Journal of General Microbiology (1980), 117, 509-520. Printed in Great Britain 509

The Use of Translocatable Genetic Elements to Construct a Fine-structure Map of the Klebsiella pneumoniae Nitrogen

Fixation (nif) Gene Cluster

By M I K E MERRICK,l* M E C H T H I L D FILSER,l RAY DIXON,l C L A U D I N E ELMERICH,2 LIONEL SIBOLD2 A N D J E A N HOUMARD2

A.R.C. Unit of Nitrogen Fixation, University of Sussex, Falmer, Brighton, Sussex BNl 9RQ

Unite' de Physiologie Cellulaire, Ddpartement de Biochimie et Gdne'tiyue Microbienne, Jnstitut Pasteur, 28 rue du Dr ROUX, 75015 Paris, France

(Received 13 August 1979; revised 5 October 1979)

The transposons Tn5, Tn7 and TnlO and bacteriophage Mu have been used to derive insertion mutations in the Klebsiella pneumoniae nifgene cluster. A large number of deletion mutants have been derived by imprecise excision of insertion mutations and these deletions have been used to construct a fine-structure map of the nifcluster. Comparison of this genetic map with a physical map of the nifcluster derived by Reidel et al. (1979) showed a very good correlation between genetic and physical mapping methods.

A new complementation group, designated nifU, has been identified and mapped between nifN and nifS. Polarity studies on the 14 nifcistrons now identified suggests that they are organized in at least seven transcriptional units and that all the multicistronic units are transcribed in the same direction.

INTRODUCTION

Genetic analysis of nif mutations in Klebsiella pneumoniae has shown that all the genes specific to nitrogenase regulation and synthesis are located in a cluster near to the histidine operon. Fourteen nif cistrons arranged in seven transcriptional units have so far been identified (Merrick et al., 1978; Elmerich et al., 1978; MacNeil et al., 1978). Insertion mutations in the nifcluster have been derived using the transposons Tn5, Tn7 and TnlO (Merrick et al., 1978) and bacteriophage Mu (MacNeil et al., 1978; Elmerich et al., 1978). These mutations have allowed analysis of the transcriptional organization of the nif genes and identification of some of the nifgene products (Roberts et al., 1978; Merrick et a/., 1978 ; Elmerich et al., 1978).

Both the transposons Tn5, Tn7 and TnlO and bacteriophage Mu can excise imprecisely from their point of insertion with the concomitant production of deletions (Kleckner et a/., 1977 ; Toussaint et al., 1977 ; Merrick et al., 1978) and this property has allowed the isolation of a large number of deletion mutants with end-points in all the known nifcistrons. Using these deletions a fine-structure map of insertion mutants in the nifcluster has been con- structed. The insertion mutants have also been the basis for a collaborative study of the physical distribution of genes in the nifgene cluster (Reidel et al., 1979) and the correlation observed between the genetic and physical maps indicates the advantages to be gained from combining these two techniques.

0022-1287/80/0OOO-8909 $02.00 @ 1980 SGM 33 M I C 117

510 M. M E R R I C K AND O T H E R S

Table 1. Bacteria, plasmids and bacteriophages Strain Genotype or phenotype

Klebsiella pnecrmoniae UNFlO7 UN 1290 UN1990 nifV4944 PC58 hisD2 hsdRl rpsL4 Mu-sensitive

rpsLI 4107 (gnd his nif) hisD4226 recA.56 srl-300: : TnlO

Plasmids

pMFlOO pCEl Phages Mucf Mucts 62 MPhl Muhl

PRDl Km Tc Cb Gnd His Nif ShiA Tra IncP Gnd His Nif ShiA Tra Mu-sensitive derivative of pRDl

Mu phage with P1 host range Mu phage with modified host range; capable of propagation on some Klebsiella strains

Reference or source

Dixon et al. (1977) MacNeil et al. (1978) MacNeil et al. (1978) This work

Dixon et al. (1976) M. Filser & F. Cannon Elmerich et al. (1978)

A. Toussaint A. Toussaint A. Toussaint This work

M E T H O D S

Bacteria, plasmids and bacteriophages. These are listed in Table 1 or have been described previously (Merrick et a/ . , 1978; Elmerich et al., 1978; MacNeil et al., 1978). Strains with nifallele numbers 8200 to 8300 were derived by ethyl methanesulphonate mutagenesis.

Media and antibiotics. These were as described by Merrick et al. (1978) and Elmerich et al. (1978). Construction of recA strains. This was performed as described by MacNeil et al. (1978) using a PlCni

clvl00 lysate grown on strain UN1290. Isolation of insertion mutants. Transposon mutagenesis with Tn5, Tn7 and Tn 10 has been described

previously (Merrick et al., 1978). Mu insertions (Muc+ or Mucfs) in plasmid pCEl were isolated as described by Elmerich et al. (1978). Klebsiella pneumoniae chromosomal insertion mutants were obtained by niuta- genesis of strain UNF5023 with phage MPhl and by mutagenesis of strain PC58 with phage Muhl. The isolation procedure was as described by Elmerich et al. (1978) to obtain M u m insertions in plasmid pCE1.

