Neuropathology and Applied Neurobiology 1991,17,441-456
Developmental profile of plasma proteins in human fetal cerebrospinal fluid and blood J. E. BELL, A. A. FRYER*, M. COLLINS?, T. MARSHALL?, P. W. JONES$, R. S T R A N G E * A N D R. H U M E t Department of Pathology, University of Edinburgh, *School of Postgraduate Medicine and Biological Sciences, University of Keele, ?Department of Child Life and Health, University of Edinburgh and $Department of Mathematics, University of Keele
BELLJ. E., FRYERA. A., COLLINS M., MARSHALL T., JONESP. W., STRANGE R. & H u m R. (199 1) Neuropathology and Applied Neurobiology 17,441-456 Developmental profile of plasma proteins in human fetal cerebrospinal fluid and blood Total protein, alphafetoprotein, albumin, prealbumin, alpha-l-antitrypsin, transferrin and ceruloplasmin levels were measured in samples of human fetal and neonatal cerebrospinal fluid (CSF) (97 cases), obtained between 12 and 41 weeks of gestation. In 39 of these cases (13 to 40 weeks gestation) plasma was also available for comparative analysis. CSF was collected from lateral ventricles in the first half of gestation and from the lumbar region in the second. Since these CSF samples were obtained from different sites, the protein levels in the lateral ventricle (early) samples could not be compared directly with those in the lumbar (later) samples. However, the mean protein levels in the lumbar samples were lower than those in the ventricular samples, which is in accord with the decline in CSF protein levels described in maturing animal fetuses. Despite a wide scatter of results, particularly in the first half of gestation, significant decline in the level of CSF alphafetoprotein was demonstrated during both first and second halves of gestation, and of albumin and prealbumin in the second half. No sex differences were found except for ceruloplasmin in lumbar CSF later in gestation, when males had higher levels than females. In fetal plasma, protein levels increased with increasing gestation apart from alphafetoprotein and prealbumin which both declined progressively. CSF/plasma ratios were dissimilar for different proteins, and changed with increasing gestation. These findings support the concept that the human fetal blood brain barrier matures early. Keywords: human fetus, cerebrospinal fluid, plasma, plasma proteins, development
INTRODUCTION It has been shown in many animal species (Ramey & Birge, 1979; Dziegielewska et al., 1980, 1981; Cavanagh et al., 1982; New, Dziegielewska & Saunders, 1983; Fielitz, Esteves & Moro, 1984) that, during crucial early stages of development, the brain is supported in cerebrospinal fluid (CSF) which contains significantly higher levels of protein than are found in the adult. In Correspondence to: Dr J. E. Bell, Neuropathology Laboratory, Western General Hospital, Crewe Road, Edinburgh EH4 2XU.
442
J . E. Bell et al.
all of these studies, CSF proteins were compared with those in plasma in order to gain better understanding of protein partitioning during ontogeny of the central nervous system (CNS) (Davson, Welch & Segal, 1987). Proteins in fetal CSF include albumin, prealbumin, alpha-lantitrypsin, transfemn and alphafetoprotein and it is likely that these are products of the same genes that encode the corresponding forms in plasma (Davson et al., 1987; Dziegielewska et al., 1980). Only limited data are available regarding protein levels in CSF and plasma in normal human fetuses. In the widely quoted study of Adinolfi and Haddad (1977) the majority of subjects were trisomic, and Brock and Sutcliffe (1972) obtained data from spina bifida fetuses. The high CSF protein concentrations found in these cases, albeit mostly abnormal, suggest that developing humans resemble animals in this regard (Adinolfi, Adinolfi & Lessof, 1975;Adinolfi et al., 1976; Seller & Adinolfi, 1975)and premature infants also show levels of CSF protein which are higher than those