TIBS 18 - MARCH 1993

THE COMPOUND 5-formylH4folate*, also known as leucovorin, folinic acid and citrovorum factor, is a ubiquitous mem- ber of biological folate pools, usually comprising between 10-25% of total cellular folates 1'2 (Fig. 1). Although it was the first one-carbon derivative of H4folate to be isolated and structurally characterized, little is known about its metabolic function. Despite this limited knowledge, 5-formylH4folate, known clinically as leucovorin, has been used for over 40 years in the treatment of cancer. It is widely administered both as a rescue agent for methotrexate tox- icity and in combination with 5-fluor- ouracil. Our limited knowledge of leuco- vorin may be partially due to a failure to pursue several key observations made in the 1940s and early 1950s. This review recounts the history of this elus- ive compound and discusses recent studies which are beginning to unravel the possible metabolic role of 5- formylH4folate.

Folic acid is a required vitamin in mammalian organisms which upon re- duction to its tetrahydro form, H4folate, functions primarily as a family of co- factors by carrying one-carbon units necessary for the synthesis of purines, pyrimidines, methionine and formyl- methionine. Shortly after its discovery in 19403, it was recognized as a necess- ary factor in cellular proliferation and almost immediately became the focus for the development of an entire class of antifolate drugs, which continue to be used clinically as antineoplastic and antimicrobial agents.

The physiologically functional form of the cofactor contains between two and nine glutamate residues linked by peptide bonds through the terminal ?-carboxylate of glutamate. While the primary function of the glutamate chain is the subject of continued debate, it does aid in cellular retention of the co- factor and greatly increases the affinity and reactivity of the cofactor for most folate-dependent enzymes. In fact,

*The term 5-formyltetrahydrofolate, as used in this article, includes all poiyglutamate derivatives.

P. Stover is at the Department of Biochemistry, Virginia Commonwealth University, Box 614 MCV Station, Richmond, VA 23298, USA. V. Schirch is at the Department of Nutritional Sciences, University of California, Morgan Hall, Berkeley, CA 94720, USA.

102

Interest in determining if leucovorin, known chemically as 5-formyltetra- hydrofolate, plays a role in one-carbon metabolism is reemerging. While investigations in the 1940s suggested it was an important donor of one- carbon units in folate-mediated biosynthetic reactions, studies between the 1950s and 1980s disproved this hypothesis and dismissed its pres- ence in biological systems as artifactual. Recently, new data has focused attention on the possible biological function of this compound that is widely used in cancer chemotherapy.

mammalian cells lacking the activity to add the additional glutamate residues display auxotrophies for glycine, thymi- dine and purines 4. The cofactor is local- ized prim.arily in the mitochondria and cytosol with its role in mitochondria still the subject of debate.

In vivo, there are six known one-car- bon substituted derivatives of H4folate, with the one-carbon moiety carried at the oxidation level of either formate, formaldehyde, or methanol. Each oxi- dation state is associated with a par- ticular metabolic cycle in the cytosol (Fig. 2). 5-MethylH4folate is required for the synthesis of methionine and other methylation reactions (cycle A) 5. 5,10- methyleneH4folate is required for the synthesis of thymidylate (cycle B), and 10-formylH4folate is required for the synthesis of purines (cycle C). There

are three additional formyl derivatives of H4folate which are not directly in- volved in biosynthetic pathways: 5-formi- minoH4folate, which carries one-carbon units derived from purine and histidine catabolism; 5,10-methenyiH4folate, a prosthetic group for the enzyme DNA photolyase; and 5-formylH4folate, for which no definitive role has been ascribed.

The reversible interconversion of Ser and H4folate to Gly and 5,10-methylene- H4folate, as catalysed by serine hydroxy- methyltransferase (SHMT), is the prin- ciple entry point of one-carbon units into one-carbon metabolism 5. SHMT is a pyridoxal phosphate-dependent enzyme present as both cytosolic and mito- chondrial isoenzymes. Recently, we have identified a second catalytic activity of SHMT, the irreversible hydrolysis of

H NH 2 .. M

HN I N,

0 I /

H / C ~ 0 N H

O~ C

I

O ~ c / O -

I CH2 I

C H 2 H I N ~ C ~ H

I CH I '

C H 2 . I

C ~ N ~ C ~ H II I 0 o .>Cxo .

