A Streamlined High-Throughput LC–MS Assay for Quantifying Peptide Degradation in Cell Culture

Abstract

Peptides are widely used in biomaterials due to their ease of synthesis, ability to signal cells, and modify the properties of biomaterials. A key benefit of using peptides is that they are natural substrates for cell-secreted enzymes, which creates the possibility of utilizing cell-secreted enzymes for tuning cell–material interactions. However, these enzymes can also induce unwanted degradation of bioactive peptides in biomaterials, or in peptide therapies. Liquid chromatography–mass spectrometry (LC–MS) is a widely used, powerful methodology that can separate complex mixtures of molecules and quantify numerous analytes within a single run. There are several challenges in using LC–MS for the multiplexed quantification of cell-induced peptide degradation, including the need for nondegradable internal standards and the identification of optimal sample storage conditions. Another problem is that cell culture media and biological samples typically contain both proteins and lipids that can accumulate on chromatography columns and degrade their performance. Removing these constituents can be expensive, time-consuming, and increases sample variability. However, loading unpurified samples onto the column without removing lipids and proteins will foul the column. Here, we show that directly injecting complex, unpurified samples onto the LC–MS without any purification enables rapid and accurate quantification of peptide concentration and that hundreds of LC–MS runs can be done on a single column without significantly diminishing the ability to quantify the degradation of peptide libraries. To understand how repeated injections degrade column performance, a model library was injected into the LC–MS hundreds of times. It was then determined that column failure is evident when hydrophilic peptides are no longer retained on the column and that failure can be easily identified by using standard peptide mixtures for column benchmarking. In total, this work introduces a simple and effective method for simultaneously quantifying the degradation of dozens of peptides in cell culture. By providing a streamlined and cost-effective method for the direct quantification of peptide degradation in complex biological samples, this work enables more efficient assessment of peptide stability and functionality, facilitating the development of advanced biomaterials and peptide-based therapies.

1 |. Introduction

Peptides are short protein sequences that can signal cells and are also natural substrates for cell-secreted enzymes [1]. Proteases are among the most abundant classes of enzymes, and they degrade peptides by hydrolyzing the amide bonds between amino acids [2]. Peptides can be easily synthesized and modified with a range of bioconjugate chemistries [3], and they are often incorporated into biomaterial systems to improve bioactivity and enable cell-mediated scaffold degradation [4, 5]. For instance, cells are unable to adhere to many of the polymers used in synthetic cell culture systems, and these polymers are often functionalized with RGD peptides to increase cell attachment [6]. Cells are also frequently cultured within hydrogels, which are highly hydrated networks of crosslinked polymers designed to mimic the extracellular matrix that surrounds cells within tissues [7]. Crosslinking these polymer networks with peptides that can be cleaved by proteases enables cells to actively degrade their local matrix, which allows them to spread and migrate within the hydrogels [8, 9]. Proteolytic degradation of peptides has also been used to modulate the signaling environment around cells [10], and release growth factors tethered to the polymer matrix [11].

There are over 600 human proteases [12], including the matrix metalloproteinases (MMPs) family of proteases, which are commonly harnessed to induce degradation of biomaterial scaffolds [13, 14]. There are also numerous exopeptidases, which can nonspecifically degrade peptides that are incorporated into biomaterials or are used therapeutically [15, 16]. The most common method for quantifying peptide degradation by proteases involves synthesizing peptides with a fluorophore and quencher at each end, then incubating the conjugate with the protease(s) of interest and measuring the fluorescence. However, many proteases, including MMPs, have promiscuous substrate specificity and a single peptide sequence is often cleaved by many proteases [17]. The total proteolytic activity near a cell or within a tissue is also incredibly complex, as cells secrete dozens of proteases, each with their own range of substrate specificities, in addition to protease inhibitors, and other cofactors that can modulate protease activity [18]. Exopeptidases target the terminal amino acids of peptides, and their kinetics are highly sensitive to the chemistry of the termini [15]. This poses a challenge for quantification methodologies that require the ends of the peptide to be modified, such as fluorophore–quencher pairs [19]. Peptides can be rapidly made on automated synthesizers, which enables the facile creation of peptide libraries to rapidly screen numerous sequences [20]. However, quantifying degradation using fluorescence-based methods limits the number of analytes in a single sample due to the need for minimally overlapping fluorescence spectra.

LC–MS is a robust analytical tool that integrates liquid chromatography (LC) to separate out and mass spectrometry (MS) to facilitate the multiplexed quantification of peptide degradation within complex samples [21]. In LC, a mixture is introduced to a chromatography column, where the analytes are separated based by a gradient of solvents. The separated mixture is then sent to a MS, which can quantify up to thousands of different peptides within a single run [22]. Since the peptides are identified by their molecular weight, there is no need to label the molecules with fluorophores or other chemistries, which may change bioactive or degradation kinetic. Furthermore, a single LC–MS run can take less than 10 minutes, which enables a high sample throughput.

