Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Anal Chem. Author manuscript; available in PMC 2010 August 15.
Published in final edited form as:
PMCID: PMC2846550

“Focusing” of low-molecular-mass heparins in polycationic polyacrylamide matrices


A novel method for separation of low-molecular-mass heparins is here reported, based on migrating the polyanionic heparins in a polycationic polyacrylamide gel, made by incorporating a gradient of positively-charged monomers (the Immobilines used for creating immobilized pH gradients) into the neutral polyacrylamide backbone. Separations can be operated either in linear or non-linear gradients of positive charges, thus modulating at whim the separation power. This allows the polydisperse heparins to reach a steady-state position along the migration path and condense (focus) in an environment inducing charge neutralization. It is shown that the separations obtained are a complex function of both size and charge distribution along the oligosaccharide chains. This novel methodology represents a marked improvement over existing techniques and appears to hold promises for applications in screening of commercial lots of heparins, also in view of possible presence of contaminants, such as those recently detected in imported heparins.


The masquerade began surreptitiously in 1967, when Shapiro et al.1 reported a curious method by which, when proteins were laden with sodium dodecyl sulphate micelles (SDS), they followed a migration path in a sieving polyacrylamide matrix showing an inverse linear relationship (on a semi-log plot) with their relative molecular mass values, thus cancelling out the contribution of surface charge, as typical of migration in disc electrophoresis2. The proteins, coated by micelles of SDS, to the point of swamping their original amphoteric charge, were forced to act like transvestites and to behave just like nucleic acids, odd macromolecules in which the charge to mass ratio becomes nearly constant above ca. 400 bp in length3. Carnival in Copacabana, not just during carnival, but all the year around! Most proteins would adsorb SDS to a magic ratio of 1.4 mg SDS/mg protein4: if one would then drive them into a porosity gradient gel5, may be even by exploiting discontinuous buffers6, one would end up with razor-blade-sharp zones, whose mobility, when plotted against the log of molecular mass, would result in a linear relationship7. It was a quantum jump in electrokinetic methodologies, and the technique became one of the most popular ones in the field of electrophoresis.

The next quantum jump occurred in 2007, when Zilberstein et al.8 reported an unorthodox method, called “SDS-PAGE focusing”, exploiting a “steady-state” process by which the SDS-protein micelles were driven to stationary zones along the migration path against a gradient of positive charges affixed to the neutral, minimally sieving, polyacrylamide matrix. As the total negative surface charge of such complexes matched the surrounding charge density of the matrix, the SDS-protein complex stopped migrating and remained stationary. Steady-state conditions were in fact demonstrated, since pattern stability could be guaranteed for at least 24 hrs of “focusing”.

A preparative version of this novel methodology was also reported, by which it was further demonstrated that such separations could be engendered not only in a chemical gel, i.e. in a solid matrix, but also on an entangled solution of cationic polymers, polymerized in the absence of cross-linking agents9. For impeding migration of these charged polymers in the electric field, viscosity gradients and low voltages had to be adopted. Unique separations were also obtained in the case of peptides, down to only 500 Da, where conventional SDS-PAGE fails, since all peptides below 6000 Da usually run with the solvent front and are lost in the anodic compartment10.

The SDS-PAGE focusing has worked so remarkably well for proteins and peptides only because they are present as mixed micelles with the SDS surfactant, which coats the polypeptide chain, effectively masking its zwitterionic character and rendering it polyanionic in nature, to a point of very-nearly constant charge to mass ratio4. Clearly this separation mechanism should work quite well also for DNA fragments, since they are intrinsically polyanionic at all pH values above pH 5 (below, in the pH 2–3 range, they would exhibit too a zwitterionic character, due to progressive protonation of three (A, C, G) out of four bases (T having no basic groups capable of being protonated11). Excellent separations of DNA fragments in our polycationic gel matrices were subsequently reported12, with some unique results, such as: (a) the possibility of achieving base-line resolution above 600 bp (base pairs), where all other gel-based electrophoretic techniques fail, (b) the possibility of separating oligo-DNAs having the same length but differing in one base along the chain (the so-called SNP, single base polymorphism), where no separation occurs by any known electrokinetic transport mechanism.

