115:412/508 Proteins and Enzymes                                                                                     spring 2002

Assays II

I want to cover some standard methods of enzyme assay fairly quickly, and then  move on to methods for assaying proteins for which there is no cata­lytic assay.

Spectrophotometry: absorption of light at a specified wave length (col­orimetry is when the wave length range is only narrowed down by filters).  If a product absorbs uniquely and the spectrophotometer is hooked to a recorder this can be a continuous assay; also, modern spectrophotometers allow you to determine the rate directly through the instrument's software.  Spectrophoto­metry typically measures compounds in the range 10-3 - 10-6 M.  Two products are very commonly observed: the coenzymes NADH and NADPH, which have a molar extinction coefficient of 6200 L/mole·cm at 340 nm, and p-nitrophenol, which has an extinction coefficient of 18,300 L/mole·cm, but a pKa of 7.15, so it should be measured at a pH of at least 8.3 so that the absorption isn't critically depend­ent on the exact pH.  Many hydrolytic enzymes are assayed using p-nitrophenyl esters and glycosides, which are artificial substrates.  Many reac­tions are coupled to production of NADH or NADPH, or their disappearance, be­cause they are so convenient to observe.

Fluorimetry: some compounds when excited by higher-energy light, UV or visible, re-emit light lower-energy of a longer wave length.  (Diagram excita­tion and emission spectra.)  Because one is observing a little light against dark­ness, this can be very sensitive, down to 10-9 M; but fewer compounds fluo­resce, and the instrument is more expensive.  It can be more specific, since the wave length of both the exciting and emitted light can be selected.  NADH and NADPH fluoresce, and can be measured more senstitively thus.  Methylumbel­liferol and methylumbelliferylamine are the fluorescent equivalents of p-nitro­phenol; they fluoresce, but parent glycosides, esters and amides don't, so assays for glycosid­ases, esterases and amidases can use them as artificial substrates.  Fluorescamine can be used as a stop-time reagent for any amine set loose by amide or peptide hydrolysis.

I should also mention quenching of fluorescence by energy transfer.  If in close proximity  - meaning 10 Ċ or so - to a fluorescing group there is another group with an absorption band overlapping the emission peak of the first group, the energy of the excited electron in the first group can be transmitted to the second, and fluorescence of the first group will be quenched.  If the second group is also fluorescent, it will fluoresce at its emission wave length, and the efficiency of this energy transfer can be used to estimate the distance between the two groups (the dependence is on 1/r6, where r is the distance between the centers of the groups.  There is also dependence on their relative orientation, parallel to perpendicular.)  This can be a useful tool for studying conformational changes in molecules and distances between sites, but here I want to mention its use in assay.  If the second group is not fluorescent, there will just be quench­ing of the fluorescence of the first group if they are reasonably close together in the same molecule, as for instance at the ends of a short peptide.  Hydrolysis of the peptide will release group A from quenching by group B, and fluorescence will therefore appear.  This effect has been used to assay proteases, particularly pro­teases specific for a particular amino acid sequence (which can be synthe­sized between the fluorescence donor and acceptor).  It could also be used as an assay for RNAses and DNAses, but so far as I know has not been, even though it would be much quicker and more sensitive than current methods.

Closely related are luminescence assays, in which light is produced by the chemical reaction without prior irradiation.  The best known in assay of ATP by the firefly tail luciferin-luciferase reaction, for which ATP is a cofactor, but now of some importance is measurement of Ca++ by the jellyfish protein aequor­in, which undergoes a light-producing oxidative reaction when it binds Ca++.

Titration may seem high school chemistry, and Analytical Chemistry Lab, but there is an instrument called a pH-stat, which is essentially a pH meter con­trolling an automatic buret, so that if the pH drops below a set value base is added to restore it, and the amount of base added is recorded.  The same can be done for acid added to counter a rising pH.  Thus reactions producing acid, such as the hydrolysis of acetyltyrosine ethyl ester, a chymotrypsin substrate, can be followed continuously, laying a straight-edge on the stair-step chart recording of base addition vs time, or recording the total amount added over the period of linearity of the rate.

