Protein Polarimetry, Perfected: Specific Rotation Measurement for the Refracto-Polarimetric Detection of Cryptic Protein Denaturation

Abstract

Protein polarimetry has been evaluated as a simple and straightforward technique to detect the cryptic denaturation of exemplary proteins. The general rules of rotation vs. amino acid and structural composition and the respective knowledge gaps were reviewed, and the specific rotation of cystine was determined in 4 M NaCl solution as [α]_D²⁰ = –302.5°. The specific rotations at 589 nm and 436 nm and the ratio were measured for several model proteins, some purified plasma-derived proteins and for three monoclonal antibodies. The immunoglobulin G concentrates all showed a narrow ratio range likely characteristic for this protein class. Heat denaturation experiments were conducted at temperatures between 50 and 85 °C both for short-time (10 min) and for prolonged periods of heat exposure (up to 210 min). Denaturation by heat resulted not only in the known levorotatory shift, but also in a shift in the specific rotation ratio. The stabilizing effect of fatty acids in bovine serum could be demonstrated by this parameter. Polarimetry thus appears to be a particularly sensitive and simple method for the characterization of the identity and the thermal stability of proteins and should therefore be added again as a complimentary method to the toolbox of protein chemistry.

1. Introduction

Protein polarimetry, as a classical optical technique and the first to detect protein denaturation, has long been superseded by the refined chiroptical method of circular dichroism (CD) spectroscopy in the ultraviolet spectral region [1]. Among such techniques to assess and analyze the storage stability of therapeutic proteins in liquid formulations [2,3,4], protein polarimetry conveniently does not require sample dilution, so that protein solutions with a concentration determined, e.g., by refractometry [5], can be measured with a minimal sample preparation workload.

The specific rotation and the rotatory dispersion of a peptide or protein macromolecule is determined by the primary structure (amino acid sequence), the secondary and tertiary structure (disulfide bridges, α-helical, β-sheet, β-turn and random coil sections), and by the glycosylation. A conclusive set of rules has not yet been established, as the research in this field of classical polarimetry was essentially abandoned in the 1960s, when circular dichroism spectroscopy was introduced, while fractionation, crystallization, crystallography, vibrational spectroscopy, amino acid analysis and partial sequencing still required considerable efforts. Protein purification by chromatography had not been established then; monoclonal antibodies were yet to be introduced, as well as recombinant protein biosynthesis for research and therapeutics.

The levorotation of purified proteins had already been investigated in the 19th century (as compiled by [6,7]), when serum albumin could be obtained by ammonium sulfate fractionation from blood serum, and fibrinogen by sodium chloride precipitation from anticoagulated blood plasma. Among the plant proteins, the cereal prolamins extractable in aqueous alcohol and soluble in dilute acetic acid show a high specific levorotation, about twice as high as plant and animal globular proteins [8].

The specific rotation [α]_l^T is expressed as ° arc/(g/mL) in a 10 cm tube and depends on the wavelength λ and the temperature T, on the solvent, and likely also on the concentration. Historically, the yellow sodium flame filtered through a solution of potassium dichromate provided a convenient and intense monochromatic light source for visual detection [6,7]. For this reason, the specific rotation [α]_D at the Na D doublet wavelength (589.44 nm) is most common in the literature; other wavelengths, such as 435.95 (g-line) or 546.22 nm (e-line) from the low-pressure mercury arc, have rather seldom been reported for proteins. The actual rotation angles measured are, however, small and require a resolution of at least 0.01°. In exploratory experiments, the suitability of this technique to detect small “cryptic” structural perturbations was examined with monoclonal antibodies, model proteins and bovine serum albumin.

