The double helix can actually take 3 different forms. And the one that's most common and was the structure predicted by Watson and Crick is the B form. But let's go ahead and talk about all of the forms. So here below in the image, we have our A, B, and Z form and I'm sure the B looks the most familiar to you. Now, it's worth noting that the A and B forms are both right-handed helices but the A form is more condensed or more squished down than the B form. So actually, in this image that we have here, each strand has the same number of bases and as you can see, the A form occupies less space than the B form. It's a little more condensed. And it turns out that double stranded RNA actually takes the A form. And as I said, most DNA is found in the B form. Most double-stranded DNA is found in the B form. The Z form is a left-handed helix unlike A and B. And the Z form is generally found near regulatory sequences. And basically, you'll have what are called the BZ junctions which are areas where the helix switches between the B and the Z forms. And more or less what happens is the nucleotides flip out from the strands. Remember the whole syn/anti conformation thing. The nucleotides flip out from the strands and they cause the actual helix to switch direction from left-handed to right-handed. And this actually puts an abrupt change in the shape of the DNA. And this is really important because this sort of noticeable shape difference is used by cells, is exploited by cells to determine where regulatory sequences are. And you'll see this pattern a lot in biology where basically, the actual physical shape and structure of DNA is used to identify various things. Now, also important to note about the structure of these different forms is what they look like if you were to, stare down them. So these images below are if you were looking I'm going to draw like an eye here. If you were looking down on the structure and if you were looking down in the A structure, you'd see that there's actually an opening in the center of it. It has a hollow space inside of it, whereas there's no opening in the B and the Z forms right here. No opening in those structures. And, that is going to be pretty important in terms of our DNA because single strands of DNA are held together not only by hydrogen bonds but also by these hydrophobic stacking forces. And these stacking forces result from the fact that the bases are actually pretty hydrophobic and as you can see in Let me actually change colors here. As you can see in the center, here all of our bases which are those blue shapes we see are mostly blue. There's some other colors in there but those mostly blue shapes we see in the center, they're all stacked on top of each other, right? So they're having these, hydrophobic, stacking forces between them that actually help maintain that structure, that B form structure of DNA. So, zooming out on that image a little bit more, if you think about the outside, we have all those phosphate groups, right? And we actually have because, you know, DNA is found in cells, it's in an aqueous environment. So there's water around DNA that helps stabilize its shape and then there's those hydrophobic stacking forces inside DNA to also help stabilize its shape. Alright, let's flip the page.
- 1. Introduction to Biochemistry4h 34m
- What is Biochemistry?5m
- Characteristics of Life12m
- Abiogenesis13m
- Nucleic Acids16m
- Proteins12m
- Carbohydrates8m
- Lipids10m
- Taxonomy10m
- Cell Organelles12m
- Endosymbiotic Theory11m
- Central Dogma22m
- Functional Groups15m
- Chemical Bonds13m
- Organic Chemistry31m
- Entropy17m
- Second Law of Thermodynamics11m
- Equilibrium Constant10m
- Gibbs Free Energy37m
- 2. Water3h 23m
- 3. Amino Acids8h 10m
- Amino Acid Groups8m
- Amino Acid Three Letter Code13m
- Amino Acid One Letter Code37m
- Amino Acid Configuration20m
- Essential Amino Acids14m
- Nonpolar Amino Acids21m
- Aromatic Amino Acids14m
- Polar Amino Acids16m
- Charged Amino Acids40m
- How to Memorize Amino Acids1h 7m
- Zwitterion33m
- Non-Ionizable Vs. Ionizable R-Groups11m
- Isoelectric Point10m
- Isoelectric Point of Amino Acids with Ionizable R-Groups51m
- Titrations of Amino Acids with Non-Ionizable R-Groups44m