A key challenge in using LC–MS to quantify the functional degradation of peptides during cell culture or within biological samples is that these types of media contain different classes of molecules that are detrimental to either LC or MS. Specifically, the presence of high concentrations of proteins and lipids will readily foul LC columns [23], and the nonvolatile salts are detrimental to reproducible ionization in MS and accumulate on the MS [24]. A variety of methods exist to isolate the analyte(s) of interest from these complex mixtures [25, 26]; however, they are often expensive and time-consuming. Furthermore, these purification techniques typically need to be optimized for a single analyte, and studies utilizing a library of peptides with a range of physiochemical properties are challenging for these methods.

In this work, we have optimized methods for quantifying the stability and degradation of peptide libraries within cell culture media (Figure 1A). Directly injecting cell culture media into an LC–MS has several advantages over methods that utilize purification steps because it requires minimal sample preparation and greatly reduces the chances of sample loss prior to analysis (Figure 1B). We removed the salts present in the samples by sending the first minute of the chromatography run to waste, which bypasses the MS. A downside to directly injecting cell culture media onto the column is that the presence of lipids and proteins within the cell culture media will degrade the performance of chromatography column over time. We quantified changes in column properties over the lifetime of a column and show that a single LC column can give reliable results for hundreds of runs without any sample purification and we identify the characteristics that are hallmarks of column degradation. This method has benefits to existing purification protocols, such as solid-phase extraction (SPE), in that it is less expensive, takes less time, and minimizes sample loss, even when studying complex peptide libraries. LC–MS is a widely available technique, and these methodologies can be easily adopted by any laboratory with access to LC–MS. Overall, this will help to advance our ability to quantitatively understand of how cells degrade peptides and improve our ability to design peptides that improve the efficacy of biomaterials and biomedical therapies.

FIGURE 1 |.

(A) Peptides and internal standards are added to cell culture media that contains numerous proteins and lipids. (B) This mixture can either be purified or remain unpurified prior to (C) running the sample on LC–MS, in which the mixture is loaded onto a column follow by mass spectrometry.

2 |. Materials and Methods

2.1 |. Materials

All peptide synthesis reagents were purchased from Chemscene or Ambeed. N,N-dimethylformamide (DMF) and dichloromethane (DCM) (both from VWR BDH Chemicals), piperidine (Millipore Sigma), trifluoroacetic acid (TFA) (Millipore Sigma), diethyl ether (Fisher Scientific), and N, N-Diisopropylethylamine (DIPEA) (VWR) were used as purchased.

2.2 |. Peptide Synthesis Procedure

Peptides for the split-and-pool libraries were synthesized using standard solid-phase peptide synthesis (SPPS) protocols standard Fmoc-protected amino acids (Chemscene) on a Rink amide resin (Supra Sciences). All amide couplings were done using O-(6-chlorobenzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HCTU) in DMF. For each coupling, the amino acid, HCTU, and DIPEA were added in a 4:4:6 M ratio to the peptide. During peptide synthesis, a ninhydrin test (Anaspec) was performed after every addition to test for the presence of free amines. Upon a positive test, the coupling was replicated until the test was negative. After successful coupling, the Fmoc group was removed by washing the resin with 20% piperidine in DMF twice for 5 min. A ninhydrin test was performed to check for a positive result. Other peptides were synthesized on an automated Liberty Blue peptide synthesizer (CEM) using ethyl cyanohydroxyiminoacetate (Oxyma, from Chemscene) and diisopropylcarbodiimide (DIC, from Chem-Impex) chemistry.

We used split-and-pool syntheses to create peptide libraries having the sequence AcβA-RGEFV-X and AcβA-X-RGEFV-AcβA, where X is each of the canonical amino acids except for cysteine (Figure S1). For split-and-pool steps, the resin was deprotected with piperidine, washed 3× with DMF, and then the entire amount of resin was weighed on a scale. This number was divided by 22 and this was split 19 ways into 19 separate tubes to ensure equal splitting of the resin. All natural amino acids (A, D, E, F, G, H, I, K, L, M, N, P, Q, R, S, T, V, W, and Y) were coupled in 19 separate 15-mL tubes. Upon successful coupling of all 19 amino acids, the 19 unique peptides were recombined to form a library. Once a peptide library was completed, it was capped with acetic anhydride (Sigma-Aldrich) in a 10:5:100 acetic anhydride:DIPEA:DMF solution twice for 5 min, and then, a ninhydrin test was performed to check for complete capping of the free amines. After capping, the resin was washed 3× with DMF and 3× with DCM.