at term (Otila, 1948;Statz & Felgenhauer, 1983).With respect to data on human fetal plasma, discrepancies in protein levels have been described between human fetuses and adults, both at term (Cochran & Good, 1974;Hill & Abramson, 1988)and, in the case of albumin, from 12 weeks of gestation (Krauer, Dayer & Anner, 1984; Nau& Krauer, 1986).In a study of 63 fetal blood specimens obtained by fetoscopy between 20 and 26 weeks of gestation, Forestier et al. (1987) found higher alphafetoprotein but lower total protein and albumin in the fetuses as compared with their mothers. We now describe the first large scale study of the developmental profiles of total protein, alphafetoprotein, albumin, prealbumin, transferrin, alpha- l-antitrypsin and ceruloplasmin in normal human fetal CSF samples, together with a number of matched plasma samples, obtained between 12 and 41 weeks of gestation. Comparison of developmental profiles of proteins in plasma and CSF is the first step towards understanding the human fetal blood/CSF barrier. MATERIALS A N D METHODS
CSF and plasma samples CSF was obtained from 97 cases between 12 and 41 weeks gestation and from 39 of these cases, plasma was also available. Blood and/or CSF samples were obtained from (a) 42 apparently normal human fetuses, including two sets of twins at 14 and 16 weeks, where pregnancy was terminated by extraamniotic prostaglandin between 12-20 weeks of gestation, and from (b) 55 neonates born between 24 and 41 weeks of gestation. In group (a), CSF and cardiac blood were obtained not more than 6 h, and usually within 2 h, of delivery. In group (b), CSF and venous blood were obtained within 24 h of birth as part of an infection screen in infants without overt neurological problems, and where CSF and blood were sterile on culture. Males represented 60% of the cases. In fetuses of 12-20 weeks gestation (group (a)), blood was obtained by cardiac puncture from 13 fetuses. Cardiac blood was centrifuged and the supernatant serum was collected. In some fetuses, cardiac blood was no longer present at the time of examination. In the same age range, CSF was aspirated from one or both lateral ventricles (LV)in 41 fetuses. The samples of CSF were obtained through the shaft of a needle (Becton Dickinson Microlance, 21 G), snapped free of its plastic end and inserted at the lateral margin of the anterior fontanelle. Very gentle advancement followed by rotation yielded clear LV CSF, not visibly contaminated by blood or by cerebral or choroid tissues. The CSF dripped steadily from the needle into a
Humanfetal CSFproteins
443
container and no suction was applied. Fetuses which did not yield crystal-clear CSF were excluded from the study. From one fetus of 20 weeks (group (a)) and 55 fetuses and infants of between 24 and 4 1 weeks of gestation (group (b)), CSF sampleswere obtained by lumbar puncture. In 26 of these cases, venous blood was collected and plasma separated by centrifugation of the heparinized samples. Samples of plasma and CSF at all ages were examined individually for protein content and no specimens were pooled. Specimens were kept at - 70°C until determination of protein concentrations. A careful estimate of developmental age was made in each fetus, based on size (including crown-heel, crown-rump and heel-toe measurements) (Scammon & Calkins, 1929), menstrual history and ultrasound dating of pregnancy. In neonates, gestational age was assessed by the Dubowitz score. Normality was confirmed by autopsy in aborted fetuses and by external examination and subsequent observation in surviving neonates. The study was approved by the Paediatric Reproductive Medicine Ethics of Medical Research Sub-committee of the Lothian Health Board.
