Amino acid sequences of cytochromes c2 and c′ from the moderately halophilic purple phototrophic bacterium Rhodospirillum salexigens

Biochimie (I 994) 76, 583-591 © Socirt6 fran~aise de biochimie et biologie mol~culaire / Elsevier, Paris

583

Amino acid sequences of cytochromes c2 and c' from the moderately halophilic purple phototrophic bacterium Rhodospirillum salexigens RP Ambler a, M Daniel a, TE Meyer b, M D Kamen c alnstitute of Cell and Molecular-Biology, University of Edinburgh, Edinburgh, EH9 3JR, UK; ,~Department of Biochemiso~y. University of Arizona, Tucson, AZ 85721; ,'300 Hot Springs Road, Montecito, California, USA (Received 20 February 1994; accepted 10 March 1994) / Summary - - Rhodospirillum salexigens is a moderately halophilic purple phototrophic bacterium which grows optimally in 8% NaCI. The amino acid sequences of the two principal soluble cytochromes c have been determined. One of mese is a cytochrome c2, similar in size to mitochondrial cytochrome c. While clearly of the same sequence class as mitochondrial cytochrome c and the proteins fi'om several other Gram-negative bacteria, it does not show particular affinity to ~ y already known sequence in terms of the percentage sequence identity. The other protein is a cytochrome c', but is also a divergent member of this widespread group. The lack of appreciable sequence identity to other species is probably due to a limit of divergence which has been reached for the majority of purple bacterial species. However, the numbers of insertions and deletions and their locations in cytochromes c2 and c' suggest that R salexigens may be related to Rhodospirillum molischianum. Like other electron transport proteins from halophiles, both of these cytochromes are notable for their high content of arginine as compared with lysine and both are acidic. However, they do not show any particular sequence homology to electron transport proteins that have been characterized from the extremely halophilic phototrophes of the genus Ectothiorhodospira. Thus, it appears that adaptation to halophilic habitats has independently occurred more than once in purple bacteria. R salexigens I cytochrome c, / cytochrome c'

Introduction There are several p o s s i b l e a p p r o a c h e s to m e a s u r e m e n t o f the relatiofiships b e t w e e n different isolates o f bacteria, r a n g i n g fl'om m o r p h o l o g y and m e t a b o l i s m to the comparison o f the whole g e n o m e at the D N A level. T h e method that is m o s t c o m m o n l y e m p l o y e d [ 1] is to c o m p a r e the partial or c o m p l e t e sequences o f 16S r R N A across a set o f organisms and m a k e generalizations about their relationships. Sneath [2] has pointed out some o f the statistical p r o b l e m s o f this approach and the desirability o f using information f r o m as m u c h o f the g e n o m e s as possible, though recognizing the lack o f precision in current hybridization methods. We have for m a n y years b e e n exploring the diversity o f soluble electron transport proteins in bacteria [3-9] and believe that the analysis o f the distribution o f the types o f proteins and a comparison o f their sequences gives information that c o m p l e m e n t s that derived f r o m other approaches. In fact, we believe that it is the method o f choice for c o m p a r i s o n o f purple phototrophic bacteria.

In the present paper, we report the amino acid sequences o f the major soluble cytochromes c from • the moderately halophilic purple phototrophic bacterium Rhodospirillum salexigens [10, 11], which was p l a c e d in the o~-1 16S r R N A group [12, 13]. Our results do not support the conclusions f r o m rRNA comparisons, but instead suggest a relationship to R mo!ischianum rather than to Rhodospirillum rubrum.

Materials and methods Materials Pseudomonad protease was a gift from Dr GR Drapeau. Staphylococcal protease was made in the Department of Molecular Biology, Edinburgh. Aminopepfidase M, carboxypeptidase A, and mouse submaxillary protease were from Boehringer Mannheim. Clostripain was from Sigma. R salexigens strain YC6. I was kindly provided by Dr Hanno Biebl, who isolated it from Solar Lake, Israel. The type strain WS68 (DSM 2132) was isolated by Dr WR Sistrom from a lake in Oregon and was a gift from Dr Gerhart Drews [10]. Bacteria were grown and cytochromes purified according to Meyer et al [Ill.

584 from strain WS68, that from strain YCt+I was not exan~ined at all. The determination of the amino acid sequences of the cytochromes c~ did not present any notable difficulties. They were available in large quantity, and when digested with any of the standard proteases, formed a readily separable mixture of soluble peptides that could be arranged to form overlapping sets and a complete sequence. The proteins are relatively rich in glutarnine, but this sometimes troublesome amino acid did not cause any problems. R salexigens cytochrome c~ does not contain a~]y particularly labile peptide bonds. There are no Asp-Pro sequences, and the AsnGlv sequence (105-106) seemed to be particularly stable. There was no suggestion during peptide purification that either of the two Asp-Gly sequences (44-45 and 77-78) had derived from very labile AsnGly sequences. A tryptic digest of the cytochrome ce from strain W$68 w~s compared with that of strain Y C t . 1. Peptides were analysed for amino acid composition, position on peptide maps, and N-terminus. The differences between the two cytochromes ce were consistent with the strain WS68 protein differing at three positions (Lys-60 (not Thr), Thr-67 (not Ala) and Thr-71 (not