- Titrations of Amino Acids with Ionizable R-Groups38m
- Amino Acids and Henderson-Hasselbalch44m
- 4. Protein Structure10h 4m
- Peptide Bond18m
- Primary Structure of Protein31m
- Altering Primary Protein Structure15m
- Drawing a Peptide44m
- Determining Net Charge of a Peptide42m
- Isoelectric Point of a Peptide37m
- Approximating Protein Mass7m
- Peptide Group22m
- Ramachandran Plot26m
- Atypical Ramachandran Plots12m
- Alpha Helix15m
- Alpha Helix Pitch and Rise20m
- Alpha Helix Hydrogen Bonding24m
- Alpha Helix Disruption23m
- Beta Strand12m
- Beta Sheet12m
- Antiparallel and Parallel Beta Sheets39m
- Beta Turns26m
- Tertiary Structure of Protein16m
- Protein Motifs and Domains23m
- Denaturation14m
- Anfinsen Experiment20m
- Protein Folding34m
- Chaperone Proteins19m
- Prions4m
- Quaternary Structure15m
- Simple Vs. Conjugated Proteins10m
- Fibrous and Globular Proteins11m
- 5. Protein Techniques14h 5m
- Protein Purification7m
- Protein Extraction5m
- Differential Centrifugation15m
- Salting Out18m
- Dialysis9m
- Column Chromatography11m
- Ion-Exchange Chromatography35m
- Anion-Exchange Chromatography38m
- Size Exclusion Chromatography28m
- Affinity Chromatography16m
- Specific Activity16m
- HPLC29m
- Spectrophotometry51m
- Native Gel Electrophoresis23m
- SDS-PAGE34m
- SDS-PAGE Strategies16m
- Isoelectric Focusing17m
- 2D-Electrophoresis23m
- Diagonal Electrophoresis29m
- Mass Spectrometry12m
- Mass Spectrum47m
- Tandem Mass Spectrometry16m
- Peptide Mass Fingerprinting16m
- Overview of Direct Protein Sequencing30m
- Amino Acid Hydrolysis10m
- FDNB26m
- Chemical Cleavage of Bonds29m
- Peptidases1h 6m
- Edman Degradation30m
- Edman Degradation Sequenator and Sequencing Data Analysis4m
- Edman Degradation Reaction Efficiency20m
- Ordering Cleaved Fragments21m
- Strategy for Ordering Cleaved Fragments58m
- Indirect Protein Sequencing Via Geneomic Analyses24m
- 6. Enzymes and Enzyme Kinetics13h 38m
- Enzymes24m
- Enzyme-Substrate Complex17m
- Lock and Key Vs. Induced Fit Models23m
- Optimal Enzyme Conditions9m
- Activation Energy24m
- Types of Enzymes41m
- Cofactor15m
- Catalysis19m
- Electrostatic and Metal Ion Catalysis11m
- Covalent Catalysis18m
- Reaction Rate10m
- Enzyme Kinetics24m
- Rate Constants and Rate Law35m
- Reaction Orders52m
- Rate Constant Units11m
- Initial Velocity31m
- Vmax Enzyme27m
- Km Enzyme42m
- Steady-State Conditions25m
- Michaelis-Menten Assumptions18m
- Michaelis-Menten Equation52m
- Lineweaver-Burk Plot43m
- Michaelis-Menten vs. Lineweaver-Burk Plots20m
- Shifting Lineweaver-Burk Plots37m
- Calculating Vmax40m
- Calculating Km31m
- Kcat46m
- Specificity Constant1h 1m
- 7. Enzyme Inhibition and Regulation 8h 42m
- Enzyme Inhibition13m
- Irreversible Inhibition12m
- Reversible Inhibition9m
- Inhibition Constant26m
- Degree of Inhibition15m
- Apparent Km and Vmax29m
- Inhibition Effects on Reaction Rate10m
- Competitive Inhibition52m
- Uncompetitive Inhibition33m
- Mixed Inhibition40m
- Noncompetitive Inhibition26m
- Recap of Reversible Inhibition37m
- Allosteric Regulation7m
- Allosteric Kinetics17m
- Allosteric Enzyme Conformations33m
- Allosteric Effectors18m
- Concerted (MWC) Model25m
- Sequential (KNF) Model20m
- Negative Feedback13m
- Positive Feedback15m
- Post Translational Modification14m
- Ubiquitination19m
- Phosphorylation16m
- Zymogens13m
- 8. Protein Function 9h 41m
- Introduction to Protein-Ligand Interactions15m
- Protein-Ligand Equilibrium Constants22m
- Protein-Ligand Fractional Saturation32m
- Myoglobin vs. Hemoglobin27m
- Heme Prosthetic Group31m
- Hemoglobin Cooperativity23m
- Hill Equation21m
- Hill Plot42m
- Hemoglobin Binding in Tissues & Lungs31m
- Hemoglobin Carbonation & Protonation19m
- Bohr Effect23m
- BPG Regulation of Hemoglobin24m
- Fetal Hemoglobin6m
- Sickle Cell Anemia24m
- Chymotrypsin18m
- Chymotrypsin's Catalytic Mechanism38m
- Glycogen Phosphorylase21m
- Liver vs Muscle Glycogen Phosphorylase21m
- Antibody35m
- ELISA15m
- Motor Proteins14m
- Skeletal Muscle Anatomy22m
- Skeletal Muscle Contraction45m
- 9. Carbohydrates7h 49m
- Carbohydrates19m
- Monosaccharides15m
- Stereochemistry of Monosaccharides33m
- Monosaccharide Configurations32m
- Cyclic Monosaccharides20m