Current State of Protein Polarimetry: Instrumentation and Data

Photoelectric “circular polarimeters” (from German “Kreispolarimeter”) with an analyzer prism driven by a digitally controlled micro-stepper motor enable a measurement resolution of at least or better than 0.001°; for the measurement of small angles up to 1°, Faraday modulation may improve this resolution by an additional half decade (to 0.0003°). This measurement is greatly simplified by the location of the detector output’s absolute minimum [9]. With electronic thermostatization by built-in Peltier elements, the temperature can be controlled to an accuracy of at least ±0.1 °C. Such high-resolution digital polarimeters (0.0001° display and 0.0003–0.0008° measurement resolution) with accurate and stable Peltier thermostatization (0.01 °C) are nowadays available from several manufacturers (Schmidt & Haensch, Berlin, Germany; Rudolph Research, Hackettstown NJ, USA; Jasco, Tokyo, Japan; Bellingham & Stanley, Turnbridge Wells, United Kingdom; Anton Paar, Graz, Austria), with thermostated polarimeter tubes (cuvettes) and traceably certified calibrated quartz plates as intrinsic reference standards. These instruments may offer several wavelengths from the Hg spectrum (365 nm, 405 nm, 436 nm, 546 nm) and the common Na wavelength (589 nm), with gas discharge lamps (e.g., Schmidt & Haensch, Jasco), incandescent tungsten halogen lamps (Rudolph Research, Bellingham & Stanley) or LEDs (Anton Paar) with narrowband interference filters as light sources. The cuvettes require considerably smaller sample volumes (0.5–5 mL) than the original glass tubes (10–20 mL), and the sample may be recovered after measurement for further analysis. Tinted or turbid samples with a residual transmission of approx. 1% can still be measured with the photoelectric detector with sufficient precision, as specified by the manufacturers. In contrast to such circular polarimeters (which measure only linear polarization) with fixed wavelengths, continuous spectropolarimeters for circular dichroism measurements are very bulky and expensive instruments [10,11].

The instrument’s accuracy is verified and re-calibrated with dextro- and levorotatory quartz crystal plates, which must be cut and polished to strict specifications and tight tolerances in flatness, thickness, parallelism, and optical axis error (OIML R14 E95). They are usually calibrated at the green ¹⁹⁸Hg isotope emission wavelength of 546.2271 nm, from which all other certified wavelengths are then calculated from a rotatory dispersion formula [12]. The specific rotation of quartz (in ° arc/mm) has been measured from the far ultraviolet to the near infrared at a reference temperature of 20 °C; in the range between 407.899 and 882.636 nm, the temperature effect has been calibrated in the range from 18 to 30 °C.

Protein concentration can be measured gravimetrically by drying and weighing, by Kjeldahl digestion or Dumas combustion with nitrogen determination (with the correct conversion factor), by refractometry (with the specific refractive index increment dn/dc calculated from the amino acid and glycan composition), or by UV absorbance measurement (with the correct extinction coefficient corrected for the glycan mass fraction). Digital refractometers, inherently insensitive to slight turbidity and opalescence, enable a measurement with up to 0.00001 n_D resolution and 0.00002 n_D accuracy corresponding to 0.1 mg protein/mL.

For polarimetric measurements, protein solutions should be as clear as possible, since turbidity causes a depolarization [13] and attenuates the intensity depending on the wavelength. Slightly turbid protein solutions may be clarified by high-speed centrifugation; opalescence may, however, be intrinsic due to the protein’s molecular weight, shape and size, as, e.g., observed with fibrinogen solutions. The overall rotation of a protein solution is a sum of the components’ rotation; amino acids and sugars in the protein-free matrix, e.g., in the formulation buffer of a purified therapeutic protein, will shift the sum value but can be subtracted if the matrix can be formulated or isolated and measured separately. Absorbing ligands (e.g., heme and bilirubin in albumin) can lead to an anomalous rotation dispersion [14]. Kosmotropes such as potassium bromide may stabilize the protein structure in solution [15], but require a correction for the refractive index.

Data from the literature on the specific rotation of amino acids and of proteins at the Na D-doublet wavelength (589.3 nm) are given in

Table 1. Data from the literature on specific rotations on constituent amino acids.

Amino Acid	Solvent	T (°C)	[α]D (°)	Reference
l-alanine	water	22	+2.7	[16]
l-arginine	water	20	+12.5	[16]
l-asparagine	water	20	–5.6	[16]
l-aspartic acid	water	18	+4.7	[16]
l-cysteine	water	23	–10.4	[16]
l-cystine	4 M NaCl	25	–302.5	this work
l-glutamic acid	water	18	+11.5	[16]
l-glutamine	water	25	+6.5	[17]
l-histidine	water	25	–39.0	[16]
l-hydroxyproline	water	22.5	–75.2	[16]
l-isoleucine	water	20	+11.3	[16]
l-leucine	water	24.7	–10.8	[16]
l-lysine	water	20	+14.6	[16]
l-methionine	water	25	–8.1	[16]
l-phenylalanine	water	20	–35.1	[16]
l-proline	water	23.4	–85.0	[16]
l-serine	water	20	–6.8	[16]
l-threonine	water	26	+28.4	[16]
l-tryptophan	water	22.7	–31.5	[16]
l-tyrosine	water	18	–13.2	[16]
l-valine	water	20	+6.4	[16]

and

Table A1. Data from the literature on specific rotations on human and animal proteins and on plant seed and cereal proteins.

