1
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s
Linking s uc u al
and composi ional changes
in a chaeological human bone
collagen: an FTIR‑ATR app oach
An onio Ma ínez Co izas1 & Olalla López‑Cos as1,2,3*
Collagen is he main s uc u al and mos abundan p o ein in he human body, and i is ou inely
ex ac ed and analysed in scien i ic a chaeology. I s deg ee o p ese a ion is, he e o e, c ucial
and se e al app oaches a e used o de e mine i . Spec oscopic echniques p o ide a cos ‑e ec i e,
non‑des uc i e me hod o in es iga e he molecula s uc u e, especially when combined wi h
mul i a ia e s a is ics (chemome ic app oach). In his s udy, we used FTIR‑ATR spec oscopy o
cha ac e ise collagen ex ac ed om skele ons eco e ed om nec opoleis in NW Spain spanning
om he B onze Age o eigh een h cen u y AD. P incipal componen s analysis was pe o med on
a selec ion o bands and s uc u al equa ion models (SEM) we e de eloped o ela e he collagen
quali y indica o s o collagen s uc u al change. Fou p incipal componen s ep esen ed: (i) Cp1,
ans o ma ions o he backbone p o ein wi h a esidual inc ease in p o eoglycans; (ii) Cp2, p o ein
ans o ma ions no accompanied by changes in p o eoglycans abundance; (iii) Cp3, a ia ions
in alipha ic side chains and (i ) Cp4, abso p ion o he OH o ca bohyd a es and amide. Highly
explana o y SEM models we e ob ained o he adi ional collagen quali y indica o s (collagen yield,
C, N, C:N), bu no ela ionship was ound be ween quali y and δ13C and δ15N a ios. The obse ed
dec ease in C and N con en and inc ease in C:N a ios is con olled by he deg ada ion o p o ein
backbone componen s and he ela i e p ese a ion o ca bon‑ ich compounds, p o eoglycans and, o
a lesse ex en , alipha ic moie ies. Ou esul s sugges ha FTIR‑ATR is an ideal echnique o collagen
cha ac e iza ion/p e‑sc eening o palaeodie , mobili y and adioca bon esea ch.
Skele al collagen is one o he mos abundan p o eins in e eb a e o ganisms, o med by a complex s uc u e o
ib es and mic o ibe s ha connec in a wis ed, ope-like assembly1,2. Collagen 3D s uc u e has been he ocus
o nume ous s udies in medical and biological sciences e.g.3–6 and special a en ion has been paid o he changes
ha occu a s uc u al le el since hey can a ec no mal molecule unc ions in body e.g. sus en ion, connec ion,
e c.7. O hopaedic in es iga ions ha e ocused upon he deg ada ion o human ca ilage/bone, pa icula ly he
de elopmen o degene a i e changes ha esul in os eoa h i is and cause modi ica ions o he 3D s uc u e,
which occu in pa allel wi h he ad ance o he disease see o example8,9.
Due o i s abundance and s eng h in skele on, i is possible o ind collagen (mainly Type I) molecules in a
human body se e al cen u ies a e dea h and e en a e millions o yea s in ossil animals10. A chaeology, o ensic
science and physical an h opology ou inely analyse ex ac ed collagen o unde s and he p e-mo em ea u es o
he deceased (e.g. die and mobili y wi h s able iso opes, animal species h ough ZooMS) and o use o adio-
ca bon da ing11–15. Mos o hese analyses a e based on iso opic composi ion ac iona ion wi h s ic con ol a
elemen al composi ion le el o disca d unsui able samples. Common conce ns wi h he ex ac ion o collagen
a e he p esence o exogenous subs ances (e.g. humid acids) and he loss o in eg i y o he collagen molecule.
The i s conce n has been alle ia ed by he imp o emen o ex ac ion me hods o p o ide an imp o ed le el o
ce ain y abou he elimina ion o non-collagen subs ances16. E o s o unde s and ex ac ed collagen in eg i y/
quali y ha e mainly ocused on applying elemen al composi ion (i.e. C, N and C:N) cu -poin s, as desc ibed in17,
o add ess he second conce n.