A logical extension of the work on DNA would be to apply our focusing technique to heparin, a naturally occurring polysaccharide, well known for its anticoagulant and antithrombotic properties. Heparin is a sulphated, linear polysaccharide belonging to the family of glycosaminoglycans and is constituted by alternating disaccharide sequences of an uronic acid and an amino sugar13. Different sulphation patterns are unevenly distributed along the heparin chains, with less charged sequences mostly concentrated towards the reducing side of the chains and the most charged ones towards the non-reducing side, with mixed domains between the two regions. On the average, the mean degree of sulphation of heparins is 2.1–2.2 sulphate residues per disaccharide unit and 1 carboxyl residue for the same disaccharide repeat. There are no free, protonatable amino groups to be found in native heparins: therefore, unlike DNAs14, that are amphoteric at very low pH and exhibit, on the average, pI values around pH 2.1–2.3, heparins present a pure polyanionic surface and therefore cannot be “focused” at any pH value.

Recently, a new family of heparins has emerged: low-molecular-mass heparins15 (LMMH). These products all have mean Mr values less than half that of standard heparin; the defining characteristic that all have in common being that 60% or more by mass must be below 8000 Da16. These fractions are prepared by depolymerization of heparin, via deaminative cleavage, or chemical or enzymatic β-elimination or by oxidative depolymerization. LMMH, as a group, share the same essential mechanism of action (binding to antithrombin) and show the following differences from unfractionated heparin: (i) higher anti factor Xa than factor IIa activity; (ii) bioavailability approaching 100%, leading to administration once or twice daily; (iii) lesser interaction with heparin-binding proteins (PF4, protamine, lipase, histidine-rich glycoproteins etc.). Therefore, due to its greater efficacy, less haemorrhagic effects and improved convenience, LMMHs are now the preferred treatment in the field of thrombosis and haemostasis as compared with heparin. Although of much reduced size as compared with heparin, LMMHs still exhibit a tremendous polydispersity in both size and charge distribution along the polymer chain, requiring more and more sophisticated approaches to decode their structure.

For fractionating and characterizing the LMM class of heparins, we report here the application of a recently-developed method8 based on “focusing” these polyanions along a gradient of polycationic charges affixed on an otherwise neutral polyacrylamide backbone. Three different preparations of LMMHs, Enoxaparin, Tinzaparin17 and γ–LMWH18, were analysed, together with a preparation of heparin derived oligosaccharides (γ–HDO)19. The results obtained are quite unique and point out to a fractionation mechanism sensitive to a subtle interplay of size and charge density of LMMHs.


Chemicals and Materials

Basic Immobiline™ with various pKa values (8.5, 9.3, 10.3, 12), were purchased from Amersham Pharmacia Biotech (Uppsala, Sweden). Fluka (Buchs, Switzerland) provided acetic acid, boric acid, StainsAll, urea, Tricine and Tris. Acrylamide/bis-acrylamide solution (Cat. Number 161-0156) was from Bio-Rad (Hercules, CA, USA). Different LMMHs (Enoxaparin Mr= 4500 Da, γ-LMWH Mr= 5000 Da, Tinzaparin Mr= 6500 Da and γ–HDO, Mr= 2200 Da) were as previously described17,18,19. For Enoxaparin and Tinzaparin, characteristic Mr values reported in the European Pharmacopoeia monograph are given. γ-LMWH and γ–HDO research samples were kindly provided by Laboratori Derivati Organici (Milan, Italy).

Casting of polycationic gels

Gels are cast with a two vessel gradient mixer, as routinely performed for pouring immobilized pH gradient (IPG) gels20. Mini gels (8 × 7 cm, 0.75 mm thick) are cast in the Bio Rad system either singly or in a six-unit casting chamber. All gels are supported by a Mylar film (Gel Bond) for easy handling when opening the cassette. The typical gel formulations consist of 4% T polyacrylamide (3.3% cross linker) in presence of a gradient of positively-charged ions, i.e. basic Immobilines. We have tried different basic species, but the preferred gel formulation adopts the pK 10.3 and pK 12 Immobilines, since this species will be fully protonated at the standard running pH (0.12 M Tris/Acetate pH 6.4). Different slopes of basic Immobiline gradients have been tested, in the intervals 2–4 mM and 0–20 mM. Immediately after casting, the gels are placed in a oven at 50°C and polymerization continued at this temperature for 1 h21. To perform electrophoresis, we have tried both vertical and horizontal systems. The preferred configuration has been a horizontal set-up in the Multiphor II chamber (Amersham-GE). The cathodic and anodic electrolytes were 0.1 M Tris/Tricine pH 8.3 as a cathodic buffer and 0.1 M Tris/Acetate pH 6.4 as an anodic buffer. Gels have been run for various periods of time, from 45 min up to 24 hrs, in order to test for steady-state “focusing” conditions. In some cases, the system has been run in the absence of a gradient of basic Immobiline, by adopting a plateau (constant concentration) level. Typical running conditions: 300 V with forced cooling at 10°C. Heparin zones were revealed by Stains All.