Determination of amino acid sequencez The methods used have been described before [ 14--16]. Amide groups were assigned from peptide electrophoretic inabilities and exopeptidase analysis. Aminopeptidase M has been used to hydrolyze several peptides te virtuai completion, and was effective even on peptides containing cysteic ~,cid or praline. One peptide was examined with an automatic sequencer (Applied Biosystems 477A). The Drapeau [17] pseudomonad pro~ease wa~ used extensively and the strain WS68 cytochrome c2 was used as a test sequence with which to assess the specificity of the arginine-splitting enzyme~ clostripain and mouse submaxillary protease.

Results

The evidence for the proposed amino acid sequences of R salexigens cytochromes c2 and c' are summarized in figures 1 and 2. Details of the purification, analysis and sequence determination experiments are available from the principal author. The primary work on R salexigens cytochrome ce was done with the protein from strain Y C t . I , although cytochrome c2 from the type strain W568 was also examined to assess the amount of interstrain variation. The cytoc:hrome c' was

,~,:F2t

:F+

+Fz

-T-

F~4~

:HZ --:H1 ~

~:H4c

:F+

F2SI~

- ;H3 -

P-'+'+

F~l*

;.j.+;. :.:p..H=~ . , :.Ic

FSlI"

/F210**

F2Sb

Sl~pltylococcal p;~taaso I Staphylococcal proteasa

t+++..i

I

~ T t S T ~ I P

--

F37b**

t+:m.~.t.:_

T26b

T21I~, .......

T3~

T)15

T26a

t

F314

F~g|*"

Thermolys|n

~

--.tgZ---t+3'

TOI9

F37C*"

tz14* t311

TPP

._~.~

Tlitt

T~111

EGDAEAGKRVYQRQCLRCHVI 0EGQQRI GPSLYGVVGREAGSVDGYLPRYSQALKDATFTWDEE KIAHYI SEKDFLDGHRHHOLGYRGGLSEQDVQDV! AY l KANGGH~I~_ IO ~0 30 40 50 60 10 8~ 90 I00 tlu = S~iuenator . c21o,

c~3

c'4

•

,

¢27+-

.__c?~b

_

....

_~:31.__o;

c3~

C~

c211at

..

~x

Staphylococcal protease

__ m,o

__,3sa _ . , l a p,2,~'~~ ; ~,2~m.,~:,_ ,,.

r..."~7.::....~"~b.

,4t~ ,]Sc_,210~" _mTb ~ X mS~ ---- ~lS,21,

-- ,16"

Fig 1. Evidence for the sequence of cytochrome c2 from R salexigens retain YC6.1. Peptides derived by digestion with trypsin (T) and staphylococcal protease (F) are shown above the sequence, and by digestion with chymotrypsin (C) or thermolysin (H) below the sequence. Pep tides from sub-digests are labelled with a second letter indicating the second method of cleavage. Continuous lines in~:~ate quantitative amino acid analyses, with substandard analyses marked * and particularly bad ones **. See Ambler and Wyhn [ 14] for a quantitative definition of these symbols. Dashed lines indicate peptides that were not analysed quantitatively. The peptide lines are doubled where the sequence was deterrained by the dansyl/phenyl isothiocyanate method, with the lower line broken where the identification was inconclusive. C-terminal residues identified as free amino acids after removal of the remainder by phenyl isothiocyanate degradation are indicated by a vertical line joining the double lines at the end of a peptide. Peptides marked t were examined by carboxypeptidase A digestion, and ~ by aminopeptidase M digestion. The double line under the start of the sequence shows results determined with a Beckman model 890A automatic sequencer. The protein from R salexigens strain WS68 is believed to have the same sequence as that shown apart from three positions (Lys-60 (not Thr), Thr-67 (not Ala) and Thr-71 (not Ser)). Double complementary differences within the same tryptic peptide would not have been detected.

585 ".IIS,

•

:m.

.,~r ~,,~ I

~

. _ _ _ _

.

.

1

.

_:r

.

_

I~n'.

- -

~m.

~c~

i~l

i

. ,,a~.

1

l

--..

tilt.

..... ".'|l.l~

Til! ~

°'

__,I.__

T,,',sllrl

__~

QDVSPEQAH~OHRQATIo • I SGSGHAGVTLLRHQQGLzo X 1 [KKflVPQLFDYHATGARLSYEAFRPDTRGADVKTEAKDAso l,O I HAHBSDFEAAHTAYAEDVEE LAALAADG KTDEA. ~ I~

Sub .....

In5

111~4" -] S~4~|°~lcgi|

IIIYI5

I~lSa

~

I~ll~t ~.1

~rotlAI!

