Alpha-fetoprotein in human fetal cerebrospinal fluid

Clinica Chimica Acta 291 (2000) 35–41www.elsevier.com/ locate /clinchim

Alpha-fetoprotein in human fetal cerebrospinal fluid

*Michael Christiansen , Claus K. Høgdall, Estrid V.S. Høgdall

Department of Clinical Biochemistry, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen S, Denmark

Received 1 June 1999; received in revised form 10 September 1999; accepted 15 September 1999

Abstract

Alpha-fetoprotein (AFP) is a fetal glycoprotein. It has been ascribed a regulatory function ofgrowth factor responses and immune functions. The concentrations of AFP and albumin (ALB) arehighly variable in fetal serum and CSF and change with gestational age. The AFP index 5

[AFP /AFP ] / [ALB /ALB ] was determined in six normal fetuses at gestationalCSF SERUM CSF SERUM

age 17–23 weeks and found to be independent of gestational age and close to unity, mean0.9060.11 (S.D.). The ratio of CSF-serum concentrations of AFP and ALB both decreasedsignificantly ( p , 0.05) with gestational age. The mean fraction of AFP being non-reactive withconcanavalin A was 1.7% in serum and 1.9% in CSF, suggesting a common hepatic origin of AFPin both compartments. In conclusion, the concentration of AFP in CSF seems to be determinedlargely by the serum-CSF concentration gradient in normal fetuses. This finding, combined withthe remarkable constancy of the AFP index compared to the highly variable absolute con-centrations of AFP in both serum and CSF should make the AFP index the marker of choice whenanalyzing for intrathecal AFP synthesis during development and in pathological conditions. 2000 Elsevier Science B.V. All rights reserved.

Keywords: Alpha-fetoprotein; Fetal; Albumin; AFP index; Cerebrospinal fluid; Intrathecalsynthesis

1. Introduction

AFP is found in fetal serum and cerebrospinal fluid (CSF) in very high and

*Corresponding author. Tel.: 1 45-32-683-657; fax: 1 45-32-683-878.E-mail address: [email protected] (M. Christiansen)

0009-8981/00/$ – see front matter 2000 Elsevier Science B.V. All rights reserved.PI I : S0009-8981( 99 )00195-3

36 M. Christiansen et al. / Clinica Chimica Acta 291 (2000) 35 –41

highly variable concentrations [1,2]. Several investigators have found evidencefor AFP synthesis in neurons at least during certain phases of mammaliandevelopment [3–6], whereas others have failed to do so [7]. The source of AFPin fetal CSF may thus be both neurons via brain interstitial fluid (brain–CSFbarrier) or plasma via the blood–CSF barrier or by retrograde transport alongmotorneurons [8]. Albumin (ALB) is also found in fetal CSF and synthesis ofALB by brain cells has been demonstrated [1]. As AFP has been ascribedimmunoregulatory [9] and, recently, modulatory function on growth factorresponses [10], the role of AFP in the development may be of significance,particularly so as one study has shown AFP to be located in the developingcortical plate during a narrow period in development [11]. At the cellular levelAFP seems to be responsible for specific receptor-mediated uptake of polyun-saturated fatty acids during proliferation and differentiation [6] and this functionmay be important for development of brain cells.

To approach the problem of the source of AFP in CSF we determined AFPindex 5 [AFP AFP ] / [ALB /ALB ] in six normal fetuses ofCSF SERUM CSF SERUM

gestational age 17–23 weeks. The AFP index is the ratio between the net resultof transport into and uptake from CSF of AFP and the net result of the samemechanisms for ALB. In adults the ratio between concentrations in CSF andserum for a specific protein is largely dependent on physico-chemical properties,particularly Stokes’ radius [12]. Due to the close physico-chemical similaritybetween AFP and ALB (40% aminoacid homology, M 69 000 and 66 000,r

respectively, and the same pI) [13] only differences in specific mechanisms, i.e.selective receptor-mediated uptake or synthesis in brain cells, should beexpected to cause deviations of the AFP index from unity.

2. Materials and methods

2.1. Fetal material

Six fetuses were obtained at abortion for socio-economic reasons, performedat the County Hospital, Hillerød, Denmark. Gestational age was determinedfrom menstruation history and confirmed by determination of crown–rumplength. The fetuses were obtained by prostaglandin-induced abortion. All fetuseswere found to be normal by autopsy. CSF was drawn with a syringe from thelateral ventricles immediately after abortion. No blood or brain tissue was visiblein the samples. Blood was sampled by cardiac puncture. Samples of serum andCSF were stored at 2 208C until analysis.

