The generation of zinc finger proteins by modular assembly

Abstract

The modular assembly (MA) method of generating engineered zinc finger proteins (ZFPs) was the first practical method for creating custom DNA-binding proteins. As such, MA has enabled a vast exploration of sequence-specific methods and reagents, ushering in the modern era of zinc finger-based applications that are described in this volume. The first zinc finger nuclease to cleave an endogenous site was created using MA, as was the first artificial transcription factor to enter phase II clinical trials. In recent years, other excellent methods have been developed that improved the affinity and specificity of the engineered ZFPs. However, MA is still used widely for many applications. This chapter will describe methods and give guidance for the creation of ZFPs using MA. Such ZFPs might be useful as starting materials to perform other methods described in this volume. Here, we also describe a single-strand annealing recombination assay for the initial testing of zinc finger nucleases.

1. Introduction

ZFPs are among the most common type of DNA-binding protein found in eukaryotes, and often comprise arrays of 8–10 fingers (1). The 1991 crystal structure of the three-finger ZFP Zif268 bound to DNA (2) revealed that each finger used amino acids in positions −1, 3 and 6 of its α-helix to contact a 3-bp subsite on the DNA. This seemingly simple mode of recognition inspired an early effort to create custom-DNA binding proteins (3). Through a combination of rational design and combinatorial methods such as phage display, individual zinc fingers were optimized to recognize one of the 64 possible 3-bp DNA subsites (4–8). In principle, these fingers could be assembled in any order to create multi-finger ZFPs to any desired sequence. Since each finger recognized an independent 3-bp subsite, ZFPs containing six fingers were expected to recognize 18 bp of DNA, a site sufficiently large to specify a unique locus in the human genome (9). The construction of ZFPs in this manner is referred to as modular assembly (MA).

Three large sets of MA fingers have been described. A widely-used and well-characterized set was developed by Carlos Barbas of The Scripps Research Institute that covers all 3-bp subsites of the type GNN, most ANN, many CNN, and some TNN (see Table 1) (4, 5, 7). Sangamo Biosciences developed a set that covers GNN subsites, often with slightly different protein sequences depending on whether the finger would be used as finger 1, 2, or 3 in a three-finger protein (see Table 2) (6). This set is not widely used by Sangamo Bioscience, and several recent studies found this set was less likely than others to generate highly active ZFNs (10, 11). ToolGen, another company, developed a set that was based on selections of zinc fingers that occurred naturally in the human genome, as opposed to synthetic variants of Zif268 (8). The final set contains more than 50 full-length fingers that collectively recognize about 25 subsites (see Tables 3 and 4).

Table 1.

The Barbas set of zinc finger modules

DNAa	Fingerb	DNA	Finger	DNA	Finger	RAc	DNA	Finger
AAA	QRANLRA	CAA	QSGNLTE	GAA	QSSNLVR	0.5	TAA
AAC	DSGNLRV +	CAC	SKKALTE +	GAC	DPGNLVR	3	TAC
AAGk	RKDNLKN	CAGk	RADNLTE +	GAGk	RSDNLVR	1	TAGk	REDNLHT +
AAT	TTGNLTV	CAT	TSGNLTE −	GAT	TSGNLVR	3	TAT
ACA	SPADLTR +	CCA	TSHSLTE +	GCA	QSGDLRRd	2	TCA
ACC	DKKDLTR	CCC	SKKHLAE −	GCC	DCRDLAR	80	TCC
ACGk	RTDTLRD	CCGk	RNDTLTE +	GCGk	RSDDLVR	9	TCG
ACT	THLDLIR +	CCT	TKNSLTE +	GCT	TSGELVRd	65	TCT
AGA	QLAHLRA	CGA	QSGHLTE	GGA	QRAHLER	3	TGA	QAGHLAS
AGC		CGC	HTGHLLE +	GGC	DPGHLVR +	40	TGC
AGGk	RSDHLTN	CGGk	RSDKLTE	GGGk	RSDKLVR	6	TGGk	RSDHLTT
AGT	HRTTLTN −	CGT	SRRTCRA +	GGT	TSGHLVR	15	TGT
ATA	QKSSLIA	CTA	QNSTLTE	GTA	QSSSLVR	2.5e	TTA
ATC		CTC		GTC	DPGALVR	40	TTC
ATGk	RRDELNV −	CTGk	RNDALTE −	GTGk	RSDELVR	15	TTG
ATT	HKNALQN	CTT	TTGALTE −	GTT	TSGSLVR	5	TTT

^aThe 3-bp subsite to which the finger was designed. A “k” following the triplet indicates that this neighboring subsite should begin with either a G or T (K in the IUPAC nomenclature). This is due to TSO interactions, as described in the I introduction.

^bThe DNA-recognition portion of the finger, consisting of amino acids −1, 1, 2, 3, 4, 5, and 6 with respect to the α-helix (based on (4, 5, 7)). These amino acids should be inserted into the appropriate positions in an Sp1C or Zif268 zinc-finger scaffold (see Fig. 1C and 1D). Also indicated is whether the finger was recommended favorably (+) or unfavorably (−) by the large-scale ToolGen study (28). Note that many fingers were not evaluated.

^cThe relative affinity of each finger in nM, as determined in a context as the middle finger of the three-finger Zif268 DNA-binding domain. Note that this information is only currently available for the GNN set of fingers (based on (7)).

^dAs described in the introduction, QSGDLRR also binds the subsite 5’-GCT-3’ with a relative affinity of 10, which is six-fold stronger than the more 5’-GCT-3’-specific finger, TSGELVR.

^eThe original relative affinity of QSSSLVR reported in (7) was determined to be erroneous. The correct value should be closer to 2.5 nM (24).

Table 2.