Protein	Solvent	T/°C	[α]D/°	Ref.
serum α-glycoprotein		25	−23	[59,60]
orosomucoid	water, pH = 5.8	25	−21.8–−22.2	[61]
colostrum M−1 glycoprotein (human)		20	−48	[62]
ovalbumin (chicken)	water		−35.5	[31]
ovalbumin (chicken)			−25.2–−42.8	[63]
ovalbumin (chicken)			−28.14–−39.23	[64]
ovalbumin (chicken)		15–17	−30.7	[65]
ovalbumin (chicken)			−28.60–−29.53	[66]
ovalbumin (chicken)			−30.3–−31.5	[67]
ovalbumin (chicken)			−30.1	[68]
ovalbumin (chicken)	1% in water	20	−37	[69]
ovalbumin (chicken)		25	−30	[59,60]
conalbumin (chicken)			−36–−39	[66]
ovomucoid (chicken)			−61.10–−61.38	[66]
lactalbumin (bovine)	water		−36.6–−38.0	[70]
β-lactoglobulin (bovine)		25	−32	[59,60]
chymotrypsinogen	0.1 M acetic acid		−76	[71]
chymotrypsinogen (bovine)	0.2 M NaCl, pH = 6.1	0	−78	[72]
chymotrypsinogen	0.1 M KCl, pH = 7	25	−73.2	[73]
transferrin	pH = 7.6	25	−44.5	[61]
transferrin			−53	[74]
fibrinogen (equine)	NaCl		−52.5	[75]
fibrinogen (equine)			−50.5	[76]
fibrinogen (bovine)		25	−49.4	[77]
globin (bovine) (apo-hemoglobin)	water		−13	[21]
globin (equine) (apo-hemoglobin)	water		−18	[21]
γ-globulin	buffer, pH = 7.8	25	−51.2	[78]
γ-globulin	glycine buffer, Γ/2 = 0.1	25	−45–−52	[61]
γ-globulin	0.15 M NaCl	22	−52.1–−54.6	[79]
serum albumin	water		−56	[31]
serum albumin	sat, NaCl		−64	[31]
serum albumin			−62.6–−64.4	[6,7]
serum albumin	3rd recrystallization		−62.6	[70]
serum albumin	further purified		−60.1	[70]
serum albumin	crystallized		−61–−61.2; −64	[80]
serum albumin (equine)	crystallized	18	−62.8	[68]
serum albumin (equine)	ammonium sulfate-fractionated, 1% in H2O	20	−54	[69]
serum albumin (bovine)	ammonium sulfate-fractionated	20.5	−63.0	[81]
serum albumin (human)	water, pH = 5.1	26	−62.2	[82]
serum albumin (human)		25–26	−64.5	[83]
serum albumin (human)		25	−63	[59,60]
clupein (herring sperm)		25	−99.7	[84]
collagen (carp swim bladder)	citrate buffer, pH = 3.7	10	−350 ± 30	[22]
collagen (carp swim bladder)	1 M acetic acid		−357	[25]
collagen (rabbit skin)	0.134 M phosphate buffer, pH = 7.4	0	−377	[23,24]
glutenin (wheat)	0.1 M acetic acid	27	−102	[85]
gliadin (wheat)	dilute acetic acid		−111	[8]
gliadin (wheat)	0.01 M acetic acid	27	−114	[85]
hordein (barley)	dilute acetic acid		−130	[8]
secalin (rye)	dilute acetic acid		−144	[8]
edestin (hemp seed)	crystallized in 10% NaCl		−41.3	[86]
excelsin (Brazil nut kernel)	crystallized in 10% NaCl		−42.94	[86]
globulin (flax seed)	crystallized in 10% NaCl		−43.53	[86]
globulin (squash seed)	crystallized in 10% NaCl		−38.73	[86]
amandin (almond kernel)	10% NaCl		−56.44	[86]
corylin (hazelnut kernel)	10% NaCl		−43.09	[86]
globulin (English walnut kernel)	10% NaCl		−45.21	[86]
globulin (American black walnut kernel)	10% NaCl		−44.43	[86]
phaseolin (horse bean seed)	crystallized in 10% NaCl		−41.46	[86]
legumin (kidney bean seed)	10% NaCl		−44.09	[86]

, respectively.

