An 18.2 MΩ·cm reading on a water polisher and a bottle labelled HPLC grade are both quality statements, and neither is a statement of cleanliness in the sense most peptide work needs. Each is a narrow guarantee about a narrow class of contaminant. The things that quietly ruin a peptide experiment — a trace metal that oxidises a cysteine, a slip agent that perturbs a protein assay, a hydroperoxide that converts methionine to its sulfoxide while the vial sits on the bench — sit largely outside what either number covers.
Resistivity is an ionic measurement and nothing more
Conductivity in water is carried by mobile charged species; resistivity is its reciprocal. The reading therefore responds to dissolved ions and to essentially nothing else. An uncharged organic molecule contributes no current path and moves the meter not at all. What a maximum resistivity reading licenses you to conclude is that the ion-exchange beds were not exhausted at the point where the sensor sits. That is worth knowing. It is not a claim about a plasticiser, a recycled column-bleed product, or anything a bacterium excreted downstream of the sensor.
The organic burden has its own parameter, total organic carbon. Regnault and colleagues compared HPLC-grade bottled water, freshly produced UV185/254-treated water and freshly produced non-UV-treated water using electrospray ionisation MS detection, on the basis that both UV photodiode-array and MS detection are sensitive to organic contaminants and that water quality has a direct impact on achievable detection limits [1]. Two independent parameters, two independent failure modes: a system can sit at full resistivity while its TOC drifts upward, and the meter will not tell you. The operational answer is a blank gradient. Run the full method with no injection before you trust a small peak, and repeat it when you change water source.
Microbial load is a third axis, and it is not visible in either of the first two. Laurence and colleagues aligned 57 sequencing runs from six centres against the four largest NCBI BLAST databases and detected reads of diverse contaminant species in every run, identifying Bradyrhizobium as the most common contaminant genus; many of the organisms they found have been reported as contaminants of ultrapure water systems, and their recommended preventive measures include extensive UV irradiation of water [2]. That was a nucleic-acid workflow, where a handful of molecules is a detectable signal. The transferable point is that a purification loop is a habitat, and organisms living in it contribute both cells and the organic material they leave behind.
Endotoxin is a fourth, and the awkward one, because the measurement itself is matrix-dependent. Correa and colleagues describe low endotoxin recovery — the loss of detectable endotoxin activity over time in compendial Limulus amebocyte lysate assays when undiluted products are spiked with a known amount of endotoxin standard — as a documented behaviour of certain matrices commonly used for recombinant proteins, and studied LPS preparations with relevant detergents by acyl-chain phase transition, aggregate structure and size distribution to work out the biophysical parameters behind it [3]. A low LAL number obtained on a formulated sample does not on its own license the conclusion that endotoxin is absent; it licenses the conclusion that the assay did not detect it in that matrix.
Trace metals have the shortest path to your peptide
Water and buffer salts are the usual entry point, and the effect is not subtle at concentrations you would never think to specify. Munday and colleagues note that problems from adventitious metals in buffers and reagents are well recognised in metal-catalysed oxidation work, and their own result runs counter to the usual expectation: the rate of copper-catalysed oxidation of cysteine increased when buffer was purified with Chelex resin or when the iron chelator desferrioxamine was added, because adventitious iron was inhibiting the copper-catalysed reaction. In purified buffer at pH 7.25 containing 0.4 µM copper, cysteine oxidised at 32 µM/min; added iron produced a dose-related decrease of up to 85%, with a 50% decrease at an iron concentration of only 11 nM, while other transition metals had no effect [4]. Two cautions on transferring that number. The analyte was free cysteine; copper-catalysed oxidation of cysteamine and of reduced glutathione was much less sensitive to inhibition by iron in the same work. And the direction is the opposite of the reflex — here chelation raised the rate — so treating a Chelex step as automatically protective is unwarranted.
For an intact peptide, the demonstration is Huang and colleagues, who presented in vitro evidence that trace levels (≤0.8 µM) of zinc, copper and iron, present as common contaminants of laboratory buffers and culture media, are the actual initiators of the classic Aβ1-42-mediated seeding process and Aβ oligomerisation. Reproducing earlier experimental conditions, they found that the in vitro precipitation and amyloid formation of 20 µM Aβ1-40 initiated by 2 µM Aβ1-42 were abolished by chelation of trace metal contaminants, and that metal chelation attenuated formation of soluble oligomers from a cell-free culture medium; they suggest self-assembly in this system is not spontaneous as previously thought [5]. That is a concentration at which a metal is a contaminant rather than a component, and it was enough to change the qualitative outcome of the experiment.
