Analytical Biochemistry 447 (2014) 82–89

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/locate /yabio

Semi-quantitative colony immunoassay for determining and optimizingprotein expression in Saccharomyces cerevisiae and Escherichia coli

0003-2697/$ - see front matter � 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.ab.2013.10.020

⇑ Corresponding author.E-mail address: e.sattlegger@massey.ac.nz (E. Sattlegger).

1 Current address: Laboratory for Evolution and Development, Department ofBiochemistry, University of Otago, Dunedin 9054, New Zealand.

2 Current address: Laboratory of Gene Regulation and Development, NationalInstitute of Child Health and Human Development, National Institutes of Health,Bethesda, MD 20892, USA.

3 Abbreviations used: SDS, sodium dodecyl sulfate; QSCI, quick semi-quacolony immunoassay; LB, Luria–Bertani; IPTG, isopropyl b-D-1-thiogalactopyPAGE, polyacrylamide gel electrophoresis; TBS, Tris-buffered saline; TBS-T,Tween 20; CCD, charge-coupled device; GST, glutathione S-transferase; Gfluorescent protein; Gcn1, general control non-derepressible; IgG, immunoglYih1, yeast IMPACT homologue; hc, high copy; lc, low copy.

Andrew G. Cridge 1, Jyothsna Visweswaraiah 2, Rashmi Ramesh, Evelyn Sattlegger ⇑Institute of Natural and Mathematical Sciences, Massey University, Auckland 0745, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history:Received 5 July 2013Received in revised form 21 September 2013Accepted 11 October 2013Available online 28 October 2013

Keywords:Colony WesternColony immunoblottingProtein overexpressionColony expression screeningSaccharomyces cerevisiaeEscherichia coli

This work describes a quick semi-quantitative colony immunoassay (QSCI) method for immunoblotdetection of intracellularly expressed proteins in both yeast and bacterial cells. After induction of proteinexpression, only 4.5 h is required for cell breakage, protein detection, and data analysis. This protocol wasused to screen and unambiguously identify Saccharomyces cerevisiae cells efficiently overexpressingglutathione S-transferase (GST)-tagged Yih1 in addition to cells expressing the myc-tagged large 297-kDa Gcn1 protein. In addition, the method was used to identify Escherichia coli cells efficiently expressingHis6-tagged Yih1 and a GST-tagged Gcn1 fragment, respectively. The protocol allows the use of bothepitope-specific and protein-specific antibodies. The same colony immunoassay can also be used todetermine the minimal concentration of inducing agent sufficient for induction of optimal proteinexpression (e.g., galactose for yeast, isopropyl b-D-1-thiogalactopyranoside [IPTG] for E. coli). To ourknowledge, this is the first report on a rapid low-cost procedure that allows the calibration of inducingagent on solid medium.

� 2013 Elsevier Inc. All rights reserved.

Many studies involve the generation of yeast and bacteria cellsthat express or overexpress a specific protein; however, tests arerequired to verify that the desired protein is indeed expressed.Cells transformed with the same overexpression plasmid some-times express proteins with various efficiencies, making it neces-sary to identify the one cell colony with the highest expressionlevels [1–4]. A collection of cells harboring a library of mutatedproteins requires screening to identify ‘‘negative’’ cells that expresstruncated versions of the protein due to a frame shift or due to theintegration of a translational stop codon in the open reading frame[5]. Such screening procedures may require the analysis of up to1000 or more clones [5]. Protein overexpression is a huge burdenon cells; therefore, mutagenic events that abolish the overexpres-sion give cells a growth advantage over the rest of the population[1,2]. As a consequence, the mutated cells outgrow the others andreduce the level of protein expressed. Thus, for some proteins, it isnecessary to routinely monitor whether cells still overexpress thedesired protein efficiently to allow validation of experimentalresults.

The standard procedure to test for protein expression is togenerate protein cell extracts and resolve them in sodium dodecylsulfate (SDS)3–polyacrylamide gels, followed by immunoblotting[5,6]. This protocol, although reliable, is laborious and time-consum-ing, in particular when large numbers of samples need to be pro-cessed. Therefore, there is a need for a fast method that canidentify cells that express protein efficiently and/or in full lengthand that allows for rapid verification of whether cells still expressthe desired protein.

When reviewing the literature, we found that for Saccharomycescerevisiae, several protocols have been published previously forcolony immunoassays. Tomlinson and Esser [7] used lyticase andDNase in the cell breakage procedure in combination with anincubation in 5% SDS at 37 �C for 30 min, a temperature that maybe advantageous to protease activities resulting in protein degra-dation. Holz and coworkers [8] used 4 M NaOH for 1 h for cellbreakage, making a subsequent neutralization step necessary.The quickest procedures available to date are based on the discov-ery that NaOH treatment in combination with heat treatment inthe presence of SDS leads to yeast cell lysis [9,10]. Cells are lysedwith 0.1% SDS, 0.2 M NaOH, and either 35 mM dithiothreitol or

ntitativeranoside;

TBS andFP, greenobulin G;

Semi-quantitative colony immunoassay / A.G. Cridge et al. / Anal. Biochem. 447 (2014) 82–89 83

0.5% b-mercaptoethanol for 30 min (e.g., see Refs. [11–13]). All ofthe above protocols, except that published by Tomlinson and Esser[7], involve growing cells on solid medium and transferring themto nitrocellulose membrane, which is then placed and incubatedon inducing medium. Using unsterile membranes in this transferincreases the risk of introducing contaminations. In addition, incontrast to our protocol described here, the process of transferringdoes not allow for the combination of colony immunoassays withphenotypic screening processes involving serial replica plating onvarious media. Furthermore, when transferring the colonies fromthe plate to the dry nitrocellulose membrane, the membrane alsotakes up media components such as glucose. Glucose may counter-act the inducing agent galactose when the membrane is trans-ferred to galactose-containing plates for induction of proteinexpression, thereby delaying or preventing efficient proteinexpression. Proteins can also be released from cells by immersingthe colony-bearing membrane in liquid nitrogen [14]; however,nitrocellulose membranes break easily, the nitrogen may dispersethe cells or proteins, and not every laboratory has access to liquidnitrogen.