Disc electrophoresis of heparins

Each heparin (Enoxaparin Mr= 4975 Da, γ–Heparin Mr= 5000 Da, Tinzaparin Mr= 8080 Da and VLMWH heparin Mr= 2200 Da) was solubilised in Tris-Acetate buffer at concentration of 10 mg/mL. Twenty µL of each sample were mixed with 16 µL of 80% glycerol solution and 0.5 µL of bromophenol blue saturated solution; the mixture was loaded in pockets pre-cast on the stacking gel. The gel was composed by a stacking gel (125 mM Tris-HCl, pH 6.8) with a large pore polyacrylamide gel (4%) cast over two different resolving gels (the first composed of 12–25% acrylamide gradient in 100 mM Tris-HCl, pH 8.45, the second of 20% acrylamide constant porosity in the same buffer). The cathodic compartment was filled with Tris-Tricine buffer and the anodic one was filled by Tris-HCl buffer pH= 8.8. This set-up is essentially identical to that of Laemmli6, with the proviso that SDS has been removed from all formulation and is, in fact, a simplified version of the classic Ornstein2 and Davis22 system in which the sample gel has been eliminated. A similar set-up was proposed by Rice et al.23, except that they used borate/HCl as a leading boundary and Gly as terminating ion and staining by either Alcian Blue or Azure A. Electrophoresis was at 100 V for 1 hour and at 250 V until the dye front reached the bottom of the gel. For staining, a solution of 0.1% Stains All (Sigma-Aldrich) was prepared in formamide (pH 7.3–7.4) and refrigerated in the dark. The staining solution is diluted 1:20 from stock with a diluent (10% formamide, 25% 2-propanol, 15 mM Tris-HCl pH 8.8 in water) and the gels were stained overnight at room temperature. For destaining, the gels were removed from the staining solution and exposed to visible light from a light box until sufficient destaining had occurred. The mono-dimensional gels were scanned with a Versa-Doc image system (Bio-Rad).

Gel permeation chromatography

The separation of oligosaccharidic components was accomplished by fractionating enoxaparin through gel permeation chromatography on Biogel P10 (Bio-Rad Laboratories) by using two columns connected in series (5 × 66 cm and 5 × 93 cm). 150 mg of Enoxaparin, dissolved in 3.5 ml of H2O, were loaded and eluted with NH4Cl 0.25 M at 0.4 mL/min. The elution profile was plotted by evaluating the absorbance of the flow-through at 210 nm. Fractions corresponding to each peak were pooled, concentrated down to about 3 mL under reduced pressure, at 30°C and finally desalted by gel permeation chromatography on TSK 40S.


Fig. 1 shows the separation of three types of LMMHs (Enoxaparin; Tinzaparin and γ-LMWH) in a 0–5 mM gradient of pK 8.5 Immobiline: it can be appreciated that a large number of bands are fully resolved in the interval of positive charges. E. g., in Enoxaparin, >15 bands can be seen sharply focused; in Tinzaparin >10 zones and even in γ-heparin, which appears as a smear, inspection of the gel with a transilluminator reveals that, underneath the apparent smear, sharply focused zones are visible. Such a smearing can be explained by the very complex oligosaccharide composition of the γ-LMWH. Actually, while both Enoxaparin and Tinzaparin are mostly composed of oligomeric chains differing for multiples of disaccharide units, γ-LMWH chains consists of both odd and even number of monomers19. As additionally visible in Fig. 2, the separation can be further modulated by using different gradients of Immobilines, e. g. in the intervals 0–5 or 0–10 mM in a linear (two upper panels) or non-linear gradient mode (two lower panels).