Isolation of nif deletion and inversion strains. Plasmid deletions were selected as described by Elmerich et al. (1978). Chromosomal deletions were isolated after transduction of specific nif: :Tn mutations to the chromosome as described by Merrick et al. (1978). Deletions were isolated following antibiotic enrichment for either drug-sensitive (Tn7, TnlO) or histidine-auxotrophic (Tn5, Tn7) derivatives of the chromosomal insertion mutants. The enrichment procedure was described by Merrick et al. (1978). Enrichtnent for histidine auxotrophs was in minimal medium supplemented with 1 :'o histidine assay medium (Difcoj.

Complementation tests. Both qualitative and quantitative tests were carried out as described by Merrick et al. (1978). In most cases recA recipients were used. The problem of high-frequency homogenotization previously encountered when attempting to analyse nif: : Tn5 mutations by complementation (Merrick et al., 1978) was observed to be red-dependent and was therefore overcome by the use of recA derivatives of the chromosomal point mutations used.

Mapping the extent of deletions. Chromosomal deletions were mapped by crossing plasmids carrying nif' point mutations into the deletion strains. For these crosses the donor plasmid strains were patched on to a nutrient agar (NA) plate to produce a master plate with up to 50 patches per plate which was incubated over- night at 37 "C. The following day 0.1 ml of an overnight culture of the recipient deletion strain was spread on a nitrogen-free (NFDM) agar plate and the master plate was replicated on to this lawn. The resultant plate was incubated anaerobically for 6 d at 30 "C. The exconjugants were scored as Nif+ when recombinants grew in the mixed patch but not in unmated controls. The mapping of pCEl nifdeletions was performed by scoring recombinants after introduction of the plasmids into a set of chromosomal point mutants and deletions as described below for the mapping of Mu insertions.

Mapping inversions. Strains carrying inversions were recognized by the fact that they failed to complement some cistrons in a plate complementation test but could be demonstrated to have the relevant nifDNA by recombination analysis with a range of nifpoint mutations. In some cases the presence of an inversion was confirmed using P1 transduction (Kleckner et al., 1979). In reciprocal crosses such strains acted as good donors for markers within the inverted segment but gave low numbers of recombinants when used as recipients.

Fine-structure map of Klebsiella nif genes 51 1 Mapping itzsertions. Tn7 insertion mutations were selected in the plasmid pRDl and were mapped against

a range of chromosomal deletion mutants using the procedure described above. Tn5 and TnlO insertions were selected in the plasmid pMFlOO and as the transfer frequency of pMFlOO is lower than that of pRDl (M. Filser, unpublished) a modified mapping procedure was used. The mutant plasmids were transferred into the recipient deletion by spotting together 0.05 ml samples of overnight cultures on selective minimal medium plates and incubating overnight at 37 "C. The exconjugants were then replicated on to NFDM plates and incubated anaerobically for 6 d at 30 "C. Alternatively, 0.1 ml of the recipient strain was spread on an NA plate and a master NA plate carrying patches of the donor strains was replicated to the lawn. This mating plate was incubated for 7 h at 37 "C and then replicated to an NFDM plate which was incubated anaerobically for 6 d at 30 "C. Mutants were scored as mapping outside a deletion when Nif+ recombinants were observed in the exconjugant patch but not in unmated controls.

Mu insertions in the plasmid pCEl were mapped against chromosomal deletions which were lysogenized with Mu derivatives (Elmerich et al., 1978) to avoid zygotic induction during matings. In addition, recipients were made histidine-requiring and streptomycin-resistant when necessary. The mapping procedure was as follows. Donors and recipients were grown overnight in complete (LB) medium at 30 "C. Donors were distributed into microculture containers (Elesa, Milan) and matings were performed by spotting with the aid of a multipoint inoculator about 10 pl of 25 different donors on an LB plate previously seeded with 0.1 5 ml of the recipient. The mating plates were incubated for 16 h at 33 "C then replicated to NFDM plates supple- mented with aspartate (100 pg ml-l), streptomycin (250 ,us ml-') and kanamycin (20 pg ml-'). The resultant plates were incubated anaerobically for 6 d at 30 "C. Mutants were scored as mapping outside a deletion when Nif+ recombinants were observed in the exconjugant patch. All mutants were checked for the produc- tion of recombinants or revertants when the plasmids were introduced into the nif total deletion UNF107.

Mu insertions in the K. pneumoniae chromosome were mapped against nifdeleted plasmids using the same procedure as above.

RESULTS

As different cistronic nomenclatures were used by Merrick et al. (1 978) and MacNeil el a!. (1978) a standard nomenclature for the nifgene cluster has been agreed and will be used for the remainder of this report and in all future publications. The original designations and the new standard designations are given in Fig. 1.

