A novel pyrroline-5-carboxylic acid and acetoacetic acid adduct in hyperprolinaemia type II

A novel pyrroline-5-carboxylic acid and acetoacetic acid adduct in

hyperprolinaemia type II

Valerie Walkera,*, Graham A. Millsb, John M. Mellorc,G. John Langleyc, R. Duncan Farrantd

aDepartment of Chemical Pathology, Southampton General Hospital, Level D, Mail Point 6, Tremona Road, Southampton, SO16 6YD, UKbSchool of Pharmacy and Biomedical Sciences, University of Portsmouth, White Swan Road, Portsmouth, PO1 2DT, UK

cDepartment of Chemistry, University of Southampton, Highfield, Southampton, SO17 1BJ, UKdPhysical Sciences, GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire, SG1 2NY, UK

Received 4 September 2002; received in revised form 15 January 2003; accepted 17 January 2003

Abstract

Background: From investigations of a child with hyperprolinaemia type II, we demonstrated in vitro that pyridoxal

phosphate forms a novel adduct with a proline metabolite, pyrroline-5-carboxylic acid, through Claisen condensation. Studies

indicated that this was a previously unsuspected generic reaction of aldehydes and some ketones. We have subsequently found

the acetoacetic acid adduct in both plasma and urine from the affected child. Methods: Mixtures of acetoacetic acid and

pyrroline-5-carboxylic acid were co-incubated at pH 7.4 and 37 jC, dried, or extracted and dried, derivatised and analysed by

gas chromatography/mass spectrometry (GC/MS). Urine and plasma from the child were analysed. Results: Fourteen new peaks

were found in derivatised pyrroline-5-carboxylic acid/acetoacetic acid co-incubates. From accurate molecular mass data, the

four largest peaks were probably diastereoisomers of tri-trimethylsilyl (tri-TMS) derivatives of alcohol adducts formed by

Claisen condensation. Eight other peaks were mono- and di-trimethylsilyl derivatives of the adduct and a decarboxylated

product. The adduct was demonstrated unequivocally in the child’s acute urine and traces in plasma. Conclusions: Pyrroline-5-

carboxylic acid forms an adduct with acetoacetic acid, which was present in urine of a sick child with hyperprolinaemia type II.

Evidence suggests it formed in vivo. The biological significance of this novel reaction of aldehydes and ketones merits

investigation.

D 2003 Elsevier Science B.V. All rights reserved.

Keywords: Pyrroline-5-carboxylic acid; Hyperprolinaemia type II; Acetoacetic acid; Claisen condensation; Proline metabolism

1. Introduction

Hyperprolinaemia type II (OMIM 239510) is a rare

inherited autosomal recessive disorder caused by a

deficiency of D1-pyrroline-5-carboxylate (P5C) dehy-

drogenase (EC 1.5.1.12) (Fig. 1). This leads to a 10-

to 15-fold increase in plasma proline, accumulation of

P5C and increased urinary proline excretion [1].

Clinical presentation is with convulsions in childhood,

usually precipitated by infection. Between episodes,

children are generally well. Most adults enjoy normal

health [2].

0009-8981/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.

doi:10.1016/S0009-8981(03)00077-9

* Corresponding author. Tel.: +44-23-8079-6419; fax: +44-23-

8070-6339.

E-mail address: [email protected] (V. Walker).

www.elsevier.com/locate/clinchim

Clinica Chimica Acta 331 (2003) 7–17

We diagnosed this disorder in a previously healthy

20-month-old girl who developed pneumonia, fed

poorly for 4 or 5 days and presented with convulsions

and a depressed conscious level and ketosis. She

recovered slowly over 5 days and was then healthy,

although the diagnostic biochemical abnormalities

persisted. She had a second acute illness 2 years later,

also precipitated by a febrile illness, from which she

recovered. The unusual feature of this case was

evidence of vitamin B6 deficiency at the time of her

first admission. This was confirmed by investigations

8 months later [3]. The deficiency was not explained

by a bizarre diet or medication.

One cause of a functional deficiency of vitamin B6

is deactivation of its co-enzyme form, pyridoxal

phosphate [4]. We proposed that P5C might do this.