M. Christiansen et al. / Clinica Chimica Acta 291 (2000) 35 –41 37

2.2. Determination of AFP and albumin in serum and CSF

AFP and ALB concentrations in CSF were determined by rocket immuno-electrophoresis [14] in 1% agarose (HMW agarose) (Litex, Glostrup, Denmark)

2using rabbit anti-human AFP (DAKO A008) (1.3 ml / cm ), rabbit anti-human2albumin (DAKO A001) (0.4 ml / cm ) and a serum standard (DAKO X908) for

albumin (DAKO A/S, Glostrup, Denmark). Electrophoresis was performedovernight at 2 V/cm. Rockets were visualized by Coomassie-staining. A pool ofamniotic fluid calibrated against the WHO AFP standard 72/225 (StatensSeruminstitut, Denmark) was used as calibrator in the assay. Intra and interassayvariation were below 4% for both assays. The AFP assay was identical to theassay used for routine evaluation of amniotic fluid samples.

2.3. Crossed-immunoelectrophoresis (CIE) and concanavalin A crossed lectinaffinity electrophoresis (Con A–CLAIE) of AFP

The procedure was performed essentially as described in Ref. [15]. Theconcentration of AFP applied at 5 ml /hole was adjusted to approx. 20 MLU/l.

2Rabbit anti-human AFP (DAKO A008) (0.4 ml / cm ) was used in the second2dimension. Free concanavalin A (con A) (Pharmacia, Denmark) (300 mg/cm )

was added to the first dimension for con A crossed line affinity immuno-electrophoresis (Con A–CLAIE). The electrophoresis buffer was 0.02 mol / lTris–barbital, pH 8.6. Electrophoretic conditions for (1) dimension were 18V/cm for 3 h and 2 V/cm for 18 h for (2) dimension. After electrophoresisplates were dried and stained with Coomassie Blue.

2.4. Statistics

All correlations were performed by Spearman rank correlation, and com-parisons between groups by paired Wilcoxon’s test.

2.5. Ethics

Sampling of fetal material was approved by the Regional Scientific EthicsCommittee of Frederiksborg County, Denmark [16].

3. Results

Individual values of serum AFP and ALB and the concentrations of AFP andALB in CSF for the six fetuses are shown in Table 1 with gestational age and


Table 1AFP and albumin concentrations in serum and CSF, con A reactivity and AFP index of six fetuses

Case Gestational S-AFP CSF-AFP S-Alb CSF-Alb AFPno. age (weeks) (MIU/ l) (MIU/ l) (g / l) (g / l) index

a a(%NR) (%NR)

1 17 1550 (1.9) 160 (2.0) 4.30 0.49 0.912 17 1800 (2.1) 185 (1.9) 4.45 0.47 0.973 20 2150 (2.0) 150 (1.8) 5.75 0.43 0.934 22 1290 (1.5) 120 (1.9) 6.45 0.58 1.035 22 1550 (2.0) 85 (2.6) 8.50 0.58 0.816 23 2850 (0.9) 60 (1.1) 11.40 0.34 0.71

a %NR, percentage of AFP non-reactive with con A.

the AFP index calculated according to the formula given in the Introduction. Nosignificant correlation was found between gestational age and AFP index. Themean AFP index was 0.9060.11 (S.D.).

The CSF-serum concentration ratios of AFP and Alb were calculated andsignificant negative correlations were found for both ratios with gestational age(R(s) 5 2 0.88, p 5 0.048, for both). The corresponding ratios are shown in Fig.1 and there was a significant Spearman rank correlation (R(s) 5 1.0, p 5 0.0028).

Con A–CLAIE gave two peaks, one reactive with con A and one non-reactive. The mean fraction of con A non-reactive AFP was 1.7% for serumAFP and 1.9% for CSF AFP. The difference between the con A non-reactive

Fig. 1. Scatterplot of CSF-serum ratio of AFP (QAFP) as a function of the CSF-serum albuminration (QAlb). The Spearman rank correlation is given with the level of significance.


proportion of AFP in CSF and serum was insignificant. The mobility in agaroseelectrophoresis was identical for AFP from either CSF or serum.

4. Discussion

The present study is, to our knowledge, the first one to examine the AFPindex in fetuses. AFP index has previously been determined in adults in order todisclose the presence of intracranial metastases when treating non-seminomatousgerm cell tumors [17]. In one other study [1] three sets of values ofcorresponding concentrations of AFP and ALB in serum and CSF at the samegestational age as in our study were given. From these values one may calculatea mean AFP index of 0.8360.22 (S.D.), in good agreement with our findings.

