLPARQLWK2485WSGNPTQR	Q9P281	BAHC1
				ELAAEDEQVFLMKQQ367SLLAK	O43237	DC1L2
22	+5	36.1 + 10.0	47.6	DK60LEHSQQKa	Q8N7P1	PLD5
				VLIGETFFQ229PSPWR	Q6UXE8	BTNL3
23	+5	35.9 + 4.1	53.7	TNGLQPAKQQNSLMoxK350CEK	Q8IU60	DCP2
				IQGSAGEIATSQ1661ERLKALLER	Q9P2E3	ZNFX1
24	+5	35.8 + 3.5	55.5	ELFEKQK402FK	Q93088	BHMT1
				LQEIYQELTQ118LK	Q9H2X3	CLC4M
25	+5	35.7 + 12.1	51.7	ELSMAKEVIAK453ELSKb	O96006	ZBED1
				SNAYQ448DLLLAK	P49758	RGS6
26	+5	35.6 + 11.5	53.6	KKSSSEAKPTSLGLAGGHK293ETR	Q96RK0	CIC
				EPISVKEQ585HK	O15226	NKRF
27	+5	35.2 + 9.3	61.7	SWSLIK140NTCPPK	P79522	PRR3
				LLQKKAYQ975PDLVK	Q6IQ55	TTBK2
28	+5	35.0 + 7.4	52.1	EIDK334IVGQLMoxDGLKa	Q13822	ENPP2
				Q106LEEEKR	Q9BQE4	SELS
29	+5	32.1 + 4.7	49.4	SLLLGKKHGLK242MLERb	Q5SVQ8	ZBT41
				GQQIGKVVQ72VYRK	P61254	RL26
30	+5	30.9 + 4.4	59.7	ASLCK535LSPCTVTR.b	Q9Y2X0	MED16
				EGSVMLQ426VDVDTVK	Q13228	SBP1
31	+5	28.4 + 2.4	45.2	SLMAIGK337RLATLPTK	Q9UBF8	PI4KB
				IAMoxAIIRIRSLQ231GR.	Q8NG92	O13H1

Ladder ions. An example of ladder ions is illustrated in Figure .

Mixed fragmentation. An example of mixed fragmentation is illustrated in Figure .

Figure 3

MS/MS spectrum of a mixed fragmentation. Lysine 48 in the green peptide (O14818 proteasome subunit alpha type-7. PSA7) is crosslinked to glutamine 1353 in the blue peptide (A6NJZ7 RIMS-binding protein 3C. RIM3C) via a zero-length isopeptide bond. The parent ion at 585.59 m/z in charge state +4 is not visible. Panel A shows the entire MS/MS mass range. The green peptide is supported by green y1 to y9 ions for the sequence SVAKLQDER from the C-terminus of the green peptide and by the ion at 1283.72 m/z showing the green peptide KSVAKLQDER linked to Q from the blue peptide. Panel B is an expansion of a region of panel A. The blue peptide is supported by blue ions a5+2, a6+2, y3+3, y4+3, y2+2, y3+2, y4+2, a5, a6, a7, and a8 and by blue b5, b6, b7, and b8 ions for peptide RVLR. The blue peptide is also supported by the six mixed fragmentation structures in panel B. Crosslink specific ions are the blue y3+3 and y4+3, blue y2+2, y3+2 and y4+2, and the six structures in panel B. Residues colored red in panel B are missing from the adjacent structure. For example, the red L in the structure at 850.49 m/z is missing from the structure at 793.95 m/z. Black lines in the Protein Prospector spectrum are for masses the software did not assign to the crosslinked peptide. We assigned three black lines to the structures at 520.83, 587.33, and 651.38 m/z.

The MS/MS spectrum in Figure fulfills our criteria for a crosslinked peptide pair. The required features are (1) the presence of a minimum of two crosslink specific ions that define an amino acid. The definition of a crosslink specific ion is an ion that contains residues from both peptides. In Figure , the crosslink specific ions are green y7+3, y8+3, and y9+3, blue y7+3, y8+3, and y9+3, and blue y8+4 and y9+4. Structures of the green y7+3, y8+3, and y9+3 ions are depicted in Figure . The green y7+3, y8+3, and y9+3 ions contain an intact blue peptide and short pieces of the green peptide. (2) A second requirement is the presence of supporting ions for both peptides. The green b3, b4, and b5 for SAPAA and green y7+3, y8+3, and y9+3 for AA support the green peptide. The blue y7+3, y8+3, and y9+3 and blue y8+4 and y9+4 ions support the blue peptide.