We subsequently demonstrated in vitro that P5C and

pyridoxal phosphate combine through a Claisen con-

densation (or Knoevenengal type of reaction) of the

activated C-4 carbon of the P5C pyrroline ring with

the aldehyde carbon of pyridoxal phosphate [5]. In

preliminary studies, we also found that P5C interacts

in vitro with a range of other aldehydes and ketones,

indicating that this is a generic, and previously unre-

ported reaction of these compounds.

We re-examined extracts of urine and plasma from

the child to see whether any possible P5C adducts were

present. Perhaps because of analytical insensitivity, we

have not identified P5C/pyridoxal phosphate adducts

by gas chromatography/mass spectrometry (GC/MS)

or nuclear magnetic resonance spectroscopy (NMR).

However, with GC/MS, we found probable peaks of a

P5C/acetoacetic acid adduct in urine samples collected

after both acute admissions. We have now investigated

the interaction of P5C and acetoacetic acid in vitro.

The adduct was demonstrated unequivocally in the

child’s urine. There were also traces of this in two of

her plasma samples and in plasma from another

unrelated child with hyperprolinaemia type II. There

was some evidence that the adduct may have formed in

vivo prior to renal excretion. The biological impor-

tance of this novel chemical reaction remains to be

explored.

2. Materials and methods

2.1. Urine and plasma

Urine and plasma were collected for diagnosis and

management of a child with hyperprolinaemia type II.

Details of the clinical presentation and biochemistry

have been reported [3]. The plasma proline concen-

tration at presentation was 2690 Amol/l (reference

range: 90–280 Amol/l) and urine proline concentra-

tion, 3700 Amol/mmol creatinine (reference range:

< 13 Amol/mmol). Over the next 2 years, plasma

proline remained extremely elevated (2290–2955

Fig. 1. Metabolic pathway showing the catabolism of L-proline. The site of the enzyme deficiency in hyperprolinaemia type II is indicated by *

(Fig. 1 based on Ref. [1]).

V. Walker et al. / Clinica Chimica Acta 331 (2003) 7–178

Amol/l, measured on six occasions). The urine excre-

tion of proline increased after recovery, reaching a

peak concentration of 47,500 Amol/mmol creatinine

after 3 months. Urine collected on the day after

admission had 4 mmol/l of ketones (Multistix SG;

Bayer, Newbury, UK). Ketonuria was not demonstra-

ble (Multistix test) subsequently. Genetic studies

(Professor David Valle, Johns Hopkins University

School of Medicine, Baltimore, USA) showed that

she does not have the genetic mutation identified in an

Irish pedigree [6].

There was evidence of vitamin B6 deficiency at

presentation, which was confirmed in samples col-

lected 3 and 8 months later, when she was well [3].

She was prescribed vitamin B6 supplementation but

was poorly compliant. She had a second acute illness,

similar to the first, aged 4 years 2 months. This

episode was curtailed by intravenous pyridoxine [3].

Urine collected 1 day after admission was negative for

ketones (Multistix test). However, 3-hydroxybutyric

acid was marginally increased in the organic acid

profile of solvent extracted urine.

Urine collected for diagnosis 24 h after both acute

admissions and four other samples collected over 27

months for monitoring when she was well were

analysed for possible P5C adducts. Plasma from the

second acute admission and on one occasion when

well was also analysed. Random urine samples were

collected into 25 ml sterile polyvinyl containers and

blood (5 ml) into lithium heparin tubes. Plasma and

urine were stored at � 20 jC.One random urine sample and a paired plasma

sample were also collected for metabolic assessment

of an unrelated child with hyperprolinaemia type II.

This 6-year-old boy was from a large Irish kindred

with the disorder [2], in which the gene mutation has

been identified [6]. His plasma proline was 2155

Amol/l (reference range: 90–280 Amol/l) and urine

proline 3375 Amol/mmol creatinine (reference range:

< 9 Amol/mmol creatinine). He was not ketonuric or

vitamin B6 deficient.

One random urine sample, collected for metabolic

investigations, was also analysed from a 3-year-old

boy with very severe fasting ketosis (urine ketones

>16 mmol/l; Multistix test). Routine tests showed no

evidence of an inherited amino acid or organic acid

disorder, and plasma and urine proline concentrations

were normal.