The Sangamo set of zinc finger modules

DNAa	Finger 1b	Finger 2	Finger 3
GAA	QRSNLVR	QSGNLAR	QSGNLAR
GAC	DRSNLTR	DRSNLTR	DRSNLTR
GAGk	RSDNLAR	RSDNLAR	RSDNLTR
GAT	QSSNLAR	TSGNLVR	TSANLSR
GCA	QSGSLTR	QSGDLTR	QSGDLTR
GCC	ERGTLAR	DRSDLTR	DRSDLTR
GCGk	RSDDLTR	RSDDLQR	RSDDLTR
GCT	QSSDLTR	QSSDLTR	QSSDLQR
GGA	QSGHLAR	QSGHLQR	QSGHLQR
GGC	DRSHLTR	DRSHLAR	DRSHLAR
GGGk	RSDHLAR	RSDHLTR	RSDHLSR
GGT	QSSHLTR	TSGHLVR	TSGHLVR
GTA	QSGALTR	QSGALAR	QRASLTR
GTC	DRSALAR	DRSALAR	DRSALAR
GTGk	RSDALRT	RSDALSR	RSDALTR
GTT	TTSALTR	TSGALTR	QSSALTR

^aThe 3-bp subsite to which the finger was designed. A “k” following the triplet indicates that this neighboring subsite should begin with either a G or T (K in the IUPAC nomenclature). This is due to TSO interactions, as described in the Introduction.

^bThe DNA-recognition portion of the finger, consisting of amino acids −1, 1, 2, 3, 4, 5, and 6 with respect to the α-helix (based on (6)). Note that this set often recommends a slightly different protein sequence depending on whether the finger would be used as finger 1, 2, or 3 in a three-finger protein. These amino acids should be inserted into the appropriate positions in an Sp1C or Zif268 zinc-finger scaffold (see Fig. 1C and 1D).

Table 3.

The ToolGen set of zinc finger modules

DNAa	Module Nameb	DNA	Module Name	DNA	Module Name		DNA	Module Name
AAA	QSNI +	CAA	QSNI	GAA	CSNR1	QSNR1 +	TAA	QSNK −
	QSNK		QSNV1		CSNR2	QSNR2		QSNV2
	QSNT		QSNV2 +		ISNR	QSNR3
	QSNV1		QSNV3		QGNR	QSNR4
	QSNV2		QSNV4		QSHR5	QSNV2
	QSNV3				QSNK	QSTV
	QSNV4
AAC	HSNK	CAC		GAC	CSNR1 +	HSNK −	TAC
					CSNR2	QSTV
					DSNR +
AAGk		CAGk		GAGk	CSNR1c	QSTVc	TAGk
					CSNR2c	RDER1
					KSNRc	RSNRc −
					QFNR c −	SSNRc
AAT	ISNV	CAT		GAT	ISNR +		TAT	VDYK +
	VSNV +
ACA		CCA		GCA	QSSR1 +	QSTR −	TCA
					QSSR2	VSTR
					QSSR3
ACC		CCC		GCC	DSCR +		TCC
ACGk		CCGk		GCGk	RDER1 +	RDER4	TCGk
					RDER2 +	RDER5
					RDER3	RDER6
ACT		CCT		GCT	VSTR+		TCT
AGA	QSHR1	CGA	QSHT −	GGA	QAHR	QSHR5	TGA	QSHT −
	QSHR3		QSHV +		QSHR1	QTHR1 −		QSHV +
	QSHT		QTHQ		QSHR2 +	QTHR2		QTHQ
	QSHV		QTHR1		QSHR3	WSNR
	QTHQ +				QSHR4
	QTHR1
AGC		CGC		GGC			TGC
AGGk	RDHT	CGGk	RDHT +	GGGk	RDHR1 +	RDKI	TGGk	RDHT +
	RDKR +				RDHR2	RDKR
					RDHT	RSHRc +
AGT		CGT		GGT	WSNR +		TGT
ATA	QNTQ +	CTA		GTA	QSSR1 +	QSTR −	TTA
					QSSR2	VSSR
					QSSR3
ATC	DSAR	CTC		GTC	DSAR +		TTC
ATGk		CTGk		GTGk	RDER1 +		TTGk
					RDER2 +
					VSSRc
ATT		CTT		GTT	HSSR −		TTT
					VSSR −

^aThe 3-bp subsite to which the finger was designed. A “k” following the triplet indicates that this neighboring subsite should begin with either a G or T (K in the IUPAC nomenclature). This is due to TSO interactions, as described in the Introduction.

^bThe unique ToolGen name for each finger (8). The name corresponds to the amino acids in positions −1, 2, 3, and 6 of the α-helix; however, the full sequence of the finger is provided in Table 4. Also indicated is whether the finger was recommended favorably (+) or unfavorably (−) by the large-scale ToolGen study (28). Note that some fingers were not evaluated.

^cThese fingers do not contain Asp in position 2 of the α-helix. Therefore, they are not expected to have a TSO interaction (i.e., the neighboring subsite could begin with any base).

Table 4.

