Genomics- The Science and Technology Behind the Human Genome Project. Charles R. Cantor, Cassandra L / genomics11-15 / 14

510

SEQUENCE-SPECIFIC MANIPULATION OF DNA

been shown to be the case for the intact L1 repeat. Given this general picture, it is not sur-

prising that certain types of repeats appear to cluster in certain genome regions. While the

exact mechanism for this clustering is not understood, a formal mechanism to explain it

would simply be to state that the transposition of the mobile elements favors genomic re-

gions with certain properties.

Alu sequences are preferentially located in Giemsa light, G

C-rich bands. Note that

the Alu’s,

themselves

are G

C

rich. Their average base composition

is 56% G

C,

and they have a CpG content that is only 64% of that expected statistically for such a G

C content. This is remarkable given the overall suppression of CpG throughout most of

the genome. Thus Alu’s behave a bit like HTF islands (Chapter 8). It

is

not surprising

then, that

like

HTF

islands. Alu’s seem to preferentially associated

with

genes. The

Giemsa light bands are very gene rich, and within many of these genes there are truly re-

markable numbers of Alu repeats (in the introns, of course). These Alu’s greatly compli-

cate attempts to sequence genomic, gene-rich regions by shotgun strategies (Chapter 10)

because of the difficulty of assembling the sequence.

L1 sequences occur preferentially in dark bands. These bands are A

T rich. The L1s

themselves

are

A

T rich (58%). They are also extremely deficient in CpG sequences

with only 13% of what would be expected statistically. Thus we can conclude from both

the patterns of

Alu’s

and L1’s that like attracts like, but the mechanism behind

this

re-

mains unclear. Earlier we indicated that the pattern of Alu distribution was really bipha-

sic. Presumably the Alu-rich phase seen in renaturation experiments corresponds to DNA

from light bands, but this has not been formally proved.

The final class of well-characterized interspersed repeats are the VNTRs. These are

preferentially

located

in telomeric light Giemsa bands, although they

are

spread

well

enough through the genome to be generally useful as genetic markers. The reason why

VNTRs cluster near the telomeres is unknown. However, it is worth noting that the fre-

quency of meiotic recombination appears to be very high in the telomeric regions and that

one mechanism of VNTR growth and shrinkage is recombination. There is no way to tell

at present whether any causal relationships existing among these observations. However,

they represent a tantalizing area for future study.

PCR BASED ON REPEATING SEQUENCES

Sequences in the Alu and L1 families are similar enough so that a single PCR primer can

be used to initiate DNA synthesis within a large number of these elements. Some of the

common Alu primers are summarized in Figure 14.38. These are chosen to try to focus on

the most-conserved regions of the repeats within known human sequences without select-

ing sequences that are also conserved in rodents. Some Alu primers are tagged with ex-

tensions to allow more efficient amplification after the first few rounds of PCR where an

inexact or very short match between primer and

target template

may

be

occurring

(Chapter

4).

The

general situation in which these primers are

applicable is

shown

in

Figure 14.39. Neighboring copies of a repeat can have inverted configurations (head to

head or tail to tail) or tandem configurations (head to tail). In the former case, a single

PCR primer will serve to amplify the DNA between the repeats. In the latter case, two

primers must be used. Inter-Alu PCR is a very powerful tool because so much of the hu-

man genome

is

dense in Alu sequences, and Alu

sequences in humans are

well

diverged

PCR BASED ON REPEATING SEQUENCES

511

from those of rodents. Thus a significant fraction of the human genome is potentially am-

plifiable by inter-Alu PCR. The L1

repeat is less useful in this regard because it

is rarer

and because its sequence is more conserved in rodents and human. Despite this limitation

inter-L1 PCR or PCR between L1

and Alu sequences can still be helpful tools.

It would take most of a chaper to

describe the myriad applications of inter-Alu or

in-

within

a

few kb of

each other. This will be quite frequent in Alu-rich regions of the

genome, much rarer in other regions. For example, inter-Alu PCR will selectively amplify

the

human

component in a rodent hybrid cell line. It will preferentially amplify the YAC

DNA

in

a

background

of yeast DNA. Combining YAC vector primers and Alu primers

512

SEQUENCE-SPECIFIC MANIPULATION OF DNA

will preferentially amplify human DNA cloned near the ends of YAC inserts. These are

exactly the samples most desirable as probes for YAC

walking techniques. Single-sided

Alu amplification procedures have also been described for regions where Alu’s are too

dilute to allow efficient inter-Alu PCR (Quereshi et al., 1994). Single-sided Alu PCR is

also a very useful method to treat chromosome-specific hncDNA libraries, as described in

Chapter 11.

