The effect of replacing bovine serum albumin (BSA) in a simple defined medium (KSOM) with polyvinyl alcohol. (PVA) and/or amino acids on the percentages of ...
Human Reproduction Update 1997, Vol. 3, No. 2 pp. 125–135
European Society for Human Reproduction and Embryology
Polyvinyl alcohol and amino acids as substitutes for bovine serum albumin in culture media for mouse preimplantation embryos John D.Biggers1,3, Michael C.Summers1,2 and Lynda K.McGinnis1 1Department of Cell Biology, Harvard Medical School, 240 2Present address: Fertility Center of New England, 20 Pond
Longwood Avenue, Boston, MA 02115, USA Meadow Drive, Reading, MA 01867, USA
TABLE OF CONTENTS Introduction Materials and methods Results Discussion Acknowledgements References
125 126 128 133 135 135
The effect of replacing bovine serum albumin (BSA) in a simple defined medium (KSOM) with polyvinyl alcohol (PVA) and/or amino acids on the percentages of mouse zygotes that develop to at least the blastocyst stage and that hatch at least partially or completely is reported. Blastocysts could form when BSA was replaced with only PVA, but at a moderately reduced rate; however, partial hatching, and hence complete hatching, were severely impaired when BSA was replaced with only PVA. The substitution of BSA with amino acids alone resulted in a high rate of blastocyst formation and moderate impairment of hatching. The addition of PVA to BSA-free KSOM supplemented with amino acids had no extra effect. BSA had significant effects when added to BSA-free KSOM supplemented with amino acids. The BSA caused a significant increase in the rate of partial hatching, and may even have had a small effect on the rate of blastocyst formation. The results also showed that glucose, at a high concentration of 5.56 mM, does not inhibit the development of mouse zygotes to hatched blastocysts when cultured in KSOM supplemented with amino acids. Key words: amino acids/embryo culture/glucose/ polyvinyl alcohol Introduction The results reported here are concerned with whether polyvinyl alcohol (PVA) is an adequate replacement for bovine 1To
whom correspondence should be addressed
serum albumin (BSA) in media that support the development of mammalian pre-implantation embryos. These studies were motivated by the need to formulate chemically defined media for the cultivation of pre-implantation embryos that are serum-free or free of proteins with undefined functions. BSA has been incorporated in all media designed to support the development of mouse pre-implantation embryos in vitro (reviewed by Biggers, 1987, 1993; Bavister, 1995). The accepted need for macromolecules in these media stems from early studies which suggested that the development of 8-cell mouse embryos in vitro to blastocysts requires the incorporation of egg white (Hammond, 1949), and the subsequent observation that the essential components in egg white are non-dialysable (Whitten, 1956). The use of egg white was dropped when it was also shown that it could be replaced with crystalline BSA (Whitten, 1956). BSA may have nutritional functions, as a source of fixed nitrogen, in culture media that support pre-implantation development. One possible function is the provision of free amino acids as the result of hydrolysis of the protein. Historically the need to substitute a non-protein macromolecule for BSA first arose in the design of experiments to determine the requirement for exogenous amino acids in mouse pre-implantation development (Brinster, 1965). In a search for a synthetic substitute for BSA, Brinster (1965) found that polyvinylpyrrolidone (mol. wt 150 000; PVP150), acacia, dextran and Ficoll were acceptable substitutes, but not methyl celluloses. However, he was unable to demonstrate a nutritional role for BSA in the development of the 2-cell mouse embryo to the blastocyst stage in vitro. He speculated that ‘some of the beneficial effects of BSA and certain amino acids may be due to their action as chelating agents, regulators of oxidation reduction potential, cell surface protectors, or enzyme protectors.’ Later work drew attention to the fact that BSA is chemically a very variable product, whose composition depends on the degree to which different small molecules remain bound to the
126
J.D.Biggers, M.C.Summers and L.K.McGinnis
