Angew. Chem. Int. Ed. Engl. 30: 658-663 (1991)
The Chemistry of Rose Pigments
Conrad Hans Eugster and Edith Märki-Fischer

Fig. 1. Carotenoids isolated from roses and their bands in the visible (λmax in nm). Formula numbers with letters indicate all-Z compounds; those without letters are E isomers The position of the E double bond is given by a wavy line from the preceding single bond that is, a wavy line from C(14) to C(15) indicates the presence of (15Z,15'Z) double bond. Note that not only mono-Z-carotenoids can occur, but that di-Z compounds, etc., are also found.

5. The Structure of Rose Carotenoids

Most of the carolenoids listed in Figure 1 also occur in other yellow, carotenoid-containing flowers, but in no other case has such a detailed analysis been made. The large number of epoxides, the many Z/E isomers, and the multitude of apocarotenoids are novel and should be stressed. The absolute configurations of chiral carotenoids given in the formulas have in some cases been determined parallel to the studies on roses; for the structure of the C(5), C(6) diols in latoxanthin (28), latochrome (29), karpoxanthin (31),[13d, 21] neoflor,[22] and other carotenoids with triol end groups.[23] This also applies to the 5,6-epoxy[14, 24] and 5,8-epoxy end groups.[14, 24c, 25]

The most fundamental of these studies was the determination of the absolute configuration of C(6) in the ε end group,[26] of OH-C(3) in the zeaxanthin end group.[24a] and of OH-C(3') in lutein (14)[27a] and 3'-epilutein (15)[27b] as well as the allene configuration in neoxanthin (25) and its relatives.[28] These are all distinguished by characteristic spectral and chiroptical data and can now be easily recognized in isolates. Positional isomerism in Z/E-isomeric carotenoids can readily be determined by NMR methods.[17] Often, this can also be inferred from the so-called cis peak [29]

6. The Synthesis of Standard Carotenoids for Comparison Purposes

As mentioned in Section 2.2. minor rose carotenoids have been synthesized for comparison in order to confirm their structures. The syntheses leading to optically active end groups will be described briefly here: more detailed information on the construction of the polyene chains can be found in Ref. [30].

Optically active carotenoids with an ε end group, such as β,ε-carotene (9), were prepared from the enantiomers of α-ionone (41 and ent-41, Scheme 1). These were always obtained by racemate resolution of the diastereomeric menthyl-hydrazones according to Ref. [31] modified as in Ref. [32a]. Straightforward transformations of 41 led to 42, which was hydrolyzed to give the corresponding glycol.[32b] For the synthesis of oxygenated carotenoids, such as zeaxanthin (13), (4R,6R)-4-hydroxy-2,2,6-trimethylcyclohexanone (43) has become important[33] and is now produced industrially. For example, 43 can be converted into 3-hydroxy-β-ionone (44),[34] which is otherwise difficult to synthesize. The synthesis of the triol end group 45 is described in Ref. [23]. The violaxanthin end group 47 was obtained from 46, which, in turn, was prepared by enantioselective epoxidation of a β-cyclogeraniol derivative,[35] using the Katsuki-Sharpless procedure.[36] Compound 46 also proved to be a suitable starting material for the synthesis of allenic end groups such as 48[37] and of the auroxanthin end group 49[38] The way was thus open for synthesis of neoxanthin (25), one of the most abundant carotenoids found in green plants. These results show that the study of rose pigments has led to results whose importance extends far beyond the boundaries of this field.