The ToolGen finger sequences and functional data

Module Namea	Finger Sequenceb		DNA-Fullc	Td	DNA-ZFNe	RAf	DNA-RAg
CSNR1	YKCKQCGKAFG	CPSNLRR HGRTH	GA(ACG)	+	GAC	2.3	GAC
CSNR2	YQCNICGKCFS	CNSNLHR HQRTH	GA(ACG)
DSAR	YSCGICGKSFS	DSSAKRR HCILH	(AG)TC	+	GTC	0.4	GTC
DSCR	YTCSDCGKAFR	DKSCLNR HRRTH	GCC	+	GCC	3.9	GCC
DSNR	YRCKYCDRSFS	DSSNLQR HVRNIH		+	GAC
HSNK	YKCKECGKAFN	HSSNFNK HHRIH	(AG)AC	−	GAC	178.6	GAC
HSSR	FKCPVCGKAFR	HSSSLVR HQRTH	GTT	−	GTT	0.8	GTT
ISNR	YRCKYCDRSFS	ISSNLQR HVRNIH	GA(AT)	+	GAT	0.1	GAT
ISNV	YECDHCGKAFS	IGSNLNV HRRIH	AAT
KSNR	YGCHLCGKAFS	KSSNLRR HEMIH	GAG			1.7	GAG
QAHR	YKCKECGQAFR	QRAHLIR HHKLH	GGA			8.4	GGA
QFNR	YKCHQCGKAFI	QSFNLRR HERTH	GAG	−	GAG	66.1	GAG
QGNR	FQCNQCGASFT	QKGNLLR HIKLH	GAA			1.2	GAA
QNTQ	YTCSYCGKSFT	QSNTLKQ HTRIH		+	ATA
QSHR1	YACHLCGKAFT	QSSHLRR HEKTH	(AG)GA
QSHR2	YKCGQCGKFYS	QVSHLTR HQKIH	GGA	+	GGA
QSHR3	YACHLCGKAFT	QCSHLRR HEKTH	(AG)GA
QSHR4	YACHLCAKAFI	QCSHLRR HEKTH	GGA
QSHR5	YVCRECGRGFR	QHSHLVR HKRTH	G(AG)A			2	GGA
QSHT	YKCEECGKAFR	QSSHLTT HKIIH	(ACT)GA	−	(CT)GA	17.1	AGA
QSHV	YECDHCGKSFS	QSSHLNV HKRTH	(ACT)GA	+	(CT)GA	64.3	CGA
QSNI	YMCSECGRGFS	QKSNLTI HQRTH	(AC)AA	+	AAA	51.8	CAA
QSNK	YKCEECGKAFT	QSSNLTK HKKIH	(AGT)AA	−	TAA	2.7	GAA
QSNR1	FECKDCGKAFI	QKSNLIR HQRTH	GAA	+	GAA	1.1	GAA
QSNR2	YVCRECRRGFS	QKSNLIR HQRTH	GAA
QSNR3	YECEKCGKAFN	QSSNLTR HKKSH	GAA
QSNR4	YECVQCGKSYS	QSSNLFR HQRRH	GAA
QSNT	YECVQCGKGFT	QSSNLIT HQRVH	AAA
QSNV1	YECNTCRKTFS	QKSNLIV HQRTH	(AC)AA
QSNV2	YVCSKCGKAFT	QSSNLTV HQKIH	(ACGT)AA	+	CAA	4.1	CAA
QSNV3	YKCDECGKNFT	QSSNLIV HKRIH	(AC)AA
QSNV4	YECDVCGKTFT	QKSNLGV HQRTH	(AC)AA
QSSR1	YKCPDCGKSFS	QSSSLIR HQRTH	G(CT)A	+	G(CT)A	0.8	GTA
QSSR2	YECQDCGRAFN	QNSSLGR HKRTH	G(CT)A
QSSR3	YECNECGKFFS	QSSSLIR HRRSH	G(CT)A
QSTR	YKCEECGKAFN	QSSTLTR HKIVH	G(CT)A	−	G(CT)A	0.9	GTA
QSTV	YECNECGKAFA	QNSTLRV HQRIH	GA(ACG)
QTHQ	YECHDCGKSFR	QSTHLTQ HRRIH	(ACT)GA	+	AGA	0.6	CGA
QTHR1	YECHDCGKSFR	QSTHLTR HRRIH	(ACG)GA	−	GGA	1.6	GGA
QTHR2	HKCLECGKCFS	QNTHLTR HQRTH	GGA
RDER1	YVCDVEGCTWKFA	RSDELNR HKKRH	G(ACT)Gk	+	G(CT)Gk	1.4	GCG
RDER2	YHCDWDGCGWKFA	RSDELTR HYRKH	GCGk	+	G(CT)Gk
RDER3	YRCSWEGCEWRFA	RSDELTR HFRKH	GCGk
RDER4	FSCSWKGCERRFA	RSDELSR HRRTH	GCGk
RDER5	FACSWQDCNKKFA	RSDELAR HYRTH	GCGk
RDER6	YHCNWDGCGWKFA	RSDELTR HYRKH	GCGk
RDHR1	FLCQYCAQRFG	RKDHLTR HMKKSH	GGGk	+	GGGk	0.9	GGG
RDHR2	CRCNECGKSFS	RRDHLVR HQRTH	GGGk
RDHT	FQCKTCQRKFS	RSDHLKT HTRTH	(ACGT)GGk	+	(CT)GGk	6.8	AGG
RDKI	FACEVCGVRFT	RNDKLKI HMRKH	GGGk
RDKR	YVCDVEGCTWKFA	RSDKLNR HKKRH	(AG)GGk	+	AGGk	4.5	GGG
RSHR	YKCMECGKAFN	RRSHLTR HQRIH	GGG	+	GGG	0.2	GGG
RSNR	YICRKCGRGFS	RKSNLIR HQRTH	GAG	−	GAG	0.7	GAG
SSNR	YECKECGKAFS	SGSNFTR HQRIH	GAG			1.3	GAG
VDYK	FHCGYCEKSFS	VKDYLTK HIRTH		+	TAT
VSNV	YECDHCGKAFS	VSSNLNV HRRIH	AAT	+	AAT	2.5	AAT
VSSR	YTCKQCGKAFS	VSSSLRR HETTH	GT(AGT)	−	GTT	2.1	GTG
VSTR	YECNYCGKTFS	VSSTLIR HQRIH	GC(AT)	+	GCT	14.5	GCT
WSNR	YRCEECGKAFR	WPSNLTR HKRIH	GG(AT)	+	GGT	1.3	GGT
Zif268	FACDICGRKFA	RSDERKR HTKIH	GCG			10	GCG

^aThe unique ToolGen name for each finger (8). The name corresponds to the amino acids in positions −1, 2, 3, and 6 of the α-helix; however, the full sequence of the finger is provided in the next column.

