Here is another nice example of the partnership between experiment and computation in ascertaining molecular structure. The Sarpong, Tantillo, Andersen, Berlinck, and Miller groups collaborated on the synthesis, characterization and biosynthesis of some metabolites from Penniculium strains.¹ I will focus here on just the structural identification component of this paper; the synthesis and the biosynthesis are very interesting too!

Cyclopiamine A 1 and cyclopiamine B 2 interconvert through an intermediate that allows for the epimerization at carbon bearing the nitro group.²

1

2

Citrinalin A 3 might also seem to undergo the same type of ring opening-ring closing reaction to produce citrinalin B. However, the original proposed structure³ of citrinalin B 4 implies an epimerization at a different carbon (at the ring fusion to the terminal 5 member ring). These authors suggested that perhaps the proper structure of citrinalin B is 5, which differs from citrinalin A only at the carbon bearing the nitro group, analogous to the relationship between 1 and 2.

3
4	5

The low energy conformations of both 4 and 5 (actually the trifluoroacetic acid salts) were optimized at B3LYP/6-31+G(d,p) and the chemical shifts for both ¹H and ¹³C were computed, Boltzmann-weighted and scaled, and then compared with the NMR spectra of authentic citrinalin B. (The lowest energy conformations of 4 and 5 are shown in Figure 1.) The corrected mean absolute deviations for the ¹H and ¹³C chemical shift for the original structure 4 are 0.45 ppm and 2.0 ppm, respectively (with the largest outliers of 2.3 ppm for H and 9.6 ppm for C). These errors are about twice what is observed in comparing the experimental and computed ¹H and ¹³C chemical shifts of 3. The agreement between the computed and experimental values using 5 are much improved, with mean deviations of 0.12 and 1.6ppm, and largest deviations of 0.38 ppm for ¹H and 4.4 ppm for ¹³C. Use of Goodman’s DP4 method indicates a 100% probability that the structure of citrinalin B is 5. This prediction is confirmed by the x-ray structure.

4

5

Figure 1. B3LYP/6-31+G(d,p) optimized lowest energy conformers of 4 and 5.

References

(1) Mercado-Marin, E. V.; Garcia-Reynaga, P.; Romminger, S.; Pimenta, E. F.; Romney, D. K.; Lodewyk, M. W.; Williams, D. E.; Andersen, R. J.; Miller, S. J.; Tantillo, D. J.; Berlinck, R. G. S.; Sarpong, R. "Total synthesis and isolation of citrinalin and cyclopiamine congeners," Nature 2014, 509, 318-324, DOI: 10.1038/nature13273.

(2) Bond, R. F.; Boeyens, J. C. A.; Holzapfel, C. W.; Steyn, P. S. "Cyclopiamines A and B, novel oxindole metabolites of Penicillium cyclopium westling," J. Chem. Soc., Perkin Trans I 1979, 1751-1761, DOI: 10.1039/P19790001751.

(3) Pimenta, E. F.; Vita-Marques, A. M.; Tininis, A.; Seleghim, M. H. R.; Sette, L. D.; Veloso, K.; Ferreira, A. G.; Williams, D. E.; Patrick, B. O.; Dalisay, D. S.; Andersen, R. J.; Berlinck, R. G. S. "Use of Experimental Design for the Optimization of the Production of New Secondary Metabolites by Two Penicillium Species," J. Nat. Prod. 2010, 73, 1821-1832, DOI: 10.1021/np100470h.

InChIs

1: InChI=1S/C26H33N3O5/c1-23(2)12-17(30)20-18(34-5)9-8-16-21(20)28(23)22(31)26(16)13-25(29(32)33)14-27-10-6-7-15(27)11-19(25)24(26,3)4/h8-9,15,19H,6-7,10-14H2,1-5H3/t15-,19+,25+,26-/m1/s1
InChIKey=LUESCPPTDJZZQN-VBPSCBQFSA-N

2: InChI=1S/C26H33N3O5/c1-23(2)12-17(30)20-18(34-5)9-8-16-21(20)28(23)22(31)26(16)13-25(29(32)33)14-27-10-6-7-15(27)11-19(25)24(26,3)4/h8-9,15,19H,6-7,10-14H2,1-5H3/t15-,19+,25-,26-/m1/s1
InChIKey=LUESCPPTDJZZQN-DCWDXPITSA-N

3: InChI=1S/C25H31N3O5/c1-22(2)11-16(29)19-17(33-22)8-7-15-20(19)26-21(30)25(15)12-24(28(31)32)13-27-9-5-6-14(27)10-18(24)23(25,3)4/h7-8,14,18H,5-6,9-13H2,1-4H3,(H,26,30)/t14-,18+,24+,25-/m0/s1
InChIKey=QVBARITWSQCBBV-DTKOROOESA-N