Figure 1
“Focusing” of Enoxaparin, Tinzaparin and γ-heparin in a 6% T polyacrylamide gel containing a 0–5 mM gradient of pK 10.3 Immobiline. Conditions: 7-cm long gel, 0.1 M Tris/Acetate , 400 V, 2 hrs. Two different loads for each ...
Figure 2
“Focusing” of γ-heparin, Tinzaparin and Enoxaparin in 6% T polyacrylamide gels containing either a linear gradient of 0–5 mM or 0–10 mM pK 8.5 Immobiline (two upper panels) or a non-linear (NL) gradient of 0–5 ...

It might be asked how classical electrophoretic methods would fare, compared to our present technology. The best method reported so far is disc electrophoresis. We have tried it in both, constant concentration (Fig. 3, left panel) or porosity gradient gels (Fig. 3, right panel): in both cases, though, only continuous smears are obtained, with only a hint at some band resolution in the Tinzaparin sample. Conversely (see Fig. 1) excellent resolution is obtained for all samples in our polycationic gradient gels.

Figure 3
Disc electrophoresis of γ-HDO, γ-LMWH, Tinzaparin and Enoxaparin in a constant porosity (20% T) polyacrylamide gel (left panel) or in a 12–25%T porosity gradient gel (right panel). For comparison of the same separations in a “focusing” ...

Since the mechanism of these separations is not quite understood, we have taken Enoxaparin and fractionated it in a Bio-gel P10. A series of peaks was obtained and the fractions corresponding to each one have been collected and analyzed by mass spectrometry (MS) to determine the average chain length of the resolved oligomeric families (data not shown). In Fig. 4, the increasing Mr values of enoxaparin components starting from a 4-mer (a tetra-saccharide unit), up to 6-mer (a hexa-saccharide unit), all the way up to a 22-mer, are indicated. The 4-mer appears to have a purity >98%, the 6-mer of 90%. Interestingly, in between the various peaks of even mers (tetra- hexa- etc.) there are small peaks, that have been identified as odd-mers (tri-mer in front of the 4-mer; penta-mer in between the 6-mer and hepta-mer peaks, etc.).

Figure 4
Size exclusion chromatography of Enoxaparin on a Bio gel P10 resin. One hundred and fifty mg of Enoxaparin, dissolved in 3.5 ml of H2O, were loaded and eluted with NH4Cl 0.25 M at 0.4 mL/min. Eluate monitored at 210 nm. The degree of polymerization of ...

The major peaks eluted have been further analyzed by our focusing technique. As shown in Fig. 5, it can be appreciated that the 4-mer appears as an almost homogeneous peak, with a couple of weak bands just above the major peak. They are due to the presence of traces of 5-mer and 6-mer, as confirmed by MS analysis. Already starting with the 6-mer lane, no single major band is visible any longer, but an increasing amount of bands of almost equal intensity (8-, 10- and 12-mers) suggesting that, superimposed to the size fractionation, our technique can distinguish subpopulations, which include both larger and smaller oligosaccharides together with oligomers of equal backbone size but with different sulfation degrees, as suggested by MS analysis of the evaluated fractions (Bisio A., manuscript in preparation). In fact, with the increase of heparin chain length, a decrease of chromatographic resolution occurs. Accordingly, the larger the oligomeric family and the higher is its heterogeneity. Such a complexity is clearly visible in both the linear (left panel) and non-linear (right panel) gradient of charges shown in Fig. 5.

Figure 5
Analysis of the Bio gel P10 fractions (Fig. 4) via “focusing” in a 0–5 mM (left panel) or 0–10 mM (right panel) polycationic gel. Tracks: 1, unfractionated Enoxaparin; 2, 4-mer; 3, 6-mer; 4, 8-mer; 5, 10-mer and 6, 12-mer. ...

Our focusing technique, just like the separation in Bio-gel P10, can be used on a small-scale preparative version, by first separating the major bands along the gradient of charges, and then eluting the bands of interest in a suitable buffer. As shown in Fig. 6, six bands have been eluted from a sample of Enoxaparin and then re-run in an analytical gel: just like the Bio-gel P10 separation, when these bands are re-run, they show microheterogeneity, suggesting here too an envelope of bands.

Figure 6
Small scale preparative fractionation of Enoxaparin in a “focusing” gel. Tracks: 1, unfractionated Enoxaparin; 2–8 eluted fractions re-run in an analytical gel, containing a linear gradient of 0–5 mM positive charges.