OPEN
1EcoPas , Facul y o Biology, Campus Vida, Uni e sidade de San iago de Compos ela, 15782 San iago de
Compos ela, Spain. 2A chaeological Resea ch Labo a o y, S ockholm Uni e si y, Wallenbe glabo a o ie ,
10691 S ockholm, Sweden. 3Labo a o y o An h opology, Depa men o Legal Medicine, Toxicology and Physical
An h opology, Facul y o Medicine, Uni e sidad de G anada, 18012 G anada, Spain. *email: [email p o ec ed]
2
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
In con as o he in i o molecule, a chaeological collagen deg ada ion models a e complex because hey
need o conside changes ha occu ed du ing pos -mo em. Some au ho s ha e used collagen quali y18,19 o
modelled linea s uc u e20 as an indica o o bone deg ada ion wi h ime, and, in con as o medical sciences,
less a en ion has been paid o s uc u al changes. E en now, he mechanisms and p ocesses ha in luence he
deg ada ion o collagen ex ac ed om a chaeological bone samples a e s ill poo ly unde s ood. To ed ess his,
i is necessa y o un a el he changes a s uc u al le el o achie e a good unde s anding o a chaeological col-
lagen p ese a ion.
Despi e i s po en ial, ew s udies ha e used spec oscopic echniques o de e mine collagen p ese a ion
in a chaeological bone21–25. Fou ie T ans o m In a ed (FTIR) spec oscopy has been ega ded as a sui able
me hod o explo e he s uc u e o collagen26–32, by ela ing FTIR abso p ion bands (o he amide I, II and III)
o speci ic chemical bonds and seconda y s uc u al ea u es (α-helix, β-shee s, β- u ns and andom coils), e en
in he mos ecen in es iga ions32. Bu , as ea ly as he mid- wen ie h cen u y, he e was a undamen al change
in he comp ehension o he collagen s uc u e led by X- ay di ac ion in es iga ions, which showed ha he
adi ional model was inco ec and he polyp oline II (PPII) model was in oduced and backed by la e in es-
iga ions, becoming he accep ed model4,33–36. Al hough FTIR does no p o ide he same le el o de ail o he
molecula composi ion compa ed wi h X- ay, Nuclea Magne ic Resonance (NMR) o Py olysis GC–MS, i can
p o ide none heless aluable insigh s abou he s uc u e o complex molecules such as p o eins37.
FTIR has many ad an ages when compa ed wi h he con en ional me hods commonly used o s udy collagen.
I is a quick, cos -e ec i e and non-in asi e me hod21,26. Mos s udies ha ha e used FTIR on ancien skele ons
ocus on he cha ac e isa ion o he bone mine al componen among o he s38–41 o on aphonomic p ocesses
such as c ema ion42–44. The collagenous po ion o bone has been analysed wi h ela i ely less equency using
FTIR15,22,45–47, and Raman spec oscopy21,48–50. S udies o bulk bone ha e also demons a ed ha i is di icul o
de ec collagen con en in poo ly p ese ed bones47, whe eas ex ac ed a chaeological collagen has only been
di ec ly analysed in ew s udies24,25,48. The e o e, p e ious esea ch ocused upon es ablishing c i e ia o pa am-
e e s o collagen p ese a ion sc eening while he changes in he s uc u e o he molecule ha e ecei ed much
less a en ion.
The objec i e o ou s udy is o cha ac e ise collagen ex ac ed om a chaeological human bone o di e en
age, une a y con ex and bu ial en i onmen , using FTIR-ATR in he mid in a ed egion (4000–400cm−1).
By using a combina ion o p incipal componen s analysis (PCA) and pa ial leas squa es-s uc u al equa ion
modelling (PLS-SEM), we (i) discuss he possible mechanisms o a chaeological bone collagen s uc u al ans-
o ma ion, (ii) he po en ial o FTIR-ATR o p edic collagen quali y indica o s (i.e. C, N, C:N, collagen yield)
and (iii) whe he collagen quali y a ec s i s iso opic (δ13C and δ 15N) composi ion, which is key o he s udy o
human palaeodie and adioca bon da ing.
Resul s
Collagen p ope ies. O he i y samples analysed, collagen yield anged be ween 25% (simila o in ac
bone) and 2% (abo e he p oposed limi o 1%17), whe eas he C:N a io was be ween 3.18 and 3.57. Ca bon and
ni ogen con en s showed a la ge ange (C: 44.3–17.9%; N: 6.1–16.1%). None o he samples analysed in his
s udy exceeded he C and N alues o esh collagen (43% and 16%, espec i ely17,19) by mo e han 3%. Eigh
and wel e samples p o ided pe cen age C and N alues below 80% o hose o esh collagen espec i ely and
wo samples (424 and 705) we e below 50%. Only one sample (424) showed a C:N a io (3.57) sligh ly abo e he
ange (3.02–3.56) p oposed as ep esen a i e o well-p ese ed collagen17.