7. Biogenesis of Carotenoids in Rose Flowers

If we attempt to summarize the results presented in Sections 2, 3, and 5, one solution is to classify them in a biogenetic sequence. This biogenesis has been worked out for plants other than roses,[39] but ought also to be valid for roses, at least in outline. The most important steps, besides the catabolic reactions discovered in roses, are shown in Scheme 2, though Z/E isomerism has been largely ignored. In recent years, however, it has become clear, thanks to improved methods of separation, that these isomers are not artifacts of workup. Phytoene (1) is genuinely a 15Z compound. In roses, we have found a mixture of (15Z)-1 and (15E)-1. With the exception of those at C(1) and C(1'), all the other trisubstituted double bonds have E configuration. Nonetheless, most of the Z/E-isomeric carotenoids that have been investigated in this respect show isomerism at one or more of the trisubstituted double bonds, possibly due to a new isomerization for each step of carotenoid biosynthesis. However, the isomerization described by Pattenden et al.[40] is also possible. This occurs during the dehydrogenation step, so that the double bond moving into conjugation can assume Z configuration. However, this theory leads to a cascade of different biosynthetic pathways, each with Z/E-isomeric intermediates.

  1. Acetyl-CoA → C401 (phytoene)
  2. Dehydrogenations: 1 2 3 45
    2.1. Degradation reactions
  3. Cyclizations
    3.1 (aIl-E)-5  → 67
    3.2. (5Z)-5(?) → 8ε,ε-carotene
    3.3. Degradation reactions
  4. Hydroxylations: for example 6 10; 912 14
    4.1 Degradation reactions
  5. Epoxidations
    5.1. Degradation reactions
  6. Epoxide transformations
    6.1. 5,8-Epoxides
    6.2. Hydrolysis
    6.3. Allene formation
    6.4. Degradation reactions

Scheme 2 The most important steps in carotenoid biosynthesis and transformations.

7.1. Biogenetic Step 2

Carotenoids 1-5 are found in most rose flowers in variable, but also characteristic amounts. In white roses, the dominant compounds are phytofluene and occasionally z-carotene, as mentioned in Section 3. Hence, the biosynthesis is strongly inhibited at the level of the desaturating enzymes. The products of carotene catabolism found in white roses, rosafluene (40) and geranylacetone (60, see Table 2) demonstrate that even z-carotene is degraded.

The colored hydrocarbons neurosporene (4) and lycopene (5) only occur in minute amounts in roses. This not only indicates a gap in the possible palette of colors in these flowers, but also shows that 4 and 5 are very rapidly consumed in the following step. Therefore, cyclases are active in nonwhite roses. They are clearly much more active than in the rosehips of many species, where lycopene (5) can be the major pigment (see Section 11).

7.2. Biogenetic Step 3

Lycopene (5) is a proven substrate for the cyclases, though neurosporene (4) may also be used.[39] Here, the pathways branch, leading to carotenoids with a β end group and those with an ε end group, clearly recognizable in 6 and 8, respectively. Our interpretation of this phenomenon is that it is due to the differing configurations at C(5): (5E)-lycopene leads to 6. whereas the 5Z isomer gives rise to 8.[41]  Various incorporation experiments in other organisms have shown[42] that the terminal methyl groups in 5 retain their identity during the proton-catalyzed ring closure, which therefore proceeds stereoselectively. The same is true for 8, though with a different folding of (5Z)-5. This ring closure generates a chiral center at C(6), which possesses R chirality in all higher plants studied to date.[43]

7.3. Biogenetic Step 4

As in all higher plants, hydroxylation occurs in roses only on cyclic end groups, at C(3) or C(3'). To date, exclusively 3R chirality has been found at the β end groups, while at ε rings the reverse configuration appears (owing to the sequence rules this is also 3R). So far, the only exception is 3'-epilutein (15)[131] which, however, always occurs together with lutein (14) and is produced by a redox process with an oxo compound as intermediate.[27b] There are as yet no indications that hydroxylations of carotenoids can take place at a later biogenetic stage. Rubixanthin (10) is found primarily in rosehips.