^bThe full sequence of the human zinc finger (based on (28, 44)). These amino acids should be inserted between canonical 5-residue linkers (TGEKP) to create new multi-finger ZFPs (see Fig. 1E). The DNA-recognition portion of the finger, consisting of amino acids −1, 1, 2, 3, 4, 5, and 6 with respect to the α-helix, is separated from the scaffold regions by spaces.

^cThe full set of 3-bp subsites to which the finger bound as determined by a yeast one-hybrid assay (8).

^dIndicates this finger was recommended favorably (+) or unfavorably (−) by the large-scale ToolGen study (28). Note that some fingers were not evaluated.

^eThe 3-bp subsite to which the finger bound when it performed favorably in the large-scale ToolGen study (28). A “k” following the triplet indicates that this neighboring subsite should begin with either a G or T (K in the IUPAC nomenclature). This is due to TSO interactions, as described in the Introduction.

^fThe relative affinity of each finger in nM, as determined in a context as finger 3 of the threefinger Zif268 DNA-binding domain. Note that these values are based on (8), but have been normalized (multiplied by a factor of 17.86) to set the value of Zif268 equal to 10 nM. This was done for easier comparison with the Barbas relative affinity values, for which Zif268 had a value of 10 nM (7).

^gThe 3-bp subsite used to test the finger in the relative affinity assay (8).

The ability to assemble the fingers in any order is based on the assumption that each finger binds its 3-bp subsite as an independent module. However, evidence soon began to emerge that, in some cases, there were “context-dependent effects”, interactions that occur when two particular fingers are assembled next to each other on a particular DNA sequence. For example, strong structural and biochemical evidence demonstrated that when the residue in position 2 of the α-helix was aspartate, its side chain made energetically important contacts that favored G or T as the first bp of the previous finger’s subsite (see Fig. 1A) (12). We can refer to this type of context-dependant effect, a helical amino acid (aa) specifying a base pair (bp) in the neighboring subsite, as target site overlap (TSO). However, there is only scattered evidence for other examples of TSO interactions. Other types of context-dependent effects, such as sequence-specific irregularities in the DNA structure due to base stacking, may in fact be more prevalent but are difficult to discover or predict.

Fig. 1.

Construction of ZFNs by MA. (A) A schematic diagram of the hydrogen bonding between the ZFP GZF3 and its 9-bp DNA target site. The ZFP binds “anti-parallel” (C-term. to N-term.) with respect to the DNA strand containing the target sequence (5’ to 3’). Vertical lines are presumed hydrogen bonds. The diagonal line represents a Target Site Overlap (TSO) by Finger 3, influencing the 5’ base of the neighboring Finger-2 subsite to be G. Fingers with potential TSO interactions are indicated in Tables 1–4. (B) A heterodimeric ZFN. Each monomer binds in an everted orientation so that the FokI cleavage domains at their C-termini (open ovals) can dimerize and cleave the DNA within the 6-bp spacer (TTTAAA) between the ZFP sites. (C) The sequence of the three-finger ZFP GZF1, created by inserting the Barbas modules (in brackets) into the Sp1C scaffold. Regions corresponding to beta strands (arrows), alpha helix (tube), and the interfinger linker (TGEKP, wavy line) are indicated. (D) The sequence for GZF1 using the Sangamo modules inserted into the Zif268 scaffold. (E) The sequence of GZF1 using the ToolGen modules, which consist of full-length fingers (in brackets) joined by interfinger linkers. Noted that D and E are for illustrative purposes; they have not been evaluated for function. (F) Additional protein sequences must be added to the ZFPs in panels C, D, or E, and must be encoded by the specific DNA sequences indicated in order to allow proper cloning into the ZFP cloning vector. (G) A schematic diagram of the ZFP cloning vector. A PGK promoter (PGK pro), HA tag (HA), SV40 Nuclear Localization Signal (NLS), ZFP (GZF1), and FokI cleavage domain (here shown to contain the obligate heterodimer modification RR), are indicated. PCR primers used for sequencing are shown in gray.

An appreciation of these context-dependent effects led to the development of what can be called “second generation” ZFP engineering methods, in which selections were performed on larger segments of the intended DNA target sequence. Yen Choo and Sir Aaron Klug developed a “bipartite” method (see Chapter 3 of this volume), allowing for rapid selection of essentially 1.5 fingers in the context of each other (13). Sangamo Biosciences eventually acquired a set of two-finger modules created by this method, and has used them as starting materials for their engineered ZFPs (14). Carl Pabo developed a “sequential selection” method, in which each new finger was selected in the context of its neighbor (15). A second method from the Pabo group, dubbed “B2H” in reference to the use of a bacterial two-hybrid system for selections, involved two steps: an enrichment of fingers on out-of-context 3-bp subsites as in classic modular assembly, followed by a final selection of the enriched fingers in the context of the full intended DNA target sequence (16). This method was extremely laborious, but was shown to frequently produce three-finger ZFPs that were superior in both affinity and specificity to three-finger ZFPs assembled by classical MA. Recently, J. Keith Joung developed a more streamlined version of this method called “oligomerized pool engineering” (OPEN) (11). In this method, the first step of the B2H method is performed by the Joung laboratory in high volume. The pools of enriched fingers are then made available to other investigators for the second step of selection on the intended target (see Chapter 2).