Analytical Total protein concentrations in plasma and CSF were determined by using the bicinchoninic acid method (Sigma Chemical Co., St Louis, Missouri, USA). The between batch coefficient of variation was 4% The concentrations of albumin, prealbumin, transferrin, alphafetoprotein, alpha- 1antitrypsin and ceruloplasmin in plasma and CSF were determined by single radialimmunodiffusion in plates (LC and M-Partigen Immunodiffusion Plates) obtained from Behring Diagnostics. Because of the wide range of protein concentrations in CSF during development it was necessary to dilute samples (between 1:3 and 1:12 v/v in 154mM NaCI) before determination of albumin, alpha- 1-antitrypsin and transferrin concentrations (LC plates). Reproducibility was determined by assaying commercially available control samples (Behring Diagnostics). For each protein, three controls were used covering high, medium and low values. The average between-batch coefficient of variation for the protein assays was 6.6%. Lowest detection limits for the various proteins were: ceruloplasmin, 1 mg/l; albumin, 130 mg/l; prealbumin, 20 mg/ml; alpha-1-antitrypsin, 1.5mg/l; transferrin, 6 mg/ml, alphafetoprotein, 1 mg/l, total protein, 50 mg/l. Plasma samples were also diluted where appropriate; dilutions of 1:300in 154 mM NaCl were made before determination of albumin concentrations. Electrophoresis was performed in the LKB midget unit using gels containing a gradient of 1&20% polyacrylamide with a stacking gel comprising 5% polyacrylamide. CSF samples were diluted 1 to 1 v/v with buffer ( 2 5 0 m ~Tris-HC1, pH 6.8 containing 8% v/v SDS, 40% v/v glycerol, 20% v/v beta-mercaptoethanol and 0.02% v/v bromophenol blue) heated (100°C for 5 min) and 10 p1 of this mixture applied to the gels. Electrophoresis was carried out at 120 V for 90 min. Gels were stained with 0.01YO v/v page blue dissolved in water containing 20% v/v methanol, 10% v/v acetic acid. Destaining was performed using a mixture of 7% v/v acetic acid and 5% v/v ethanol. CSF viscosity was determined at 25°C by the capillary tube method using a Coulter Hackness Viscometer (Coulter Electronics Ltd, Luton, UK). Statistical methods
Correlation analysis (Pearson product moment) was used to investigate possible linear relationships between protein concentrations in CSF and plasma with time. F tests for equality of
~
~~
L
L
L
7
5
10
36-37
38-39
40-41
30 I26 (60-192) 32.2 (3 1.5-33) 85f31.5 (24-234) 41.7 (36-45) 57.8 f 7.7 (1 8-1 31) 65.4 & 7.2 (51-93) 54f 13.5 (IS-105) 31.8f4.3 (21-46) 30k7.1 (1.5-74)
180
293.1k53.1 (60447) 244.5 f 33.8 (57-579) 235.5+ 57.6 (72-372) I53 f42.5 (60-348)
Antir
3210f 526.4 (13445592) 2964.6i428.6 (6724744) 3862 f89 I .2 (12484384) 2988 f672.8 (1 8966120) 5484 1014 2928 (14884368) 954 (936-972) I243 f 254.1 (504-2238) 2537.3 (528-5500) 1328.6+ 195.3 (606-3558) 1228.8k212.2 (804-1848) 999.64119.1 (714-1296) 627.6 k 99.9 (426-1008) 594.6k88.3 (132-1068)
Alb
AFP
1484i355.2 (324-3630) 808.9 It: 130.I (55-2040) 730.8 k 149.9 (220-1 110) 357.5f 144.2 (I 17-1032) 288 71 91.5 (54- 129) 40 (30-50) 19.2k4.7 (4.8-36) 18.1 (8.8-26) 18.1 k3.3 (6.8-53) 14.2& 5 (7.2-34) 6.4f 2.9 (0-20) 0.8 f 0.8 (0-3.8) 1.1k0.6 (0-5.3)
Prealb
201.3 49.7 (164-239) 163.6+ 13.1 (59-279) I45 f26.6 (75-230) 149.1 k45.4 (65-368) 38 61 55
(4842) 43 (3947) 45.3?; 2.1 (35.5-49) 41 (3-5) 40.5 ?; 2.8 (2245) 38.4k5.8 (24-52.5) 41.8k3.8 (25-53) 37.7 2.4 (30.5-45) 40.2 f2.1 (30.5-52.5)
Trairs
160.4f3 I. I (15-3 18) 146.1f20.2 (42-378) 146.8k34.8 (45-261) 118 k 29.6 (36-240) 96 42 110.5
(71-150) 35.2 (7.5-63) 52.5 f 14.9 (20-1 19) 62.3 (4S-81) 67.8& 11.8 (7.5- I 98) 69.2f 10.8 (45-102) 66.2 f 10.8 (38- 108) 45.6 19.2 (18 4 6 ) 37.517.7 (M6)
(780-3780) I990 (1250-2730) 2578.3 k620.8 ( 1200-9 130) 2360f 134.5 (2100-2550) 1646& 274.7 (850-2550) 144Of440 (1 OW1880) 1798.3 f 348.8 (780-3300)
( 1930-1930) 2035 f 63 I. 1
1930
5216k 754.5 (3400-7130) 3945 f 710.1 (1050-9800) 7442 f2560 (2630-1 3880) 5136f 1234.2 (3050-9750) 6850 I650 ND
Total proteiii
Key: Cer, ceruloplasrnin; Antit, alpha-I-antitrypsin;Alb, albumin; Trans, transferrin;Prealb, prealbumin; AFP, alphafetoprotein;LV,lateral ventricle; L,lumbar; ND, not determined.
