Attaching poly(ethylene glycol) to a peptide changes almost everything measurable about the molecule except its sequence: the mass, the hydrodynamic size, the viscosity of a concentrated solution, the elution position on every column you own, and — if you used amine chemistry — the isoelectric point. It also turns one defined species into a population, because PEG itself is a population. This guide covers where the polymer attaches, how you steer that choice, and why the analytical work afterward is harder than for an unmodified peptide.
PEG is a distribution, not a molecule
Poly(ethylene glycol) is a chain of repeating -CH2-CH2-O- units with hydroxyl termini; the methoxy-capped form, mPEG, leaves one end reactive, and most conjugation reagents are built from it. The repeat unit has an average mass of 44.05 Da, so a reagent sold as "20 kDa" carries on average about 454 of them (20,000 / 44.05 = 454). Chains one unit shorter and longer are in the same bottle, 44 Da apart.
How wide that spread is gets reported as dispersity, the ratio of the weight-average to the number-average molar mass. For a distribution with mean Mn and standard deviation of the number distribution s, the ratio equals 1 + (s / Mn) squared. Run the arithmetic on a nominal 20 kDa chain: a dispersity of 1.01 implies s = 0.1 x 20,000 = 2,000 Da, which is about 45 repeat units. Two standard deviations either side of the mean therefore spans roughly 180 distinct chain lengths, and those 180 masses hold something like 95% of the material. A "narrow" polymer is still an ensemble of a couple of hundred molecular species.
Real reagents sit in that range. Kemptner and colleagues characterized four mPEG-succinimidyl succinate derivatives by two independent methods, gas-phase electrophoretic mobility molecular analysis and MALDI-TOF mass spectrometry, obtaining a dispersity of 1.02 for the 2 kDa material and 1.01 for the 5, 10 and 20 kDa materials, which they took to agree with the supplier's gel-permeation figures of 1.05 to 1.10 [1]. Note what a band of that width means physically: by the identity above, 1.10 on a 20 kDa chain implies s of about 6,300 Da against 2,000 Da at 1.01 — a threefold difference in the width of the distribution, for a reagent described by one nominal mass.
Amine conjugation: many sites, one chemistry
The default route targets primary amines: the alpha-amine at the N-terminus and the epsilon-amines of lysine side chains. Activated esters — most commonly N-hydroxysuccinimidyl or succinimidyl esters carried on the PEG terminus — acylate those amines to give an amide. Two consequences follow immediately. First, the reaction consumes the amine's positive charge, so the conjugate is more acidic than the parent; Minshull and colleagues used exactly this to quantify extent of reaction, noting that NHS chemistry reacts with free primary amines with a resulting loss of charge and reduction in isoelectric point, and reading the distribution of PEGylation states off a denaturing capillary isoelectric focusing electropherogram as a bell-shaped curve [2].
Second, amine chemistry is not as narrow as the textbook picture. The same study found its method sensitive enough to resolve non-lysine adducts of PEG, including species conjugated to histidine residues [2] — a reminder that "amine-reactive" describes a preference, not a rule. Activated esters also hydrolyze in water in parallel with aminolysis, so reagent excess and reaction time are always traded against a competing background reaction. That is standard practice rather than a finding.
The bigger problem with amine chemistry is arithmetic. Cai and Yue put it plainly: random PEGylation usually results in product mixtures composed of mono-PEGylated isomers and multi-PEGylated attachments [3]. Take a peptide with four lysines. Counting the N-terminal alpha-amine there are five amine sites, and the reaction can populate all of them:
- five distinct mono-conjugates, all of the same mass
- ten di-conjugates, again all of one mass
- ten tri-, five tetra- and one penta-conjugate
- thirty-one distinct modified species in one pot
Species of different conjugation degree differ by whole PEG chains. Positional isomers of the same degree differ by nothing at all in mass, which is the case that defeats most of the analytical toolkit.
Steering to the N-terminus
The alpha-amine of the N-terminal residue and the epsilon-amine of lysine differ in basicity, the alpha-amine being the less basic of the two, which means that as pH falls the alpha-amine retains a usable fraction of free base after the lysines are largely protonated. That pKa gap is the lever for site selection with amine chemistry, and it is general textbook chemistry rather than a result attributable to a particular paper.