X~a

xllmi

~

....

,,

ul)

iMlr

ill4

_ _ l ~ . I lm.. . . .

~

I

I~11

Illr~b •

]

diP* m

1

- t~,.

--

i~r~,

r.tm"

1

~QIDll'lrplll

:ulm..

,ns,,'~

l~l II|n

I-IP"

~.u.t+lJ~.. --

~

HGPIFKH ~RGCHHOYRT.~.

1

II'ITI" --

Nllllb

[.,lift"

,tit,-

•

~

-

:roT..~t7._

.~-.,-

-

Fig 2. Evidence for the sequence of cytochrome c' from R salexigens strain WS68. Peptides derived by digestion with trypsin (1") are shown above the sequence, and by digestion with thermolysin (H) or cleavage with cyanogen bromide (X) below the sequence. Peptides from sub-digests are labelled with a second letter indicating the second method of cleavage (D, pseudomonad protease, C, chymotrypsin, F, staphylococcal protease). Continuous lines indicate quantitative amino acid analyses, with substandard analyses marked * and particularly bad ones **. See Ambler and Wynn [14] for a quantitative definition of these symbols. Dashed lines indicate peptides that were not analysed quantitatively. The peptide lines are doubled where the sequence was determined by the dansyl/phenyl isothiocyanate method, with the lower line broken where the identification was inconclusive. C-terminal residues identified as free amino acids after removal of the remainder by phenyl isothiocyanate degradation are indicated by a vertical line joining the double lines at the end of a peptide. Peptides marked -~were examined by carboxypeptidase A digestion. The thick line represents a peptide sequence detemfined in an Applied Biosystems model 477A automatic sequencer.

Ser)). Double complementary differences within the same tryptic peptide would not have been detected and it is possible that the two strains of cytochrome c2 could differ by more than three positions. The proteins are richer in arginine than other known cytochromes c2 and opportunity was taken of this to assess the effectiveness of two reputed arginine-only specific proteases, clostripain [18] and mouse submaxillary protease [19] at producing fragments for sequencing. Indications of some hydrolysis at lysine was observed with both enzymes, but peptide 50--80, that contained four unhydrolysed lysine bonds, was recovered in 5% yield from the clostripain digest. The sequence of the cytochrome c' presented more difficulties, particularly at the N-terminus. Peptides derived from the N-terminus were blocked, probably by cyclization of glutamine during purification. There is no lysine residue near the N-terminus, so small peptides that include the region were not detectable with ninhydrin. For no obvious reason, trypsin did not seem to hydrolys¢ the Arg-Gln bond (13-14), but peptide T-0 (1-29) was readily obtainable in a pure s~te as the largest fragment separated by gel-filtration through Sephadex G-25 in 5% (v/v) formic acid at pH 2. The other large fragment (T-core, 73-104) was completely insoluble at this pH. Cyanogen bromide cleavage was not very helpful, as the largest fragments were formed by breaking a Met-Gin bond, with the result ~hat the peptides had become blocked by the time they had been purified. Another of the methio-

nine residues precedes a threonine (86--87), and it is known that such bonds do not cleave well with CNBr [20]. In the only large CNBr fragment that was isolated in a recognizable form (X-II, 31-112) this bond had not been broken. Atternpts to obtain an N-terminal sequence of cytochrome c' in an automatic sequencer, either before or after treatment with pyroglutamate aminopeptidase have not yet been successful. Despite these difficulties, we have deduced the sequence shown in figure 2. The weakest overlap is at positions 52-53, but no unaccountable peptides were isolated from any of the digests so we believe it to be correct. The evidence whether residue 2 is asparagine or aspartic acid is tenuous, being derived from the electrophoretic mobility at pH 6.5 of the histidine-containing peptide T-0D6. Residue 14 was clearly identified as glutamine rather than glutamic acid in an automatic sequencer degradation, despite the expectation from protease specificity that it be the latter. The evidence from the electrophoretic mobility at pH 6.5 of peptide H35d (+0.30, Ala-Asp-His-Arg-Glx) also supports this assignment. Discussion

Identification of cytochrome c2 Physical and spectral methods are not yet adequate to assign a bacterial cytochrome c to the appropriate

586 structural class [7] and the complete determination of amino acid sequence is the only reliable method for doing so. For example, there are a number of low-spin cytoehromes c with high redox potentials and similar absorption spectra which have been assigned to a number of either homologous or unrelated groups based on amino acid sequence [5, 7]. The R salexigens cytochrome c_, was however identified prior to sequence determination by a novel method, which takes advantage of the fact that all cytochromes c2 have positive charge at the site of reduction even w h e n the overall charge of the protein is negative. Thus, the kinetics of reduction of R salexigens cytochrome c2 by FMN semiquinone were measured as a function of ionic strength and it was found that there was a plusminus interaction due to a plus one charge at the active site of the cytochrome even though the oxidized protein has a net charge of minus 5 [ 111. The alignment of figure 3 shows that the high redox potential cytoehrome from R salexigens is in fact homologous to the cytochromes c_,. The proximal charges due to arginines at positions 13, 17, 27, 80, and 87 apparently dominate the larger electrostatic field due to acidic residues on the 'backside" of the protein away from the active site. Note that arginines in the R salexigens cytochrome c2 replace the normal lysines in the mitochondriai cytochromes c and non-halophilic cytochromes c,.. This will be discussed more fully below. The cytochrome c' from R salexigens also is homologous to those of purple bacteria as shown in figure 4.