Peptide libraries and all peptides containing tryptophan were cleaved using 92.5% TFA, 2.5% H₂O, 2.5% triisopropylsilane (TIPS), and 2.5% dithiothreitol (DTT). Peptides not containing a tryptophan were cleaved using 95% TFA, 2.5% H₂O, and 2.5% TIPS. Peptides were typically cleaved for 2–3 h at room temperature using approximately 25 mL of cleavage solution per millimole of peptide. After cleaving peptide(s) from the resin, they were precipitated in diethyl ether, centrifuged for 5 min at 4000 rpm, and the supernatant was discarded. The peptide pellet was washed with diethyl ether, centrifuged two more times, and then dried. Once dried, the peptide(s) were dissolved in water and neutralized with ammonium hydroxide prior to purification.

All peptides were purified using high-performance liquid chromatography (HPLC) using a Phenomenex Gemini 5 μm NX-C18 110 Å LC Column 150 × 21.2 mm. Gradients were run from 95% Mobile Phase A (water with 0.1% TFA) and 5% Mobile Phase B (acetonitrile with 0.1% TFA) to 100% Mobile Phase B. A typical HPLC run featured a two-minute equilibration step, followed by a 10-min ramp from 95% Mobile Phase A to 100% Mobile Phase B, and then 2 minutes of equilibration at 100% Mobile Phase B, before ramping back down to the starting conditions. Notably, the split-and-pool libraries were ramped up to 100% Mobile Phase B over 2 minutes, since these libraries consisted of approximately 19 different peptides, which were not intended to be separated from each other. After purification, all peptides were lyophilized and were ready to use.

2.3 |. Quantification of Sample Storage Conditions on Peptide Degradation

We quantified cell-mediated peptide degradation at defined time points to confirm that degradation was caused by cells and not storage conditions. Human mesenchymal stem cells (hMSCs) (Rooster Bio, LOT 310268, a 19-year-old Eritrean/East African male) at Passage 3 were seeded into 24-well plates at a seeding density of 75,000 cells per well in 1 mL of RoosterBasal-MSC-CC (RoosterBio, SU-022) containing RoosterBooster-MSC (RoosterBio SU-003). After 24 h, the media was changed, and the peptide library AcβA-RGEFV-X-NH₂ was added to the cell media for a final concentration of 37 μM per peptide. Each Library was tested in three times, each in triplicate with three technical repeats for a total of 27 wells. 40 μL samples were collected from the media at hours 0, 1, 4, 8, 24, and 48 in duplicate.

Samples were taken at desired time points, with one set of samples being loaded directly on to the LC–MS and measured as soon as possible, then stored at 4°C. The second set of samples was immediately frozen at −80°C. After 48 h, samples were kept frozen and thawed just prior to LC–MS, and 4 μL of acetic acid was added to each well. After three weeks, the samples were tested again. A nondegradable internal standard, NH₂-βF(βA)₆-Amide, where βA is a β-alanine and βF is β-homophenylalanine, was included at a 37 μM concentration for all studies.

2.4 |. Peptide Calibration Curve Studies

The peptide library AcβA-X-RGEFV-NH₂ was dissolved in RoosterBasal cell culture media containing RoosterBooster-MSC at the following concentrations: 2.6, 13, 26, 52, 78, 130, 56, 182, 208, 234, and 260 μM per peptide. Each concentration was injected onto the LC–MS three times.

2.5 |. Sample Injection Solvent Studies

The AcβA-X-RGEFV-βA peptide library was added to solvents: water, 50% acetonitrile in water, PBS, RoosterBasal-MSC-CC (RoosterBio, SU-022) containing RoosterBooster-MSC supplement (RoosterBio SU-003), VascuLife Basal Media (Llifeline Cell Technology, LM-0002), and macrophage serum-free media with L-Glutamine (Gibco 12,065–074) with 1% antibiotic–antimycotic (Gibco, 15,240–062) at a concentration of 37 μM per peptide. This was also repeated with each solvent containing 10% acetic acid at a concentration of 37 μM for a total of 12 different solvents tested. The library within any given solvent was tested three times, each in triplicate with three technical repeats for a total of 27 wells. Samples were immediately loaded into the LC–MS for data collection.

2.6 |. Solid-Phase Extraction Peptide Purification

The samples were dissolved in 1 mL of water with 0.1% acetic acid and purified using Sep-Pak C18 Plus Short Cartridges. The Sep-Pak cartridges were conditioned with 6 mL of acetonitrile with 0.1% acetic acid. Once the 0.1% acetic acid in water had been flushed through, the samples were loaded, and it was washed with water containing 4% acetonitrile and 0.1% acetic acid to wash the salts through the Sep-Pak cartridge. The peptides were then eluted with 50:50 acetonitrile:water and 0.1% acetic acid. Both the liquid that was flushed through (the wash) and the eluted peptides were collected separately.