Lee and colleagues used it directly, conjugating PEG derivatives of 2,000 and 5,000 Da bearing a reactive terminal aldehyde group to epidermal growth factor at slightly acidic pH 5.5, targeting the alpha-amine at the N-terminus, and confirming that the mono-PEGylation had occurred there by tryptic digestion mapping combined with MALDI-TOF mass spectrometry [4]. The pattern — steer with pH, then prove the site by digest plus mass spectrometry rather than by inference — is the durable part of that work.
Two site-specific routes to the same protein make a useful comparison. Da Silva Freitas and colleagues PEGylated recombinant human growth hormone twice over: chemically, with a 20 kDa PEG-aldehyde on the alpha-amine of Phe1, giving a monoconjugate of 44,152.2 Da by MALDI-TOF; and enzymatically, with microbial transglutaminase and a 20 kDa PEG-NH2, which coupled only at Gln141, giving 44,064.9 Da [5]. The two conjugates differ by 87.3 Da — far less than the roughly 2,000 Da spread a 20 kDa chain carries at a dispersity of 1.01, the figure worked out above, which is a compact illustration of why an intact mass alone cannot tell you which site reacted. Circular dichroism showed both strategies preserved the native-like secondary structures of the protein [5].
Thiol conjugation: fewer sites, a less permanent bond
Cysteine offers the opposite trade. The thiol is a strong nucleophile in its thiolate form and free cysteines are comparatively scarce in a folded protein, so selectivity comes from scarcity rather than from tuning conditions. Maleimides are the usual partner, adding the thiol across the double bond to give a succinimidyl thioether.
Manjula and colleagues exploited a naturally reactive site this way, describing the unusually high reactivity of Cys-93(beta) of oxy hemoglobin toward maleimide and using maleidophenyl-PEG chains of 5, 10 and 20 kDa to build conjugates carrying two PEG copies per tetramer, isolated as homogeneous preparations by ion exchange chromatography [6]. That is what site-selective thiol chemistry buys: a countable number of products rather than a combinatorial mixture.
What it does not buy is a permanent linkage. Fontaine and colleagues measured the two competing fates of the adduct — disruptive cleavage by thiol exchange, and stabilizing ring-opening hydrolysis of the succinimide — across a series of N-substituted succinimide thioethers, and reported that ring-opening of conjugates prepared with commonly used maleimides was too slow to serve as prevention against thiol exchange, while ring-opening rates were greatly accelerated by electron-withdrawing N-substituents and the ring-opened products had half-lives of over two years [7]. Their proposal is that conjugates made with electron-withdrawing maleimides may be purposefully hydrolyzed to the ring-opened form before use, which is a deliberate extra step, not a property the chemistry gives you for free.
A second route changes the ring rather than the substituent. Gober and colleagues, working with peptides bearing an N-terminal cysteine, report that conjugation to a maleimide rearranges to a thiazine; their studies indicated the thiazine linker degrades markedly slower than the thioether conjugate across a broad pH range and is over 20 times less susceptible to glutathione adduct formation, and their NMR work confirmed the rearrangement is stereoselective, yielding a single diastereomer [8]. That last point is easy to overlook: the ordinary thiol-maleimide adduct creates a new stereocenter, so a "single" conjugate is generally a diastereomeric pair before polydispersity is considered at all.
What polydispersity does to characterization
Polydispersity is not merely an aesthetic problem. Mero and colleagues stated the analytical cost directly — a drawback of the polymer lies in its polydispersity, which makes identification of the sites of protein modification difficult — and worked around it by derivatizing with a monodisperse Boc-PEG-NH2 instead, which let the transglutaminase-modified glutamine residues in three model proteins be identified directly by electrospray ionization mass spectrometry and tandem MS [9]. Using a single-mass polymer to find the site, then the real reagent for the conjugate, is a legitimate and common strategy.
For intact analysis you are stuck with the distribution. Work the numbers for a 3,000 Da peptide carrying a 20 kDa PEG, total roughly 23,000 Da. Neighboring oligomers are 44.05 Da apart, so at charge state 20+ they appear 44.05 / 20 = 2.2 m/z apart. Adjacent charge states of a 23,000 Da species are separated by approximately 23,000 / (20 x 21) = 55 m/z. Only about 25 oligomer peaks fit between one charge state and the next — but the oligomer distribution is roughly 180 masses wide, so a single charge-state envelope spans about 400 m/z, some seven charge-state steps, and overlaps three or four neighbors on each side. That overlap, not sensitivity, is why intact PEG-conjugate spectra are hard to read.