7.4. Biogenetic Step 5

The 5,6-epoxides antheraxanthin (16) and violaxanthin (19) are found in large amounts in roses: the other 5.6-epoxides 25, 28, and 37 are products of subsequent reactions. Epoxidation thus occurs preferentially at hydroxylated β end groups. We have not yet been able to identify epoxides of hydrocarbons such as β,β-carotene.[44] The epoxide oxygen atom is derived from O2 and is presumably introduced via the so-called violaxanthin cycle [Eq.(a)].[45]

Since some yellow roses have total epoxide levels of  > 80%, the violaxanthin cycle appears to be disrupted. The 5,6-epoxides always have 5R,6S chirality; introduction of the oxygen atom therefore takes place from the opposite face to the hydroxyl group.

Although hydroxylation at C(3) does not alter the light absorption of the polyene system, the epoxidation shifts it significantly to shorter wavelengths, that is from the orange region of the visible into the yellow. At the same time, epoxidation makes the molecule labile, especially towards acids.

7.5. Biogenetic Step 6

The acid-catalyzed transition from the 5,6- to the 5,8-epoxides is an important epoxide rearrangement (Scheme 3). It can take place even in the intact plant, for example, during chromoplast senescence, but occurs very readily on insufficiently careful workup (cf. Section 2.2). The configuration at C(5) is retained during the rearrangement, except for compounds like 47, where hydrolysis occurs with inversion at C(6).[32b, 46] A new chiral center is formed at C(8); hence, all acid-catalyzed, nonenzymatic epoxide rearrangements afford a mixture of diastereomers. C(8) is especially easily oxidized in the furanoid epoxides 51 and 52. Peroxidation and rearrangement leads to lactones of type 53, which contribute, for example, to the aroma of tea.

The rearrangement of the 5.6- to the 5,8-epoxides also leads to a shortening of the chromophoric system. The first step. from 19 to 20/21, shifts the absorption maximum 20 nm to shorter wavelengths, while the second step, leading to the auroxanthins 22-24, yields a further 25-nm shift. This affords a pure, somewhat greenish yellow of high saturation. For pure yellow colors, without any trace of red, the epoxide rearrangement is of great importance.

Recently, the hydrolysis of carotenoid-5,6-epoxides in flowers and fruits has also been discovered. The 5,6-diols produced are highly polar compounds. This is one reason for their late discovery; latoxanthin (28) and latochrome (29) in 1983,[13b,13d] karpoxanthin (31) in 1985[13b,21,47a] (for others see Refs. [47b, c]. The 5,6-diols are presumably substrates of further catabolic reactions.

The transformation of violaxanthin (19) into allenes 25-27 is a very strange one and as yet mechanistically unexplained. Any theories proposed must take into account the uniform S-axial chirality observed in the allene bond to date.

7.6. Carotenoid Catabolism in Roses

Our findings show that, besides citrus fruits,[48] rose flowers are also a relatively rich source of degraded carotenoids. In total, we have isolated fifteen apocarotenoids, including stereoisomers, and characterized their structures. These compounds include β-citraurin (32), usually a trace component in the flowers of various roses. We were able to isolate quite large amounts from the Chinese shrub rose R. hugonis and from the bush roses 'Piccadilly' (S. McGredy, 1960) and 'Alexander' (Harkness, 1972).[12] The 10'-apolycopene derivatives 33 and 34 were first found by us in the famous old climber 'Maréchal Niel', whose large, gloriously scented flowers are primrose yellow owing to the presence of 33, as long as the plants flower in a greenhouse. In the open air (i.e., in direct sunlight), the blooms are yellow with a tinge of green, owing to the replacement of 33 by 34.[13f] Both compounds are especially prevalent in old noisette hybrids and in tea roses such as 'Safrano' (Beauregard, 1839), but even modern roses such as 'Alexander' and 'Elina' (= 'Peaudouce', P. Dickson, 1985) contain significant quantities.

10'-Apo-β-caroten-10'-ol (35) and 10'-apozeaxanthin-10-ol (36) have been found in several old and modern garden roses.[12,13c,13g] Compound 35 is novel, as are 33 and 34, but 36 is already known under the name galloxanthin and has previously been isolated from the retina of birds. Apparently, birds and roses possess similar carotenoid-cleaving enzymes.