Evolutionary relationships A percentage identity matrix is often used to quantify relationships among protein sequences and~ by extrapolation, to the host organisms as well. However, Meyer et al [21] found that the cytochromes c, and other redox proteins from purple p|~,,~c:,,rophicbacteria have generally diverged to the maximum extent which still permits them to function. This means that divergent mutations are balanced by convergent (the sum of back and parallel) mutations for the majority of comparisons. The matrix of sequence identities then gives a measure of the percentage of residues which are practically invariant because of the necessity to maintain the structure/function relationship. This region of balanced change is 40:!: 15% for cytochromes c2 and 25 + 15% for the cytochromes c ~.Only identities greater than 55% and 40% for cytochromes c: and c', respectively, represent significantly closer relationships. The R salexigens cytochromes average 35% and 21% similarity to all other cytochromes c2 and c'. R salexigens cytochrome c2 is 35-42% identical to the most closely related sequences shown in figure 3. R salexigens cytochrome c2 is most similar to Euglena gracilis mitochondrial cytochrome c (42%),

but that is too close to the average and is not significant. Based upon this criterion aloac, R salexigens cytochrome Cz appears to be equidistant from all other cytochromes c2. However, we do not believe that this is the true relationship of R salexigens to other purple bacterial species, but that the difference matrix does not provide sufficient data to evaluate such relationships for distantly related species. R salexigens cytochrome c' is 20-22% identical to the sequences shown in figure 4. It is most similar to Rhodobacter capsulatus (25%) but this is also not significant since it is the same as the average for all cytochromes c'. R salexigens cytochrome c' thus appears to be equidistant from all others based upon this criterion. The matrix of sequence identities therefore gives no clue as to the quantitative relationship of R salexigens to other purple bacteria, except that it is a very distinct species. To ati~mpt any further conclusions would be to overinterpret the matrix data. Insertions and deletions may actually give a better representation of relationships than do matrices of sequence identities for distantly related proteins provided that they are precisely located [21]. Insertions and deletions are particularly well-characterized in cytochromes c2 based upon three-dimensional structures [22]. Thus, cytochromes c2 may be separated into large and small categories based upon shared 3 and 8 residue insertions and a single residue deletion (at positions 62-64, 86-93, and 105). R salexigens cytochrome c2 is 'small' like those of R molischianum [3] and Rhodopila globiformis [23] and appears to share a single-residue insertion with Rp globiformis just before the heme (at position 18 of figure 3) and a single-residue deletion with both species just after the heine (at position 29). A three-residue insertion at positions 99-101 just after the sixth heine ligand methionine is in the same location as a two-residue insertion in R molischianum iso-1 cytochrome c2 [3]. Because these insertions differ in size and sequence, it is possible that they may have been independent events or that successive insertions occurred at the same location. A two-residue insertion at positions 53-54 in the midsection is in the same location as a single-residue insertion in the 'large' eytochrome c, from Rhodobacter sphaeroides [3]. These are almost certainly independent events, because 'large' and 'small' cytochromes c2 are separated by three gaps as illustrated by the comparison with R rubrum shown in figure 3. A single-residue deletion at residue 81 is unique to R salexigens. We tentatively conclude from analysis of insertions and deletions that R salexigens cytochrome c2 may be distantly related to that of R molischianum based on possible shared gaps at positions 29 and 99-101. There are several insertions and deletions in the cytochromes c', which have been positioned based

3 J.~

.,~

1"11 ~,

? ~ ,,;..

~.~

~ ~..~_.~

I~,,, ~_. ~

,,-,. ~ , ~

~-

~.

,,~

~'~,~

~,~..

~.~

t'a., M=

aE~7

~;~'~

~=.~_.~>

~~.

I

I I

I I

I

i

I

I~I =

I~ I io

~I~

3.

H

L~

~L°

~@

L'*

~

H

L'~

p,

°1

¢)

i - ~ ~ i ~ ~1 ~

I

!