In principle reactions producing acid or base can be measured spectro­photometrically if an indicator is present, which in effect serves as a buffer of an otherwise unbuffered assay mixture, takes up or releases protons and changes color, the increase or decrease of absorb­ance being followed spectrophotomet­rically.  This is linear over only short pH ranges, and the equivalence between reaction and absorption change gets complicated if there is also a buffer pres­ent.  It is best used when high sensitivity is needed, such as stoichiometric displacement of a proton from a protein by a metal ion being bound.

Some now rarely used assays are manometric assays, measuring uptake or release of a gas in a closed system, the oxygen electrode measuring O2 concentration in solution, and viscometric assays measuring change, usual­ly decrease, in the viscosity of a solution as a polymer is broken down.  This last can be useful if you have a glycosidase which releases reducing sugar from a polysaccharide, but you don't know whether it is an endoglycosidase, breaking the polymer in the middle, or an exoglycosidase, chewing on the ends.  The endogly­cosidase will reduce the vis­cosity of a solution of the polymer, the exoglycosidase won't.  The same could apply to an aminopeptidase vs. a true protease (endo cleaving).

The next group of assays all depend on separation of product and start­ing material, the method of detection usually being non-specific.  Therefore they must be stop-time assays, the reaction must be stopped to allow separation.

Gas chromatography and liquid chromatography are similar in prin­ciple: reactant and product differ in the time they take coming through a column, whether as a volatile molecule in a flow of carrier gas or as a solute in a flowing liquid.  Gas chromatographs measure material passing the detector by flame ion­ization, or by capture of an electron from a radioactive source if they contain nitrogen or phosphorus; liquid chromatographs can detect spectrophotomet­rically, fluorimetrically, by other methods I don't know, or least sensitively but most broadly by change in the refractive index of the solvent when there is a solute.  The recorder can integrate the area under each peak to quantitate it.  The power and versatility is great; such methods are particularly useful when substrate and product are very similar chemically, as for instance cello­biose and glucose.  The drawback is the time each measurement takes, but in an enzyme assay, where you are looking for separation of only two compounds, this can be decreased.  But turnaround time is not likely to be less than 5 or 10 minutes.

Various other separation methods are in practice similar, such as capillary electrophor­esis and even gel electrophoresis, as for assay of a restriction endo­nuclease, or the gel shift assay.  But the longer the separation takes, the more you'd like to find another way.

Radioactive assays are very common.  A radioactive substrate is acted on by an enzyme; the product is in some way separated from the substrate, very carefully, and measured by its radioactivity.  This must therefore be a stop-time assay.  The counting takes a while, but can be done in a counter overnight.  Rad­ioactive assays are best when the separation is quick and complete, for instance incorporation of a small molecule into an easily precipitated polymer, as in protein and nucleic acid synthesis.

You should know the most important radioactive isotopes in biochemistry, and their half-lives: 3H, t1/2 = 12 years; 14C, t1/2 =5700 years; 35S, t1/2 = 87 days; 32P, t1/2 = 14.3 days; 33P, t1/2 = 25 days.  Note that N and O have no useful radioisotopes; the stable isotopes 15N and 18O can be measured accurately in mass spec­trometers, but this is not a routine assay procedure.  These isotopes all emit g radiation (electrons), in energy ratio H:C:S:P 0.1:1:1:10; tritium is therefore most difficult to detect, though in modern counters it can be counted at efficiency >60%; and 32P is the most hazardous to work with, because it can irradiate you right through usual vessels.  32P is usually used as a label for nucleic acids, though 33P is now also used, it has energy more like 35S, which is also sometimes used in the form of thionucleotides.  On the other hand, if you want to measure, say, protein synthesis in whole cells, you would use [35S]-methionine, because it wouldn't go into nucleic acids.

