Elecrrophorc~si~ 1992, 13, 587-595

'Laterally aggregated' polyacrylamide gels for electrophoresis

587

Pier Giorgio Righetti' SiIvia Caglio' Marco Saracchi2 Sergio Quaroni'

'Laterally aggregated' polyacrylamide gels for electrophoresis

'Department of Biomedical Sciences and Technologies, University of Milano 'Institute of Plant Pathology, University of Milano

Anew method is described for producing highly porous polyacrylamide matrices: polymerization in presence of a preformed hydrophilic polymer. If a standard mixture of monomers (e.g., 5%T, 4YoC) is polymerized in presence of, e.g., polyethylene glycol (PEG) 10 kDa, lateral chain aggregation occurs, with formation of large pore sizes. In PEG 10 kDa, the transition from a small- to a large-pore gel is clearly apparent at 0.5% PEG addition and reaches a pleateau already at 2.5% PEG. Even with shorter PEG fragments (6,2 and 1 kDa) this transition occurs, but with progressively largeramounts ofPEG in solution (up to 25 O/o forthe 1kDa species). Other polymers such as hydroxymethyl cellulose (1000 kDa) and polyvinylpyrrolidone (360 kDa and 25 kDa) are also able to elicit this phenomenon. It appears that lateral chain aggregation (before the cross-linking event) is induced via intra-chain hydrogen bonding, since urea and temperature strongly inhibit it, whereas tetramethylurea (an agent quenching hydrophobic interactions) does not hamper it. By scanning electron microscope, it is found that the maximum pore size obtained in a 5%T, 4 % c gel in presence of 2.5% PEG 10 kDA is of the order of 0.5 pm,whereas the same 5%T, 4 % c control gel would have an average pore diameter of 5 nm. Thus, an increment of pore size of about 2 orders of magnitude is obtained: in these new matrices, a 21000 bp DNA fragment exhibits a much greater migration than in a control gel in which the sample is entrapped at the application site.

1 Introduction

they could be used for immunoelectrophoretic techniques (which rely exclusively on the use of agarose matrices) and Since their inception as a matrix for electrophoresis 11-31, for isotachophoresis [9], where nonrestrictive media are polyacrylamide gels have gained enormous importance in mandatory. The Ogston theory [lo] predicts that, if the averbiochemical separations, especially in the field of protein age polyacrylamide fiber were to be reduced in length and analysis and characterization. This is because the average thickened in diameter (believed to be of the order of 0.5pore size of such matrices is similar to the magnitude of the 1.0 nm), the average pore size would increase progressively. average diameter of globular proteins (4-10 nm), thus in- Curiously, this was achieved since the inception of disc electroducing molecular sieving in an otherwise pure surface trophoresis, as Ornstein and Davis [2,3] proposed gel plugs charge fractionation in gel-free media. Due to restriction of of 20%C for the sample and stacking gel segments, where molecular diffusion, sieving media also offer substantial an isotachophoretic process would occur. Intuitively, it was improvements in resolution. Over the years, a dichotomy understood that such highly cross-linked gels would be has developed between the two most popular gel media in highly porous, a fact confirmed by Fawcett and Morris [7], electrophoretic separations: agarose gels (exhibiting aver- who found that in such high %C gels the fiber diameter age pore diameters in the order of several hundredths nm would increase from 0.5 up to 6 nm, with a concomitant up to pm size) are used almost exclusively for nucleic acid shortening of the average fiber 1ength.This is thought to be fractionations [4], whereas polyacrylamides have become due to 'bundling' of individual fibers into 'pillar-like' aggrethe matrix of choice for protein analysis [5]. gates [ll]. The increase in pore size in highly cross-linked gels is quite impressive; thus with N,N-methylenebisWith some exceptions, most authors today agree that the acrylamide (Bis) as cross-linker, Righetti et al. 112,131 have average pore diameter in standard polyacrylamides is of the measured, in 3%T, 60%C,,, gels, an average pore diameter order of a few nm. Thus, White 161 has reported an average as large as 600 nm. In a corresponding series of 3%T- with diameter of 3.6 nm in 5%T, 5%C gels; Fawcett and Morris N,N'-(1,2-dihydroxyethylene) bisacrylamide (DHEBA) 171 have given a 4 nm diameter for a 6.5 O/oT, 5 %C gel and ~OO/OC,,,,, gels, the pore diameter was close to 500 nm. A Grimaud ed a/. [8] have proposed a range of 15-20 nm pore pore size in the submicron range is essentially nonsieving diameter in Bio-Gel P-300 spheres for chromatography. for large proteins such as ferritin (a globule of 9.4 nm diameter) 1141, whose mobility (in 30°/oc gels) was found to However, for a host of separations, polyacrylamides of large be almost identical to that measured in free-zone electropore size would be extremely useful because, for example, phoresis. However, such very high %C gels suffer from severe limitations: they are opaque, collapse in the gravitaCorrespondence: Prof. P. G. Righetti, University of Milano,Via Celoria 2, tional field and exude water (possibly also as a result of the Milano 20133, Italy much increased hydrophobicity). Clearly, their use is very limited and has been largely abandoned, except for the Abbreviations: %C,g cross-linker/O/oT;OhT, (g acrylamide + g Bis)/100 mL; hoax of highly cross-linked N,N-diallyltartardiamide Bis, N,N-methylenebisacrylamide; bp, base pairs; PEG, polyethylene gly(DATD) gels. With DATD as an inhibitor of gel polymerizacol; HMC, hydroxymethyl cellulose; PVP, polyvinylpyrrolidone; TMU, tetrarnethylurea; DHEBA, N,N-(1,2-dihydroxyethylene)bisacrylamide; tion, such gels are not highly cross-linked but highly unpoDATD, N,N-diallyltartardiamide lymerized [15]. Q VCH Verlagsgesellschaft