2.2. Materials and instrumentation

DL-D1-Pyrroline-5-carboxylic acid (as 2,4-dini-

trophenylhydrazine hydrochloride double salt),

N,O-bis(trimethyl)trifluoroacetamide (BSTFA), N-

tert-butyldimethylsilyl-N-methyltrifluoroacetamide

(MTBSTFA) derivatising agents and urease (type

III) were from Sigma, and acetoacetic acid (as

lithium salt) and acetophenone (99% pure) were

from Aldrich (Sigma-Aldrich, Poole, UK). Deuter-

ated (d9)-BSTFA was from MSD Isotopes, NJ, USA.

Diethyl ether (Analar) was from BDH, Poole, UK.

AG 50W-X8 200–400 mesh (hydrogen ion form)

cation-exchange resin was from Bio-Rad, Watford,

UK. Water was deionised by reverse osmosis.

A bench-top GC/MS (5890 series 2 GC linked to a

5971A quadrupole MS) from Agilent (Bracknell,

Berkshire, UK) with electron impact (EI) ionisation

(70 eV) was used routinely for profiling the sample

extracts. This was fitted with a non-polar BPX-5

fused-silica capillary column (30 m� 0.22 mm I.D.,

film thickness 0.25 Am) (Scientific Glass Engineering,

Milton Keynes, UK). Helium was the carrier gas at a

flow rate of 1 ml/min. Data was analysed with

Chemstation software (Agilent).

For molecular mass determinations, chemical ion-

isation (CI) GC/MS (ThermoFinnigan Trace GC/MS;

Hemel Hempstead, Herts, UK) with ammonia

(99.99%) reagent gas was used. The GC/MS was

fitted with an Optima Delta 3 stationary phase fused-

silica capillary column (30 m� 0.4 mm O.D., film

thickness 0.25 Am) (Macherey-Nagel, Middleton-

Cheney, Oxon, UK). A VG Analytical 70-250-SE

double focusing MS with a MASPEC II32 data

system (VG, Manchester, UK) was used for the

high-resolution (HR) direct probe accurate mass

measurements.

2.3. Preparation of P5C and acetoacetic acid co-

incubates

P5C was prepared from its 2,4-dinitrophenylhy-

drazine hydrochloride double salt [7], but using a

more concentrated starting solution of the hydrazine

and a diethyl ether wash [5]. P5C prepared according

to the original method [7] was polymerised and

contaminated with toluene and acetophenone. The

pH of the extracted P5C was adjusted to 7.4 with

V. Walker et al. / Clinica Chimica Acta 331 (2003) 7–17 9

NaOH and the preparation stored in the dark at 4 jC.P5C was stable under these conditions for at least 4

weeks. Acetoacetic acid, 5 mg/ml in water, was

prepared immediately before use and the pH adjusted

to 7.4 with HCl. A total of 250 Al of acetoacetic acid

solution (approximately 9 Amol) and 200 Al of P5C(approximately 6 Amol) were placed into 1.8 ml glass

vials (Radleys, Saffron Walden, Essex, UK). The

precise amount of P5C used could not be determined

because of losses of the aqueous extract during the

wash stages of its preparation. The pH of the mixture

was readjusted to 7.4 if necessary and vials were

incubated at 37 jC overnight (16 h). The mixtures

were then freeze-dried and derivatised with either 50

Al of BSTFA, d9-BSTFA or MTBSTFA at 70 jC for

30 min. Some co-incubates were acidified (pH 1–2)

and extracted with a cation-exchange resin before

drying (see below).

In two experiments, two batches of P5C were

diluted serially to concentrations ranging from

approximately 6.50–0.40 mmol/l and acetoacetic acid

to concentrations over the range 10.00–0.60 mmol/l.

Mixtures of P5C (200 Al) and acetoacetic acid (250

Al) were co-incubated overnight at pH 7.4 and 37 jC,then freeze-dried and derivatised with BSTFA (50 Al).

2.4. Extraction of urine and plasma

We used a cation-exchange method to extract P5C

adducts from plasma and urine, similar to procedures

used to extract amino acids from biological fluids [8].