The pattern of bands amplified by inter-Alu PCR from a whole human chromosome is

often too complex to analyze (although inter-L1 PCR is helpful in such applications).

Sometimes, with particular primers and conditions, the number of amplified bands can be

reduced to of the order to 40. In such cases a significant fraction of these bands appears to

be polymorphic in the population, and thus the Alu PCR provides a very convenient and

easily used set of multiplexed genetic markers (Fig. 14.40).

In contrast to most attempts to

amplify DNA from the whole genome, the pattern of

bands from a few Mb of DNA is

usually quite clear and

diagnostic. Thus inter-Alu

PCR

is a powerful fingerprinting method. The added incentive is that the PCR-amplified bands

seen in an electrophoretic analysis can be cut out and used as single-copy hybridization

probes (competing any residual Alu sequences, as needed). Thus inter-Alu PCR finger-

printing has been applied to YACs,

chromosome fragments, radiation hybrids, and PFG

gel slices, and it has been used to isolate single-copy DNA probes from all of these kinds

of samples. The products of inter-Alu PCR reactions are very useful for FISH mapping.

They are usually complex and

concentrated enough to yield good results and allow

the

rough map position of the sample to be identified. Inter-Alu PCR products are also very

useful for cross-connecting libraries.

An example of inter-Alu PCR

applied to

the analysis

of PFG gel

slices

is shown in

Figure 14.41. Here a

Not

I-digested hybrid cell line containing chromosome 21 as its only

human component was fractionated by PFG under a number of different conditions to

maximize the resolution of indivdiual size regions. Each gel lane was sliced into 40 frag-

ments, and each fragment was subjected to inter-Alu PCR. In most cases lanes showed

discrete patterns of several amplified bands, and adjacent slices often showed quite differ-

ent patterns (Fig. 14.42). Thus the carryover

of material from slice to slice during

the

PFG and subsequent steps is not so

serious as to obscure the fractionation. This means

that individual PCR products from analytical gels like the one shown in Figure 14.42 can

be cut out, reamplified, and used as

immortal specific single-copy DNA probes. Even if

the PFG gel slice contained more than one genomic human

Not

I fragment, the individual

PCR products from that slice are each likely to derive from only a single genomic frag-

ment. Thus they constitute a very

convenient

source

of

new

single-copy

human

DNA

probes.

The method illustrated in Figures 14.41 and 14.42 is very helpful in the later stages of

physical mapping where most fragments of a chromosome are located and the goal is to

obtain new probes for unassigned bands as efficiently as possible. One problem with the

kind of results shown in Figure 14.42 is that the number of new probes provided by a sin-

gle experiment is very large, frequently a hundred or more. Before selecting probes for

further study, usually one would like to know something about their regional location on

the chromosome of interest. The standard way to do this is to take a probe of interest and

hybridize it to a mapping panel of chromosome deletions as we described in Chapter 8.

However, this is far too inefficient

when a hundred or

more probes must be mapped at

once. An alternative approach, useful

for YACs or slices of PFG fractionations,

is shown

in Figure

14.43. Ideally

what one would like to

do

is take DNA from

hybrid

cell lines

PCR BASED ON REPEATING SEQUENCES

515

Figure 14.43

Example of how inter-Alu PCR products can be assigned, en masse, to chromosome

regions. Inter-Alu probes generated from DNA from cell lines 8q

(A), R2-10W (B), or 21q (C)

were used to assign inter-Alu gel slice products regionally. Conventional gel lanes (lanes 1 and 2)

containing inter-Alu products generated from template DNA contained in two different PFG slices

are shown. Note that each cell line specific probe is hybridized to a different set of inter-Alu PCR

products. From Wang et al. (1995).

complex to analyze on a whole genome level. The situation is much more favorable with

the 5 L1 sequence. To estimate the number of PCR products expected with a single 5

L1

primer, Yoshiyuki Sakaki assumed that there were about 3000 copies of the L1 repeat or

one per Mb. This estimate is on the low side of the range reported by others (Table 14.3).

Suppose that the PCR range is 2 kb. Only a quarter of the L1’s within this range will be

APTAMER SELECTION STRATEGIES

517

APTAMER SELECTION STRATEGIES

A relatively recently developed set of strategies combines

physical purification and

PCR

to select DNAs (or RNAs or proteins) with

desired

sequences

or

binding

properties.