Note that, for equal concentrations of the radioisotopes, the one with the short half life will decay faster, more counts per minute per µmole, so that if you want extreme sensitivity you use a tritiated substrate at high specific activity rather than a 14C, as long as you have a good detector.  'Specific activity' of course is how much radioactivity per molar amount or weight of material; the specific activity on a molar basis of the product is normally the same as for the substrate, though as I shall mention in talking about kinetics this is not exactly so, especially for tritiated substrate, if the labeled atom is invcolved in the reac­tion.  One's raw data are 'counts' recorded by the counter, converted into dis­integrations per minute or Becquerels by knowing how efficiently the counter counts a standard in the same scintillation solvent.

I'm passing out an example of a radioactive assay, which shows some types of corrections that may have to be made.  They are assaying formation of geranyl pyrophosphate from dimethylallyl pyrophosphate and isopentenyl pyro­phosphate, using isopentenyl pyrophosphate labeled in the 1, the hydroxyl, position.  In the product the pyrophosphate ester linkage, allylic to the double bond, can be hydrolyzed by added acid, while in the substrate the ester is not allylic to the double bond and is not hydrolyzed.  Thus the [14C]geraniol formed by enzymatic reaction and hydrolysis is specifically extracted into heptane and counted.  (They also state that free isopentenol is not extracted into heptane, because ethanol as well as HCl is added for hydrolysis.  It is extracted in the phosphatase assay, in which they add neither ethanol nor HCl.  If it is not ex­tracted by heptane in the standard assay, this gives protection against apparent activity due to isopentenol production by phosphatases.)  They have two prob­lems: first, the substrates are hydrolyzed by phosphatases present, decreasing the substrate concentra­tion; second, the isopentenyl pyrophosphate is isomer­ized into di­methylallyl pyrophosphate, which is allylic and hydrolyzed, putting extra counts into the heptane phase.  They can deal with the first problem by including in the assay fluoride ion, which is a fairly general inhibitor of phosphat­ases, but could not inhibit the isomerization except by purifying their enzyme away from it.

Notice that they defined a unit - 1 nmol of product per minute - but didn't use it in Table I, where product is reported in Becquerels = disintegrations per minute.  They were using very little radioactivity, only 167 Becquerels, so a for­mation of 81.3 Bq is about 50% conversion of substrate to product.  I wondered, is the isopentenyl pyrophosphate concentration, initially 4.5 nmol/0.1 ml, below the Km of the enzyme?  4.5x10-9 mol/10-4 L = 4.5 x 10-5 M.  No.

Another category are biological assays, which measure a more speci­fic biological action of the protein rather than a chemical action.  Biological assays are particularly important when the chemical event carried out is unremarkable, the importance being in its specificity - such as the many proteases which cause specific biological effects by cleavage of one or a few bonds in specific protein substrates.  I gave you one example, assay of growth hormone by thickness of the knees of hypophysectomized rats.  Another is the blood clotting system, the result of a cascading series of proteolytic events, the product of each reaction being the enzyme for the next, culminating in the cleavage of fibrinogen to fib­rin and the formation of the clot.  The clotting assay is timing how long it takes a clot to form; it may be used as an assay for any of the factors involved if you have a serum deficient in that factor so that the sam­ple added supplies what is needed to bring about clotting.  This includes Factor VIII, the antihemophilia factor which many of Queen Victoria's descendents lacked, and Factor V; these are both protein cofactors of the system, not enzymes.