2.7 |. Quantification of Column Performance

A new column (ProntoSIL C18 AQ, 120 Å, 3 μm, 2.0 × 50 mm HPLC column, PN 0502F184PS030) was first primed by flushing 0.1% acetic acid in ultrapure water, followed by 0.1% acetic acid in acetonitrile, and last with 0.1% acetic acid in ultrapure water all at a volumetric flow rate of 300 μL/min for 20 min. Peptide Library AcβA-X-RGEFV-βA was dissolved in ultrapure water containing 10% acetic acid and was used to track performance of the HPLC column. Conditioned media consisting of high-glucose DMEM supplemented with 1% vol/vol anti/anti, 7.5% wt/vol bovine serum albumin (BSA), 0.1% vol/vol L-proline (Sigma-Aldrich), 1% vol/vol insulin–transferrin–sodium selenite (ITS; Sigma-Aldrich), 0.2% vol/vol dexamethasone (Sigma-Aldrich), 0.1% vol/vol ascorbic acid, 0.1% vol/vol transforming growth factor-β1 (TGF-β1; Peprotech), and 0.1% vol/vol linoleic acid (Sigma-Aldrich) was used [27]. After every injection of the peptide library into the HPLC, the fouling agent was injected four times. The amount of each of the 19 peptides present in the library was quantified for each injection, and the column was determined to have degraded when the calculated values for peptides began to deviate by more than 10% from their original values.

2.8 |. LC–MS Data Acquisition

From each sample, 10 μL of crude solution was introduced by the LC–MS through an Thermo Scientific Vanquish LC System (Thermo Fisher Scientific) which outputted to a Thermo Scientific LTQ XL Linear Ion Trap Mass Spectrometer (Thermo Fisher Scientific). The sampled mixture was trapped on a column (ProntoSIL C18 AQ, 120 Å, 3 μm, 2.0 × 50 mm HPLC column, PN 0502F184PS030, MAC-MOD Analytical Inc.). The samples were loaded onto the column with a solvent containing acetonitrile/water, 5:95 (v/v) containing 1% acetic acid at a flow rate of 300 μL/min and held for 1 minute. The sample was then eluted from the column with a linear gradient of 5–40–Solvent B (1% acetic acid in acetonitrile) at the same flow rate for 5 minutes. This was followed by a 1-min ramp-up to 100% solvent B, where it was re-equilibrated with solvent A (1% acetic acid) to 5% solvent B over the course of 1 min and held there for 2 min. The column temperature was a constant 29°C. The MS was operated in positive ion mode. Using a heated ESI, the source voltage was set to 4.1 kV, and the capillary temperature was 350°C.

2.9 |. LC–MS Analysis

Data analysis was performed on Xcalibur Software (Thermo Scientific). Peptides were identified automatically using the Thermo Xcalibur Processing Setup window where the mass (m/z) ±0.5 AMU and expected retention times for each quantified peptide were inputted. This was later used to isolate individual peptide species from the total ion count (TIC) trace using the Thermo Xcalibur Quan Setup window, where the area under the curve was calculated and visually inspected for accuracy. All peptides were normalized to a nonchanging internal standard, NH₂-βF(βA)₆-Amide, where βA is a β-alanine and βF is β-homophenylalanine. The integrated peak area of the peptide of interest was divided by the NH₂-βF(βA)₆-Amide internal standard to create an area ratio. Relative amounts of a peptide of peptide were then calculated by normalized all values to their corresponding time zero area ratio. Data were visualized using RStudio and Excel.

2.10 |. Statistical Analysis

All statistical analysis was done in RStudio. Data with two conditions were analyzed with a Student’s t-test, and data with more than two conditions were analyzed using an ANOVA followed by a Tukey’s post hoc test.

3 |. Results and Discussion

3.1 |. Identification of Nondegradable Internal Standards for Peptides

Peptide degradation during culture was assessed by measuring peptide concentrations at different time points and comparing them to the initial concentration at time zero. LC–MS is often employed for quantifying the concentration of biological compounds by injecting a defined sample volume onto a column, followed by chromatographic separation and MS.

However, a variety of factors can influence the concentration of peptides within the samples independently of peptide degradation. This includes sample evaporation during culture, which increases the concentration of all solutes, dilution during sample processing, or errors in the amount of peptide injected due to factors such as air bubbles. Furthermore, the amount of peptide that is measured on the MS can vary due to changes in instrument parameters over time [7]. To compensate for this variability, quantitative studies in LC–MS typically benchmark each of the analytes to nondegradable internal standards [28]. Internal standard selection is important to ensure accurate and reliable analysis of LC–MS-based quantification of peptide-based systems to compensate for variability in external factors on the LC–MS [29, 30]. To that end, a systematic approach was taken to choosing and assessing the appropriateness of the standard based on the requirements that it is stable throughout culture and is well retained on the chromatography column.