These methods appear to be powerful enough that in some cases one can start with all 4

n

possible nucleic acids of length

n and

find the one

or

few with

the optimal affinity for a

given target. Among the potential applications are:

It is relatively easy to make all 4

n

possible DNAs of length

n just by adding all four

dpppN’s at each step in automated DNA synthesis. More restricted mixtures can be made

by an obvious extension of this approach.

The general principle behind select strategies is shown in Figure 14.45. A complex

mixture is allowed to bind to an immobilized target of interest, usually at fairly low strin-

gency. Those species that do bind are eluted, and PCR amplification is used to regenerate

a population of molecules comparable to

the initial total concentration. Now, however,

this population should be enriched for

molecules that

have some affinity for the target.

The cycles of affinity purification and amplification can be repeated as often as needed,

until the complexity of the mixture becomes small enough to analyze. In the perfect case

only a single species would remain, and

if necessary, the stringency of the affinity step

could be progressively increased during successive cycles. In actual cases a mixture of

molecules will be seen, but this will eventually attain a small enough complexity so that

individual components can be cloned and examined. Alternatively, a powerful approach is

to sequence the mixture of molecules remaining after a large number of cycles of select

purification. If certain positions within the DNA (or RNA or protein) are required for

affinity, and others are not, the sequence of the mixture will show conserved residues

at

Figure 14.46

Select strategies that have been proposed for purifying DNAs, and DNAs coding for

RNAs, and proteins with particular properties

that can be converted

to

differential

affinities.

Adapted from Irvine et al. (1991).

TABLE

14.4 High-Affinity Aptamers

Target

Library

Nucleotides

a

Rounds

b

Motif

K d (nM)

E. coli

rho factor

RNA

30

8

Hairpin

1

E. coli

metJ protein

RNA

40

15

Unknown

1

HIV-1 rev protein

RNA

32

10

Bulge

1

sPLA2

RNA

30

12

Complex

1

Modified

30

10

Pseudoknot

1

RNA

Basic fibroblast

RNA

30

13

Hairpin

0.20

growth factor

Vascular endothelial

RNA

30

13

Hairpin/bulge

0.20

growth factor

SLE monoclonal

ssDNA

40

8

Unknown

1

antibody

APTAMER

SELECTION STRATEGIES

519

tive for finding consensus nucleic acid binding sequences to known proteins such as tran-

scription factors (Pollack and Treisman, 1990; Nallur, et al., 1996). For proteins the select

strategy shown may well be too cumbersome to use in practice. However, as an alterna-

tive to PCR amplification, one can use in vivo amplification for proteins instead. An ele-

gant way to do this is bacteriophage display, shown schematically in Figure 14.47. Here

the random coding sequence of interest is subcloned as a fusion with a surface coat pro-

tein of the bacteriophage M13. Either a minor coat protein is used with only five copies or

the major coat protein

with

thousands

of copies

is used. Each bacteriophage plaque

will

be a clonal population representing one particular variant. Mixtures of plaques can be

subjected to cycles of selection by physical affinity to the target of interest, and the suc-

cessive

populations

of

bacteriophage that remain will begin

to

be

populated

more

and

more with clones with the desired affinity properties.

Note that there is no reason why one must start with totally random sequences in select

or bacteriophage display strategies. In many cases there will be existing structures with

properties similar to the optimum behavior desired. In this

case random

mutagensis of

just a small portion of an existing macromolecule can be used as a starting point try to se-

lect a more desirable variant. Despite the intrinsic attractiveness of bacteriophage display,

this approach does have a number of limitations. Only monomeric target proteins can be

examined by bacteriophage display. The protein targets must

be

able

to

fold

properly

within

E. coli,

and they must be oriented in the fusion

so

that

the

site

that

generates their

affinity for the ligand or target is accessible.

A generalization of select strategies has been proposed

by

Sydney

Brenner

and

Richard Lerner; it is called encoded combinatorial chemistry. The basic idea involves tag-

ging linear oligomers of any type of residue, with a PCR-amplifiable specific DNA se-

quence that is a unique identifier of the particular oligomer. The general chemical struc-

ture needed is shown in Figure 14.48. Here the variable DNA identifier is placed between

two constant PCR primers, and one of these is attached via a hub to the oligomer, which

could be a nucleic acid, a peptide, an oligosaccharide,

or really any kind of organic

species

that can

be

built

up in a

stepwise

fashion. The

number

of

different

types

of