Currently, there are three major sources of ZFPs for investigators interested in engineered zinc finger technology, such as zinc finger nucleases (ZFNs). Sangamo Biosciences uses proprietary methods to create ZFNs that are highly potent and can target an impressively diverse spectrum of DNA sequences. Custom ZFNs from Sangamo are available through Sigma-Aldrich as CompoZr ZFN Technology, albeit at considerable cost (http://www.compozrzfn.com). MA is still one of the most popular methods for engineering ZFPs, and has been referred to in several chapters in this volume (see Chapters 6–11, 17, 23, and 24). Due to its large repertoire of targetable subsites, low cost, and ease of use, MA has been used to produce numerous regulators of endogenous genes, targetable live-cell imaging probes, recombinases, integrases, transposases, and methylases (17–21). Indeed, the first zinc finger nuclease to cleave an endogenous site was created using MA, as was the first artificial transcription factor to enter phase II clinical trial (22, 23). However, many have reported difficulty creating highly active ZFNs using MA. The Zinc Finger Consortium (ZFC), a confederation of academic laboratories, recently found that three-finger MA ZFPs composed of any set of modules could only produce a “positive” binding signal in a B2H reporter assay for 24% of 104 DNA targets (i.e., an “unexpected” failure rate of 76%) (10). The OPEN system has become the method of choice for the ZFC, and has been shown consistently to create high quality ZFNs (11). However, it is important to consider that the current version of OPEN is still a fairly complex procedure and only allows selection of fingers that recognize GNN and a few TNN subsites. If three-finger ZFPs engineered by OPEN were evaluated on the same 104 target sequences as used in the MA study, they would have an “expected” failure rate of 87%. Further advances are therefore required to develop methods for engineering high quality ZFPs that are simultaneously inexpensive, accessible, comprehensive, and robust.

In that regard, several recent reports offer helpful guidance for practitioners of MA. Many of the Barbas set of fingers were selected based on high specificity rather than high affinity. A more recent study by the ZFC showed that three-finger MA ZFPs containing one or more of the four lowest affinity Barbas fingers (GGC, GTC, GCT, GCC) tended to be poor binders in a B2H assay (24). These results argue against the proposition that MA is fundamentally flawed because it fails to address context-dependent effects (25), and suggest that one approach to improving performance might be to replace low affinity fingers with higher affinity variants. An interesting example supporting this approach is TSGELVR, a Barbas finger engineered to recognize 5’-GCT-3’ with high specificity but relatively low affinity (7). QSGDLRR, a Barbas finger designed to recognize 5’-GCA-3’, also binds 5’-GCT-3’ with six-fold higher affinity than TSGELVR. Several successful ZFNs in the literature actually use fingers essentially identical to QSGDLRR to recognize 5’-GCT-3’ (26, 27). Another approach might be to construct ZFPs containing 4, 5, or 6 fingers, which may have increased affinity compared to the three-finger MA ZFPs examined in the ZFC study (10). In support of this approach, a recent large-scale analysis by ToolGen found that ZFN pairs composed of two four-finger MA ZFPs were more frequently active than those composed of three-finger ZFPs (28). The authors also recommended a subset of 37 ToolGen and Barbas modules to produce active ZFNs based on their study (see Table 1 and 3).

In this chapter we describe methods and design considerations for creating ZFPs by MA (see Method 3.1). Once constructed, it is often useful to test some performance characteristics of the ZFPs (i.e., affinity, specificity, or activity). Many such assays are described in other chapters of this volume, including in vitro methods such as EMSA (see Chapters 6 and 16), in vitro ZFN cleavage assays (see Chapter 13), cell-based transcription reporter assays such as B2H or luciferase (see Chapters 2 and 7), and expression assays such as westerns (see Chapters 11 and 16). Here we describe a single-strand annealing (SSA) recombination assay for the initial analysis of ZFNs (see Method 3.2). Once active ZFN pairs are identified, assays at the endogenous site can be performed, such as the Surveyor mutagenesis detection assay (see Chapter 15), the restriction site modification assay (see Chapter 19), the chromatin immunoprecipitation (ChIP) binding assay (see Chapters 11 and 27). Upon achieving activity at the endogenous site, more laborious characterizations of the ZFN are warranted, such as in vitro target site selection assays (SELEX or Bind-n-Seq, (29, 30)), in vivo genome-wide binding site assays (ChIP-seq, see Chapter 27), or other target-specific activity assays (e.g., see Chapter 18).

In the SSA recombination assay, a gene encoding firefly luciferase is divided into two segments, separated by a stop codon and a ZFN target site. Both segments contain an 870-bp region of homology in direct repeat orientation (see Fig. 2). A ZFN-induced DSB between the segments allows efficient SSA homologous recombination, resulting in an active luciferase gene. Firefly luciferase activity is therefore proportional to ZFN activity. The assay has several features that make it well suited as an initial test of ZFN activity. First, the assay is performed in the biologically-relevant environment of a human cell (or could be any easily transfected cell type). Adverse events, such as cytotoxicity due to off-target cleavages, will result in the death of transfected cells. Loss of firefly luciferase signal due to toxicity-dependant cell death can be distinguished from poor ZFN activity by the inclusion of a Renilla luciferase control plasmid. Second, there are many special challenges to the success of a ZFN at an endogenous target site, including limited chromatin accessibility, competing endogenous binding factors, and unexpected local DNA structures. If ZFN activity is not observed at an endogenous site, knowing that the ZFN was functional in an SSA assay will help to differentiate complications at the endogenous site from problems with the ZFN. Third, because it is difficult to anticipate the challenges at the endogenous site, a pragmatic approach would be to create 5–10 ZFNs that target different sites in the region of interest. The SSA assay can be easily reconfigured with a new ZFN target site with a single oligonucleotide. Finally, ZFNs are composed of a heterodimer of two ZFP-FokI cleavage domain fusion proteins ((31), and see Fig. 1B). Often only one monomer of the ZFN pair can be the cause of heterodimer failure. It is therefore important that the functional assay identify potential underperforming monomers, which can be easily done by configuring the SSA assay with homodimer ZFN target sites. Once identified, improvements can be directed towards the deficient monomer, such as replacing poorly-binding fingers with stronger variants or adding additional fingers.