Coupled assays are those in which at least one additional reaction is in­cluded.  Most commonly this is done so that a product is generated which can be measured specrophotometrically, often continuously.  Two examples:

d-alanine + O2 _ pyruvate + NH4+ + H2O2 (d-amino acid oxidase)
H2O2 + chromogen _ colored product + H2O  (peroxidase)

ATP + H2O _ ADP + Pi  (ATPase: any enzyme using ATP)
phospho(enol)pyruvate + ADP € pyruvate + ATP (pyruvate kinase)
pyruvate + NADH + H+ €  lactate + NAD+ (lactate dehydrogenase)

In each case one product of the reaction which you want to measure is used as a substrate for one or more subsequent reactions, culminating in one which can be measured easily and continuously, usually spectrophotometrically or fluorimetrically.  If you want to do this continuously you need a relatively large amount of the subsequent enzymes, so that they will operate efficiently even though the concentration of their substrate, the product of the prior reaction, is well below their Km.  Immediately after the reaction is started the observed reaction will proceed more slowly, as the concentration of the inter­mediate product, peroxide or ADP, builds up to its steady state level at which it is being produced and utilized at the same rate ((draw).  Scopes gives a rule of thumb that to achieve 98% of maximum rate within 5 min an amount of coup­ling enzyme calculated by Vm/Km = 5 min-1  is needed, but suggests using more; see p. 66.  The article by Rudolph et al. in Methods in Enzymology v. 63 gives full treatment of calculations for this, but it may be simpler just to try decreas­ing amounts of coupling enzyme until you find the minimum amount giving maximum rate in what you consider a short enough time.  If the coupling enzymes cost more than you want to spend, you might use them in a stop-time assay rather than a continuous assay, because you need much less.

Two other reasons for coupled assays are illustrated in the assay of GPDH, glyceraldehyde-3-phosphate dehydrogenase, which we used to do in the lab:

fructose-1,6-bisphosphate €  glyceraldehyde-3-P + dihydroxyacetone-P
glyceraldehyde-3-P €  dihydroxyacetone-P
glyceraldehyde-3-P + NAD+ + Pi €  1,3-diphosphoglycerate + NADH + H+
1,3-diphosphoglycerate+ ADP €  3-phosphoglycerate + ATP

The glyceraldehyde-3-phosphate dehydrogenase reaction produces NADH directly, but 1) glyceraldehyde-3-phosphate is an expensive & labile substrate, 2) the GPDH reaction does not go far toward completion, indeed the equilibrium lies toward the reactants.  We deal with the first problem by gener­ating the substrate in the reaction mixture from cheaper, more stable fructose-1,6-bis P, and with the second problem by coupling the GPDH reaction with the phosphoglycerokinase reaction whose equilibrium does lie far to the right, pulling the over­all reaction toward completion.  A cheaper way to do this is to use arsenate rather than phosphate; 1-arseno-3-phosphoglycerate is formed but is unstable, breaks down as fast as it is formed, so that the reverse reaction cannot occur and pro­duction of NADH continues without equilibrium being reached.

Cycled assays use a small amount of a compound as rate-limiting inter­mediate in reactions going both ways.  Strictly these are assays for the com­pound rather than for an enzyme, but the amount of compound started with might be the product of an enzyme reaction carried out on a very small scale, say one cell.  An example, from a paper by Valero & García-Carmona, Anal. Biochem. 239:47-52 [1996], which derives equations to optimize the assay:

pyruvate + NADH + H+ Ĉ l-lactate + NAD+ (lactate dehydrogenase)
pyruvate + H2O2 ¨ l-lactate + O2 (lactate oxidase)

In this case the amount of pyruvate (0-2 nmol, in a volume of 1 ml) controls the rate of the observed reaction (oxidation of NADH), in presence of 0.25 mM NADH, 1.8 µg lactate dehydrogenase and 60 µg lactate oxidase.  If the reaction is started by addition of pyruvate rather than enzymes, a 'blank' rate of NADH oxidation is observed, indicating that one of the enzymes either contains pyruvate or lactate or has NADH oxidase activity.