macromolecule during its preparation. Thus extensive literature arose describing the variable effects produced by different BSA preparations in pre-implantation embryo culture media (Kane, 1983; McKiernan and Bavister, 1992). Brinster (1965) favoured the use of PVP150 as a substitute for BSA in his work on the need of 2-cell mouse embryos for a fixed nitrogen source in vitro. He emphasized the beneficial effect of this polymer in facilitating handling of the embryos, presumably by influencing the physico-chemical properties of the medium. Later, an alternative PVA was also found to be an acceptable substitute for BSA in media used for fertilizing hamster ova in vitro (Bavister, 1981). Since the publication of these results, the polyvinyl polymers PVP and PVA have been commonly used as substitutes for BSA in media for the culture of mammalian pre-implantation embryos, with PVA being the preferred substitute in recent times because of the reported toxicity of PVP (Ashwood-Smith and Warby, 1971). Nevertheless, results obtained with BSA-substituted media have not been consistent. In experiments using PVP150 as a substitute for BSA, Brinster (1965) found that outbred 2-cell mouse embryos required a fixed nitrogen source to develop into blastocysts. This source could be in the form of BSA, an amino acid mixture simulating the composition of hydrolysed BSA (Brinster, 1965) or glutathionine (Brinster, 1968). In contrast, Cholewa and Whitten (1970), also using PVP as a substitute for BSA, were unable to show a need by 2-cell F1 hybrid mouse embryos for a fixed nitrogen source. Moreover, outbred 8-cell mouse embryos did not require a fixed nitrogen source to develop into blastocysts (Brinster and Thomson, 1966). Recently, interest has been renewed on the need for amino acids in media used to culture mouse pre-implantation embryos (reviewed in Gardner, 1994; Bavister, 1995). Gardner and Lane (1994) and Lane and Gardner (1993) reported that the pre-implantation development of 1-cell F1 hybrid mouse embryos was enhanced when a mixture of 20 amino acids was added to medium MTF (Gardner and Leese, 1990). Ho et al. (1995) have reported a similar enhancement using 1-cell embryos from C57BL/6J inbred mice when amino acids were added to KSOM (Lawitts and Biggers, 1993). The rate of development of blastocysts increased, as well as the rate of hatching and the blastocyst total cell number. The results of Gardner and Lane (1993) and Ho et al. (1995) are of particular interest because the addition of free amino acids enhanced development, even in the presence of BSA in the medium (4 mg/ml in MTF; 1 mg/ml in KSOM). We have therefore compared the effect of adding free amino acids to KSOM, in which the macromolecule is either BSA or PVA, on the development of
outbred mouse zygotes to the hatching stage of blastocyst development. Our results also suggest that glucose does not inhibit the enhanced effects of amino acids and albumin on the development of the pre-implantation mouse embryos. These results are in contrast to the reported inhibitory effects of glucose when added to medium CZB (Chatot et al., 1989).
Materials and methods Animals
Outbred CF1 female mice, 6-8 weeks old, initially from Charles River Laboratories (CRL) (Wilmington, MA, USA) and subsequently from Harlan Sprague Dawley (HSD) (Indianapolis, IN, USA) were used in this work. The source of the mice was changed when the strain from CRL became unavailable. Separate tests showed that the HSD mice were adequate substitutes. The females were mated to F1 hybrid B6D2F1 males, 2–11 months old, from CRL. Females were stimulated with 5 IU pregnant mare’s serum gonadotrophin (Sigma Chemical Company, St Louis, MO, USA) and superovulated with 5 IU human chorionic gonadotrophin (HCG; Sigma) 48 h later. All females were killed 18–22 h post-HCG and any ova present recovered. Only those embryos showing two pronuclei were allocated to the experiments. Thus the embryos used were outbred prior to and after zygotic activation. Table I. Composition of medium designated as KSOMb (KSOM without bovine serum albumin) Component
Concentration (mM)
NaCl KCl KH2PO4 MgSO4 Lactate Pyruvate Glucose NaHCO3 CaCl2 L-Glutamine EDTA
95 2.5 0.35 0.2 10.0 0.20 0.20 25 1.71 1.0 0.01
Media
The base medium used was KSOM minus the BSA, denoted KSOMb (Table I). It was supplemented with several macromolecules, which are denoted by superscripts. For example, KSOMbPVA indicates that the medium was supplemented with PVA. The base medium was also supplemented with a mixture of 19 amino acids (Table II), and is denoted KSOMbAA.