A wide dis ibu ion o iso opic esul s has been ound in his s udy, especially o δ13C, which is in e p e ed
as he esul o palaeodie a y p e e ences. Fo example, he obse ed di e ences in δ13C can be ela ed o geo-
g aphical loca ion, whe he coas al o inland, and δ13C was ound o be in luenced by he consump ion o ma ine
esou ces. P e e ence o he use o C4 plan s in human and domes ic animal die and a s ong eliance on sea ood
and ish—on he coas —occu s in No h-Wes e n Spain51. His o ical and a chaeological da a ag ee wi h he
ob ained iso opic signa u es and we e discussed in de ail o he analysed popula ions52.
Fo he samples used in his s udy, collagen yield shows signi ican , al hough low, co ela ions only wi h C,
N and he C:N a io ( 0.44, 0.49 and −0.48, espec i ely; P < 0.01). Ca bon and ni ogen con en s a e highly
co ela ed ( 0.99; P < 0.01) wi h each o he and a e nega i ely co ela ed wi h he C:N a io (−0.76 and −0.79,
espec i ely; P < 0.01). Collagen composi ional p ope ies a e no signi ican ly co ela ed wi h he iso ope a ios.
Despi e his, bo h iso ope a ios a e mode a ely co ela ed ( 0.55; P < 0.01), caused by he inpu o ma ine
esou ces in luencing some o he samples see51.
Collagen FTIR‑ATR spec a. The a e age spec um o he samples shows he cha ac e is ic band dis ibu-
ion o collagen, wi h high abso bance in he egions 1500–1700cm−1 and 2800–3500cm−1, mode a e abso b-
ance a 1300–1500cm−1 and ela i ely low a e age abso bance a 800–1200cm−1 (Fig.1a). The s anda d de ia-
ion spec um is simila o he a e age one bu shows a ela i ely la ge a ia ion be ween samples in he egion
800–1200cm−1, despi e i s low a e age abso bance (Fig.1a); whe eas he 2800–3500cm−1 egion only p esen s
a peak a ound 3300cm−1.
The mos ele an peaks ob ained om he second de i a i e spec a, in he egion 800–1800cm−1, a e shown
in Fig.1b. Assignmen o he selec ed bands can be ound in SI_Table2. All spec a p esen ed abso p ions a
897, 918, 947, 974, 1030, 1059, 1080 and 1121cm−1 ha a e cha ac e is ic o ca bohyd a e moie ies (CO s and
COC s ); 1236cm−1, o he amide III (CN s and NH d); 1337 and 1450cm−1, a ibu able o me hylene (CH2
d and CH3 d; he eon named as alipha ic) abso p ions; 1545, 1624 and 1719cm−1, due o amide II (CN s and
NH bd) and amide I (mos ly C = 0 s ), espec i ely; 2874 and 2930cm−1 assigned o alipha ics (CH s and CH3
s ); 3070cm−1, o he amide B (NH s ); and a b oad band 3500–3300cm−1 ela ed o amide A (NH s ) and OH
3
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
ib a ions. Abso p ions a 947, 974 and 1624cm−1 showed he la ges a iabili y; while abso p ions a 1030, 1059,
1080, 1121, 1200, 1236, 1337, 1450, 1545 and 1655cm−1 displayed mode a e a iabili y (Fig.1b).
Main spec oscopic signals o collagen. We selec ed 24 bands, which a e ep esen a i e o he di e en
spec al egions o he ype I collagen spec um (ca bohyd a es, amide III, miscellaneous—mainly alipha ics—
egion, amide II, amide I, alipha ics, amide B, amide A/OH; o a de ini ion o hese egions see o example27,31),
o pe o m he PCA. Fou p incipal componen s accoun ed o 95.5% o he a iance (Table1). The i s com-
ponen , Cp1, explains 45.5% o he o al a iance and i is cha ac e ised by la ge posi i e loadings (0.73–0.94)
o abso p ions o ca bohyd a es (i.e. collagen p o eoglycans) and la ge nega i e loadings (−0.86 o −0.72) o
abso p ions o he amides (I, II and III) and he miscellaneous egion (Table1).