As expected we found low concentrations of ALB in fetal serum and veryhigh concentrations of AFP in fetal serum and CSF compared with adult values[1,2,18]. This difference may be explained by a comparatively higher activepassage, by pinocytosis, of plasma proteins through endothelial cells into CSF[19]. The decrease in CSF-serum concentration ratios found for both proteinswith increasing gestational age is consistent with the known ontogeny of theblood–CSF barrier [19].

The low variation of the AFP index and its closeness to unity indicates thatAFP and ALB concentrations in CSF, at the examined time in development, arepredominantly determined by AFP and ALB concentrations in serum, making itunlikely that selective synthesis or receptor-mediated transport is of significantimportance. However, the value of the AFP index is consistently slightly lowerthan unity, which could be explained by the finding that neuronal AFP synthesisceases after the 14th week, whereas ALB is still synthesized by brain cells in the24th week as judged from the presence of ALB mRNA in fetal brain tissue [5].

AFP exhibits microheterogeneity dependent on the tissue of origin [13]. Thisis evaluated by con A affinity analysis, as AFP synthesized in fetal liver has alow fraction of con A non-reactive AFP as opposed to AFP synthesized in fetalyolk sac or fetal intestine and kidney. The very low con A-binding found in thisstudy supports that AFP found in CSF is derived from the liver via plasma. Theidentical mobility in agarose electrophoresis shows that AFP in fetal CSF is notsubject to major changes in charge, as is seen for other glycoproteins present inCSF, e.g. transferrin, a-2-macroglobulin, haptoglobin and prealbumin [20].

The biological function of AFP has not been established with certainty. AFPhas been ascribed an immunoregulatory [9] and an apoptosis inhibitory [21] rolein several systems, and has been found to promote growth and growth-factorresponses in other systems [10], whereas it inhibits cellular proliferation in yetother systems [22]. AFP has been found to bind certain eicosanoids, and hasbeen suggested to exert its biological functions by internalising eicosanoids [6].


The existence of specific receptors has been partly documented [23,24], anduptake of AFP by activated lymphocytes, and some malignant hemopoietic cells[25] and by, among others, neuroblastoma cells [26] has been demonstrated. Anautocrine loop, where cells synthesise AFP, which is then internalised by thesame cells has been described in HT-29 carcinoma cells [27]. As growth, growthregulation and apoptosis are very important processes in fetal development, it isof major interest to ascertain what role AFP plays in the development of thebrain. When studying variations in the concentration of AFP in CSF it should beremembered, that this concentration is dependent on the blood–CSF barrier, andfurthermore, is subject to gestational age variation. The use of an AFP index willeliminate the first of these confounding factors, whereas the use of gestationalage-dependent reference ranges for the AFP index also will eliminate the lastsource of variation.

Acknowledgements

Jette Rasmussen is thanked for excellent technical assistance.

References

[1] Bell JE, Fryer AA, Collins M et al. Developmental profile of plasma proteins in human fetalcerebrospinal fluid and blood. Neuropathol Appl Neurobiol 1991;17:441–56.

[2] Nørgaard-Pedersen B, Christiansen M, Jensen SP. alpha;-Fetoprotein in fetal pathology. In:Pasqualii JR, Scholler R, editors, Hormones and fetal pathophysiology, New York: MarcelDekker, 1992, pp. 369–412.

[3] Ali M, Balapure AK, Singh DR, Shukla RN, Sahib M. Ontogeny of alpha-fetoprotein inhuman fetal brain. Brain Res 1981;207:459–64.

[4] Dziegielewska KM, Saunders NR, Schejter EJ et al. Synthesis of plasma proteins in fetal,adult, and neoplastic human brain tissue. Dev Biol 1986;115:93–104.

˚[5] Møllgard K, Dziegielewska KM, Saunders NR, Zakut H, Soreq H. Synthesis and localizationof plasma proteins in the developing human brain. Dev Biol 1988;128:207–21.

[6] Naval J, Calvo M, Laborda J et al. Expression of mRNAs for a-fetoprotein (AFP) andalbumin and incorporation of APP and docosahexenoic acid in baboon fetuses. J Biochem1992;111:649–54.

[7] Schachter BS, Toran-Allerand CD. Intraneuronal a-fetoprotein and albumin are not syn-thesized locally in developing brain. Dev Brain Res 1982;5:93–8.