Isolation of deletion strains Tetracycline-sensitive derivatives of TnlO insertions either retained their original Nif-

phenotype or were deletions or inversions with one end-point apparently at the original insertion site. In no case did tetracycline-sensitive derivatives give a Nif+ phenotype suggest- ing that precise excision of the transposon was rare.

With Tn5 and Tn7, selection for histidine auxotrophy allowed independent screening for deletion or inversion formation and for loss or retention of the transposons. A variety of excision events were observed including three classes of deletion: class I, with an end-point at the original insertion site; class 11, deletions extending through the original insertion site; class 111, where the deletion did not extend as far as the insertion site. Examples of the three classes of excision event are given in Fig. 2. All three classes couId be obtained from a single insertion mutation. In class I and class I1 derivatives the antibiotic resistance markers of the transposon were often retained but no attempt was made to determine whether the trans- poson was still nif-linked or whether it had been relocated on the chromosome. Whenever possible, deletions which had lost the drug resistance phenotype of the transposon were used

Merrick et al. (1978) 1 Elmerich et al. (1978)f - A/L - I

MacNeil et al. (1978) Q B A/L F M V S - N E K D H J

Standard nomenclature Q B A/L F M V S U N E K D H J

Fig. 1. Standard nomenclature for K. pneumoniae ng cistrons. 33-2

512 M. M E R R I C K A N D OTHERS

Plasmid nif mutation

H i s ' B A F M V S U N E K D H J Tp

nifE2182::Tn7 + + + + + + + + rec rec + + + + R Parent UNFl l42

Class I Class I1 Class I11

- rec + + + + R o r S

R or S

- - - - - - - - - - - - _ - _ - _ _ - d _ -

+ + + + + r e c r e c + + + + R - - - Fig. 2. Examples of the three classes of deletion obtained from Tn7 or Tn5 insertions. The parent strain UNFll42 and his derivatives were tested in a plate complementation assay against plasmid- borne point mutants for each nif cistron: + , complementation; -, no complementation; rec, some Nif+ recombinants. All strains were scored for trimethoprim (Tp) resistance (R) or sensitivity (S).

for mapping purposes in preference to deletions which had retained the phenotype. Class I11 derivatives always retained the drug resistance markers of the transposon.

Three categories of heat-resistant survivors were obtained from strains carrying Much plasmid insertions: (1) mutations with partial nifdeletions with an end-point at the original insertion site (analogous to class I above), these strains were either Mu-immune or Mu- sensitive; (2) mutants with total or partial nif deletions extending through the original insertion site (analogous to class II), these were all Mu-sensitive and, where examined, total nif deletions had lost 50 to 70% of the plasmid DNA; (3) mutants with no nif deletion which were apparently impaired in Mu-killing functions and were all Mu-immune.

Reversion analysis N i f : :Tn5 mutants reverted to Nif+ at a frequency of lov6 to lo-' and all revertants tested

retained kanamycin resistance. Spontaneous kanamycin-sensitive derivatives of Tn5 occurred at a frequency of less than 5 x and it is not possible to enrich for kanamycin- sensitive derivatives.

Only 5076 of the nif::Tn7 mutants tested reverted to Nif+ and the reversion frequency was lo-* to Again all revertants retained the antibiotic resistance markers of the transposon. Precise excision of Tn7 was never observed following selection for trimethoprim sensitivity.

Not all TnlO mutants reverted to Niff but of those which did the reversion frequency varied from to The majority of Nif+ revertants retained tetracycline resistance. Hence, with all three transposons (Tn5, Tn7 and TnlO), when precise excision was selected by reversion of the nif::Tn mutation to Nif+, nearly all revertants retained the antibiotic resistance markers of the element but no attempt was made to determine the location of the transposon in these revertants.

Mutations caused by the insertion of wild-type Mu do not revert, at least in Escherichia coli (Taylor, 1963). Several Muc+ insertions in each cistron were tested for reversion. Nif+ revertants were obtained only in the case of nifA8502: : Mucf, with a frequency of about

A proportion of the survivors from Mucts insertions should carry MuX prophages (Bukhari, 1975) which are known to revert. However, no Nif+ revertants were obtained from those survivors tested.

Mapping Insertion mutations in the plasmids pRD1, pMFlOO and pCEl were mapped against a

large number of chromosomal deletion strains as described in Methods. When an insertion was located within a deletion interval covering two adjacent cistrons assignment of the insertion to a specific cistron was based on complementation analysis. A total of 397 insertion mutations were mapped into 84deletionintervals in the nifgene cluster (Figs 3 and4).