(0-10.5)
L
5
34-35
30-3 1 5 . 6 i 1.8 (0-20.5) 4.8 f3 (0-13) 9.5 f5.2 (0-33) 4k2.5 (0-12) 1.6k 1.1
28-29
L
(0-1 1.5) 0
L
6
26-27
LV L L
15
2
20-2 1 20-2 I 24-25
LV
32-33
2
18-19
LV
L
6
16-17
LV
3
6
14-15
4.5k1.7 (0-12.5) 3.8k 1 (0-12.5) 6.3f2.3 (0-13.5) 1.2k 1.2 (0-7) 0 0 5.7
LV
4.9f3.2 (0-17.5) 0
20
12-13
Cer
Site
L
8
1 1
n
Age
Table 1. Protein levels in human fetal CSF by fortnight of gestation (mg/litre). Mean & s.e. (ifn > 3) (range)
Humanfetal CSFproteins Table 2. Protein content (mg/l) and viscosity (NMS-2x and lumbar CSF (all samples)
445
of lateral ventricle
n
mean
SD
95% CI
Lateral ventricle Cer Antit Alb Trans Prealb AFP Total protein Viscosity
39 38 39 39 38 39 28 9
3.70 236.40 3304.30 141.90 159.90 857.00 5676.80 9.26
4.4 139.7 1809.1 81 66.7 711.4 3507.9 0.18
2.2-5.1 190.4-282.3 271 7.9-3890.7 115.6-168.1 138.0-181.8 626.4-1087.6 43 16.7-7036.9 9.1-9.41
Lumbar Cer Antit Alb Trans Prealb AFP Total protein Viscosity
55 55 53 55 56 54 36 43
4.4 60.1 1085.3 58.2 41.7 16.6 2108.9 9.05
7 60.3 729.6 35.6 9.2 22.2 1387.6 0.5
2.5-6.3 43.8-76.4 884.2-1286.4 48.6-67.8 39.2f44.1 10.6-22.7 1639.4-2578.4 9.0-9.09
For abbreviations, see Table 1. Note: one twin randomly excluded from each of two pairs in this analysis.
variances were carried out to compare scatter in protein concentrations before and after 20 weeks of gestation. Independent t-tests were used to compare protein concentrations in CSF samples from males and females. CSF/plasma ratios were calculated for each gestational stage, and the means of CSF and plasma proteins in matched samples were also compared. All tests' were based on the set of independent data with one twin of each of two sets being excluded from the analysis. Descriptions of the data (e.g. Tables 1 and 2, and Figure 1) are based on the full data set, including the two sets of twins. RESULTS Relationship between CSF protein concentrations and gestational time
To review the expression of the selected CSF proteins during the developmental period 12-41 weeks gestation, data from both CSF sites were first combined. Figure 1 (a-g) shows the concentrations of proteins in lateral ventricular (LV) and lumbar CSF during this period. Table 1 summarizes mean CSF protein values for each gestational stage (including the twin sets). Table 2 summarizes the data for all the lateral ventricle (excluding one of each set of twins) and lumbar samples of CSF. Inspection of Figure 1 and Tables 1 and 2 shows that, except for ceruloplasmin, the mean concentration of proteins was lower in the lumbar samples from the second half of gestation than in the LV samples collected in the first half of gestation. Figure 1 shows the generally greater scatter among samples obtained pre-20 weeks gestation than among samples obtained
J . E. Bell et al.
446
b, Alpha-l-Antitrypsin
a, Ceruloplasmin
301
450
25
. Gestational age (weeks)
Gestational age (weeks)
d, Transferrin
c, Albumin
35010 300
5000
.-c
.-c
5
0
L
z
3000
c v)
m C
$ 2000
100
1000
50 I
15
I
23 27 31 35 Gestational age (weeks) 19
I
0
39
0
Gestational age (weeks)
Figure l.(a-d) Figure l.(a-g) Concentrations of individual and total proteins in CSF from lateral ventricle (0)and lumbar (0)site during developmentalperiod 12-41 weeks gestation. Individual female (W) and male (0)ceruloplasmin values in lumbar CSF are identified to illustrate significantdifferencesbetween the sexes. Regression lines are fitted to the scattergrams where a significantfall in protein value was detected in LV or L CSF.