Internal standards should have physiochemical properties akin to the analyte(s) of interest, ensuring similar behavior during purification and analysis, while being resistant to degradation during the experiment. A polypeptide is an ideal internal standard for peptide degradation studies; however, the susceptibility of peptides to enzymatic degradation was a significant concern. β-Amino acids are noncanonical amino acids that have two carbons between amide bonds instead of one and have been shown to be resistant to most proteolytic degradation [31–33]. β-Alanine (βA) was chosen because it has a chemical structure similar to the amino acid glycine, and the Fmoc-protected β-alanine used during synthesis is inexpensive. Internal standards must not only resist degradation but also remain on the chromatography column and ionize effectively in the MS for accurate quantification. NH₂-βFβA₆-Amide was chosen for its highly repeatable peptide elution time of 2.68 min, its repeatability in measurement within 48 h of incubation with within the sample matrix (Figure 2), and its net charge of +1 at in the LC–MS mobile phase.

FIGURE 2 |.

A NH₂-βFβA₆-Amide internal standard has physiochemical properties that are similar to peptides but is stable from proteolytic degradation for 48 h in cell culture media.

3.2 |. Validation of Sample Linearity Within Complex Mixtures

A central goal of this work is to identify a simple methodology that requires minimal effort in obtaining quantitative degradation results from complex mixtures. Standard curves are widely used within biological assays to convert measured values to absolute amounts of analyte. However, generating the standard curves can be laborious, especially for multiplexed assays containing numerous analytes. Ideally, there would be a linear dependence of the measured peptide amount in MS compared to amount of peptide present in the solution over the concentration range used in degradation studies.

We synthesized two libraries each containing 19 peptides with a range of charge states and hydrophobicities to ensure that the results are robust and not specific to an individual peptide. These libraries have the form AcβA-RGEFV-X-βA and AcβA-X-RGEFV-βA, where AcβA is an acetylated β-alanine on the N-terminus, and the position X contains peptides with every canonical amino acid except cysteine, which forms disulfide bonds under physiological conditions. The two different libraries have the variable amino acid position on either the N-terminus or C-terminus to account for any influence of variable chemistry at each terminus. We made a series of dilutions of the peptide libraries that kept the concentration of the NH₂-βFβA₆-Amide internal standard constant and varied the levels of analyte libraries (Figure 3). The concentration of the peptides in the library were varied over two orders of magnitude, from 2.6 μmolar to 260 μmolar. A graph containing a representative subset of the data indicates that the relationship between the amount in the sample and calculated amount in MS was linear for all peptides, typically having an R-squared value above 0.98, indicating a high degree of correlation (Figure 3). Table 1 details the R-squared values for all samples, showing that while the AcβA-X-RGEFV-βA peptides tended to have higher R-squared values than AcβA-RGEFV-X peptides, all the tested peptides had a high degree linearity.

FIGURE 3 |.

Calibration curves for a subset of the AcβA-X-RGEFV-βA peptide library show that the linear relationship between the measured amount on the LC–MS and the concentration of peptide in the sample exists within the experimental range. The area ratio is the ratio of the area of the peptide peak in mass spectroscopy divided by the area of the internal standard peak. Error bars are ± standard deviation.

TABLE 1 |.

A simple linear regression was fit to each of the peptides within two different libraries. Every peptide has an R-squared value close to one, indicating the linear relationship between the measured and actual amounts in MS.

AcβA-RGEFV-X
Amino Acid	A	D	E	F	G	H	I/L	K	M
R2 value	0.999	0.994	1.000	0.999	0.999	0.997	1.000	0.995	1.000
Amino Acid	N	P	Q	R	S	T	V	W	Y
R2 value	0.999	0.986	1.000	0.999	0.998	0.999	0.999	0.999	1.000
AcβA-X-RGEFV-βA
Amino acid	A	D	E	F	G	H	I/L	K	M
R2 value	1.000	0.995	0.997	0.999	0.999	0.993	1.000	0.999	1.000
Amino acid	N	P	Q	R	S	T	V	W	Y
R2 value	0.999	0.999	0.999	0.999	0.999	0.999	0.999	1.000	1.000

3.3 |. Effect of Sample Storage on Peptide Stability

In our degradation studies, we incubate peptide libraries with cells for a defined period, after which peptide degradation is quantified using LC–MS. Sample injection onto the LC–MS is not typically done immediately after culture, and in most cases, the samples are stored in a − 80°C freezer prior to analysis. Before analysis, the samples are thawed and the plates are loaded into the LC–MS, which is kept at 4°C; however, for extended runs, an individual sample may not be injected onto the LC–MS for more than 72 h. We cultured our peptide library with hMSCs, which are widely used within biomaterials and have importance both in understanding tissue biology and in regenerative medicine. Cell culture media contains proteases, such as MMPs [34] and others that have the potential to further degrade peptide libraries after the sample has been collected, which is undesirable. The peptide library AcβA-RGEFV-X was previously shown to degrade in the presence of hMSCs over a 48-h period, and we used this library to quantify the effects that sample storage has on further peptide degradation. The overall goal of these experiments was to identify simple storage conditions that minimize sample degradation over a three-week storage period.