- Hemiacetal vs. Hemiketal19m
- Anomer14m
- Mutarotation13m
- Pyranose Conformations23m
- Common Monosaccharides33m
- Derivatives of Monosaccharides21m
- Reducing Sugars21m
- Reducing Sugars Tests19m
- Glycosidic Bond48m
- Disaccharides40m
- Glycoconjugates12m
- Polysaccharide7m
- Cellulose7m
- Chitin8m
- Peptidoglycan12m
- Starch13m
- Glycogen14m
- Lectins16m
- 10. Lipids5h 49m
- Lipids15m
- Fatty Acids30m
- Fatty Acid Nomenclature11m
- Omega-3 Fatty Acids12m
- Triacylglycerols11m
- Glycerophospholipids24m
- Sphingolipids13m
- Sphingophospholipids8m
- Sphingoglycolipids12m
- Sphingolipid Recap22m
- Waxes5m
- Eicosanoids19m
- Isoprenoids9m
- Steroids14m
- Steroid Hormones11m
- Lipid Vitamins19m
- Comprehensive Final Lipid Map13m
- Biological Membranes16m
- Physical Properties of Biological Membranes18m
- Types of Membrane Proteins8m
- Integral Membrane Proteins16m
- Peripheral Membrane Proteins12m
- Lipid-Linked Membrane Proteins21m
- 11. Biological Membranes and Transport 6h 37m
- Biological Membrane Transport21m
- Passive vs. Active Transport18m
- Passive Membrane Transport21m
- Facilitated Diffusion8m
- Erythrocyte Facilitated Transporter Models30m
- Membrane Transport of Ions29m
- Primary Active Membrane Transport15m
- Sodium-Potassium Ion Pump20m
- SERCA: Calcium Ion Pump10m
- ABC Transporters12m
- Secondary Active Membrane Transport12m
- Glucose Active Symporter Model19m
- Endocytosis & Exocytosis18m
- Neurotransmitter Release23m
- Summary of Membrane Transport21m
- Thermodynamics of Membrane Diffusion: Uncharged Molecule51m
- Thermodynamics of Membrane Diffusion: Charged Ion1h 1m
- 12. Biosignaling9h 45m
- Introduction to Biosignaling44m
- G protein-Coupled Receptors32m
- Stimulatory Adenylate Cyclase GPCR Signaling42m
- cAMP & PKA28m
- Inhibitory Adenylate Cyclase GPCR Signaling29m
- Drugs & Toxins Affecting GPCR Signaling20m
- Recap of Adenylate Cyclase GPCR Signaling5m
- Phosphoinositide GPCR Signaling58m
- PSP Secondary Messengers & PKC27m
- Recap of Phosphoinositide Signaling7m
- Receptor Tyrosine Kinases26m
- Insulin28m
- Insulin Receptor23m
- Insulin Signaling on Glucose Metabolism57m
- Recap Of Insulin Signaling in Glucose Metabolism6m
- Insulin Signaling as a Growth Factor1h 1m
- Recap of Insulin Signaling As A Growth Factor9m
- Recap of Insulin Signaling1m
- Jak-Stat Signaling25m
- Lipid Hormone Signaling15m
- Summary of Biosignaling13m
- Signaling Defects & Cancer20m
- Review 1: Nucleic Acids, Lipids, & Membranes2h 47m
- Nucleic Acids 19m
- Nucleic Acids 211m
- Nucleic Acids 34m
- Nucleic Acids 44m
- DNA Sequencing 19m
- DNA Sequencing 211m
- Lipids 111m
- Lipids 24m
- Membrane Structure 110m
- Membrane Structure 29m
- Membrane Transport 18m
- Membrane Transport 24m
- Membrane Transport 36m
- Practice - Nucleic Acids 111m
- Practice - Nucleic Acids 23m
- Practice - Nucleic Acids 39m
- Lipids11m
- Practice - Membrane Structure 17m
- Practice - Membrane Structure 25m
- Practice - Membrane Transport 16m
- Practice - Membrane Transport 26m
- Review 2: Biosignaling, Glycolysis, Gluconeogenesis, & PP-Pathway3h 12m
- Biosignaling 19m
- Biosignaling 219m
- Biosignaling 311m
- Biosignaling 49m
- Glycolysis 17m
- Glycolysis 27m
- Glycolysis 38m
- Glycolysis 410m
- Fermentation6m
- Gluconeogenesis 18m
- Gluconeogenesis 27m
- Pentose Phosphate Pathway15m
- Practice - Biosignaling13m
- Practice - Bioenergetics 110m
- Practice - Bioenergetics 216m
- Practice - Glycolysis 111m
- Practice - Glycolysis 27m
- Practice - Gluconeogenesis5m
- Practice - Pentose Phosphate Path6m
- Review 3: Pyruvate & Fatty Acid Oxidation, Citric Acid Cycle, & Glycogen Metabolism2h 26m
- Pyruvate Oxidation9m
- Citric Acid Cycle 114m
- Citric Acid Cycle 27m
- Citric Acid Cycle 37m
- Citric Acid Cycle 411m
- Metabolic Regulation 18m
- Metabolic Regulation 213m
- Glycogen Metabolism 16m
- Glycogen Metabolism 28m
- Fatty Acid Oxidation 111m
- Fatty Acid Oxidation 28m
- Citric Acid Cycle Practice 17m
- Citric Acid Cycle Practice 26m
- Citric Acid Cycle Practice 32m
- Glucose and Glycogen Regulation Practice 14m
- Glucose and Glycogen Regulation Practice 26m
- Fatty Acid Oxidation Practice 14m
- Fatty Acid Oxidation Practice 27m
- Review 4: Amino Acid Oxidation, Oxidative Phosphorylation, & Photophosphorylation1h 48m