Rotatory dispersion data of ammonium sulfate fractions obtained from alcohol-delipidated plasma [18] were tentatively interpolated with a 3rd order polynomial regression to calculate the value at 589.3 nm and the ratio between the rotations at 436 and at 589 nm. The light at 660 nm was isolated from a Nernst lamp with a crystal violet filter and may most likely not have been strictly monochromatic (

Table A2. Rotation dispersion data from [18].

λ/nm	Albumin (°)	“Pseudoglobulin” (°)	“Euglobulin” (°)
435.9	−151.5	−118.2	−105.2
546.1	−78.1	−68.7	−61.2
578	−67.0	−59.8	−54.0
660	−47.3	−43.1	−39.4
589.3 (calc.)	−63.5	−52.5	−57.6
ratio 436/589	2.38	2.25	1.83

).

The following factors have been elucidated to contribute to the total specific rotation of a protein [19]:

(i)

Strongly levorotatory amino acids such as proline ([α]_D = −85°) and cystine (disulfide bridges, [α]_D = −265° in phosphate buffer) [20] shift the value to the left, as shown for bovine globin (α-helical, no cystine) and equine globin (1 cystine) [21].

(ii)

Right-handed α-helices are dextrorotatory, β-sheets and -turns levorotatory; proline breaks the structure and induces a turn; left-handed polyproline helical structures as in collagen [22,23,24,25] and in proline-rich and S-poor prolamins [8,26,27,28] shift the rotation to the left.

(iii)

Ligands (such as bilirubin bound to albumin) may shift the rotation depending on the amount present [29].

(iv)

Glycans will most likely shift the rotation to the right (as evident for α-glycoprotein and orosomucoid).

(v)

The denaturation of a globular protein, both reversibly by chaotropic agents [29,30] or irreversibly by acid or alkali [31], organic solvents [21], disulfide reduction [20], radiolysis [32,33], UV photolysis [30,33,34,35,36], heat [30,37,38,39], heat and acid or alkali [40], or pressure [41], shifts the rotation; in the case of reduction, to the right [20], with all other mechanisms to the left, to about −100° at most.

Recently, a concept of “optical calorimetry” has been proposed based on a laser polarimeter, demonstrating a denaturation temperature measurement from a rotation thermogram [42]. Gelatin, however, shows a temperature-dependent specific rotation in the native state caused by the coil-to-helix transition [43].

This work aims to bridge historical insights with contemporary analytical capabilities, thereby reintroducing polarimetry as a valuable, complementary tool in the biophysical characterization of therapeutic proteins. We systematically investigated the specific rotation and rotatory dispersion of various model proteins, plasma-derived proteins and monoclonal antibodies. We further examined the effects of thermal denaturation and formulation excipients on these optical parameters. Our findings suggest that polarimetry can provide valuable complementary insights into protein conformation and stability.

2. Materials and Methods

2.1. Polarimetry

Rotations were measured at 20 or 25 °C with a Peltier-thermostated digital polarimeter (Rudolph Research Autopol IV with 589 (Na D-doublet, α_D²⁵) and 436 nm (Hg g-line, α_g²⁵) wavelengths, 0.0001° display and 0.0007–0.0008° measurement resolution) in thermostatable 2, 5, or 10 cm cuvettes (Rudolph Research Temp-Trol). The polarimeter was verified and calibrated, if necessary, with NIST-traceable certified calibrated dextrorotatory (α_D²⁰ = +11.507°) single and levorotatory double (α_D²⁰ = −1.158°) quartz plates. The polarimeter cuvette lengths were referenced with a saccharimetric standard sucrose solution (26 g/100 mL at 20 °C, 23.7 mass-%) on the 10 cm cuvette as unity. Results from the 2 cm cuvette must be multiplied by 4.9953, from the 5 cm cuvette by 2.001, and from the 20 cm cuvette by 0.50007.

2.2. Refractometry

Protein concentrations were measured with a digital refractometer (Rudolph Research J357, 0.00001 n_D resolution) at 20 °C. The overall instrumental accuracy of 0.00002 n_D as specified by the manufacturer corresponds to a protein concentration uncertainty of (±) 0.1 mg mL⁻¹. Verification with aqueous sucrose solutions as defined in the OIML R142 verification guideline confirmed this specification [44]. For day-to-day stability, the accuracy of the refractometer was validated with water and 80 mass-% aqueous ethanol (n_D²⁰ = 1.36495 ± 0.00005).