For Pichia pastoris, the so-called ‘‘yeastern’’ colony blot assay isavailable and also involves the transfer of cells from solid mediumto membranes, followed by immunoblotting [15]; however, in con-trast to our protocol, it detects only secreted proteins.

Several protocols for rupturing membrane-attached Escherichiacoli cells have been published; however, these require the use ofenzymes [16], require exposure to chloroform followed by a 12-to 16-h incubation in the presence of lysozyme [5], or requirethe sequential incubation of the colony-bearing membrane inSDS, denaturation solution containing NaOH, 2� neutralizationsolution, and 2� SSC (sodium chloride–sodium citrate) buffer(from QIAexpressionist, Qiagen).

In an effort to find a fast and cost-efficient way to screen multi-ple colonies for the ones with satisfactory (over)expression of a de-sired protein, we found that the current protocols can be furthersimplified in that incubation times could be shortened further; thatincubations at increased temperature are unnecessary, therebyreducing the risk of protein degradation; that only a few standardchemicals are needed; and that costly enzymes are not required.For simplicity, our refined protocol is referred to as quick semi-quantitative colony immunoassay (QSCI). In addition, we estab-lished that the same procedure could be used to determine theoptimal concentration of agents used to induce protein overex-pression, thereby reducing the consumption of sometimes expen-sive inducing agents. We optimized our method for S. cerevisiae,and when testing the protocol on E. coli we have discovered thatit also works well for this organism.

Materials and methods

Strains and plasmids

E. coli and S. cerevisiae strains and plasmids used in this studyare listed in Tables 1 and 2, respectively. Vectors used werepRS316 and pRS426 [17], pES128-9 [18], pEMBLyex4 [19], pET-28a (Novagen), pETDuet (EMD Millipore), and pGEX-6p-3 (GEHealthcare Life Sciences).

Plasmid pES21-3 was generated by extracting the SacII–SalIfragment from p1831 [20] and inserting it into similarly digestedp1827 [18].

Growth conditions

Yeast strains harboring plasmids for the overexpression of adesired protein were grown on synthetic dextrose (SD) medium

containing 1.9 g/L yeast nitrogen base lacking amino acids andammonium sulfate (Formedium), 5 g/L ammonium sulfate (Ajax,Thermo Fisher Scientific), supplements as required by the yeaststrain, 20 g/L glucose (Sigma), and (if required) 20 g/L agar (Forme-dium). For the galactose-induced overexpression of proteins, theinduction medium that was used contained the same ingredientsas above except that 20 g/L galactose (Formedium) was used as acarbon source instead of glucose. In addition, 20 g/L raffinose (For-medium) was added to the medium if the amount of galactose wascalibrated for protein induction (i.e., if the galactose concentrationwas <20 g/L).

E. coli strains harboring a plasmid for the induced expression ofthe desired protein were grown on Luria–Bertani (LB) mediumcontaining 10 g/L NaCl (Ajax, Thermo Fisher Scientific), 10 g/L tryp-tone (Formedium), 5 g/L yeast extract (Formedium), standardamounts of antibiotics for plasmid maintenance (ampicillin andkanamycin [5]), and (if required) 20 g/L agar. Protein expressionwas induced by adding isopropyl b-D-1-thiogalactopyranoside(IPTG, Formedium) to the medium at the indicated concentration.

Induction of protein expression

To induce protein expression, single colonies were patched ontosolid growth medium. Alternatively, 5 ll of saturated overnightcultures derived from single colonies were transferred onto themedium. The cells were incubated overnight at 30 �C (yeast) or37 �C (E. coli) and replica plated onto inducing medium, incubatedagain overnight, and then immediately lysed to ensure efficient cellbreakage.

To optimize the concentration of inducing agent, plates wereprepared on a level surface, each containing exactly 25 ml of solidmedium with increasing amounts of inducing agent. Liquid over-night cultures of single colonies were prepared, and 5 ll of the cul-tures was transferred in a single row onto each inducing plate anda non-inducing plate as a control (see Fig. 3A, image 1, in Resultsand Discussion below). The plates were incubated overnight at30 �C (yeast) or 37 �C (E. coli), such that the cells had just grownup (�12–15 h for yeast and 8 h for E. coli). Using a scalpel, the rowsof grown cells were cut out, and each strip of solid medium wastransferred onto a glass plate with a spatula (see Fig. 3A, images2 and 3). The agar pieces were aligned next to each other from low-est to highest concentration of inducing agent. This allowed simul-taneous transfer of all colonies onto one nitrocellulose membrane(see Fig. 3A, image 4), as described in more detail in the nextsection.

Cell lysis

An outline of this procedure is shown in Fig. 1A. In step 1, a drynitrocellulose membrane (e.g., Bio-Rad, 162-0150) was carefullyplaced onto the colonies, and light pressure was applied to bringthe cells into direct contact with the nitrocellulose membrane. Instep 2, after 5 min, the nitrocellulose membrane was carefullypeeled off the plate, placed colony side up onto a paper towel,and left to dry for at least 5 min. In step 3, for cell lysis, two layersof 3MM filter paper (Whatman) or a 3-mm stack of well-absorbingpaper was soaked in lysis buffer (10� Laemmli SDS–PAGE [poly-acrylamide gel electrophoresis] running buffer: 250 mM Tris base,1.92 M glycine, and 1% [w/v] SDS) or Tris-buffered saline (TBS:10 mM Tris–HCl [pH 7.3] and 100 mM NaCl supplemented with1% [w/v] SDS). Excess liquid was removed, and the well-driednitrocellulose membrane was placed on top of the damp filter pa-per colony side up for 5 min. The membrane was left to absorb thelysis buffer, resulting in cell lysis that was visible by the coloniesadopting a shiny and slimy appearance. In step 4, the membranewas carefully lifted away from the filter paper and placed on top

Table 1Strains used in this study.