To quantify the effects of storage conditions on sample quality, 40 μL of cell culture media was taken and a portion was immediately injected onto the LC–MS to quantify the amount of peptide initially present in the samples. The plates in the “untreated” condition were then stored at 4°C and “acidified” samples were frozen at −80°C until after the 48-h timepoint, where they were treated with 4 μL of acetic acid, injected onto the LC–MS, and then stored again at −80°C. It is worth noting that the untreated and acidified data are paired, with both being drawn from the sample at the same time. After 23 days, the samples were injected again onto the LC–MS. The amount of each peptide in a sample is quantified by dividing the area of the peptide peak in LC–MS by the area of the NH₂-βFβA₆-Amide internal standard. Since each peptide has a different propensity to undergo ionization, the integrated area of different peptide sequences can vary by an order of magnitude, even if the peptides are present in the sample at the same concentration. To better understand sample variability across different peptides, we quantified the coefficient of variation for each sample. The coefficient of variation is calculated by dividing the standard deviation of the sample by the sample mean, which normalizes the variance to be the fraction of the sample mean, irrespective of the magnitude of the data. In our studies, we found that the untreated samples showed more statistically significant variability after 24 h across all amino acids in the library (Figure 4).

FIGURE 4 |.

Samples were either left untreated and stored at 4°C, or were stored at −80°C and acidified to 10% acetic acid prior to LC–MS. We then ran the samples on LC–MS and found that storing the samples at −80°C and treating them with acetic acid-reduced variability within technical replicates. Pairwise comparisons were made using a Student’s t-test. * indicates p < 0.05, and ** indicates p < 0.01. Error bars are ± standard deviation.

3.4 |. Effects of Sample Solvent Conditions on Chromatography

Reversed-phase chromatography is a highly effective technique for separating peptides [35]. The use of a hydrophobic sorbent as the stationary phase results in affinity-based partitions, determined by the interactions of the solute with the mobile phase and the stationary phase [35]. Ideally, samples should be dissolved in solvents having the same composition as the mobile phase that is present within the system when the sample is being loaded onto the column. However, this can require processing steps to remove the initial solvents in which the samples are dissolved, and replacing them with the mobile phase, which can be both time-consuming and increases the chances of sample loss.

To study effects of initial peptide solvent on column performance, the peptide library AcβA-X-RGEFV-βA was chosen, where X is every amino acid except cysteine. Previous work has shown that chemistry of peptide termini is the most effective means by which nonspecific degradation of peptides may be controlled as compared to concentration or even PEGylation [15]. In previous studies, peptides modified with an N-terminal acetylated β-alanine were shown to be the most proteolytically resistant to degradation within a 48-h period when incubated with a variety of cell types [15]. This resistance to proteolytic degradation makes the AcβA-X-RGEFV-βA library a good model system to study the effects of the sample matrix on peptide affinity to the stationary phase of the LC–MS, as well as performance on the MS.

To better understand the effects that different sample solvent conditions have on chromatography, we dissolved the AcβA-X-RGEFV-βA library in 12 different solvents (Figure 5 and S1). This includes pure water, phosphate-buffered saline (PBS), and three different cell culture medias. We also tested conditions with 10% acetic acid to account for the acidification that is used to prevent protease degradation, and acetonitrile, which can deactivate proteases and is used during sample purification. For each of the 12 conditions, the peptide libraries were tested in triplicate, each with three technical repeats, for a total of nine wells per solvent. In previous work, salt has been shown to alter relative separation and retention times of species within the sample matrix due to changes in adsorption due to the electrical double layer repulsion [36]. In our studies, the presence of salt in the sample has a minimal effect on the area ratio, but reduces retention time on the stationary phase, particularly for peptides where the X position contains positively charged residues H, K, and R.

FIGURE 5 |.

(A) The AcβA-X-RGEFV-βA peptide library was dissolved in a variety of common media conditions and solvents, and run on the LC–MS. The area ratio is the ratio of the area of the peptide peak in mass spectroscopy divided by the area of the internal standard peak. The amount of each peptide was quantified, and it was seen that having significant amounts of organic solvents such as acetonitrile, reduces the reliability of the results. This is due to the poor retention on the column, which is mostly seen in the form of peak splitting. When a peptide is well-adsorbed to the column, there is a single peak (B), while those with poor chromatography end up in both the breakthrough peak, which is typically sent to the waste, in addition to the normal peak (C).

We found that the presence of acetonitrile in the injected solution had a profound effect on LC–MS performance. Acetonitrile is often mixed with water to improve the solubility of hydrophobic peptides, but we found that 50% acetonitrile in water caused peak splitting in the internal standard such that a significant fraction of the internal standard failed to adsorb to the column and eluted prior to the start of the solvent gradient (Figure 5B,C).