Figu e1. (A) A e age (black line) and s anda d de ia ion (g ey line) mid in a ed FTIR-ATR spec a o he
whole se o collagen samples analysed in his s udy. (B) A e age spec um o he second de i a i e spec a o
he analysed samples in he 1800–800cm−1 egion; g ey ba s co espond o he s anda d de ia ion o he main
abso p ions. Ve ical dashed lines sepa a e he main collagen spec al egions (see “Collagen FTIR-ATR spec a”
sec ion).
4
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
The second componen , Cp2, explains 38.3% o he o al a iance. Amide (I, II, III, A and B) and alipha ic
(CH2 and CH3) abso p ions show la ge posi i e loadings (Table1). O he collagen abso p ion bands, Cp2
accoun s o a la ge pe cen age o he 1690cm−1 (76%) and a mode a e pe cen age o 1624cm−1 (45%) a i-
ance o amide I, and 1200cm−1 (86%) o amide III. I also con ains a low (24%) pe cen age o he a iance he
1655cm−1 abso p ion.
Componen s Cp3 and Cp4 accoun o a mino pa o he o al a iance, 6.7 and 5.2%, espec i ely (Table1).
Abso p ions ela ed o alipha ics (2874, 2930 and 2982cm−1) ha e he la ges (albei mode a e o low) loadings
in Cp3. While abso p ions o he amide A/OH egion (3320 and 3458cm−1) and one o he ca bohyd a es bands
(1030cm−1) show mode a e and opposed (nega i e and posi i e, espec i ely) loadings in Cp4 (Table1).
Cp1 is highly co ela ed (P < 0.01) o he PGI, C and N con en , and he C:N a io (Table2). Collagen yield is
signi ican ly co ela ed o Cp1 and Cp3, and he CI is nega i ely co ela ed wi h Cp3, al hough he co ela ion
coe icien s a e low.
Table 1. Fac o loadings o he IR band o he ex ac ed componen s. WN: wa enumbe ; Eig : eigen alue;
Va : p opo ion o a iance. The la ges loading o each abso p ion band is in bold.
WN cm−1 Cp1 Cp2 Cp3 Cp4
897 0.93 0.33 0.07 − 0.05
918 0.90 0.38 0.12 0.06
947 0.94 0.29 0.09 0.01
974 0.91 0.29 0.10 0.22
1030 0.73 0.22 0.34 0.50
1059 0.88 0.21 0.34 0.23
1080 0.91 0.31 0.25 0.00
1121 0.91 0.33 0.14 − 0.07
1236 − 0.75 0.64 0.06 0.09
1337 − 0.72 0.62 0.16 0.08
1450 − 0.72 0.58 0.19 0.21
1522 − 0.86 0.41 0.25 0.15
1545 − 0.80 0.52 0.23 − 0.04
1655 − 0.76 0.49 0.19 − 0.11
1200 − 0.28 0.93 −0.03 0.04
1624 − 0.59 0.67 0.37 0.07
1690 − 0.35 0.87 0.05 0.00
1719 0.09 0.84 − 0.33 0.02
2874 0.33 0.76 − 0.53 0.07
2930 0.02 0.88 − 0.44 0.13
2982 0.16 0.89 − 0.40 0.06
3070 0.20 0.91 − 0.30 0.07
3320 0.22 0.73 0.21 − 0.60
3458 0.46 0.58 0.15 − 0.63
Eig 10.9 9.2 1.6 1.2
Va 45.4 38.2 6.7 5.2
Table 2. Co ela ion be ween de ex ac ed IR p incipal componen s. Collagen yield (Coll_yield), IR indices
(CI and PGI), elemen al composi ion (C and N), C:N mola a ios and iso opic composi ion o he collagen.
Cp1 Cp2 Cp3 Cp4
Coll_yield 0.42 0.21 0.41 − 0.22
CI 0.07 − 0.14 − 0.41 0.31
PGI 0.98 − 0.11 0.02 0.10
C − 0.93 0.05 0.11 0.09
N − 0.92 0.08 0.13 0.08
C:N 0.73 0.18 − 0.21 − 0.06
δ13C 0.00 − 0.15 − 0.09 0.17
δ 15N 0.23 − 0.07 − 0.17 0.18
5
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
Modelling collagen quali y and iso opic composi ion. The PCA esul s sugges ha he spec o-
scopic na u e o ex ac ed bone collagen can p o ide insigh s in o he main ans o ma ions o i s composi-
ion and s uc u e, which may be ela ed o collagen p ese a ion. To do so, we applied PLS-SEM modelling o
de e mine (i) whe he ans o ma ions o he collagen s uc u e a e coupled o changes in collagen quali y (i.e.