Strain Genotype Source

E. coli strainBL21(DE3) F-ompT hsdSB(rB-, mB-) gal dcm (DE3) Novagen

Saccharomyces strainsH1511 strain backgroundH1511 MATa ura3-52 trp1-63 leu2-3,112 GAL2+ [33]H2556 as H1511, and gcn1D [18]

BY4741 strain backgroundBY4741 MATa his3D0 leu2D0 met15D ura3D0 Life TechnologiesMSY-1-1 MATa leu2D0 met15D ura3D0 gcn1D::KanMX [22]GCN1-GFP as BY4741 and GCN1-GFP::HIS3MX6 Life TechnologiesPGK1-GFP as BY4741 and PGK1-GFP::HIS3MX6 Life Technologies

W303 strain backgroundKW195 MATa ura3 leu2 (his3) HTB2-mCherry::His5+ O’Cohen–Fix

Table 2Plasmids used in this study.

Plasmid Gene Relevant features Vector Source

Yeast genes under their own promoterp2367 GCN1-myca Amp, URA3, CEN6/ARSH4 pRS316 [20]p1834 GCN1-myca Amp, URA3, 2l pRS426 [18]

Yeast gene fusions, under GAL1–CYC1 promoterpES187-B1 GSTb-YIH1 Amp, URA3, 2l pES128-9 [22]pES21-3 GCN1–myca Amp, URA3, leu2d, 2l pEMBLyex4 This study

Yeast genes for expression in E. colipES189-D1A His6b-YIH1 Kan pET28a [22]pES123-B1 GST-Gcn1[2052–2428]c Amp pGEX-6p-3 [18]

a Epitope tag at the C terminus of the open reading frame.b Epitope tag at the N terminus of the open reading frame.c Numbers in brackets indicate amino acids encoded be the respective gene.

84 Semi-quantitative colony immunoassay / A.G. Cridge et al. / Anal. Biochem. 447 (2014) 82–89

of a dry paper towel and dried for 5 min or longer to ensure thatthe liberated proteins came in contact with the membrane. In step5, the well-dried membrane was then vigorously washed (highestrevolutions per minute [rpm] possible that does not lead to spillageof liquid, i.e., �120 rpm) in distilled water for 1 min. The washingwas repeated until all visible cell debris was removed from themembrane and the water remained clear. In step 6, the membranewas washed (using a shaker) three times in TBS-T (10 mM Tris–HCl[pH 7.3], 100 mM NaCl, and 0.1% [v/v] Tween 20) for 3 min toensure that all cell debris was removed from the membrane. Instep 7, to assess whether sufficient cell breakage had occurred,the membrane was immersed in 0.1% (w/v) Ponceau S solution(Sigma–Aldrich, P3504) and 5% (v/v) acetic acid for 10 min andthen washed in 5% (v/v) acetic acid until no background stainingwas visible. The result was documented using a scanner, a copymachine, or a camera.

Immunoblotting and semi-quantitative analysis of protein expressionlevels

The nitrocellulose membrane carrying the proteins of the lysedcells was first blocked in TBS-T buffer containing 5% (w/v) low-fatmilk powder (standard household non/low-fat dry milk powdersuch as Pam’s Instant Skim Milk Powder 0.5% fat [Pam’s Products]or Carnation non-fat milk powder [Nestlé Canada]) for 30 min atroom temperature with gentle agitation. Then the membranewas incubated with the primary antibody diluted in TBS-T contain-ing 5% (w/v) milk powder for 1 h at room temperature with gentleagitation. After washing the membrane three times with TBS-T for5 min, the membrane was incubated with the horseradish peroxi-dase-linked secondary antibody diluted in TBS-T containing 5% (w/

v) milk powder (following the manufacturer’s protocol for anti-body dilution) for 1 h at room temperature with gentle agitation,followed by three 5-min washing steps and one 10-min washingstep using TBS-T. It is important that the primary antibody haslow cross-reactivity to proteins of the host organism.

For detection of the horseradish peroxidase, the membrane wastransferred into a plastic bag, the bag was sealed on three sides,and the excess liquid was removed. The chemiluminescencesubstrate (e.g., SuperSignal West Pico, Thermo Scientific, Pierce-24080) was prepared according to the manufacturer’s protocol(usually 0.01 ml/cm2) and transferred into the bag, air bubblesbetween the blot and the bag were removed, and the bag wassealed. The membrane was incubated with the substrate for2 min while moving the substrate solution back and forth in thebag, and the blot was imaged using X-ray films (X-Omat BT Film,Kodak, 866 9400) or a cooled digital camera (charge-coupled de-vice [CCD] camera: FujiFilm LAS-4000). The manufacturer’s manualinstructs to remove the excess substrate from the membranebefore imaging, however, we discovered that when using a CCDcamera this is not necessary as long as the substrate volume is keptto a minimum and all air bubbles were removed from the bag.