The epoxides sinensiaxanthin (37) and the sinensiachromes 38 and 39 are widespread in fruits, where they give rise to the yellow color of the flesh in apple varieties such as "Golden Delicious", for example. The structures, synthesis, and purification of the multiple stereoisomers present have, however, only been studied in connection with roses.[13g, 38] Sinensiaxanthins and sinensiachromes are found in many white, yellow, or orange roses, usually together with the stereoisomeric violaxanthins and auroxanthins, from whose cleavage they arise.

The colorless, highly fluorescent C14 compound rosafluene (40), mentioned briefly in Section 3, is most remarkable. Rosafluene occurs as an ester and possesses extremely similar characteristics to phytofluene (2). It is very unstable, especially as the free alcohol. We have found it in numerous old and modern roses, with relatively high amounts in 'Piccadilly', 'Alexander', and 'Penelope' (Pemberton, 1924).

With the exception of 34 and 40, the apocarotenoids found are all C27 alcohols. They are definitely carotenoid metabolites and arise by cleavage at C(9) and/or C(9'). The apocarotenoids from roses are not aldehydes but the corresponding alcohols, just as is the case with Car(15,15') dioxygenase. This enzyme cleaves β,β-carotene (7) to retinal, a C20 compound which is stored after reduction as the alcohol (retinol, Vitamin A1). Presumably aldehydes also occur, but because of the very small amounts present, they have not yet been identified.[12]

An analysis of the products shows that the postulated Car(9,10:9',10') dioxygenase is not specific for end groups. It cleaves excentrically according to Equation(b). The widespread occurrence of rosafluene (40) indicates that the enzyme can also cleave the C27 compounds a second time.

C40 → C13 + C27 → C13 + C14              (b)

Recently, we have begun to identify the postulated C13 compounds in Equation (b).[50] Our results (Table 2) convincingly show the other side of the carotenoid catabolism under discussion. The detection of acyclic C13 ketones and alcohols is also important. If the latter were stored in the form of glycosides, they would not be detectable by gas chromatography.[51]

In summary, the degradation starts with ζ-carotene (3) and includes all the end groups in Figure 1 found in roses to date[52] (Table 3).

The results presented in Tables 2 and 3 complete the picture of our pigment analyses of carotenoid-containing roses. Originally, carotenoids may have served in the plant's showpiece as a means of communication with potential pollinators. Then followed the degradation of carotenoids to aroma compounds and to other substances with many different physiological qualities, whose significance is still unclear to us. If one also remembers the fundamental role of carotenoids in photosynthesis, then it is clear that all in all they are the most multifaceted of all natural product groups.

7.7. Carotenoids and the Genealogy of Garden Roses

Old European garden roses, that is, those before 1830, had flowers that were white or red (in many shades). The only exceptions were a few botanical roses which had been introduced to Europe centuries earlier: R. hemisphaerica with its full, sulfur-yellow flowers, R. foetida   (also called R. lutea by later authors, the so-called Austrian briar), with a simple, bright yellow flower, and R. foetida bicolor (Austrian copper), also with a simple flower, but in two colors, yellow outside and red within. They all originated in central Asia and possibly represent cultivated forms of the original species. R. hemisphaerica is often to be seen in old floral still-life paintings. Remarkably, no hybrids between these roses and the old European garden roses are known, although they have been grown in gardens for centuries. The same is true of the yellowish-flowering varieties of R. pimpinellifolia. Only after 1800, when several cultivated forms of Chinese roses, including the yellow 'Park's Yellow Tea-scented China'[53] were introduced, did crossings lead to a gradual expansion of the color scale (Fig. 4). 'Park's Yellow' was of crucial importance for the yellow shades. Unfortunately, it has died out[54] and is not available for analysis. On looking through old rose lists one notices that yellow flowers were commoner in noisette roses than in tea roses. These were usually pink on a yellow base or vice versa. Noisette roses arose from crosses with R. moschata, which had an as yet unknown effect on the flower color. Famous yellow roses were 'Elise Sauvage' (Millez, 1818, tea rose, pale yellow with an orange center[55]), 'Jaune Desprez' (Desprez, 1830, noisette, yellow and pink), 'Solfaterre' (Boyau, 1843, noisette, sulfur yellow), and the varieties listed in Table 4.