C, N, C:N, collagen yield), and (ii) i changes in collagen quali y a ec he iso opic (δ13C and δ 15N) composi-
ion. The model was ini ially designed wi h ou p edic o LV (amides, backbone lipids, side-chain lipids, and
ca bohyd a es; SI_Figu e3), one p ima y esponse LV (collagen quali y) and a seconda y esponse LV (collagen
iso opic composi ion; his one depending exclusi ely on collagen quali y). As indica o s, we used ep esen a i e
abso p ion bands o he p edic o LV, analysed p ope ies and indices (C, N, C:N, collagen yield, CI and PGI)
and iso opic a ios (δ13C, δ 15N). Al hough he model p edic ed 92% o he collagen quali y a iance (SI_Fig-
u e3), he lipids LV ailed o pass he collinea i y es s as i sha ed 88% o i s a iance wi h he amide LV and i s
o al e ec coe icien on collagen quali y was e y low (−0.04). As a esul , o he inal model we me ged his
LV wi h he amides in o one LV, named as “s uc u al componen s”.
O he 24 abso p ion bands used in he PCA, 17 me he c i e ia o good indica o s (absolu e alue o he
loading > 0.7, Table3) and we e kep in he model. I is wo h emembe ing ha he squa e o he ou e load-
ing accoun s o he p opo ion o a iance o he indica o ha is cap u ed by he LV in PLS-SEM e lec i e
mode. The loadings o he FTIR abso bances, wi h only one excep ion (1200cm−1, in he s uc u al componen s
LV), show ha almos all hei a iance is cap u ed by he modelled LV. Ca bon, N, C:N and PGI also mee he
c i e ia o good indica o s o collagen quali y, bu collagen yield has a mode a e loading and he CI a e y low
one (Table3). While he PGI highly co- a ies wi h he common collagen quali y pa ame e s and maybe a alid
indica o , collagen yield and CI a e no . Fo his speci ic model, collagen quali y is hus ela ed o he o me .
Collagen yield has some dependence on ope a o p ocessing (inaccu acy in pipe ing, il e ing, e c.).
The o al e ec s’ coe icien s (Fig.2) show ha he s uc u al componen s ha e he s onges , posi i e e ec
(0.79) on collagen quali y, while ca bohyd a es and side-chain lipids ha e nega i e o al e ec s (−0.43 and −0.22
espec i ely). The weigh o he s uc u al componen s on collagen quali y is almos wo and ou imes highe
han he weigh s o he o he wo LVs. This simple PLS-SEM model explains 92% o he a ia ion in collagen
quali y (Fig.2), in ol ing as much as 92–94% o he C and N, 85% o he PGI and 70% o he C:N a iance. Fig-
u e3 shows he ela ionship be ween obse ed and expec ed alues o he collagen quali y indica o s ob ained
wi h he PLS-SEM model. To al C and N con en s and he PGI a e accu a ely es ima ed, C:N a ios also show
a good albei lowe pe o mance, es ima ion o collagen yield is mode a e and ha o he CI is no signi ican .
Table 3. Loadings o he indica o s o he LV (p edic o s and esponses) o he PLS-SEM model. LVsc:
s uc u al componen s, LVcb: ca bohyd a es, LVsc: side chains, LVcq: collagen quali y, LVis: collagen iso opic
composi ion.
Indica o LVsc LVcb LVsc LVcq LVis
897cm−1 0.98
918cm−1 0.99
947cm−1 0.99
974cm−1 0.98
1059cm−1 0.95
1080cm−1 0.98
1200cm−1 0.78
1236cm−1 0.99
1337cm−1 0.97
1450cm−1 0.95
1522cm−1 0.96
1545cm−1 0.97
1624cm−1 0.94
1655cm−1 0.91
2874cm−1 0.97
2930cm−1 0.99
2982cm−1 0.99
C 0.96
N 0.97
C:N −0.84
Coll_yield 0.61
CI −0.19
PGI −0.92
δ13C 0.68
δ 15N 0.98
6
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
Addi ionally, a his le el, collagen quali y seems o ha e no signi ican e ec on he iso opic composi ion: i s
o al e ec coe icien on he iso opic composi ion is low and he explained a iance is almos negligible (4%).
Discussion
The esul s o he PCA a e in ag eemen wi h p e ious in es iga ions ha use FTIR spec a o p o ide addi ional
insigh s on p o ein, in pa icula collagen, composi ion and s uc u e29,31,32,37,53–55. Di e en collagen ypes can
be iden i ied/disc imina ed e icien ly using abso bances om selec ed egions o he spec um27.