Urine samples were pre-incubated with urease to

avoid a large urea peak in the GC/MS profiles. Urine

containing 1 Amol creatinine (water added to make up

to 1 ml if necessary) was incubated with approxi-

mately 1 mg of urease at 37 jC for at least 2 h. After

acidification (pH 1–2), the samples were applied to a

1� 0.5 cm column of AG 50W-X8 cation-exchange

resin, washed four times with water (1 ml), eluted

with 2� 0.8 ml of 5 mol/l ammonium hydroxide into

1.8 ml glass vials and freeze-dried. The residue was

derivatised with 100 Al of BSTFA or MTBSTFA (one

analysis) at 70 jC for 30 min. According to avail-

ability, 500 Al to 1 ml of plasma was applied to the

columns without acidification (three samples) and one

sample, after acidification (pH 2.0). The extracts were

eluted as for urine, dried and derivatised with 50 Al ofBSTFA.

2.5. Gas chromatography/mass spectrometry

BSTFA derivatives were analysed routinely by

GC/MS with EI ionisation (scan range: 40–600 Da).

Usually, 2 Al of extract was injected with a 1:15 split

ratio. Splitless injection was used for plasma extracts.

The following conditions were used: solvent delay,

13.5 min; injector, 250 jC; interface transfer line, 280jC; oven temperature programme, 50 jC (5 min) then

5 jC/min to 270 jC (15 min). For MTBSTFA

derivatives, the solvent delay was 12.5 min and the

oven temperature programme, 100 jC (5 min) then 5

jC /min to 270 jC (15 min).

Dried co-incubates of P5C and acetoacetic acid (as

BSTFA derivatives) were analysed by CI-GC/MS.

The oven temperature programme was 50 jC (5

min) then 5 jC/min to 270 jC, then 20 jC/min to

320 jC (5 min). The reagent gas source pressure was

adjusted to produce both protonated quasi-molecular

ions (M+H)+. and ammoniated (M+NH4)+ adducts

from which the molecular mass (M) of the derivatives

could be determined.

2.6. High-resolution probe mass spectrometry

In order to confirm the accurate molecular mass

and molecular formula assignments, a dried co-incu-

bate of P5C and acetoacetic acid (as BSTFA deriva-

tive) was analysed by HR/MS. The measurements

were carried out using high voltage scans at 10,000

resolution, internally calibrated with perfluorokero-

sine (EI measurements) and polyethyleneglycol

(PEG-200 and PEG-400 mixture) for CI measure-

ments.

3. Results

3.1. Interaction of P5C and acetoacetic acid in vitro

Thirteen new chromatographic peaks were found

by GC/MS when D/L P5C was incubated with aceto-

acetic acid, dried and derivatised with BSTFA (Fig.

2a). Seven of these plus an additional new peak

(shown as B3) were found when the co-incubates

were extracted with a cation-exchange resin before

drying (the procedure used for urine and plasma) (Fig.

2b). With serial dilution, the A, B and D peaks were


Fig. 2. (a) Total ion current chromatogram obtained for co-incubates of pyrroline-5-carboxylic acid and acetoacetic acid at pH 7.4 and 37 jC (16

h), dried and derivatised with 50 Al of N,O-bis(trimethyl)trifluoroacetamide at 70 jC for 30 min to form TMS derivatives. Key to proposed peak

identification: A1 and A2 mono-TMS derivatives of the decarboxylated adduct (structures 8 and/or 9 in Fig. 3); B1, B2, B4 di-TMS derivatives

of the decarboxylated adduct (structures 6 and/or 7 in Fig. 3); C1 and C2 unknown compounds; D1, D2 di-TMS derivatives of the complete

adduct (structures 3 and/or 4 and/or 5 in Fig. 3); E1, E2, E3, E4 tri-TMS derivatives of the complete adduct (structure 2 in Fig. 3). (b) Total ion

current chromatogram obtained for co-incubates of pyrroline-5-carboxylic acid and acetoacetic acid as described in (a), then acidified (pH 1–2)

and extracted with a cation-exchange resin, dried and derivatised with 50 Al of N,O-bis(trimethyl)trifluoroacetamide at 70 jC for 30 min to form

TMS derivatives. Key to peak identification as for (a) and B3 di-TMS derivative of the decarboxylated adduct (structure 6 and/or 7 in Fig. 3).


still identifiable in dried preparations at the lowest

concentrations tested (final incubate concentrations of

approximately 0.18 mmol/l of P5C and 0.35 mmol/l

of acetoacetic acid).