Figure 1 Structure~of the diglutamyl form of 5-formylH4folate.

© 1993, Elsevier Science Publishers, (UK)

TIBS 18 -MARCH 1993

5,10-methenylH4folate to 5-formylH¢ folate, and demonstrated that this reac- tion is the probable source of this com- pound in vivo 6. While these studies may explain the occurrence of 5-formyl- H4folate in biological systems, the properties of the biosynthetic pathway for 5-formylH4folate has increased re- cent speculation about an active role for this compound in one-carbon metab- olism.

The history of 5-formylH4folate, 1948-1954 5-FormylH4folate was first identified

in 1948 as citrovorum factor, an acid- labile, base-stable growth factor pres- ent in yeast, ricebran and liver extracts which was found to promote growth for Leuconostoc citrovorum [now known as Pediococcus acidilactici (ATCC 8081)] 7. A relationship between folic acid and 5-formylH4folate was suggested by the observation that human and rat sub- jects fed increased levels of folic acid markedly increased their urinary ex- cretion of 5-formylH4folate 8. This re- lationship was further substantiated when partially purified 5-formylH4folate effectively replaced folic acid as a growth requirement for Staphylococcus lactis R [now known as Enterococcus hirae (ATCC 8043)] and Lactobacillus casei (ATCC 7469) 9. Subsequent ob- servations that Leuconostoc citrovorum required xanthine, guanine or hypoxan- thine in the absence of 5-formylH4folate implicated 5-formylH4folate in purine metabolism '°. A functional relationship between folic acid, thymidine and 5- formylH4folate was shown when thymi- dine or folic acid were only partially effective in promoting growth of L. citro- vorum, while their simultaneous ad- dition permitted marked growth.

Continued studies suggested that folic acid must first be converted to 5- formylH4folate before it can. be used for catalysis. Early evidence demonstrated that folic acid could be converted to 5- formylH4folate enzymatically in rat-liver slices u. A relationship between 5-formyl- H4folate and serine was demonstrated using pigeon-liver extracts. Incubation of [14C]formate with H4folate, ATP, NADP ÷ and glucose 6-phosphate resulted in [14C]serine formation. However, omis- sion of the glucose 6-phosphate resulted in the rapid formation of 5-formylH 4- folate 12. If it is assumed that the glucose 6-phosphate maintained a pool of re- ducing equivalents in the form of NADPH, formate is converted to serine under reducing conditions, otherwise, 5-formylH4folate is formed. An expla-

I .4,0..,. I I ,,,olat. I so~ B ~ - ~ Thymidylate

Ser,,~," [methyleneH4folate ] ] H4folate ]__ Purines

A _ L I,o-,orm,..4,o,.,.I NADP÷'~rJ ~ "w" NADPH /

Methionine ~-methylH4folate f f~methenyl.__H,folate" ]

ATP" ~ i [ 5-formylH4folate I

Figure 2 Metabolic pathways associated with cytosolic one-carbon metabolism.

nation for the role of glucose 6-phos- phate in diverting formate incorpor- ation from 5-formylH4folate to serine remained unexplained for nearly 40 years.

The first evidence that 5-formylH4- folate was not the principle one-carbon donor for the biosynthesis of these compounds came during the study of purine ring biosynthesis 13. 5-NH2-4-imi- dazole-carboxamide-5'-phosphoribotide was converted to inosinic acid in pigeon-liver extracts with the addition of ATP and boiled yeast extract. The substitution of 5-formylH4folate for yeast extract and ATP did not produce inosinic acid, while the addition of 5- formylH4folate with ATP was effective for inosinic acid formation. The authors postulated that 5-formylH4folate was converted to an active form in an ATP- dependent reaction, and this activated form was subsequently converted to 10- formylH4folate, which alone showed maximal activity towards inosinic acid formation. The authors further postu- lated that the active form of the co- enzyme was anhydroleucovorin (5,10- methenylH4folate). Thus, only six years after its initial discovery, the role of 5- formylH4folate in one-carbon metab- olism was called into question.