A differential technique based on the measurement of the protein solution and a protein-free permeate obtained by centrifugal ultrafiltration [4] enabled the subtraction of the protein-free matrix refractive index. The specific refractive index increments dn/dc were either calculated from the amino acid composition by the SEDFIT software (version 16.50) [45], or they had been calibrated experimentally previously either in-house as for albumin, α₁-proteinase inhibitor, and IgG, or externally as for the NIST BSA reference material SRM 926.

2.3. Analysis of Cystine

l-Cystine from a non-animal source was obtained as a powder (Sigma-Aldrich C7602, St. Louis, MO, USA). A solution containing 200 mg L⁻¹ in 4 M NaCl [46] was prepared by dissolving the powder with gentle microwave heating and ultrasonication for measurement in the 20 cm cuvette.

2.4. Analysis of Protein Solutions

From protein solutions, the protein-free matrix was separated by centrifugal ultrafiltration (Millipore Amicon Ultra-4 10 kDa tubes, Darmstadt, Germany) in a refrigerated swinging bucket centrifuge (Thermo Heraeus Multifuge 1S, Thermo Electron LED GmbH, Osterode am Harz, Germany) at 4500 rpm. To avoid a Donnan shift, only a small permeate volume (~150–250 µL) was obtained from 4 mL protein solution for refractometry, which then was recovered; for polarimetry, which required at least 0.55 mL in the 5 cm cuvette, the centrifugation was continued, and the rotation angle corrected by the refractive index ratio between the refractometry and the polarimetry permeate. The following samples were investigated:

• A set of 14 structural model proteins [47];
• A set of purified proteins:

  o

  Human serum albumin ultra-diafiltration concentrate from fractionation by cold ethanol (citrate-depleted by addition of caprylate before the ultra-diafiltration step, before pasteurization with no further stabilizers added);

  o

  Recombinant human serum albumin from rice (Sigma-Aldrich, solid powder);

  o

  Bovine serum albumin (Carl Roth, solid powder);

  o

  Bovine serum albumin, defatted (Sigma-Aldrich, solid powder);

  o

  α₁-proteinase inhibitor, plasma-derived by cold ethanol fractionation and chromatographic purification, formulated in phosphate-buffered saline (concentrated by centrifugal ultrafiltration);

  o

  Immunoglobulin G, plasma-derived by cold ethanol fractionation and chromatographic purification, formulated in 0.25 M glycine;

  o

  Immunoglobulin G, plasma-derived by cold ethanol fractionation and chromatographic purification, formulated in 0.25 M proline (specified);

  o

  mAb 1, monoclonal IgG₁ antibody in a citrate-phosphate-buffered saline formulation containing 45 mM histidine;

  o

  mAb 2, monoclonal IgG₁ antibody in a buffered formulation containing 50 mM histidine, 125 mM arginine and 5 mM methionine;

  o

  mAb 3, monoclonal IgG₁ antibody formulated in 0.25 M glycine, pH = 5.

The monoclonal antibodies were purified from recombinant cell cultures by the common process chromatographic techniques. Ovalbumin, lysozyme, chymotrypsino-gen A, concanavalin A, glucose oxidase and plasma-derived human serum albumin solutions with a sufficiently high protein concentration (>80 mg mL⁻¹) were measured in the 10 cm polarimeter cuvette, all other samples in the 5 cm cuvette. The optically active formulation excipients histidine, arginine, methionine and proline are optically active and required correction for the matrix rotation.

2.5. Thermal Denaturation Measurements

The short-time heat denaturation experiment of mAb 3 was carried out in the original vials containing 2 mL of the 5% protein solution each, which were immersed into a thermostated water bath (Haake F/Haake S3, Haake, Karlsruhe, Germany) at temperatures of 50, 60, 65, 70, 75, 80, and 85 °C for 10–11 min and then chilled in an ice/water bath. The stable rotation of these protein solutions at 20 °C was then measured in the 5 cm cuvette at 436 nm and 589 nm.

For the long-time denaturation experiments, the solutions of BSA with fatty acid (50.0 mg mL⁻¹) and defatted BSA (51.0 mg mL⁻¹) at 60 °C, and mAb 3 at 70, 75, 80 and 85 °C) were heated in borosilicate glass tubes for NMR spectroscopy (Wilmad, Vineland, NJ, USA, 17.8 (L) × 0.5 (OD)/0.4 (ID) cm, 2 × 1.25 mL solution in duplicate) immersed into a thermostated water bath (Haake F3) for defined times (5, 10, 20, 30, 45, 60, 90, 150 and 210 min) and chilled in an ice-water bath immediately afterwards. The stable rotation was measured in the 10 cm cuvette at 25 °C.