Molecular masses of most of the peaks were

confirmed by CI-GC/MS. These were: 243 Da for

peaks A1 and A2; 315 Da for peaks B1, B2 and B4;

359 Da for peaks D1 and D2; and 431 Da for peaks

E1, E2, E3 and E4. The molecular mass of the C

peaks could not be assigned by observation of the

parent ions. Compounds with the same molecular

mass were likely to be isomers. The following pairs

of peaks had similar EI fragmentation patterns: A1

and A2; B1 and B3; B2 and B4; C1 and C2; D1 and

D2; E1 and E2; E3 and E4, respectively.

Using d9-BSTFA increases the molecular mass of

trimethylsilyl (TMS) derivatives by 9 Da for each

TMS group added. With d9-BSTFA, the A peaks were

found to be mono-TMS derivatives, the B and D

peaks, di-TMS derivatives, and the E peaks, tri-TMS

derivatives. With MTBSTFA, no E peaks were found

and only one small B peak, suggesting that derivati-

sation of one available site by the relatively large tert-

butyldimethylsilyl (t-BDMS) group (molecular mass:

115–42 Da greater than a TMS group) was prevented

by steric hindrance. The molecular mass of the A

Fig. 3. Proposed reaction pathway of pyrroline-5-carboxylic acid and acetoacetic acid by Claisen condensation to form adduct (1) with

stereochemical centres (a–c) and proposed trimethylsilyl derivatives (2–9) formed by reaction with N,O-bis(trimethyl)trifluoroacetamide

derivatising reagent. Me =methyl (CH3) group.


peaks increased by 42 Da, confirming that one site on

these compounds was derivatised, and of the D peaks

by 84 Da, confirming derivatisation at two sites.

Using BSTFA, the largest compounds identified (E

peaks), with a molecular mass of 431 Da, were likely to

be tri-TMS derivatives of a conjugate of P5C and

acetoacetic acid. Three hydrogen atoms would be

replaced by TMS groups. The molecular formula

would be C18H37NO5Si3. If such a conjugate were

formed by a Claisen condensation, the reaction would

occur as in Fig. 3 and the adducts would be the alcohols

(1) leading to tri-derivatised products (2). Each would

have three stereocentres (labelled as a, b and c in Fig. 3)

and four pairs of diastereoisomers would be possible.

This would yield a maximum of four independent

peaks on an achiral GC stationary phase.

The electron impact GC/MS fragmentation pattern

of peak E1 is shown in Fig. 4. The 314 ion was likely

to be due to the loss of a-COOTMS (C4H9O2Si) group

from the parent molecule. With probe HR/MS, the

accurate mass of ion 431 Da (molecular ion) was

found to be 431.19842 Da and the predicted compo-

sition, C18H37NO5Si3 (1.1 ppm error). For the 314

ion, the accurate mass was 314.16064 Da and the

predicted molecular formula was C14H28NO3Si2 (0.4

ppm error). Thus, the accurate mass data provided

strong evidence that the E1 peak was the tri-TMS

derivative of the P5C/acetoacetic acid adduct (struc-

ture 2 in Fig. 3). Peaks E2, E3 and E4 were probably

diastereoisomers.

By comparing the fragmentation patterns of

BSTFA and d9-BSTFA derivatives, we deduced that

peaks D1 and D2 were di-TMS derivatives of the

intact P5C/acetoacetic acid adduct (structures 3, 4 or 5

in Fig. 3). Peaks A and B were probably decarboxy-

lated products of the adduct. Possible structures for

A1 and A2 are 8 or 9 and for peaks B1 to B4, 6 or 7 in

Fig. 3. In separate experiments, we demonstrated that

the carboxyl group of underivatised P5C is lost

following injection into the GC/MS. On this basis,

structure 6 is tentatively assigned to the B series of

peaks.