Electrophoretic techniques, except for separations by capillary zone electrophoresis24, have been largely abandoned for analysis of heparins, due to their tremendous complexity arising from both size and varying charge distribution along the polysaccharide chain. In fact, today, most high-resolution separations of heparins are achieved via HPLC in the reversed phase mode, via charge neutralization of the highly sulphated species by ion-pairing. Our technique brings to the limelight electrophoresis, in a “focusing” version, as a high-resolution method for fractionating, both analytically and in a small-scale preparative version, LMMH heparins. Although the mechanism of our separation has not been fully unravelled, it appears that it is a mixed-mode fractionation, in which both size and charge distribution along the polymer play an important role. From this point of view, more studies will have to be conducted with refined techniques, such as NMR and mass spectrometry of isolated fractions. It is in any event known, via chromatographic approaches, that isolated, homogenous fractions (as obtained, e.g., by gel filtration) would give rise to a number of bands when further analyzed by RP-HPLC (not shown).

An important historical fact will be here recalled to show how, indeed, the present methodology is brand new and has nothing to do with conventional focusing techniques as implemented, in the past, by exploiting soluble amphoteric carrier buffers. In 1974 and 1975, Nader et al.25 and McDuffie et al.26 reported that heparin could be fractionated into 21 components, with apparent pI values between pH 3.2 and 5.0. The basis of the fractionation was associated with their Mr polydispersity, the “low pI" components exhibiting the smallest size (3000 Da), while the “high pI” species had the highest Mr (37500 Da), with graded intervals of ca. 2000 Da for species in between. This separations were quite puzzling since, as stated in the introduction, whereas nucleic acids are at least amphoteric, although with identical pI values around pH 2.1 to 2.3, heparins are not, since they are pure polyanions, with no positive counter-charged covalently affixed along the polysaccharide chain. The mechanism of this fractionation was unravelled by Righetti and Gianazza27 and Righetti et al.28 On the basis of spectra taken in solution and in the focused gel, of repeated runs of isolated, focused bands, of IEF fractionations performed in the ‘presence and absence of urea, and variable Ampholine-heparin ratios and with heparins of varying degrees of carboxylation and sulphation, they demonstrated the IEF heparin profile to be artefactually elicited by interaction with carrier ampholytes. The 21 IEF fractions indeed represented complexes of heparin (independent form its true nature and polydispersity) with 21 Ampholine molecules. The complexes differed markedly in binding strength, the high pI species forming the weakest interactions, which could be extensively split by Toluidine blue, while the lowest pI species exhibited the strongest aggregates, which could not be disaggregated by excess basic dye. Thus, paradoxically, heparin was used as a reporter molecule in order to find the distribution of the major Ampholine species focusing in the pH 3.2–5.0. It was then understood that all polyanions, such as polysulphates, polyphosphates and polycarboxylates, would give rise to the very same artefacts29. On the contrary, in the present technique, a genuine focusing is achieved, since the polyanion meets and binds an equivalent number of positive charges along the gradient of the polycationic gel, thereby attaining charge neutralization and thus condensation at a pI value given by the balancing of the two opposite charges. This arrest of migration and pI condensation is at the basis of the very high resolution given by all focusing techniques.


With the present report, we are adding a new technique to the panoply of methodologies already available for heparin analysis and fractionation: a focusing technique able to discriminate complex LMMH heparins via a subtle interplay of mass and charge fractionation. The resolving power can be enhanced by adopting gradient of positive charges of different slopes (e.g., 0–5 mM or 0–10 mM, or higher, as here reported) as well as linear and non-linear gradients of such charges. It is hoped that this will open a new window in analysis of such complex polysaccharide mixture. This new method could be useful also for detecting potential contaminants of heparins and LMMHs30.


PGR is supported by the Bilateral Project "Novel Methods for Top-down Analysis of Macromolecular and Nanosized Samples of Biotechnological and Environmental Interest" within the VIII Executive Programme of Scientific and Technological Co-operation between Italy and Korea for the years 2007–2009. GT and AB were supported by National Institutes of Health grant 1-R01-HL080278-01. The authors thank Leo-Pharma (Denmark) and Sanofi-Aventis (France and Italy) for providing Tinzaparin and Enoxaparin samples, respectively.