Concurrent with these studies were advances in understanding the chem- istry of folic acid and its reduced formyl derivatives. In 1951 there were two reports of the organic synthesis of a crystalline substance, one called folinic

acid and the other leucovorin, which had biological and chemical properties identical to 5-formylH4folate. Both were structurally identified to be 5-formyl- H4folate. The chemical interconversion of 5-formylH4folate and 10-formylH4- folate was also elucidated at this time 14-'6 (Fig. 3). 5-FormylH4folate was found to be stable at neutral or alkaline pH, but underwent dehydration under acidic conditions forming an imidazo- lium-type compound identified as anhy- droleucovorin (5,10-methenylH4folate) (Fig. 3, reactions C and B). Incubation of 5,10-methenyIH4folate at neutral pH resulted in the formation of 10-formyl- H4folate (Fig. 3, reactions B and A), which unlike 5-formylH4folate, was sen- sitive to oxidative degradation and lost its formyl group above pH 11.0. One of these studies in 1952 reported the for- mation of a hydrated form of anhydro- leucovorin named anhydroleucovorin B, formed by incubation of anhydro- leucovorin at pH 4.0, 100°C 16. This com- pound was similar to anhydroleuco- vorin, except in its degree of hydration, and was reconverted to anhydroleuco- vorin at pH 2.0. This observation was again not pursued and no subsequent report of this compound appeared until 1992.

The stability of 5-formylH4folate and the relative ease with which 5-formyl- and 10-formylH4folate could be chemi- cally interconverted not only cast fur- ther doubt on the possible metabolic role of 5-formylH4folate, but even its

103

TIBS 1 8 - MARCH 1993

,o / • % / . - %

I mt

methenylH4folate (anhydroleucovorin)

\N N ( H I

10-formylH4folate

/

B \5 / J ' - - -~ ,o / C

< ~ " ~ c / " - ~ _ : =

(11 S)-hydroxymethyleneH4folate

°IT l

N ~ G / N ' - ~ ~,~" ~'H

(11 Rl-hydroxymet hyleneH4folate (anhydroleucovorin B)

S / ~ N l O

."%N

5-formylH4folate

I

5-formylH4folate

Figure 3 Proposed mechanisms for the interconversion of the formyl derivatives of H4folate.

presence in biological systems. Some believed that the conditions used to extract and measure folates from cells resulted in the non-enzymatic forma- tion of 5-formylH4folate and, thereby, accounted for its presence in biological systems. This assumption was probably one of the major reasons the study of the biological role of 5-formylH4folate was neglected for nearly three decades.

1954-1989 During the following 35 years, many of

the folate-dependent reactions were eluci- dated and the associated enzymes puri- fied and characterized. Meanwhile, little progress was made regarding the bio- logical chemistry of 5-formylH4folate. Two enzymes that could metabolize 5- formylH4folate were identified. The first was an enzyme partially purified from porcine liver which catalysed reaction 1:

5-formylH4folate + L-GIu --+ H4folate + N-formyl-L-Glu (1)

However, this activity was demon- strated to be a less-specific activity of the enzyme N-formimino-L-glutamate: tetrahydrofolate formiminotransferase 17,

104

and the properties of this side reaction suggested it had no physiological sig- nificance.