Metals also act alongside oxidants rather than alone. Hemenway and colleagues monitored neat and aqueous PEG 400 at 40 °C and 50 °C and found that the presence of water increased the rate of formation of all measured reactive impurities, especially formic acid, as did the presence of hydrogen peroxide or trace iron metal [6]. Both accelerated the same degradation chemistry in that reagent, so the metal question and the peroxide question arrive together rather than as two separate checklist items.
Peroxides and the solvents that carry them
Ethers accumulate hydroperoxides on storage in contact with air — the reason ether bottles carry opening dates and the reason distilling from an aged bottle is treated as a hazard. The peptide-relevant consequence is quantified best in a polyether. Kumar and Kalonia showed that peroxide content in polyethylene glycols increased on storage, more so at higher temperature and with light exposure, and that vacuum drying at 0.1 mm Hg for 48 h at 25 °C gave a greater than 90% decrease in peroxide level in all cases except when high-peroxide-containing 25% w/v PEG 20000 or 50% w/v PEG 1450 solutions were dried, where the reductions were 88% and 52% respectively. Critically, oxidation of methionine in the peptide Met-Leu-Phe was significantly reduced when vacuum-dried PEGs were used [7]. The chain runs from storage condition, to reagent peroxide content, to a modified residue in a tripeptide.
The rate of that final step is measurable. Sjöberg and colleagues followed oxidation of the methionine residue of the tripeptide GlyMetGly by hydrogen peroxide using Raman spectroscopy and DFT, observing the appearance of the S=O vibration and modification of the C–S vibrations near 700 cm-1, and used changes in the intensity of the C–S stretching band to calculate a rate constant of 7.9 ± 0.6 × 10-3 dm3 mol-1 s-1 [8]. That constant belongs to the free tripeptide; the same work also determined the solvent-accessible surface area of the methionine residues of β-lactoglobulin and compared rate constants and energy barriers between tripeptide and protein, to get at the role of the molecular environment in the oxidation process.
Stabilisers are the other half of the solvent problem, and they are deliberate additions. Antioxidants including butylated hydroxytoluene and butylated hydroxyanisole were among those effective at limiting reactive impurity formation in PEG 400 [6], and the same compounds are used as stabilisers in peroxide-forming solvents. Their concentration is set by the stabiliser specification rather than by your experiment, so they can appear in a chromatogram or a mass spectrum as something you did not put in the tube. If you switch to an unstabilised solvent to remove the peak, you have taken on the peroxide problem the stabiliser was preventing. Record which stabiliser system the bottle carries; it is a method variable.
Leachables: the vessel is a reagent
McDonald and colleagues demonstrated that manufacturing agents in disposable plasticware — lubricating slip agents exemplified by oleamide, and cationic biocides included to prevent bacterial colonisation of the plastic surface — leach into a standard aqueous buffer, into dimethyl sulfoxide and into methanol, and can have profound effects on proteins and thus on results from bioassays of protein function [9]. The organic solvents matter here: the demonstrated scope covers DMSO and methanol, not only aqueous buffer, so a peptide stock held in DMSO in a plastic tube sits inside it.
The scale has since been mapped. Canez and Li found that contaminant profiles varied widely between microcentrifuge tubes from different manufacturers, and that the most suitable polypropylene tubes they tested still introduced 847 labware-originating contaminant m/z values across three manufacturing batches in Folch lipid extractions, including 21 primary amide and fatty acid surfactants identical to endogenous lipids, 16 of which had not been previously reported as polypropylene leachables. Borosilicate glassware with PTFE-lined screw caps introduced 98 contaminant m/z values over three batches, and current databases and literature allowed identification of only 32 of the contaminants found [10]. That was a lipidomics workflow, and the species involved are lipid-like rather than peptide-like — but the glass-versus-plastic ratio, the batch-to-batch variability and the identification gap are properties of the labware, not of the analyte.