Fine-structure map of Klebsiella n$ genes 513

514 M. M E R R I C K A N D OTHERS

Muc'

Mucts

TnlO

Tn7

Tn5

Point mutations

I I I

I I I

I I I

I I I I I

I I I

I

I I

8501 8080

I

I I I

I I I I I I I I I I I I I I

I I I I I I I I I I I I

1 I I

I I

I I

I I I I I I I 1 I I I I I I I I I I I I

I I I I I I I I

126121 I 5741261 11255312581

I I I I I I I I I I I I I I I I I I I I I

I I I I

I I I I I I I I

I I I I I I I I I I I I I

I I I

I

8503 8186 8182 SOP7

I I I I I I I I I I I I I I

I831 266 I I I

I I

589 556l

I I I

8d03 4066

8 5'44

!ii202 -4303 I

I ' 2474, 2632

2476 -2515 2475

2496 x 7 0 G 2630 2493

I I I I I I I I I I I I

I I 1 I I I

I

I I I I I I

I I

I I I

I I

I I I 1 I I I

I I I I

I l l 1

5691 2559 I I I l l 1

I I I I I

I I I I I I I I I I I

X I X I I

07 I' 054;x< I

p I

8571

85hO

I I

I I I I

hXd h73l hh3#22(

1 I I I I

I?(h I I I I I I

m 5

1 2 5 1

I246

i

Fig. 4. Map of nifdeletions in the K. pneumoniae chromosome. Deletions were mapped against point and insertion mutations in plasmids pRD1, pMFlOO and pCEl and allele numbers for these mutations are listed above each gene. Allele numbers for deletions are given at the end-point of the dele tion.

To facilitate comparison with the physical map of Reidel et al. (1979) the following should be noted: for Mu inserticns 8000 should be added to the figures given by Reidel et al. (1979; Fig. 5 ) to derive relevant allele numbers; for Tn5 insertions 2300 should be added to pMF numbers to derive allele numbers; for TnlO insertions pMF125 is nifB2018, pMFlO6 is nifA2013, pMF103 and

All nij'insertion mutants reported in Figs 3 and 4 gave rise to Niff recombinants when introduced into recipients carrying a point mutation in the gene in which the insertion was located. We therefore assumed that extensive deletions were not created upon insertion.

TnlO insertions in the nif gene cluster were significantly non-random, with the majority of insertions occurring in two deletion intervals within nifJ. Tn5, Tn7 and Mu insertions appeared to be random and insertions were obtained in 13 cistrons. No insertions were obtained in nifQ and this was probably due to the leaky nature of the mutations or a problem of definition (see later).

The plasmid pRD209 was found to carry a mutation nif-2209: : Tn7 mapping in ngU and

I I I I I l l ) I l l 1 I I 121901 1225 11221 71 21 811 I l l ) I 126091 I

-. . . I l l 1 I I I I I I I I I I I I

Fine-structure map of Klebsiella n$ genes

I I I I I I I I I I I I I 1 I I I t I f I I I 1 I I I I I I

5081 I 5071 I

;&;8;; 061180871 053180691

I I I I

0021 80401

I I 1 1

108' ; I I I I

027: I

I 1 I I I I

1841 I 182122581

9 1 I I I I

2555 I

I I I I I I 1 I I I

I I I I I I I I I I I

I I I I I

I I I I I

121! I

I I 1 I

2:/

I l l I l l I l l

I I I

I I I I I I I I I I I I I I

I I I

530: I 1 I I I I I I I I I I I

5651

I I I I I

24n

sz$ I I I I

I I I I

I I 1 I I I

I I I I I 1 I I I I

I I I I I I I 1 I I

I I I I I I I I I I

_ _ . . 12574

2.5'2 254

515

Much } } TnlO

} Tn7

) Tn5

Point mutations

2411

2415 8702

2472

-7482

!499

UNF2038 are nifM2011, pMF104 and UNF2040 are nifS2012, pMFll3 and UNF143 are nifiV2014, pMF114 is nifE2027, pMF124 is nifK2038, pMF107 is nifJ2016, pMF102 and UNF2027 are nifJ2009.

not in n$S as previously reported. Nif-2 184 : : Tn7 was confirmed as being in nifEas suggested by complementation data (Merrick et al., 1978). The plasmid pLS23, previously reported as a n$E::Mu insertion on the basis of complementation data, was reassigned as nifS8023 : : Muc+.

Complementation analysis and polarity Eleven complementation groups [NifB, nifA, nifF, n$M, n$S (previously nifiv), niJN

(previously n$I), n$E, n$K, nifo, nifH and nifJ] were defined by Merrick et af. (1978). A further two cistrons nifQ and n f V were described by MacNeil et al. (1978). We have now

Tab

le 2

. Com

plem

enta

tion

anal

ysis

of

nifU

2461

R

esul

ts a

re n

orm

aliz

ed to

a p

erce

ntag

e of

the

activ

ity g

iven

by

the

Nif+

pla

smid

(pR

DI,

pM

FlO

O o

r pC

E1) i

n ea

ch m

utan

t bac

kgro

und.