I

=

~

I

=

~

I-I

o

o

o

I-,

~1 ~ ~

t~

5"1

L'*

~ H


L~

I~

I L~

~

L'*IH

H

~ ~

~

H

~J

~D o

(D

-.I

~1° ~ ~1 ~

L~

~l~r-~

L~

L~

I I

~3~

I

"~'

I

I

I~

I

i

d°r

=I~

I

I

I

~

I

I

~

I

I

I

~

I

I

I

I

I

L"

~I

~'~

° °1

~ ~

i

<~

I

~

I

t'=l

~"

I-~-EI,

tlj

~

I

I

¢1

<~

I

~CI

L"'

)

~)

I

',~

I-~%-I

~

1,4

o

t~a

"

i-i

us

..a

L/I

~°

•

~"

~.~ I ~

~I~

~

~

Ill

n

~

~

~

~

{~

LTJ

t

~

Ill

~II

I

n

n

~

~

<

:~'~

~

~

~

~

t

I ~

~

~

~'~

~ -

~1

L'I'J

~"~.-.

. ~

~'

~

,°°°°°°,

<1"

~

~-

~

F;'~I~'

~

m

i

~..,.~

a~

~e

~

~:~:_

~'~

r~ ~ "

•

o

Ill

n

~

~

~

H

~

~:l

t

~'

~'

•~.

I

~

~I-~

I=

n

~l,~

~,

~

~

~1

I

t

~'

~'

~,

o

nl~

~

1::2

I-~

~

~

o

I-"

~'1o ~

~'1 ~'1

I~l

~,

~

~

~

2:

I

¢

~

~1

I

|--i

I-~-I
~,

,.e

2:

I

~1

[]

t~

I~

IO

•

i

I

I

I

~

~

m

t~

I

III

I

I

I

I

~

~

L~

I=l

rJ)

t~

I

I

I

~I

~

-~

2:

I

¢

!

i

°°

~D

00

,.4

o

~

~

Ill

l I

L~

~

I-I

O~

~

~

~

I

~

{1 I

m

I..~..~4

>

~

t¢

4:.

I:l

~

~

~

~'~

Un

~

~,

I

~

1,1

>

~

I~

~

1

~

~

I

I

~

3:

L~

~

~

~,

~

I

I

L~

~3

I-~

I~

1

=

I,!1 1.11

o

,1:.

W

,

~)

Ii

I-~

~

,

~

<

II

x

<

~

,

~

~

M

x

~

H

,

~

~

H

xl~

~

<

~

~

~

H

~

~

i0

,

~o

I~

a

>

<

i0

o

W

M

o

OQ O0

589 upon comparison of three-dimensional structures [24]. R salexigens cytochrome c' appears to share a single residue insertion at position 108 with that of the halophilic Paracoccus sp [25]. This is the only possible relationship to another halophile although it is not convincing in itself. R salexigens cytoehrome c' shares a deletion at position 54 with that of R molischianum. Therefore, insertions and deletions in both cytochromes c2 and c' indicate a relationship to R molischianum. Perhaps the best indication of the relationship of R salexigens to other purple bacteria is in the types of redox proteins found [6]. The presence of both cytochromes c2 and c', but not HiPW is typical of a number of purple bacteria, such as R molischianum and R rubrum. However, most species with this pattern contain a 'large' cytochrome c2 (eg R rubrum). R molischianum is the only species, besides R salexio gens which has been found to have a small cytochrome c2 plus a c', but not HiPIP. If this observation is combined with that of two possibly shared insertions and deletions with R molischianum cytochrome c2 and one with cytochrome c', then the conclusion that R salexigens and R molischianum may be related is reinforced. One should keep in mind that even these modest conclusions may not be warranted until genetic analyses are used to show that there is only one copy of each cytochrome gene and that isozymes are not expressed under special growth conditions, which might be more closely related to the genes of other species than to that of the main type. Examples are the small cytochrome c2 isozymes discovered following deletion of the main type large cytochromes c2 in Rb sphaeroides [26] and Rb capsulatus [271. These cytochromes have very different evolutionary histories from those of the main type proteins. Although these isozymes presumably have different functions, they could replace the main type protein (and isozymes may have done so in other species) leading to false conclusions about evolution of the genome. Thus, single genes are not necessarily representative of the whole genome. Furthermore, negative results of protein distribution analyses should not be weighted as heavily as positive results since some genes may not be regulated the same in all species. R salexigens strain YC6.1 was originally thought to be a strain of Ectothiorhodospira mobilis, but the sequence of the cytochrome c2 shows that only three substitutions (K60, T67, and T71) were identified in the type strain WS68 relative to strain YC6.1. Furthermore, there are more differences than similarities between the types of redox proteins found in R salexigens and Ectothiorhodospira species [6], such as Ectothiorhodospb'a shaposhnikovii [28, 29] and Ectothiorhodosph'a halophila [30] which have HiPIP but