Fig. 2.

The SSA assay. (A) A schematic diagram of the SSA reporter plasmid. The sequence at the junction site between the homologous repeats is shown at the top. The GZF1 and GZF3 heterodimer target site is indicated. PCR primers used in construction are shown as black arrows on the plasmid diagram below. PCR primers used for amplification of the junction region and sequencing are shown in gray. (B) Schamatics of the four plasmids that will be transfected into the test cells for the SSA assay. (C) The expressed ZFN pair will bind at the target site and create a double strand break (DSB). The break will be efficiently repaired by the SSA pathway to reconstruct an active firefly luciferase reporter gene.

2. Materials

2.1 The construction of ZFNs by MA

1. The list of Barbas zinc finger modules (see Table 1).

2. The list of Sangamo zinc finger modules (see Table 2).

3. The list of ToolGen zinc finger modules (see Tables 3 and 4).

4. pPGK-GZF1-wt ZFN-expression plasmid: this plasmid contains a phosphoglycerate kinase promoter driving expression of a gene encoding an HA-tag (YPYDVPDYA), the nuclear localization signal of SV40 large T antigen (PKKKRKV), a three-finger ZFP (GZF1), a 6-aa linker (TGAAAR), and a wildtype FokI endonuclease catalytic domain (32). The expressed ZFN is active as a homodimer. All non-commercial plasmids are available from the authors upon request.

5. pPGK-GZF1-RR ZFN-expression plasmid: This plasmid is similar to pPGK-GZF1-wt, except the FokI cleavage domain contains a modification that prevents activity as a homodimer. The expressed ZFN is only active as a heterodimer with a DD-modified ZFN (i.e., GZF3-DD; see below).

6. pPGK-GZF3-DD ZFN-expression plasmid: This plasmid is similar to pPGK-GZF3-wt, except the FokI cleavage domain contains a modification that prevents activity as a homodimer. The expressed ZFN is only active as a heterodimer with an RR-modified ZFN (i.e., GZF1-RR; see above).

7. Oligonucleotide primers for sequencing the pPGK ZFN expression plasmid:

   1. HAtag-f : 5’-ATGTACCCATACGATGTCCCAGACTACG-3’
   2. FokLinkSeq-r: 5’-TCTATCCTGAGTGGAATTTCTGGC-3’

8. Standard equipment and reagents for agarose gel electrophoresis.

9. Restriction endonucleases and appropriate buffers: XholI and NotI.

10. T4 DNA ligase.

2.2 The SSA recombination assay

2.2.1 Preparation of the SSA reporter plasmid

1. Oligonucleotide primers for SSA reporter plasmid construction:

   1. RvP3-f: 5’-CTAGCAAAATAGGCTGTCCC-3’
   2. SSA-1-3-r: 5’-CTC GAATTC C – [GTT ATC CCT TTTAAA GAA GAT GGT]-CTCACATAGGACCTCTCACACACAG –3’
   3. Additional custom SSA-r oligonucleotides, in which the desired ZFN target site replaces the bracketed sequence in SSA-1-3-r.
2. Oligonucleotide primers for sequencing the SSA reporter plasmid:

   1. LucRep-f: 5’-CGAAGGTTGTGGATCTGGATACC-3’
   2. LucRep-r: 5’-TAGCTGATGTAGTCTCAGTGAGC-3’

3. pGL3-Control plasmid: This plasmid expresses full-length firefly luciferase under the control of a SV40 promoter and enhancer (Invitrogen). It serves as the template for construction of the SSA reporter plasmid, and can be used as a positive control.

4. pSSA-1-3 reporter plasmid: This plasmid contains the target site for the ZFN target site for the heterodimer of GZF1 and GZF3, flanked by a split firefly luciferase reporter gene. This is the exemplary plasmid that will be constructed in Method 3.2.1, but it is also used for the construction of new SSA reporter plasmids. This and all other non-commercial plasmid vectors are available from the investigator upon request.

5. 2× Taq-Pro Complete polymerase mix with dNTPs and 2.0 mM MgCl₂ (Denville Scientific, Cat. CB4070-4). Thaw the 3-mL solution on ice, aliquot 250-µL volumes into 1.5-µL tubes, and store at −20°C.

6. Standard equipment and reagents for agarose gel electrophoresis.

7. PCR purification kit (e.g., QIAquick PCR purification kit, Qiagen).

8. Restriction endonucleases and appropriate buffers: BglII and EcoRI.

9. T4 DNA ligase.

2.2.2 Performing the SSA assay

1. Appropriate ZFN expression plasmids and SSA reporter plasmids (see Note 1).

2. pPGK-empty vector: This plasmid is similar to pPGK-GZF1-wt but lacks the ZFP and FokI domains. It can be used as a negative control.

3. pRL-TK vector: This plasmid expresses Renilla luciferase under control of a HSVthymidine kinase promoter. It can be used to assess cytotoxicity.

4. 24-well polystyrene tissue culture plates.

5. 2× Poly-d-Lysine: Poly-d-Lysine (Poly-K; MP Biomedicals, Cat. 102694) is supplied as a 10 mg dessicant that is resuspended in 50 mL of filtered autoclaved water. This 2× stock should be stored at 4°C. Fresh 1× Poly-K is made by diluting the 2× stock in filtered autoclaved water.

6. HEK 293-T cells (ATCC, Cat. CRL-11268).

7. 75-cm² tissue culture flasks.