Primary antibodies used in this study were anti-GST (glutathi-one S-transferase) (1:1000, Santa Cruz Biotechnology, Z-5,sc-454), anti-His6 (1:200, Santa Cruz Biotechnology, H-15, sc-803), anti-myc (1:500, Roche, clone 9E10, order code 11 667 203001, or Santa Cruz Biotechnology, 1:500, 9E10, sc-40), anti-GFP(green fluorescent protein) (1:200, Santa Cruz Biotechnology, sc-8334), anti-yeast Pgk1 (1:5000, Life Technologies, 459250), andanti-Gcn1 (general control non-derepressible) (HL1405, 1:1000[21]). Secondary antibodies linked to horseradish peroxidase weregoat anti-mouse immunoglobulin G (IgG) peroxidase conjugate

Fig.1. Fast screening for cells overexpressing a desired protein. (A) Overview of the quick semi-quantitative colony immunoassay (QSCI) procedure. For more detail, see text.(B) Yeast strain H1511 harboring empty vector (pRS316), or a plasmid that allowed the galactose-inducible overexpression of GST–Yih1 (plasmid pES187-B1) or GST alone(pES128-9), was patched on solid medium and grown overnight. This master plate was then replica plated onto inducing medium containing 2% galactose and grown for1 day. The cells were then subjected to the QSCI assay, and cell lysis was assessed via Ponceau S staining (top panel) before immunodetection using antibodies against the GSTepitope (bottom panel). (C) Single colonies of E. coli BL21(DE3) cells harboring empty vector (pET28a) or expressing His6–Yih1 (pES189-D1A), GST–Gcn1[2052–2428](pES123-B1), or GST alone (pGEX-6p-3) were grown on LB plates with antibiotics to maintain the plasmid, replica plated on medium containing IPTG, and then transferred tonitrocellulose membrane. Cells were lysed, and the lysis was assessed by Ponceau S staining (top panels) before immunodetection using antibodies against the His6 and GSTepitopes (bottom panels). It is evident that the majority of cells in one cell colony patch failed to overexpress the desired protein (black circle). The Ponceau S staining in theleft panel shows that a crease in the membrane partially impaired transfer of the colonies in the second row, but this did not affect the immunoassay analysis.

Semi-quantitative colony immunoassay / A.G. Cridge et al. / Anal. Biochem. 447 (2014) 82–89 85

(1:100,000, Thermo Scientific, Pierce-31430) and goat anti-rabbitIgG peroxidase conjugate (1:50,000, Thermo Scientific, Pierce-31460).

Results and discussion

Fast semi-quantitative screening for cells overexpressing a desiredprotein

Because SDS is known to affect protein folding as well as mem-brane integrity, we attempted to break yeast cells by simply expos-ing them to 10� Laemmli SDS–PAGE running buffer containing 1%SDS. Single colonies of yeast strains verified previously to overex-press GST-tagged Yih1 (yeast IMPACT homologue), or GST alone,from a galactose-inducible promoter [22–24] were patched andgrown on selective medium containing glucose as a carbon sourceand then replica plated on selective medium containing galactoseas a carbon source to induce protein expression. The cells werethen transferred to nitrocellulose membranes and lysed by simplyplacing the membrane on absorbent paper containing 10� Lae-mmli SDS–PAGE running buffer (Fig. 1A). After removing cell

debris, the membrane was stained with Ponceau S in order to as-sess whether sufficient cell breakage had occurred and whetherthe cells from all colonies released similar amounts of protein. Pon-ceau S staining has previously been reported to be a suitable ‘‘load-ing control’’ because it is a nonspecific protein dye that colors allproteins attached to the membrane in a quantitative manner[25]. The membrane was then subjected to immunoblotting usinganti-GST antibodies. We found that the cells left a strong Ponceau Sstained footprint (Fig. 1B, top panel), suggesting that cell lysis hadoccurred. As expected, immunoblotting signals were detected onlywhen the cells expressed GST alone or GST–Yih1 (Fig. 1B, bottompanel). We repeated the experiment using TBS buffer supple-mented with 1% SDS instead of Laemmli SDS–PAGE running bufferand obtained similar results (data not shown). Together, our re-sults demonstrate that a solution containing 1% SDS is sufficientfor lysing yeast cells efficiently.

Next we tested whether our method was also applicable forE. coli. E. coli was transformed with plasmids that allowed theIPTG-inducible expression of a GST-tagged fragment of the yeastprotein Gcn1 (amino acids 2052–2428), GST alone, His6-taggedYih1, and vector alone, respectively [18,22,23]. Five independentcolonies from each transformation were subjected to the colony

86 Semi-quantitative colony immunoassay / A.G. Cridge et al. / Anal. Biochem. 447 (2014) 82–89

immunoassay as described above for yeast except that LB mediumwas used and IPTG was used for the induction of protein expres-sion. Ponceau S staining showed that E. coli lysed well and the pro-teins were transferred successfully onto the membrane (Fig. 1C,top panels). As expected, anti-His6 antibodies led to a signal onlywhen cells expressed His6-tagged proteins, in this case with theHis6–Yih1-expressing cells (Fig. 1C, bottom panel). Conversely,the GST antibody led to a signal only when cells expressed GSTalone or GST–Gcn1[2051–2428]. From the transformants express-ing GST–Gcn1[2051–2428], it appears that the left colony ex-presses the protein at higher levels as compared with the otherfour transformants. It is evident that one of the five E. coli coloniestransformed with the His6–Yih1-expressing plasmid did not ex-press His6–Yih1 (see circle in Fig. 1C).

Together, our results demonstrate that this method clearly al-lows the identification of yeast or E. coli cells that (over)expressthe desired proteins.