The countermeasure is to reduce the charge. Bagal and colleagues reacted gas-phase superbases with electrospray ions in the source region of time-of-flight instruments and reported that charge stripping decreases the spectral congestion typically observed in ESI mass spectra of high molecular weight polydisperse PEGylated molecules, recovering average molecular weights and distributions; the average mass they measured for PEGylated rh-GCSF was 40,726.2 Da, and a 16 Da shift corresponding to an oxidized form was easily observed in the spectrum of that heterogeneous conjugate [10]. They also report that low-abundance free PEG in a PEGylated peptide preparation, not directly detectable by mass spectrometry, could be observed and accurately quantified once ion mobility separation was added [10]. Unreacted polymer is a real impurity class and an easy one to miss.
One bookkeeping point causes avoidable confusion: number-average, weight-average and monoisotopic mass are three different quantities, and for a polydisperse conjugate they do not coincide. A reported mass is uninterpretable unless the record says which one it is.
Separations behave differently too
Size-exclusion chromatography sorts by hydrodynamic volume, and PEG's contribution to that volume is out of proportion to its mass. Akbarzadehlaleh and colleagues PEGylated human serum albumin site-specifically with mPEG-maleimide and found a linear relation between the molecular weight of the attached PEG and the hydrodynamic radius of the conjugate, with no significant difference between the radii of proteins modified with linear and branched PEG [11]. Their practical conclusion is the one that bites: PEGylated albumin and the albumin dimer have almost the same size and elute at very close retention times, so ion exchange proved more effective than size exclusion here, and branched PEG-albumin eluted earlier than linear on anion exchange [11]. They also observed the conjugate to be smaller than the sum of the hydrodynamic radii of the native protein and the polymer, which they suggest could result from shielded attachment of the polymer around the protein — a hypothesis offered, not a demonstrated mechanism.
Gel electrophoresis misleads in a specific, correctable way. Kurfürst showed that PEG-modified protein can be stained directly after SDS-PAGE with barium iodide, which complexes the polymer and detects conjugate at concentrations not visible by Coomassie staining, and that molecular weight can be obtained by calibrating the gel with PEGs of different molecular weight, since PEG showed linear mobility under those conditions; using non-PEGylated proteins as standards gave incorrect values, because of the lower mobility of the PEGylated protein [12]. The albumin study saw the same effect, with branched PEG delaying migration more than linear [11].
Positional isomers are the hardest case, and they get harder as the polymer grows. Cai and Yue report that when peptides or proteins are PEGylated with polymer as large as 40 kDa, the hydrophobicity and molecular size of the isomers become too similar for the routine methods — reversed-phase HPLC, size exclusion, SDS-PAGE and capillary isoelectric focusing — to work, and they resolved the mono-PEGylated isomers of a peptide analogue by cation exchange instead [3]. The large polymer masks the small differences the other techniques depend on.
What PEGylation actually changes about the molecule
Size and solution behavior scale differently from each other. In the hemoglobin series, both the hydrodynamic volume and the molecular radius increased on surface decoration with PEG and exhibited a linear correlation with the mass of the PEG chain attached, while the viscosity and the colloidal osmotic pressure of the same conjugates increased exponentially with PEG chain length [6]. Doubling the polymer does not double the solution properties.
Charge changes only for amine chemistry: acylation of a primary amine removes a positive charge and lowers the isoelectric point [2], whereas thiol conjugation to a neutral cysteine leaves the charge ladder intact. Minshull and colleagues accordingly conclude that biopharmaceutical developers should consider capillary isoelectric focusing for extent-of-reaction monitoring and measurement of conjugates of free amine groups [2].
Conformational stability is the least settled of these. Lawrence and Price describe the impact of PEGylation on protein conformational stability as incompletely understood, and note that aside from avoiding active sites and binding interfaces there are few guidelines for selecting optimal PEGylation sites [13]. The growth-hormone work above found native-like secondary structure preserved at both sites tested [5], but one protein at two sites is not a rule.
What is not known
Two gaps are worth naming. Site selection remains largely empirical: beyond avoiding active sites and binding interfaces there are still few guidelines, and Lawrence and Price present structure-based guidelines for choosing stabilizing sites as work under development rather than a settled method [13]. Thiol-maleimide stability is an active area rather than a solved one, with ring-opening [7] and ring rearrangement [8] offered as competing fixes for the same exchange reaction, and no consensus that either is the general answer.
Those gaps set the questions worth asking of any PEGylated material: which site, established by what evidence; which linkage chemistry, and what is known about its persistence; and what dispersity, measured how. An intact mass and a purity percentage answer none of the three.