The partial purification and charac- terization of the second 5-formylH4- folate metabolizing enzyme, methenyl- H,folate synthetase, was reported in 1957 TM (reaction 2):

5-formylH4folate + ATP -+ 5,10-methenylH4folate + ADP + P~ (2)

Since 1984 this enzyme has been puri- fied to homogeneity from Lactobacillus casei 19, rabbit liver 2° and, most recently, human liver 21. The enzyme-catalysed reaction was demonstrated to be essen- tially irreversible. This is the only enzyme known to date which uses 5- formylH4folate as a substrate. The effec- tiveness of leucovorin therapy as a res- cue agent for methotrexate toxicity is explained by the product of reaction 2,5,10-methenylH4folate, being rapidly converted to 10-formylH4folate and the other one-carbon derivatives of H4folate according to the reactions shown in Fig. 2. Since reaction 2 is irreversible it does not account for the cellular source of 5-formylH4folate.

During this time a body of literature describing the relative concentrations of the different derivatives of tetrahydro- folate in various biological materials ac- cumulated. 5-FormylH4folate accounted for between 10-25% of most cellular folate pools 1,2. However, the levels were found to be particularly high in soy b e a n s , 70% 22 , and may be the predomi- nant folate derivative in uredospores 23, promoting speculation that 5-formyl- H4folate was a storage form of reduced, substituted folates.

The mechanism for the non-enzymatic interconversion of 5-formyl-, 10-formyl- and 5,10-methenylH4folate was also ex- tensively investigated 24 (Fig. 3). Evi- dence was presented suggesting that a transient hydrated intermediate, be- lieved to be (11S)-hydroxymethyleneH4- folate, was formed during the hydrolysis of 5,10-methenylH4folate (Fig. 3, reaction B). Breakdown of the intermediate was under prototropic control, that is either 10-formylH4folate or 5-formylH4folate was formed depending upon protonation of N-5 (pKa= 4.5) or N-10 (pKa = -1.2) (Fig. 3, reactions A and C). This mechanism accounted for the observation that 5,10- methenylH4folate initially hydrolyses almost exclusively to 10-formylH4folate. 5-FormylH4folate, however, is the more thermodynamically favored compound, and therefore accumulates with time or at elevated temperature.

1989-1992 The first evidence for the possible

biological origin of 5-formylH4folate was not reported until 19906 during the study of an in vitro enzymatic system containing enzymes necessary for the conversion of formate to serine (Fig. 4). Like the results obtained in 1954 (Ref. 12) in liver extract, it was found that formate was incorporated into serine in the presence of reducing equivalents in the form of NADPH. However, once the supply of reducing equivalents was exhausted, formate was incorporated into 5-formylH4folate. The reaction re- sponsible for the formation of 5- formylH4folate was demonstrated to be the irreversible hydrolysis of 5,10- methenylH4folate by SHMT. The de- pletion of NADPH results in an increase in methenylH4folate, the apparent sub- strate for the SHMT-catalysed hydroly- sis to 5-formylH4folate.

In this reconstituted system, the 5,10- methenylH4folate was generated en- zymatically. However, when chemically synthesized 5,10-methenylH4folate was incubated with SHMT, a burst of 5-

TIBS 18 - MARCH 1993

formylH4folate was observed followed by a slower steady-state rate of forma- tion. The rapid phase of the reaction was demonstrated to be the prefer- ential conversion of an unidentified compound present in the chemically synthesized 5,10-methenylH4folate sol- utions. Further studies 25,2~ of both the non-enzymatic- and SHMT-catalysed formation of 5.formylH4folate suggested that they occur by the same mechanism and that the unknown compound giving rise to the burst was anhydroleucovorin B, first described in 1952 (Ref. 16). Mech- anistic arguments suggested that an- hydroleucovorin B is (llR)-hydroxy- methyleneH4folate (Fig. 3), a C-11 hydrated derivative of 5,10-methenyl- H4folate distinct from the hydrated derivative described in the formation of 10-formylH4folate which is believed to be (llS)-hydroxymethyleneH4folate (Fig. 3). A number of experimental obser- vations suggested that the llR isomer is an intermediate in both the SHMT- catalysed and non-enzymatic formation of 5-formylH4folate and is formed by the isomerization of the initially formed 11S isomer. The isomerization appears to involve an unusual ylide carbanion inversion. The breakdown of the puta- tive anhydroleucovorin B results only in the formation of 5-formylH4folate by both enzymatic and non-enzymatic reactions (reaction E, Fig. 3). Anhydro- leucovorin B is stable at neutral pH and has a much higher affinity for SHMT than 5,10-methenylH4folate, suggesting that it may be the in vivo substrate for the SHMT hydrolysis t o 5-formylH4- folate. It cannot be ruled out that there is an enzyme which catalyses the for- mation of anhydroleucovorin B (reac- tion D, Fig. 3), adding this compound to the list of cellular one-carbon deriva- tives of H4folate.