not cytochrome c2. It appears that R salexigens and Ectothiorhodospira species evolved independently for life in halophilic e:avironments. Eventual sequence determination of Ectothiorhodospira cytochromes c' should shed more light on this comparison. The genera and species of purple bacteria were rearranged by Imhoff et al [31] who suggested that further revision would be required in the future. Furthermore, it was recommended that taxonomy reflect genetic relationships which should take precedence over phenotypic properties, although new species would have to be phenotypically distinct [32]. The genus Rhodospiriilum is thus in need of revision, with at least four different kinds of purple bacteria included. By our analysis of electron transfer proteins [3, 4, 6], R rubrum and Rhodospirillum photometricum are the only two species which should be retained in the genus. R molischianum and Rhodospirillum fulvum are so closely related as to be strains of the same species, thus R fulvum should be made a synonym for R molischianum. It is problematic whether R molischianum and R salexigens should be considered members of the same or different genera although they are obviously different species. Rhodospirillum salinarum is clearly different ft'om all of the above. Rhodospirillum centenum appears to he related to Rhodopseudomonas palustris (Van Beeumen et al, unpublished) and the two last mentioned species should be the only members of the genus Rhodopseudomonas, which, like Rhodospirillum, is currently a collection of distantly related species. It therefore appears that many new genera of purple phototrophic bacteria may be necessary. The placement of R salex~'gens in the or-1 group of 10S rRNA along with R rub,,'um, Rp globiformis, and R molischianum agrees only partially with the relationships established by the sequences of electron transfer proteins. R salexigens cytochromes c2 and c' may be related to those of R molischianum, but neither species is very close to R rubrum. The sequences of cytochrome c2 [231 and HiPIP [33] from Rp globiformis show that this species is actually closer to Rhodomicrobium vannielii, Rhodopseudomonas acidophila, and Rhodopseudomonas viridis than to any of the above. These latter species were placed in the tz-2 group of 16S rRNA [12, 13]. This suggests to us that the rRNA results have been overinterpreted and that much more sequence data will be required to es'ablish the best means of analyzing the data and to determine precise relationships. It is probable that convergent mutations in rRNA lirnii use of the matrix of sequence identities in the same way as fournd for cytoctnomes c2 and c'. Thus, the only points of agreement to be expected between cytochrome c2 trees and 16S rRNA trees is where the species are particularly closely related, such as the comparison of R molischianum and R fidvum or Rb capsutatus and Rb sphaeroides.

590

Adaptation to the haiophilic environment

References

It has been observed that proteins from halophiles are more acidic than are those of non-halophiles and have elevated levels of arginine [25]. We previously found that HiPIPs from a halophilic Paracoccus and E halophila have a net charge which averages minus ! 3 + 1, but that of fresh-water species average minus 4 + 4 [34, 35]. In the present study, R salexigens cytochrome c_, was found to have a net charge of minus 4, whereas the average oxidized cytochrome c2 has a net charge of plus 1 4- 2 [31. Paracoccus denitrificans cytochrome c,. which has a net charge of minus 7 [36, 37] is the only fresh-water species containing a cytochrome c, more acidic than that of R salexigens. R salexigens cytochrome c' has a charge of minus 10, whereas the average non-halophilic cytochrome c' has a charge of minus 2.5 + 4 [4]. The halophilic Paracoccus sp cytochrome c' has a charge of minus 21 [251. The only fresh-water cytochrome c' homolog which has a larger charge than that of R salexigens is Rb sphaeroides cytochrome e-554, which has a charge of minus 14 [38]. Although R salexigens is only a moderate halophile, it thus appears to conform to the pattern of acidic proteins observed with more extreme halophiles. An elevated arginine content relative to lysine was also observed in halophiles [25]. Arginines are apparently more effective than are lysines in binding anions [39]. Whether that might confer an advantage in halophiles is presently unknown. Arginine also increases thermal stability of proteins [40]. In the present study., we have found that there are five lysines and eight arginines in R salexigens cytochrome c2, whereas the average cytochrome c, (excluding R salexigens) has 13 4- 2 lysines and 2 + 1 arginines. R sale. xigens cytochrome c' has six lysines and seven arginines and the average cytochrome c' (excluding R salexigens and Paracoccus sp) has 14 + 3 lysines and 3 + 1 arginines. R salexigens grows at temperatures as high as 45°C, thus an elevated arginine content in cytochromes c_, and c' may increase their stability at elevated growth temperatures. The histidine content of the halophilic cytochromes c' is also high (four versus an average 2 + 1), although that of halophilic cytochromes c~ and HiPIP is average and may not be of general significance.