In he samples analysed he e, Cp1 and Cp2 seems o e lec a loss o p o ein backbone componen s. As mos
o he a ia ion o he cha ac e is ic abso p ion o he alipha ic bonds (a 1337, 1450, 2874, 2934, and 2982cm−1)
a e also con ained in Cp1 and Cp2, and only a smalle p opo ion is cap u ed by Cp3 (Table2), i is likely ha
ib a ions in he i s wo componen s a e ela ed o he me hylene p esen in he backbone pep ide s uc u e
whe eas Cp3 may co espond o he alipha ic side chains. Cp4 seems o disc imina e be ween he OH abso p ion
o ca bohyd a es and ha o he amide A.
Figu e4 ep esen s a p ojec ion o samples’ sco es o Cp1 and Cp2. Mos samples (28 ou o 50) show nega i e
Cp1 sco es and posi i e o sligh ly nega i e Cp2 sco es. These may ep esen collagen wi h a mo e in ac , PPII-
like, molecula s uc u e. Twel e samples show posi i e Cp1 sco es and posi i e o sligh ly nega i e Cp2 sco es,
sugges ing some deg ee o collagen ans o ma ion no a ec ing he main p o ein backbone s uc u es. Samples
wi h posi i e Cp1 and nega i e Cp2 sco es may co espond o hose wi h mo e in ense s uc u al modi ica ions.
Collagen quali y pa ame e s (C, N and C:N) wi h he mos p onounced depa u e om hose o esh collagen
occu in he wo samples wi h he la ges Cp1 alues (424 om Ou igo and 705 om Capela do Pila ; Fig.4).
No e idence o soil con amina ion (i.e. humic acids) was de ec ed. Ou esul s a e consis en wi h indings in
a p e ious molecula s udy which used py olysis-GC–MS on 28 o he samples analysed he e16. Al hough a
de ailed compa ison wi h he molecula da a canno be done, he e is an o e all ag eemen in he classi ica ion
o collagen as well o poo ly p ese ed (20 samples ou o 28).
The PLS-SEM model (Fig.2) sugges s ha he mo e in ac he collagen backbone s uc u e ( e lec ed by LVs ),
he highe collagen quali y (highe C and N con en s and, o some ex en , collagen yield), while lowe quali y
(highe C:N a ios and PGI alues) is cha ac e ised by he ela i e abundance o ca bohyd a es (LVcb) and, o a
limi ed ex en , lipidic side chains (LVsc). Collagen ans o ma ion esul s in an o e all dec ease in C and N, and
Figu e2. To al e ec s coe icien s o he inal PLS-s uc u al equa ion model, including h ee p edic o LVs
(s uc u al componen s, polysaccha ides, side chain lipids), one p ima y esponse LV (collagen quali y) and a
seconda y esponse LV (iso opic composi ion). P oxies o he p edic o LVs a e iden i ied by he wa enumbe s
o he main abso p ions o collagen componen s.
7
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
Figu e3. Co ela ion (coe icien o de e mina ion) be ween p edic ed (PLS-SEM model) and obse ed
(s anda dized)- alues o he main quali y c i e ia indica o s o collagen quali y, plus de PGI and CI indices.
8
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
an inc ease in C:N a ios and he PGI. This poin s o selec i e bac e ial deg ada ion o he p o ein componen
(amides and backbone lipids) and he ela i e p ese a ion o ca bohyd a es and lipidic side chains. In ac , he
nega i e coe icien o he in e ac ion be ween he s uc u al componen s (LVs ) and he ca bohyd a es (LVcb)
accoun s o he inc ease in ca bohyd a es as he p o ein componen dec eases, which is consis en wi h he
esul s o he PCA. The a io 1660/1690cm−1, ela ed o ma u i y o collagen c oss-links56, is nega i ely co ela ed
o collagen quali y (LVcq; −0.77, P < 0.01) and posi i ely co ela ed o ca bohyd a es (LVcb) and side-chain
lipids (LVsc) ( 0.67 and 0.79, P < 0.01, espec i ely), also consis en wi h he PCA esul s. I has been p oposed
ha he loss o spec al in ensi y o collagen backbone s uc u es is mos likely ela ed o he agmen a ion o
he molecule due o bac e ial p e e ence o he ela i ely high-ene gy amide bonds21. Al oge he , his ein o ces
he idea ha he main collagen ans o ma ion in he samples analysed he e is con olled by he deg ada ion
o he amide backbone s uc u e. Howe e , i is no possible o assess whe he bac e ial deg ada ion occu ed
du ing body pu e ac ion o la e soil con ac .