Further support for the enzymatic synthesis and possible metabolic role of 5.formylH4folate was provided by the in vivo studies of Bertrand and Jolivet 27. Using MCF-7 human breast-cancer cells, they found that inhibition of meth- enylH4folate synthetase (reaction 2) by 5-formyltetrahydrohomofolate resulted in a twofold increase in cellular 5-formyl- H4folate concentration with a concomi- tant and equal decrease in cellular 5- methylH4folate concentration. The in- crease in cellular 5-formylH4folate levels was correlated with decreases in both cell growth and rate of de novo purine biosynthesis. These effects were ex- plained by showing that the increased cellular levels of 5-formylHJolate de-

ADP + Pi

ATP @

Formate + H4folate

serine ~ ( a ) "-~ 5,1

glycine

10-formylH4folate

H +

H20

5,10-methen rlH4folate

(d) NADPH

0-methyleneH4folate . ~ 1 NADP +

Figure 4 Reactions involved in the enzymatic conversion of formate to serine. The enzymes in the cycle are: (a) serine hydroxymethyltransferase (SHMT); (b) lO-formylH4folate synthetase; (c) methenylH4folate cyclohydrolase; (d) methyleneH4folate dehydrogenase.

creased purine biosynthesis by inhibiting the folate requiring enzyme phosphoribo- sylaminoimidazolecarboxamide formyl- transferase (AICAR formyltransferase).

Interestingly, 5-formylH4folate was also observed to be a slow tight-binding inhibitor of SHMT. Therefore, the cata- lytic formation of 5-formylH4folate may represent a unique form of regulation whereby an enzyme utilizes a second, unrelated catalytic activity to produce a powerful inhibitor of its principle physiological activity 2s. In addition to inhibiting SHMT and AlCAR formyl- transferase, 5-formylH4folate has been observed to inhibit other enzymes in one-carbon metabolism including sar- cosine dehydrogenase, dimethylglycine dehydrogenase, methionyl-tRNA formyl- transferase, 5,10-methenylH4folate cyclo- hydrolyase, 5,10-methyleneH4folate dehydrogenase and dihydrofolate re- ductase (see Ref. 26 and references therein).

The only enzyme that uses 5- formylH4folate as a substrate is meth- enylH4folate synthetase, which cata- lyses the irreversible ATP-dependent conversion of 5-formylH4folate to methenylH4folate in what has been termed a salvage reaction. Together, SHMT and methenylH4folate synthetase constitute an apparent futile cycle that can buffer the concentration of 5.formyl- H4folate in the cell (Fig. 2, cycle D). As noted above, the true substrate for SHMT may be (11R)-hydroxymethyl- eneH4folate and other enzymes may be involved in its production. However, the existence and properties of this

putative futile cycle in regulating 5- formylH4folate concentrations in the cell is at the forefront of research regarding the metabolic role of this compound.