mbH, D-6940 Weinheim, 1992

0173-0835/92/0910-0587 $3.50+.25/0

Elcctroiniioresrr 1992. 13. 587-595

We propose a novel method for producing highly porous polyacrylaniide gels: the use of ‘laterally aggregating’ agents. In this procedure, gels of regular %C (e.g., 4-5 %C) are polymerized, but in the presence of a preformed, hydrophilic polymer, which induces formation of thick gel fibers by lateral aggregation of the nascent chains.

2 Materials and methods 2.1 Materials Acrylamide, ammonium persulfate and (N,N,N’,N’-tetramethylethylenediamine (TEMED) were from Bio-Rad Laboratories (Richmond, CA). The polyethylene glycols (PEG) 2000, 6000 and 10000 as well as polyvinylpyrrolidone (PVP) 25000 were from Merck (Darmstadt, Germany), while PEG 20000 was from Baker Chemicals (Deventer, The Netherlands) and PVP 360000 and hydroxymethyl cellulose (HMC) 1000 kDa were from Aldrich (Steinheim, Germany). All chemicals were of analytical grade. The following DNA M, markers: 1(6 fragments from 3530 to 21266 bp),II (8 fragments,frorn 125 to 23130 bp),III (13 fragments, from 125 to 21266 bp) and V (22 fragments, from 8 to 587 bp) were from Boehringer Mannheim (Penzberg, Germany).

2.2 Spectrophotometric reading The turbidity of‘laterally aggregated’ gels was studied with a Varian DMS-90 UV-Vis spectrophotometer (Varian, Palo Alto, CA) at 600 nm against a plain water blank. All gels were prepared as 5%T, 4%C matrices in presence of variable amounts of preformed linear polymers, as follows: for the PEG series, the concentrations added ranged from 0.1 Yo up to 25 O/o w/v; for PVP, the range was in the 0.2O/o up to 15% interval and for HMC 1000 kDa the amount incorporated ranged from 0.01% up to 0.1%. All solutions were supplemented with 1 pL/mL of T E M E D and 1 pL/mL of 40% ammonium persulfate. Each point was run in triplicate cuvettes and was allowed to polymerize 1 h at room temperature

