Jan 28 Embryology Lecture: (chapter 4 etc.) Biology 2005 Albert Harris   Important methods used in embryological research: I) Descriptive anatomy of embryos & Serial histological sections Tables of normal stages made for many species (much descriptive work 1870-1920 to confirm phylogenies) (For example, H. V. Wilson's monograph on the embryology of the Black Sea Bass: Earned him his professorship at UNC. II) Tissue grafting : Organ development at abnormal locations? Chorio-allantoic grafting in chicken embryos; early organ precursors can be grafted to extraembryonic membranes of other chicken embryos; blood vessels develop into grafts Tissue culture ; cut groups of cells out of embryos, or from tumors in adults, and grow them in sterile nutrient medium. This method was invented by Ross Harrison, in about 1907, specifically in order to prove that nerves are formed by locomotion of nerve growth cones, forming a strand of cytoplasm (= axon). Before that, most people thought nerve fibers were secreted by surrounding tissues, or some other theories. Probably they were misinterpreting the formation of myelin sheaths. III) Tracing cell lineages ; and making "fate maps " Vital staining , with nile blue sulfate or other non-toxic dyes; injection of fluorescent dextran; horseradish peroxidase enzyme. graft tissues from other species with distinguishable nuclei Japanese quail grafted to chicken embryos (also frog mutants with more or fewer nucleoli than normal) J E. Sulston shared the 2002 Nobel Prize, partly for tracing the entire cell lineage of Caenorhabditis elegans IV) Selective poisons to block certain molecular processes (& test how this changes embryonic development) actinomycin D blocks gene transcription (RNA synthesis) (despite what its name sounds like, it is NOT an actin inhibitor!) puromycin & cycloheximide block translation of messenger RNA cytochalasin blocks actin assembly. colchicine , nocodazole, etc. block microtubule assembly. Many specific inhibitors of particular kinases & phosphatases V) Genetic analysis of mutant organisms : Find mutant strains of animals (or plants, etc.); (Temperature sensitive mutants are especially useful); Using genetic criteria to show that a single protein was changed. Studying combinations of different mutations in the same animal. Logically analogous to use of selective poisons, except that greater specificity and unlimited range of application (doesn't depend on whether a poison exists); Try to figure out how the non-mutant proteins interact together normally by seeing what goes wrong when the proteins are abnormal. Also notice that genes themselves are usually named after the effect produced by mutation of the gene; a gene needed for heart development would probably be named the "heartless gene" etc.) VI) Transgenic organisms : insertion of genes, by various tricks. Introduction of reporter genes , "downstream" of promoter regions. Transgenic = "genetically modified " in news articles & TV Genes for rare needed proteins can be "transformed" into cows, so that their milk will contain the needed protein, etc. GFP (green fluorescent protein is a certain sequence of amino acids, that will make any (?) protein fluoresce green in UV light, if the DNA coding for that amino acid sequence is inserted into its gene) "Knock-out experiments " in which a certain gene is selectively inactivated by homologous recombination with a DNA sequence similar to its own. Infection of cells with viruses into which animal genes have been added. VII) Use of specific antibodies (most often monoclonal antibodies) * In what patterns do they bind (shows where in the embryo a given gene is expressed) * What processes do they block? (antibody binding can block the function of all copies of one specific protein) VIII) in situ hybridization of RNAs of genes of special interest. (Map where in the developing embryo these genes are expressed) IX) Nuclear transplantation cell fusion, cloning Dolly the sheep etc. X) Don't confuse this with making chimeric mammals by fusing early embryos, even of different species. "tetraparental mice " etc. XI) DNA and RNA identification methods: The underlying principle for several methods is this: Suppose you are interested in finding the location of some messenger RNAs coded by certain genes; if they contain regions with the base sequence AUUCCGGCGAUA..(that I just typed at random). This will pair with UAAGGCCGCUAU.. . Single stranded nucleic acids will spontaneously "find" and bind to any other single-stranded nucleic acid (RNA or DNA!) that has regions with complementary sequences. So if you can make, buy or borrow some highly radioactive, or otherwise labeled, single stranded DNA (or RNA) then you can use this labeled stuff as a "probe" to find out the location of RNA (or DNA) with the complementary base sequence. By random diffusion, and bumping into it, the probe will find and bind tightly to the target sequences. When you find RNA in fixed tissue sections, that's an "in situ " When you separate nucleic acids by electrophoresis , then "blot" them onto a material to which they will bind & not diffuse then it's "a southern " if you are locating DNA and it's "a northern " if you are locating RNA. (I hope everyone realizes the origin of these names? A Dr. Southern invented the method for blotting electrophoresis gels, and probing them with complementary sequences. Therefore, when other people used the same basic method to study RNAs, they called their method "A northern". Imagine the helicopter being invented by the Left Brothers. Blotting proteins from gels, and probing them with antibodies is called "the western" technique. If it were worth while to electrophorese polyshaccharides, and blot them, and probe them with lectins, or