after 20 weeks gestation. In some cases, a decline of CSF protein value with time was also shown within the first half, or within the second half of gestation. Thus, in the pre-20 weeks gestation group (a), alphafetoprotein (rj9= - 0.41, P < 0.05); and in the post-20 weeks gestation group (b), aIphafetoprotein (r%= -0.69, P
Developmental profile of plasma proteins in human fetal cerebrospinal fluid and blood J. E. BELL, A. A. FRYER*, M. COLLINS?, T. MARSHALL?, P. W. JONES$, R. S T R A N G E * A N D R. H U M E t Department of Pathology, University of Edinburgh, *School of Postgraduate Medicine and Biological Sciences, University of Keele, ?Department of Child Life and Health, University of Edinburgh and $Department of Mathematics, University of Keele
BELLJ. E., FRYERA. A., COLLINS M., MARSHALL T., JONESP. W., STRANGE R. & H u m R. (199 1) Neuropathology and Applied Neurobiology 17,441-456 Developmental profile of plasma proteins in human fetal cerebrospinal fluid and blood Total protein, alphafetoprotein, albumin, prealbumin, alpha-l-antitrypsin, transferrin and ceruloplasmin levels were measured in samples of human fetal and neonatal cerebrospinal fluid (CSF) (97 cases), obtained between 12 and 41 weeks of gestation. In 39 of these cases (13 to 40 weeks gestation) plasma was also available for comparative analysis. CSF was collected from lateral ventricles in the first half of gestation and from the lumbar region in the second. Since these CSF samples were obtained from different sites, the protein levels in the lateral ventricle (early) samples could not be compared directly with those in the lumbar (later) samples. However, the mean protein levels in the lumbar samples were lower than those in the ventricular samples, which is in accord with the decline in CSF protein levels described in maturing animal fetuses. Despite a wide scatter of results, particularly in the first half of gestation, significant decline in the level of CSF alphafetoprotein was demonstrated during both first and second halves of gestation, and of albumin and prealbumin in the second half. No sex differences were found except for ceruloplasmin in lumbar CSF later in gestation, when males had higher levels than females. In fetal plasma, protein levels increased with increasing gestation apart from alphafetoprotein and prealbumin which both declined progressively. CSF/plasma ratios were dissimilar for different proteins, and changed with increasing gestation. These findings support the concept that the human fetal blood brain barrier matures early. Keywords: human fetus, cerebrospinal fluid, plasma, plasma proteins, development
INTRODUCTION It has been shown in many animal species (Ramey & Birge, 1979; Dziegielewska et al., 1980, 1981; Cavanagh et al., 1982; New, Dziegielewska & Saunders, 1983; Fielitz, Esteves & Moro, 1984) that, during crucial early stages of development, the brain is supported in cerebrospinal fluid (CSF) which contains significantly higher levels of protein than are found in the adult. In Correspondence to: Dr J. E. Bell, Neuropathology Laboratory, Western General Hospital, Crewe Road, Edinburgh EH4 2XU.
442
J . E. Bell et al.
all of these studies, CSF proteins were compared with those in plasma in order to gain better understanding of protein partitioning during ontogeny of the central nervous system (CNS) (Davson, Welch & Segal, 1987). Proteins in fetal CSF include albumin, prealbumin, alpha-lantitrypsin, transfemn and alphafetoprotein and it is likely that these are products of the same genes that encode the corresponding forms in plasma (Davson et al., 1987; Dziegielewska et al., 1980). Only limited data are available regarding protein levels in CSF and plasma in normal human fetuses. In the widely quoted study of Adinolfi and Haddad (1977) the majority of subjects were trisomic, and Brock and Sutcliffe (1972) obtained data from spina bifida fetuses. The high CSF protein concentrations found in these cases, albeit mostly abnormal, suggest that developing humans resemble animals