2.5 Electron microscopy The following procedure has proven optimal for sample preparation for the scanning electron microscope: the polymerized gels (as prepared in a spectrophotometric cuvette, in presence of increasing amounts o f P E G 10 kDa,from 0% up to 2.5%) are first fractured in several segments (not by cutting, but by breaking them), which are then equilibrated for 2 days in 2.5% PEG 10 kDa (i.e., the plateau concentration giving maximum turbidity and thus maximal chain aggregation).The fragments are quicklyfrozen in liquid nitrogen and immediately lyophilized, without allowing melting to occur during the entire process. It was found that. in the absence of the PEG 10 kDa equilibration step, all samples containing either no PEG or lower amounts of PEG presented variable degrees of disintegration of gel structure during the various sample manipulations prior to microscopic observations. The samples were then spread with a thin gold/palladium layer in an Edwards 306 metallizer (Edwards High Vacuum Ltd. Crawley U K ) and then observed by a scanning electron microscope (SEM) Stereoscan 250 from Cambridge Sci. Instr. Ltd,Cambridge,UK,at 20 kV. Photographs were taken with a Kodak film TRI-X PAN 120. 2.6 DNA electrophoresis

DNA fractionation was performed in 6%T, 4 %C polyacrylamide gel slabs in 89 mM Tris, 89 mM borate and 2 mM EDTA, pH 8.3.The slabs were 1.5 m m thick and 12 cm long and were run at 180 V for 3.5 h. In addition to control gels, two other types of gels were run: a series containing a PEG 10 kDa gradient from 0 to 1O/o and another with a PEG gradient from 0 to 2%. All solutions were supplemented with 1 pL/mL o f T E M E D and 1 pL/mL of 40°/0 ammonium persulfate just prior to casting. Polymerization was allowed to ,>roceedfor 1 h at room temperature. DNA samples were loaded by diluting 1.5 pL DNA with 13.5 pL sample buffer. Staining was with ethidium bromide.

3 Results

2.3 Disaggregation studies

3.1 Gel polymerization in presence of PEG

In order to investigate the type of bond formed in ‘laterally aggregated’ gels, the same series of experiments as in Section 2.2 was run, except that the solutions were made up in increasing concentrations of either urea (up to 6 M ) or tetramethylurea (from 0.86 M up to 4.31 M, c.e. 50% v/v). After gelling, readings of turbidity were performed at 600 nm as described in Section 2.2. In the urea series, two sets of experiments were run - one at room temperature and the other at 70°C -to check the added effects of urea and temperature.

Figure 1 shows the process of ‘lateral chain aggregation’as monitored by turbidity during gel polymerizaation in presence of increasing amounts of preformed polymer. In this case, the PEG series was utilized,ranging in mass from 2 up to 20 kDa. All gels were 5 %T,4%C and polymerized with the same amounts of catalysts as in Section 2.2. Note that the transition occurs at progressively lower amounts of PEG with increasing PEG chain length: thus, while only 1.5% PEG 20 kDA is needed for plateauing, ca. 10% PEG 2000 Da is required for the same transition

2.4 Centrifugation studies

Figure 2 shows that the same phenomenon can be obtained even with such a short chain as PEG 1000 Da, but with concentrations that are much too high; 25% PEG being required to reach plateau levels.At such hjgh levels, another phenomenon becomes important: dehydration of the gel matrix, with contraption of the gel phase and water exudation. Due to such unwanted effects, most studies are now conducted with PEG 10 kDa, since only 2.5% is required to

To gather information on the type of structural elements formed in ‘laterally aggregated’ gels, a series of polymerization experiments was conducted in a centrifuge (Biofuge 28RS, Heraeus Sepatech GmbH, Osterode Germany) at 25°C at different centrifugal speeds ranging from 100 to 6000 revolution5 per minute (RPM).