8. Phosphate Buffered Saline, sterile for tissue culture (e.g., Dulbecco’s PBS, Invitrogen, Cat. 14190).

9. Trypsin-EDTA (Invitrogen, Cat. 25300-054).

10. Pen-Strep (Invitrogen, Cat. 15140-122), contains 100 U/mL of penicillin and 100 µg/mL streptomycin. 5-mL aliquots are stored at −20°C.

11. Bovine Calf Serum (JR Scientific, Cat. WBJR050). 50-mL aliquots are stored at −20°C.

12. DMEM Complete Medium: DMEM medium (Invitrogen, Cat. 11995) is supplemented with 5 mL of Pen-Strep and 50 mL of Bovine Calf Serum (10% final). This mixture should be stored at 4°C.

13. Hemocytometer.

14. OPTI-MEM reduced serum medium (Invitrogen, Cat. 31985). This should be stored at 4°C.

15. Lipofectamine2000 (1 mg/mL solution, Invitrogen, Cat. 11668-019). This should be stored at 4°C.

16. 1× Reporter Lysis Buffer. Dilute the 5× stock solution (Promega, Cat. E397A) with filtered autoclaved water to make a 1× working solution. The 1× solution can be stored at room temperature.

17. 25× Protease Inhibitor Cocktail: Suspend one Complete Protease Inhibitor Cocktail Tablet without EDTA (Roche, Cat. 11873580001) in 2 mL of filtered autoclaved water. Prepare small aliquots and store at −20°C.

18. Bright-Glo Luciferase Assay System (Promega, Cat. E2620). Aliquot in 15-mL tubes and store at −80°C. This substrate is light-sensitive, keep it wrapped in aluminum foil.

19. Renilla Luciferase Assay System (Promega, Cat. E2820). Aliquot in 15-mL tubes and store at −80°C. This substrate is light-sensitive, keep it wrapped in aluminum foil.

20. Opaque white 96-well plates (Denville, Cat. P9725).

21. Microplate luminometer (e.g., Veritas Microplate Luminometer, Turner Biosystems).

3. Methods

3.1 The construction of ZFNs by MA

1. Find 5–10 appropriate target sites within your DNA region of interest (see Note 2).

2. Verify that the target sites are reasonably unique (not more than 1–2 mismatches) in the target genome. Usually, this can be achieved using BLAST (see Note 3).

3. Design the ZFP coding region based on the information in Tables 1–4 and Fig. 1. For the Barbas and Sangamo module sets, only the DNA-contacting regions corresponding to positions −1, 1, 2, 3, 4, 5, and 6 of the α-helix are inserted into a Sp1C (see Fig. 1C) or Zif268 (see Fig. 1D) three-finger scaffold. It is not clear if either scaffold has any advantage compared to the other. The ToolGen modules listed in Table 4 are connected with TGEKP linkers to create a finger-specific scaffold (see Fig. 1E and Note 4).

4. The coding regions can be physically constructed by commercial synthesis (see Note 5). Typically this is rapid and relatively inexpensive, especially for new users. In addition, the codons can be optimized to the target organism. Be sure to request that the DNA sequences are appended to the protein coding region as indicated in Fig. 1F to facilitate cloning into the pPGK-based ZFN expression plasmids (see Fig. 1G).

5. Digest both the ZFP coding region and a pPGK-based ZFN expression vector with XhoI and NotI. Subclone the ZFP into the expression vector. To create obligate heterodimer ZFN pairs, one ZFP will replace GZF1 in pPGK-GZF1-RR and the other will replace GZF3 in pPGK-GZF3-DD. For homodimer ZFNs, each ZFP will replace GZF1 in pPGK-GZF1-wt.

6. Verify the new constructs by sequencing using primers HAtag-f and/or FokLinkSeq-r.

3.2 The SSA recombination assay

3.2.1 Preparation of the SSA reporter plasmid

1. Prepare the left arm of the split luciferase gene in a 50-µL PCR reaction consisting of the following components (see Fig. 2 and Note 6):

   1. 19 µL of H₂O
   2. 1.0 µL of 10 ng/µL pGL3-Control template
   3. 2.5 µL of 5 µM RvP3 primer
   4. 2.5 µL of 5 µM SSA-1-3-r primer (or other SSA-r primer with a desired ZFN site)
   5. 25 µL of 2× TaqPro polymerase mix

   The reactions conditions are as follows: 95°C for 30 s; 25 cycles of 95°C for 30 s, 58°C for 30 s, 72°C for 30 s; 72°C for 2 min; hold at 4°C.

2. Check for correct amplification by analyzing 10 µL of the reaction using agarose gel electrophoresis. Purify the amplicon using a PCR purification kit.

3. Digest both the amplicon and pSSA-1-3 using BglII and EcoRI (see Note 7).

4. Ligate the amplicon into the pSSA vector using T4 DNA ligase.

5. Because the duplicated region also makes sequencing of the final plasmid difficult, the junction of the repeated region (containing the ZFN target site) should be first amplified in the following 50-µL PCR reaction:

   1. 19 µL of H₂O
   2. 1.0 µL of 10 ng/µL pSSA reporter plasmid
   3. 2.5 µL of 5 µM LucRep-f primer
   4. 2.5 µL of 5 µM LucRep-r primer
   5. 25 µL of 2× TaqPro polymerase mix

   The reactions conditions are as follows: 95°C for 30 s; 25 cycles of 95°C for 30 s, 58°C for 30 s, 72°C for 30 s; 72°C for 2 min; hold at 4°C.

6. Check for correct amplification by analyzing 10 µL of the reaction using agarose gel electrophoresis. Purify the amplicon using a PCR purification kit.