- Amino Acid Oxidation 15m
- Amino Acid Oxidation 211m
- Oxidative Phosphorylation 18m
- Oxidative Phosphorylation 210m
- Oxidative Phosphorylation 310m
- Oxidative Phosphorylation 47m
- Photophosphorylation 15m
- Photophosphorylation 29m
- Photophosphorylation 310m
- Practice: Amino Acid Oxidation 12m
- Practice: Amino Acid Oxidation 22m
- Practice: Oxidative Phosphorylation 15m
- Practice: Oxidative Phosphorylation 24m
- Practice: Oxidative Phosphorylation 35m
- Practice: Photophosphorylation 15m
- Practice: Photophosphorylation 21m
Nucleic Acids 4 - Online Tutor, Practice Problems & Exam Prep
The double helix structure of DNA exists in three forms: A, B, and Z. The B form, identified by Watson and Crick, is the most common, while the A form is more condensed and typically found in double-stranded RNA. The Z form is a left-handed helix found near regulatory sequences, where BZ junctions occur. These structural variations are crucial for cellular functions, as the physical shape of DNA aids in identifying regulatory regions. Hydrophobic stacking forces and phosphate groups stabilize these structures in an aqueous environment, emphasizing the importance of conformation in biological processes.
Nucleic Acids 4
Video transcript
Here’s what students ask on this topic:
What are the different forms of the DNA double helix?
The DNA double helix can exist in three different forms: A, B, and Z. The B form is the most common and was the structure predicted by Watson and Crick. The A form is more condensed and is typically found in double-stranded RNA. The Z form is a left-handed helix and is generally found near regulatory sequences. These forms are crucial for cellular functions, as the physical shape of DNA aids in identifying regulatory regions. Hydrophobic stacking forces and phosphate groups stabilize these structures in an aqueous environment.
What is the significance of BZ junctions in DNA?
BZ junctions are areas where the DNA helix switches between the B form and the Z form. This switch involves nucleotides flipping out from the strands, causing the helix to change direction from right-handed (B form) to left-handed (Z form). This abrupt change in shape is important because it helps cells identify regulatory sequences. The noticeable shape difference at BZ junctions is exploited by cells to determine where these regulatory regions are located, playing a crucial role in gene expression and regulation.
How do hydrophobic stacking forces contribute to the stability of DNA?
Hydrophobic stacking forces contribute to the stability of DNA by helping to maintain its structure. These forces arise because the bases in the DNA are hydrophobic and tend to stack on top of each other in the center of the helix. This stacking minimizes their exposure to water, creating a stable, compact structure. Additionally, the phosphate groups on the outside of the DNA helix interact with the aqueous environment, further stabilizing the DNA structure. Together, these interactions ensure the integrity and functionality of the DNA molecule.
Why is the B form of DNA the most common?
The B form of DNA is the most common because it is the most stable under physiological conditions, such as those found in the aqueous environment of cells. The B form allows for optimal hydrogen bonding between base pairs and efficient hydrophobic stacking of the bases, which contributes to its stability. Additionally, the B form's structure is well-suited for interactions with proteins and other molecules involved in DNA replication, transcription, and repair, making it the preferred conformation for most cellular processes.
What are the structural differences between A, B, and Z forms of DNA?
The A, B, and Z forms of DNA differ in their helical structure and handedness. The A form is a right-handed helix that is more condensed than the B form, with a hollow center. The B form, also right-handed, is the most common and has a more extended structure without a hollow center. The Z form is a left-handed helix and is generally found near regulatory sequences. These structural differences are important for the various functions and interactions of DNA within the cell.