Oliver Lowry, who invented coupled assays and wrote a book about them (A Flexible System of Enzymatic Analysis, Academic Press: New York, 1972), used them more as stop-time assays, for instance of NAD+ (which is destroyed by heating in base) or NADH (which is destroyed by heating in acid); enzymes and oxalacetate are also destroyed by boiling.  One would incubate NAD+ or NADH with 0.3 M ethanol, 2 mM alcohol dehydrogenase, 50 µg/ml alcohol dehydrogenase and 5 µmg/ml malate dehydrogenase: the reactions are

EtOH + NAD+ Ĉ CH3CHO + NADH + H+
oxalacetate + NADH + H+ Ĉ malate + NAD+

These reactions are let run for 1 hr, then the reaction is stopped by boiling, and the amount of malate present is measured using malate dehydrogenase in pres­ence of a system to remove the oxalacetate formed so that the reaction will go:

malate + NAD+ _ oxalacetate + NADH + H+
oxalacetate + glutamate Ĉ aspartate + a-ketoglutarate (transaminase)
or oxalacetate + H2NNH2 Ĉ H2N-NH=C(COO-)-CH2COO- (non-enzymatic)

NADH is then measured spectrophotometrically or fluorimetrically.  For greatest sensitivity, the 1st reaction is carried out in a volume of 0.4 µl under oil, boiled, the second reaction carried out, then boiled in 0.1 M NaOH to destroy NAD+, and the first and second reactions repeated in larger volumes to generate a measura­ble amount of NADH.  This can measure 10-15 mole NAD+ or NADH.  Further ref­erences: Lowry et al., J. Biol. Chem. 236:2746-2755 (1961); Kato et al., Anal. Bio­chem. 53:86-97 (1973); McDougal and Dargar, Anal. Biochem. 97:103 (1979).

The same thinking is involved in assay of enzymes which activate other enzymes, for instance plasminogen activators such as urokinase: the activating enzyme is incubated with a small volume of plasminogen for a period of time, then this is used in a plasmin assay, for instance hydrolysis of _-methyl-_-tosyl-lysine p-nitrophenyl ester: each action of the activator on plasminogen results in many p-nitrophenol molecules being produced.

Immunological assays

I have mentioned methods which are specific for a particular protein without being specific for its function, such as observation of a specific band in gel elec­trophoresis; I’d call them non-functional assays, while those based on the real activity of the protein would be functional assays.  They are in principle stoichi­ometric rather than catalytic, but some can be made catalytic by using an enzyme as the eventual reporter of the amount of protein present.  The most obvious are assays based on recognition of the protein by antibodies.  One may use either polyclonal or monoclonal antibodies; monoclonal are more specific, but polyclonal anti­bodies, which are a mixture recognizing different parts of the antigen protein, are sometimes preferable, as well as easier to produce.  If you are looking for a cloned protein produced in E. coli or other host cell, you would pretreat the antibody preparation with an extract of E. coli not contain­ing the cloned protein, to remove any antibodies against normal E. coli proteins.

Originally such methods were limited by the need to have a sample of pure protein to use as antigen to induce antibodies, but this is now evaded in two ways.  If you have partially purified protein, say 30% pure, you can have mono­clonal antibodies made, and check individual clones until you find one which binds your protein (this is most easily done if you do have a functional assay for your protein, and can determine that it has been removed from a mixture by bound antibody).  Then you use this, or if you have two use them both in a 'sand­wich assay' as I shall describe, as a more general assay for the protein.

Secondly, if you have cloned an open reading frame which represents some protein, either of unknown function or unassayable, you can have peptides syn­thesized which represent parts of the sequence and raise antibodies to these peptides.  Since there are 9.5 x 1013 possible sequences for pentapeptides, it is easy to make peptides which should be fairly specific for one protein, though of course you would pick hydrophilic sequences expected to be on the surface of the protein.  The immune system responds better to whole proteins rather than small molecules, which in this context are called haptens, so it is best to attach the peptide or other small molecule to some common protein, raise antibodies, and treat with the unmodified protein to remove those which bind to it.