in this regard (Adinolfi, Adinolfi & Lessof, 1975;Adinolfi et al., 1976; Seller & Adinolfi, 1975)and premature infants also show levels of CSF protein which are higher than those at term (Otila, 1948;Statz & Felgenhauer, 1983).With respect to data on human fetal plasma, discrepancies in protein levels have been described between human fetuses and adults, both at term (Cochran & Good, 1974;Hill & Abramson, 1988)and, in the case of albumin, from 12 weeks of gestation (Krauer, Dayer & Anner, 1984; Nau& Krauer, 1986).In a study of 63 fetal blood specimens obtained by fetoscopy between 20 and 26 weeks of gestation, Forestier et al. (1987) found higher alphafetoprotein but lower total protein and albumin in the fetuses as compared with their mothers. We now describe the first large scale study of the developmental profiles of total protein, alphafetoprotein, albumin, prealbumin, transferrin, alpha- l-antitrypsin and ceruloplasmin in normal human fetal CSF samples, together with a number of matched plasma samples, obtained between 12 and 41 weeks of gestation. Comparison of developmental profiles of proteins in plasma and CSF is the first step towards understanding the human fetal blood/CSF barrier. MATERIALS A N D METHODS
CSF and plasma samples CSF was obtained from 97 cases between 12 and 41 weeks gestation and from 39 of these cases, plasma was also available. Blood and/or CSF samples were obtained from (a) 42 apparently normal human fetuses, including two sets of twins at 14 and 16 weeks, where pregnancy was terminated by extraamniotic prostaglandin between 12-20 weeks of gestation, and from (b) 55 neonates born between 24 and 41 weeks of gestation. In group (a), CSF and cardiac blood were obtained not more than 6 h, and usually within 2 h, of delivery. In group (b), CSF and venous blood were obtained within 24 h of birth as part of an infection screen in infants without overt neurological problems, and where CSF and blood were sterile on culture. Males represented 60% of the cases. In fetuses of 12-20 weeks gestation (group (a)), blood was obtained by cardiac puncture from 13 fetuses. Cardiac blood was centrifuged and the supernatant serum was collected. In some fetuses, cardiac blood was no longer present at the time of examination. In the same age range, CSF was aspirated from one or both lateral ventricles (LV)in 41 fetuses. The samples of CSF were obtained through the shaft of a needle (Becton Dickinson Microlance, 21 G), snapped free of its plastic end and inserted at the lateral margin of the anterior fontanelle. Very gentle advancement followed by rotation yielded clear LV CSF, not visibly contaminated by blood or by cerebral or choroid tissues. The CSF dripped steadily from the needle into a
Humanfetal CSFproteins
443
container and no suction was applied. Fetuses which did not yield crystal-clear CSF were excluded from the study. From one fetus of 20 weeks (group (a)) and 55 fetuses and infants of between 24 and 4 1 weeks of gestation (group (b)), CSF sampleswere obtained by lumbar puncture. In 26 of these cases, venous blood was collected and plasma separated by centrifugation of the heparinized samples. Samples of plasma and CSF at all ages were examined individually for protein content and no specimens were pooled. Specimens were kept at - 70°C until determination of protein concentrations. A careful estimate of developmental age was made in each fetus, based on size (including crown-heel, crown-rump and heel-toe measurements) (Scammon & Calkins, 1929), menstrual history and ultrasound dating of pregnancy. In neonates, gestational age was assessed by the Dubowitz score. Normality was confirmed by autopsy in aborted fetuses and by external examination and subsequent observation in surviving neonates. The study was approved by the Paediatric Reproductive Medicine Ethics of Medical Research Sub-committee of the Lothian Health Board.