Polyvinyl alcohol and mouse embryo culture
127
Table II. Concentrations of amino acids added to KSOM and KSOMb to give KSOMAA and KSOMbAA respectively Amino acid
Concentration (mM)
Amino acid
Concentration (mM)
L-Alanine–HCl
0.05 0.3 0.05 0.05 0.05 0.05 0.05 0.1 0.1 0.2
L-Leucine
0.2 0.2 0.2 0.05 0.1 0.05 0.05 0.2 0.025
L-Arginine–HCl L-Asparagine–H2O L-Aspartic
acid
L-Cystine L-Glutamic
acid
Glycine L-Histidine–HCl–H2O L-Tyrosine L-Valine
All culture media were formulated from KSOM with 1.0 mg/ml BSA (Fraction V; Sigma, Cat. no. A9647, lot no. 15H0672) or 0.1 mg/ml PVA (10 kDa; Sigma; P-8136). The KSOM used in these experiments was prepared as a 2× solution (complete KSOM without the CaCl2, BSA or PVA), divided into aliquots in 50 ml culture tubes and frozen at –70°C for up to 3 months. Calcium chloride, BSA and PVA were prepared as 1 M stocks and either frozen (CaCl2 and BSA) or stored at 4°C (PVA). At 1 day before embryo collection, one aliquot of 2× KSOM stock was thawed and supplemented with CaCl2 and H2O. All chemicals used in the preparation of KSOM were from Sigma. Non-essential (NEAA) and essential (ESAA) amino acids (Eagle, 1959) were added in the concentrations used by Ho et al. (1995). The NEAA and ESAA stock amino acid solutions were purchased from Gibco BRL (Life Technologies Inc., Grand Island, NY, USA).
Embryo culture
Ova were flushed from the oviduct using a modification of the flushing medium described by Lawitts and Biggers (1993), called FHM, in which the BSA was replaced with 0.1 mg/ml PVA. The zygotes were then washed in 0.3 mg/ml hyaluronidase in the modified FHM to remove cumulus cells. At this step, the ova with two pronuclei were selected for culture. Embryos were cultured for 5 days (144 h post-HCG) in groups of 12 per 50 µl droplet of medium overlayered with embryo-tested light mineral oil (Sigma; M8410). The cultures were incubated at 37°C in modular incubator chambers (Billups-Rothenberg Inc., Del Mar, CA, USA), which were gassed with a mixture of 5% O2, 6% CO2 and 89% N2 (Lawitts and Biggers, 1993). Culture plates (60 mm non-tissue culture-treated; Corning Inc., Corning, NY, USA) were prepared 1 day before embryo collection and equilibrated in the module overnight.
L-Isoleucine L-Lysine–HCl L-Methionine L-Phenylalanine L-Proline
l-Serine L-Threonine L-Tryptophan
Embryo evaluation
Embryos were observed at ×100 on a warmed microscope stage (35°C; Wild dissecting microscope) and graded for stage of development, including compaction, blastocoel formation and hatching, at 96, 120 and 144 h post-HCG. Cell counts
Embryos were fixed 144 h post-HCG in 3% formaldehyde for 15 min at 37°C. After fixation, nuclei were stained with the fluorochrome, Hoechst 33258 (1 µg/ml) in Dulbecco’s phosphate-buffered saline for 15 min at room temperature. Groups of one to four blastocysts were mounted onto glass slides and covered with a mounting medium (50% glycerol, 50% sodium azide and 1 µg/ml Hoechst 33258). Stained embryos were covered with a glass coverslip and sealed with clear nail polish. Nuclei were counted at ×40 on an inverted Zeiss epifluorescence microscope with a 365 nm band pass excitation filter and a 420 nm long pass barrier filter. Statistical methods Experimental design
Five experiments were performed using randomized block experimental designs. Experiments 1 and 2 were preliminary experiments, the results of which led to the design of the major experiments (nos. 3–5). In all experiments the experimental unit was the set of ova, containing two pronuclei, that was randomly allocated to each drop. In experiments 3, 4 and 5, each drop contained 12 zygotes. Each block in each experiment corresponded to a single replicate of the experiment. Within each replicate, the experimental units were allocated at random to the treatments being compared in each experiment. The number of experimental units allocated to each treatment within a replicate was the same, and ranged from one to four depending on the number of zygotes collected and the number of treatments. In experiments 1 and 2, the number of zygotes in a
128
J.D.Biggers, M.C.Summers and L.K.McGinnis
drop, and the number of drops allocated to a treatment, varied between replicates. Developmental responses