Electrophoresis 1992, 13, 587-595

complete the transition. Figure 3 summarizes the data obtained with the PEG series: the concentrations required as a function of the polymer length are seen to approach both axes asymptotically.

Chemical gel 5 X T , 4%C

/J/:.o 0

0

1

2

589

‘Laterally aggregated’ polyacrylamide gels for electrophoresis

3 4 5 6 % Polyethylenglycol

7

8

9

Figure I . Gel turbidityvs. type and percent oflaterally aggregating agents. Five O/oT, 4%cgels were polymerized in presence of increasing amounts of PEG having the following molecular mass: 2000 (2K), 6000 (6K), 10000 (10K) and 20000 (20K). ’he gel opacity was read in aVarian spectrophotometer at 600 nm.

3.2 Gel polymerization in presence of other polymers We then tried to generalize this phenomenon to other types of polymers. As shown in Fig. 4,HMC (1 million Da) produces very similar effects, although in a much reduced concentration range (barely from 0.01 O/o and flattening at 0.1 Vo). In order to show the difference with the PEG series, the results with PEG 10 kDa are plotted on the same scale. It is also apparent that the structural changes in the gel matrix thus obtained should be different from the PEG series because the upper level of turbidity is only 1.5 A, as compared with 2.3 A for the PEG-aggregated matrices. The lateral chain aggregation is a quite general phenomenon which can apparently be elicited by a number of polymers. As shown in Fig. 5,PVP can also reproduce the same event and here, too, the extent of the effect is directly proportional to the polymer mass: a 360 kDa PVP is about 5 times more effective (on a O/o basis) than the corresponding 25 kDa polymer. It also cannot be excluded that matrices produced with the different polymers (PEG, HMC or PVP) will have slightly different topological features because their plateau levels of turbidity differ to some extent.

3.3 Absence of cross-linkers In order to understand the mechanism of the structural changes induced by linear polymers in a gelling solution, we have tried to see if the phenomenon could be elicited

Chemical gel 5%T, 4%C

PEG-1 OK

HMC-1 OOOK AeOO nm

/ 0

5

0

10 15 X Polyethylenglycol 1 K

20

25

Figure 2. Gel turbidity as a function of addition of PEG 1000 Da. Note how plateau opacity is reached only at very high (25%) levels of PEG. At concentrations greater than 10% PEG l K , there is a progressive loss of water from the gel matrix, up to about 50% of the gel volume at the highest (25%) level of PEG (solid triangles, right ordinate).

2

1.6

Figure 4. Lateral-chain aggregation as induced by HMC polymer (1 million Da) added in solution. As a comparison, the turbidity curve in presence of PEG 10K is shown. All other conditions as in Fig. 1.

15T

i Ii

Lateral agprspatlon vs. PEG tlza

Chemical gel 2 --

./’,.-.

5%T, 4%C ,AyA

360K

5t

OJ 0

25K

\ I

5000

10000 Type of PEG

15000

0.5

20000

Figure 3. Lateral-chain aggregation as a function of molecular mass of the PEG polymers. The values reported for the various PEG types (lK, 2K, 6K, 10K and 20K), represent the inflection points (not the plateau values) of the curves in Figs. 1 and 2.

A’

0

2

4

6 a 10 % Polyvlnylpyrrolldone

12

Figure 5. Lateral-chain aggregation as induced by PVP (360K and 25K) added in solution. All other conditions as in Fig. 1.

590

P. G . Righetti et

Electrophoresis 1992, 13, 881-898

a/.

even in the absence of a cross-linker (in all cases Bis).Thus, we have run a series dubbed ‘physical gels’, as opposed to ‘chemical gels’ permanently cross-linked in the threedimensional network. It is possible to trigger a similar phenomenon even in the absence of cross-linkers, although about twice as much linear polymer is needed in solution, as compared with a‘chemical gel’(Fig. 6).The ‘physical gel’, in reality, is not a proper gel: it is an ensemble of possibly very large bundles of linear strings, which slowly sediment in the gravitational field and exude much of the water embedding the polymer - about */, of the volume will be lost as supernatant upon overnight rest (not shown).