3. Results and Discussion

3.1. Analysis of Cystine

The specific rotation of l-cystine under isoelectric conditions (−302.5°, see Table 1) is still lower than the value obtained in 0.0156 M HCl (−270°) [48]. The particularly low solubility of l-cystine in the physiological pH range (~110 mg L⁻¹ in water at 25 °C) [49,50] may be slightly enhanced with neutral salts such as NaCl or Na₂SO₄ [46,51], but requires such a high instrumental resolution and accuracy as possible only with digital polarimeters so that the actual rotation (in this case only −0.1210° in the 20 cm polarimeter cuvette at 589 nm for a solution containing 200 mg L⁻¹ in 4 M NaCl) can be measured at least with a relative instrumental uncertainty of ~±1%.

3.2. Analysis of Proteins

The model proteins showed the α-helix-dependent specific rotation ratio as described previously [47]. As the hemoproteins myoglobin and hemoglobin preclude any rotation measurement, any determination of the rotation ratio would require a delicate purification of the apoproteins [37]. For lipoxidase and β-galactosidase, the centrifugal ultrafiltration permeates showed a rotation indicating the presence of optically active formulation reagents such as sugars (+0.7530° and +0.3897° at 436 and 589 nm, resp, for β-galactosidase) or amino acids (−0.0390° and–0.0183°, resp., for lipoxidase) as measured in the 10 cm polarimeter tube. The correction calculation is the same as described below for the proline-formulated IgG.

Among the purified proteins investigated, both the polyclonal plasma-derived immunoglobulin G and the IgG monoclonal antibodies revealed a close range of the rotation ratio likely characteristic for this protein class. The proline-formulated IgG concentrate showed a rotation at 25 °C of −13.1410° and −6.3856° at 436 and 589 nm, resp., in the 5 cm cuvette. The rotation of the proline had therefore to be subtracted. 0.8 mL permeate was finally obtained with a refractive index n_D²⁰ = 1.33936 and a rotation of −3.1162° and −1.5501°at 436 and 589 nm, resp., in the 5 cm cuvette. By correction to the initial permeate refractive index of 1.33925, these values were multiplied the ratio of the net refractive indices (against water) of 0.00627/0.00637 = 0.98273 to −3.0624° and −1.5233°. The net rotation of the protein in the 5 cm cuvette was thus calculated as −13.1410–−3.0624° = −10.0786° and −6.3856–−1.5233° = −4.8623° at 436 and 589 nm, respectively. With the proline specific rotation of [α]_D = −85° (23.7 °C), the permeate rotation of −3.0481° (589 nm, 10 cm) corresponds to a concentration of 35.86 g L⁻¹, i.e., 0.311 mol L⁻¹.

The mAb 1 formulation buffer showed a weak levorotation (−0.3913° and −0.2011° at 436 and 589 nm, resp., and 10 cm path length). This value had to be subtracted from the mAb 1 protein solution (−8.3708° and −4.0791° at 436 and 589 nm, resp., in the 5 cm cuvette). The mAb 2 formulation buffer showed a weak dextrorotation (0.1821° and 0.4299° at 436 and 589 nm, resp.).

The results are listed in

Table 2. Specific rotation values of structural model proteins (more information in Table S1).

Model Protein	Origin	c/mg mL−1	αg25, 10 cm	αD25, 10 cm	[α]g25	[α]D25	[α]g/[α]D
albumin, bovine	serum	83.2	−12.5121	−5.3187	−150.4	−64.0	2.35
albumin, human	serum	94.9	−13.7865	−5.9168	−145.3	−62.3	2.33
lipoxidase *	soybean	33.0	−2.4092	−1.0383	−73.0	−31.5	2.32
ovalbumin	chicken	97.8	−4.7096	−2.0747	−48.2	−21.2	2.27
lysozyme	chicken	86.5	−5.0279	−2.2232	−58.1	−25.7	2.26
conalbumin	chicken	42.1	−3.5420	−1.5780	−84.0	−37.4	2.24
β-galactosidase	AO	25.9	−2.0896	−0.9403	−79.4	−35.7	2.22
β-lactoglobulin	bovine	53.3	−4.0282	−1.4858	−83.9	−30.9	2.17
chymotrypsinogen A	bovine	82.5	−7.0018	−3.2379	−84.9	−39.2	2.17
ribonuclease A	bovine	39.6	−6.3650	−2.9525	−160.8	−74.6	2.16
concanavalin A	JB lectin	8.9	−0.7487	−0.3480	−84.1	−39.1	2.15
pepsin A	porcine	31.4	−7.0599	−3.3931	−224.9	−105.8	2.13
peroxidase *	HR	37.6	n. m.	−2.3700	---	−63.0	---
glucose oxidase **	AN	78.8	n. m.	−0.3425	---	−8.8	---