Fragmentation of the parent ion by loss of residues from the N-terminus of one peptide, while retaining the charge state of the parent ion, is illustrated in Figure for blue y8+4 and blue y9+4 ions. The blue y8+4 and y9+4 ions have an intact green peptide and short pieces of the blue peptide. This type of fragmentation, called ladder sequencing, has been reported by others. We found ladder sequences in 11 crosslinked peptides in Table .

Figure is another example of a crosslinked peptide pair identified in SH-SY5Y cells. Figure panel A highlights a 9 amino acid y-ion sequence from the C-terminus of the green peptide and a crosslink specific ion at 1283.72 m/z, both of which support the crosslinked peptide interpretation. Additional masses supporting the presence of the blue peptide can be seen at a5+2, a6+2, y3+3, y4+3, y2+2, y3+2, y4+2, a5, a6, a7, a8, and b5, b6, b7, and b8.

The MS/MS spectrum in Figure panel B shows an unusual fragmentation. The mass spectrometer fragmented the crosslinked peptide pair causing the loss of residues from both peptides. Figure panel B shows structures for each mass in the series, with the lost residue in red font. Peaks in the mass spectrum associated with each fragment are identified with an arrow. Most of the y- and b-ion designations in the mass spectrum do not apply to this type of fragmentation. The starting fragment for the series is the y4+2 crosslink specific blue ion at 850.49. It lost L from the blue peptide to yield the 793.95 y3+2 blue ion. The 793.95 ion lost R from the C-terminus of the green peptide to yield 715.90, which lost E from the green peptide to yield the 651.38 ion. The 651.38 ion lost K from the C-terminus of the blue peptide to yield the 587.33 ion. The 587.33 ion lost D from the green peptide to yield the 520.83 ion. The structures in panel B were identified without the aid of software. We have named this type of fragmentation from both peptides “mixed” fragmentation. Having recognized that mixed fragmentation occurs, we found mixed fragmentation in six entries in Table .

Our mixed fragmentation interpretation supports the crosslink in Figure . Had we relied on the Protein Prospector interpretation, we would have assigned blue y2+2, y3+2, and y4+2 as crosslink specific ions encoding RL. However, we would have reported RL appended to three residues in peptide KQERL (523.26, 587.33, 651.38, 715.90, and 793.95). Since KQERL does not fit the candidate crosslink, we would have disqualified the blue crosslink specific ions. With no blue crosslink specific ions to support the blue peptide, we would have discarded this crosslinked peptide pair. The identification of this crosslink relied on manual evaluation. Search programs do not report mixed fragmentation from both peptides in a crosslink. Protein Prospector leaves one peptide intact and reports fragment ions from the second peptide.

Another unusual fragmentation spectrum is shown in Figure . A series of +3 ions from 774.12 to 908.19 echoes the series of +4 ions from 670.90 to 771.45. Both series yield the sequence IAMoxA. The +3 ion at 774.12 has lost five residues from the N-terminal of the green peptide, two residues from the C-terminal of the green peptide, and one residue from the C-terminal of the blue peptide. The 774.12 m/z ion has also lost ammonia. The structure of the 774.12 m/z ion is shown. Masses associated with the +3 series are marked by arrows in the MS/MS spectrum. The unusual +3 ion series was identified by manual evaluation. It was not identified by Protein Prospector, whereas the comparable +4 series was identified by Protein Prospector.

Figure 4

MS/MS spectrum showing lysine 337 in the blue peptide (Q9UBF8. Phosphatidylinositol 4-kinase beta. PI4KB) crosslinked to glutamine 231 in the green peptide (Q8NG92 olfactory receptor 13H1. O13H1) via a zero-length isopeptide bond. The full MS/MS range has been expanded to allow better presentation of the masses of interest. The parent ion MH+5 is at 639.98 m/z (not shown). The green peptide is supported by green crosslink specific ions y9+4, y10+4, y11+4, y12+4, and y13+4 and by the +3 series 774.12 to 908.19 m/z. Both series define the sequence IAMoxA. The blue peptide is supported by blue y10+3 and y11+3 encoding the crosslink specific amino acid isoleucine, by blue b2 and b3 ions (not shown), and by the series 774.12 to 908.19 m/z.