Fig.2. The QSCI method produces semi-quantitative results. (A, B) Yeast wild-typestrain H1511 (A) and gcn1D strain H2556 (B) harboring empty vector (plasmidpRS316), or expressing myc-tagged Gcn1 from its own promoter and from a low-copy plasmid (lc Gcn1–myc, plasmid p2367) or high-copy plasmid (hc Gcn1–myc,plasmid p1834), or from a galactose-inducible promoter (Galp Gcn1–myc, pES21-3),were subjected to the QSCI assay as outlined in Fig. 1A using antibodies against themyc epitope (A) or polyclonal antibody specifically against the Gcn1 protein (B).Cells failing to efficiently express the desired protein can be easily detected (blackcircles). (C) The four transformants on the left side of panel B were grown toexponential phase in minimal medium containing 2% galactose, harvested, andsubjected to SDS–PAGE and immunoblotting using antibodies against Gcn1 andagainst Pgk1 as control for equal loading. The results suggest that the amount ofGcn1 increases with plasmid copy number (low copy vs. high copy) and that thehighest protein level is achieved with the galactose-inducible promoter. (D, E)Strains with chromosomally GFP-tagged or mCherry-tagged genes as indicated(HTB2 codes for histone H2B) and a gcn1D strain as control (strains from top tobottom: panel D, PGK1–GFP, GCN1–GFP, MSY-1-1, KW195; panel E, GCN1–GFP,BY4741, MSY-1-1) were grown to saturation in liquid culture, Then 5 ll of thecultures was transferred to solid medium and after no more than 15 h of incubationwas subjected to the QSCI assay using antibodies against GFP or Gcn1 as indicated.

Semi-quantitative analysis of protein expression

We next tested whether this protocol could be used to quantifythe level of protein expression and whether large proteins could bereadily detected as well such as the 297-kDa protein Gcn1 [26]. Wetransformed wild-type yeast strain H1511 with various plasmidsexpressing myc-tagged Gcn1. Because the myc tag is located atthe C terminus of the protein, immunoblotting signals obtainedwith the anti-myc antibody would indicate that a full-length pro-tein was expressed. The plasmids used in this experiment werelow copy (lc) and high copy (hc) plasmids expressing Gcn1–mycfrom its native promoter, where the plasmid copy number in thecell determines the total amount of Gcn1 protein expressed inthe cell. We also included an hc plasmid expressing Gcn1 fromthe strong galactose-inducible promoter. The yeast transformantswere subjected to our colony immunoassay using plates that con-tain galactose as a carbon source. Using the myc antibody, wefound that the signal intensity correlated with the expectedcellular protein levels (see also below), and this was not due to dif-ferences in the amount of protein released from lysed cells asjudged by Ponceau S staining (Fig. 2A). The strongest signal wasfound with cells expressing Gcn1 from the galactose-inducible pro-moter, followed by cells expressing Gcn1 from an hc plasmid andan lc plasmid, and only a very weak signal was detected with cellsharboring empty vector. These results clearly show that our meth-od lysed yeast cells efficiently to even liberate large intracellularproteins such as Gcn1. From the colony Westerns, it can be clearlyseen that one colony patch failed to efficiently express Gcn1. Takentogether, our assay clearly allows the semi-quantitative evaluationof protein expression; therefore, we called it the quick semi-quan-titative colony immunoassay.

Interestingly, Gcn1–myc expressed from an lc plasmid and itsown promoter still led to a signal that was significantly higher thanthat of cells lacking Gcn1–myc (Fig. 2A). We found that the expres-sion level of this plasmid-borne Gcn1 was approximately twotimes lower than that of endogenous Gcn1 (R. Shanmugam andE. Sattlegger, unpublished), and together with the fact that Gcn1is not a highly expressed protein (7330 molecules/cell [27]), thissuggested that the QSCI method is also suitable for detecting pro-teins expressed from chromosomal genes (see, e.g., Refs. [28,29]).In fact, GFP-tagged proteins expressed from genes in their naturalchromosomal location and from their native promoter were de-tected with GFP antibodies (Fig. 2D), and the signal intensities cor-related with the abundance of the protein (Pgk1 is more abundantin the cell than Gcn1 [27]). In addition, Gcn1 expressed from itschromosomal gene, untagged or GFP tagged, could be detectedwith the protein-specific Gcn1 antibody (Fig. 2E).

Histone H2B fused to the mCherry red fluorescent protein haspreviously been shown to be located in the nucleus [30], and wewere able to detect histone H2B in our assay (Fig. 2D, bottompanel), suggesting that the QSCI assay was also applicable fordetecting an abundant nuclear protein.

Protein-specific antibodies can be used with this method

Next we repeated the above experiment with an independentset of transformants that derived from the strain H2556 lackingthe chromosomal GCN1 gene and using anti-Gcn1 antiserum in-stead of anti-myc antibodies. This Gcn1 antiserum can detectGcn1 proteins that have been subjected to denaturing gel electro-phoresis and subsequent immunoblotting, and it is capable ofimmunoprecipitating Gcn1 from cell extracts, suggesting that thisantiserum can bind to natively folded as well as denatured Gcn1[18,24]. The Gcn1 antiserum has previously been verified to specif-ically detect Gcn1 with low cross-reactivity for other yeast proteins