HPLC grade and MS grade answer different questions
A solvent specification written for chromatography is largely about what the solvent does to a baseline: UV transmission, non-volatile residue, gradient response. An MS-grade specification addresses a different quantity — what the solvent contributes to the background ion population — and a specification written in bulk terms does not predict which ions an ion source will register. Keller and colleagues reviewed reported contaminants and introduced further interferences identified in their own laboratories over a decade, covering both proteinaceous and non-proteinaceous compounds, and supplied a searchable ion list; they also make the constructive point that knowing a background ion's identity enables its use as an internal calibrant for accurate mass measurement [11].
Grade labels are also not interchangeable between suppliers. Cajka and colleagues investigated how LC-MS-grade isopropanol from different vendors may influence the quality of the mobile phase in untargeted lipidomic profiling, and separately reported an unusual but highly abundant ethylamine adduct of the form M + 46.0651 forming for specific lipid subclasses in positive electrospray when acetonitrile is part of the mobile phase, with potential for misidentification [12]. The adduct case is the sharper lesson: nothing there is an impurity at all, and the artefact still appears.
Metal in the flow path, not only in the bottle
Metal exposure does not end at the buffer. DeLano and colleagues attribute peak tailing and loss of analyte signal in HPLC to adsorption of certain analytes on the metal oxide layer of metal components, and characterised an ethylene-bridged hybrid organic–inorganic surface applied to UHPLC instruments and columns, demonstrating improved performance for separations of nucleotides, a phosphopeptide and an oligonucleotide, with higher and more consistent peak areas particularly at low analyte mass loads and acidic mobile phases; they also observed reduced abundances of iron adducts in the mass spectrum of a peptide [13].
Gilar and colleagues localised part of the effect to the frits. Making repetitive injections of acidic small molecules or an oligonucleotide onto individual column frits, they saw losses on stainless steel and titanium, with adsorption greatest at acidic pH because of the positive charge on the metal oxide surface, and nearly complete recovery once the metal adsorptive sites were saturated with the analyte. Conditioning the frits with phosphoric, citric or etidronic acid achieved a similar effect, but that effect is temporary because the conditioning agent is gradually removed by the running mobile phase [14]. Note the analytes: acidic small molecules and a 25-mer phosphorothioate oligonucleotide, not peptides. What transfers is the mechanism and the caveat that a conditioning step is a decaying fix.
Containers, and the acid-wash reflex
Storage vessels are usually chosen by habit. Reimann and colleagues tested that habit directly, collecting groundwater samples in parallel into five clear plastic bottle types — HDPE, polypropene from two manufacturers, FEP and PFA — with polyethene and polypropene sampled into both factory-new unwashed and acid-washed bottles, and analysing 62 elements by ICP-MS down to the ng/L range. Acid-washing of factory-new flasks had no clear systematic beneficial effect on analytical result; for the PP bottles, concentrations of Pb and Sn were clearly elevated in the acid-washed bottles. For the vast majority of elements bottle type was of no importance, with systematic differences tentatively discernible for six (Al, Cr, Hf, Hg, Pb, Sn), the most pronounced being clearly elevated Cr from HDPE bottles regardless of washing; for those six, FEP or PFA seemed marginally preferable [15]. That work is about ultra-trace inorganic analysis of groundwater, not about peptide storage, and it does not license skipping cleaning where organic residue is the concern. It does mean a cleaning step should be justified by a measurement rather than assumed to help.
What to specify and what to record
Water and solvent quality behaves like an assay panel, not a single score. Resistivity covers ions at the sensor. TOC covers the organic burden that sets UV and MS detection limits [1]. Bacterial and endotoxin limits are separate again, and an endotoxin figure is conditional on the matrix it was measured in [3]. Trace metals are typically outside all of those parameters at the concentrations that matter to a cysteine or a methionine [4, 5]. Peroxide content is a function of a bottle's age and storage history rather than its grade [7].
The corresponding laboratory record is short: water source and the date of the last cartridge change; solvent grade, supplier, lot, stabiliser system and opening date; vessel material and whether it was cleaned; whether the buffer was metal-treated. Those entries are what let you separate a real result from a reagent artefact six months later. They also feed the two downstream questions this library treats separately — elemental impurities, where trace metals are quantified against defined limits, and residual solvents, where the solvent that touched the peptide has to be accounted for in the final material. Both of those assessments inherit whatever came in through the water and the bottle.