One can see that breeders have always aimed for a purer and more saturated yellow, a development which culminated in 'Maréchal Niel'. Nonetheless, the yellow color in most of these roses was unstable and faded a great deal during flowering. One exception was R. x harisonii (x indicates a hybrid), which only flowers once. The red roses developed much more rapidly than the yellow varieties and gave rise to the new rose classes of portland, bourbon, and hybrid perpertuals. Among the roughly 4000 hybrid perpetual varieties, there was not a single yellow rose! Even the tea hybrids, which developed slowly from ca. 1850—one of the most important classes today—displayed only very few good yellow varieties before 1900. The emergence of varieties with an intense, highly saturated yellow did not begin till 1900, with the breeding of 'Soleil d'Or' (Jean Pernet-Ducher, Lyon), a hybrid of R. foetida persiana (Fig. 5) and a red hybrid perpetual. Predecessors such as 'Gottfried Keller' (Hermann Müller, 1894) had practically no effect on the development of modern yellow varieties.

Fig. 4. Old yellow garden roses with carotenoid biogenesis steps 1-3 and pronounced degradation reactions. Left: 'Park's Yellow' (1:3.5, died out: photo taken from P. J. Redouté: Les Roses, 3rd ed. 1828; a specimen from the Jardin botanique de la Ville de Genéve, parent of the old yellow garden roses). Middle: 'Maréchal Niel' (1:2.5; climber, Pradel, 1864; most famous of the old yellow garden roses). Right: 'William Allen Richardson' (1:2.2; rambler, Ducher, l878); carotenoids concentrated in the center; become almost white owing to degradation.
[The Redouté illustration is actually 'Knight's Yellow China', a seedling of 'Hume's Blush Tea-scented China' that was introduced in 1823.]

We have analyzed forty old and modern varieties of yellow garden rose and several important species.[12] In Table 4, some results of this study are presented. The data allow the following conclusions to be drawn:

Fig. 5. Representatives of foetida wild roses that were most important for the breeding of modern, yellow roses. Left: R. foetida bicolor (1: 2.2), known in Europe since the Middle Ages. Middle: R. foetida (1:3), originated in middle Asia, first described by Conrad Gessner in 1561 and Matthias Lobelius in 1581. Right: R. foetida persiana (1: 3). introduced from middle Asia in 1837. Characteristic of all is the high content of carotenoids, especially those arising at biogenetic step 6, and a weak catabolism. Thus, these carotenoids are very stable toward light. The main anthocyanin in R. foetida bicolor is peonin.

1. Old, yellow, garden roses that have the Chinese cultivars and above all 'Park's Yellow' and R. moschata in their pedigree display an incomplete carotenoid biosynthetic pathway; hydrocarbons at step 2 are accumulated, though usually without neurosporene (4) and lycopene (5). As a rule, there is a pronounced inhibition of the desaturating enzymes, and catabolic reactions are prominent.

2. One exception is R. odorata x pseudindica, which has the ability to carry out cyclization (step 3). Unfortunately, it was never used for breeding.

3. Garden roses with pure, stable and strongly saturated yellows emerged only by crossing in of roses of the foetida group (Fig. 6). especially R. foetida persiana. They contain a complete carotenoid biogenetic sequence; step 5 is pronounced. Degradation reactions are minimal.

4. Carotenoid analysis shows that R. x harisonii is descended from R. pimpinellifolia and R. foetida.

5. Modern, light yellow rose varieties such as 'Mme Meilland' (Meilland, 1945) or 'Elina' combine a complete carotenoid biogenetic pathway with enhanced Car(9,10:9',10') dioxygenase activity.

Previous section

Eugster & Märki-Fischer: Anthocyanins (1991)