The technology has evolved from the original competitive radioimmunoassay, through enzyme-linked immunoassays (ELISA for short) to sandwich-type assays which respond in direct proportional to the amount of antigen present.

The competitive assays require a sample of the antigen which is labeled in some specific way, with radioactivity or by attaching an enzyme.  Proteins can be labeled by acetylating with 14C-acetic anhydride or iodinating using Chlor­amine T (as described in Rosenberg p. 31), and raising antibodies to labeled protein; or they can be labeled by chemically attaching some enzyme, whose eventual pro­duction of product is measured.  Horseradish peroxidase and alka­line phosphatase are the enzymes used for such purpose, since they produce products with very high extinction coefficients.  All the methods require that the antigen-antibody complex can be separated from the solution phase, either by precipitation, using Protein A from Staphylococcus aureus or Protein G from streptococci, attached to agarose beads or cells, or prior immobilization of the antibody.

Radioimmunoassay is diagrammed in the handout.  In drawing A, an amount of radiolabeled antigen just equivalent to the amount of antibody (here expres­sed as 6 sites) is added; all the antigen is bound, and hence after precipitation of antibody no antigen is left in the supernatant.  In B, an amount of unlabeled ant­igen half the amount of labeled antigen is added, making a total of 9 equivalents; the antibody binds 6 of these, and of course does not distinguish between unlab­eled and labeled.  So one-third of the labeled antigen remains in solution.  In C, 6 equivalents each of labeled and unlabeled antigen have been labeled; the anti­body binds 3 of each, and hence 3 equivalents of labeled antigen, half the total, are left in solution.  In D 12 equivalents of unlabeled antigen are added, plus 6 labeled, so 2/3 of the label remains in solution.  In each case the bound and free antigen are separated as mentioned above, and the amount of labeled antigen in either phase measured.  In radioimmunoassay the solution phase is generally counted, but in ELISA carried out in a 96-well plate the wells are washed to remove unbound antigen, and then substrate for the attached enzyme is added; the amount of product produced by bound antigen-linked enzyme is measured, often by a "plate reader" which can measure the absorbance of solution in the wells.  The results are generally presented as amount of label bound, even if determined as total labeled antigen minus free.  Figure 1 at the right shows that direct plot of this vs. amount of unlabeled antigen added yields a curved line approaching but not reaching zero.  The results can be made a straight line by plotting free label over bound, as in Fig. 3, or plotting natural log of % bound over %free, the logit function, as in Fig. 4.

Radioimmunoassay tends to be limited by the specific radioactivity of the labeled antigen.  ELISA, enzyme-linked immunoassay, using enzyme attached to the antigen in a way which does not affect the enzyme's activity, can be made more sensitive by increasing the time of incubation with substrate.

Sandwich-type immunoassays require antibodies to the antigen produced in two different animals, say rabbit and mouse, plus antibodies to immunoglobu­lins of one of these animals.  (I'm giving you the example I know; I think there are various other ways of doing this.)  One antibody, say a monoclonal made in mouse cells, is immobilized in the wells of the plate by overnight incubation.  Free antibody is washed away, and solution of the antigen being measured is added and allowed to bind for two hours or so.  The wells are washed again, and the second antibody solution is added, a rabbit polyclonal antibody which binds to other parts of the antigen's surface.  After incubation free antibody is washed away; the amount of this antibody bound is proportional to the amount of anti­gen bound to the first antibody.  Then goat anti-rabbit immunoglobulin, with an enzyme linked to it, is added; this binds in proportion to the amount of rabbit antibody bound.  After another wash, substrate for the enzyme is added.  The amount of product produced is proportional to the amount of enzyme bound, which is proportional to the amount of goat anti-rabbit antibody bound, which is proportional to … - you get the idea.  "Western blot" technology for measuring a protein after gel electrophoresis, or simply after adsorbing to a surface such as a nitrocellulose membrane, is similar, except that the first antibody isn't needed, the antigen binds directly to the surface, the second antibody binds to it, etc.