R

ecip

ient

alle

le

7

h_

7

Plas

mid

A

107

dono

r (g

nd h

ir

alle

le

nifB

4106

ni

fA22

63

nifF

4066

nif

M40

93 n

ifS2

442

nifU

2461

nif

N24

14

nifE

205

nifK

4026

ri

ifD

2290

ni

jH22

60

riifJ

2408

n

if)

Non

e 0.

03

0.02

3-

5 5.

4 4.

0 6.

0 0.

6 0.

2 0.

04

0.01

0.

02

2.2

0.06

N

if +

10

0 10

0 10

0 10

0 10

0 10

0 10

0 10

0 10

0 10

0 10

0 10

0 10

0 ni

fU24

61

82

72

47

107

85

6-1

86

18

0 65

40

56

69

5.

5

Tab

le 3

. Com

plem

enta

tion

anal

ysis

of

inse

rtio

n m

utat

ions

in t

he n

ifU

SV

M g

ene

clus

ter

Res

ults

are

nor

mal

ized

to a

per

cent

age

of t

he a

ctiv

ity g

iven

by

the

Nif

+ pl

asm

id (p

RD

1, p

MFl

OO

or

pCE1

) in

each

mut

ant b

ackg

roun

d,

Dat

a fo

r nif

: :Tn

5 m

utat

ions

are

from

com

plem

enta

tion

anal

ysis

with

recA

der

ivat

ives

of t

he re

cipi

ent m

utan

ts li

sted

. R

ecip

ient

alle

le

h

Plas

mid

-

\ r

dono

r al

lele

N

one

Nif+

ni

fM80

46 : :

Muc

f ni

fM25

77 : : Tn

5 ni

fM26

62 : :

Tn7

nifV

2249

: : Tn

7 nifS8031: : M

uc+

nifS

2569

: : T

n5

nifS

219.

5 : :

Tn7

nif

U81

16::

Muc

ts

nifU

2562

: : Tn

5 ri

ifU

.220

9: : T

n7

A10

7 (g

nd hi

s n

if)

0.02

10

0 6 7 11

46 0-7

0.8

3 3.3

0.7

5

nifF

4066

4 10

0 47

78

45

27

125 68

81

95

25

55

nifM

2I 0

4 ni

fM40

93

0-2

5.4

100

100 4.

4

8 j

-

17

48

76

-

i8

55

49

-

62

10

-

-

-

-

nif V

4994

45

10

0 85

105

137 42

51

73

51

-

-

-

nifS

2442

ni

fS20

12

nifU

2461

100

100

100

15

-

28

40

-

62

34

-

54

0.9

95

64 7

-

1.0

-

-

-

I" '1

44

48

-

4.5

nifN

24 I4

100 99

0.6

-

-

-

53

-

-

85

10s -

Fine-structure mgp of Klebsiella nif genes 517

recognized a new complementation group defined by the point mutation nif-2461 and designated nijU (Table 2). NifU2461 was 38 % cotransducible with hisD2 and mapping data indicated a location between n$S and nifN (Fig. 4). Six n f U : : Tn and ten nifU: : Mu insertion mutations have been isolated. All these insertions were strongly polar on nifs (Table 3) . No polarity of nifU insertions on nifN or n$E or of n$N insertions on nifU was observed and hence we conclude that nifU and n$S belong to a single transcriptional unit transcribed from n f U to niJS.

The effect of either nifU or n$S insertions on both nifV and nifM is very difficult to ascertain. All complementation studies with nifV are complicated by the leakiness of the nifV mutations (see above) and no clear polar effects of nifU or nifs insertions can be observed. No polarity of n i fu or n$S insertions on nifM is observable by acetylene reduction (Merrick et al., 1978; and Table 3), but some effect is apparent by plate complementation. The same situation applies to deletion nif-2472 starting outside nifJ and ending in nifS which has no significant effect on nifM activity in assays but some effect in plate complementation.

Some effect of nifM: : Tn5 insertions on nifF complementation has consistently been observed in plate complementation but not by acetylene reduction (Merrick et al., 1978). This effect was confirmed with all nifM: : Tn5 mutants reported here but no such effect was observed with other insertions in n$M.

No unambiguous nifQ mutants were identified in this study but three insertion mutants (nif-2739: : Tn7, nif-2740: : Tn7 and nif-2574 : : Tn5) were isolated which are nifB in com- plementation and are leaky in acetylene reduction. The two Tn7 insertions map within the nif-4303 deletion of MacNeil et al. (1978) and are therefore withinnifQ as defined by MacNeil et al. (1978). Leaky Mu-induced nifB mutations were also described by MacNeil et al. (1 978) and hence it is not clear to us whether nifQ is definitely an independent cistron.