[8] Fishman PS, Farrand DA, Krisst DA. Penetration and internalization of plasma proteins inthe human spinal cord. J Neurol Sci 1991;104:166–75.

[9] Laan-Putsep K, Wigzell H, Cotran P, Gidlund M. Human alpha-fetoprotein (APP) causes aselective down regulation of monocyte MHC class II molecules without altering otherinduced or noninduced monocyte markers or functions in monocytoid cell lines. CellImmunol 1991;133:506–18.


[10] Keel BA, Eddy KB, Cho S, May JV. Human alpha-fetoprotein purified from amniotic fluidenhances growth-factor mediated cell proliferation in vitro. Mol Reprod Devel 1991;30:112–8.

˚[11] Mølgard K, Jacobsen M. Immunohistochemical identification of some plasma proteins inhuman embryonic and fetal forebrain with particular reference to the development of theneocortex. Dev Brain Res 1984;13:49–63.

[12] Felgenhauer K. Protein size and cerebrospinal fluid composition. Klin Wschr 1974;52:1158–64.

[13] Christiansen M, Ishiguro T, Høgdall C, Nørgaard-Pedersen B. Alpha-fetoprotein. In: BallestaAM et al., editor, Updating on tumor markers in tissues and in biological fluids. basic aspectsand clinical applications, Torino: Edizioni Minerva Medica, 1993, pp. 245–69.

[14] Axelsen NH, Bock E. Electroimmunoassay (Rocket immunoelectrophoresis). Scand JImmunol 1983;17(Suppl 10):103–6.

[15] Nørgaard-Pedersen B, Toftager-Larsen K, Philip J, Hindersson P. Concanavalin A reactivitypattern of human amniotic fluid APP examined by crossed affino-immunoelectrophoresis. Adefinite test for neural tube defect? Clin Genet 1980;17:355–62.

[16] Hansen M, Høgdall CK, Kryger-Baggesen N. Cobalamin binding proteins in the humanfetus. Scand J Clin Lab Invest 1989;49(Suppl 194):23–6.

[17] Marrink J, Sleijfer DTH, Kok AJ, Ockhuizen TH, Schraffordt Koops H. AFP index, aparameter for the detection of brain metastases in non-seminomatous germ cell tumours ofthe testis. Eur J Cancer Clin Oncol 1984;20:1207–8.

[18] Adinolfi M, Haddad SA. Levels of plasma proteins in human and rat fetal CSF and thedevelopment of the blood–CSF barrier. Neuropaediatrie 1977;8:345–53.

˚[19] Møllgard K, Saunders NR. The development of the human blood–brain and blood–CSFbarriers. Neuropathol Appl Neurobiol 1986;12:337–58.

[20] Stibler H. The normal cerebrospinal fluid proteins identified by means of thin-layerisoelectric focusing and crossed immunoelectrofocusing. J Neurol Sci 1978;36:273–88.

[21] Laderoute MP. A new perspective on the nature of the cancer problem: Anti cellularsenescence. Mol Carcinog 1994;10:125–33.

[22] Mizejewski GJ. New insights into alpha-fetoprotein structure and function: potentialbiomedical applications. In: Mizejewski GJ, editor, Alpha-fetoprotein and congenitaldisorders, New York: Academic Press, 1985, pp. 5–34.

[23] Suzuki Y, Zeng CQ, Alpert E. Isolation and partial characterization of a specific alpha-fetoprotein receptor on human monocytes. J Clin Invest 1992;90:1530–6.

[24] Laderoute M, Willans D, Wegmann T, Longenecker M. The identification, isolation andcharacterization of a 67 kilodalton PNA-reactive autoantigen commonly expressed in humanadenocarcinomas. Anticancer Res 1994;14:1233–45.

[25] Esteban C, Trojan J, Macho A, Mishal Z, Lafarge-Frayssinet C, Uriel J. Activation of analphafetoprotein / receptor pathway in human normal and malignant peripheral blood mono-nuclear cells. Leukemia 1993;7:1807–16.

[26] Hajeri-Germond M, Naval J, Trojan J, Uriel J. The uptake of alpha-foetoprotein by C-1300mouse neuroblastoma cells. Br J Cancer 1985;51:791–7.

[27] Esteban C, Geuskens M, Uriel J. Activation of an alpha-fetoprotein (AFP) / receptor autocrineloop in HT-29 human colon carcinoma cells. Int J Cancer 1991;49:425–30.

Documents

Alpha-fetoprotein in human fetal cerebrospinal fluid