Similar technology is used for other molecules for which there are specific tight-binding proteins, such as cyclic AMP-binding protein, avidin and streptavi­din for biotinylated proteins, lectins for glycosylated proteins, etc.  The binding proteins themselves can be meas­ured in such an assay using immobilized ligand to adsorb them from solution.

Assay by radioactive tracer

The essential idea is that if you have a sample of the protein you want to purify which is radioactively labeled, you can add it to crude extract and follow the purification of the protein by the label.  This derives from procedures such as purifying proteins containing coenzyme B12 by growing the bacteria in pres­ence of 60Co++ and following the radioactivity through purification.  C.C. Marvel & H.O. Kammen, in a paper "Purification of Plasmid-Expressed Proteins Which Lack Functional Assay Systems", Anal. Biochem. 181, 336-340 (1989), applied this to the purification of a protein corresponding to an unknown cloned open reading frame, whether of known genetic but unknown biochemical function, or simply a neigh­bor in the same operon.  They use as marker for purification, from the normal source or from cells in which the gene has been cloned, radiolabeled protein, prepared in vitro from the cloned gene.  To save time, I shall skip over most of the worries involved - if you are really interested, consult the paper.

The radiolabeled protein might be produced in a maxicell or minicell system, but they work with a cell-free invitro expression system.  This is good for bac­terial proteins [they work with E. coli proteins]; eukaryotic proteins will not be appropriately modified post-translationally.  Presumably appropriate eukaryotic systems (baculovirus in insect cells, CHO cells) could be used for eukaryotic pro­teins, but there is no way to guarantee that the protein produced in vitro will have the same modifications as that in the natural tissue source.  The system will produce also other proteins coded in the plasmid, e.g. b-lactamase if the plasmid carries ampicillin resistance; this is a problem if the desired protein is expected to be the same size (31,000 d for b-lactamase).  If so, use a plasmid with a different resistance marker.  Production of a protein of the molecular weight expected from the length and sequence of the open reading frame should be verified by SDS gel electro­phoresis [occasionally, aberrant mobility is a prob­lem.  Run-on or run-off trans­lation - i.e. including some of the vector sequence - should also be guarded against.]  The radiolabeled protein should then be puri­fied by standard methods, but only enough to rid it of other radioactive proteins produced in the same system.  Since one is generally working with a small volume, hplc is a useful technique here.

The radiolabeled protein is then used as a marker for the non-radioactive pro­tein dur­ing purification, either adding it at the beginning and following throughout, or at each indi­vidual step.  To get a lot one would probably use as source E. coli or other cells trans­formed with the plasmid containing the gene for the protein, but one would also like to know whether the protein is actually expressed in the original tissue from which the gene was isolated.  Eventually the protein being purified should become evident as a band in gel electrophore­sis corresponding to the radiolabeled band.  Thereafter, the gel becomes the criter­ion of purity, until only the one band is seen.  [Its N-terminal amino acid sequence is then determined by gas phase protein sequenator analysis (which will also tell if more than one protein is still present, by giving more than one amino acid at each step).  The sequence should correspond to that predicted from the DNA sequence.  (It is assumed that both radiolabeled and cold protein have undergone the same process­ing; the amino terminus may not correspond to the beginning of the open reading frame, but should be findable.)  If necessary the purified protein is cleaved and peptide sequences analyzed.]

The hard part is trying to identify a function for the protein.  If one has puri­fied other proteins coded in the same operon, one can test whether it binds to any of them, or affects their action on their substrates.  Using radiolabeled pro­tein, one could determine whether it binds any other proteins in a crude extract, e.g. by determining whether electrophoretic mobility in a non-denaturing gel is decreased by presence of crude extract, or mobility in gel filtration increased.  Similarly, one could test whether it binds to DNA or RNA, or has nuclease, prot­ease or ATPase activity.  Other possibilities might be suggested by the other pro­teins in the operon.