something, then probably they will call that "an Eastern" XII) Mathematical and computer simulations Most people seem to misunderstand the need for simulations. For example, on page 5, our textbook says "A fourth method...approach is mathematical modeling , which seeks to describe developmental phenomena in terms of equations" That is what most embryologists think; but I think it's wrong. I think modeling is the only practical method to discover (and prove) the testable predictions of theories, when the theories are even slightly complicated. Remember Turing's wave-generating reaction diffusion systems: Then can you figure out how it should change these waves if an experimenter added different amounts of one of the diffusing chemical "morphogens"? Suppose that water were slowly trickled past a tissue in which such a reaction-diffusion mechanism was making waves? Suppose that flakes of impermeable mica were inserted into tissues while a reaction-diffusion mechanism was acting? In each case, how would you know what Turing's theory would predict about how the color pattern should be changed? How could you invent experiments that would give different results depending on which of 3 or 4 alternative classes of pattern-generating mechanisms is the true cause of stripes, etc. The purpose is NOT just to produce something realistic-looking. The purpose is to test whether theories accurately predict what cells actually do; and also (by experimental changes in programs) to find experimentally testable differences between the predictions of alternative theories. The purposes of the "cellular automata" exercise were to learn: a* Simple rules can produce complicated patterns. b* Until you try a set of rules, it's hard to know what will happen. c* Seeing such a program in operation, no one is smart enough to "see" what rules are being obeyed. d* By experimenting with different combinations of rules, you could discover which rules can produce a given result. Cellular automata are so abstract, people doubt their relevance. Imagine a cellular automaton in which each square has a number. 123444333221 112343332211 111233222110 This could represent a diffusion gradient. 011122211100 001111110000 To simulate diffusion , for each square add up the numbers in the surrounding 8 squares, and divide by 8. When this average is more than the number in the center square, then add 1 to the number in the center square When the average of the neighbors is less than the center number, then decrease the center number. It is easy to program computers to obey simple rules; and computers love to do the same thing, over & over. Better simulations of diffusion have the center number change different amounts, depending on how different it is from the average of the surrounding numbers. The degree of change is the diffusion constant. But you can also simulate non-linear diffusion. Think about this : 1* If there were a class in which each student always sat in the same seat, every day, 2** And these students are conformists, who get their hair cut to be about the same length as the hair of the people sitting closest to them, in the adjacent seats 3*** then if somebody in the back had long hair, and somebody in the front left corner had a crew cut.... The results would resemble a diffusion gradient. Developing embryos, and regenerating planaria, etc. often contain gradients of certain properties, or of behavior. Almost everyone concludes "Oh, there must be chemical diffusion gradients inside them; and the diffusing chemicals must control all the properties that vary gradually with location!" What else could it be? It could be that each cell detects properties of its neighbors, and tends to change toward the average (like the hair-cuts) By what experiments could you prove that one idea is true and that other alternatives are false? Not so easy! A good method is to write a different computer program for each possible hypothesis, that predicts what would happen if that theory were correct. One advantage of this approach is that programming languages are the closest thing to "a notation " for abstract possibilities. (like music notation represents sounds) Just like drawing a picture of something gives you a better idea what it looks like, and you notice details, writing each program clarifies what a theory really "says". Then you experiment with each different program to find out what it predicts should happen in response to lots of different conditions that might occur, or situations that you might impose on the system. What you want to find is situations in which the different theories make different predictions about what ought to happen in response to some experimental manipulation you can impose. Do the theories (simulation programs) make different predictions about what ought to happen if you turn the embryo upside down? Then that's your experiment! Just because somebody happens to be the first person to propose a certain hypothesis doesn't give them any greater ability to know what the theory predicts in different experimental situations. On the other hand, the trouble with mathematicians is that they want to apply familiar (simple) rules, like making diffusion rates linear functions of the local steepness of a diffusion gradient, "D'arcy's Law" or making solids obey "Hooke's Law" or fluids obey "Newton's Law". Almost all those "Laws" are just idealized simplifications. They don't have to be true of biological materials, and not even for simple materials. Disobeying such laws is often the main cause of phenomena. Consider the "Ideal Gas Law"; PV=nRT (don't worry if you forgot) Liquids don't exist according to this idealization. So don't use it to try to predict whether it will rain.

 

back to syllabus

back to index page