40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 4200

500000

1000000

m/z

73

314

181

147

416

45258

100

Fig. 4. Electron impact (70 eV) mass spectrum of chromatographic peak E1 for dried co-incubates of pyrroline-5-carboxylic acid and acetoacetic

acid. Reaction conditions as described for Fig. 2a.


3.2. P5C/acetoacetic acid adducts in clinical samples

3.2.1. Samples from children with hyperprolinaemia

type II

We identified significant amounts of the P5C/

acetoacetic acid adduct in the BSTFA-derivatised

extract of urine collected 1 day after our child’s first

acute admission (Fig. 5). There were prominent peaks

of the intact adduct (D and E peaks) together with the

decarboxylated adduct (A and B peaks). The electron-

impact-mass spectrum of peak E1 (Fig. 6) closely

resembled that reported in Fig. 4. With MTBSTFA,

the intact and decarboxylated adduct was found.

In urine from her second acute admission, there

were small peaks of the decarboxylated adduct but

only a trace of the intact compound. Traces of the

decarboxylated adduct only were found in two of the

four random urine samples collected when she was

well and not ketotic.

The decarboxylated adduct was also found in low

concentrations in both plasma samples collected—one

from her second acute admission and one when she

was well. The intact form was not identified. The

adduct was not identified in urine from the other child

with hyperprolinaemia type II, but there was a trace of

the decarboxylated compound in his plasma.

3.2.2. Urine from a ketotic child

The P5C/acetoacetic acid adduct found in urine

might have been produced in the body and excreted

through the kidneys. Alternatively, it could have

formed in urine accumulating in the bladder. To find

out whether the second possibility was feasible, we

incubated P5C with urine from a child with normal

proline metabolism but severe fasting ketosis (urine

ketones >16 mmol/l). A volume of urine containing 1

Al of creatinine was incubated with 200 Al of P5C or

water (control) for 16 h at 37 jC and pH 7.4. The

samples were then incubated with urease for 2 h,

acidified (pH 2.0) and extracted with cation-exchange

resin. The adduct was not found in the control

sample. With P5C, the intact adduct was identified,

Fig. 5. Total ion current chromatogram of urinary metabolites in the affected child’s urine after the first acute admission. Urine containing 1

Amol of creatinine was incubated with urease at 37 jC for 2 h, acidified (pH 1–2) and extracted on a cation-exchange resin. The extract was

dried and derivatised with 100 Al of N,O-bis(trimethyl)trifluoroacetamide 70 jC for 30 min. Key to proposed peak identifications as for Fig. 2a

and b.


together with the decarboxylated derivatives and peak

C2.

4. Discussion

This study evolved from a novel observation that a

child with hyperprolinaemia type II was also vitamin

B6 deficient. As this could not be explained by diet or

medication, we proposed that P5C deactivates vitamin

B6 by combining with its co-enzyme, pyridoxal

phosphate. There was foundation for this, since an

analytical test for P5C uses its reaction with another

aromatic aldehyde, 2-aminobenzaldehyde [9]. In sol-

ution, P5C and L-glutamic acid–g-semialdehyde (Fig.

1) are in equilibrium. P5C is favoured at physiological

pH. We have defined a mechanism for the reaction of

P5C with pyridoxal phosphate [5]. We also observed

that at pH 7.4, P5C reacts with a range of other

aldehydes: acetaldehyde, benzaldehyde, butyralde-

hyde, formaldehyde, propionaldehyde, pyruvic alde-

hyde, valeraldehyde and glyoxylic acid, and with

some ketones: acetoacetic acid, pyruvic acid, oxalo-

acetic acid and 2-oxobutyric acid. No reaction was

demonstrable for acetone, 2-oxoadipic acid or 2-

oxoglutaric acid. Some of these compounds are bio-

logically important.

We then reviewed the GC/MS data for samples

from the child to look for possible P5C adducts.

Because of her gross amino acid disturbance, the

NMR spectra for her plasma and urine were too

complex for this type of search. We did not find the

P5C/pyridoxal phosphate adduct. However, in urine

collected during her first acute admission, we tenta-

Fig. 6. Electron impact (70 eV) mass spectrum of the chromatographic peak E1 found for the affected child’s urine at the first acute admission.