I Woese CR (1987) Bacterial evolution. Microbiol Rev 51.22 !-27 I 2 Sneath PHA (1989) Analysis and interpretation of sequence data for bacterial systematics: the view of a numerical taxonomist. Syst Appl Microbiol 12, 15-31 3 Ambler RP, Daniel M0 Hermoso J, Meyer TE. Bartsch RG, Kamen MD 0979) Cytochrtnne c= sequence variation among the recognised species of purple nonsulphar photosynthetic bacteria. Nature 278. 659-660 4 Ambler Ri~. Bartsch RG, Daniel M. Kamen MD. McLellan L, Meyer TE. Van Beeumen J 0981 ) Amino acid sequences of bacterial cytochmmes c' and c556. Proc Natl Acad Sci USA 78. 6854-6857 5 Ambler RP (1991) Sequence variability in bacterial eytochromes c. Bioehim Biophys Acta 1058. 42-47 6 Barlsch RG (1991) The distribution of soluble metallo-redox proteins in purple phototmphic bacteria. Biochim Biophys Acta 1058. 28-30 7 Meyer TE, Kamen MD (1982) New perspectives on c-type cytoehromes. Adv Pn~t Chem 35. 105-212 8 Meyer TE (199tl Evolution of cytochmmes and ph{~tosynthesis. Biochim Biophys Acta 1058, 31-34 9 Van Bceumen J (1991) Primary structure diversity of pmkaryotic diheme cytochromes c. Biochim Biophys Acta 1058. 56-60 I0 [hews G (1981) Rhodospirillum salexigens, spe¢ nov. an obligatory halophilic phototrophic bacterium. Asx'h Mierobiol 130. 325-327 I I Meyer TE. Fitch JC. Bartsch RG. Tollin G. Cusanovich MA 0990) Soluble cytochromes and a photoaetive yellow protein isolated from the moderately halophilic purple phototrophic bacterium. Rhodospirilhm~ salexigens. Bioehim Biophys A¢'ta 1016. 364-370 12 Woese CR. Stackebrandt E. Weisbmg WG. Paster BJ. Madigan MT. Fowler VL Hahn CM. Blanz P. Gupta R. Nealson KH. Fox GE (1984) The phylogeny of purple bacteria: the alpha subdivision. Syst Appi Microbiol 5. 315326 13 Kawasaki H. Hoshino Y. Yamasato K 0993) Phylogenetic diversity of phototrophic purple non-sulfur bacteria in the PlvJteobac'teria ¢~ group. FEMS MicrobiolLett 112. 61-66 14 Ambler RP. Wynn M 0973) The amino acid sequences of cytochromes e-551 from three species of Pseodomonas. Biochem J 13 I. 485-498 15 Ambler RP. Daniel M. Meyer TE. Barlsch RG. Kaman MD (1979) The amino acid .sequence of cytochmme ¢' from the purple sulphur bacterium Chrnma. tium vim~sum. Biochem J 177.819-823 16 Ambler RP. Daniel M. Melts K. Stout CD (1984) The amino acid sequence of the dihaem cytochromc c. from Ihe bacterium A'~owba('ter vilu.kuulii. Bin. c'hem J 222. 217-222 17 [hapeau. GR (1980) Substrate specilicity of a proleolytie enzyme isolated from a mutant of Pseudomonasfragi. J Bird Chem 255. 839-840 18 Mitchell WM. Harringlon WF 0970) CIostripain. Methods En'ytnol 19. 635--64 I 19 Levy M. Fishman L. Schenkein I (1970) Mou~ submaxillary gland proteases. Methods Enrymol 19. 672-68 I 20 Sehroeder WA. Shelton ]B. Shelton JR (1969) An examination of conditions for the cleavage of polypeptidc chains with cyanogen bromide: application to catalase. Arch Biochem Binphys 130. 551-556 21 Meyer TE. Cusanovich MA. Kamen MD (1986) Evkience against use of bacterial amino acid sequence data for construction of all-inclusive phylogenetie trees. Proc Natl Acod Set USA 83. 217-220 22 Benning MM. Wesenberg G, Caffn,'y MS. Bartsch RG, Meyer TE. Cusanorich MA. Raymem I. Holdcn HM (1991 ) Molecular structure of cytochrome e, isolated from Rhodobacter capsulatns determined at 2.5 A resolution. J 114o1Bio1220. 673-685 23 Ambler RP. Meyer TE. Cusanovich MA. Kamen MD 0987) The amino acid sequence of the cytochrome c~ from the phototrophic bacterium Rhodopseudomonas globiformis. Bin,'hem J 24h. 115-120 24 Pen Z. Meyer TE. McRee DE 0993) Atomic structure of a cytochrome e' with an unusual ligand-controlled dimer dissociation at 1.8 A resolution. J Mol Biol 234. 433--445. 25 Ambler RP. Daniel M. McLellan L. Meyer TE. Cusanovieh MA. Kamen MD (1987) Amino acid sequences of cytochrome c-554(548) and cytochrome c' from a halnphilic denitrifying bacterium of the genus Pat'aeoc'cus. Bioehemt J 248. 365-371

Acknowledgments This work was supported in part by National Institutes of Health grant GM 21277 and the Virginia Swanson Trust. The automatic sequencer experiments were done at the WELMET Protein Characterization Facility, University of Edinburgh, which is supported by the Weilcome Trust and the Salvesen Foundation.