5%T, 0Y.C

Physical gel

1.6.-

1.2-

am^ nm 0.8.-

0.4-

0 0

1

2

3

4 5 6 7 % Polyothylenglycol

8

9

10

3.4 Polymerization in a centrifugal field

To gain insight into the possible size of the gel elements as induced by the lateral-aggregating agents, we performed a series of polymerizations in a centrifuge, in a range from 100 to 6000 RPM.As seen in Fig. 7, the gels polymerized in presence of preformed linear polymers are strongly gravitysensitive, since a substantial portion precipitates in the force field (just prior to the cross-linking event).On the contrary, control gels, polymerized in the absence of such polymers, appear to be quite insensitive to the centrifugal field. Note that the y-axis (labeled O/o unpolymerized) does not indicate unreacted monomers: the supernatant was checked by capillary zone electrophoresis for the presence of free monomers, and none could be detected (not even large amounts of added catalysts could trigger any additional polymerization event). It is thus apparent that the nascent polymeric structures formed had precipitated in the force field before gelation and thus had formed a more concentrated gel in the lower portions.

Figure 6. Lateral-chain aggregation in physical gels (i.e., solutions ofpure monofunctional monomer, in the absence of cross-linker, O%C). The experiments were run with PEG 20000, 10000 and 6000. Note that about twice as much polymer is needed to bring about chain aggregation (compare with Fig. 1).

*-/

30T

PEG-aggregated gels

i 251 // 5 P 20

ir1 5 10- I

Control

00

3.5 The effect of urea and tetramethylurea It still remained obscure by what mechanism the gel would become opaque and form this ‘lateral aggregation’.We have hypothesized that, just before gelation, the nascent chains would be forced, by the preformed polymer present in solution, to ‘bundle’and form large super-coils either by hydrogen bonding or by hydrophobic interaction. In order to check for that, three series of experiments were run: in increasing urea malarities (up to 6 M) either at 25°C or at 70°C and in tetramethylurea (TMU) up to 50% v/v concentration (corresponding to 4.31 M). It appears that the ‘bundling’ phenomenon is undisturbed by the presence of TMU, while being strongly sensitive to the combined effects ofurea and temperature (Fig. 8). Since TMU is known to quench hydrophobic interactions [ 161,whereas urea and temperature are classic H-bond breakers, we attribute the formation of ‘supercoiling’ to H-bond intra-chain formation occurring just prior to the cross-linking event.

3.6 Electrophoresis of DNA restriction fragments In the past, formation of turbid polyacrylamide gels has always been associated with progressive enlarging of the pore size: in fact, Righetti et al. [12] demonstrated that 3 %T gels, with 6O%C cross-linker (Bis), exhibit porosities as high as 600 nm; with a 6O%C,,,,,, average pore size of ca. 500 nm are obtained. We have thus assumed that ‘laterally aggregated gels’should also offer large pore sizes since they

1000

2000

4000

3000

6000

5000

RPM

Figure 7. Gravity sensitivity of laterally aggregated gels. A 5%T, 4 % c gel,

in presence of 2.5 O/o PEG 10K,was polymerized in a centrifuge at progressively higher speeds (from 100 to 6000 RPM). Note how the nascent chains precipitate in the centrifugal field, whereas a control gel (5%T, 4%c, no PEG added) seems to be unaffected by centrifugal forces, suggesting the presence of thin filaments.

1.J

\ --A-

I

0.64 0

I

1

2

3

4

5

6

Molarlty

Figure 8. Disaggregation of‘laterally aggregated‘gekThree series of gels

(all 5%T,4%C),containing2.5%PEGlOK,were polymerized in presence of increasing amounts of urea (from 2 to 6 M) or TMU (from 0.86 to 4.31 ~ ) , e i t h e r a t 2 5 ’ c o r a t70’C.Note howureaand temperature strongly affect the process of bundle formation, whereas TMU leaves the process unaltered.