4: InChI=1S/C25H31N3O5/c1-22(2)11-16(29)19-17(33-22)8-7-15-20(19)26-21(30)25(15)12-24(28(31)32)13-27-9-5-6-14(27)10-18(24)23(25,3)4/h7-8,14,18H,5-6,9-13H2,1-4H3,(H,26,30)/t14-,18-,24-,25+/m1/s1
InChIKey=QVBARITWSQCBBV-KUCRYEJDSA-N

5: InChI=1S/C25H31N3O5/c1-22(2)11-16(29)19-17(33-22)8-7-15-20(19)26-21(30)25(15)12-24(28(31)32)13-27-9-5-6-14(27)10-18(24)23(25,3)4/h7-8,14,18H,5-6,9-13H2,1-4H3,(H,26,30)/t14-,18+,24-,25-/m0/s1
InChIKey=QVBARITWSQCBBV-QBDGMKJCSA-N

Carbenes remain an active area of interest for computational chemists, as seen in Chapter 5 of my book. For many carbenes, the triplet is the ground state, and that is true of diphenylcarbene 1. Can solvent play a role in the stability of carbene spin states? Surprisingly, the answer, provided in a recent paper by Sander,¹ is yes!

In the gas phase, the singlet-triplet gap of 1 is computed to be 5.62 kcal mol^-1 at (U)B3LYP/6-311++G(d,p) (and this reduces to 5.06 at (U)B3LYP+D3/6-311++G(d,p)) with the ground state as a triplet. If a single methanol molecules is allowed to approach 1, then the complex involving the singlet has a short hydrogen bond distance of 1.97 Å but the triplet has a much longer distance of 2.33 Å. These structures are shown in Figure 1. This manifests in a dramatic change in the relative stability, with the singlet complex now 0.26 kcal mol^-1 (0.44 with the D3 correction) lower in energy than the triplet.

Singlet-1:methanol

Triplet-1:methanol

Figure 1. (U)B3LYP/6-311++G(d,p) optimized geometries of the compelxes of methanol with singlet or triplet 1.

IR spectroscopy of 1 in an argon matrix doped with a small amount of methanol confirms the presence of the singlet carbene, and detailed description of the difference in the reactivities of the singlet and triplet are provided.

References

(1) Costa, P.; Sander, W. "Hydrogen Bonding Switches the Spin State of Diphenylcarbene from Triplet to Singlet," Angew. Chem. Int. Ed. 2014, 53, 5122-5125, DOI: 10.1002/anie.201400176.

InChIs

1: InChI=1S/C13H10/c1-3-7-12(8-4-1)11-13-9-5-2-6-10-13/h1-10H
InChIKey=XMGMFRIEKMMMSU-UHFFFAOYSA-N

Dihdroxycarbene was the subject of a post a few years ago relating to how this carbene does not undergo tunneling,¹ while related hydroxycarbene do undergo a tunneling rearrangement.

Now we have a gas-phase microwave determination of the trans,cis isomer of dihydroxycarbene.² The computed CCSD(T)/cc-pCVQZ structure is shown in Figure 1. What is truly remarkable here is the amazing agreement between the experimental and computed structure – as seen in Table 1.The bond distance are in agreement within 0.001 Å and the bond angles agree within 0.3°! Just further evidence of the quality one can expect from high-level computations. And computing this structure was certainly far easier than the experiments!

Figure 1. CCSD(T)/cc-pCVQZ optimized geometry of dihydroxycarbene.

Table 1. Experimental and computed (CCSD(T)/cc-pCVQZ) geometric parameters of dihydroxycarbene.^a

	Expt.	Comp.
C-O	1.335	1.336
C-O	1.309	1.309
O-Htrans	0.961	0.960
O-Hcis	0.976	0.975
O-C-O	107.30	107.25
C-O-H­­trans	106.8	106.8
C-O-H­­cis	110.7	110.4
aDistances in Å and angles in deg.

References

(1) Schreiner, P. R.; Reisenauer, H. P. "Spectroscopic Identification of Dihydroxycarbene," Angew. Chem. Int. Ed. 2008, 47, 7071-7074, DOI: 10.1002/anie.200802105.

(2) Womack, C. C.; Crabtree, K. N.; McCaslin, L.; Martinez, O.; Field, R. W.; Stanton, J. F.; McCarthy, M. C. "Gas-Phase Structure Determination of Dihydroxycarbene, One of the Smallest Stable Singlet Carbenes," Angew. Chem. Int. Ed. 2014, 53, 4089-4092, DOI: 10.1002/anie.201311082.

InChIs

Dihydroxycarbene: InChI=1S/CH2O2/c2-1-3/h2-3H
InChIKey=VZOMUUKAVRPMBY-UHFFFAOYSA-N