In experiments 3, 4 and 5, the embryos were cultured for 144 h post-HCG. At 96, 120 and 144 h post-HCG the embryos were classified into one of a series of developmental states: pre-blastocyst, unhatched blastocyst, partially hatched blastocyst and completely hatched blastocyst. Thus statistically the observations are repeated measurements which, at each time, are a set of ordinal categorical responses (McCullagh and Nelder, 1989; Clogg and Shihadeh, 1994). For the statistical analysis, the observations were re-expressed as cumulative totals, i.e. the numbers of embryos developing to at least the blastocyst stage, at least the partially hatched blastocyst stage and at least the completely hatched blastocyst stage. Since the group size was constant throughout the data sets (n = 12), each of these cumulative sums was transformed before being used in analyses of variance (ANOVA) using the two-term inverse sine transformation proposed by Laubscher (1961): t4 = n1/2sin–1(x/n)1/2 + (n + 1)1/2sin–1[(x + 3/4)/(n + 3/2)]1/2, where x = the number of responders (blastocysts) and n = the group size. The advantage of this transformation is that it is more stable for proportions 0.9, which occur frequently in our data. The theoretical variance of this distribution approaches 1 as n increases. Repeat measurement ANOVA
Data from experiments 3, 4 and 5 were analysed by repeated measures ANOVA using the transformed data for each of the cumulative sums and the NCSS 6.0 Statistical Program (Jerry L.Hintze, Kaysville, UT, USA) [see Diggle et al. (1994) and Kshirsagar and Smith (1995) for the theory and assumptions of this analysis]. The statistical analysis is in two parts, each with separate error terms. One part analyses the effects of the experimental factors and their interactions [error (a)], while the other analyses the effect of time and the interactions between time and the experimental factors [error (b)]. The mean square for replicates was significant in several of the analyses. It is assumed that there were no large differences between the treatment effects within the replicates of each experiment. This assumption was checked graphically. Furthermore, the error mean squares used in the tests of significance include the variation attributable to the differential effects of the treatments between replicates. Data presentation
The raw data from the experiments are not shown because of their extensive size. The data can be obtained from the
first author (J.D.B.). Instead, the relevant information is summarized by the ANOVA of the transformed data which identify the significant effects. These significant effects are then illustrated graphically by plotting the relevant means of the transformed data after retransformation back to the percentage scale. This retransformation was performed using an iterative program written in QuickBasic. In all experiments, partial hatching and complete hatching had not started at 96 and 120 h post-HCG respectively. The data for these stages were therefore not included in the graphical and statistical analyses. Thus the number of repeat measurements for each response category, shown in Figures 1–3, diminishes from one category to the next. Other statistical procedures
The preliminary results obtained in experiments 1 and 2 consisted of observations on the number of blastocysts formed (a single categorical variable) in drops that did not always contain the same number of embryos. The results were analysed after logistic transformation by a logistic regression analysis (Hosmer and Lemeshow, 1989) and the LogXact-Turbo Computer Program (Cytel Software Corporation, Cambridge, MA, USA). Experiment 2 also provided data on the number of cells that developed in the blastocysts formed. Data on cell counts were analysed by a two-way ANOVA using the NCSS 6.0 Statistical Program (Jerry L.Hintze). All effects with a probability value ≤0.05 were considered to be statistically significant.
Results Preliminary observations (experiments 1 and 2)
In experiment 1 the effects of adding either BSA (1.0 mg/ ml) or PVA (0.1 mg/ml) to KSOMbAA before or after freezing the medium were compared. Thus, there were four experimental treatments. Three replications were performed. Two of the replicates contained 10 embryos in each drop and the third replicate contained 11 embryos per drop. The embryos were cultured for 120 h post-HCG and the number of blastocysts recorded. A logistical regression analysis showed no significant replicate × treatment interaction, so the data have been pooled over replicates. The results are summarized in Table III. Although there was a suggestion that the percentage of blastocysts that developed was less when PVA was substituted for BSA, and through freezing of the medium, the differences were not statistically significant. Analyses of the total cell number in blastocysts that developed under the four different conditions showed no statistically significant differences (Table III).