For comparison, the structure of the green y9+4 ion is shown so that the difference between a standard y9+4 ion and the comparable, unusual +3 fragment at 774.12 m/z can be visualized. Search programs correctly identify the fragmentation pattern for a series where one peptide remains intact, while the second peptide loses amino acids from one end, such as the +4 series. The green y9+4 ion at 670.90 m/z has an intact blue peptide linked to nine residues of the green peptide, having lost five residues from the N-terminal of the green peptide.

The dipeptides in Table are naturally occurring crosslinked peptides produced by human transglutaminase activity inside human SH-SY5Y neuroblastoma cells. The cells had not been permeabilized and treated with calcium. The cell lysate had not been treated with calcium chloride.

Procedure for Evaluating Candidate Crosslinked Dipeptides

The Search Compare page in Protein Prospector lists hundreds of candidate crosslinked peptides. We screened the candidates and made a list of those that have (a) Score + Score difference greater than 30 and (b) matched intensity greater than 40%. This reduced the number of candidate crosslinks to 89.

We examined 89 screened MS/MS spectra for crosslink specific ions. Crosslink specific ions are defined as ions that contain amino acids from both peptides. An acceptable candidate crosslink peptide was required to have a minimum of two crosslink specific ions in a series that defined an amino acid. We did not accept random crosslink specific ions. Crosslink specific ions can be b-ions, y-ions, ladder-ions, or ions from mixed fragmentation. Figure shows structures of crosslink specific y-ions and ladder ions. Figure shows structures of crosslink specific ions produced by mixed fragmentation. Figure shows an MS/MS spectrum for a crosslinked peptide pair in which the crosslink specific ions are b-ions.

Figure 5

MS/MS spectrum of a crosslinked peptide pair supported by crosslink specific b-ions. Glutamine 511 in the green peptide (Q9Y2G4 Ankyrin repeat domain-containing protein 6. ANKR6) is crosslinked to lysine 241 in the blue peptide (Q92541 RNA polymerase-associated protein RTF1 homolog. RTF1) via an isopeptide bond. The MH+4 parent ion is at 694.12 m/z. The crosslink specific ions are green b5+3, b6+3, and b7+3 (not marked), green b5+2, b6+2, and b7+2, and blue b6+3, b7+3, b8+3, and b9+3. Blue b6+4 (561.06) and b7+4 (593.32) and blue b9+4 (657.60) (not marked) are ladder ions that lose residues from the C-terminus. These are referred to as peeling sequence ions by Protein Prospector or [bn-1 + 18] fragments.[27] Additional support for the blue peptide includes the blue y2 to y5 series EEQEK. Additional support for the green peptide includes the green y6 to y9 series EQK.

Protein Prospector MS/MS spectra have red lines for masses assigned to the candidate crosslink and black lines for unassigned masses. The Peak Matches page lists masses and charge states for all red peaks but masses only (no charge state) for black peaks. Protein Prospector automatically subtracts background peaks from MS/MS spectra, which in some cases are important for a peptide sequence. To find the missing peaks and determine the charge state of all peaks, we examined MS/MS spectra retrieved from Thermo Scientific Xcalibur Qual Browser.

We calculated the mass difference between all peaks in a given charge state. The calculations included the unassigned black peaks. Mass differences were converted to specific amino acids. A table of dehydroamino acids in charge states +1, +2, +3, and +4 was useful (Table S1). For N- and C-terminal amino acids, the calculations took note of the fact (a) the mass of an N-terminal b-ion is 1 Da larger than the dehydro mass, (b) the mass of a C-terminal b-ion is 18 Da larger than the dehydro mass, (c) the mass of an N-terminal y-ion is the same as the dehydro mass, and (d) the mass of a C-terminal y-ion is 19 Da larger than the dehydro mass. We checked our calculations using the Proteomics Toolkit (Institute for Systems Biology), taking into consideration that a KQ crosslink loses NH3 (−17), whereas a standard peptide bond loses H2O (−18). Our calculations also took into account that +4 is written on each peptide in a crosslinked dipeptide by Protein Prospector, suggesting eight extra protons, whereas a crosslinked dipeptide has only four extra protons.

Criteria for Accepted Crosslinked Peptides

The MS/MS spectrum must contain amino acid sequence information from both peptides. A sequence is defined as a series of adjacent amino acids. Sequences of masses that include the crosslink specific ion and ladder sequences, that is, neutral losses from the parent ion, need only be one amino acid long. N-terminal or C-terminal sequences or sequences from internal fragmentation such as at proline must be at least three amino acids long.