Analytical Total protein concentrations in plasma and CSF were determined by using the bicinchoninic acid method (Sigma Chemical Co., St Louis, Missouri, USA). The between batch coefficient of variation was 4% The concentrations of albumin, prealbumin, transferrin, alphafetoprotein, alpha- 1antitrypsin and ceruloplasmin in plasma and CSF were determined by single radialimmunodiffusion in plates (LC and M-Partigen Immunodiffusion Plates) obtained from Behring Diagnostics. Because of the wide range of protein concentrations in CSF during development it was necessary to dilute samples (between 1:3 and 1:12 v/v in 154mM NaCI) before determination of albumin, alpha- 1-antitrypsin and transferrin concentrations (LC plates). Reproducibility was determined by assaying commercially available control samples (Behring Diagnostics). For each protein, three controls were used covering high, medium and low values. The average between-batch coefficient of variation for the protein assays was 6.6%. Lowest detection limits for the various proteins were: ceruloplasmin, 1 mg/l; albumin, 130 mg/l; prealbumin, 20 mg/ml; alpha-1-antitrypsin, 1.5mg/l; transferrin, 6 mg/ml, alphafetoprotein, 1 mg/l, total protein, 50 mg/l. Plasma samples were also diluted where appropriate; dilutions of 1:300in 154 mM NaCl were made before determination of albumin concentrations. Electrophoresis was performed in the LKB midget unit using gels containing a gradient of 1&20% polyacrylamide with a stacking gel comprising 5% polyacrylamide. CSF samples were diluted 1 to 1 v/v with buffer ( 2 5 0 m ~Tris-HC1, pH 6.8 containing 8% v/v SDS, 40% v/v glycerol, 20% v/v beta-mercaptoethanol and 0.02% v/v bromophenol blue) heated (100°C for 5 min) and 10 p1 of this mixture applied to the gels. Electrophoresis was carried out at 120 V for 90 min. Gels were stained with 0.01YO v/v page blue dissolved in water containing 20% v/v methanol, 10% v/v acetic acid. Destaining was performed using a mixture of 7% v/v acetic acid and 5% v/v ethanol. CSF viscosity was determined at 25°C by the capillary tube method using a Coulter Hackness Viscometer (Coulter Electronics Ltd, Luton, UK). Statistical methods
Correlation analysis (Pearson product moment) was used to investigate possible linear relationships between protein concentrations in CSF and plasma with time. F tests for equality of
~
~~
L
L
L
7
5
10
36-37
38-39
40-41
30 I26 (60-192) 32.2 (3 1.5-33) 85f31.5 (24-234) 41.7 (36-45) 57.8 f 7.7 (1 8-1 31) 65.4 & 7.2 (51-93) 54f 13.5 (IS-105) 31.8f4.3 (21-46) 30k7.1 (1.5-74)
180
293.1k53.1 (60447) 244.5 f 33.8 (57-579) 235.5+ 57.6 (72-372) I53 f42.5 (60-348)
Antir
3210f 526.4 (13445592) 2964.6i428.6 (6724744) 3862 f89 I .2 (12484384) 2988 f672.8 (1 8966120) 5484 1014 2928 (14884368) 954 (936-972) I243 f 254.1 (504-2238) 2537.3 (528-5500) 1328.6+ 195.3 (606-3558) 1228.8k212.2 (804-1848) 999.64119.1 (714-1296) 627.6 k 99.9 (426-1008) 594.6k88.3 (132-1068)
Alb
AFP
1484i355.2 (324-3630) 808.9 It: 130.I (55-2040) 730.8 k 149.9 (220-1 110) 357.5f 144.2 (I 17-1032) 288 71 91.5 (54- 129) 40 (30-50) 19.2k4.7 (4.8-36) 18.1 (8.8-26) 18.1 k3.3 (6.8-53) 14.2& 5 (7.2-34) 6.4f 2.9 (0-20) 0.8 f 0.8 (0-3.8) 1.1k0.6 (0-5.3)
Prealb
201.3 49.7 (164-239) 163.6+ 13.1 (59-279) I45 f26.6 (75-230) 149.1 k45.4 (65-368) 38 61 55
(4842) 43 (3947) 45.3?; 2.1 (35.5-49) 41 (3-5) 40.5 ?; 2.8 (2245) 38.4k5.8 (24-52.5) 41.8k3.8 (25-53) 37.7 2.4 (30.5-45) 40.2 f2.1 (30.5-52.5)
Trairs
160.4f3 I. I (15-3 18) 146.1f20.2 (42-378) 146.8k34.8 (45-261) 118 k 29.6 (36-240) 96 42 110.5
(71-150) 35.2 (7.5-63) 52.5 f 14.9 (20-1 19) 62.3 (4S-81) 67.8& 11.8 (7.5- I 98) 69.2f 10.8 (45-102) 66.2 f 10.8 (38- 108) 45.6 19.2 (18 4 6 ) 37.517.7 (M6)
(780-3780) I990 (1250-2730) 2578.3 k620.8 ( 1200-9 130) 2360f 134.5 (2100-2550) 1646& 274.7 (850-2550) 144Of440 (1 OW1880) 1798.3 f 348.8 (780-3300)
( 1930-1930) 2035 f 63 I. 1
1930
5216k 754.5 (3400-7130) 3945 f 710.1 (1050-9800) 7442 f2560 (2630-1 3880) 5136f 1234.2 (3050-9750) 6850 I650 ND
Total proteiii
Key: Cer, ceruloplasrnin; Antit, alpha-I-antitrypsin;Alb, albumin; Trans, transferrin;Prealb, prealbumin; AFP, alphafetoprotein;LV,lateral ventricle; L,lumbar; ND, not determined.