Raman analysis o collagen has shown ha dec easing yield is accompanied by disappea ance o amide peaks
bu no necessa ily o alipha ic (C-H) componen s, since poo ly p ese ed collagen samples p oduced spec a
wi h well-de ined alipha ic peaks21. Ano he s udy ound ha changes in amino acid composi ion alone could
no accoun o he ele a ed C:N a ios in low collagen bone om expe imen ally aged human bones18. Mo eo e ,
low-collagen samples a e mo e likely o show ele a ed a ios han con amina ed samples17. Ou esul s a e in line
wi h hese obse a ions since he less in ac collagen samples a e en iched in C- ich compounds (ca bohyd a es
om p o eoglycans and side chain lipids) and hus he C:N is expec ed o inc ease as deg ada ion p og esses.
Al hough he p esence o small amoun s o non-ca bon and non-ni ogen ich con aminan s, as de ec ed in
o he s udies57, canno be dismissed, hei quan i y was no deemed la ge enough o p oduce a de ec able signal
in he spec a.
Ano he in e es ing ea u e is ha he bes -p ese ed samples cha ac e ised by nega i e Cp1 sco es (Fig.4)
show a high co ela ion ( 0.91; P < 0.01) be ween he CI and he PGI (Fig.5): he ela i e abundance o alipha ics
and ca bohyd a es o he amide componen ends o emain cons an . In ou opinion, his esul has po en ial
o he assessmen o collagen ans o ma ion and in eg i y using FTIR-ATR; he la ge he depa u e om he
end he mo e deg aded he collagen s uc u e.
The model also sugges s ha collagen quali y (i.e. C, N, C:N and collagen yield) has no signi ican e ec on
he iso opic composi ion o he collagen. This is also consis en wi h he PCA and co ela ion esul s ob ained
he e and in p e ious in es iga ions, since no co ela ion was ound be ween molecula indica o s o collagen
diagenesis and iso opic composi ion16. O he esea ch also ound ha he iso opic alues (δ13C and δ 15N) and
C:N a ios o he insoluble ac ion emained almos s able un il collagen yield ep esen ed less han 1%18.
We pe o med ANOVA es s on he LV sco es o he PLS-SEM model, using he nec opoleis, a chaeologi-
cal pe iod (B onze Age o Mode n pe iod), bu ial en i onmen (acidic o alkaline), sex (male o emale), ype
o bone and age-a -dea h (< 19, 20–39, 40–59, > 60 es ima ed yea s old) as g ouping a iables. No signi ican
di e ences we e ound o any o he LV sco es (s uc u al componen s, ca bohyd a es, side-chain lipids, and
Figu e4. Cp1-Cp2 p ojec ion o he PCA samples sco es. FC: mass g a e om pos -medie al imes
(se en een h o eigh een h cen u y AD); SMM: San a Ma ía chu ch (Pon e ed a), medie al ceme e y
( hi een h–se en een h cen u y AD); SBA: San Ba olomé medie al chu chya d ( hi een h– i een h cen u y
AD); CP: Capela do Pila , inhuma ions om a chapel o he Lugo Ca hed al (ele en h– ou een h cen u y AD);
CR: Rúa Real, pos -Roman ( i h–se en h cen u y AD) inhuma ion nec opolis; OUV: Ou igo, Ea ly-medie al
( en h– wel h cen u y AD) ceme e y wi h a mino phase o bu ials om he Roman pe iod (second– hi d
cen u y AD); LNZ: A Lanzada, inhuma ions om Roman and pos -Roman imes ( i s –se en h cen u y AD);
CS: Co a do San o (nine een h–six een h cen u y BC), B onze Age human emains ound a a ca e n su ace.
9
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:17888 | h ps://doi.o g/10.1038/s41598-020-74993-y
www.na u e.com/scien i ic epo s/
collagen quali y) o a chaeological pe iod, ype o bone, sex and age-a -dea h (SI_Table3). A chaeological
si e and bu ial con ex p esen ed signi ican di e ences o collagen quali y (LVcq) and s uc u al componen s
(LVs ) and ca bohyd a es (LVcb) o a chaeological si e only. S uc u al componen s (amides and backbone
lipids) con en was highe and ca bohyd a es con en lowe in Co a do San o and Rúa Real compa ed o Capela
do Pila samples; he o he nec opoleis showing in e media e alues be ween hese wo ex emes. As a esul ,
collagen quali y was signi ican ly highe in Co a do San o and Rúa Real han in Capela do Pila . In he la e
case, he good mac oscopic p ese a ion o he skele ons does no ag ee wi h ha sugges ed by he deg ee o
in eg i y o he collagen s uc u e.