Since the mobile phase prior to the start of the gradient is typically sent to the waste, the presence of the early elution of the internal standard will adversely impact the quantification of every analyte within the sample. This is problematic because the effects of acetonitrile on the AcβA-X-RGEFV-βA peptide library was not consistent, and a portion of the hydrophilic peptides also failed to adsorb to the column, while the hydrophobic peptides had good retention across all 12 solvent conditions (Figure S1).

3.5 |. Quantification of Chromatography Column Failure

A key benefit of LC–MS is that individual runs are often less than 12 min, and sometimes less than 5 min. Sample injection and collection are typically automated and hundreds of samples can be loaded at a time, which enables LC–MS instruments to run dozens to hundreds of samples per day. While the combination of high throughput and the powerful characterization capabilities of LC–MS is advantageous, this workflow requires preparation of numerous samples for analysis. The presence of lipids and proteins in biological samples is detrimental to the longevity of reverse phase chromatography columns, and a variety of approaches exist to separate these compounds from the analytes of interest prior to LC–MS. For instance, trichloroacetic acid is frequently mixed with samples to precipitate soluble proteins [37], and disposable SPE columns, whose properties are similar LC–MS columns, are used to load the crude sample mixture followed by elution of the desired analytes [38].

A downside of using SPE columns is that each column is typically single use, and generally cost $3–5 per sample, which can be a significant cost burden when analyzing hundreds of samples per week. More generally, both precipitation-based and SPE-based methods have drawbacks in that they need to be optimized for a specific analyte to minimize sample loss and ensure reproducibility, and this is often not possible for studies utilizing libraries of peptides with significantly different physiochemical properties. We validated this by loading our AcβA-X-RGEFV-βA library onto an SPE column and purified the peptides from the proteins, salts, and lipids using the manufacturer’s instructions. In this process, a sample is loaded onto a disposable column, constituents which have poor interactions with the column, such as salts, are washed through, and then the sample is eluted, leaving chemical species such as proteins and lipids on the column. We then ran LC–MS on both the crude mixture prior to purification and the initial wash waste and found that all peptides in the library were present in the waste, but that hydrophilic peptides are enriched compared to hydrophobic peptides (Figure 6). This highlights the difficulty in purifying libraries of analytes with different physiochemical properties prior to analysis. Furthermore, performing these purification steps on hundreds of samples is a significant time and experimental burden.

FIGURE 6 |.

The AcβA-X-RGEFV-βA library was dissolved in water and loaded onto a solid-phase extraction column. We then performed the purification steps according to the protocol from the manufacturer. We first primed the column using a 50:50 acetonitrile–water mixture with 0.1% acetic acid, then an aqueous-rich mixture with 4% acetonitrile and 0.1% acetic acid. The peptide solution was then loaded onto the column and we then performed an aqueous-rich washing step, followed by eluting the peptides from the column with an acetonitrile–water mixture. We performed LC–MS on both the aqueous-rich washing step and the eluted fraction. We found that every peptide was present in the washing step fraction; however, hydrophilic peptides were enriched, highlighting that complex mixtures with different physiochemical properties are a challenge for sample purification methods. Error bars are ± standard deviation. N = 3.

We sought to greatly simplify sample preparation by injecting the biological samples directly on the LC column without any purification steps. This prevents any sample losses prior to analysis, requires minimal effect for even for hundreds of samples, and negates the need for costly consumables. A significant downside to this approach is that the proteins and lipids present in biological samples are detrimental to the performance of the LC column. While these columns are not typically treated as disposable, they can be purchased for under $300, which is not prohibitively expensive if each column can be used for hundreds of runs. We quantified changes in the column properties after repeated sample injection by recording changes the column pressure, retention time of each of the analytes on the column, and their integrated area in MS to better understand how repeated sample injections influence column performance (Figure 7). It should be noted that specific types of media are used when culturing a cell type. Media formulations frequently incorporate fetal bovine serum (FBS), which is rich in proteins having total concentrations between 30 and 45 mg/mL [39]. In these studies, we used media with a high protein concentration 7.5% wt/vol (75 mg/mL) of BSA to quantify the influence that repeated sample injections have on column performance. Due to the variability inherent in FBS, there has been an emphasis on the use of fully defined media formulations [40]. Fully defined medias typically have lower protein content than those with FBS and peptides dissolved in these medias should reduce column fouling, which will likely increase the number of LC–MS samples that can be run before the column needs to be replaced.

FIGURE 7 |.