Conclusions 5-FormylH4folate, known clinically as

leucovorin, has been widely used for over 30 years in cancer chemotherapy, yet little is known about its metabolic role. After initial studies suggesting it as a one-carbon donor were disproven, and chemical studies indicated it could be formed non-enzymatically during cell extraction procedures, some doubted that it was even present as a cellular folate compound. In addition, several experimental observations from the 1940s and early 1950s that were key to understanding the metabolic role and chemistry of 5-formylH4folate were apparently forgotten. For nearly 35 years little was learned about the metabolic function of this compound. Recent methods have confirmed that 5- formylH4folate is present in cells and is not an artifact of isolation. Also, identi- fication of enzymatic reactions which both use and form 5-formylH4folate have been identified. These observations, and evidence that 5-formyiH4folate is an inhibitor of many folate-dependent en- zymes, have reawakened an interest in determining a possible metabolic role for this folate derivative.

Acknowledgement This work was supported by Grant

GM28143 from the NIH, USA.

105

TIBS 1 8 - MARCH 1 9 9 3

References I Cossins, E. A. (1984) in Folates and Pterins

(Vol. 1) (Blakely, R. L. and Benkovic, S. J., eds), pp. 1-60, Wiley

2 Horne, D. W., Patterson, D. and Cook, R. J. (1989) Arch. Biochem. Biophys. 270, 729-733

3 Snell, E. E. and Peterson, W. H. (1940) J. Bacteriol. 39, 273-278

4 Shane, B. (1989) Vitam. Horm. (N.Y.)45, 263-335

5 MacKenzie, R. E. (1984) in Folates and Pterins (Vol. 1) (Blakely, R. L. and Benkovic, S. J., eds), pp. 255-306, Wiley

6 Stover, P. and Schirch, V. (1990) J. Biol. Chem. 265, 14227-14233

7 Sauberlich, H. E. and Baumann, C. A. (1948) J. Biol. Chem. 176, 165-173

8 Sauberlich, H. E. (1949) J. Biol. Chem. 181, 467-471

9 Keresztesy, J. C. and Silverman, M. C. (1950) J. Biol. Chem. 183, 473-479

10 Sauberlich, H. E. (1949) Fed. Proc. 8, 247-248 11 Broquist, H. P., Kohier, R., Hutchison, D. J.,

Burchenal, J. H. (1952) J. Biol. Chem. 202, 59--66

12 Kisliuk, R. L. and Sakami, W. (1954) J. Am. Chem. Soc. 76, 1456-1457

13 Greenberg, G. R. (1954) J. Am. Chem. Soc. 76, 1458-1459

14 Cosulich, D. B. eta/. (1952) J. Am. Chem. Soc. 74, 3252-3262

15 May, M. eta/. (1951) J. Am. Chem. Soc. 73, 3067-3075

16 Roth, B. et al. (1952) J. Am. Chem. Soc. 74, 3247-3452

17 Bortoluzzi, L. and MacKenzie, R. E. (1983) Can. J. Biochem. Cell. Biol. 61, 248-253

18 Greenberg, D. M., Wynston, L. K. and Nagabhushanam, A. (1965) Biochemistry 4, 1872-1878

19 Grimshaw, C. E. et al. (1984) J. Biol. Chem. 259, 2728-2733

20 Hopkins, S. and Schirch, V. (1984) J. Biol. Chem. 259, 5618-5622

21 Bertrand, R., MacKenzie, R. E. and Jolivet, J. (1987) Biochim. Biophys. Acta 911, 154-161

22 Shin, Y. S., Kim, J. E., Watson, J. E. and Stokstad, E. L. R. (1975) Can. J. Biochem. 53, 338-343

23 Jackson, A. 0., Samborski, D. J., Rohringer, R. and Kim, W. K. (1970) Can. J. Biochem. 48, 1617-1623

24 Benkovic, S. J. (1980) Annu. Rev. Biochem. 49, 227-251

25 Stover, P. and Schirch, V. (1992) Biochemistry 31, 2148-2155

26 Stover, P. and Schirch, V. (1992) Biochemistry 31, 2155-2164

27 Bertrand, R. and Jolivet, J. (1989) J. Biol. Chem. 264, 8843-8846

28 Stover, P. and Schirch, V. (1991) J. Biol. Chem. 266, 1543-1550

COMPUTER CORNER

Not much to malign

Multalin 4.0

by Florence Corpet, Cherwell Scientif ic Publishing, 1992. £ 2 9 9 . 0 0 , £ 2 4 9 . 0 0 (academic)