Fig.3. Determining the optimal concentration of inducing agent. (A) Outline of the experimental procedure. (1) First, 5 ll of overnight culture was transferred in rows on solidmedium containing various amounts of inducing agent. (2, 3) After overnight incubation, slices of agar containing the cells were cut out (2) and transferred onto a clean glassplate and aligned next to each other from highest to lowest concentration of inducing agent (3). (4) Then a nitrocellulose membrane was carefully placed on the cells, and theQSCI assay was conducted as described above. (B, C) Yeast gcn1D strain MSY-1-1 (B) and wild-type H1511 (C) harboring empty vector (plasmid pEMBLyex4), oroverexpressing from a galactose-inducible promoter Gcn1–myc (pES21-3), GST–Yih1 (pES187-B1), or GST alone (pES128-9), were subjected to the assay outlined in panel Ausing plates with 2% raffinose and increasing concentrations of galactose as indicated, and a plate with glucose alone, and using antibodies against myc (B) or GST (C). (D)Wild-type H1511 from panel C expressing GST–Yih1 from a galactose-inducible promoter was grown to exponential phase in liquid medium containing increasing amountsof galactose (as in panel B), harvested at OD600nm = 1, and then subjected to SDS–PAGE and immunoblotting using antibodies against GST and Pgk1 as control for equalloading. The optical density of a duplicate set of cultures was measured and plotted in a graph. (E) E. coli BL21(DE3) harboring empty vector (pET28a), or expressing His6–Yih1(pES189-D1A) from an IPTG-inducible promoter, was subjected to the assay shown in panel A using plates with increasing concentrations of IPTG as indicated and usingantibodies against His6. (F) E. coli BL21(DE3) expressing GST–Gcn1[2051–2428] (pES123-B1), empty vector (pETDuet), or GST alone (pGEX-6p-3) from an IPTG-induciblepromoter, was subjected to the assay shown in panel A using plates with increasing concentrations of IPTG as indicated, and using antibodies against GST. (G) E. coliBL21(DE3) from panel F expressing GST–Gcn1[2051–2428] (pES123-B1), or GST alone (pGEX-6p-3), was grown to exponential phase in liquid medium. At OD600nm = 0.7, IPTGwas added to a final concentration as indicated in panel F, and after 10 or 60 min equal volumes of each sample were taken and the cell pellet was subjected to SDS–PAGE andimmunoblotting using antibodies against GST. The optical density of the cultures was measured and plotted in a graph. The growth curves for cells expressing GST alone andGST–Gcn1[2051–2428] were similar, and a representative result is shown.

Semi-quantitative colony immunoassay / A.G. Cridge et al. / Anal. Biochem. 447 (2014) 82–89 87

[31]. Again, it can be clearly seen that one colony patch failed toefficiently express the desired protein (Fig. 2B). As found above,Gcn1 was expressed the highest from the galactose-inducible pro-moter, followed by Gcn1 expressed from its own promoter and anhc plasmid, with the lowest expression being from its own

promoter and from an lc plasmid. These findings were comparableto those obtained from standard Western procedures for scoringprotein expression levels [5,6] (Fig. 2C), indicating that the signalintensity in the QSCI assay reflected the cellular abundance of pro-tein. Thus, the QSCI allows the use of protein-specific antibodies.

88 Semi-quantitative colony immunoassay / A.G. Cridge et al. / Anal. Biochem. 447 (2014) 82–89

Determining the optimal concentration of inducing agent

Because the amount of inducing agent is relevant for the effi-cient induction of protein expression, we endeavored to testwhether the QSCI assay could be used to optimize the concentra-tion of inducing agent. To test this, we generated saturated over-night cultures from yeast cells harboring plasmid-borne andgalactose-inducible Gcn1–myc, GST–Yih1, GST alone, or cells car-rying empty vector. Then 5 ll of each culture (�105 cells [32])was transferred to solid medium containing various amounts ofinducing agent (Fig. 3A) and incubated at 30 �C. Because the cellswere aligned on the medium in a single row, it was possible tocut out a strip of solid medium carrying the cells atop. The agarpieces were then aligned next to each other from lowest to highestconcentration of inducing agent, the cells were transferred tonitrocellulose membranes, and the QSCI assay continued asdescribed above. From the QSCI result, it is evident that galact-ose-dependent overexpression of Gcn1–myc occurs at galactoseconcentrations ranging from 0.1% to 10% (Fig. 3B). When compar-ing the anti-myc signals with the Ponceau S image, it appears thatincreasing the galactose concentration from 0.1% to 0.5% may haveslightly improved Gcn1 expression, but a further increase in galact-ose concentration did not significantly increase Gcn1 expression,implying that 0.5% galactose is sufficient for the highest possiblelevel of Gcn1–myc expression on solid medium.

For GST–Yih1 overexpression galactose concentrations between0.1 and 10% were necessary, and for GST alone concentrationsbetween 2% and 10% were necessary, whereas for both proteins2% galactose was sufficient to obtain the highest expression(Fig. 3C). It can be clearly seen that at each galactose concentrationGST–Yih1 is expressed at a greater level than GST alone, a phenom-enon we have observed previously when growing the same yeastcells in liquid cultures and subjecting their cell extracts to standardimmunoblotting procedures [23].

To verify that in the QSCI assay an increase in signal strengthcorrelated with a higher amount of expressed protein, we deter-mined the GST–Yih1 protein levels in strains grown to exponentialphase in liquid medium containing increasing amounts of galact-ose (Fig. 3D, Western panels). An identical second set of strainswas used to determine whether high galactose concentration af-fects their growth rate (Fig. 3D, graph). Similar to the QSCI assay,we found that galactose concentrations between 0.1% and 10%were necessary for good GST–Yih1 expression, whereas someamount of GST–Yih1 was expressed at 0.02% galactose (Fig. 3Cvs. Fig. 3D). It appeared that 10% galactose led to a reduced growthrate, but despite this GST–Yih1 was well expressed. Taken to-gether, these results suggest that the signal intensity in the QSCIassay correlates with the amount of protein expressed.

In this experiment, we used strain MSY-1-1 (Fig. 3B), which isisogenic to the commonly used yeast strain background BY4741.In addition, we used the H1511 yeast strain background [33],which is different from that of MSY-1-1 or BY4741 (Fig. 3C), dem-onstrating that the applicability of the QSCI assay is not limited to aspecific strain background.

Next we tested whether the QSCI assay could also be used forcalibrating inducing agent concentrations in E. coli. This in factwas the case; our results clearly show that His6–Yih1 was ex-pressed at IPTG concentrations of 9 lg/ml IPTG or higher, withthe highest expressions being between 18 and 44 lg/ml IPTG(Fig. 3E). Interestingly, IPTG concentrations of higher than 44 lg/ml led to reduced His6–Yih1 expression even though similaramounts of protein were released from lysed cells as judged bythe strong Ponceau S signals observed for all cells on all inducingplates (Fig. 3E, top panel).