AN = Aspergillus niger; AO = Aspergillus oryzae; HR = horseradish (Armoracia rusticana); JB = Jack bean (Canavalia ensiformis); n. m. = not measurable; * hemoprotein; ** flavoprotein; (cytochrome c, hemoglobin and myoglobin could neither be measured at 589 nor at 436 nm, due to intense absorption in this spectral region by the heme prosthetic groups).

and

Table 3. Specific rotation of purified proteins (more information in Table S2).

Purified Protein	Origin	Formulation	c/mg mL−1	αg25, 10 cm	αD25, 10 cm	[α]g25	[α]D25	[α]g/[α]D
albumin, human *	plasma	water	50.1	−7.4819	−3.1911	−149.34	−63.7	2.34
recombinant albumin **	rice	water	96.2	−13.4413	−5.7467	−139.72	−59.7	2.34
albumin, bovine ***	serum	water	68.7	−9.5616	−4.0666	−139.18	−59.2	2.35
α1-proteinase inhibitor	plasma	PBS	56.6	−5.4227	−2.4168	−95.81	−42.7	2.24
immunoglobulin G	plasma	glycine	99.8	−10.7884	−5.2028	−108.10	−52.1	2.07
immunoglobulin G	plasma	Gly, NaCl	162.2	−16.8216	−8.1223	−103.71	−50.1	2.07
immunoglobulin G	plasma	proline	196.3	−20.1673	−9.7295	−102.74	−49.6	2.07
mAb 1 ****		(His)	164.8	−16.3587	−7.9513	−99.26	−48.2	2.06
mAb 2 ****		(His, Arg)	50.7	−5.9973	−2.8873	−118.29	−56.9	2.08
mAb 3 *****			52.9	−6.0786	−2.9253	−114.91	−55.3	2.08

* cold ethanol precipitation-fractionated, bilirubin-depleted by ethanol; ** human serum albumin from transgenic rice, contains heme, but no bilirubin; *** heat precipitation-fractionated; **** dn/dc calculated with SEDFIT for monoclonal antibodies; ***** measurement performed at 20 °C.

. The ratio between the specific rotations at 436 and 589 nm clearly differs between α-helical proteins (albumin, α₁-proteinase inhibitor = antitrypsin) and the β-structure-dominated immunoglobulins. A contribution of the albumin-bound bilirubin (or of porphyrins) to an anomalous rotation dispersion around 436 nm may not be ruled out [14].

3.3. Short-Time Thermal Denaturation Measurements

The short-time heat denaturation experiment with the glycine-formulated mAb 3 showed a temperature-dependent increase of the specific rotation and of the ratio between the values at 436 and 589 nm, as summarized in

Table 4. Short-time heat denaturation experiments with mAb 3 (more information in Table S3).

T/°C	Duration	c */mg mL−1	αg20, 5 cm	αD20, 5 cm	[α]g20	[α]D20	[α]g20 rel.	[α]D20 rel.	[α]g/[α]D
37	native	52.9	−3.0378	−1.4619	−114.91	−55.3	1.0000	1.0000	2.078
50	10 min	52.8	−3.0420	−1.4627	−115.28	−55.4	1.0033	1.0024	2.080
60	10 min	52.9	−3.0465	−1.4636	−115.24	−55.4	1.0029	1.0012	2.082
65	10 min	52.8	−3.0510	−1.4679	−115.63	−55.6	1.0062	1.0060	2.078
70	11 min	52.8	−3.0836	−1.4828	−116.86	−56.2	1.0170	1.0162	2.080
75	10 min	52.8	−3.2812	−1.5738	−124.35	−59.6	1.0822	1.0786	2.085
80	10 min	52.8	−4.0611	−1.9369	−153.91	−73.4	1.3394	1.3274	2.097
85	10 min	52.8	−5.0107	−2.3746	−189.89	−90.0	1.6526	1.6274	2.110

* dn/dc = 0.000189.

.