7. The sequence of the newly constructed SSA reporter plasmid can then be verified by sequencing the PCR amplicon using LucRep-f as the sequencing primer.

3.2.2 Performing the SSA assay

Day 0: Seed the cells in the 24-well plates

• 1Pretreat the assay tissue culture plates with Poly-K to help loosely-adherent cells such as HEK 293-T to remain bound during washing procedures. Add 300 µL of 1× Poly-K to each well of a 24-well plate. Allow the Poly-K solution to incubate at room temperature (RT) for 30 min. Steps 1–19 should be performed in a biosafety cabinet to maintain sterility and biosafety.
• 2Prepare the HEK 293-T cells to seed on 24-well plates. Aspirate the medium from the cells growing in a 75-cm² flask.
• 3Wash the cells with 5 mL of PBS. Swirl the PBS around to wash away remaining medium, then aspirate the PBS.
• 4Add 2 mL of Trypsin-EDTA and swirl the flask to ensure the entire surface is covered. Incubate at RT for 1 min.
• 5Resuspend the cells using 8 mL of DMEM Complete Medium.
• 6Place 15 µL of the above suspension onto a hemocytometer. Count the number of cells using an inverted microscope. After use, decontaminate the hemocytometer in 10% final bleach solution.
• 7Aspirate the 1× Poly-K solution from the 24-well plate.
• 8Add 100,000 – 150,000 cells to each well.
• 9Incubate for 24 h in a 37°C CO₂ incubator before transfection.

Day 1: Transfection

• 10To prepare the DNA for transfection, add an appropriate amount of DNA (25 ng of Renilla reporter plasmid, 25 ng of SSA reporter plasmid, and 100 ng of each ZFN plasmid) to 250 µL of OPTI-MEM in a 1.5-mL tube, and incubate at RT for 5 min.
• 11Add 2 µL of Lipofectamine2000 to 250 µL of OPTI-MEM in a separate 1.5-mL tube incubate at RT for 5 min (see Note 8).
• 12Combine the two and mix gently. Incubate at RT for 20 min.
• 13To transfect the cells, remove the medium from cells in the 24-well plate from the previous day. Add the DNA-Lipofectmine2000-OPTI-MEM solution to wells.
• 14Incubate for 4–5 h in a 37°C incubator.
• 15Add 1 mL of DMEM Complete Medium.
• 16Incubate the cells in a 37°C incubator, then allow 24 h for ZFN expression and activity (see Note 9).

Day 2: Harvesting

• 17To preparing the cell lysates, remove the media from wells.
• 18Wash twice with 0.5 mL of PBS. Carefully aspirate the PBS.
• 19Supplement the 1× Reporter Lysis Buffer with 1× Protease Inhibitor Cocktail (4 µL of the 25× Protease Inhibitor Cocktail stock for every 100 µL of 1× Reporter Lysis Buffer). Add 100 µL of supplemented Reporter Lysis Buffer to each well of the 24-well plate.
• 20Allow to incubate on ice for 5 min.
• 21Harvest the cell lysates by scraping the cells from the bottom of the well with pipette tip. Keep the 24-well plate on ice to prevent any degradation of proteins. Wear appropriate personal protective equipment, such as a lab coat and eye goggles.
• 22Transfer the lysates to 1.5-mL tubes on ice.
• 23Centrifuge the tubes at 20,817 × g for 5 min at 4°C to pellet cellular debris. The supernatants can be used directly in the luciferase assays (see Note 10).
• 24Thaw the Bright-Glo solution. This luciferase substrate is light sensitive so keep it wrapped in aluminum foil.
• 25For each sample, pipette 20 µL of cell lysate into a well of an opaque white 96-well plate. The opaque walls ensure that the luminescence generated by one sample does not interfere with measurements of the neighboring samples.
• 26Add 100 µL/well of Bright-Glo Luciferase Assay solution and pipette to mix. Cover the plate with aluminum foil.
• 28Read the plate using a microplate luminometer. Integrate the signal of each sample for 1 s.
• 29To read the Renilla luciferase signal, pipette another 20 µL of cell lysate into unused wells of the opaque white 96-well plate. Add 100 µL/well of Renilla Luciferase Assay solution, mix, and read using the microplate luminometer (see Note 11).
• 30Calculate and plot the average and standard deviation of the firefly and Renilla luciferase signals for each set of duplicates or triplicates (see Fig. 3). The standard deviation of the replicates will be an indication of well-to-well variation, which should be small. The gain in firefly luciferase signal is an indicator of ZFN activity. The loss of Renilla luciferase signal is an indicator of ZFN toxicity, due to significant off-target cleavage activity (typically leading to apoptosis of the transfected cells). Note that a low firefly luciferase signal may be due either to low ZFN activity or high ZFN toxicity. A good example is the ZFN GZF3-wt (see Fig. 3). The presence of this monomer, either as a GZF3-wt homodimer or as a GZF1-wt/GZF3-wt heterodimer, leads to significant toxicity (reduced Renilla signal) and a corresponding apparent loss of ZFN activity (reduced firefly signal). In fact, the nuclease is active, but is causing the transfected cells to die. This conclusion is supported by the obligate heterodimer modifications, GZF1-RR/GZF3-DD. The GZF3 monomer is no longer active by itself, and the heterodimer pair now displays reduced toxicity and increased activity.

Fig. 3.

Exemplary data from an SSA assay. Firefly (SSA recombination due to ZFN activity) and Renilla (constitutively high in the absence of ZFN toxicity) luciferase data is shown for some of the experiments described in Note 1, including the analysis of GZF1-wt and GZF3-wt homodimers, as well as GZF1-wt/GZF3-wt and GZF1-RR/GZF3-DD heterodimers. Luciferase activity was measured 24 h post-transfection. Error bars indicate the standard deviation of triplicate experiments.