591 26 Roll MA, Witthuhn VC. Schilke BA. Soranno M, All A. Donohue TJ (1993) Genetic evidence for the role of isocytochrome c: in photosynthetic growth of Rhodobacter sphaeroides Spd mutants. J Baeteriol 175, 358-366 27 Jenney FE Jr, Daldal F (1993) A novel membrane-associated c-type cytochrome, cyt c . can mediate the photosynthetic growth ofRhodobacter capsulatus and Rhodobacter sphaeroiae'x. EMBO J 12, 1283-1292 28 Kusche WH. Truper HG (1984) Cytnchromes of the purple ~ulfur bacterium Eetothiorhodospira shaposhnikoviL Z I~tu.~j~rsch 39c, 894--901 29 Kusche WH. Tmper HG (1984) Iron sulfur proteins of the purple sulfur bacterium Ectothiorhodospira shaposhn~kovii. Arch Mierobiol 137. 266 -27 I 30 Meyer TE (1985) Isolation and chzaacterization of soluble cytochromes, ferredoxins and other chromophoric proteins from the hatophilic phototrophic bacterium Ectothiorbodospira haA'~phila. Biochim Biophys Acta 806, 175183 31 lmhoff JF. Troper HG. Pfennig N (1984) Rearrangement of the species and genera of the phototrophic 'purple nonsulfur bacteria*. Int J Syst Bacteriol 34. 340-343 32 Wayne LG. Brenner DJ. Colwell RR, Grimont PAD. Kandler O, Krichevsky MI, Moore LH, Moore WEC. Murray RGE, Stackebrandl E, Start MP. Troper HG (1987) Report of the ad hot: committee on reconciliation of approaches to bacterial systematics, lutJ Syst Bacterio137, 463--464 33 Ambler PP. Meyer TE, Kamen MD (1993) Amino acid sequence of a high redox potential ferredoxin (HiPIP) from the purple phototrophic bacterium Rhodopila globiformis, which has the highest known redox potential of its class. Arch Biochem Biophys 306, 215-222 34 Tedro SM. Meyer TE. Kamen MD (1977) Primary stmcture of a high potential iron-sulfur protein from a moderately halophilic denitrifying coccus. Y Biol Chum 252, 7826-7833 35 Tedro SM, Meyer TE, Kmnen MD (1985) Amino acid sequence of highredox-polential ferredoxin (HiPIP) isoaymes from the extremely halophilic purple phototrophic bacterium, Eetothiorhodospira halophila. Arch Biochem Biophys 241. 656-664

36 Ambler RP, Meyer TE, Kamen MD, Schichman SA. Sawyer L ( 1981 ) A reassessment of the structure of Paracoccus cytochrome c-550, J Mol Biol 147. 351-356 37 Van Spanning RJM. Wansell C, Harms N, Oltmann LF, Stouthamer AH (1990) Mutagenesis of the gene encoding cylochrome c-550 of Paracoccus denitrificans and analysis of the resultant physiological effects. J' Bacteriol 172, 986-996 38 Bartsch RG, Ambler RP, Meyer "rE. Cusanovich MA (1989) Effect of aerobic growth conditions on the soluble cytochrome content of the purple phototrophic bacterium Rhodobacter sphaeroides: Induction of cytochromc c-554~ Arch Biochem Biapi~ys 271,433-440 39 Riordan JF 0979) Arginyl residues and anion binding sites in proteins. Mol Cell Biochem 26, 71-92 40 Mmbet NT, Van den Broeck A, "Van den Bra~;iie I, Sl~ssens 13. Laroche Y, Lambeir AM, Matthijssens G, Jenkins J, Chiadmi M, Van Tilbeurgh H, Rey F, $anin J, Quax WJ, Laslers L DeMaeyer M, Wodak SJ (1992) Arginine residues as stabilizing elements in proteins. Biochemistry 3 I, 239-253 41 Pettigrew GW (1973) The amino acid sequence of cytochrome c from Euglena gracilis. Natut~ 241, 531-533 42 Dus K, Slettcn K, Kamen MD (1968) Cylochrome c: of RhodospiriUum rubrum. 11. Complete amino acid sequence and phylogenelic relationships. J Bio! Cbem 243, 5507-5518 43 Bhatia GE ( 1981 ) Refinement of the c~'st, al structure of oxidized Rhodospirillum rnbrum cytochrome c~. PhD Thesis, University Of. California at San Diego 44 Meyer TE, Ambler RP, Barlsch RG, Kamen MD (19=/5) Amino acid sequence of cytochrome c' ii'om the purple photosynthetic bacterium Rhodospirillum rubrum SI. J Biol Chum 250, 8416---842I 45 Finzel BC. Weber PC, Hardman KD, Salemme FR 0985) Structure of ferricytochrome c' from Rhodospirillum molischianum at 1.67 A i~esolution. J Mol Biol 186, 627-643 46 Yasui M, Harada S, Kai Y, Kasai N, Kusunoki M, Matsuura Y (1992) Threedimensional structure of ferricytochrome c' from Rhodospirillum rubrum at 2.8 A resolution. J Biochem 11 !, 317-324