As o he bu ial con ex , he alkaline en i onmen s ( he ca e on limes one and he palaeodunes wi h biogenic
ca bona es) showed be e collagen p ese a ion han he acidic ones as ound in p e ious esea ch e.g.58,59.
Al hough no signi ican a P < 0.05, s uc u al componen s and ca bohyd a es we e highe and lowe (P < 0.10)
espec i ely in he alkaline en i onmen s. Thus, alkaline condi ions seem o be he main eason o he good
quali y o he collagen o samples om Co a do San o (limes one ca e) and hose o Rúa Real and A Lanzada
(bu ials on palaeodunes). This is pe haps su p ising gi en he sensi i i y o collagen o hyd olysis unde alkaline
condi ions20. The easons o his appa en disag eemen may be explained by (i) ela i ely low alkalini y in he
bu ial con ex s (pH < 9), he a e o collagen hyd olysis la gely inc easing abo e pH 1120; (ii) well-d ained/ae a ed
condi ions p edomina e; (iii) low decomposi ion o collagen ma ix p e en ing pos -mo em al e a ion in bone
mine al c ys al60,61; and (i ) he dissolu ion o he bone mine al phase is e a ded, limi ing collagen exposi ion
o enzyma ic a ack.
Recen esea ch a A Lanzada concluded ha he in ensi y o bone diagenesis was la ge in bu ials in acidic
soils han hose on palaeodunes, ega dless o he pe iod (Roman o pos -Roman)59. The con ined en i onmen
o Co a do San o ca e could ha e had a la ge e ec han he high pH, as i was obse ed on esea ch made in
ca acombs62. Howe e , he pa icula mine al con en o g oundwa e in his ca e could also ha e p omo ed
collagen p ese a ion63. In ou p e ious s udy o collagen molecula composi ion16, we iden i ied a depolyme i-
za ion p ocess ha di e ed depending on bu ial en i onmen : acidic (soils/sedimen s) showing highe deg ee
o depolyme iza ion han alkaline (sand dunes and limes one ca e) en i onmen s. Acidic condi ions, which
ha e been ound o be he main cause o bioapa i e al e a ion41,59 and p omo ion o collagen dissolu ion64, seem
o be also impo an in he p ese a ion o he p o ein s uc u e— ega dless o he ch onological age. The old-
es bones we e he ones wi h he bes p ese a ion in ou s udy. Finally, pH has been conside ed as pa o “ he
si e hyd ology”—including also he mine al con en o g oundwa e —a much mo e gene al ac o ha con ols
bone p ese a ion65. In ou s udy, well-d ained si es (e.g. palaeodunes, such as he ones om Calle Real and A
Lanzada) and places wi h cons ained wa e mo emen (ca es, as Co a do San o) p o ided he bes condi ions o
p ese a ion. In bo h a eas, g oundwa e is p obably o e sa u a ed o calcium phospha e, which would explain
he good p ese a ion o mine al and o ganic phases o he bone. The humidi y o he soil can also p omo e
bone deg ada ion h ough mic obial and ungal a ack since al e a ion by mic oo ganisms seems o domina e
in empe a e egions63: p.114. Humid condi ions in NW Spain a ou ungi in hose soils nei he well-d ained no
anoxic. In addi ion, bones om Co a do San o we e exposed (no bu ied), which may ha e esul ed in di e en
pos mo em changes61,63.
Despi e hese ese a ions, we conclude ha he e is no single ac o o explain he changes in collagen
s uc u e. All nec opoleis p esen ed ela i ely la ge a ia ions in hei samples´ collagen s uc u al componen s
(Fig.3); i.e. we ound a ange o p ese a ion wi hin popula ions a he han be ween popula ions o well/poo ly
p ese ed collagen. This may indica e ha wi hin any gi en geochemical en i onmen condi ions occu ing
Figu e5. Co ela ion be ween he (s anda dized)- alues o he PGI and CI indices. Samples showing good
collagen in eg i y a e highly co ela ed ( hose i ing he dashed line).