Electrophoresis

1992. / 3 , 587-595

‘Laterally aggregated’ polyacrylamide gels for electrophoresis

180

V,

59 1

3.5 h

21 k bp*

421 k b p

con t rol

0-1 % P E G 6% T ,4 % C

0-2%

PEG

Figure 9. Fractionation of DNAfraginents in control and laterally aggregatcd gels.All gels were 6%T,4%Cin the absence (control, left) or presence of a linear gradient from 0 to 1 O/o (central) or from 0 to 2 Yo (right panel) o f P E G 10K. All gels were run for 3.5 h at 180 V in Tris-borate-EDTA buffer, pH 8.3. Staining with ethidium bromide. Note how the 21000 bp fragment precipitates at the origin in control gels, whereas it migrates down 15% of the gel length in presence of the 0-1 O/o PEG gradient (central gel) and 30% of the gel length in the 0-2% PEG gradient (right gel). Samples (1) M , marker 111; (2) M, marker 11; ( 3 ) M, marker 1 and (4) M, markerV.

Figure 10. Scanning electron micrographs ofcontrol (A) and ‘laterally aggregated’gels. (B) 5%T,4%Cgel in presence of0.5% PEG 10 kDa; (C) same, but polymerized after adding 2 % PEG 10 kDa; (D) same, but with addition of2.5% PEG 10 kDa. Note the smooth surface ofthe control gel,with no visible pores and the progressive opening ofthe pore sizes in (C) and (D). In the last gel (D) the surface is seen to be made by regularbundles of2-300 nm thickness. delimiting ‘holes’ of ca. 500 nm diameter.

appear to have similarturbidityvalues. Gels were h e n polymerized in presence of gradients of linear polymers, ranging from 04’0to l’%and from 0% to 2%. Upon running a series of DNA restriction fragments, large DNA strands are seen to be able to penetrate the gel matrix to much greater extents than in control gels of 6O/oT, 4%C (Fig. 9). Thus, while in a control gel a 21000 bp fragment is entrapped at the deposition site. in a 0-1% PEG 10 kDagradient gel it is seen migrating down 15% of the gel length and in a 0-2% PEG 10 kDa gradient it is seen traveling down 30% of the gel length, suggesting a substantial increase in pore size.

3.7 Scanning electron microscopy Yet, no matter how accurate these data could be, the fact remains that no direct measurement of pore size has been given.We have attempted this using a scanning electron microscope (SE M), as previously reported by Blank and Reimschuessel [17], by Riichel et a/. [18] and Hsu and Cohen [ 191 (the latter two in the transmission mode). The results in Fig. 10 show that a control gel (5%T,4%C,no linear polymers added) presents fracture surfaces which are smooth and much like a solid wall, with no porous elements in sight. At a 0.5% added PEG 10 KDa (Fig. IOB), the gel fracture now appears pitted and irregular, but again with no visible clear ‘pores’. Finally, at a 2 % PEG kDa addition, porous surfaces are quite evident, with porosities in the 300-500 nm range. The most interesting ‘surface’ appears to be the one generated at a 2.5% PEG 10 kDa addition (the plateau concentrationj.In these gels (Fig. l0D) two intriguing features are most prominent: (i) the gel structural elements are organized in very thick fibers or bundles (with an average diameter of 200-300 nm) and (ii) ‘holes’ or ‘pores’ are now seen to be regularly distributed through the surface, with an average diameter of 500 nm. These structural changes are quite dramatic and they can be easily correlated with the sigmoidal turbidity transitions shown in Figs. 1-4. By correlating this evidence with the mobility of large DNA fragments, we have to surrender to the proof that such structural changes are real and not an artifact of the sample preparation procedure prior to SEM (see also Section 4).