Each peptide in the crosslink must contain a minimum of five amino acids.

A minimum of two crosslink specific ions must be present in a series that defines a crosslink specific amino acid.

We often find a peptide sequence in the MS/MS spectrum that is unrelated to the crosslinked peptides. If the unrelated peptide sequence includes none of the crosslink specific ions, we ignore the extra peptide because the extra peptide does not invalidate the crosslinked peptide.

Sometimes, the green and blue ions have the same mass. For example, in Figure , green y10+4 and blue y11+4 ions both have a mass of 699.17+4. The series continues having the same mass for green y11+4 and blue y12+4 ions of 716.93. However, green ions y9+4, y12+4, and y13+4 are present that have no matching blue ion. This means that the y9+4 to y13+4 series supports the green peptide but not the blue peptide.

Criteria for Rejecting a Candidate Crosslinked Dipeptide

Exclude crosslinks in charge states +6 and +2.

A single crosslink specific ion is not sufficient evidence to support a crosslinked peptide.

If two or more amino acids are appended to a convincing crosslinked series, we do not accept the crosslink specific ions as real. See Figure .

Figure 6

Example of a rejected crosslinked peptide pair. Manual evaluation showed that this MS/MS spectrum has no support for the green peptide. The green y11+4, y12+4, y14+4, and y16+4 series is interrupted by ions at 1033.05 and 1093.58. The interrupting ions change the sequence from a potential green ERATE to ENLET. The ENLET sequence does not fit the crosslink. The green y11+4 and y12+4 ions are disqualified as crosslink specific ions because they link to four residues that do not fit the green peptide sequence. The blue peptide is supported by blue y14+3, y16+3, y17+3, and y18+3, but without support for the green peptide, the candidate crosslink cannot be accepted.

We do not accept a crosslink if the candidate crosslinked lysine is at the C-terminus because trypsin does not cleave a modified lysine.

Crosslinks between two peptides from the same protein that involve closely spaced residues are rejected if the proposed crosslink can be fit to a linear peptide. The 1 Da difference between a KQ crosslink and a peptide bond can be an artifact introduced during the fragmentation process.

When the interval between two ions, for example, y7+4 (506.04) and y8+4 (545.06), corresponds to the mass of R+4 (39.02), we look for a possible intervening ion. An intervening ion at 530.80 (+4) breaks the R(39.02) into V(24.76) + G(14.26). In this case, we accept V + G as the correct sequence and reject R. This would disqualify a candidate crosslinked peptide if the R were a crosslink specific ion. Other instances where the interval mass for an amino acid is the same as that for a pair of amino acid intervals are as follows:

R = V + G

K = G + A

Q = G + A

W = G + E = A + D = S + V

N = G + G

Discussion

Intracellular Crosslinked Proteins Are Present

The prevailing view is that intracellular TG2 has no crosslinking activity under normal physiological conditions because the cell has high GTP (around 100 μM) and low Ca2+ (around 100 nM) concentrations.[26] At 100 μM, GTP binds to TG2 and stabilizes the closed form, which blocks access to the active site and therefore blocks crosslinking activity. Activation of TG2 to the open form is achieved by binding calcium.[20] However, at 100 nM, Ca2+ is unable to compete with GTP. Thus, it is argued that TG2 cannot attain the open conformation capable of crosslinking proteins.

The conclusion that TG2 has no crosslinking/transamidase activity inside a cell under physiological conditions is challenged by the following experimental results. Fesus and Tarcsa[28] found epsilon-(gamma-glutamyl)-lysine isodipeptides, the product of TG2 crosslinking activity, in Chinese Hamster Ovary cells under basal conditions. A fluorescence resonance energy transfer study found the open TG2 conformation beneath the cell membrane in human lung adenocarcinoma A549 cells, mouse fibroblasts 3T3, and Chinese hamster ovary cells.[29] Yamaguchi and Wang[30] detected incorporation of biotinylated pentylamine in HCT116 cells, in activity assays performed in the presence of the calcium chelators EGTA and EDTA. These results support the conclusion that cytoplasmic TG2 has low but detectable crosslinking activity. Low but detectable transamidase activity, measured by incorporation of radiolabeled putrescine, was found in WI-38 fibroblasts and MDA-MB-231 cells.[31] A cell permeable fluorescent inhibitor specific for TG2 was bound to the open conformation of TG2 inside the low Ca2+ intracellular environment of HUVEC and NIH3T3 cells.[32]