Urine was extracted and derivatised as described in Fig. 5.


tively identified a conjugate of P5C and acetoacetic

acid.

We have now confirmed that P5C reacts in vitro

with acetoacetic acid at pH 7.4 and 37 jC. This is anew observation. With GC/MS, we found a total of 14

new peaks (BSTFA derivatives) formed from a P5C/

acetoacetic acid adduct. The largest products were

shown to be tri-TMS derivatives. From the accurate

molecular mass measurements, the predicted compo-

sition matched that expected for an adduct formed by

a similar mechanism to the P5C/pyridoxal phosphate

conjugate. The proposed structure would have

resulted from an addition of the C-4 carbon to the

ketone group of acetoacetic acid, reminiscent of the

Knoevenagel reaction, with the imine function of P5C

behaving like the carbonyl in the more usual ketone–

aldehyde reaction (Fig. 3). Because the adduct has

three stereocentres, there are four possible pairs of

diastereoisomers.

All 14 peaks identified in vitro were found in urine

collected from our child 24 h after her first acute

admission. She was ketotic with approximately 4

mmol/l of ketones (mainly acetoacetic acid) in her

urine. She may have been more ketotic when first

admitted. The adduct was also identifiable in urine

collected 24 h after her second acute admission,

although in smaller amounts. On that occasion, the

Multistix test was negative for ketones, but a small

increase in 3-hydroxybutyric acid in the GC/MS pro-

file was evidence of mild, possibly resolving, ketosis.

This adduct has not been reported to be present in

urine before. It is important to know its origin. If it

were formed from P5C and acetoacetic acid in urine

collecting in the bladder or in vitro after sample

collection, its production would have no metabolic

significance. We demonstrated that this is feasible,

since we found the adduct after co-incubating P5C

with urine from a severely ketotic child who had

normal proline metabolism. If, however, the adducts

were produced in vivo prior to excretion by the

kidney, its formation within the metabolic pool might

have biological implications. This would also apply to

other P5C adducts.

P5C is an intracellular metabolite produced in liver,

kidney and brain by the mitochondrial enzyme proline

oxidase (no EC number assigned) [1]. Normally, it is

converted to glutamic acid by D1-pyrroline-5-carbox-

ylic acid dehydrogenase in the mitochondrial matrix

(Fig. 1; Refs. [10–12]). Some, however, diffuses into

the cytoplasm where it is reduced back to proline by

P5C reductase (EC 1.5.12), or passes out of the cells

into the circulation [1,13]. Deficiency of P5C dehy-

drogenase in hyperprolinaemia type II leads to a

considerable increase in cytoplasmic flux of P5C to

proline, explaining the very high plasma proline

concentrations observed [1]. There is also a 10- to

40-fold increase in plasma P5C concentrations above

the normal of 0.2–2.0 Amol/l [1,2,14].

Acetoacetic acid is synthesised in liver mitochon-

dria. P5C and acetoacetic acid would be present

together at high concentration in liver mitochondria

and cytoplasm should an individual with hyperproli-

naemia type II become ketotic. Children with this

disorder are not at increased risk for ketosis. However,

like other young children, they can be expected to

become ketotic readily during intercurrent illnesses if

they are catabolic and not feeding well. The adduct

might form in vivo during such circumstances and,

being small, should be excreted via the kidney. We

know that acetoacetic acid production was increased

(urine ketones 4 mmol/l; with plasma acetoacetic acid

probably around 1 mmol/l [15]) and the very high

plasma proline concentration (2690 Amol/l) indicates

a large P5C flux. In vitro, we demonstrated that the

adduct was produced through co-incubation of P5C at

a concentration of around 180 Amol/l with 0.35 mmol/

l of acetoacetic acid (the lowest tested). These con-

centrations are close to biological values.

There is some evidence that the adduct may have

been produced in vivo prior to excretion and not

merely as an artefact formed in bladder urine. Firstly,

it was found in urine collected during her second

acute admission, in the absence of significant keto-

nuria. Secondly, we were able to detect it in plasma

collected from this child and from another, unrelated

affected child, when they were not ketotic. Thirdly,

the lowest proline concentrations recorded were for

urine collected during both her acute admissions. This

is unlike most inherited amino acid disorders, in

which excretion of the affected amino acid is greatest

during acute decompensation. Concentrations of her

other urine amino acids were not reduced. An explan-

ation is that, instead of being recycled back to proline,

P5C was consumed in an alternative reaction. For-

mation of an adduct with acetoacetic acid is one

possibility.