Polyvinyl alcohol and mouse embryo culture
129
Table III. Results of experiment 1: the number of zygotes developing into blastocysts and the blastocyst cell number, after freezing and thawing of KSOMbAA containing either bovine serum albumin (BSA; 1 mg/ml) or polyvinyl alcohol (PVA; 0.1 mg/ml) Treatment
No. of
Blastocyst cell counta
Blastocysts
n
Number ± SEMc
zygotes
n
%
Pb
BSA
31
28
90.3
–
18
88.30 ± 6.22
PVA
31
24
77.4
0.301
18
86.30 ± 6.22
BSA (frozen)
31
23
74.2
0.182
16
87.80 ± 6.59
PVA (frozen)
31
24
77.4
0.301
18
91.30 ± 6.22
KSOM = base medium; KSOMb = KSOM minus BSA; KSOMbAA = KSOMb supplemented with a mixture of 19 amino acids. aProbability that the treatment means are different = 0.953. bFisher’s exact test. cCalculated from the analysis of variance pooled error mean square: 695.72, df = 64. Table IV. Results of experiment 2: the number of zygotes developing into blastocysts and the blastocyst cell counts, when cultured in KSOMbAA containing various macromolecular supplements Macromolecular supplement
No. of
Blastocysts
Blastocyst cell count %
Pa
n
Number ± SEMb
P
44
78.6
–
12
88.80 ± 6.69
–
56
42
75.0
0.823
11
55.9±6.99
0.001
Polyvinyl alcohol (0.1 mg/ml)
56
42
75.0
0.823
14
80.50 ± 6.20
0.364
Polyvinyl alcohol (1.0 mg/ml)
56
35
62.5
0.097
14
77.7 ± 6.20
0.226
Ficoll (1.0 mg/ml)
56
34
60.7
0.064
13
68.50 ± 6.43
0.033
zygotes
n
BSA (control)
56
None
KSOMbAA: see Table III. aFisher’s exact test. bCalculated from the analysis of variance pooled error mean square: 537.63, df = 54.
In experiment 2 the effects of adding BSA, PVA or Ficoll to KSOMbAA were compared. The five treatments were KSOMAA containing 1.0 mg/ml BSA (positive control), KSOMbAA with no BSA (negative control), KSOMAA containing either 0.1 or 1.0 mg/ml PVA and KSOMAA containing 1.0 mg/ml Ficoll. Three replications were performed. The first replicate contained 12 embryos in a single drop, the second replicate contained 12 embryos in each of two drops, and the third replicate contained 10 embryos in each of two drops. The embryos were cultured for 120 h postHCG and the number of blastocysts recorded. The statistical analyses of the separate replicates and examination of the replicate × treatment interactions showed that the data were sufficiently homogeneous to be pooled (data not shown). The results of the pooled data are summarized in Table IV. After 120 h post-HCG in culture there were no significant differences in the percentage of embryos reaching the blastocyst stage, although the percentage was close to being significantly reduced in the group containing the higher concentration of PVA. The response to Ficoll was particularly variable, but overall the result was close to significance (P = 0.06). The total numbers of cells in the blastocysts that developed were significantly less in the group cultured without a macromolecular component (P =
0.001). In contrast, the total numbers of cells in the blastocyst were unaffected by any of the macromolecular additions, with the exception of Ficoll in which the total cell number was reduced (P = 0.03). Effects of an amino acid supplement to KSOM and KSOMb PVA
In experiment 3 the effects of amino acid supplements to KSOM, and KSOM in which BSA was replaced with PVA, were examined. The experiment was a randomized block design with the treatments arranged in a 2 × 2 factorial array. The factors were BSA (1.0 mg/ml) versus PVA (0.1 mg/ml) and the presence or absence of the amino acids listed in Table II. The numbers of unhatched, partially hatched and completely hatched blastocysts were recorded at 120 and 144 h post-HCG for each of the four treatments. Five replications were performed. The results of adding the amino acids to KSOM or to KSOMb, at the concentrations shown in Table II, demonstrated clearly that there were large differences in the percentage of embryos that developed at the times of observation (120 and 144 h) at least to the zona-enclosed blastocyst, the partially and completely hatched blastocyst (Figure 1 and Table V).
130
J.D.Biggers, M.C.Summers and L.K.McGinnis
Figure 1. The effects of bovine serum albumin (BSA) or polyvinyl alcohol (PVA) and the presence or absence of amino acids (AA) in KSOMb (base medium minus BSA) on the percentages of zygotes that develop to at least (A) zona-enclosed blastocysts, (B) partially hatched blastocysts and (C) completely hatched blastocysts at different times of culture after human chorionic gonadotrophin injection (experiment 3).
Table V. Results of experiment 3: analysis of variance tables of the data shown in Figure 1 Variation
df
At least blastocysts Mean square
Replicates Amino acids (A) Macromolecules (M) AM Error (a) Time (T) AT MT AMT
4 1 1 1 60 1 1 1 1
11.779 7.368 15.447 6.705 3.521 5.277 0.366 0.308 0.668
Error (b)
64
0.835
P 0.0113 0.141 0.0347 0.160 – 0.0208 0.535 0.569 0.402 –
At least partially hatched
Completely hatched
Mean square
P
Mean square
P
19.983 27.236 33.245 6.735 2.017 83.068 5.880 0.208 3.477