4 Discussion So much gossip has been spread about polyacrylamide matrices since their use in electrokinetic methodologies that it is worthwhile to attempt to discern between facts and myths. Although we cannot guarantee that we will succeed, we shall nevertheless attempt to highlight some interesting general features. 4.1 Highly cross -linked or laterally aggregated gels?

Since the proposal of Ornstein and Davis [2,3], the only obvious way to produce large pore size gels (in a way, shifting the structure of a polyacrylamide gel towards that of an agarose gel) has been to prepare matrices with a high degree ofcross-links (typically> 20%C). It has been accepted up to now that high %C gels offer high porosities and this has always been associated to a transition from a clear to a highly turbid gel phase [12, 131. Moreover, highly cross-

linked gels are known to contain inelastic structures, consisting preferentially of intramolecular cycles, in turn associated with ‘clustering’or aggregation of individual thin fibers into ‘bundles’ [20].This occurs spontaneously during gelation of an agarose matrix: double helices are formed, which upon further cooling aggregate laterally into ‘pillarlike’ structures consisting of 7-11 helices [21]. As discovered by architects in the transition from Romanesque to Gothic buildings, abandoning a column in favor of a pillar (which can be envisioned as a n aggregate of columns) has a double effect: (a) it strengthens the supporting beams, thus allowing for creation of taller buildings, and (b) it is compatible with the production of larger spaces (or cavities) between the supporting beams. Agarose matrices spontaneously form such structures, while maintaining a strong coordination with surrounding water and avoiding opacity. On the contrary, polyacrylamides camouflaged as agaroses suffer from major drawbacks: (i) they are strongly opaque, (ii) they collapse in the gravitational field and (iii) they become strongly hydrophobic, exuding water and causing severe short circuits during electrophoresis.The last phenomenon is also correlated with the increased hydrophobicity of the cross-linker [22]: water is no longer coordinated along the polymer coil and keeps being spurted out from the gel matrix. Quite different is the situation oflaterallyaggregated gels: while they are also opaque, they are made with standard amounts of cross-linker (4-5 %C); therefore, they are not hydrophobic and do not lose solvent upon standing or in an electric field. There is thus a clear advantage associated with this last structure for u5e in electrophoresis.

4.2 On the pore size of hydrophilic gels We have tried to gather and summarize data on the real pore size of polyacrylamide gels. In the sixties, there was enormous interest in this topic due to the widespread use of polyacrylamides in protein separations. It was immediately understood that, because protein particles were effectively sieved in 5-10%T (typically S%C) matrices, their average pore diameter should be of the order of a handful of nanometers. We have collected a series of data in Table 1: there is reasonable agreement that a standard gel formulation ( 5 %T, 5%C) should have an average pore diameter of 5 nm, as measured by a wide variety of experimental procedures. There is only one notable exception: Blank and Reimschuessel[17] give arange of2-15 pm as average pore diameters for a 4%T, 5 %C gel. This seems to be off by some 3 orders of magnitude and does not quite make sense, considering that a more dilute gel (2.5 O/oT, 5 O/oC), which by definition should be more porous, is assigned a pore size of 1-10 prn (smaller!). In addition, Riichel e t a / . [18] who also used electron microscopy), while keeping tight-lipped about any pore size value, nevertheless clearly state that a 130 nm diameter viral particle could not penetrate a 5%T, 5%cgel. However, during additional work with the scanning electron microscope, they also found ‘cellular structures’ in the 1-2 pm diameter range, which, nonetheless, they refuse to accept as ‘sieving’ elements [31]. If Blank and Reimschuessel [ 171 were correct, even an E. colr cell (a particle of 0.5 x 2 pm size) would easily penetrate a 4%T, 5%C gel. This is obviously not the case; in fact, bare penetration of E. roll in any gel media has been reported only by Serwer el a / . [25] by using extremely dilute, almost liquid agarose