These observations suggest that the low intracellular Ca2+ concentration is sufficient to activate TG2 in some cases or that factors other than Ca2+ can stimulate TG2 to acquire the open conformation associated with crosslinking activity. Interaction with proteins or lipids may activate TG2 crosslinking/transamidase activity.[32,33]

Our finding of a limited number of crosslinked proteins in the cell lysate of human neuroblastoma cells is consistent with low, but real, intracellular TG2 crosslinking activity. Cells were lysed in the presence of EDTA, thus eliminating the possibility of activating transglutaminase crosslinking activity by calcium during handling. Isopeptide crosslinked proteins have a half-life of about 3 h in living cells, as measured by Fesus and Tarcsa with radioactive lysine.[28] The observation that isopeptide crosslinked proteins do not accumulate supports our finding of a low level of crosslinked proteins inside a living cell.

Crosslinked Proteins

We had expected to find crosslinking between abundant intracellular proteins, for example, tubulin and actin. However, this was not the case. With one exception, the crosslinked proteins were low abundance proteins. The exception was P10809, the 60 kDa heat shock protein, an abundant protein in SH-SY5Y cell lysate.

We had also expected to find internally crosslinked peptides, where the KQ crosslink is between two residues on the same protein. We rejected the few potential candidates because the difference in the parent ion mass between a linear peptide and a KQ crosslinked peptide is 1 Da. The mass spectrometer can delete 1 Da during the fragmentation process, making the assignment of an internal crosslink uncertain without additional evidence.

We specifically searched for crosslinks that included a RhoA peptide (UniProt P61586) because of a report that retinoic acid treatment of SH-SY5Y cells results in increased expression of TGase activity, leading to incorporation of radiolabeled putrescine into RhoA.[25] However, we found no RhoA peptide in either a crosslink with another protein or internally crosslinked. This does not invalidate the results of Singh et al. but suggests a low abundance of crosslinked RhoA. A targeted mass spectrometry method aimed at RhoA may detect the RhoA crosslink.

Hitomi Peptides

The rate-limiting step in a transglutaminase-catalyzed crosslinking reaction is formation of an acyl enzyme intermediate between the active site cysteine of TG and the γ-carboxyamide group of glutamine in the substrate. Glutamine-containing peptides that are preferred as substrates for transglutaminase have been identified.[34−37] Preferred glutamine donor peptides have a consensus sequence that depends on the TG isozyme. In Table S2, we compare the consensus sequences of preferred glutamine donor peptides to the glutamine donor peptides in the crosslinked proteins in SH-SY5Y cells. The glutamine donor peptides in SH-SY5Y cells have no consensus sequence.

Limitations

We found crosslinked peptides in SH-SY5Y cells that had been treated with 10 μM retinoic acid and 10 μM dichlorvos in the serum-free medium for 2 days before harvesting cells for our study. TG2 protein levels and transamidase activity are reported to be induced by retinoic acid in SH-SY5Y cells.[25] Our retinoic acid-treated cells have very low quantities of soluble, crosslinked proteins. We also found crosslinked peptides in SH-SY5Y cells that had been grown in 10% fetal bovine serum and had not been treated with retinoic acid. Thus, we suggest that our results have not been biased by retinoic acid.

We searched the cell lysates for soluble crosslinked proteins. We did not search for crosslinked proteins in the insoluble protein pellet. Highly crosslinked proteins are likely to form insoluble aggregates that we missed in our study of soluble proteins.

Dichlorvos is an organophosphorus pesticide that makes adducts on the side chains of tyrosine, serine, threonine,[38] and possibly lysine though adducts on lysine have not been reported. Organophosphorus toxicants do not inhibit TG2 transamidase activity.[39]

Conclusions

We describe a method for evaluating crosslinked proteins produced by the action of tissue transglutaminase, TG2. The method includes immunopurification of tryptic peptides, separation of peptides by liquid chromatography, acquisition of fragmentation spectra by mass spectrometry, Protein Prospector database search of MS/MS data, and manual evaluation of candidate crosslinks. Our criteria for accepting candidate crosslinks are conservative. Our method for evaluating zero-length isopeptide crosslinks will be useful for understanding the mechanism of therapeutic drugs aimed at inhibiting TG2 activity in cancer and other diseases.