NifV was described by MacNeil et al. (1978) and it was reported that all non-polar nifV point mutations are very leaky. One such strain UN1990 (nifV4944) had 45% wild-type activity when assayed by us for acetylene reduction and grew quite well anaerobically on nitrogen-free medium. The allocation of mutations to the nifV cistron was therefore impossible by acetylene reduction assays and extremely difficult by assessment of growth on nitrogen-free medium. Assignment of mutations to nifV was based on plate complementa- tion, leaky phenotype and map position. NifV: :Tn7 mutations had activities of 50 to 100 ;h by acetylene reduction even though such mutations should completely inactivate the nifV gene product.

Using complementation analysis, polarity studies have been carried out with all the insertion mutations described here. The results suggest that the nif gene cluster contains seven or eight transcriptional units :

nyJ, nifHDK, nifEN, nifUS( V), n$M, nifF, nifAIL, nifB(Q) This is essentially in agreement with previous reports (Elmerich et al., 1978; MacNeil et al., 1978 ; Merrick et al., 1978), with the exception of the nifusVM cluster which may contain two transcripts (see Discussion).

It is not yet clear whether nifA/L comprises one or two genes. MacNeil et al. (1978) suggest that nifL [originally designated as a separate gene by Kennedy (1977) on the basis of transductional mapping] is not required for growth on nitrogen and hence mutations in nifL are only detected if they are polar on nifA. There is therefore no precise phenotype for all nifL mutations and we have not attempted to define a boundary between nifA and nifL (Figs 3 and 4). The ability of some Mu insertions to give Niff revertants was the property used by MacNeil et al. (1978) to divide the nifA complementation group into two genes. Mutations in the five his-distal deletion intervals of the nifA region were assigned to n f L on the basis that the majority of Mu insertions in this region reverted to Niff. A similar distinction could not be made with Tn5, Tn7 and TnlOmutations as these transposons

518 M. M E R R I C K AND O T H E R S

can excise precisely. Reversion to Nif+ was observed with one Mu insertion nifA8502: : Mucf isolated in this study and located in the his-distal part of the nifA/L region.

D I S C U S S I O N

Virtually all studies of transposition and related phenomena to date have been carried out in Escherichia coli and/or Salmonella typhimurium, but there appear to be some dif- ferences in behaviour of the insertion elements used in this study when examined in Kleb- siella pneumoniae.

The site specificities of integration into the nqgene cluster are similar to those reported for other genetic systems in E. coli using Tn5 (Berg, 1977), TnlO (Kleckner et a/., 1977) and Mu (Bukhari & Zipser, 1972). Tn7 has been reported to integrate preferentially at one site in the E. coli chromosome (Barth et al., 1976) but, in contrast, no significant specificity of Tn7 integration into nif was observed. Furthermore, the transposition of Tn7 into RP4 has been reported to cause frequent deletions of up to 2 megadaltons at the point of insertion (Barth et al., 1978). No such deletions were detected when Tn7 was inserted into the nif gene cluster.

The characteristics of deletions and inversions formed by TnlO in K. pneumoniae appeared to be similar to those described in S. typhimurium (Kleckner et a/., 1977) although the deletion end-points may be more random than observed in Salmonella (Dale Noel & Ferro-Luzzi Ames, 1978). The generation of deletions with one end-point always at the original insertion site is also observed with Tn3 (Nisen et al., 1977).

In contrast, Mu produces two classes of deletions in both E. coli and K. pneumoniae (Toussaint et al., 1977; and this paper): deletions with one end-point at the insertion site (class I) and deletions extending to both sides of the insertion site (class 11). Deletion forma- tion by Tn5 and Tn7 has not previously been reported, but in the nifgene cluster these transposons can produce three classes of deletion: class I and class 11, as described above for Mu, and class I11 deletions where the transposon is always retained but the deletion is not continuous with the original site of the insertion. In class I11 deletions we did not analyse whether or not the DNA between the original insertion site and the deletion end-point was inverted. If an inversion is involved in the generation of class I11 strains the transposon could be located at either end of the inverted segment. Such strains are probably not stable and can apparently become deleted for the inverted segment (M. J. Merrick & M. Filser, unpublished observations).

The reversion frequencies of nifmutations generated by Tn5, Tn7 and TnlO are similar to those quoted for other genes in E. coli and S. typhimurium, i.e. lacZ::Tn5 (Berg, 1977), tra::Tn7 (Barth & Grinter, 1977), hisG::TnlO (Kleckner et al., 1979) and lac2::TnlO (Foster, 1977). However, with Tn5 and TnlO in E. coli and S. typhimurium the majority of revertants are drug-sensitive, presumably as a result of precise excision and loss of the transposon. Around 1 % of Tn5 insertions in E. coli lac are kanamycin-resistant and it is suggested that in these revertants the transposon was re-inserted in a new site prior to the reversion event. By comparison, all Niff revertants obtained in K. pneumoniae from nif: : Tn5 and nif: : Tn7 mutations and most revertants from n$: : TnlO mutations retained the trans- poson. In the majority of cases the phenotype of the nif: : Tn mutation precludes reversion by a suppressor mutation since most nifgenes are believed to code for structural proteins and Mu insertions in all nif genes except nifL cannot be suppressed. Therefore, these revertants must either have arisen in a clone in which transposition had occurred prior to the reversion event, or reversion must be linked to transposition. These differences in precise excision of the transposon between E. coli and K. pneumoniae may reflect differences in host proteins involved in the transposition process.