There are several reasons for wanting to align protein or nucleic acid sequences for comparison. The three most common are perhaps (1) DNA sequencing projects, (2) the deduction of evolutionary relationships and (3) the identification of potentially homologous domains in proteins and RNA molecules. Such alignments are definitely a job for a computer program (as anyone who has tried to do this kind of thing by hand will know), and several programs have been developed to deal with some or all of these tasks. Multalin 4.0 is a stand-alone program for IBM PC and compatible computers designed with the last of these needs in mind. It performs similar functions to the GCG programs Lineup and Pretty, or the protein alignment module of the [ntelliGenetics GeneWorks program for the Macintosh, and can align a large number of protein sequences of average size in a very reasonable time.

Multalin works by first making a rough-and-ready pair-wise comparison of the sequences to be aligned, using the FASTP algorithm of Lipman and Pearson 1. The similarity scores are then used to calculate a hierarchical clustering of the sequences by the UPGMA (unweighted pair-group method using arithmetic averages) method z.

Multalin then applies a modification of the Needleman-Wunsch algorithm a,4 to

106

find the optimal alignments between pairs of sequences and between pairs of clusters until all the sequences are aligned. After this alignment is complete, the similarity scores and clustering are recalculated. If the new clustering is different from the previous one, it may be worth repeating the alignment process, since the Needleman-Wunsch algorithm makes use of clustering information in deciding which sequences to align first. Iteration of the alignment process in this way can improve the alignment slightly, but it takes time.

Similarity between pairs of sequences is scored using comparison matrices, together with a penalty for gaps in the alignment. The matrices and the gap penalty can both be adjusted by the user. Multalin can deal with sequences based on any alphabet of less than 32 symbols, but most users will probably rely on the various standard matrices supplied for protein sequence alignment (Dayhoff, Genetiq, Risler and Staden). Predefined matrices are also provided for comparing ambiguity-coded nucleic acid sequences.

How well does Multalin perform? In a side-by-side comparison between Multalin running on a modest 286-based PC, and lntelliGenetics GeneWorks 2.0 on a Macintosh IIci, Multalin came out clearly ahead. A simple alignment of a dozen proteins of 300 residues was completed by Multalin in less than two minutes, almost ten times faster than its rival. As sequence length goes up, and more sequences are compared, the time required grows rapidly since the number of comparisons increases approximately

2 as n(n-1)m where n is the number of sequences, and m their length 4. Any alignment algorithm based on clustering is likely to show a similar dependence.

In addition to its welcome speed, Multalin makes very sensible use of colour. Once an alignment has been created, it is displayed above a consensus sequence. At a given position, residues that match at 90% are displayed in one colour, 50% matching in another, and less than 50% in a third colour. All these colours (and the background colour) can be altered to taste, though the thresholds are fixed. Gaps can easily be added or subtracted to adjust the alignment, and doing so causes the consensus to be recalculated and the colours of residues to change appropriately. This represents far better use of colour than the lntelligenetics package.

Although Multalin has a commendably good and easy-to-learn user interface (which is usable either with or without a mouse), it has the odd annoying quirk. For example, to open several sequence files simultaneously, a file of file names must be created. It is not possible to specify a set of file names using wild-cards. Frustratingly, wild-cards are acceptable to the program in just about every other situation.

Apart from this, Multalin will accept sequences as unformatted text and in Genbank, EMBL, PIR or GCG format. Sequences are not recognized automatically however; the program must be told explicitly which format to expect.

Multalin can save its alignments in either its own simple text-based format or as standard GCG output, which facilitates further manipulation. The program will output to Epson- or HP-compatible printers, and can highlight conserved residues in a configurable manner, which should allow owners of colour printers hours of fun and frustration with their printer manuals.

© 1993, Elsevier Science Publishers, (UI O