GST–Gcn1[2052–1418] expression appeared to be the highestat 10 and 20 lg/ml IPTG (Fig. 3F). GST alone was already expressed

efficiently at 10 lg/ml IPTG. As observed before, high IPTG concen-trations led to reduced expression of GST–Gcn1[2052–2428] orGST even though strong Ponceau S signals were observed for allcells on all inducing plates, which would support the idea thatthe cells grew equally well (Fig. 3E, top panel). In agreement withthe latter, cells grown in liquid medium showed the same growthbehavior regardless of the IPTG concentrations to which the cellswere exposed (Fig. 3G, graph). It can be clearly seen that at eachIPTG concentration, GST alone is expressed at a greater level thanGST–Gcn1[2052–1418] (Fig. 3F), a phenomenon we have observedpreviously for these proteins when growing the same E. coli cells inliquid cultures and subjecting their cell extracts to standard immu-noblotting procedures [23].

To verify that the signal intensities in the QSCI assay are indic-ative of the expression levels of the detected protein, we deter-mined proteins levels using standard Western assays. For this,we grew cells in liquid medium to exponential phase and thenadded various amounts of ITPG, followed by SDS–PAGE of the cellextract and immunoblotting. We found that for an induction timeof 1 h, any IPTG concentration led to efficient protein expression;however, when shortening the induction time to 10 min, it canbe clearly seen that for induction of GST–Gcn1[2052–1418]expression higher IPTG concentrations were necessary than forGST expression (Fig. 3G), similar to what was found in the QSCI as-say. This demonstrates that our QSCI assay is also quantitative forE. coli.

Taken together, our method allows the determination of theoptimal concentration of inducing agents for maximal proteinexpression on solid medium. Furthermore, our data clearly showthat high amounts of inducing agent can reduce the amount ofprotein induction, possibly due to the metabolic burden imposedon the cells, a phenomenon previously reported for E. coli [34],demonstrating the necessity of adjusting IPTG concentrations foroptimal protein expression.

Conclusion

We have developed a rapid and cost-effective method for thesemi-quantitative screening of multiple strains for those thatexpress recombinant protein most efficiently. With this method,after incubating cells on inducing medium, the final result can beobtained in only 4.5 h (transfer and lysis of cells and Ponceau Sstaining: 1 h; immunoblotting: 3 h; scanning and analysis:30 min). Furthermore, we showed that the optimal concentrationof inducing agent can be determined easily, reducing the consump-tion of the sometimes very expensive inducing agents. Here, thismethod was optimized using S. cerevisiae and E. coli, but this meth-od may also be applicable for other yeast and bacterial cells.

Acknowledgments

We thank Orna Cohen-Fix for the strain harboring mCherry-tagged H2B, Su Jung Lee and Renuka Sastry for technical support,and Renuka Shanmugam for supplying the image for Fig. 2C. Thiswork was funded by Marsden Fund Council Grant MAU0607(administered by the Royal Society of New Zealand) and the Auck-land Medical Research Foundation to E.S. and by Massey Universitydoctoral scholarships to J.V. and R.R.

References

[1] F. Silva, J.A. Queiroz, F.C. Domingues, Evaluating metabolic stress and plasmidstability in plasmid DNA production by Escherichia coli, Biotechnol. Adv. 30(2012) 691–708.

[2] Z. Zhang, M. Moo-Young, Y. Chisti, Plasmid stability in recombinantSaccharomyces cerevisiae, Biotechnol. Adv. 14 (1996) 401–435.

Semi-quantitative colony immunoassay / A.G. Cridge et al. / Anal. Biochem. 447 (2014) 82–89 89

[3] B. Miroux, J.E. Walker, Over-production of proteins in Escherichia coli: mutanthosts that allow synthesis of some membrane proteins and globular proteinsat high levels, J. Mol. Biol. 260 (1996) 289–298.

[4] M. Zoonens, B. Miroux, Expression of membrane proteins at the Escherichia colimembrane for structural studies, Methods Mol. Biol. 601 (2010) 49–66.

[5] J. Sambrook, D.W. Russell, Molecular Cloning: A Laboratory Manual, ColdSpring Harbor Laboratory Press, Cold Spring Harbor, NY, 2001.

[6] H. Towbin, T. Staehelin, J. Gordon, Electrophoretic transfer of proteins frompolyacrylamide gels to nitrocellulose sheets: procedure and some applications,Proc. Natl. Acad. Sci. USA 76 (1979) 4350–4354.

[7] S. Tomlinson, A.F. Esser, Rapid immunological screening for protein expressionin yeast transformants, Biotechniques 13 (1992) 710–711.

[8] C. Holz, O. Hesse, N. Bolotina, U. Stahl, C. Lang, A micro-scale process for high-throughput expression of cDNAs in the yeast Saccharomyces cerevisiae, ProteinExpr. Purif. 25 (2002) 372–378.

[9] H. Riezman, T. Hase, A.P. van Loon, L.A. Grivell, K. Suda, G. Schatz, Import ofproteins into mitochondria: a 70 kilodalton outer membrane protein with alarge carboxy-terminal deletion is still transported to the outer membrane,EMBO J. 2 (1983) 2161–2168.

[10] V.V. Kushnirov, Rapid and reliable protein extraction from yeast, Yeast 16(2000) 857–860.

[11] S. Lyons, N. Nelson, An immunological method for detecting gene expressionin yeast colonies, Proc. Natl. Acad. Sci. USA 81 (1984) 7426–7430.

[12] A.A. McCracken, I.V. Karpichev, J.E. Ernaga, E.D. Werner, A.G. Dillin, W.E.Courchesne, Yeast mutants deficient in ER-associated degradation of the Zvariant of a-1-protease inhibitor, Genetics 144 (1996) 1355–1362.