Whilst the polar effects of the insertion mutations described here suggest a transcriptional

Fine-structure map of Klebsiella nifgenes 519 organization which is essentially in agreement with previous reports (Merrick et al., 1978 ; Elmerich et al., 1978; MacNeil et al., 1978), the precise transcriptional organization of the central cluster nifUSVM is difficult to determine. The absence of any strong polar effect of nifU and nifS insertions on nifM may indicate that either there is a secondary promotor between nifS and nifM which can allow transcription initiation after nifS or that nifUS and nifM represent separate transcriptional units. Both possibilities are supported by the complementation phenotypes of deletions extending into nifS or nifV from the right-hand side of the nifcluster. Such deletions, which have been described in this paper and which were also described by MacNeil et al. (1978), complement nifM mutants even though the nifUS promotor is deleted. MacNeil et al. (1978) suggest that this result is due to the fusion of nifA4 to other promotors within or outside the nifcluster. However, the majority of their deletions extend outside nif and would therefore have to be fused either to constitutive promotors or to promotors under similar control to the n i f u s promotor.

The insertion of a transposon into a nifgene increases the size of the EcoRI restriction fragment on which the gene is located by an amount equal to the size of the transposon. This increase can be exploited in conjunction with suitable radioactive probes derived from amplifiable plasmids carrying cloned nif DNA fragments (Cannon et al., 1979) to determine the distribution of the nifgenes on EcoRI restriction fragments. Using this approach 85 nif insertion mutations, whose genetic mapping and complementation analysis is described in this paper, have also been assigned physical locations in the K. pneurnoniae nifgene cluster (Reidel et al., 1979).

The availability of both a genetic and a physical map of a large number of insertion mutations in the nifgene cluster allows comparison to be made between the two mapping methods. The physical map was constructed quite independently of the genetic map reported here and is virtually identical to the genetic map in the relative allocations of the insertion mutants. However, when making comparisons between the two maps the following dis- crepancies should be noted. The plasmids pMF320 and pMF321 carry Tn5 insertions in nifN (nifN2620 and nif'2621, respectively) and not in nifE as indicated by Reidel et al. (1 979). Plasmids pMF308 and pMF3 18 were inadvertantly interchanged on the physical map and hence pMF308 and pMF318 refer to nifJ2618 and nifJ2608 on the genetic map.

Some of the TnlO strains mapped by Reidel et al. (1979) carried the same nif: : TnlO allele either on the plasmid or on the chromosome; these are UNF2038 and pMF103 both carrying nifA42011::Tn10, UNF2040 and pMF104 both carrying nifS2012::Tn10, UNF143 and pMFll3 both carrying nifN2014: :TnlO and UNF2027 and pMFlO2 both carrying nifJ2009::TnlO. In the case of nifM2011::TnlO the orientation of the transposon was apparently reversed as a result of transduction from the plasmid to the chromosome. In all these cases the two insertions were mapped to within 0.2 kilobase of each other (as estimated from the physical map) and hence if the location of the transposon was unchanged by transduction from the plasmid to the chromosome this is an indication of the sensitivity of the physical mapping technique.

When the above points are taken into consideration it is clear that the agreement between the genetic and physical maps of the nifcluster is extremely good. Whilst these two methods are not interchangeable they are clearly complementary. For example, if one wishes to clone small fragments of the nif cluster, such as particular nif promotors, suitable mutants can be chosen which will be within the required fragment, thus allowing selection of the required clones by marker rescue.

The authors wish to thank Dawn Allitt, Simon Harrison and Nicole Charpin for skilled technical assistance. L. S. was the recipient of a predoctoral fellowship from the DClCgation Gknbrale A la Recherche Scientifique et Technique (D.G.R.S.T.). The investigation per- formed by C. E., L. S. and J. H. was supported by research funds from the University Paris 7 and by a grant from the D.G.R.S.T. (no. 78-7-0440).

520 M. M E R R I C K A N D OTHERS

R E F E R E N C E S

BARTH, P.T. & GRINTER, N. J. (1977). Map of plasmid RP4 derived by insertion of transposon C. Journal of Molecular Biology 113, 455-474.

BARTH, P. T., DATTA, N., HEDGES, R. W. & GRINTER, N. J. (1976). Transposition of a deoxyribonucleic acid sequence encoding trimethoprim and strepto- mycin resistances from R483 to other replicons. Journal of Bacteriology 125, 800-810.