1. Shapiro AL, Vinuela E, Maizel JV. Biochem. Biophys. Res. Commun. 1967;28:815–829. [PubMed]
2. Ornstein L. Ann. N. Y. Acad. Sci. 1964;121:321–349. [PubMed]
3. Stellwagen N, Gelfi C, Righetti PG. Biopolymers. 1997;42:687–703. [PubMed]
4. Pitt-Rivers R, Impiombato FSA. Biochem. J. 1968;109:825–831. [PubMed]
5. Margolis J, Kenrick KG. Anal. Biochem. 1968;25:347–355. [PubMed]
6. Laemmli UK. Nature. 1970;227:680–681. [PubMed]
7. Chrambach A, Rodbard D. Science. 1971;172:440–445. [PubMed]
8. Zilberstein G, Korol L, Antonioli P, Righetti PG, Bukshpan S. Anal Chem. 2007;79:821–827. [PubMed]
9. Zilberstein G, Korol L, Righetti PG, Bukshpan S. Anal Chem. 2007;79:8624–8630. [PubMed]
10. Zilberstein G, Korol L, Shlar I, Righetti PG, Bukshpan S. Electrophoresis. 2008;29:1749–1752. [PubMed]
11. De Wachter R, Fiers W. In: Gel Electrophoresis of Nucleic Acids: a Practical Approach. Rickwood D, Hames BD, editors. Oxford: IRL Press; 1982. pp. 77–116.
12. Zilberstein G, Korol L, Znaleziona J, Sebastiano R, Righetti PG, Shlar I, Baskin E, Bukshpan S. Anal. Chem. 2008;80:5031–5035. [PubMed]
13. Casu B. In: Chemistry and Biology of Heparin and Heparan Sulphate. Garg HG, Linhardt RJ, Hales CA, editors. Amsterdam: Elsevier; 2005.
14. Drysdale JW, Righetti PG. Biochemistry. 1972;11:4044–4052. [PubMed]
15. Casu B, Lindahl U. Adv. Carbohydr. Chem. Biochem. 2001;57:159–206. [PubMed]
16. Gray E, Mulloy B, Barrowcliffe TW. Thromb. Haemost. 2008;99:807–818. [PubMed]
17. Guerrini M, Guglieri S, Naggi A, Sasisekharan R, Torri G. Semin. Thromb. Hemost. 2007;33:478–487. [PubMed]
18. Bisio A, De Ambrosi L, Gonella S, Guerrini M, Guglieri S, Maggia G, Torri G. Arzneim. Forsch. Drug Res. 2001;51:806–813. [PubMed]
19. Bisio A, Guglieri S, Frigerio M, Torri G, Vismara E, Cornelli U, Bensi D, Gonella S, De Ambrosi L. Carbohydr. Polym. 2004;55:101–112.
20. Righetti PG. Immobilized pH Gradients: Theory and Methodology. Amsterdam: Elsevier; pp. 127–139.
21. Righetti PG, Ek K, Bjellqvist B. J. Chromatogr. 1984;291:31–42.
22. Davis B. Ann. N. Y. Acad. Sci. 1964;121:404–427. [PubMed]
23. Rice KG, Rottink MK, Linhardt RJ. Biochem. J. 1987;244:515–522. [PubMed]
24. Chiesa C, O’Neill RA, Horvath CG, Oefner PJ. In: Capillary Electrophoresis in Analytical Biotechnology. Righetti PGZ, editor. Boca Raton: CRC Press; 1996. pp. 277–430.
25. Nader HB, McDuffie NM, Dietrich CP. Biochem. Biophys. Res. Commun. 1974;57:488–493. [PubMed]
26. McDuffie NM, Dietrich CP, Nader HB. Biopolymers. 1975;14:1473–1486. [PubMed]
27. Righetti PG, Gianazza E. Biochim. Biophys. Acta. 1978;532:137–146. [PubMed]
28. Righetti PG, Brown RP, Stone AL. Biochim. Biophys. Acta. 1978;542:232–244. [PubMed]
29. Gianazza E, Righetti PG. Biochim. Biophys. Acta. 1978;540:357–364. [PubMed]
30. Guerrini M, Beccati D, Shriver Z, Naggi A, Viswanathan K, Bisio A, et al. Nature Biotech. 2008;26:669–675. [PMC free article] [PubMed]