This observation agrees with the incremental short-time heat effect on BSA solutions measured visually at 589 nm and shown only in a graph [38], but photoelectric polarimetry [9] at 436 nm (or even 365 nm) appears to enable an about twice (or three-times) as sensitive detection of early “cryptic” structural perturbations. It is, however, noteworthy that the protein denaturation kinetics depends both on the time and on the temperature [37,38,39,52]. Even at 80 °C, it still took approx. 2.5 h to achieve a constant rotation of a solution containing 5 mg mL⁻¹ β-lactoglobulin heated directly in a water-jacketed polarimeter tube, while at 60 °C, the rotation increase was still in progress after 6 h [39]. The short-term thermal stress of 10 min will thus most likely not perturb the protein structure to the maximum extent.

3.4. Long-Time Thermal Denaturation Measurements

To further investigate this effect, long-term denaturation experiments on mAb 3 were conducted wherein the antibody was exposed to temperatures from 70 to 85 °C for time durations up to 3.5 h. Samples filled in NMR tubes were thermally stressed in a water bath, followed by quenching of the aggregation through immersion in an ice water bath. mAb 3 is a monoclonal antibody of type IgG₁, a class which typically denatures irreversibly at temperatures over 65 °C [53,54,55], with the antibody’s Fab domain unfolding initially, followed by an aggregation mechanism [53,56]. Figure 1 (and Tables S4–S7) show the progression of the specific rotation of the [α]_g²⁵ line with increasing denaturation time at different denaturation measurements. At temperatures of 80 °C and 85 °C, the specific rotation of mAb 3 reaches a plateau early after a rapid initial increase, indicating completion of the denaturation process. At a temperature of 75 °C, the slope of the polarimetric signal is lower, but almost reaches maximum denaturation after 3.5 h. At a temperature of 70 °C, there occurs a very gradual decrease of the optical rotation with a comparatively low degree of irreversible denaturation even after a denaturation time of 3.5 h. These findings align with a study by Sawyer et. al. [39] on the β-sheet rich protein β-lactoglobulin. Rosenqvist et. al. [57] found similar results for the thermal aggregation of human IgG, controlled by dynamic light scattering, which showed that the aggregation process of this IgG already starts at temperatures over 50 °C and the aggregation rate increases rapidly with increasing temperature. Monitoring the thermal stability of mAbs is of major interest for the biopharmaceutical industry, as denaturation and aggregation lead to a loss of the therapeutic effect. This study demonstrates the suitability of polarimetry to detect and monitor perturbation of the protein structure.

Polarimetry was further used to investigate small differences in the denaturation behavior between BSA with fatty acids compared to defatted BSA. Ma et al. employed far-UV circular dichroism spectroscopy to show that fatty acids can stabilize BSA by inserting into hydrophobic pockets, whereas defatted BSA exhibits greater conformational flexibility and is therefore more susceptible to unfolding [58]. Figure 2 (and Tables S8 and S9) show the progression of the ratio of the specific rotation [α]_g²⁵ and [α]_D²⁵ of BSA with and without fatty acids with increasing denaturation time at a denaturation temperature of 60 °C. These results show a faster and more extensive denaturation of defatted BSA compared to BSA with fatty acids and confirm the previously reported stabilization effect of fatty acids in protein denaturation.

4. Conclusions

After decades of inattention, the technique of digitally controlled high accuracy polarimetry has demonstrated potential applications with native and denatured proteins, likely valuable for biological therapeutics in particular. In contrast to the techniques available from the late 19th century until about 60 years ago, the progress in protein purification achieved over the past decades will now enable an uncomplicated and accurate characterization of proteins.

The specific rotation ratio between 436 and 589 nm has only recently been demonstrated to indicate structural properties, in particular the part of α-helices, with surprisingly reasonable accuracy. The narrow specific rotation ratio range of immunoglobulin G with its β-structure-dominated structure may constitute a specific property of this protein class to distinguish it from other proteins and may thus contribute to confirm the authenticity of such protein concentrates. Further investigations on a wider diversity of purified plasma-derived polyclonal and recombinant monoclonal immunoglobulins are therefore highly desirable to corroborate this assumption.

The apparent sensitivity of polarimetry to detect cryptic protein denaturation warrants a thorough investigation into a wider variety of purified proteins with combined orthogonal techniques such as infrared and Raman spectroscopy for neat, and as CD, UV, and fluorescence spectroscopy as well as light scattering for diluted samples. Advantageously, the neat sample solution can be recovered from the polarimeter tube and then subjected to the other analytical procedures proposed.