To evaluate the lifespan and performance of LC columns during degradation studies, we conducted a repeated injection protocol totaling hundreds of injections. The protocol involved injecting the AcβA-X-RGEFV-βA peptide library once, followed by four consecutive injections of media with 7.5% wt/vol BSA; this entire sequence was repeated 124 times until column failure (620 total injections). The arrows indicate when the first injection was performed after at least 24 h of nonuse. The red arrow marks the onset of column failure where failure of the column is initiated. (A) The LC–MS pressure at the beginning of sample runs increases steadily for the first 300 injections and then plateaus. (B) The retention time for each analyte decreases slightly over the course of hundreds of injections. (C) The area ratio of most peptides, which is the area of the analyte divided by the area of the internal standard, decreases slightly over the series of injections, except for hydrophilic molecules. (D) Normalizing the area ratio to the values from the initial injections shows that column failure occurs when hydrophilic peptides are no longer retained on the LC column.

To study the effects of repeatedly injecting cell culture samples containing proteins on column lifetime and consistency in data collection, we performed hundreds of injections of media containing 7.5% BSA. During these runs, we ran the AcβA-X-RGEFV-βA peptide library every fifth injection to quantify column performance over time, including sample and column characteristics. The accumulation of lipids and proteins on the column will physically clog the column and increase the pressure at any given flow rate [41]. We observed a steady increase in pressure over the first 200 injections, with the pressure increasing by ~100 bar (Figure 7A). Injections were repeated for over 600 runs, during which time the retention time of each peptide was also found to be consistent (Figure 7B). An exception to this is a few notable instances in which the column was used after days of disuse, where the retention time was evenly shifted across all analytes (black arrows in Figure 7B).

In our studies, the hallmark of column failure is the inability for hydrophilic peptides to be retained on the column. We started to observe column failure after 348 injections of the media containing 7.5% BSA and 87 injections of the library, for a totally of 435 injections (Figure 7C). To better quantify changes in column performance, we normalized the area ratio of each peptide during the initial runs on the column as a baseline and then divided all runs by this normalized value (Figure 7D). It should be noted that the first two or three injections of a series of runs, as depicted by the arrows on Figure 7, can have substantially altered chromatographic characteristics compared to subsequent injections, and this effect is independent of column health. We found that after 435 injections, the most hydrophilic sequences, which are the AcβA-X-RGEFV-βA peptides containing the positively charged amino acids lysine and arginine in the X position, started showing decreased area ratio due to poor retention on the column. These results show that column failure is typified by the poor retention of hydrophilic peptides but highlight that using libraries with hydrophilic sequences are effect for assessing column health.

3.6 |. Quantification of Peptide Degradation

In Figure 3 and Table 1, we showed that the amount of peptide quantified by LC–MS is directly proportional within the concentration range examined within biological samples. As a result, we can divide the amount of peptide measured at specific timepoints in culture by the amount initially present to generate the fraction of peptide remaining in the sample to quantify degradation. We synthesized a series of six peptides with a range of physiochemical properties and cultured them with hMSCs for 48 h (Figure 8). These samples were run on the LC-MS and eluted from the column ranging in time from 3 to 6.5 min. We were able to quantify the degradation of all the peptides in the library and found that individual peptide sequences can have substantially different degradation rates. Proteases are one of the largest classes of proteins, and proteolytic activity is a complex and highly regulated network of enzymes, inhibitors, and cofactors. This functional methodology is versatile way to quantify how individual peptide sequences are degraded under desired conditions. The only major limitations are that very hydrophilic peptides are poorly retained on C18 chromatography columns, and very hydrophobic peptides are not soluble in cell culture media. It should be noted that while proteases are a major source of peptide degradation, any peptides that are internalized by cells or modified via nonproteolytic mechanisms during culture will count as degraded.

FIGURE 8 |.

A library of six different peptides were incubated with hMSCs for 48 h. Time points were taken at 1, 4, 8, 24, and 48 h and the fraction of peptide remaining was quantified using LC–MS. These results show that this methodology is compatible with peptides having a range of physiochemical properties and that the extent of peptide varies widely between different sequences. N = 3, and error bars are ± standard deviation.

4 |. Conclusions

In this work, we have developed a method for quantifying the degradation of peptide libraries within complex biological samples. Biological samples contain proteins and lipids, which can both be difficult to easily remove from samples and will foul chromatography columns. We found that hundreds of injections of biological samples without any purification can be run on an LC–MS column and that column failure happens when hydrophilic peptides fail to be retained on the column. Adding 10% acetic acid to the samples was also found to be sufficient to prevent further peptide degradation after the sample was collected. This method utilizes widely available instrumentation and was designed to minimize the need for consumables, while reducing the time-consuming preparation steps. Peptides can be easily made on automated synthesizer and show great promise as therapeutics and as components within biomaterials. Library-based peptide assays that quantify peptide degradation under physiological conditions are needed to improve our control over cell-peptide interactions, and we found that we could quantify the degradation of a library of peptides over 48 h when they were incubated with stem cells. This work will help scientists and engineers use functional approaches to improve biomedical therapies and platforms.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.