So far, the P5C/acetoacetic acid conjugate is the

only adduct that we have identified in samples from

the two hyperprolinaemic children. We do not know

whether P5C forms adducts with other biologically

important aldehydes and ketones, including pyridoxal

phosphate, in vivo. Further studies are indicated using

more sensitive analyses.

References

[1] Scriver CR, Beaudet AL, Sly WS, Valle D, editors. The meta-

bolic basis of inherited disease. New York: McGraw-Hill;

2001. p. 1821–38.

[2] Flynn MP, Martin MC, Moore PT, et al. Type II hyperproli-

naemia in a pedigree of Irish travellers (nomads). Arch Dis

Child 1989;64:1699–707.

[3] Walker V, Mills GA, Peters SA, Merton WL. Fits, pyridoxine,

and hyperprolinaemia type II. Arch Dis Child 2000;82:236–7.

[4] Klosterman HJ. Vitamin B6 antagonists and antimetabolites.

In: Dolphin D, Poulson R, Avramovic O, editors. Coenzymes

and cofactors, volume 1. Vitamin B6. Pyridoxal Phosphate.

New York: Wiley; 1986. p. 391–415.

[5] Farrant RD, Walker V, Mills GA, Mellor JM, Langley GJ.

Pyridoxal phosphate de-activation by pyrroline-5-carboxylic

acid. Increased risk of vitamin B6 deficiency and seizures in

hyperprolinaemia type II. J Biol Chem 2001;276:15107–16.

[6] Geraghty MT, Vaughn D, Nicholson AJ, et al. Mutations in

the D1-pyrroline 5-carboxylate dehydrogenase gene cause type

II hyperprolinaemia. Hum Mol Genet 1998;7:1411–5.

[7] Mezl VA, Knox WE. Properties and analysis of a stable de-

rivative of pyrroline-5-carboxylic acid for use in metabolic

studies. Anal Biochem 1976;74:430–40.

[8] Ersser RS, Smith I. Amino acids and related compounds. In:

Smith I, Seakins JWT, editors. Chromatographic and electro-

phoretic techniques. Bath, UK: William Heinemann Medical

Books; 1976. p. 75–108.

[9] Vogel HJ, Davis BD. Glutamic g-semialdehyde and D1-pyrro-

line-5-carboxylic acid, intermediates in the biosynthesis of

proline. J Am Chem Soc 1952;74:109–22.

[10] Strecker HJ. The interconversion of glutamic acid and proline:

III. D1-Pyrroline-5-carboxylic acid dehydrogenase. J Biol

Chem 1960;235:3218–23.

[11] Forte-McRobbie CM, Pietruszko R. Purification and charac-

terization of human liver ‘‘high Km’’ aldehyde dehydrogenase

and its identification as glutamic g-semialdehyde dehydrogen-

ase. J Biol Chem 1986;261:2154–63.

[12] Small WC, Jones ME. Pyrroline 5-carboxylate dehydrogenase

of the mitochondrial matrix rat liver. J Biol Chem 1990;265:

18668–72.

[13] Hagedorn CH, Phang JM. Catalytic transfer of hydride ions

from NADPH to oxygen by the interconversions of proline

and delta 1-pyrroline-5-carboxylate. Arch Biochem Biophys

1986;248:166–74.

[14] Fleming GA, Hagedorn CH, Granger AS, Phang JM. Pyrro-

line-5-carboxylate in human plasma. Metabolism 1984;33:

739–42.

[15] Heinemann L, Asche W, Withold W, Berger M. Quantitative

relationship between ketonuria as determined by four rapid

tests and simultaneously present ketonaemia. Diab Stoffw

1994;3:339–42.


A novel pyrroline-5-carboxylic acid and acetoacetic acid adduct in hyperprolinaemia type II

Documents

Transcript of A novel pyrroline-5-carboxylic acid and acetoacetic acid adduct in hyperprolinaemia type II