Elecrrophoresie

'1,;itcraIly dggregated' pnlgacrylarnide gels for electrophoresis

1992, / 3 , 587-595

gels (0.03 Oio). Howcvcr, it must be cmphasizcd that the effective pore sizes determined by electrophoresis of particles and by electron microscopy could be significantly different in the case of polyacrylamides: e.g., relatively large pores could be interconnected by relatively small channels, which would ultimately be the limiting factor for particle sieving (the "lakes and dire straits" model of Zimm, impersonal communication). In addition to highly diluted agarose matrices, the only other materials which could be compatible with high pore sizes are rigid ceramic bodies [24], which can reach 1-2 pm and even larger channels. In the case o f hydrophilic gels, polyacrylamides could be made substantially more porous by producing mixed-bed, polyacrylamide/agarose beads (Ultrogel AcA 54) 181, reaching porosities of ca. 250 nm. Thus, the unreasonably high pore values ofBlank and Reiinschuessel[17] could be due to artifacts or sample preparation prior to electron microscopy.

593

negligible thickness, i.e. r = 0, r being the fiber radius). In our case, given the substantial thickness of the gel fibers, their radius has to be subtracted form the computed value of the most frcqucnt pore size population [26].Thus:

R=~/GL-~ There appear to be two ways for accommodating larger objects in a gel network (thus increasing the pore size): either by reducing the fiber length or by decreasing the numbcr of fibers per unit volume (or both!). The manipulation of pore sizes in polyacrylamides by maintaining a constant OhT and progressively increasing%C seems to rely on the simultaneous shortening of L and thickening of the fiber diameter. Thus Fawcett and Morris [7] suggest that, while a 5 OhT gel has an average fiber diameter of 0.5 nm, a high %C gel has an upper fiber diameter value of 6 nm.

The procedure adopted by us is based on a quite different strategy. We drastically diminish the number of fibers per unit gel volume, while possibly not reducing the average L Ifwe now return to polyacrylamide matrices, how can their value (since %C is low and constant).Tliis is accomplished pore size be manipulated? According to the Ogston model by gelling the monomers'in presence of a hydrophilic poly[I01 mer. We believe that such a polymer (notably PEG 10 or R = 1/d%L 20 kDa) acts by sequestering the water to the growing chains and forcing them to form large bundles. held towhere R is the radius of a sphere which can be acco~nino- gether by. preferentially (but perhaps not solely), interdated within the 'open'spaces of a gel, Lis I/?the length of chain H-bonds. It is known that PEG coordinates large the gel fiber and n is the number of fibers/cm3 of gel vol- amounts ofwater around its coil (on the average, 1.5 water molecules per oxygen atom in the chain).This perturbation ume Note, however, that the above equation can only be of the solvent phase could force the growing polyacrylapplied for netwvrks consisting o r cerq long fibers having a

4.3 On the manipulation of gel pore size

Table 1. Average pore diameters in different types of gels Gel tvne 5 %IT,5 O

K

I 0 %T, 5 %C I S YaT, 5 9/oC 2 0 9/07, 5 "O 'C 35o:,T, 5%C 6.5 %T, 5 h C 8 8 '/Or,5 %C 12 %IT,5 %C 15 ?'or,5 YOC 20%T, 5 (!'oC 5 OioT. 5 OioC 7%T, 5 OhC 10 %T. 5 OO /C 12%T.S%C 18%IT,5 %C 7.5 %IT,5 YoC 30?laT. Su/cC 5 o:oT, 5 O K SOioT, 2.664'oC SephadCx G-200

Pore diameter

Met hod

3.6 nni 2.6 n m 2.0 n m 1.6 nm 0.5 n m 4.0 nm 3.2 11111

H20 permeability under prcssurc

I61

Chromatography o f marker proteins

I71

Electrophoresis of marker proteins

[231

Computed

121

Virus migration TEM~) Pellet ration of Hg under prcssurc SEM~'

1181 [I91

2.8

niii

1.1 n m 1.2 n m 8.5 n m 6.1 n m 50nm

4.2 tin1 3.2 n m 5.0 n m 2.0 n m