[13] L.J. Wuestehube, R. Duden, A. Eun, S. Hamamoto, P. Korn, R. Ram, R. Schekman,New mutants of Saccharomyces cerevisiae affected in the transport of proteinsfrom the endoplasmic reticulum to the Golgi complex, Genetics 142 (1996)393–406.

[14] P.L. Bartel, S. Fields, Analyzing protein–protein interactions using two-hybridsystem, Methods Enzymol. 254 (1995) 241–263.

[15] J.M. Cregg, I. Tolstorukov, A. Kusari, J. Sunga, K. Madden, T. Chappell,Expression in the yeast Pichia pastoris, Methods Enzymol. 463 (2009) 169–189.

[16] R. Repaske, Lysis of gram-negative bacteria by lysozyme, Biochim. Biophys.Acta 22 (1956) 188–191.

[17] R.S. Sikorski, P. Hieter, A system of shuttle vectors and yeast host strainsdesigned for efficient manipulation of DNA in Saccharomyces cerevisiae,Genetics 122 (1989) 19–27.

[18] E. Sattlegger, A.G. Hinnebusch, Separate domains in GCN1 for binding proteinkinase GCN2 and ribosomes are required for GCN2 activation in amino acid-starved cells, EMBO J. 19 (2000) 6622–6633.

[19] G. Cesarini, A.G. Murray, Plasmid vectors carrying the replication origin offilamentous single-stranded phages, in: J.K. Setlow (Ed.), Genetic Engineering:Principles and Methods, Plenum, New York, 1987, pp. 135–154.

[20] M.J. Marton, C.R. Vazquez de Aldana, H. Qiu, K. Chakraburtty, A.G. Hinnebusch,Evidence that GCN1 and GCN20, translational regulators of GCN4, function onelongating ribosomes in activation of eIF2a kinase GCN2, Mol. Cell. Biol. 17(1997) 4474–4489.

[21] C.R. Vazquez de Aldana, M.J. Marton, A.G. Hinnebusch, GCN20, a novel ATPbinding cassette protein, and GCN1 reside in a complex that mediatesactivation of the eIF-2a kinase GCN2 in amino acid-starved cells, EMBO J. 14(1995) 3184–3199.

[22] E. Sattlegger, M.J. Swanson, E.A. Ashcraft, J.L. Jennings, R.A. Fekete, A.J. Link,A.G. Hinnebusch, YIH1 is an actin-binding protein that inhibits protein kinaseGCN2 and impairs general amino acid control when overexpressed, J. Biol.Chem. 279 (2004) 29952–29962.

[23] E. Sattlegger, J.A.R.G. Barbosa, M.C.S. Moraes, R.M. Martins, A.G. Hinnebusch,B.A. Castilho, Gcn1 and actin binding to Yih 1: implications for activation of theeIF2 kinase Gcn2, J. Biol. Chem. 286 (2011) 10341–10355.

[24] T. Waller, S.J. Lee, E. Sattlegger, Evidence that Yih1 resides in a complex withribosomes, FEBS J. 279 (2012) 1761–1776.

[25] I. Romero-Calvo, B. Ocon, P. Martinez-Moya, M.D. Suarez, A. Zarzuelo, O.Martinez-Augustin, F.S. de Medina, Reversible Ponceau staining as a loadingcontrol alternative to actin in Western blots, Anal. Biochem. 401 (2010) 318–320.

[26] M.J. Marton, D. Crouch, A.G. Hinnebusch, GCN1, a translational activator ofGCN4 in Saccharomyces cerevisiae, is required for phosphorylation ofeukaryotic translation initiation factor 2 by protein kinase GCN2, Mol. Cell.Biol. 13 (1993) 3541–3556.

[27] S. Ghaemmaghami, W.K. Huh, K. Bower, R.W. Howson, A. Belle, N. Dephoure,E.K. O’Shea, J.S. Weissman, Global analysis of protein expression in yeast,Nature 425 (2003) 737–741.

[28] M.S. Longtine, A. McKenzie III, D.J. Demarini, N.G. Shah, A. Wach, A. Brachat, P.Philippsen, J.R. Pringle, Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae, Yeast 14(1998) 953–961.

[29] M. Knop, K. Siegers, G. Pereira, W. Zachariae, B. Winsor, K. Nasmyth, E.Schiebel, Epitope tagging of yeast genes using a PCR-based strategy: more tagsand improved practical routines, Yeast 15 (1999) 963–972.

[30] M.T. Webster, J.M. McCaffery, O. Cohen-Fix, Vesicle trafficking maintainsnuclear shape in Saccharomyces cerevisiae during membrane proliferation, J.Cell Biol. 191 (2010) 1079–1088.

[31] C.R.V. deAldana, M.J. Marton, A.G. Hinnebusch, Gcn20, a novel ATP bindingcassette protein, and GCN1 reside in a complex that mediates activation of theeIF-2-a kinase GCN2 in aminoacid-starved cells, EMBO J. 14 (1995) 3184–3199.

[32] D. Burke, D. Dawson, T. Stearns, Methods in Yeast Genetics: A Cold SpringHarbor Laboratory Course Manual, Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY, 2000.

[33] M. Foiani, A.M. Cigan, C.J. Paddon, S. Harashima, A.G. Hinnebusch, GCD2, atranslational repressor of the GCN4 gene, has a general function in theinitiation of protein synthesis in Saccharomyces cerevisiae, Mol. Cell. Biol. 11(1991) 3203–3216.

[34] C.P. Chou, Engineering cell physiology to enhance recombinant proteinproduction in Escherichia coli, Appl. Microbiol. Biotechnol. 76 (2007) 521–532.