BARTH, P. T., GRINTER, N. J. & BRADLEY, D. E. (1978). Conjugal transfer system of plasmid RP4: analysis by transposon 7 insertion. Journal of Bacteriology 133, 43-52.

BERG, D.E. (1977). Insertion and excision of the transposable kanamycin resistance determinant Tn5. In DNA Insertion Elements, Plasmids and Episomes, pp. 205-212. Edited by A. I. Bukhari, J. A. Shapiro & S. L. Adhya. New York: Cold Spring Harbor Laboratory.

BUKHARI, A. 1. (1975). Reversal of mutator phage Mu integration. Journal of Molecular Biology 96,

BUKHARI, A. I. & ZIPSER, D. (1972). Random insertion of Mu-1 DNA within a single gene. Nature New Biology 236, 240-243.

CANNON, F. C., REIDEL, G. E. & AUSUBEL, F. M. (1 979). Overlapping sequences of Klebsiella pneumoniae nif DNA cloned and characterized. Molecular and General Genetics 174, 59-66.

DALE NOEL, K. & FERRO-LUZZI AMFS, G. (1978). Evidence for a common mechanism for the insertion of the TnlO transposon and for the generation of TnlO-stimulated deletions. Molecular and General Genetics 166, 217-223.

DIXON, R. A., CANNON, F. & KONDOROSI, A. (1976). Construction of a P plasmid carrying nitrogen fixation genes from Klebsiella pneumoniae. Nature, London 260, 268-271.

DIXON, R., KENNEDY, C., KONDOROSI, A., KRISHNAPILLAI, V. & MERRICK, M. (1977). Complementation analysis of Klebsiella pneu- moniae mutants defective in nitrogen fixation. Molecular and General Genetics 157, 189-198.

ELMERICH, C., HOUMARD, J., SIBOLD, L., MANHEIMER, I. & CHARPIN, N. (1978). Genetic and biochemical analysis of mutants induced by bacteriophage Mu DNA integration in to Klebsiella pneumoniae nitrogen fixation genes. Molecular and General Genetics 165, 181-189.

FOSTER, T. J. (1977). Insertion of the tetracycline resistance translocation unit TnlO into the lac

87-99.

operon of Escherichia coli K12. Moleculur and General Genetics 154, 305-309.

KENNEDY, C. (1977). Linkage map of the nitrogen fixation (nif) genes in Klebsiella pneumoniae. Molecular and General Genetics 157, 195-204.

KLECKNER, N., ROTH, J. & BOTSTEIN, D. (1977). Genetic engineering in vivo using translocatable drug-resistance elements. Journal of Molecrrlar Biology 116, 125-159.

KLECKNER, N., REICHARDT, K. & BOTSTEIN, D. (1979). Inversions and deletions of the Salmonella chromosome generated by the translocatable tetracycline resistance element Tn 10. Journal of Molecular Biology 127, 89-1 15.

MACNEIL, T., MACNEIL, D. , ROBERTS, G. P., SUPIANO, M. & BRILL, W. J. (1978). Fine- structure mapping and complementation analysis of nif (nitrogen fixation) genes in Klebsiella pneumoniae. Journal of Bacteriology 136, 253-266.

MERRICK, M., FILSER, M., KENNEDY, C. & DIXON, R. (1978). Polarity of mutations induced by insertion of transposons Tn5, Tn7 and TnlO into the nif gene cluster of Klebsiella pneumoniae. Molecular and General Genetics 165, 103- 1 1 1 .

NISEN, P. D., KOPECKO, D. J., CHOU, J. & COHEN, S. N. (1977). Site-specific DNA deletions occur- ring adjacent to the termini of a transposable ampicillin resistance element (Tn3). Journal of Moleciilar Biology 117, 975-998.

REIDEL, G. E., AUSUBEL, F. M. & CANNON, F. C. (1979). Physical map of chromosomal nitrogen fixation (nif) genes of Klebsiella pneumoniae. Proceedings of the National Academy of Sciences of the United States of America 76, 2866-2870.

ROBERTS, G. P., MACNEIL, T., MACNEIL, D. & BRILL, W. J. (1978). Regulation and characterisa- tion of protein products coded by the nif(nitrogen fixation) genes of Klebsiella pneumoniae. Journal of Bacteriology 136, 267-279.

TAYLOR, A. L. (1963). Bacteriophage induced mutations in E. coli. Proceedings of the National Academy of Sciences of the United State.\ of’ America 50, 1043-1051.

TOUSSAINT, A., FAELEN, M. & BUKHARI, A. I. (1 977). Mu-mediated illegitimate recombination as an integral part of the Mu life cycle. In DNA Insertion Elements, Plasmids and Episomes, pp. 275-285. Edited by A. I. Bukhari, J. A. Shapiro & S. L. Adhya. New York: Cold Spring Harbor Laboratory.