(0-10.5)
L
5
34-35
30-3 1 5 . 6 i 1.8 (0-20.5) 4.8 f3 (0-13) 9.5 f5.2 (0-33) 4k2.5 (0-12) 1.6k 1.1
28-29
L
(0-1 1.5) 0
L
6
26-27
LV L L
15
2
20-2 1 20-2 I 24-25
LV
32-33
2
18-19
LV
L
6
16-17
LV
3
6
14-15
4.5k1.7 (0-12.5) 3.8k 1 (0-12.5) 6.3f2.3 (0-13.5) 1.2k 1.2 (0-7) 0 0 5.7
LV
4.9f3.2 (0-17.5) 0
20
12-13
Cer
Site
L
8
1 1
n
Age
Table 1. Protein levels in human fetal CSF by fortnight of gestation (mg/litre). Mean & s.e. (ifn > 3) (range)
Humanfetal CSFproteins Table 2. Protein content (mg/l) and viscosity (NMS-2x and lumbar CSF (all samples)
445
of lateral ventricle
n
mean
SD
95% CI
Lateral ventricle Cer Antit Alb Trans Prealb AFP Total protein Viscosity
39 38 39 39 38 39 28 9
3.70 236.40 3304.30 141.90 159.90 857.00 5676.80 9.26
4.4 139.7 1809.1 81 66.7 711.4 3507.9 0.18
2.2-5.1 190.4-282.3 271 7.9-3890.7 115.6-168.1 138.0-181.8 626.4-1087.6 43 16.7-7036.9 9.1-9.41
Lumbar Cer Antit Alb Trans Prealb AFP Total protein Viscosity
55 55 53 55 56 54 36 43
4.4 60.1 1085.3 58.2 41.7 16.6 2108.9 9.05
7 60.3 729.6 35.6 9.2 22.2 1387.6 0.5
2.5-6.3 43.8-76.4 884.2-1286.4 48.6-67.8 39.2f44.1 10.6-22.7 1639.4-2578.4 9.0-9.09
For abbreviations, see Table 1. Note: one twin randomly excluded from each of two pairs in this analysis.
variances were carried out to compare scatter in protein concentrations before and after 20 weeks of gestation. Independent t-tests were used to compare protein concentrations in CSF samples from males and females. CSF/plasma ratios were calculated for each gestational stage, and the means of CSF and plasma proteins in matched samples were also compared. All tests' were based on the set of independent data with one twin of each of two sets being excluded from the analysis. Descriptions of the data (e.g. Tables 1 and 2, and Figure 1) are based on the full data set, including the two sets of twins. RESULTS Relationship between CSF protein concentrations and gestational time
To review the expression of the selected CSF proteins during the developmental period 12-41 weeks gestation, data from both CSF sites were first combined. Figure 1 (a-g) shows the concentrations of proteins in lateral ventricular (LV) and lumbar CSF during this period. Table 1 summarizes mean CSF protein values for each gestational stage (including the twin sets). Table 2 summarizes the data for all the lateral ventricle (excluding one of each set of twins) and lumbar samples of CSF. Inspection of Figure 1 and Tables 1 and 2 shows that, except for ceruloplasmin, the mean concentration of proteins was lower in the lumbar samples from the second half of gestation than in the LV samples collected in the first half of gestation. Figure 1 shows the generally greater scatter among samples obtained pre-20 weeks gestation than among samples obtained
J . E. Bell et al.
446
b, Alpha-l-Antitrypsin
a, Ceruloplasmin
301
450
25
. Gestational age (weeks)
Gestational age (weeks)
d, Transferrin
c, Albumin
35010 300
5000
.-c
.-c
5
0
L
z
3000
c v)
m C
$ 2000
100
1000
50 I
15
I
23 27 31 35 Gestational age (weeks) 19
I
0
39
0
Gestational age (weeks)
Figure l.(a-d) Figure l.(a-g) Concentrations of individual and total proteins in CSF from lateral ventricle (0)and lumbar (0)site during developmentalperiod 12-41 weeks gestation. Individual female (W) and male (0)ceruloplasmin values in lumbar CSF are identified to illustrate significantdifferencesbetween the sexes. Regression lines are fitted to the scattergrams where a significantfall in protein value was detected in LV or L CSF.
after 20 weeks gestation. In some cases, a decline of CSF protein value with time was also shown within the first half, or within the second half of gestation. Thus, in the pre-20 weeks gestation group (a), alphafetoprotein (rj9= - 0.41, P < 0.05); and in the post-20 weeks gestation group (b), aIphafetoprotein (r%= -0.69, P
